Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

F/W 4.14.x causes dwc_otg_handle_mode_mismatch_intr:68: messages on boot with C-Media Electronics Inc. USB PnP Sound Device #2408

Closed
hofnarr132 opened this Issue Mar 1, 2018 · 109 comments

Comments

Projects
None yet
@hofnarr132
Copy link

hofnarr132 commented Mar 1, 2018

Looks like an issue with dwc_otg was introduced with 4.14.x versions - 4.9.80 does not show issue.

When booting RPi 3, 2 or ZeroW with 4.14.x versions with a "C-Media Electronics Inc. USB PnP Sound Device" connected** get the following boot messages (extracted from dmesg):

[    3.292306] usb 1-1.3.4: new full-speed USB device number 7 using dwc_otg
[    3.320586] uart-pl011 3f201000.serial: no DMA platform data
[    3.450503] usb 1-1.3.4: New USB device found, idVendor=0d8c, idProduct=013c
[    3.467395] usb 1-1.3.4: New USB device strings: Mfr=1, Product=2, SerialNumber=0
[    3.484570] usb 1-1.3.4: Product: USB PnP Sound Device          
[    3.500214] usb 1-1.3.4: Manufacturer: C-Media Electronics Inc.      
[    3.522058] input: C-Media Electronics Inc.       USB PnP Sound Device           as /devices/platform/soc/3f980000.usb/usb1/1-1/1-1.3/1-1.3.4/1-1.3.4:1.3/0003:0D8C:013C.0003/input/input2
[    3.622573] hid-generic 0003:0D8C:013C.0003: input,hidraw2: USB HID v1.00 Device [C-Media Electronics Inc.       USB PnP Sound Device          ] on usb-3f980000.usb-1.3.4/input3
.
.
.
[    6.170183] bcm2835_alsa bcm2835_alsa: card created with 8 channels
[    6.374754] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode

[    6.374820] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode

[    6.376756] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode

[    6.406816] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode

[    6.410315] usbcore: registered new interface driver snd-usb-audio
[    7.518782] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode

[    7.520726] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode

[    7.712309] usb 1-1.5: reset high-speed USB device number 5 using dwc_otg
.
.
.
[    7.913820] usbcore: registered new interface driver rt2800usb
.
.
.
[   24.342952] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode

[   24.343077] usb 1-1.3.4: 1:1: cannot set freq 44100 to ep 0x1
[   24.432846] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode

[   24.512751] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode

[   24.512823] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode

[   24.526869] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode

[   24.527122] usb 1-1.3.4: 2:1: cannot get freq at ep 0x82
[   24.528811] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode

[   24.576791] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode

[   24.576821] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode

Tested and errors with 4.14.18 (and 4.14.22).

uname -a
Linux raspberrypi 4.14.18-v7+ #1093 SMP Fri Feb 9 15:33:07 GMT 2018 armv7l GNU/Linux

Back revved to 4.9.80 where the error messages do not show:

pi@raspberrypi:~ $ uname -a
Linux raspberrypi 4.9.80-v7+ #1092 SMP Fri Feb 9 13:56:00 GMT 2018 armv7l GNU/Linux

** note: the "C-Media Electronics Inc. USB PnP Sound Device" is actually embedded in a 4-Port HDMI & USB KVM Switch (this one from ModMyPi) - there is a push button switch to turn off audio which disables the Sound Device - the RPi will then boot without error (but then no sound..)

@ajw99uk

This comment has been minimized.

Copy link

ajw99uk commented Mar 10, 2018

Experiencing similar symptoms, at first resolved by unplugging printer from USB hub to boot, but it has occurred without that connected (also unplugging the mouse and keyboard then enabled booting).
Sometimes the error appears a dozen or so times but booting proceeds as normal, sometimes it keeps appearing (requiring a power-off).
My current lsusb shows
Bus 001 Device 009: ID 0d49:3200 Maxtor Personal Storage 3200
Bus 001 Device 008: ID 05e3:0608 Genesys Logic, Inc. Hub
Bus 001 Device 007: ID 05e3:0608 Genesys Logic, Inc. Hub
Bus 001 Device 006: ID 045e:0040 Microsoft Corp. Wheel Mouse Optical
Bus 001 Device 005: ID 04d9:1503 Holtek Semiconductor, Inc. Keyboard
Bus 001 Device 004: ID 05e3:0608 Genesys Logic, Inc. Hub
Bus 001 Device 003: ID 0424:ec00 Standard Microsystems Corp. SMSC9512/9514 Fast Ethernet Adapter
Bus 001 Device 002: ID 0424:9514 Standard Microsystems Corp. SMC9514 Hub
Bus 001 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub

Research suggested this may be a PSU issue, but (i) the USB hub has its own power, (ii) the error appears on a Pi2B with a 2.0A PSU, which should be adequate and (iii) I see no indications of under-powering during use - the system, once booted, is stable.
uname -ar
Linux whitepi 4.14.24-v7+ #1097 SMP Mon Mar 5 16:42:05 GMT 2018 armv7l GNU/Linux

@popcornmix

This comment has been minimized.

Copy link
Collaborator

popcornmix commented Mar 11, 2018

@P33M any ideas on this one?

@P33M

This comment has been minimized.

Copy link
Contributor

P33M commented Mar 12, 2018

Mode mismatch interrupts occur when a Device mode register is read/written while in host mode or vice versa - which is weird because the driver code hasn't changed between 4.9 and 4.14.

As the issue occurs across several Pi models, this is even more confusing.

You have several devices attached. Does the issue occur with just the C-media device attached?

@Hellorheaven

This comment has been minimized.

Copy link

Hellorheaven commented Mar 13, 2018

same problem here
Linux serverhome 4.14.24-v7+ #1097 SMP Mon Mar 5 16:42:05 GMT 2018 armv7l GNU/Linux

@pelwell

This comment has been minimized.

Copy link
Contributor

pelwell commented Mar 13, 2018

Please post the entire dmesg output.

@Hellorheaven

This comment has been minimized.

Copy link

Hellorheaven commented Mar 13, 2018

Mar 11 06:27:03 smarthome liblogging-stdlog:  [origin software="rsyslogd" swVersion="8.24.0" x-pid="306" x-info="http://www.rsyslog.com"] rsyslogd was HUPed
Mar 11 14:00:06 smarthome kernel: [84656.159348] hrtimer: interrupt took 188854 ns
Mar 11 20:57:35 smarthome kernel: [    0.000000] Booting Linux on physical CPU 0x0
Mar 11 20:57:35 smarthome kernel: [    0.000000] Linux version 4.14.24-v7+ (dc4@dc4-XPS13-9333) (gcc version 4.9.3 (crosstool-NG crosstool-ng-1.22.0-88-g8460611)) #1097 SMP Mon Mar 5 16:42:05 GMT 2018
Mar 11 20:57:35 smarthome kernel: [    0.000000] CPU: ARMv7 Processor [410fd034] revision 4 (ARMv7), cr=10c5383d
Mar 11 20:57:35 smarthome kernel: [    0.000000] CPU: div instructions available: patching division code
Mar 11 20:57:35 smarthome kernel: [    0.000000] CPU: PIPT / VIPT nonaliasing data cache, VIPT aliasing instruction cache
Mar 11 20:57:35 smarthome kernel: [    0.000000] OF: fdt: Machine model: Raspberry Pi 3 Model B Rev 1.2
Mar 11 20:57:35 smarthome kernel: [    0.000000] Memory policy: Data cache writealloc
Mar 11 20:57:35 smarthome kernel: [    0.000000] cma: Reserved 8 MiB at 0x3dc00000
Mar 11 20:57:35 smarthome kernel: [    0.000000] random: fast init done
Mar 11 20:57:35 smarthome kernel: [    0.000000] percpu: Embedded 17 pages/cpu @be59f000 s38720 r8192 d22720 u69632
Mar 11 20:57:35 smarthome kernel: [    0.000000] Built 1 zonelists, mobility grouping on.  Total pages: 253242
Mar 11 20:57:35 smarthome kernel: [    0.000000] Kernel command line: 8250.nr_uarts=1 bcm2708_fb.fbwidth=1824 bcm2708_fb.fbheight=984 bcm2708_fb.fbdepth=16 bcm2708_fb.fbswap=1 vc_mem.mem_base=0x3f000000 vc_mem.mem_size=0x3f600000  dwc_otg.lpm_enable=0 console=tty1 root=/dev/sda2 rootfstype=ext4 elevator=deadline fsck.repair=yes rootwait
Mar 11 20:57:35 smarthome kernel: [    0.000000] PID hash table entries: 4096 (order: 2, 16384 bytes)
Mar 11 20:57:35 smarthome kernel: [    0.000000] Dentry cache hash table entries: 131072 (order: 7, 524288 bytes)
Mar 11 20:57:35 smarthome kernel: [    0.000000] Inode-cache hash table entries: 65536 (order: 6, 262144 bytes)
Mar 11 20:57:35 smarthome kernel: [    0.000000] Memory: 990992K/1021952K available (7168K kernel code, 548K rwdata, 2028K rodata, 1024K init, 706K bss, 22768K reserved, 8192K cma-reserved)
Mar 11 20:57:35 smarthome kernel: [    0.000000] Virtual kernel memory layout:
Mar 11 20:57:35 smarthome kernel: [    0.000000]     vector  : 0xffff0000 - 0xffff1000   (   4 kB)
Mar 11 20:57:35 smarthome kernel: [    0.000000]     fixmap  : 0xffc00000 - 0xfff00000   (3072 kB)
Mar 11 20:57:35 smarthome kernel: [    0.000000]     vmalloc : 0xbe800000 - 0xff800000   (1040 MB)
Mar 11 20:57:35 smarthome kernel: [    0.000000]     lowmem  : 0x80000000 - 0xbe600000   ( 998 MB)
Mar 11 20:57:35 smarthome kernel: [    0.000000]     modules : 0x7f000000 - 0x80000000   (  16 MB)
Mar 11 20:57:35 smarthome kernel: [    0.000000]       .text : 0x80008000 - 0x80800000   (8160 kB)
Mar 11 20:57:35 smarthome kernel: [    0.000000]       .init : 0x80b00000 - 0x80c00000   (1024 kB)
Mar 11 20:57:35 smarthome kernel: [    0.000000]       .data : 0x80c00000 - 0x80c89244   ( 549 kB)
Mar 11 20:57:35 smarthome kernel: [    0.000000]        .bss : 0x80c90ed4 - 0x80d417b4   ( 707 kB)
Mar 11 20:57:35 smarthome kernel: [    0.000000] SLUB: HWalign=64, Order=0-3, MinObjects=0, CPUs=4, Nodes=1
Mar 11 20:57:35 smarthome kernel: [    0.000000] ftrace: allocating 24646 entries in 73 pages
Mar 11 20:57:35 smarthome kernel: [    0.000000] Hierarchical RCU implementation.
Mar 11 20:57:35 smarthome kernel: [    0.000000] NR_IRQS: 16, nr_irqs: 16, preallocated irqs: 16
Mar 11 20:57:35 smarthome kernel: [    0.000000] arch_timer: cp15 timer(s) running at 19.20MHz (phys).
Mar 11 20:57:35 smarthome kernel: [    0.000000] clocksource: arch_sys_counter: mask: 0xffffffffffffff max_cycles: 0x46d987e47, max_idle_ns: 440795202767 ns
Mar 11 20:57:35 smarthome kernel: [    0.000007] sched_clock: 56 bits at 19MHz, resolution 52ns, wraps every 4398046511078ns
Mar 11 20:57:35 smarthome kernel: [    0.000022] Switching to timer-based delay loop, resolution 52ns
Mar 11 20:57:35 smarthome kernel: [    0.000273] Console: colour dummy device 80x30
Mar 11 20:57:35 smarthome kernel: [    0.000817] console [tty1] enabled
Mar 11 20:57:35 smarthome kernel: [    0.000857] Calibrating delay loop (skipped), value calculated using timer frequency.. 38.40 BogoMIPS (lpj=192000)
Mar 11 20:57:35 smarthome kernel: [    0.000897] pid_max: default: 32768 minimum: 301
Mar 11 20:57:35 smarthome kernel: [    0.001221] Mount-cache hash table entries: 2048 (order: 1, 8192 bytes)
Mar 11 20:57:35 smarthome kernel: [    0.001254] Mountpoint-cache hash table entries: 2048 (order: 1, 8192 bytes)
Mar 11 20:57:35 smarthome kernel: [    0.002203] Disabling memory control group subsystem
Mar 11 20:57:35 smarthome kernel: [    0.002300] CPU: Testing write buffer coherency: ok
Mar 11 20:57:35 smarthome kernel: [    0.002710] CPU0: thread -1, cpu 0, socket 0, mpidr 80000000
Mar 11 20:57:35 smarthome kernel: [    0.003104] Setting up static identity map for 0x100000 - 0x10003c
Mar 11 20:57:35 smarthome kernel: [    0.003238] Hierarchical SRCU implementation.
Mar 11 20:57:35 smarthome kernel: [    0.003887] smp: Bringing up secondary CPUs ...
Mar 11 20:57:35 smarthome kernel: [    0.004539] CPU1: thread -1, cpu 1, socket 0, mpidr 80000001
Mar 11 20:57:35 smarthome kernel: [    0.005253] CPU2: thread -1, cpu 2, socket 0, mpidr 80000002
Mar 11 20:57:35 smarthome kernel: [    0.005956] CPU3: thread -1, cpu 3, socket 0, mpidr 80000003
Mar 11 20:57:35 smarthome kernel: [    0.006057] smp: Brought up 1 node, 4 CPUs
Mar 11 20:57:35 smarthome kernel: [    0.006127] SMP: Total of 4 processors activated (153.60 BogoMIPS).
Mar 11 20:57:35 smarthome kernel: [    0.006148] CPU: All CPU(s) started in HYP mode.
Mar 11 20:57:35 smarthome kernel: [    0.006166] CPU: Virtualization extensions available.
Mar 11 20:57:35 smarthome kernel: [    0.007036] devtmpfs: initialized
Mar 11 20:57:35 smarthome kernel: [    0.017488] VFP support v0.3: implementor 41 architecture 3 part 40 variant 3 rev 4
Mar 11 20:57:35 smarthome kernel: [    0.017736] clocksource: jiffies: mask: 0xffffffff max_cycles: 0xffffffff, max_idle_ns: 19112604462750000 ns
Mar 11 20:57:35 smarthome kernel: [    0.017780] futex hash table entries: 1024 (order: 4, 65536 bytes)
Mar 11 20:57:35 smarthome kernel: [    0.018354] pinctrl core: initialized pinctrl subsystem
Mar 11 20:57:35 smarthome kernel: [    0.019126] NET: Registered protocol family 16
Mar 11 20:57:35 smarthome kernel: [    0.021676] DMA: preallocated 1024 KiB pool for atomic coherent allocations
Mar 11 20:57:35 smarthome kernel: [    0.026383] hw-breakpoint: found 5 (+1 reserved) breakpoint and 4 watchpoint registers.
Mar 11 20:57:35 smarthome kernel: [    0.026415] hw-breakpoint: maximum watchpoint size is 8 bytes.
Mar 11 20:57:35 smarthome kernel: [    0.026633] Serial: AMBA PL011 UART driver
Mar 11 20:57:35 smarthome kernel: [    0.028280] bcm2835-mbox 3f00b880.mailbox: mailbox enabled
Mar 11 20:57:35 smarthome kernel: [    0.028758] uart-pl011 3f201000.serial: could not find pctldev for node /soc/gpio@7e200000/uart0_pins, deferring probe
Mar 11 20:57:35 smarthome kernel: [    0.060384] bcm2835-dma 3f007000.dma: DMA legacy API manager at be80d000, dmachans=0x1
Mar 11 20:57:35 smarthome kernel: [    0.061834] SCSI subsystem initialized
Mar 11 20:57:35 smarthome kernel: [    0.061981] usbcore: registered new interface driver usbfs
Mar 11 20:57:35 smarthome kernel: [    0.062050] usbcore: registered new interface driver hub
Mar 11 20:57:35 smarthome kernel: [    0.062150] usbcore: registered new device driver usb
Mar 11 20:57:35 smarthome kernel: [    0.070074] raspberrypi-firmware soc:firmware: Attached to firmware from 2018-03-09 19:56
Mar 11 20:57:35 smarthome kernel: [    0.071340] clocksource: Switched to clocksource arch_sys_counter
Mar 11 20:57:35 smarthome kernel: [    0.147714] VFS: Disk quotas dquot_6.6.0
Mar 11 20:57:35 smarthome kernel: [    0.147827] VFS: Dquot-cache hash table entries: 1024 (order 0, 4096 bytes)
Mar 11 20:57:35 smarthome kernel: [    0.148032] FS-Cache: Loaded
Mar 11 20:57:35 smarthome kernel: [    0.148240] CacheFiles: Loaded
Mar 11 20:57:35 smarthome kernel: [    0.157064] NET: Registered protocol family 2
Mar 11 20:57:35 smarthome kernel: [    0.157810] TCP established hash table entries: 8192 (order: 3, 32768 bytes)
Mar 11 20:57:35 smarthome kernel: [    0.157940] TCP bind hash table entries: 8192 (order: 4, 65536 bytes)
Mar 11 20:57:35 smarthome kernel: [    0.158141] TCP: Hash tables configured (established 8192 bind 8192)
Mar 11 20:57:35 smarthome kernel: [    0.158288] UDP hash table entries: 512 (order: 2, 16384 bytes)
Mar 11 20:57:35 smarthome kernel: [    0.158350] UDP-Lite hash table entries: 512 (order: 2, 16384 bytes)
Mar 11 20:57:35 smarthome kernel: [    0.158595] NET: Registered protocol family 1
Mar 11 20:57:35 smarthome kernel: [    0.159046] RPC: Registered named UNIX socket transport module.
Mar 11 20:57:35 smarthome kernel: [    0.159070] RPC: Registered udp transport module.
Mar 11 20:57:35 smarthome kernel: [    0.159089] RPC: Registered tcp transport module.
Mar 11 20:57:35 smarthome kernel: [    0.159107] RPC: Registered tcp NFSv4.1 backchannel transport module.
Mar 11 20:57:35 smarthome kernel: [    0.160523] hw perfevents: enabled with armv7_cortex_a7 PMU driver, 7 counters available
Mar 11 20:57:35 smarthome kernel: [    0.163370] workingset: timestamp_bits=14 max_order=18 bucket_order=4
Mar 11 20:57:35 smarthome kernel: [    0.171432] FS-Cache: Netfs 'nfs' registered for caching
Mar 11 20:57:35 smarthome kernel: [    0.172047] NFS: Registering the id_resolver key type
Mar 11 20:57:35 smarthome kernel: [    0.172099] Key type id_resolver registered
Mar 11 20:57:35 smarthome kernel: [    0.172118] Key type id_legacy registered
Mar 11 20:57:35 smarthome kernel: [    0.174004] Block layer SCSI generic (bsg) driver version 0.4 loaded (major 251)
Mar 11 20:57:35 smarthome kernel: [    0.174153] io scheduler noop registered
Mar 11 20:57:35 smarthome kernel: [    0.174174] io scheduler deadline registered (default)
Mar 11 20:57:35 smarthome kernel: [    0.174468] io scheduler cfq registered
Mar 11 20:57:35 smarthome kernel: [    0.174489] io scheduler mq-deadline registered
Mar 11 20:57:35 smarthome kernel: [    0.174509] io scheduler kyber registered
Mar 11 20:57:35 smarthome kernel: [    0.177235] BCM2708FB: allocated DMA memory fdd10000
Mar 11 20:57:35 smarthome kernel: [    0.177279] BCM2708FB: allocated DMA channel 0 @ be80d000
Mar 11 20:57:35 smarthome kernel: [    0.203483] Console: switching to colour frame buffer device 228x61
Mar 11 20:57:35 smarthome kernel: [    0.219406] Serial: 8250/16550 driver, 1 ports, IRQ sharing enabled
Mar 11 20:57:35 smarthome kernel: [    0.220832] bcm2835-rng 3f104000.rng: hwrng registered
Mar 11 20:57:35 smarthome kernel: [    0.221051] vc-mem: phys_addr:0x00000000 mem_base=0x3f000000 mem_size:0x3f600000(1014 MiB)
Mar 11 20:57:35 smarthome kernel: [    0.221684] vc-sm: Videocore shared memory driver
Mar 11 20:57:35 smarthome mtp-probe: checking bus 1, device 3: "/sys/devices/platform/soc/3f980000.usb/usb1/1-1/1-1.1"
Mar 11 20:57:35 smarthome mtp-probe: checking bus 1, device 7: "/sys/devices/platform/soc/3f980000.usb/usb1/1-1/1-1.5"
Mar 11 20:57:35 smarthome mtp-probe: bus: 1, device: 7 was not an MTP device
Mar 11 20:57:35 smarthome mtp-probe: checking bus 1, device 8: "/sys/devices/platform/soc/3f980000.usb/usb1/1-1/1-1.4/1-1.4.3"
Mar 11 20:57:35 smarthome mtp-probe: checking bus 1, device 6: "/sys/devices/platform/soc/3f980000.usb/usb1/1-1/1-1.4/1-1.4.2"
Mar 11 20:57:35 smarthome mtp-probe: bus: 1, device: 8 was not an MTP device
Mar 11 20:57:35 smarthome mtp-probe: bus: 1, device: 6 was not an MTP device
Mar 11 20:57:35 smarthome mtp-probe: checking bus 1, device 5: "/sys/devices/platform/soc/3f980000.usb/usb1/1-1/1-1.4/1-1.4.1"
Mar 11 20:57:35 smarthome mtp-probe: bus: 1, device: 3 was not an MTP device
Mar 11 20:57:35 smarthome mtp-probe: bus: 1, device: 5 was not an MTP device
Mar 11 20:57:35 smarthome kernel: [    0.222038] gpiomem-bcm2835 3f200000.gpiomem: Initialised: Registers at 0x3f200000
Mar 11 20:57:35 smarthome kernel: [    0.231715] brd: module loaded
Mar 11 20:57:35 smarthome kernel: [    0.240434] loop: module loaded
Mar 11 20:57:35 smarthome kernel: [    0.240530] Loading iSCSI transport class v2.0-870.
Mar 11 20:57:35 smarthome kernel: [    0.241106] usbcore: registered new interface driver smsc95xx
Mar 11 20:57:35 smarthome kernel: [    0.241222] dwc_otg: version 3.00a 10-AUG-2012 (platform bus)
Mar 11 20:57:35 smarthome kernel: [    0.467042] Core Release: 2.80a
Mar 11 20:57:35 smarthome kernel: [    0.467112] Setting default values for core params
Mar 11 20:57:35 smarthome kernel: [    0.467222] Finished setting default values for core params
Mar 11 20:57:35 smarthome kernel: [    0.667568] Using Buffer DMA mode
Mar 11 20:57:35 smarthome kernel: [    0.667636] Periodic Transfer Interrupt Enhancement - disabled
Mar 11 20:57:35 smarthome kernel: [    0.667733] Multiprocessor Interrupt Enhancement - disabled
Mar 11 20:57:35 smarthome kernel: [    0.667828] OTG VER PARAM: 0, OTG VER FLAG: 0
Mar 11 20:57:35 smarthome kernel: [    0.667911] Dedicated Tx FIFOs mode
Mar 11 20:57:35 smarthome kernel: [    0.668315] WARN::dwc_otg_hcd_init:1046: FIQ DMA bounce buffers: virt = 0xbdd04000 dma = 0xfdd04000 len=9024
Mar 11 20:57:35 smarthome kernel: [    0.668491] FIQ FSM acceleration enabled for :
Mar 11 20:57:35 smarthome kernel: [    0.668491] Non-periodic Split Transactions
Mar 11 20:57:35 smarthome kernel: [    0.668491] Periodic Split Transactions
Mar 11 20:57:35 smarthome kernel: [    0.668491] High-Speed Isochronous Endpoints
Mar 11 20:57:35 smarthome kernel: [    0.668491] Interrupt/Control Split Transaction hack enabled
Mar 11 20:57:35 smarthome kernel: [    0.668882] WARN::hcd_init_fiq:459: FIQ on core 1 at 0x805c68fc
Mar 11 20:57:35 smarthome kernel: [    0.668991] WARN::hcd_init_fiq:460: FIQ ASM at 0x805c6c64 length 36
Mar 11 20:57:35 smarthome kernel: [    0.673127] WARN::hcd_init_fiq:486: MPHI regs_base at 0xbe878000
Mar 11 20:57:35 smarthome kernel: [    0.677294] dwc_otg 3f980000.usb: DWC OTG Controller
Mar 11 20:57:35 smarthome kernel: [    0.681468] dwc_otg 3f980000.usb: new USB bus registered, assigned bus number 1
Mar 11 20:57:35 smarthome kernel: [    0.685633] dwc_otg 3f980000.usb: irq 62, io mem 0x00000000
Mar 11 20:57:35 smarthome kernel: [    0.689776] Init: Port Power? op_state=1
Mar 11 20:57:35 smarthome kernel: [    0.693865] Init: Power Port (0)
Mar 11 20:57:35 smarthome kernel: [    0.698089] usb usb1: New USB device found, idVendor=1d6b, idProduct=0002
Mar 11 20:57:35 smarthome kernel: [    0.702199] usb usb1: New USB device strings: Mfr=3, Product=2, SerialNumber=1
Mar 11 20:57:35 smarthome kernel: [    0.706279] usb usb1: Product: DWC OTG Controller
Mar 11 20:57:35 smarthome kernel: [    0.710338] usb usb1: Manufacturer: Linux 4.14.24-v7+ dwc_otg_hcd
Mar 11 20:57:35 smarthome kernel: [    0.714445] usb usb1: SerialNumber: 3f980000.usb
Mar 11 20:57:35 smarthome kernel: [    0.719176] hub 1-0:1.0: USB hub found
Mar 11 20:57:35 smarthome kernel: [    0.723332] hub 1-0:1.0: 1 port detected
Mar 11 20:57:35 smarthome kernel: [    0.728196] usbcore: registered new interface driver usb-storage
Mar 11 20:57:35 smarthome kernel: [    0.732444] mousedev: PS/2 mouse device common for all mice
Mar 11 20:57:35 smarthome kernel: [    0.736541] IR NEC protocol handler initialized
Mar 11 20:57:35 smarthome kernel: [    0.740543] IR RC5(x/sz) protocol handler initialized
Mar 11 20:57:35 smarthome kernel: [    0.744564] IR RC6 protocol handler initialized
Mar 11 20:57:35 smarthome kernel: [    0.748557] IR JVC protocol handler initialized
Mar 11 20:57:35 smarthome kernel: [    0.752543] IR Sony protocol handler initialized
Mar 11 20:57:35 smarthome kernel: [    0.756483] IR SANYO protocol handler initialized
Mar 11 20:57:35 smarthome kernel: [    0.760374] IR Sharp protocol handler initialized
Mar 11 20:57:35 smarthome kernel: [    0.764243] IR MCE Keyboard/mouse protocol handler initialized
Mar 11 20:57:35 smarthome kernel: [    0.768138] IR XMP protocol handler initialized
Mar 11 20:57:35 smarthome kernel: [    0.772739] bcm2835-wdt 3f100000.watchdog: Broadcom BCM2835 watchdog timer
Mar 11 20:57:35 smarthome kernel: [    0.776982] bcm2835-cpufreq: min=600000 max=1200000
Mar 11 20:57:35 smarthome kernel: [    0.781354] sdhci: Secure Digital Host Controller Interface driver
Mar 11 20:57:35 smarthome kernel: [    0.785378] sdhci: Copyright(c) Pierre Ossman
Mar 11 20:57:35 smarthome kernel: [    0.789767] mmc-bcm2835 3f300000.mmc: could not get clk, deferring probe
Mar 11 20:57:35 smarthome kernel: [    0.794170] sdhost-bcm2835 3f202000.mmc: could not get clk, deferring probe
Mar 11 20:57:35 smarthome kernel: [    0.798344] sdhci-pltfm: SDHCI platform and OF driver helper
Mar 11 20:57:35 smarthome kernel: [    0.803496] ledtrig-cpu: registered to indicate activity on CPUs
Mar 11 20:57:35 smarthome kernel: [    0.807717] hidraw: raw HID events driver (C) Jiri Kosina
Mar 11 20:57:35 smarthome kernel: [    0.811906] usbcore: registered new interface driver usbhid
Mar 11 20:57:35 smarthome kernel: [    0.815893] usbhid: USB HID core driver
Mar 11 20:57:35 smarthome kernel: [    0.820418] vchiq: vchiq_init_state: slot_zero = bdd80000, is_master = 0
Mar 11 20:57:35 smarthome kernel: [    0.825819] [vc_sm_connected_init]: start
Mar 11 20:57:35 smarthome kernel: [    0.838188] [vc_sm_connected_init]: end - returning -1
Mar 11 20:57:35 smarthome kernel: [    0.843009] Initializing XFRM netlink socket
Mar 11 20:57:35 smarthome kernel: [    0.847062] NET: Registered protocol family 17
Mar 11 20:57:35 smarthome kernel: [    0.851170] Key type dns_resolver registered
Mar 11 20:57:35 smarthome kernel: [    0.855573] Registering SWP/SWPB emulation handler
Mar 11 20:57:35 smarthome kernel: [    0.860177] registered taskstats version 1
Mar 11 20:57:35 smarthome kernel: [    0.870086] uart-pl011 3f201000.serial: cts_event_workaround enabled
Mar 11 20:57:35 smarthome kernel: [    0.874258] 3f201000.serial: ttyAMA0 at MMIO 0x3f201000 (irq = 87, base_baud = 0) is a PL011 rev2
Mar 11 20:57:35 smarthome kernel: [    0.879984] 3f215040.serial: ttyS0 at MMIO 0x0 (irq = 166, base_baud = 31250000) is a 16550
Mar 11 20:57:35 smarthome kernel: [    0.884982] mmc-bcm2835 3f300000.mmc: mmc_debug:0 mmc_debug2:0
Mar 11 20:57:35 smarthome kernel: [    0.889003] mmc-bcm2835 3f300000.mmc: DMA channel allocated
Mar 11 20:57:35 smarthome kernel: [    0.952002] sdhost: log_buf @ bdd07000 (fdd07000)
Mar 11 20:57:35 smarthome kernel: [    0.957275] Indeed it is in host mode hprt0 = 00021501
Mar 11 20:57:35 smarthome kernel: [    1.057689] mmc1: queuing unknown CIS tuple 0x80 (2 bytes)
Mar 11 20:57:35 smarthome kernel: [    1.063323] mmc1: queuing unknown CIS tuple 0x80 (3 bytes)
Mar 11 20:57:35 smarthome kernel: [    1.068875] mmc1: queuing unknown CIS tuple 0x80 (3 bytes)
Mar 11 20:57:35 smarthome kernel: [    1.075688] mmc1: queuing unknown CIS tuple 0x80 (7 bytes)
Mar 11 20:57:35 smarthome kernel: [    1.101363] mmc0: sdhost-bcm2835 loaded - DMA enabled (>1)
Mar 11 20:57:35 smarthome kernel: [    1.106328] of_cfs_init
Mar 11 20:57:35 smarthome kernel: [    1.110348] of_cfs_init: OK
Mar 11 20:57:35 smarthome kernel: [    1.114742] Waiting for root device /dev/sda2...
Mar 11 20:57:35 smarthome kernel: [    1.191381] usb 1-1: new high-speed USB device number 2 using dwc_otg
Mar 11 20:57:35 smarthome kernel: [    1.195441] Indeed it is in host mode hprt0 = 00001101
Mar 11 20:57:35 smarthome kernel: [    1.289132] mmc1: new high speed SDIO card at address 0001
Mar 11 20:57:35 smarthome kernel: [    1.451607] usb 1-1: New USB device found, idVendor=0424, idProduct=9514
Mar 11 20:57:35 smarthome kernel: [    1.455555] usb 1-1: New USB device strings: Mfr=0, Product=0, SerialNumber=0
Mar 11 20:57:35 smarthome kernel: [    1.460048] hub 1-1:1.0: USB hub found
Mar 11 20:57:35 smarthome kernel: [    1.484149] hub 1-1:1.0: 5 ports detected
Mar 11 20:57:35 smarthome kernel: [    1.801366] usb 1-1.1: new high-speed USB device number 3 using dwc_otg
Mar 11 20:57:35 smarthome kernel: [    1.931578] usb 1-1.1: New USB device found, idVendor=0424, idProduct=ec00
Mar 11 20:57:35 smarthome kernel: [    1.935562] usb 1-1.1: New USB device strings: Mfr=0, Product=0, SerialNumber=0
Mar 11 20:57:35 smarthome kernel: [    1.942080] smsc95xx v1.0.6
Mar 11 20:57:35 smarthome kernel: [    2.034419] smsc95xx 1-1.1:1.0 eth0: register 'smsc95xx' at usb-3f980000.usb-1.1, smsc95xx USB 2.0 Ethernet, b8:27:eb:0a:36:17
Mar 11 20:57:35 smarthome kernel: [    2.131370] usb 1-1.4: new high-speed USB device number 4 using dwc_otg
Mar 11 20:57:35 smarthome kernel: [    2.261589] usb 1-1.4: New USB device found, idVendor=0409, idProduct=0059
Mar 11 20:57:35 smarthome kernel: [    2.265802] usb 1-1.4: New USB device strings: Mfr=0, Product=0, SerialNumber=0
Mar 11 20:57:35 smarthome kernel: [    2.270476] hub 1-1.4:1.0: USB hub found
Mar 11 20:57:35 smarthome kernel: [    2.274774] hub 1-1.4:1.0: 4 ports detected
Mar 11 20:57:35 smarthome kernel: [    2.591363] usb 1-1.4.1: new full-speed USB device number 5 using dwc_otg
Mar 11 20:57:35 smarthome kernel: [    2.754624] usb 1-1.4.1: New USB device found, idVendor=0658, idProduct=0200
Mar 11 20:57:35 smarthome kernel: [    2.758908] usb 1-1.4.1: New USB device strings: Mfr=0, Product=0, SerialNumber=0
Mar 11 20:57:35 smarthome kernel: [    2.861369] usb 1-1.4.2: new full-speed USB device number 6 using dwc_otg
Mar 11 20:57:35 smarthome kernel: [    3.022992] usb 1-1.4.2: New USB device found, idVendor=0403, idProduct=6001
Mar 11 20:57:35 smarthome kernel: [    3.027259] usb 1-1.4.2: New USB device strings: Mfr=1, Product=2, SerialNumber=3
Mar 11 20:57:35 smarthome kernel: [    3.031539] usb 1-1.4.2: Product: Interface USB -> Compteur
Mar 11 20:57:35 smarthome kernel: [    3.035762] usb 1-1.4.2: Manufacturer: Cartelectronic
Mar 11 20:57:35 smarthome kernel: [    3.039941] usb 1-1.4.2: SerialNumber: C10344
Mar 11 20:57:35 smarthome kernel: [    3.091370] usb 1-1.5: new high-speed USB device number 7 using dwc_otg
Mar 11 20:57:35 smarthome kernel: [    3.222955] usb 1-1.5: New USB device found, idVendor=174c, idProduct=55aa
Mar 11 20:57:35 smarthome kernel: [    3.227129] usb 1-1.5: New USB device strings: Mfr=2, Product=3, SerialNumber=1
Mar 11 20:57:35 smarthome kernel: [    3.231295] usb 1-1.5: Product: AS2115
Mar 11 20:57:35 smarthome kernel: [    3.235449] usb 1-1.5: Manufacturer: ASMedia
Mar 11 20:57:35 smarthome kernel: [    3.239561] usb 1-1.5: SerialNumber: 12345678A106
Mar 11 20:57:35 smarthome kernel: [    3.244245] usb-storage 1-1.5:1.0: USB Mass Storage device detected
Mar 11 20:57:35 smarthome kernel: [    3.248746] usb-storage 1-1.5:1.0: Quirks match for vid 174c pid 55aa: 400000
Mar 11 20:57:35 smarthome kernel: [    3.253093] scsi host0: usb-storage 1-1.5:1.0
Mar 11 20:57:35 smarthome kernel: [    3.321367] usb 1-1.4.3: new full-speed USB device number 8 using dwc_otg
Mar 11 20:57:35 smarthome kernel: [    3.478868] usb 1-1.4.3: New USB device found, idVendor=0403, idProduct=6015
Mar 11 20:57:35 smarthome kernel: [    3.483041] usb 1-1.4.3: New USB device strings: Mfr=1, Product=2, SerialNumber=3
Mar 11 20:57:35 smarthome kernel: [    3.487193] usb 1-1.4.3: Product: FT230X Basic UART
Mar 11 20:57:35 smarthome kernel: [    3.491352] usb 1-1.4.3: Manufacturer: FTDI
Mar 11 20:57:35 smarthome kernel: [    3.495490] usb 1-1.4.3: SerialNumber: DO00KH2J
Mar 11 20:57:35 smarthome kernel: [    4.312116] scsi 0:0:0:0: Direct-Access     ASMT     2115             0    PQ: 0 ANSI: 6
Mar 11 20:57:35 smarthome kernel: [    4.318038] sd 0:0:0:0: [sda] 117231408 512-byte logical blocks: (60.0 GB/55.9 GiB)
Mar 11 20:57:35 smarthome kernel: [    4.322586] sd 0:0:0:0: [sda] Write Protect is off
Mar 11 20:57:35 smarthome kernel: [    4.327045] sd 0:0:0:0: [sda] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA
Mar 11 20:57:35 smarthome kernel: [    4.333258]  sda: sda1 sda2
Mar 11 20:57:35 smarthome kernel: [    4.339315] sd 0:0:0:0: [sda] Attached SCSI disk
Mar 11 20:57:35 smarthome kernel: [    4.374750] EXT4-fs (sda2): mounted filesystem with ordered data mode. Opts: (null)
Mar 11 20:57:35 smarthome kernel: [    4.379029] VFS: Mounted root (ext4 filesystem) readonly on device 8:2.
Mar 11 20:57:35 smarthome kernel: [    4.388397] devtmpfs: mounted
Mar 11 20:57:35 smarthome kernel: [    4.397216] Freeing unused kernel memory: 1024K
Mar 11 20:57:35 smarthome kernel: [    4.748058] NET: Registered protocol family 10
Mar 11 20:57:35 smarthome kernel: [    4.753864] Segment Routing with IPv6
Mar 11 20:57:35 smarthome kernel: [    4.769508] ip_tables: (C) 2000-2006 Netfilter Core Team
Mar 11 20:57:35 smarthome kernel: [    5.471265] Installing knfsd (copyright (C) 1996 okir@monad.swb.de).
Mar 11 20:57:35 smarthome kernel: [    6.021729] EXT4-fs (sda2): re-mounted. Opts: (null)
Mar 11 20:57:35 smarthome kernel: [    6.572012] random: crng init done
Mar 11 20:57:35 smarthome kernel: [    6.591177] snd_bcm2835: module is from the staging directory, the quality is unknown, you have been warned.
Mar 11 20:57:35 smarthome kernel: [    6.601764] bcm2835_alsa bcm2835_alsa: card created with 8 channels
Mar 11 20:57:35 smarthome kernel: [    6.721929] usbcore: registered new interface driver brcmfmac
Mar 11 20:57:35 smarthome kernel: [    7.066945] usbcore: registered new interface driver usbserial
Mar 11 20:57:35 smarthome kernel: [    7.067017] usbcore: registered new interface driver usbserial_generic
Mar 11 20:57:35 smarthome kernel: [    7.067076] usbserial: USB Serial support registered for generic
Mar 11 20:57:35 smarthome kernel: [    7.080609] sd 0:0:0:0: Attached scsi generic sg0 type 0
Mar 11 20:57:35 smarthome liblogging-stdlog:  [origin software="rsyslogd" swVersion="8.24.0" x-pid="300" x-info="http://www.rsyslog.com"] start
Mar 11 20:57:35 smarthome kernel: [    7.087742] usbcore: registered new interface driver ftdi_sio
Mar 11 20:57:35 smarthome kernel: [    7.087814] usbserial: USB Serial support registered for FTDI USB Serial Device
Mar 11 20:57:35 smarthome kernel: [    7.088305] ftdi_sio 1-1.4.2:1.0: FTDI USB Serial Device converter detected
Mar 11 20:57:35 smarthome kernel: [    7.088463] usb 1-1.4.2: Detected FT232RL
Mar 11 20:57:35 smarthome kernel: [    7.090384] usb 1-1.4.2: FTDI USB Serial Device converter now attached to ttyUSB0
Mar 11 20:57:35 smarthome kernel: [    7.099593] ftdi_sio 1-1.4.3:1.0: FTDI USB Serial Device converter detected
Mar 11 20:57:35 smarthome kernel: [    7.099778] usb 1-1.4.3: Detected FT-X
Mar 11 20:57:35 smarthome kernel: [    7.100882] smsc95xx 1-1.1:1.0 enxb827eb0a3617: renamed from eth0
Mar 11 20:57:35 smarthome kernel: [    7.101754] usb 1-1.4.3: FTDI USB Serial Device converter now attached to ttyUSB1
Mar 11 20:57:35 smarthome kernel: [    7.115314] cdc_acm 1-1.4.1:1.0: ttyACM0: USB ACM device
Mar 11 20:57:35 smarthome kernel: [    7.116533] usbcore: registered new interface driver cdc_acm
Mar 11 20:57:35 smarthome kernel: [    7.116546] cdc_acm: USB Abstract Control Model driver for USB modems and ISDN adapters
Mar 11 20:57:35 smarthome kernel: [    8.005503] uart-pl011 3f201000.serial: no DMA platform data
Mar 11 20:57:35 smarthome kernel: [    8.321757] ip6_tables: (C) 2000-2006 Netfilter Core Team
Mar 11 20:57:36 smarthome kernel: [    8.942215] IPv6: ADDRCONF(NETDEV_UP): wlan0: link is not ready
Mar 11 20:57:36 smarthome kernel: [    8.942231] brcmfmac: power management disabled
Mar 11 20:57:36 smarthome kernel: [    9.476796] smsc95xx 1-1.1:1.0 enxb827eb0a3617: hardware isn't capable of remote wakeup
Mar 11 20:57:38 smarthome kernel: [   10.969636] IPv6: ADDRCONF(NETDEV_CHANGE): wlan0: link becomes ready
Mar 11 20:57:38 smarthome kernel: [   10.981280] smsc95xx 1-1.1:1.0 enxb827eb0a3617: link up, 100Mbps, full-duplex, lpa 0xC1E1
Mar 11 20:57:40 smarthome kernel: [   13.123216] Bluetooth: Core ver 2.22
Mar 11 20:57:40 smarthome kernel: [   13.123346] NET: Registered protocol family 31
Mar 11 20:57:40 smarthome kernel: [   13.123355] Bluetooth: HCI device and connection manager initialized
Mar 11 20:57:40 smarthome kernel: [   13.124941] Bluetooth: HCI socket layer initialized
Mar 11 20:57:40 smarthome kernel: [   13.124971] Bluetooth: L2CAP socket layer initialized
Mar 11 20:57:40 smarthome kernel: [   13.125032] Bluetooth: SCO socket layer initialized
Mar 11 20:57:40 smarthome kernel: [   13.133157] Bluetooth: HCI UART driver ver 2.3
Mar 11 20:57:40 smarthome kernel: [   13.133171] Bluetooth: HCI UART protocol H4 registered
Mar 11 20:57:40 smarthome kernel: [   13.133176] Bluetooth: HCI UART protocol Three-wire (H5) registered
Mar 11 20:57:40 smarthome kernel: [   13.276277] Bluetooth: BNEP (Ethernet Emulation) ver 1.3
Mar 11 20:57:40 smarthome kernel: [   13.276289] Bluetooth: BNEP filters: protocol multicast
Mar 11 20:57:40 smarthome kernel: [   13.276309] Bluetooth: BNEP socket layer initialized
Mar 11 20:57:45 smarthome kernel: [   17.949447] NFSD: starting 90-second grace period (net 80c74080)
Mar 11 20:57:46 smarthome kernel: [   18.720817] Adding 1048572k swap on /var/swap.  Priority:-2 extents:13 across:2154492k FS
Mar 11 21:11:21 smarthome kernel: [  713.041179] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 11 21:11:21 smarthome kernel: [  713.041179]
Mar 11 22:37:02 smarthome kernel: [ 5853.458563] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 11 22:37:02 smarthome kernel: [ 5853.458563]
Mar 11 22:37:02 smarthome kernel: [ 5853.458636] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 11 22:37:02 smarthome kernel: [ 5853.458636]
Mar 11 22:49:30 smarthome kernel: [ 6601.769915] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 11 22:49:30 smarthome kernel: [ 6601.769915]
Mar 11 22:50:30 smarthome kernel: [ 6661.320009] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 11 22:50:30 smarthome kernel: [ 6661.320009]
Mar 11 22:55:57 smarthome kernel: [ 6989.040563] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 11 22:55:57 smarthome kernel: [ 6989.040563]
Mar 11 23:34:20 smarthome kernel: [ 9291.384885] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 11 23:34:20 smarthome kernel: [ 9291.384885]
Mar 11 23:40:37 smarthome kernel: [ 9669.005406] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 11 23:40:37 smarthome kernel: [ 9669.005406]
Mar 11 23:53:40 smarthome kernel: [10451.596965] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 11 23:53:40 smarthome kernel: [10451.596965]
Mar 11 23:58:00 smarthome kernel: [10712.027418] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 11 23:58:00 smarthome kernel: [10712.027418]
Mar 12 00:09:13 smarthome kernel: [11384.789115] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 00:09:13 smarthome kernel: [11384.789115]
Mar 12 00:23:54 smarthome kernel: [12265.530363] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 00:23:54 smarthome kernel: [12265.530363]
Mar 12 00:41:22 smarthome kernel: [13313.321993] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 00:41:22 smarthome kernel: [13313.321993]
Mar 12 01:06:54 smarthome kernel: [14846.055807] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 01:06:54 smarthome kernel: [14846.055807]
Mar 12 01:19:36 smarthome kernel: [15607.956223] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 01:19:36 smarthome kernel: [15607.956223]
Mar 12 01:19:36 smarthome kernel: [15607.956273] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 01:19:36 smarthome kernel: [15607.956273]
Mar 12 01:57:09 smarthome kernel: [17860.331118] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 01:57:09 smarthome kernel: [17860.331118]
Mar 12 02:01:06 smarthome kernel: [18097.771757] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 02:01:06 smarthome kernel: [18097.771757]
Mar 12 02:01:06 smarthome kernel: [18097.771800] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 02:01:06 smarthome kernel: [18097.771800]
Mar 12 02:14:10 smarthome kernel: [18881.653917] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 02:14:10 smarthome kernel: [18881.653917]
Mar 12 02:55:03 smarthome kernel: [21334.319156] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 02:55:03 smarthome kernel: [21334.319156]
Mar 12 03:12:41 smarthome kernel: [22393.031847] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 03:12:41 smarthome kernel: [22393.031847]
Mar 12 03:12:41 smarthome kernel: [22393.031915] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 03:12:41 smarthome kernel: [22393.031915]
Mar 12 04:05:37 smarthome kernel: [25569.135394] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 04:05:37 smarthome kernel: [25569.135394]
Mar 12 04:06:44 smarthome kernel: [25635.847919] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 04:06:44 smarthome kernel: [25635.847919]
Mar 12 05:19:55 smarthome kernel: [30026.411032] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 05:19:55 smarthome kernel: [30026.411032]
Mar 12 05:36:17 smarthome kernel: [31008.634138] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 05:36:17 smarthome kernel: [31008.634138]
Mar 12 05:50:45 smarthome kernel: [31877.103407] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 05:50:45 smarthome kernel: [31877.103407]
Mar 12 05:50:45 smarthome kernel: [31877.103449] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 05:50:45 smarthome kernel: [31877.103449]
Mar 12 06:13:59 smarthome kernel: [33270.414486] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 06:13:59 smarthome kernel: [33270.414486]
Mar 12 06:29:28 smarthome liblogging-stdlog:  [origin software="rsyslogd" swVersion="8.24.0" x-pid="300" x-info="http://www.rsyslog.com"] rsyslogd was HUPed
Mar 12 06:56:49 smarthome kernel: [35840.406674] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 06:56:49 smarthome kernel: [35840.406674]
Mar 12 07:21:30 smarthome kernel: [37321.168066] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 07:21:30 smarthome kernel: [37321.168066]
Mar 12 07:22:19 smarthome kernel: [37370.448012] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 07:22:19 smarthome kernel: [37370.448012]
Mar 12 07:49:23 smarthome kernel: [38994.418866] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 07:49:23 smarthome kernel: [38994.418866]
Mar 12 07:55:16 smarthome kernel: [39347.849264] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 07:55:16 smarthome kernel: [39347.849264]
Mar 12 08:11:43 smarthome kernel: [40335.090385] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 08:11:43 smarthome kernel: [40335.090385]
Mar 12 08:23:34 smarthome kernel: [41045.590953] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 08:23:34 smarthome kernel: [41045.590953]
Mar 12 08:25:55 smarthome kernel: [41186.931208] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 08:25:55 smarthome kernel: [41186.931208]
Mar 12 09:08:43 smarthome kernel: [43754.806000] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 09:08:43 smarthome kernel: [43754.806000]
Mar 12 09:20:38 smarthome kernel: [44469.297325] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 09:20:38 smarthome kernel: [44469.297325]
Mar 12 09:28:19 smarthome kernel: [44930.858020] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 09:28:19 smarthome kernel: [44930.858020]
Mar 12 09:28:31 smarthome kernel: [44942.378090] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 09:28:31 smarthome kernel: [44942.378090]
Mar 12 09:29:14 smarthome kernel: [44985.848195] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 09:29:14 smarthome kernel: [44985.848195]
Mar 12 09:43:47 smarthome kernel: [45858.259722] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 09:43:47 smarthome kernel: [45858.259722]
Mar 12 09:49:13 smarthome kernel: [46184.360186] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 09:49:13 smarthome kernel: [46184.360186]
Mar 12 09:51:40 smarthome kernel: [46331.630412] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 09:51:40 smarthome kernel: [46331.630412]
Mar 12 11:11:14 smarthome kernel: [51105.430045] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 11:11:14 smarthome kernel: [51105.430045]
Mar 12 11:40:13 smarthome kernel: [52844.874471] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 11:40:13 smarthome kernel: [52844.874471]
Mar 12 11:47:15 smarthome kernel: [53266.355478] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 11:47:15 smarthome kernel: [53266.355478]
Mar 12 12:02:53 smarthome kernel: [54204.771277] systemd: 35 output lines suppressed due to ratelimiting
Mar 12 12:09:33 smarthome kernel: [54604.652519] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 12:09:33 smarthome kernel: [54604.652519]
Mar 12 12:32:16 smarthome kernel: [55967.856088] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 12:32:16 smarthome kernel: [55967.856088]
Mar 12 13:53:36 smarthome kernel: [60847.414932] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 13:53:36 smarthome kernel: [60847.414932]
Mar 12 14:05:39 smarthome kernel: [61570.416414] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 14:05:39 smarthome kernel: [61570.416414]
Mar 12 14:09:26 smarthome kernel: [61797.946921] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 14:09:26 smarthome kernel: [61797.946921]
Mar 12 14:11:16 smarthome kernel: [61907.897132] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 14:11:16 smarthome kernel: [61907.897132]
Mar 12 14:37:10 smarthome kernel: [63461.883575] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 14:37:10 smarthome kernel: [63461.883575]
Mar 12 14:47:07 smarthome kernel: [64058.641748] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 14:47:07 smarthome kernel: [64058.641748]
Mar 12 16:13:13 smarthome kernel: [69224.729119] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 16:13:13 smarthome kernel: [69224.729119]
Mar 12 16:18:26 smarthome kernel: [69537.649532] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 16:18:26 smarthome kernel: [69537.649532]
Mar 12 16:32:13 smarthome kernel: [70364.631398] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 16:32:13 smarthome kernel: [70364.631398]
Mar 12 17:00:51 smarthome kernel: [72082.454549] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 17:00:51 smarthome kernel: [72082.454549]
Mar 12 19:02:25 smarthome kernel: [79376.696106] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 19:02:25 smarthome kernel: [79376.696106]
Mar 12 19:12:49 smarthome kernel: [80001.147119] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 19:12:49 smarthome kernel: [80001.147119]
Mar 12 19:39:33 smarthome kernel: [81605.199815] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 19:39:33 smarthome kernel: [81605.199815]
Mar 12 19:39:33 smarthome kernel: [81605.199866] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 19:39:33 smarthome kernel: [81605.199866]
Mar 12 19:47:13 smarthome kernel: [82064.970574] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 19:47:13 smarthome kernel: [82064.970574]
Mar 12 20:08:52 smarthome kernel: [83363.993214] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 20:08:52 smarthome kernel: [83363.993214]
Mar 12 20:24:12 smarthome kernel: [84283.795090] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 20:24:12 smarthome kernel: [84283.795090]
Mar 12 20:49:00 smarthome kernel: [85771.757647] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 20:49:00 smarthome kernel: [85771.757647]
Mar 12 20:57:39 smarthome kernel: [86290.696049] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 20:57:39 smarthome kernel: [86290.696049]
Mar 12 21:36:25 smarthome kernel: [88617.012500] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 21:36:25 smarthome kernel: [88617.012500]
Mar 12 21:44:39 smarthome kernel: [89110.963238] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 21:44:39 smarthome kernel: [89110.963238]
Mar 12 22:13:20 smarthome kernel: [90831.986151] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 22:13:20 smarthome kernel: [90831.986151]
Mar 12 22:25:50 smarthome kernel: [91581.467472] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 22:25:50 smarthome kernel: [91581.467472]
Mar 12 22:30:46 smarthome kernel: [91878.228218] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 22:30:46 smarthome kernel: [91878.228218]
Mar 12 22:36:50 smarthome kernel: [92241.480137] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 22:36:50 smarthome kernel: [92241.480137]
Mar 12 22:41:49 smarthome kernel: [92540.649626] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 22:41:49 smarthome kernel: [92540.649626]
Mar 12 23:58:12 smarthome kernel: [97123.979548] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 12 23:58:12 smarthome kernel: [97123.979548]
Mar 13 00:46:56 smarthome kernel: [100048.105875] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 00:46:56 smarthome kernel: [100048.105875]
Mar 13 00:56:22 smarthome kernel: [100613.427055] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 00:56:22 smarthome kernel: [100613.427055]
Mar 13 00:58:43 smarthome kernel: [100754.931707] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 00:58:43 smarthome kernel: [100754.931707]
Mar 13 01:21:52 smarthome kernel: [102143.700108] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 01:21:52 smarthome kernel: [102143.700108]
Mar 13 01:24:23 smarthome kernel: [102294.380194] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 01:24:23 smarthome kernel: [102294.380194]
Mar 13 01:27:17 smarthome kernel: [102468.920563] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 01:27:17 smarthome kernel: [102468.920563]
Mar 13 01:42:11 smarthome kernel: [103363.052300] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 01:42:11 smarthome kernel: [103363.052300]
Mar 13 01:44:45 smarthome kernel: [103516.712491] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 01:44:45 smarthome kernel: [103516.712491]
Mar 13 02:03:28 smarthome kernel: [104639.304894] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 02:03:28 smarthome kernel: [104639.304894]
Mar 13 02:03:29 smarthome kernel: [104641.256114] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 02:03:29 smarthome kernel: [104641.256114]
Mar 13 02:03:29 smarthome kernel: [104641.256156] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 02:03:29 smarthome kernel: [104641.256156]
Mar 13 02:03:32 smarthome kernel: [104643.734810] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 02:03:32 smarthome kernel: [104643.734810]
Mar 13 02:12:35 smarthome kernel: [105186.796247] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 02:12:35 smarthome kernel: [105186.796247]
Mar 13 02:12:35 smarthome kernel: [105186.796290] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 02:12:35 smarthome kernel: [105186.796290]
Mar 13 02:24:52 smarthome kernel: [105923.947586] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 02:24:52 smarthome kernel: [105923.947586]
Mar 13 02:45:34 smarthome kernel: [107165.499765] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 02:45:34 smarthome kernel: [107165.499765]
Mar 13 02:52:55 smarthome kernel: [107606.580630] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 02:52:55 smarthome kernel: [107606.580630]
Mar 13 04:13:16 smarthome kernel: [112428.186243] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 04:13:16 smarthome kernel: [112428.186243]
Mar 13 04:15:10 smarthome kernel: [112541.786451] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 04:15:10 smarthome kernel: [112541.786451]
Mar 13 04:19:11 smarthome kernel: [112782.986601] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 04:19:11 smarthome kernel: [112782.986601]
Mar 13 04:38:20 smarthome kernel: [113931.400161] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 04:38:20 smarthome kernel: [113931.400161]
Mar 13 04:38:20 smarthome kernel: [113931.400184] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 04:38:20 smarthome kernel: [113931.400184]
Mar 13 04:47:21 smarthome kernel: [114472.988145] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 04:47:21 smarthome kernel: [114472.988145]
Mar 13 04:53:00 smarthome kernel: [114812.011190] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 04:53:00 smarthome kernel: [114812.011190]
Mar 13 04:59:39 smarthome kernel: [115210.743696] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 04:59:39 smarthome kernel: [115210.743696]
Mar 13 05:39:37 smarthome kernel: [117608.712103] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 05:39:37 smarthome kernel: [117608.712103]
Mar 13 05:58:16 smarthome kernel: [118727.791094] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 05:58:16 smarthome kernel: [118727.791094]
Mar 13 06:27:14 smarthome liblogging-stdlog:  [origin software="rsyslogd" swVersion="8.24.0" x-pid="300" x-info="http://www.rsyslog.com"] rsyslogd was HUPed
Mar 13 08:03:00 smarthome kernel: [126211.355312] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 08:03:00 smarthome kernel: [126211.355312]
Mar 13 08:03:01 smarthome kernel: [126212.825464] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 08:03:01 smarthome kernel: [126212.825464]
Mar 13 08:43:39 smarthome kernel: [128651.031593] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 08:43:39 smarthome kernel: [128651.031593]
Mar 13 08:55:35 smarthome kernel: [129366.553641] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 08:55:35 smarthome kernel: [129366.553641]
Mar 13 09:29:46 smarthome kernel: [131417.579257] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 09:29:46 smarthome kernel: [131417.579257]
Mar 13 09:36:34 smarthome kernel: [131826.220425] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 09:36:34 smarthome kernel: [131826.220425]
Mar 13 11:09:21 smarthome kernel: [137393.294420] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 11:09:21 smarthome kernel: [137393.294420]
Mar 13 11:27:42 smarthome kernel: [138493.656707] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 11:27:42 smarthome kernel: [138493.656707]
Mar 13 11:27:42 smarthome kernel: [138493.656761] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 11:27:42 smarthome kernel: [138493.656761]
Mar 13 11:52:02 smarthome kernel: [139954.319801] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 11:52:02 smarthome kernel: [139954.319801]
Mar 13 12:08:32 smarthome kernel: [140944.201772] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 12:08:32 smarthome kernel: [140944.201772]
Mar 13 12:09:14 smarthome kernel: [140986.321875] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 12:09:14 smarthome kernel: [140986.321875]
Mar 13 12:43:41 smarthome kernel: [143052.845997] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 12:43:41 smarthome kernel: [143052.845997]
Mar 13 12:51:10 smarthome kernel: [143502.107420] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 12:51:10 smarthome kernel: [143502.107420]
Mar 13 12:54:12 smarthome kernel: [143683.477312] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 12:54:12 smarthome kernel: [143683.477312]
Mar 13 13:41:59 smarthome kernel: [146550.833308] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 13:41:59 smarthome kernel: [146550.833308]
Mar 13 14:13:10 smarthome kernel: [148421.497255] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 14:13:10 smarthome kernel: [148421.497255]
Mar 13 14:56:37 smarthome kernel: [151029.288323] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 14:56:37 smarthome kernel: [151029.288323]
Mar 13 16:22:16 smarthome kernel: [156168.375240] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
Mar 13 16:22:16 smarthome kernel: [156168.375240]

I boot on a USB SSD . I never get this message before the firmware update of 11/03.

Bus 001 Device 008: ID 0403:6015 Future Technology Devices International, Ltd Bridge(I2C/SPI/UART/FIFO)
Bus 001 Device 006: ID 0403:6001 Future Technology Devices International, Ltd FT232 USB-Serial (UART) IC
Bus 001 Device 005: ID 0658:0200 Sigma Designs, Inc.
Bus 001 Device 004: ID 0409:0059 NEC Corp. HighSpeed Hub
Bus 001 Device 003: ID 0424:ec00 Standard Microsystems Corp. SMSC9512/9514 Fast Ethernet Adapter
Bus 001 Device 002: ID 0424:9514 Standard Microsystems Corp. SMC9514 Hub
Bus 001 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub ```
@pelwell

This comment has been minimized.

Copy link
Contributor

pelwell commented Mar 13, 2018

Thanks - there's nothing obviously wrong there, but it's still useful to have a more complete picture.

@6by9

This comment has been minimized.

Copy link
Contributor

6by9 commented Mar 13, 2018

I've seen the same error message. I currently have a local patch to nuke the error message, and everything still works OK.
The only USB devices I have connected are a mouse, keyboard and network on a Pi2. It does have an NFS file system.
I'll see if I still get the error if I revert the patch.

@hofnarr132

This comment has been minimized.

Copy link
Author

hofnarr132 commented Mar 28, 2018

Please don't close an issue that is not resolved. Saying "there's nothing obviously wrong" does not mean it's fixed!

Have now experienced the problem on a RPi 3B+ after the latest standard update pushed 4.14.30-v7+.

The new RPi 3B+ had been fresh installed with from NOOBS 2.7.0 and worked fine with 4.9.80.

The updated firmware caused the RPi to fail to complete the init startup processes with 100s of errors:

WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode

Only resolved by disabling the C-Media USB sound device in the USB HDMI KVM Switch and rebooting - RPi then proceeded to do a clean boot without problem.

Re-enabling the C-Media USB Sound device produces this from 'dmesg' (note that [ 18.476520] is the last entry from boot up, 133 onwards relates to turning on the USB device):

[   16.871637] Bluetooth: RFCOMM socket layer initialized
[   16.871662] Bluetooth: RFCOMM ver 1.11
[   18.175922] EXT4-fs (sda8): mounted filesystem with ordered data mode. Opts: (null)
[   18.303320] EXT4-fs (sda5): mounted filesystem with ordered data mode. Opts: (null)
[   18.476520] EXT4-fs (sda10): mounted filesystem with ordered data mode. Opts: (null)
[  133.775588] usb 1-1.2.5.4: new full-speed USB device number 9 using dwc_otg
[  133.913936] usb 1-1.2.5.4: New USB device found, idVendor=0d8c, idProduct=013c
[  133.913951] usb 1-1.2.5.4: New USB device strings: Mfr=1, Product=2, SerialNumber=0
[  133.913961] usb 1-1.2.5.4: Product: USB PnP Sound Device          
[  133.913970] usb 1-1.2.5.4: Manufacturer: C-Media Electronics Inc.      
[  133.919250] input: C-Media Electronics Inc.       USB PnP Sound Device           as /devices/platform/soc/3f980000.usb/usb1/1-1/1-1.2/1-1.2.5/1-1.2.5.4/1-1.2.5.4:1.3/0003:0D8C:013C.0003/input/input2
[  133.986156] hid-generic 0003:0D8C:013C.0003: input,hidraw2: USB HID v1.00 Device [C-Media Electronics Inc.       USB PnP Sound Device          ] on usb-3f980000.usb-1.2.5.4/input3
[  134.096558] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode

[  134.096585] usb 1-1.2.5.4: 2:1: cannot get freq at ep 0x82
[  134.096604] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode

[  134.096612] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode

[  134.104537] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode

[  134.112536] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode

[  134.112641] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode

[  134.120551] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode

[  134.120585] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode

[  134.120594] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode

[  134.120623] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode

[  134.120631] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode

[  134.122656] usbcore: registered new interface driver snd-usb-audio
[  134.144539] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode

[  134.144644] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode

lsusb output:

pi@raspberrypi:~ $ lsusb
Bus 001 Device 009: ID 0d8c:013c C-Media Electronics, Inc. CM108 Audio Controller
Bus 001 Device 008: ID 040b:2013 Weltrend Semiconductor 
Bus 001 Device 007: ID 05e3:0608 Genesys Logic, Inc. Hub
Bus 001 Device 005: ID 05e3:0718 Genesys Logic, Inc. IDE/SATA Adapter
Bus 001 Device 004: ID 1a40:0201 Terminus Technology Inc. FE 2.1 7-port Hub
Bus 001 Device 006: ID 0424:7800 Standard Microsystems Corp. 
Bus 001 Device 003: ID 0424:2514 Standard Microsystems Corp. USB 2.0 Hub
Bus 001 Device 002: ID 0424:2514 Standard Microsystems Corp. USB 2.0 Hub
Bus 001 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub
pi@raspberrypi:~ $ uname -a
Linux raspberrypi 4.14.30-v7+ #1102 SMP Mon Mar 26 16:45:49 BST 2018 armv7l GNU/Linux
@pelwell

This comment has been minimized.

Copy link
Contributor

pelwell commented Mar 28, 2018

Easy there - the issue is still open.

@hofnarr132

This comment has been minimized.

Copy link
Author

hofnarr132 commented Mar 28, 2018

Apologies - saw the 'Closed' state against the cross referenced issue and jumped to wrong conclusion!

@hofnarr132

This comment has been minimized.

Copy link
Author

hofnarr132 commented Mar 28, 2018

I have also noticed that dmesg continues to log the WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: message at a rate of about 150 lines per second! This did not happen with the previous version.

@macmpi

This comment has been minimized.

Copy link

macmpi commented Mar 29, 2018

@popcornmix has just pushed a corruption fix: does it help?

@hofnarr132

This comment has been minimized.

Copy link
Author

hofnarr132 commented Mar 31, 2018

Tested on 1 x RPi 3B and found much improved. Will test more..

Prep for f/w 4.14.31 2148253d340d7c646583cb8f0be6403511522cc4

Tested on RPi 3B with 16GB SD-Card (boot device) and directly connected 4GB USB memory stick as storage device (log capture etc).
Also a 4x1 USB HDMI KVN Switch connected for HDMI display and provides a USB2 hub in which a C-Media USB sound device is embedded (this can be turned On/Off with a push button)
and a wireless mouse/keyboard is connected.

1). Started with NOOBS 2.7.0 2018-03-14 / Raspian 2018-03-13 Kernel 4.9 (firmware 4.9.80)
2). Applied all updates EXCEPT firmware and kernel related packages shown below:

pi@raspberrypi:~ $ apt list --upgradable
Listing... Done
base-files/stable 9.9+rpi1+deb9u4 armhf [upgradable from: 9.9+rpi1+deb9u3]
libraspberrypi-bin/stable 1.20180328-1 armhf [upgradable from: 1.20180313-1]
libraspberrypi-dev/stable 1.20180328-1 armhf [upgradable from: 1.20180313-1]
libraspberrypi-doc/stable 1.20180328-1 armhf [upgradable from: 1.20180313-1]
libraspberrypi0/stable 1.20180328-1 armhf [upgradable from: 1.20180313-1]
raspberrypi-bootloader/stable 1.20180328-1 armhf [upgradable from: 1.20180313-1]
raspberrypi-kernel/stable 1.20180328-1 armhf [upgradable from: 1.20180313-1]
raspberrypi-sys-mods/stable 20180328+1 armhf [upgradable from: 20180312]
rc-gui/stable 1.16 armhf [upgradable from: 1.15]
rpd-icons/stable 0.15 all [upgradable from: 0.14]

3). reboot and test USB sound device (OK)

pi@raspberrypi:~ $ uname -a
Linux raspberrypi 4.9.80-v7+ #1098 SMP Fri Mar 9 19:11:42 GMT 2018 armv7l GNU/Linux

4). rpi-update to 4.14.31 firmware
5). reboot and test USB sound device (OK)

pi@raspberrypi:~ $ uname -a
Linux raspberrypi 4.14.31-v7+ #1104 SMP Thu Mar 29 16:52:18 BST 2018 armv7l GNU/Linux

6). 'sudo apt-get upgrade' to apply latest firmware and kernel
7). reboot and test USB sound device (OK)

pi@raspberrypi:~ $ uname -a
Linux raspberrypi 4.14.30-v7+ #1102 SMP Mon Mar 26 16:45:49 BST 2018 armv7l GNU/Linux

Hmm. not sure how to check that latest firmware is actually being used - any pointers?
Tried re-applying firmware update - it reports it is the latest...

pi@raspberrypi:~ $ time sudo rpi-update 2148253d340d7c646583cb8f0be6403511522cc4
 *** Raspberry Pi firmware updater by Hexxeh, enhanced by AndrewS and Dom
 *** Performing self-update
 *** Relaunching after update
 *** Raspberry Pi firmware updater by Hexxeh, enhanced by AndrewS and Dom
 *** Your firmware is already up to date

In all cases I get between 12 and 25 WARN messages and then no more (looks like the 'usb_core' clears or resolves the 'dwc_otg_handle_mode_mismatch_intr:68'

usbcore: registered new interface driver snd-usb-audio

dmesg logs available if required.

@kauailabs

This comment has been minimized.

Copy link

kauailabs commented Apr 4, 2018

I'm still seeing this issue w/4.14.31, which I installed via rpi-update. I have a USB headset, mouse and keyboard plugged into an IOgear USB switch, connected to the RPI. I can't get it to boot due to the continual messages during boot - unless I disconnect the Logitech USB headset. And then even after I do disconnect the Logitech USB headset and reboot, I'm still having significant performance problems. Here's the last few seconds of entry from dmesg, followed by uname -a to display kernel version number.

This problem is occurring both on a Rpi3B and Rpi3B+.

Last few lines of output from dmesg:

[ 422.847026] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode

[ 422.847142] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode

[ 422.862955] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode

[ 422.863034] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode

pi@raspberrypi:~ $ uname -a
Linux raspberrypi 4.14.31-v7+ #1104 SMP Thu Mar 29 16:52:18 BST 2018 armv7l GNU/Linux

** Additional info, added 4/4/2018 **

Have done some additional testing, same 4.14.31-v7+ kernel, latest rpi-update as of this date.

USB Devices attached to RPI are a Microsoft keyboard, Microsoft optical mouse, and an IOgear Model GU5434 USB switch.

When connecting mouse to RPI via switch, disconnecting it and then reconnecting it, on 2 out of 3 attempts, the dwc_otg Mode Mismatch Interrupt message occurs. The errors are displayed when the device is reconnected.

About 1 out of every 4 times, the messages are generated continually (on the order of 100 a second) and don't stop - and the system appears more sluggish; the remainder of the times that messages appear, only 2 or 3 appear.

Then, if the USB switch is removed from the picture, and the USB mouse is connected directly, the dwc_otg Mode Mismatch interrupt messages are not seen.

Note: unlike the intial report above which focused on system boot behavior, these results are observed during the normal operation after the system, post-boot.

@hofnarr132

This comment has been minimized.

Copy link
Author

hofnarr132 commented Apr 7, 2018

I've carried out further testing with 4.14.32 and also still see problems, but hopefully help narrow this down a bit to 'daisy chained' USB hubs..

Tested system: RPi 3B+ booting from Corsair SSD which USB connected - using a PlusBerryPi USB device (SATA + USB Hub) and connected to a USB/HDMI KVM Switch with embedded USB hub, emdedded C-Media USB Sound Device and USB Wireless Keyboard/Mouse.

When the USB/HDMI KVM switch is plugged into to the PlusBerryPi board with the C-Media USB device active/connected (ie. 'daisy chained' RPi USB Hub --> PlusBerryPi Hub --> USB/HDMI KVM hub), I get many more WARN::dwc_otg_handle_mode_mismatch_intr:68: messages - this often seem to overwhelm the system on a startup and cause fail or time out (unless USB Sound device disconnected.

If the USB/HDMI KVM switch is connected directly to the RPi, then I get a few WARN messages as per my previous post on testing with a RPi 3B.

usb-device output:


T:  Bus=01 Lev=00 Prnt=00 Port=00 Cnt=00 Dev#=  1 Spd=480 MxCh= 1
D:  Ver= 2.00 Cls=09(hub  ) Sub=00 Prot=01 MxPS=64 #Cfgs=  1
P:  Vendor=1d6b ProdID=0002 Rev=04.14
S:  Manufacturer=Linux 4.14.32-v7+ dwc_otg_hcd
S:  Product=DWC OTG Controller
S:  SerialNumber=3f980000.usb
C:  #Ifs= 1 Cfg#= 1 Atr=e0 MxPwr=0mA
I:  If#= 0 Alt= 0 #EPs= 1 Cls=09(hub  ) Sub=00 Prot=00 Driver=hub

T:  Bus=01 Lev=01 Prnt=01 Port=00 Cnt=01 Dev#=  2 Spd=480 MxCh= 4
D:  Ver= 2.00 Cls=09(hub  ) Sub=00 Prot=02 MxPS=64 #Cfgs=  1
P:  Vendor=0424 ProdID=2514 Rev=0b.b3
C:  #Ifs= 1 Cfg#= 1 Atr=e0 MxPwr=2mA
I:  If#= 0 Alt= 1 #EPs= 1 Cls=09(hub  ) Sub=00 Prot=02 Driver=hub

T:  Bus=01 Lev=02 Prnt=02 Port=00 Cnt=01 Dev#=  3 Spd=480 MxCh= 3
D:  Ver= 2.00 Cls=09(hub  ) Sub=00 Prot=02 MxPS=64 #Cfgs=  1
P:  Vendor=0424 ProdID=2514 Rev=0b.b3
C:  #Ifs= 1 Cfg#= 1 Atr=e0 MxPwr=2mA
I:  If#= 0 Alt= 1 #EPs= 1 Cls=09(hub  ) Sub=00 Prot=02 Driver=hub

T:  Bus=01 Lev=03 Prnt=03 Port=00 Cnt=01 Dev#=  6 Spd=480 MxCh= 0
D:  Ver= 2.10 Cls=ff(vend.) Sub=00 Prot=ff MxPS=64 #Cfgs=  1
P:  Vendor=0424 ProdID=7800 Rev=03.00
C:  #Ifs= 1 Cfg#= 1 Atr=e0 MxPwr=2mA
I:  If#= 0 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=00 Prot=ff Driver=lan78xx

T:  Bus=01 Lev=02 Prnt=02 Port=01 Cnt=02 Dev#=  4 Spd=480 MxCh= 6
D:  Ver= 2.00 Cls=09(hub  ) Sub=00 Prot=02 MxPS=64 #Cfgs=  1
P:  Vendor=1a40 ProdID=0201 Rev=01.00
S:  Product=USB 2.0 Hub [MTT]
C:  #Ifs= 1 Cfg#= 1 Atr=e0 MxPwr=100mA
I:  If#= 0 Alt= 1 #EPs= 1 Cls=09(hub  ) Sub=00 Prot=02 Driver=hub

T:  Bus=01 Lev=03 Prnt=04 Port=00 Cnt=01 Dev#= 14 Spd=480 MxCh= 0
D:  Ver= 2.00 Cls=00(>ifc ) Sub=00 Prot=00 MxPS=64 #Cfgs=  1
P:  Vendor=090c ProdID=1000 Rev=11.00
S:  Manufacturer=SMI Corporation
S:  Product=USB DISK
C:  #Ifs= 1 Cfg#= 1 Atr=80 MxPwr=500mA
I:  If#= 0 Alt= 0 #EPs= 2 Cls=08(stor.) Sub=06 Prot=50 Driver=usb-storage

T:  Bus=01 Lev=03 Prnt=04 Port=01 Cnt=02 Dev#=  5 Spd=480 MxCh= 0
D:  Ver= 2.00 Cls=00(>ifc ) Sub=00 Prot=00 MxPS=64 #Cfgs=  1
P:  Vendor=05e3 ProdID=0718 Rev=00.09
S:  Product=USB Storage
S:  SerialNumber=000000000033
C:  #Ifs= 1 Cfg#= 1 Atr=c0 MxPwr=96mA
I:  If#= 0 Alt= 0 #EPs= 2 Cls=08(stor.) Sub=06 Prot=50 Driver=usb-storage

T:  Bus=01 Lev=03 Prnt=04 Port=04 Cnt=03 Dev#=  7 Spd=480 MxCh= 4
D:  Ver= 2.00 Cls=09(hub  ) Sub=00 Prot=01 MxPS=64 #Cfgs=  1
P:  Vendor=05e3 ProdID=0608 Rev=77.63
S:  Product=USB2.0 Hub
C:  #Ifs= 1 Cfg#= 1 Atr=e0 MxPwr=100mA
I:  If#= 0 Alt= 0 #EPs= 1 Cls=09(hub  ) Sub=00 Prot=00 Driver=hub

T:  Bus=01 Lev=04 Prnt=07 Port=01 Cnt=01 Dev#=  8 Spd=1.5 MxCh= 0
D:  Ver= 1.10 Cls=00(>ifc ) Sub=00 Prot=00 MxPS= 8 #Cfgs=  1
P:  Vendor=040b ProdID=2013 Rev=02.00
S:  Product=GiGa HiD
C:  #Ifs= 2 Cfg#= 1 Atr=a0 MxPwr=100mA
I:  If#= 0 Alt= 0 #EPs= 1 Cls=03(HID  ) Sub=01 Prot=01 Driver=usbhid
I:  If#= 1 Alt= 0 #EPs= 1 Cls=03(HID  ) Sub=01 Prot=02 Driver=usbhid

Process to update RPi 3B+:

  1. USB Sound Device disconnected.
  2. sudo apt-get update && sudo apt-get upgrade -y
  3. Reboot - dmesg checked - all fine
  4. Kernel version check:
pi@raspberrypi:~ $ uname -a
Linux raspberrypi 4.14.30-v7+ #1102 SMP Mon Mar 26 16:45:49 BST 2018 armv7l GNU/Linux
  1. Update kernel/firmware to latest (kernel 4.14.32, firmware last commit 06Apr18):
pi@raspberrypi:~ $ sudo rpi-update 064219e329252abd8532b7fa8030ed9a9f391dd3
 *** Raspberry Pi firmware updater by Hexxeh, enhanced by AndrewS and Dom
 *** Performing self-update
  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed
100 13403  100 13403    0     0  59290      0 --:--:-- --:--:-- --:--:-- 59568
 *** Relaunching after update
 *** Raspberry Pi firmware updater by Hexxeh, enhanced by AndrewS and Dom
 *** We're running for the first time
 *** Backing up files (this will take a few minutes)
 *** Backing up firmware
 *** Backing up modules 4.14.30-v7+
#############################################################
This update bumps to rpi-4.14.y linux tree
Be aware there could be compatibility issues with some drivers
Discussion here:
https://www.raspberrypi.org/forums/viewtopic.php?f=29&t=197689
##############################################################
 *** Downloading specific firmware revision (this will take a few minutes)
  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed
100   168    0   168    0     0    313      0 --:--:-- --:--:-- --:--:--   314
100 55.6M  100 55.6M    0     0  1605k      0  0:00:35  0:00:35 --:--:-- 1464k
 *** Updating firmware
 *** Updating kernel modules
 *** depmod 4.14.32-v7+
 *** depmod 4.14.32+
 *** Updating VideoCore libraries
 *** Using HardFP libraries
 *** Updating SDK
 *** Running ldconfig
 *** Storing current firmware revision
 *** Deleting downloaded files
 *** Syncing changes to disk
 *** If no errors appeared, your firmware was successfully updated to 064219e329252abd8532b7fa8030ed9a9f391dd3
 *** A reboot is needed to activate the new firmware
  1. Reboot. (USB Sound Device still disconnected)
  2. Kernel version check:
pi@raspberrypi:~ $ uname -a
Linux raspberrypi 4.14.32-v7+ #1107 SMP Fri Apr 6 16:36:46 BST 2018 armv7l GNU/Linux
  1. Follow dmesg:
    pi@raspberrypi:~ $ dmesg --follow | tee -a /media/pi/E8AA-82A6/RPi3B+post-dmesg.txt

  2. Connect USB Sound Device:
    Detected with some WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode messages
    These clear/stop moments after usbcore: registered new interface driver snd-usb-audio

  3. Reboot with USB Sound Device connected:

WARN messages for approx 10 seconds during boot.

dmesg > /media/pi/E8AA-82A6/RPi3B+post1-dmesg.txt

[    4.021524] usb 1-1.2.5.4: new full-speed USB device number 9 using dwc_otg
[    4.029870] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
.
.  (More WARN messages)
.
[    4.351018] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode

[    4.374657] usb 1-1.2.5.4: New USB device found, idVendor=0d8c, idProduct=013c
[    4.384921] usb 1-1.2.5.4: New USB device strings: Mfr=1, Product=2, SerialNumber=0
[    4.395137] usb 1-1.2.5.4: Product: USB PnP Sound Device          
[    4.405308] usb 1-1.2.5.4: Manufacturer: C-Media Electronics Inc.      
[    4.421768] input: C-Media Electronics Inc.       USB PnP Sound Device           as /devices/platform/soc/3f980000.usb/usb1/1-1/1-1.2/1-1.2.5/1-1.2.5.4/1-1.2.5.4:1.3/0003:0D8C:013C.0003/input/input2
[    4.451079] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
.
.  (Lots more WARN messages!)
.
[   14.258226] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode

[   14.258323] usb 1-1.2.5.4: 1:1: cannot get freq at ep 0x1
[   14.266069] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
.
.  (More WARN messages)
.
[   14.314102] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode

[   14.321911] usbcore: registered new interface driver snd-usb-audio
[   16.723415] uart-pl011 3f201000.serial: no DMA platform data
  1. Subsequent reboots and WARN messages displaying for more than 600 seconds with many being dropped and causing boot up to fail. If the C-Media USB device is disconnected while in bootup, WARN messages eventually stop and bootup continues.

  2. Connect KVM USB directly to RPi (this is requires the case lid to be removed).
    Successfully boots C-Media USB device detected (tested with a mutiple reboots - WARN messages appear for between 4 and 10 seconds).

@P33M

This comment has been minimized.

Copy link
Contributor

P33M commented Apr 9, 2018

Can you try booting with the following parameter settings in /boot/cmdline.txt?

dwc_otg.fiq_fsm_mask=0x0
dwc_otg.fiq_fsm_mask=0x1
dwc_otg.fiq_fsm_mask=0x3
dwc_otg.fiq_fsm_mask=0x7

Try each one in turn and see if it generates the errors in question. Note that the audio device may become unreliable (garbled audio) when testing these options.

@citlee

This comment has been minimized.

Copy link

citlee commented Apr 10, 2018

I am also seeing this issue (syslog flooded with WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode), on a Pi 3B+ freshly imaged with 2018-03-13-raspbian-stretch-lite and then apt update/upgrade'd and rebooted. The only attached USB devices are a keyboard, which is also a one port USB hub, which has a mouse attached. Unplugging the mouse from the keyboard hub stops these messages.

uname -a reports:
Linux raspberrypi 4.14.30-v7+ #1102 SMP Mon Mar 26 16:45:49 BST 2018 armv71 GNU/Linux

@pelwell

This comment has been minimized.

Copy link
Contributor

pelwell commented Apr 10, 2018

Please work through the four cmdline.txt settings (one at a time) as described by @P33M above.

@dcs3spp

This comment has been minimized.

Copy link

dcs3spp commented Apr 10, 2018

Hi,

I have originally logged a separate issue #2494 related to a usb bluetooth device not working in an externally powered hub. However, I am also receiving the dwc_otg warning messages and have been directed to this issue.

In short, using dwc_otg.fiq_fsm_mask=0x0 in cmdline.txt cleared the dwc_otg warning messages, however bluetooth controller remains unavailable.

I have an Asus BT-400 usb bluetooth device that is plugged into a self powered hub, Logik L4THUB10.

Details for the usb bluetooth device are:

Bus 001 Device 006: ID 0b05:17cb ASUSTek Computer, Inc. Broadcom BCM20702A0 Bluetooth

Other usb devices running on the system are:

Bus 001 Device 007: ID 1005:b113 Apacer Technology, Inc. Handy Steno/AH123 / Handy Steno 2.0/HT203
Bus 001 Device 006: ID 0b05:17cb ASUSTek Computer, Inc. Broadcom BCM20702A0 Bluetooth
Bus 001 Device 005: ID 7392:7711 Edimax Technology Co., Ltd EW-7711UTn nLite Wireless Adapter [Ralink RT2870]
Bus 001 Device 004: ID 05e3:0608 Genesys Logic, Inc. Hub
Bus 001 Device 003: ID 0424:ec00 Standard Microsystems Corp. SMSC9512/9514 Fast Ethernet Adapter
Bus 001 Device 002: ID 0424:9512 Standard Microsystems Corp. SMC9512/9514 USB Hub
Bus 001 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub

The usb tree is reported as follows:

/:  Bus 01.Port 1: Dev 1, Class=root_hub, Driver=dwc_otg/1p, 480M
    |__ Port 1: Dev 2, If 0, Class=Hub, Driver=hub/3p, 480M
        |__ Port 1: Dev 3, If 0, Class=Vendor Specific Class, Driver=smsc95xx, 480M
        |__ Port 3: Dev 4, If 0, Class=Hub, Driver=hub/4p, 480M
            |__ Port 1: Dev 5, If 0, Class=Vendor Specific Class, Driver=rt2800usb, 480M
            |__ Port 3: Dev 6, If 3, Class=Application Specific Interface, Driver=, 12M
            |__ Port 3: Dev 6, If 1, Class=Vendor Specific Class, Driver=btusb, 12M
            |__ Port 3: Dev 6, If 2, Class=Vendor Specific Class, Driver=btusb, 12M
            |__ Port 3: Dev 6, If 0, Class=Vendor Specific Class, Driver=btusb, 12M
            |__ Port 4: Dev 7, If 0, Class=Mass Storage, Driver=usb-storage, 480M

I have performed the following actions.

  1. Downloaded and installed latest updates
    sudo pacman -Syu
  2. This downloaded a later kernel version
    uname -a

Linux raspberrypi.ubuntu.private 4.14.33-1-ARCH #1 SMP Tue Apr 10 00:59:56 UTC 2018 armv6l GNU/Linux

  1. Powered off hub and Pi, removed bluetooth from external usb hub and then
    rebooted Pi.
    No dwc_otg mode mismatch interrupt warning messages are logged when the
    bluetooth device is removed from the external usb hub.
  2. Followed dmesg:
    dmesg --follow
  3. Inserted bluetooth device into external usb hub. The dwc_otg warning messages are now logged. The bluetooth device patch command fails,
    command 0x0c03 tx timeout occurs and reset fails.
[ 1115.741505] usb 1-1.3.3: new full-speed USB device number 7 using dwc_otg
[ 1115.880547] usb 1-1.3.3: New USB device found, idVendor=0b05, idProduct=17cb
[ 1115.888158] usb 1-1.3.3: New USB device strings: Mfr=1, Product=2, SerialNumber=3
[ 1115.896100] usb 1-1.3.3: Product: BCM20702A0
[ 1115.900556] usb 1-1.3.3: Manufacturer: Broadcom Corp
[ 1115.905791] usb 1-1.3.3: SerialNumber: 5CF370607CCB
[ 1116.500200] Bluetooth: Core ver 2.22
[ 1116.509339] NET: Registered protocol family 31
[ 1116.514202] Bluetooth: HCI device and connection manager initialized
[ 1116.524615] Bluetooth: HCI socket layer initialized
[ 1116.529716] Bluetooth: L2CAP socket layer initialized
[ 1116.552891] Bluetooth: SCO socket layer initialized
.
. WARN messages
.
[ 1116.701903] usbcore: registered new interface driver btusb
.
. WARN messages
.
[ 1116.790356] Bluetooth: hci0: BCM: chip id 63
[ 1116.796280] Bluetooth: hci0: BCM: features 0x07
.
. WARN messages
.
[ 1116.842102] Bluetooth: hci0: BCM20702A
[ 1116.848285] Bluetooth: hci0: BCM20702A1 (001.002.014) build 0000
[ 1118.477926] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
.
. WARN messages
.
[ 1120.136307] Bluetooth: BNEP (Ethernet Emulation) ver 1.3
[ 1120.196923] Bluetooth: BNEP filters: protocol multicast
.
. WARN messages
.
[ 1120.354359] Bluetooth: BNEP socket layer initialized
[ 1121.131625] Bluetooth: hci0 command 0xfc4c tx timeout
.
. WARN messages
.
[ 1129.458311] Bluetooth: hci0: BCM: Patch command fc4c failed (-110)
[ 1131.531535] Bluetooth: hci0 command 0x0c03 tx timeout
.
. WARN messages
.
[ 1139.691414] Bluetooth: hci0: BCM: Reset failed (-110)
[ 1139.711576] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
  1. Run bluetoothctl and controller not listed:
[alarm@raspberrypi ~]$ bluetoothctl
Agent registered
[bluetooth]# list
[bluetooth]# show
No default controller available

Trying tests with kernel settings in /boot/cmdline.txt, as suggested by @P33M:

dwc_otg.fiq_fsm_mask=0x0
dwc_otg.fiq_fsm_mask=0x1
dwc_otg.fiq_fsm_mask=0x3
dwc_otg.fiq_fsm_mask=0x7

Prior to testing the contents of /boot/cmdline.txt are:

root=/dev/mmcblk0p2 rw rootwait console=ttyAMA0,115200 console=tty1 selinux=0 plymouth.enable=0 smsc95xx.turbo_mode=N dwc_otg.lpm_enable=0 kgdboc=ttyAMA0,115200 elevator=noop

  1. Test dwc_otg.fiq_fsm_mask=0x0:
    No warn messages are visible on boot. The bluetooth controller is visible in
    the bluetoothctl utility.
[   25.896265] Bluetooth: HCI device and connection manager initialized
[   25.990257] Bluetooth: HCI socket layer initialized
[   26.068415] Bluetooth: L2CAP socket layer initialized
[   26.162070] Bluetooth: SCO socket layer initialized
[   26.440699] usbcore: registered new interface driver btusb
[   26.572965] Bluetooth: hci0: BCM: chip id 63
[   26.587316] Bluetooth: hci0: BCM: features 0x07
[   26.620312] Bluetooth: hci0: rpi
[   26.634007] Bluetooth: hci0: BCM20702A1 (001.002.014) build 1467
[   26.781169] random: crng init done
[   27.700022] usb 1-1.3.1: reset high-speed USB device number 5 using dwc_otg
[   27.757857] FAT-fs (mmcblk0p1): Volume was not properly unmounted. Some data may be corrupt. Please run fsck.
[   27.791029] Bluetooth: hci0: BCM20702A1 (001.002.014) build 1467
[   27.856997] Bluetooth: hci0: Broadcom Bluetooth Device
[   27.862176] ieee80211 phy0: rt2x00_set_rt: Info - RT chipset 3070, rev 0201 detected
[   27.931187] ieee80211 phy0: rt2x00_set_rf: Info - RF chipset 0005 detected
[   28.068422] ieee80211 phy0: Selected rate control algorithm 'minstrel_ht'
[   28.113335] usbcore: registered new interface driver rt2800usb
[   29.929969] Bluetooth: hci0 command 0x0c23 tx timeout
[   32.010014] Bluetooth: hci0 command 0x0c14 tx timeout
[   34.090082] Bluetooth: hci0 command 0x0c25 tx timeout
[   34.221643] Bluetooth: BNEP (Ethernet Emulation) ver 1.3
[   34.234781] Bluetooth: BNEP filters: protocol multicast
[   34.396530] Bluetooth: BNEP socket layer initialized
[   36.169989] Bluetooth: hci0 command 0x0c38 tx timeout
[   36.474961] ieee80211 phy0: rt2x00lib_request_firmware: Info - Loading firmware file 'rt2870.bin'
[   36.540069] ieee80211 phy0: rt2x00lib_request_firmware: Info - Firmware detected - version: 0.36
[   37.844694] IPv6: ADDRCONF(NETDEV_UP): wlan0: link is not ready
[   38.250385] Bluetooth: hci0 command tx timeout

After running sudo reboot no warning messages are displayed. However, the
controller is no longer accessible in bluetoothctl. Removing the usb device
from the hub and using dmesg --follow shows no additional dwc_otg warning
messages.

  1. Test dwc_otg.fiq_fsm_mask=0x1:
    Some dwc_otg warning messages are displayed and the bluetooth controller is
    not visible in bluetoothctl utility.
[   27.425779] Bluetooth: Core ver 2.22
[   27.490174] NET: Registered protocol family 31
[   27.490188] Bluetooth: HCI device and connection manager initialized
[   27.495269] Bluetooth: HCI socket layer initialized
[   27.495295] Bluetooth: L2CAP socket layer initialized
[   27.495382] Bluetooth: SCO socket layer initialized
[   27.693875] usbcore: registered new interface driver btusb
[   27.843585] Bluetooth: hci0: BCM: chip id 63
[   27.858701] Bluetooth: hci0: BCM: features 0x07
[   27.893516] Bluetooth: hci0: BCM20702A
[   27.895626] Bluetooth: hci0: BCM20702A1 (001.002.014) build 0000
[   27.919836] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
[   27.990622] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
[   27.991896] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
[   28.439861] random: crng init done
[   29.480148] usb 1-1.3.1: reset high-speed USB device number 5 using dwc_otg
[   29.629604] ieee80211 phy0: rt2x00_set_rt: Info - RT chipset 3070, rev 0201 detected
[   29.900655] ieee80211 phy0: rt2x00_set_rf: Info - RF chipset 0005 detected
[   30.010001] Bluetooth: hci0 command 0xfc4c tx timeout
[   30.127876] ieee80211 phy0: Selected rate control algorithm 'minstrel_ht'
[   30.194904] usbcore: registered new interface driver rt2800usb
[   34.907443] Bluetooth: BNEP (Ethernet Emulation) ver 1.3
[   34.921552] Bluetooth: BNEP filters: protocol multicast
[   35.100146] Bluetooth: BNEP socket layer initialized
[   37.196852] ieee80211 phy0: rt2x00lib_request_firmware: Info - Loading firmware file 'rt2870.bin'
[   37.264364] ieee80211 phy0: rt2x00lib_request_firmware: Info - Firmware detected - version: 0.36
[   38.249934] Bluetooth: hci0: BCM: Patch command fc4c failed (-110)
[   38.533400] IPv6: ADDRCONF(NETDEV_UP): wlan0: link is not ready
[   40.330113] Bluetooth: hci0 command 0x0c03 tx timeout
[   48.489842] Bluetooth: hci0: BCM: Reset failed (-110)

Removing bluetooth device and reinserting yields some dwc_otg warning messages.
Bluetooth controller remains unavailable in bluetoothctl utility.

[  347.283443] Bluetooth: hci0: BCM: chip id 63
[  347.289461] Bluetooth: hci0: BCM: features 0x07
[  347.311400] Bluetooth: hci0: BCM20702A
[  347.317502] Bluetooth: hci0: BCM20702A1 (001.002.014) build 0000
[  347.395704] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
.
. WARN messages
.
[  349.448336] Bluetooth: hci0 command 0xfc4c tx timeout
[  357.607945] Bluetooth: hci0: BCM: Patch command fc4c failed (-110)
[  359.768077] Bluetooth: hci0 command 0x1001 tx timeout
[  367.847951] Bluetooth: hci0: BCM: Reading local version info failed (-110)
[  368.914277] usb 1-1.3.3: USB disconnect, device number 8
[  382.517952] usb 1-1.3.3: new full-speed USB device number 9 using dwc_otg
[  382.655417] usb 1-1.3.3: New USB device found, idVendor=0b05, idProduct=17cb
[  382.662853] usb 1-1.3.3: New USB device strings: Mfr=1, Product=2, SerialNumber=3
[  382.670714] usb 1-1.3.3: Product: BCM20702A0
[  382.675135] usb 1-1.3.3: Manufacturer: Broadcom Corp
[  382.680375] usb 1-1.3.3: SerialNumber: 5CF370607CCB
[  382.852544] Bluetooth: hci0: BCM: chip id 63
[  382.860597] Bluetooth: hci0: BCM: features 0x07
[  382.882542] Bluetooth: hci0: BCM20702A
[  382.890156] Bluetooth: hci0: BCM20702A1 (001.002.014) build 0000
.
WARN messages
.
[  385.047937] Bluetooth: hci0 command 0xfc4c tx timeout
[  393.447846] Bluetooth: hci0: BCM: Patch command fc4c failed (-110)
[  395.607791] Bluetooth: hci0 command 0x1001 tx timeout
[  403.687805] Bluetooth: hci0: BCM: Reading local version info failed (-110)

After issuing a sudo reboot the same behaviour occurs.

  1. Test dwc_otg.fiq_fsm_mask=0x3:
    The dwc_otg warning messages are logged during system boot. Bluetooth
    controller is unavailable in bluetoothctl utility.
[   29.060393] Bluetooth: Core ver 2.22
[   29.182569] NET: Registered protocol family 31
[   29.196181] Bluetooth: HCI device and connection manager initialized
[   29.651470] Bluetooth: HCI socket layer initialized
[   29.993793] Bluetooth: L2CAP socket layer initialized
[   30.277412] Bluetooth: SCO socket layer initialized
[   30.848027] usbcore: registered new interface driver btusb
[   30.973840] Bluetooth: hci0: BCM: chip id 63
[   30.988854] Bluetooth: hci0: BCM: features 0x07
[   31.022809] Bluetooth: hci0: BCM20702A
[   31.036942] Bluetooth: hci0: BCM20702A1 (001.002.014) build 0000
[   31.095355] random: crng init done
[   31.192453] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
.
. WARN messages
.
[   32.249922] usb 1-1.3.1: reset high-speed USB device number 5 using dwc_otg
[   32.492064] ieee80211 phy0: rt2x00_set_rt: Info - RT chipset 3070, rev 0201 detected
[   32.766237] ieee80211 phy0: rt2x00_set_rf: Info - RF chipset 0005 detected
[   33.097485] ieee80211 phy0: Selected rate control algorithm 'minstrel_ht'
[   33.180891] usbcore: registered new interface driver rt2800usb
[   33.449848] Bluetooth: hci0 command 0xfc4c tx timeout
[   36.149976] Bluetooth: BNEP (Ethernet Emulation) ver 1.3
[   36.164661] Bluetooth: BNEP filters: protocol multicast
[   36.219966] Bluetooth: BNEP socket layer initialized
[   38.310894] ieee80211 phy0: rt2x00lib_request_firmware: Info - Loading firmware file 'rt2870.bin'
[   38.374711] ieee80211 phy0: rt2x00lib_request_firmware: Info - Firmware detected - version: 0.36
[   39.623060] IPv6: ADDRCONF(NETDEV_UP): wlan0: link is not ready
[   41.449932] Bluetooth: hci0: BCM: Patch command fc4c failed (-110)
[   41.466976] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode

After removing bluetooth device and reinserting, dmesg --follow yields:

[  340.791354] usb 1-1.3.3: new full-speed USB device number 8 using dwc_otg
[  340.929396] usb 1-1.3.3: New USB device found, idVendor=0b05, idProduct=17cb
[  340.936900] usb 1-1.3.3: New USB device strings: Mfr=1, Product=2, SerialNumber=3
[  340.944796] usb 1-1.3.3: Product: BCM20702A0
[  340.949243] usb 1-1.3.3: Manufacturer: Broadcom Corp
[  340.954484] usb 1-1.3.3: SerialNumber: 5CF370607CCB
[  341.125844] Bluetooth: hci0: BCM: chip id 63
[  341.132945] Bluetooth: hci0: BCM: features 0x07
[  341.154702] Bluetooth: hci0: BCM20702A
[  341.160890] Bluetooth: hci0: BCM20702A1 (001.002.014) build 0000
[  341.169912] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
[  343.221492] Bluetooth: hci0 command 0xfc2e tx timeout
[  351.222157] Bluetooth: hci0: BCM: Download Minidrv command failed (-110)
[  353.302050] Bluetooth: hci0 command 0x0c03 tx timeout
[  361.462996] Bluetooth: hci0: BCM: Reset failed (-110)

The bluetooth controller remains unavailable in bluetoothctl utility. When
sudo reboot is issued, warning messages and unavailable bluetooth controller
in bluetoothctl utility remains an issue:

[   27.953488] Bluetooth: hci0: BCM: chip id 63
[   27.969584] Bluetooth: hci0: BCM: features 0x07
[   28.004469] Bluetooth: hci0: BCM20702A
[   28.019486] Bluetooth: hci0: BCM20702A1 (001.002.014) build 0000
[   28.138582] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
.
. WARN messages
.
[   28.704780] random: crng init done
[   29.400084] usb 1-1.3.1: reset high-speed USB device number 5 using dwc_otg
[   29.611893] ieee80211 phy0: rt2x00_set_rt: Info - RT chipset 3070, rev 0201 detected
[   29.871300] ieee80211 phy0: rt2x00_set_rf: Info - RF chipset 0005 detected
[   30.076870] ieee80211 phy0: Selected rate control algorithm 'minstrel_ht'
[   30.107524] usbcore: registered new interface driver rt2800usb
[   30.169992] Bluetooth: hci0 command 0xfc4c tx timeout
[   34.522420] Bluetooth: BNEP (Ethernet Emulation) ver 1.3
[   34.536574] Bluetooth: BNEP filters: protocol multicast
[   34.709994] Bluetooth: BNEP socket layer initialized
[   35.557181] ieee80211 phy0: rt2x00lib_request_firmware: Info - Loading firmware file 'rt2870.bin'
[   35.719196] ieee80211 phy0: rt2x00lib_request_firmware: Info - Firmware detected - version: 0.36
[   37.050196] IPv6: ADDRCONF(NETDEV_UP): wlan0: link is not ready
[   38.249991] Bluetooth: hci0: BCM: Patch command fc4c failed (-110)
[   40.330000] Bluetooth: hci0 command 0x0c03 tx timeout
[   48.490027] Bluetooth: hci0: BCM: Reset failed (-110)
  1. Test dwc_otg.fiq_fsm_mask=0x7: The dwc_otg warning messages are
    logged during system boot. Bluetooth controller is unavailable in bluetoothctl
    utility.
[   27.356020] Bluetooth: Core ver 2.22
[   27.420375] NET: Registered protocol family 31
[   27.420389] Bluetooth: HCI device and connection manager initialized
[   27.443424] Bluetooth: HCI socket layer initialized
[   27.443450] Bluetooth: L2CAP socket layer initialized
[   27.443552] Bluetooth: SCO socket layer initialized
[   27.597111] usbcore: registered new interface driver btusb
[   27.743706] Bluetooth: hci0: BCM: chip id 63
[   27.758745] Bluetooth: hci0: BCM: features 0x07
[   27.791734] Bluetooth: hci0: BCM20702A
[   27.805840] Bluetooth: hci0: BCM20702A1 (001.002.014) build 0000
[   27.922385] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
.
. WARN messages
.
[   28.421650] Bluetooth: hci0: BCM: Patch command fc4c failed (-61)
[   28.422669] Bluetooth: hci0: BCM: Reset failed (-61)
[   29.279963] usb 1-1.3.1: reset high-speed USB device number 5 using dwc_otg
[   29.491413] ieee80211 phy0: rt2x00_set_rt: Info - RT chipset 3070, rev 0201 detected
[   29.630572] random: crng init done
[   29.692530] ieee80211 phy0: rt2x00_set_rf: Info - RF chipset 0005 detected
[   29.856966] ieee80211 phy0: Selected rate control algorithm 'minstrel_ht'
[   29.894896] usbcore: registered new interface driver rt2800usb
[   30.489899] Bluetooth: hci0 command tx timeout
[   33.930802] Bluetooth: BNEP (Ethernet Emulation) ver 1.3
[   33.944763] Bluetooth: BNEP filters: protocol multicast
[   34.093821] Bluetooth: BNEP socket layer initialized

After removing bluetooth device and reinserting, dmesg --follow yields:

[  284.744567] usb 1-1.3.3: new full-speed USB device number 8 using dwc_otg
[  284.883684] usb 1-1.3.3: New USB device found, idVendor=0b05, idProduct=17cb
[  284.891189] usb 1-1.3.3: New USB device strings: Mfr=1, Product=2, SerialNumber=3
[  284.899062] usb 1-1.3.3: Product: BCM20702A0
[  284.903484] usb 1-1.3.3: Manufacturer: Broadcom Corp
[  284.908718] usb 1-1.3.3: SerialNumber: 5CF370607CCB
[  285.078520] Bluetooth: hci0: BCM: chip id 63
[  285.088645] Bluetooth: hci0: BCM: features 0x07
[  285.110532] Bluetooth: hci0: BCM20702A
[  285.116497] Bluetooth: hci0: BCM20702A1 (001.002.014) build 0000
[  285.724300] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
.
. WARN messages
.
[  288.254770] Bluetooth: hci0 command 0x0c03 tx timeout
[  296.815178] Bluetooth: hci0: BCM: Reset failed (-110)
[  300.837538] usb 1-1.3.3: USB disconnect, device number 8
[  301.235396] usb 1-1.3.3: new full-speed USB device number 9 using dwc_otg
[  301.375674] usb 1-1.3.3: New USB device found, idVendor=0b05, idProduct=17cb
[  301.383192] usb 1-1.3.3: New USB device strings: Mfr=1, Product=2, SerialNumber=3
[  301.391130] usb 1-1.3.3: Product: BCM20702A0
[  301.395666] usb 1-1.3.3: Manufacturer: Broadcom Corp
[  301.400827] usb 1-1.3.3: SerialNumber: 5CF370607CCB
[  301.559778] Bluetooth: hci0: BCM: chip id 63
[  301.567701] Bluetooth: hci0: BCM: features 0x07
[  301.589601] Bluetooth: hci0: BCM20702A
[  301.596658] Bluetooth: hci0: BCM20702A1 (001.002.014) build 0000
[  301.678092] WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode
[  304.735610] Bluetooth: hci0 command 0x0c03 tx timeout
[  312.815834] Bluetooth: hci0: BCM: Reset failed (-110)

The bluetooth controller remains unavailable in bluetoothctl utility. When
sudo reboot is issued, warning messages and unavailable bluetooth controller
in bluetoothctl utility remains an issue. This time there is a failure to
recognise devices plugged into the external usb hub:

[   12.349777] usb 1-1.3.3: device descriptor read/64, error -110
[   27.709778] usb 1-1.3.3: device descriptor read/64, error -110
[   27.939778] usb 1-1.3.3: new full-speed USB device number 7 using dwc_otg
[   33.069757] usb 1-1.3.3: device descriptor read/64, error -110
[   48.429772] usb 1-1.3.3: device descriptor read/64, error -110
[   48.560144] usb 1-1.3-port3: attempt power cycle
[   49.229746] usb 1-1.3.3: new full-speed USB device number 8 using dwc_otg
[   54.329819] usb 1-1.3.3: device descriptor read/8, error -110
[   59.529823] usb 1-1.3.3: device descriptor read/8, error -110
[   59.759770] usb 1-1.3.3: new full-speed USB device number 9 using dwc_otg
[   64.809854] usb 1-1.3.3: device descriptor read/8, error -110
[   70.009839] usb 1-1.3.3: device descriptor read/8, error -110
[   70.140167] usb 1-1.3-port3: unable to enumerate USB device
[   70.249827] usb 1-1.3.4: new high-speed USB device number 10 using dwc_otg
[   70.403949] usb 1-1.3.4: New USB device found, idVendor=1005, idProduct=b113
[   70.419578] usb 1-1.3.4: New USB device strings: Mfr=1, Product=2, SerialNumber=3
[   70.435529] usb 1-1.3.4: Product: USB FLASH DRIVE
[   70.448729] usb 1-1.3.4: Manufacturer:
[   70.461227] usb 1-1.3.4: SerialNumber: 0C31B95092B2D9EC
[   70.480610] usb-storage 1-1.3.4:1.0: USB Mass Storage device detected

As a final test, reverted back to using dwc_otg.fiq_fsm_mask=0x0 in
/boot/cmdline.txt. Powered off and on the external hub and Pi. There are no
dwc_otg messages logged. However the bluetooth controller is unavailable
in bluetoothctl utility:

[   27.837227] Bluetooth: Core ver 2.22
[   27.900081] NET: Registered protocol family 31
[   27.900095] Bluetooth: HCI device and connection manager initialized
[   27.919973] Bluetooth: HCI socket layer initialized
[   27.919999] Bluetooth: L2CAP socket layer initialized
[   27.920096] Bluetooth: SCO socket layer initialized
[   28.110364] usbcore: registered new interface driver btusb
[   28.252345] Bluetooth: hci0: BCM: chip id 63
[   28.267263] Bluetooth: hci0: BCM: features 0x07
[   28.300349] Bluetooth: hci0: BCM20702A
[   28.314360] Bluetooth: hci0: BCM20702A1 (001.002.014) build 0000
[   28.455001] random: crng init done
[   29.382297] Bluetooth: hci0: BCM20702A1 (001.002.014) build 1467
[   29.414260] Bluetooth: hci0: Broadcom Bluetooth Device
[   29.990067] usb 1-1.3.1: reset high-speed USB device number 5 using dwc_otg
[   30.202778] ieee80211 phy0: rt2x00_set_rt: Info - RT chipset 3070, rev 0201 detected
[   30.484892] ieee80211 phy0: rt2x00_set_rf: Info - RF chipset 0005 detected
[   30.778463] ieee80211 phy0: Selected rate control algorithm 'minstrel_ht'
[   30.850836] usbcore: registered new interface driver rt2800usb
[   31.449967] Bluetooth: hci0 command 0x0c14 tx timeout
[   33.530066] Bluetooth: hci0 command 0x0c25 tx timeout
[   35.610049] Bluetooth: hci0 command 0x0c38 tx timeout
[   35.642500] Bluetooth: BNEP (Ethernet Emulation) ver 1.3
[   35.656074] Bluetooth: BNEP filters: protocol multicast
[   35.811972] Bluetooth: BNEP socket layer initialized
[   37.689918] Bluetooth: hci0 command 0x0c05 tx timeout
[   37.952531] ieee80211 phy0: rt2x00lib_request_firmware: Info - Loading firmware file 'rt2870.bin'
[   38.020437] ieee80211 phy0: rt2x00lib_request_firmware: Info - Firmware detected - version: 0.36
[   39.363581] IPv6: ADDRCONF(NETDEV_UP): wlan0: link is not ready
[   39.769911] Bluetooth: hci0 command tx timeout

Following this, removed the bluetooth device from the external hub and then used
dmesg --follow. No further dwc_otg warning messages are logged. The bluetooth
controller is available from bluetoothctl utility. After issuing a sudo reboot,
no dwc_otg warning messages displayed. Bluetooth controller is
available in bluetoothctl utility with some evidence of command timeouts.

[   26.890096] Bluetooth: Core ver 2.22
[   26.921487] NET: Registered protocol family 31
[   26.934660] Bluetooth: HCI device and connection manager initialized
[   27.020109] Bluetooth: HCI socket layer initialized
[   27.104937] Bluetooth: L2CAP socket layer initialized
[   27.119181] Bluetooth: SCO socket layer initialized
[   27.415433] usbcore: registered new interface driver btusb
[   27.432684] FAT-fs (mmcblk0p1): Volume was not properly unmounted. Some data may be corrupt. Please run fsck.
[   27.532660] Bluetooth: hci0: BCM: chip id 63
[   27.534768] Bluetooth: hci0: BCM: features 0x07
[   27.551679] Bluetooth: hci0: rpi
[   27.553759] Bluetooth: hci0: BCM20702A1 (001.002.014) build 1467
[   27.778272] random: crng init done
[   28.701704] Bluetooth: hci0: BCM20702A1 (001.002.014) build 1467
[   28.733774] Bluetooth: hci0: Broadcom Bluetooth Device
[   28.810426] usb 1-1.3.1: reset high-speed USB device number 5 using dwc_otg
[   28.991239] ieee80211 phy0: rt2x00_set_rt: Info - RT chipset 3070, rev 0201 detected
[   29.160969] ieee80211 phy0: rt2x00_set_rf: Info - RF chipset 0005 detected
[   29.486631] ieee80211 phy0: Selected rate control algorithm 'minstrel_ht'
[   29.540633] usbcore: registered new interface driver rt2800usb
[   33.829436] Bluetooth: BNEP (Ethernet Emulation) ver 1.3
[   33.842938] Bluetooth: BNEP filters: protocol multicast
[   33.999814] Bluetooth: BNEP socket layer initialized
[   34.886385] ieee80211 phy0: rt2x00lib_request_firmware: Info - Loading firmware file 'rt2870.bin'
[   34.998550] ieee80211 phy0: rt2x00lib_request_firmware: Info - Firmware detected - version: 0.36
[   36.338398] IPv6: ADDRCONF(NETDEV_UP): wlan0: link is not ready
[   37.049702] Bluetooth: hci0 command 0x0c14 tx timeout
[   39.129833] Bluetooth: hci0 command 0x0c25 tx timeout
[   41.209858] Bluetooth: hci0 command 0x0c38 tx timeout
[   43.289789] Bluetooth: hci0 command 0x0c05 tx timeout
[   45.369722] Bluetooth: hci0 command tx timeout

In summary, setting dwc_otg.fiq_fsm_mask=0x0 in /boot/cmdline.txt, appears to
clear up the dwc_otg messages. However, the issue logged at #2494 remains with
evidence that the bluetooth controller is unavailable. The only way to make the
controller available is to take it out from the usb hub and reinsert after a few
seconds. Issuing sudo reboot results in the controller becoming unavailable
again.

I have also tried powering off the Pi and usb hub and then inserting the
bluetooth device into the Pi usb port, as suggested by @pelwell. The first time, this worked. However, for a
second and third attempt, after powering off and on (external
hub and Pi), the bluetooth controller is unavailable. This occurs with and without
the presence of dwc_otg.fiq_fsm_mask=0x0 in /boot/cmdline.txt.

[   25.000236] Bluetooth: Core ver 2.22
[   25.060142] NET: Registered protocol family 31
[   25.073411] Bluetooth: HCI device and connection manager initialized
[   25.199965] Bluetooth: HCI socket layer initialized
[   25.260031] Bluetooth: L2CAP socket layer initialized
[   25.351568] Bluetooth: SCO socket layer initialized
[   25.570883] usbcore: registered new interface driver btusb
[   26.671208] random: crng init done
[   27.119880] usb 1-1.3.1: reset high-speed USB device number 6 using dwc_otg
[   27.271010] ieee80211 phy0: rt2x00_set_rt: Info - RT chipset 3070, rev 0201 detected
[   27.322489] ieee80211 phy0: rt2x00_set_rf: Info - RF chipset 0005 detected
[   27.471682] ieee80211 phy0: Selected rate control algorithm 'minstrel_ht'
[   27.500641] usbcore: registered new interface driver rt2800usb
[   27.610078] Bluetooth: hci0 command 0x0c03 tx timeout
[   28.197981] FAT-fs (mmcblk0p1): Volume was not properly unmounted. Some data may be corrupt. Please run fsck.
[   32.970289] Bluetooth: BNEP (Ethernet Emulation) ver 1.3
[   32.983712] Bluetooth: BNEP filters: protocol multicast
[   33.136518] Bluetooth: BNEP socket layer initialized
[   35.097168] ieee80211 phy0: rt2x00lib_request_firmware: Info - Loading firmware file 'rt2870.bin'
[   35.230395] ieee80211 phy0: rt2x00lib_request_firmware: Info - Firmware detected - version: 0.36
[   35.689887] Bluetooth: hci0: BCM: Reset failed (-110)
@hofnarr132

This comment has been minimized.

Copy link
Author

hofnarr132 commented Apr 15, 2018

@P33M and @pelwell - apologies not been able to run thru /boot/cmdline.txt switches yet.

Update on daisy chained hubs... Power related? The setup that is giving me most grief is as follows:

RPi 3B+ (with USB Hub) --> PlusBerryPi *Powered Hub (with SSD drive) --> USB/HDMI KVM hub

The PlusBerryPi is powered by a 3.5 AMP 5 Volt supply with powers the RPi3B+ and provides a 7 port USB hub (1 port connected to RPi 3B+, a USB to SATA interface with a SSD drive attached, 1 port to the KVM switch). I'm using USB booting from the SSD drive (no SD Card). Additionally I'm using the onboard Wifi and HDMI.

As I was getting the occasional low power warning, I considered that the PlusBerryPi was not getting enough power through to the RPi - sadly I am not equipped to measure USB power draw, so I've replaced the 3B+ with a 3B (as it should draw less power) which is configured for USB boot (no change to the OS on the SSD) and now seem to have a 'power' stable system - however I still get WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode messages on boot or when activating (on/off button on KVM) the C-Media sound device on the KVM.

As to the RPi 3B+, I have put that in the 3B case with a 2.5 AMP 5.25 Volt supply directly to the RPi 3B, connected directly to the KVM switch (RPi 3B USB Hub -> KVM Switch hub) and booting from SD Card. This system is now 'power' stable so far and almost never showing the WARN message.

I will continue to test when able and report back - I hope to be able to run the cmdline.txt options on both RPis this week.

@pelwell

This comment has been minimized.

Copy link
Contributor

pelwell commented Apr 16, 2018

Thanks for the feedback. As you may have guessed, we still don't know why those warnings have started appearing with the 4.14 kernel. I seem the occasionally at boot time, but I usually only have a mouse and keyboard attached.

@sdcoil

This comment has been minimized.

Copy link

sdcoil commented Apr 17, 2018

FYI. I too am experiencing the same on a Zero W. And I also have a USB/HDMI KVM, however without sound. If you wish, I can post whatever you need.

@kewlrich

This comment has been minimized.

Copy link

kewlrich commented Apr 22, 2018

I have the same error WARN::dwc_otg_handle_mode_mismatch_intr:68: Mode Mismatch Interrupt: currently in Host mode" but its when I plug in an Xbox 360 Wireless dongle into a Pi case that has a USB hub, I opened issue 2357 but they closed it and told me to put it on the forums

@TheNetStriker

This comment has been minimized.

Copy link

TheNetStriker commented Apr 26, 2018

I also have this problem when plugging in a Xbox 360 Wireless dongle into my RPI 3+ using a USB hub. And if I plug the donge directly into the Pi the dongle doesn't work most of the time. (I think because of usb power issues) When plugging directly into the Pi I the the folloging errors in dmesg most of the time:

[ 1343.069199] usb 1-1.3: new full-speed USB device number 17 using dwc_otg
[ 1343.169189] usb 1-1.3: device descriptor read/64, error -32
[ 1343.389184] usb 1-1.3: device descriptor read/64, error -32
[ 1343.609187] usb 1-1.3: new full-speed USB device number 18 using dwc_otg
[ 1343.709195] usb 1-1.3: device descriptor read/64, error -32
[ 1343.939209] usb 1-1.3: device descriptor read/64, error -32
[ 1344.059342] usb 1-1-port3: attempt power cycle
[ 1344.719194] usb 1-1.3: new full-speed USB device number 19 using dwc_otg
[ 1345.159204] usb 1-1.3: device not accepting address 19, error -32
[ 1345.259192] usb 1-1.3: new full-speed USB device number 20 using dwc_otg
[ 1345.699188] usb 1-1.3: device not accepting address 20, error -32
[ 1345.699314] usb 1-1-port3: unable to enumerate USB device

Any ideas what's different regarding the usb host power on the RPI 3+?

@P33M

This comment has been minimized.

Copy link
Contributor

P33M commented Apr 26, 2018

Ah. That's a clue. Can you post a full sudo lsusb -v with the dongle plugged into the USB hub?

@TheNetStriker

This comment has been minimized.

Copy link

TheNetStriker commented Apr 26, 2018

Here is the output of the command:

pi@retropie:~ $ sudo lsusb -v

Bus 001 Device 009: ID 045e:0291 Microsoft Corp. Xbox 360 Wireless Receiver for Windows
Device Descriptor:
  bLength                18
  bDescriptorType         1
  bcdUSB               2.00
  bDeviceClass          255 Vendor Specific Class
  bDeviceSubClass       255 Vendor Specific Subclass
  bDeviceProtocol       255 Vendor Specific Protocol
  bMaxPacketSize0         8
  idVendor           0x045e Microsoft Corp.
  idProduct          0x0291 Xbox 360 Wireless Receiver for Windows
  bcdDevice            1.07
  iManufacturer           0 
  iProduct                0 
  iSerial                 0 
  bNumConfigurations      1
  Configuration Descriptor:
    bLength                 9
    bDescriptorType         2
    wTotalLength          321
    bNumInterfaces          8
    bConfigurationValue     1
    iConfiguration          0 
    bmAttributes         0xa0
      (Bus Powered)
      Remote Wakeup
    MaxPower              260mA
    Interface Descriptor:
      bLength                 9
      bDescriptorType         4
      bInterfaceNumber        0
      bAlternateSetting       0
      bNumEndpoints           2
      bInterfaceClass       255 Vendor Specific Class
      bInterfaceSubClass     93 
      bInterfaceProtocol    129 
      iInterface              0 
      ** UNRECOGNIZED:  14 22 00 01 13 81 1d 00 17 01 02 08 13 01 0c 00 0c 01 02 08
      Endpoint Descriptor:
        bLength                 7
        bDescriptorType         5
        bEndpointAddress     0x81  EP 1 IN
        bmAttributes            3
          Transfer Type            Interrupt
          Synch Type               None
          Usage Type               Data
        wMaxPacketSize     0x0020  1x 32 bytes
        bInterval               1
      Endpoint Descriptor:
        bLength                 7
        bDescriptorType         5
        bEndpointAddress     0x01  EP 1 OUT
        bmAttributes            3
          Transfer Type            Interrupt
          Synch Type               None
          Usage Type               Data
        wMaxPacketSize     0x0020  1x 32 bytes
        bInterval               8
    Interface Descriptor:
      bLength                 9
      bDescriptorType         4
      bInterfaceNumber        1
      bAlternateSetting       0
      bNumEndpoints           2
      bInterfaceClass       255 Vendor Specific Class
      bInterfaceSubClass     93 
      bInterfaceProtocol    130 
      iInterface              0 
      ** UNRECOGNIZED:  0c 22 00 01 01 82 00 40 01 02 20 00
      Endpoint Descriptor:
        bLength                 7
        bDescriptorType         5
        bEndpointAddress     0x82  EP 2 IN
        bmAttributes            3
          Transfer Type            Interrupt
          Synch Type               None
          Usage Type               Data
        wMaxPacketSize     0x0020  1x 32 bytes
        bInterval               2
      Endpoint Descriptor:
        bLength                 7
        bDescriptorType         5
        bEndpointAddress     0x02  EP 2 OUT
        bmAttributes            3
          Transfer Type            Interrupt
          Synch Type               None
          Usage Type               Data
        wMaxPacketSize     0x0020  1x 32 bytes
        bInterval               4
    Interface Descriptor:
      bLength                 9
      bDescriptorType         4
      bInterfaceNumber        2
      bAlternateSetting       0
      bNumEndpoints           2
      bInterfaceClass       255 Vendor Specific Class
      bInterfaceSubClass     93 
      bInterfaceProtocol    129 
      iInterface              0 
      ** UNRECOGNIZED:  14 22 00 01 13 83 1d 00 17 01 02 08 13 03 0c 00 0c 01 02 08
      Endpoint Descriptor:
        bLength                 7
        bDescriptorType         5
        bEndpointAddress     0x83  EP 3 IN
        bmAttributes            3
          Transfer Type            Interrupt
          Synch Type               None
          Usage Type               Data
        wMaxPacketSize     0x0020  1x 32 bytes
        bInterval               1
      Endpoint Descriptor:
        bLength                 7
        bDescriptorType         5
        bEndpointAddress     0x03  EP 3 OUT
        bmAttributes            3
          Transfer Type            Interrupt
          Synch Type               None
          Usage Type               Data
        wMaxPacketSize     0x0020  1x 32 bytes
        bInterval               8
    Interface Descriptor:
      bLength                 9
      bDescriptorType         4
      bInterfaceNumber        3
      bAlternateSetting       0
      bNumEndpoints           2
      bInterfaceClass       255 Vendor Specific Class
      bInterfaceSubClass     93 
      bInterfaceProtocol    130 
      iInterface              0 
      ** UNRECOGNIZED:  0c 22 00 01 01 84 00 40 01 04 20 00
      Endpoint Descriptor:
        bLength                 7
        bDescriptorType         5
        bEndpointAddress     0x84  EP 4 IN
        bmAttributes            3
          Transfer Type            Interrupt
          Synch Type               None
          Usage Type               Data
        wMaxPacketSize     0x0020  1x 32 bytes
        bInterval               2
      Endpoint Descriptor:
        bLength                 7
        bDescriptorType         5
        bEndpointAddress     0x04  EP 4 OUT
        bmAttributes            3
          Transfer Type            Interrupt
          Synch Type               None
          Usage Type               Data
        wMaxPacketSize     0x0020  1x 32 bytes
        bInterval               4
    Interface Descriptor:
      bLength                 9
      bDescriptorType         4
      bInterfaceNumber        4
      bAlternateSetting       0
      bNumEndpoints           2
      bInterfaceClass       255 Vendor Specific Class
      bInterfaceSubClass     93 
      bInterfaceProtocol    129 
      iInterface              0 
      ** UNRECOGNIZED:  14 22 00 01 13 85 1d 00 17 01 02 08 13 05 0c 00 0c 01 02 08
      Endpoint Descriptor:
        bLength                 7
        bDescriptorType         5
        bEndpointAddress     0x85  EP 5 IN
        bmAttributes            3
          Transfer Type            Interrupt
          Synch Type               None
          Usage Type               Data
        wMaxPacketSize     0x0020  1x 32 bytes
        bInterval               1
      Endpoint Descriptor:
        bLength                 7
        bDescriptorType         5
        bEndpointAddress     0x05  EP 5 OUT
        bmAttributes            3
          Transfer Type            Interrupt
          Synch Type               None
          Usage Type               Data
        wMaxPacketSize     0x0020  1x 32 bytes
        bInterval               8
    Interface Descriptor:
      bLength                 9
      bDescriptorType         4
      bInterfaceNumber        5
      bAlternateSetting       0
      bNumEndpoints           2
      bInterfaceClass       255 Vendor Specific Class
      bInterfaceSubClass     93 
      bInterfaceProtocol    130 
      iInterface              0 
      ** UNRECOGNIZED:  0c 22 00 01 01 86 00 40 01 06 20 00
      Endpoint Descriptor:
        bLength                 7
        bDescriptorType         5
        bEndpointAddress     0x86  EP 6 IN
        bmAttributes            3
          Transfer Type            Interrupt
          Synch Type               None
          Usage Type               Data
        wMaxPacketSize     0x0020  1x 32 bytes
        bInterval               2
      Endpoint Descriptor:
        bLength                 7
        bDescriptorType         5
        bEndpointAddress     0x06  EP 6 OUT
        bmAttributes            3
          Transfer Type            Interrupt
          Synch Type               None
          Usage Type               Data
        wMaxPacketSize     0x0020  1x 32 bytes
        bInterval               4
    Interface Descriptor:
      bLength                 9
      bDescriptorType         4
      bInterfaceNumber        6
      bAlternateSetting       0
      bNumEndpoints           2
      bInterfaceClass       255 Vendor Specific Class
      bInterfaceSubClass     93 
      bInterfaceProtocol    129 
      iInterface              0 
      ** UNRECOGNIZED:  14 22 00 01 13 87 1d 00 17 01 02 08 13 07 0c 00 0c 01 02 08
      Endpoint Descriptor:
        bLength                 7
        bDescriptorType         5
        bEndpointAddress     0x87  EP 7 IN
        bmAttributes            3
          Transfer Type            Interrupt
          Synch Type               None
          Usage Type               Data
        wMaxPacketSize     0x0020  1x 32 bytes
        bInterval               1
      Endpoint Descriptor:
        bLength                 7
        bDescriptorType         5
        bEndpointAddress     0x07  EP 7 OUT
        bmAttributes            3
          Transfer Type            Interrupt
          Synch Type               None
          Usage Type               Data
        wMaxPacketSize     0x0020  1x 32 bytes
        bInterval               8
    Interface Descriptor:
      bLength                 9
      bDescriptorType         4
      bInterfaceNumber        7
      bAlternateSetting       0
      bNumEndpoints           2
      bInterfaceClass       255 Vendor Specific Class
      bInterfaceSubClass     93 
      bInterfaceProtocol    130 
      iInterface              0 
      ** UNRECOGNIZED:  0c 22 00 01 01 88 00 40 01 08 20 00
      Endpoint Descriptor:
        bLength                 7
        bDescriptorType         5
        bEndpointAddress     0x88  EP 8 IN
        bmAttributes            3
          Transfer Type            Interrupt
          Synch Type               None
          Usage Type               Data
        wMaxPacketSize     0x0020  1x 32 bytes
        bInterval               2
      Endpoint Descriptor:
        bLength                 7
        bDescriptorType         5
        bEndpointAddress     0x08  EP 8 OUT
        bmAttributes            3
          Transfer Type            Interrupt
          Synch Type               None
          Usage Type               Data
        wMaxPacketSize     0x0020  1x 32 bytes
        bInterval               4
Device Status:     0x0007
  Self Powered
  Remote Wakeup Enabled
  Test Mode

Bus 001 Device 008: ID 05e3:0608 Genesys Logic, Inc. Hub
Device Descriptor:
  bLength                18
  bDescriptorType         1
  bcdUSB               2.10
  bDeviceClass            9 Hub
  bDeviceSubClass         0 Unused
  bDeviceProtocol         1 Single TT
  bMaxPacketSize0        64
  idVendor           0x05e3 Genesys Logic, Inc.
  idProduct          0x0608 Hub
  bcdDevice           60.52
  iManufacturer           0 
  iProduct                1 USB2.0 Hub
  iSerial                 0 
  bNumConfigurations      1
  Configuration Descriptor:
    bLength                 9
    bDescriptorType         2
    wTotalLength           25
    bNumInterfaces          1
    bConfigurationValue     1
    iConfiguration          0 
    bmAttributes         0xe0
      Self Powered
      Remote Wakeup
    MaxPower              100mA
    Interface Descriptor:
      bLength                 9
      bDescriptorType         4
      bInterfaceNumber        0
      bAlternateSetting       0
      bNumEndpoints           1
      bInterfaceClass         9 Hub
      bInterfaceSubClass      0 Unused
      bInterfaceProtocol      0 Full speed (or root) hub
      iInterface              0 
      Endpoint Descriptor:
        bLength                 7
        bDescriptorType         5
        bEndpointAddress     0x81  EP 1 IN
        bmAttributes            3
          Transfer Type            Interrupt
          Synch Type               None
          Usage Type               Data
        wMaxPacketSize     0x0001  1x 1 bytes
        bInterval              12
Hub Descriptor:
  bLength               9
  bDescriptorType      41
  nNbrPorts             4
  wHubCharacteristic 0x00e9
    Per-port power switching
    Per-port overcurrent protection
    TT think time 32 FS bits
    Port indicators
  bPwrOn2PwrGood       50 * 2 milli seconds
  bHubContrCurrent    100 milli Ampere
  DeviceRemovable    0x00
  PortPwrCtrlMask    0xff
 Hub Port Status:
   Port 1: 0000.0100 power
   Port 2: 0000.0100 power
   Port 3: 0000.0100 power
   Port 4: 0000.0103 power enable connect
Binary Object Store Descriptor:
  bLength                 5
  bDescriptorType        15
  wTotalLength           32
  bNumDeviceCaps          2
  USB 2.0 Extension Device Capability:
    bLength                 7
    bDescriptorType        16
    bDevCapabilityType      2
    bmAttributes   0x00000006
      Link Power Management (LPM) Supported
  Container ID Device Capability:
    bLength                20
    bDescriptorType        16
    bDevCapabilityType      4
    bReserved               0
    ContainerID             {7019bf3a-baae-e411-ab27-0800200c9a66}
Device Status:     0x0001
  Self Powered

Bus 001 Device 007: ID 05e3:0608 Genesys Logic, Inc. Hub
Device Descriptor:
  bLength                18
  bDescriptorType         1
  bcdUSB               2.10
  bDeviceClass            9 Hub
  bDeviceSubClass         0 Unused
  bDeviceProtocol         1 Single TT
  bMaxPacketSize0        64
  idVendor           0x05e3 Genesys Logic, Inc.
  idProduct          0x0608 Hub
  bcdDevice           60.52
  iManufacturer           0 
  iProduct                1 USB2.0 Hub
  iSerial                 0 
  bNumConfigurations      1
  Configuration Descriptor:
    bLength                 9
    bDescriptorType         2
    wTotalLength           25
    bNumInterfaces          1
    bConfigurationValue     1
    iConfiguration          0 
    bmAttributes         0xe0
      Self Powered
      Remote Wakeup
    MaxPower              100mA
    Interface Descriptor:
      bLength                 9
      bDescriptorType         4
      bInterfaceNumber        0
      bAlternateSetting       0
      bNumEndpoints           1
      bInterfaceClass         9 Hub
      bInterfaceSubClass      0 Unused
      bInterfaceProtocol      0 Full speed (or root) hub
      iInterface              0 
      Endpoint Descriptor:
        bLength                 7
        bDescriptorType         5
        bEndpointAddress     0x81  EP 1 IN
        bmAttributes            3
          Transfer Type            Interrupt
          Synch Type               None
          Usage Type               Data
        wMaxPacketSize     0x0001  1x 1 bytes
        bInterval              12
Hub Descriptor:
  bLength               9
  bDescriptorType      41
  nNbrPorts             4
  wHubCharacteristic 0x00ed
    Per-port power switching
    Compound device
    Per-port overcurrent protection
    TT think time 32 FS bits
    Port indicators
  bPwrOn2PwrGood       50 * 2 milli seconds
  bHubContrCurrent    100 milli Ampere
  DeviceRemovable    0x10
  PortPwrCtrlMask    0xff
 Hub Port Status:
   Port 1: 0000.0100 power
   Port 2: 0000.0100 power
   Port 3: 0000.0100 power
   Port 4: 0000.0503 highspeed power enable connect
Binary Object Store Descriptor:
  bLength                 5
  bDescriptorType        15
  wTotalLength           32
  bNumDeviceCaps          2
  USB 2.0 Extension Device Capability:
    bLength                 7
    bDescriptorType        16
    bDevCapabilityType      2
    bmAttributes   0x00000006
      Link Power Management (LPM) Supported
  Container ID Device Capability:
    bLength                20
    bDescriptorType        16
    bDevCapabilityType      4
    bReserved               0
    ContainerID             {7019bf3a-baae-e411-ab27-0800200c9a66}
Device Status:     0x0001
  Self Powered

Bus 001 Device 004: ID 0424:7800 Standard Microsystems Corp. 
Device Descriptor:
  bLength                18
  bDescriptorType         1
  bcdUSB               2.10
  bDeviceClass          255 Vendor Specific Class
  bDeviceSubClass         0 
  bDeviceProtocol       255 
  bMaxPacketSize0        64
  idVendor           0x0424 Standard Microsystems Corp.
  idProduct          0x7800 
  bcdDevice            3.00
  iManufacturer           0 
  iProduct                0 
  iSerial                 0 
  bNumConfigurations      1
  Configuration Descriptor:
    bLength                 9
    bDescriptorType         2
    wTotalLength           39
    bNumInterfaces          1
    bConfigurationValue     1
    iConfiguration          0 
    bmAttributes         0xe0
      Self Powered
      Remote Wakeup
    MaxPower                2mA
    Interface Descriptor:
      bLength                 9
      bDescriptorType         4
      bInterfaceNumber        0
      bAlternateSetting       0
      bNumEndpoints           3
      bInterfaceClass       255 Vendor Specific Class
      bInterfaceSubClass      0 
      bInterfaceProtocol    255 
      iInterface              0 
      Endpoint Descriptor:
        bLength                 7
        bDescriptorType         5
        bEndpointAddress     0x81  EP 1 IN
        bmAttributes            2
          Transfer Type            Bulk
          Synch Type               None
          Usage Type               Data
        wMaxPacketSize     0x0200  1x 512 bytes
        bInterval               0
      Endpoint Descriptor:
        bLength                 7
        bDescriptorType         5
        bEndpointAddress     0x02  EP 2 OUT
        bmAttributes            2
          Transfer Type            Bulk
          Synch Type               None
          Usage Type               Data
        wMaxPacketSize     0x0200  1x 512 bytes
        bInterval               0
      Endpoint Descriptor:
        bLength                 7
        bDescriptorType         5
        bEndpointAddress     0x83  EP 3 IN
        bmAttributes            3
          Transfer Type            Interrupt
          Synch Type               None
          Usage Type               Data
        wMaxPacketSize     0x0010  1x 16 bytes
        bInterval               4
Binary Object Store Descriptor:
  bLength                 5
  bDescriptorType        15
  wTotalLength           22
  bNumDeviceCaps          2
  USB 2.0 Extension Device Capability:
    bLength                 7
    bDescriptorType        16
    bDevCapabilityType      2
    bmAttributes   0x00000006
      Link Power Management (LPM) Supported
  SuperSpeed USB Device Capability:
    bLength                10
    bDescriptorType        16
    bDevCapabilityType      3
    bmAttributes         0x00
    wSpeedsSupported   0x000e
      Device can operate at Full Speed (12Mbps)
      Device can operate at High Speed (480Mbps)
      Device can operate at SuperSpeed (5Gbps)
    bFunctionalitySupport   1
      Lowest fully-functional device speed is Full Speed (12Mbps)
    bU1DevExitLat          10 micro seconds
    bU2DevExitLat        1500 micro seconds
Device Status:     0x0001
  Self Powered

Bus 001 Device 003: ID 0424:2514 Standard Microsystems Corp. USB 2.0 Hub
Device Descriptor:
  bLength                18
  bDescriptorType         1
  bcdUSB               2.00
  bDeviceClass            9 Hub
  bDeviceSubClass         0 Unused
  bDeviceProtocol         2 TT per port
  bMaxPacketSize0        64
  idVendor           0x0424 Standard Microsystems Corp.
  idProduct          0x2514 USB 2.0 Hub
  bcdDevice            b.b3
  iManufacturer           0 
  iProduct                0 
  iSerial                 0 
  bNumConfigurations      1
  Configuration Descriptor:
    bLength                 9
    bDescriptorType         2
    wTotalLength           41
    bNumInterfaces          1
    bConfigurationValue     1
    iConfiguration          0 
    bmAttributes         0xe0
      Self Powered
      Remote Wakeup
    MaxPower                2mA
    Interface Descriptor:
      bLength                 9
      bDescriptorType         4
      bInterfaceNumber        0
      bAlternateSetting       0
      bNumEndpoints           1
      bInterfaceClass         9 Hub
      bInterfaceSubClass      0 Unused
      bInterfaceProtocol      1 Single TT
      iInterface              0 
      Endpoint Descriptor:
        bLength                 7
        bDescriptorType         5
        bEndpointAddress     0x81  EP 1 IN
        bmAttributes            3
          Transfer Type            Interrupt
          Synch Type               None
          Usage Type               Data
        wMaxPacketSize     0x0001  1x 1 bytes
        bInterval              12
    Interface Descriptor:
      bLength                 9
      bDescriptorType         4
      bInterfaceNumber        0
      bAlternateSetting       1
      bNumEndpoints           1
      bInterfaceClass         9 Hub
      bInterfaceSubClass      0 Unused
      bInterfaceProtocol      2 TT per port
      iInterface              0 
      Endpoint Descriptor:
        bLength                 7
        bDescriptorType         5
        bEndpointAddress     0x81  EP 1 IN
        bmAttributes            3
          Transfer Type            Interrupt
          Synch Type               None
          Usage Type               Data
        wMaxPacketSize     0x0001  1x 1 bytes
        bInterval              12
Hub Descriptor:
  bLength               9
  bDescriptorType      41
  nNbrPorts             3
  wHubCharacteristic 0x000d
    Per-port power switching
    Compound device
    Per-port overcurrent protection
    TT think time 8 FS bits
  bPwrOn2PwrGood       50 * 2 milli seconds
  bHubContrCurrent      1 milli Ampere
  DeviceRemovable    0x02
  PortPwrCtrlMask    0xff
 Hub Port Status:
   Port 1: 0000.0503 highspeed power enable connect
   Port 2: 0000.0100 power
   Port 3: 0000.0100 power
Device Qualifier (for other device speed):
  bLength                10
  bDescriptorType         6
  bcdUSB               2.00
  bDeviceClass            9 Hub
  bDeviceSubClass         0 Unused
  bDeviceProtocol         0 Full speed (or root) hub
  bMaxPacketSize0        64
  bNumConfigurations      1
Device Status:     0x0001
  Self Powered

Bus 001 Device 002: ID 0424:2514 Standard Microsystems Corp. USB 2.0 Hub
Device Descriptor:
  bLength                18
  bDescriptorType         1
  bcdUSB               2.00
  bDeviceClass            9 Hub
  bDeviceSubClass         0 Unused
  bDeviceProtocol         2 TT per port
  bMaxPacketSize0        64
  idVendor           0x0424 Standard Microsystems Corp.
  idProduct          0x2514 USB 2.0 Hub
  bcdDevice            b.b3
  iManufacturer           0 
  iProduct                0 
  iSerial                 0 
  bNumConfigurations      1
  Configuration Descriptor:
    bLength                 9
    bDescriptorType         2
    wTotalLength           41
    bNumInterfaces          1
    bConfigurationValue     1
    iConfiguration          0 
    bmAttributes         0xe0
      Self Powered
      Remote Wakeup
    MaxPower                2mA
    Interface Descriptor:
      bLength                 9
      bDescriptorType         4
      bInterfaceNumber        0
      bAlternateSetting       0
      bNumEndpoints           1
      bInterfaceClass         9 Hub
      bInterfaceSubClass      0 Unused
      bInterfaceProtocol      1 Single TT
      iInterface              0 
      Endpoint Descriptor:
        bLength                 7
        bDescriptorType         5
        bEndpointAddress     0x81  EP 1 IN
        bmAttributes            3
          Transfer Type            Interrupt
          Synch Type               None
          Usage Type               Data
        wMaxPacketSize     0x0001  1x 1 bytes
        bInterval              12
    Interface Descriptor:
      bLength                 9
      bDescriptorType         4
      bInterfaceNumber        0
      bAlternateSetting       1
      bNumEndpoints           1
      bInterfaceClass         9 Hub
      bInterfaceSubClass      0 Unused
      bInterfaceProtocol      2 TT per port
      iInterface              0 
      Endpoint Descriptor:
        bLength                 7
        bDescriptorType         5
        bEndpointAddress     0x81  EP 1 IN
        bmAttributes            3
          Transfer Type            Interrupt
          Synch Type               None
          Usage Type               Data
        wMaxPacketSize     0x0001  1x 1 bytes
        bInterval              12
Hub Descriptor:
  bLength               9
  bDescriptorType      41
  nNbrPorts             4
  wHubCharacteristic 0x000d
    Per-port power switching
    Compound device
    Per-port overcurrent protection
    TT think time 8 FS bits
  bPwrOn2PwrGood       50 * 2 milli seconds
  bHubContrCurrent      1 milli Ampere
  DeviceRemovable    0x02
  PortPwrCtrlMask    0xff
 Hub Port Status:
   Port 1: 0000.0503 highspeed power enable connect
   Port 2: 0000.0100 power
   Port 3: 0000.0503 highspeed power enable connect
   Port 4: 0000.0100 power
Device Qualifier (for other device speed):
  bLength                10
  bDescriptorType         6
  bcdUSB               2.00
  bDeviceClass            9 Hub
  bDeviceSubClass         0 Unused
  bDeviceProtocol         0 Full speed (or root) hub
  bMaxPacketSize0        64
  bNumConfigurations      1
Device Status:     0x0001
  Self Powered

Bus 001 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub
Device Descriptor:
  bLength                18
  bDescriptorType         1
  bcdUSB               2.00
  bDeviceClass            9 Hub
  bDeviceSubClass         0 Unused
  bDeviceProtocol         1 Single TT
  bMaxPacketSize0        64
  idVendor           0x1d6b Linux Foundation
  idProduct          0x0002 2.0 root hub
  bcdDevice            4.14
  iManufacturer           3 Linux 4.14.34-v7+ dwc_otg_hcd
  iProduct                2 DWC OTG Controller
  iSerial                 1 3f980000.usb
  bNumConfigurations      1
  Configuration Descriptor:
    bLength                 9
    bDescriptorType         2
    wTotalLength           25
    bNumInterfaces          1
    bConfigurationValue     1
    iConfiguration          0 
    bmAttributes         0xe0
      Self Powered
      Remote Wakeup
    MaxPower                0mA
    Interface Descriptor:
      bLength                 9
      bDescriptorType         4
      bInterfaceNumber        0
      bAlternateSetting       0
      bNumEndpoints           1
      bInterfaceClass         9 Hub
      bInterfaceSubClass      0 Unused
      bInterfaceProtocol      0 Full speed (or root) hub
      iInterface              0 
      Endpoint Descriptor:
        bLength                 7
        bDescriptorType         5
        bEndpointAddress     0x81  EP 1 IN
        bmAttributes            3
          Transfer Type            Interrupt
          Synch Type               None
          Usage Type               Data
        wMaxPacketSize     0x0004  1x 4 bytes
        bInterval              12
Hub Descriptor:
  bLength               9
  bDescriptorType      41
  nNbrPorts             1
  wHubCharacteristic 0x0008
    Ganged power switching
    Per-port overcurrent protection
    TT think time 8 FS bits
  bPwrOn2PwrGood        1 * 2 milli seconds
  bHubContrCurrent      0 milli Ampere
  DeviceRemovable    0x00
  PortPwrCtrlMask    0xff
 Hub Port Status:
   Port 1: 0000.0503 highspeed power enable connect
Device Status:     0x0001
  Self Powered

This time the dongle also worked when plugging in directly to the Pi. Maybe this is only a problem when the device is running for a while or when the temperature is higher.

popcornmix added a commit that referenced this issue Feb 12, 2019

Add dwc_otg driver
Signed-off-by: popcornmix <popcornmix@gmail.com>

usb: dwc: fix lockdep false positive

Signed-off-by: Kari Suvanto <karis79@gmail.com>

usb: dwc: fix inconsistent lock state

Signed-off-by: Kari Suvanto <karis79@gmail.com>

Add FIQ patch to dwc_otg driver. Enable with dwc_otg.fiq_fix_enable=1. Should give about 10% more ARM performance.
Thanks to Gordon and Costas

Avoid dynamic memory allocation for channel lock in USB driver. Thanks ddv2005.

Add NAK holdoff scheme. Enabled by default, disable with dwc_otg.nak_holdoff_enable=0. Thanks gsh

Make sure we wait for the reset to finish

dwc_otg: fix bug in dwc_otg_hcd.c resulting in silent kernel
	 memory corruption, escalating to OOPS under high USB load.

dwc_otg: Fix unsafe access of QTD during URB enqueue

In dwc_otg_hcd_urb_enqueue during qtd creation, it was possible that the
transaction could complete almost immediately after the qtd was assigned
to a host channel during URB enqueue, which meant the qtd pointer was no
longer valid having been completed and removed. Usually, this resulted in
an OOPS during URB submission. By predetermining whether transactions
need to be queued or not, this unsafe pointer access is avoided.

This bug was only evident on the Pi model A where a device was attached
that had no periodic endpoints (e.g. USB pendrive or some wlan devices).

dwc_otg: Fix incorrect URB allocation error handling

If the memory allocation for a dwc_otg_urb failed, the kernel would OOPS
because for some reason a member of the *unallocated* struct was set to
zero. Error handling changed to fail correctly.

dwc_otg: fix potential use-after-free case in interrupt handler

If a transaction had previously aborted, certain interrupts are
enabled to track error counts and reset where necessary. On IN
endpoints the host generates an ACK interrupt near-simultaneously
with completion of transfer. In the case where this transfer had
previously had an error, this results in a use-after-free on
the QTD memory space with a 1-byte length being overwritten to
0x00.

dwc_otg: add handling of SPLIT transaction data toggle errors

Previously a data toggle error on packets from a USB1.1 device behind
a TT would result in the Pi locking up as the driver never handled
the associated interrupt. Patch adds basic retry mechanism and
interrupt acknowledgement to cater for either a chance toggle error or
for devices that have a broken initial toggle state (FT8U232/FT232BM).

dwc_otg: implement tasklet for returning URBs to usbcore hcd layer

The dwc_otg driver interrupt handler for transfer completion will spend
a very long time with interrupts disabled when a URB is completed -
this is because usb_hcd_giveback_urb is called from within the handler
which for a USB device driver with complicated processing (e.g. webcam)
will take an exorbitant amount of time to complete. This results in
missed completion interrupts for other USB packets which lead to them
being dropped due to microframe overruns.

This patch splits returning the URB to the usb hcd layer into a
high-priority tasklet. This will have most benefit for isochronous IN
transfers but will also have incidental benefit where multiple periodic
devices are active at once.

dwc_otg: fix NAK holdoff and allow on split transactions only

This corrects a bug where if a single active non-periodic endpoint
had at least one transaction in its qh, on frnum == MAX_FRNUM the qh
would get skipped and never get queued again. This would result in
a silent device until error detection (automatic or otherwise) would
either reset the device or flush and requeue the URBs.

Additionally the NAK holdoff was enabled for all transactions - this
would potentially stall a HS endpoint for 1ms if a previous error state
enabled this interrupt and the next response was a NAK. Fix so that
only split transactions get held off.

dwc_otg: Call usb_hcd_unlink_urb_from_ep with lock held in completion handler

usb_hcd_unlink_urb_from_ep must be called with the HCD lock held.  Calling it
asynchronously in the tasklet was not safe (regression in
c4564d4).

This change unlinks it from the endpoint prior to queueing it for handling in
the tasklet, and also adds a check to ensure the urb is OK to be unlinked
before doing so.

NULL pointer dereference kernel oopses had been observed in usb_hcd_giveback_urb
when a USB device was unplugged/replugged during data transfer.  This effect
was reproduced using automated USB port power control, hundreds of replug
events were performed during active transfers to confirm that the problem was
eliminated.

USB fix using a FIQ to implement split transactions

This commit adds a FIQ implementaion that schedules
the split transactions using a FIQ so we don't get
held off by the interrupt latency of Linux

dwc_otg: fix device attributes and avoid kernel warnings on boot

dcw_otg: avoid logging function that can cause panics

See: raspberrypi/firmware#21
Thanks to cleverca22 for fix

dwc_otg: mask correct interrupts after transaction error recovery

The dwc_otg driver will unmask certain interrupts on a transaction
that previously halted in the error state in order to reset the
QTD error count. The various fine-grained interrupt handlers do not
consider that other interrupts besides themselves were unmasked.

By disabling the two other interrupts only ever enabled in DMA mode
for this purpose, we can avoid unnecessary function calls in the
IRQ handler. This will also prevent an unneccesary FIQ interrupt
from being generated if the FIQ is enabled.

dwc_otg: fiq: prevent FIQ thrash and incorrect state passing to IRQ

In the case of a transaction to a device that had previously aborted
due to an error, several interrupts are enabled to reset the error
count when a device responds. This has the side-effect of making the
FIQ thrash because the hardware will generate multiple instances of
a NAK on an IN bulk/interrupt endpoint and multiple instances of ACK
on an OUT bulk/interrupt endpoint. Make the FIQ mask and clear the
associated interrupts.

Additionally, on non-split transactions make sure that only unmasked
interrupts are cleared. This caused a hard-to-trigger but serious
race condition when you had the combination of an endpoint awaiting
error recovery and a transaction completed on an endpoint - due to
the sequencing and timing of interrupts generated by the dwc_otg core,
it was possible to confuse the IRQ handler.

Fix function tracing

dwc_otg: whitespace cleanup in dwc_otg_urb_enqueue

dwc_otg: prevent OOPSes during device disconnects

The dwc_otg_urb_enqueue function is thread-unsafe. In particular the
access of urb->hcpriv, usb_hcd_link_urb_to_ep, dwc_otg_urb->qtd and
friends does not occur within a critical section and so if a device
was unplugged during activity there was a high chance that the
usbcore hub_thread would try to disable the endpoint with partially-
formed entries in the URB queue. This would result in BUG() or null
pointer dereferences.

Fix so that access of urb->hcpriv, enqueuing to the hardware and
adding to usbcore endpoint URB lists is contained within a single
critical section.

dwc_otg: prevent BUG() in TT allocation if hub address is > 16

A fixed-size array is used to track TT allocation. This was
previously set to 16 which caused a crash because
dwc_otg_hcd_allocate_port would read past the end of the array.

This was hit if a hub was plugged in which enumerated as addr > 16,
due to previous device resets or unplugs.

Also add #ifdef FIQ_DEBUG around hcd->hub_port_alloc[], which grows
to a large size if 128 hub addresses are supported. This field is
for debug only for tracking which frame an allocate happened in.

dwc_otg: make channel halts with unknown state less damaging

If the IRQ received a channel halt interrupt through the FIQ
with no other bits set, the IRQ would not release the host
channel and never complete the URB.

Add catchall handling to treat as a transaction error and retry.

dwc_otg: fiq_split: use TTs with more granularity

This fixes certain issues with split transaction scheduling.

- Isochronous multi-packet OUT transactions now hog the TT until
  they are completed - this prevents hubs aborting transactions
  if they get a periodic start-split out-of-order
- Don't perform TT allocation on non-periodic endpoints - this
  allows simultaneous use of the TT's bulk/control and periodic
  transaction buffers

This commit will mainly affect USB audio playback.

dwc_otg: fix potential sleep while atomic during urb enqueue

Fixes a regression introduced with eb1b482. Kmalloc called from
dwc_otg_hcd_qtd_add / dwc_otg_hcd_qtd_create did not always have
the GPF_ATOMIC flag set. Force this flag when inside the larger
critical section.

dwc_otg: make fiq_split_enable imply fiq_fix_enable

Failing to set up the FIQ correctly would result in
"IRQ 32: nobody cared" errors in dmesg.

dwc_otg: prevent crashes on host port disconnects

Fix several issues resulting in crashes or inconsistent state
if a Model A root port was disconnected.

- Clean up queue heads properly in kill_urbs_in_qh_list by
  removing the empty QHs from the schedule lists
- Set the halt status properly to prevent IRQ handlers from
  using freed memory
- Add fiq_split related cleanup for saved registers
- Make microframe scheduling reclaim host channels if
  active during a disconnect
- Abort URBs with -ESHUTDOWN status response, informing
  device drivers so they respond in a more correct fashion
  and don't try to resubmit URBs
- Prevent IRQ handlers from attempting to handle channel
  interrupts if the associated URB was dequeued (and the
  driver state was cleared)

dwc_otg: prevent leaking URBs during enqueue

A dwc_otg_urb would get leaked if the HCD enqueue function
failed for any reason. Free the URB at the appropriate points.

dwc_otg: Enable NAK holdoff for control split transactions

Certain low-speed devices take a very long time to complete a
data or status stage of a control transaction, producing NAK
responses until they complete internal processing - the USB2.0
spec limit is up to 500mS. This causes the same type of interrupt
storm as seen with USB-serial dongles prior to c8edb23.

In certain circumstances, usually while booting, this interrupt
storm could cause SD card timeouts.

dwc_otg: Fix for occasional lockup on boot when doing a USB reset

dwc_otg: Don't issue traffic to LS devices in FS mode

Issuing low-speed packets when the root port is in full-speed mode
causes the root port to stop responding. Explicitly fail when
enqueuing URBs to a LS endpoint on a FS bus.

Fix ARM architecture issue with local_irq_restore()

If local_fiq_enable() is called before a local_irq_restore(flags) where
the flags variable has the F bit set, the FIQ will be erroneously disabled.

Fixup arch_local_irq_restore to avoid trampling the F bit in CPSR.

Also fix some of the hacks previously implemented for previous dwc_otg
incarnations.

dwc_otg: fiq_fsm: Base commit for driver rewrite

This commit removes the previous FIQ fixes entirely and adds fiq_fsm.

This rewrite features much more complete support for split transactions
and takes into account several OTG hardware bugs. High-speed
isochronous transactions are also capable of being performed by fiq_fsm.

All driver options have been removed and replaced with:
  - dwc_otg.fiq_enable (bool)
  - dwc_otg.fiq_fsm_enable (bool)
  - dwc_otg.fiq_fsm_mask (bitmask)
  - dwc_otg.nak_holdoff (unsigned int)

Defaults are specified such that fiq_fsm behaves similarly to the
previously implemented FIQ fixes.

fiq_fsm: Push error recovery into the FIQ when fiq_fsm is used

If the transfer associated with a QTD failed due to a bus error, the HCD
would retry the transfer up to 3 times (implementing the USB2.0
three-strikes retry in software).

Due to the masking mechanism used by fiq_fsm, it is only possible to pass
a single interrupt through to the HCD per-transfer.

In this instance host channels would fall off the radar because the error
reset would function, but the subsequent channel halt would be lost.

Push the error count reset into the FIQ handler.

fiq_fsm: Implement timeout mechanism

For full-speed endpoints with a large packet size, interrupt latency
runs the risk of the FIQ starting a transaction too late in a full-speed
frame. If the device is still transmitting data when EOF2 for the
downstream frame occurs, the hub will disable the port. This change is
not reflected in the hub status endpoint and the device becomes
unresponsive.

Prevent high-bandwidth transactions from being started too late in a
frame. The mechanism is not guaranteed: a combination of bit stuffing
and hub latency may still result in a device overrunning.

fiq_fsm: fix bounce buffer utilisation for Isochronous OUT

Multi-packet isochronous OUT transactions were subject to a few bounday
bugs. Fix them.

Audio playback is now much more robust: however, an issue stands with
devices that have adaptive sinks - ALSA plays samples too fast.

dwc_otg: Return full-speed frame numbers in HS mode

The frame counter increments on every *microframe* in high-speed mode.
Most device drivers expect this number to be in full-speed frames - this
caused considerable confusion to e.g. snd_usb_audio which uses the
frame counter to estimate the number of samples played.

fiq_fsm: save PID on completion of interrupt OUT transfers

Also add edge case handling for interrupt transports.

Note that for periodic split IN, data toggles are unimplemented in the
OTG host hardware - it unconditionally accepts any PID.

fiq_fsm: add missing case for fiq_fsm_tt_in_use()

Certain combinations of bitrate and endpoint activity could
result in a periodic transaction erroneously getting started
while the previous Isochronous OUT was still active.

fiq_fsm: clear hcintmsk for aborted transactions

Prevents the FIQ from erroneously handling interrupts
on a timed out channel.

fiq_fsm: enable by default

fiq_fsm: fix dequeues for non-periodic split transactions

If a dequeue happened between the SSPLIT and CSPLIT phases of the
transaction, the HCD would never receive an interrupt.

fiq_fsm: Disable by default

fiq_fsm: Handle HC babble errors

The HCTSIZ transfer size field raises a babble interrupt if
the counter wraps. Handle the resulting interrupt in this case.

dwc_otg: fix interrupt registration for fiq_enable=0

Additionally make the module parameter conditional for wherever
hcd->fiq_state is touched.

fiq_fsm: Enable by default

dwc_otg: Fix various issues with root port and transaction errors

Process the host port interrupts correctly (and don't trample them).
Root port hotplug now functional again.

Fix a few thinkos with the transaction error passthrough for fiq_fsm.

fiq_fsm: Implement hack for Split Interrupt transactions

Hubs aren't too picky about which endpoint we send Control type split
transactions to. By treating Interrupt transfers as Control, it is
possible to use the non-periodic queue in the OTG core as well as the
non-periodic FIFOs in the hub itself. This massively reduces the
microframe exclusivity/contention that periodic split transactions
otherwise have to enforce.

It goes without saying that this is a fairly egregious USB specification
violation, but it works.

Original idea by Hans Petter Selasky @ FreeBSD.org.

dwc_otg: FIQ support on SMP. Set up FIQ stack and handler on Core 0 only.

dwc_otg: introduce fiq_fsm_spin(un|)lock()

SMP safety for the FIQ relies on register read-modify write cycles being
completed in the correct order. Several places in the DWC code modify
registers also touched by the FIQ. Protect these by a bare-bones lock
mechanism.

This also makes it possible to run the FIQ and IRQ handlers on different
cores.

fiq_fsm: fix build on bcm2708 and bcm2709 platforms

dwc_otg: put some barriers back where they should be for UP

bcm2709/dwc_otg: Setup FIQ on core 1 if >1 core active

dwc_otg: fixup read-modify-write in critical paths

Be more careful about read-modify-write on registers that the FIQ
also touches.

Guard fiq_fsm_spin_lock with fiq_enable check

fiq_fsm: Falling out of the state machine isn't fatal

This edge case can be hit if the port is disabled while the FIQ is
in the middle of a transaction. Make the effects less severe.

Also get rid of the useless return value.

squash: dwc_otg: Allow to build without SMP

usb: core: make overcurrent messages more prominent

Hub overcurrent messages are more serious than "debug". Increase loglevel.

usb: dwc_otg: Don't use dma_to_virt()

Commit 6ce0d20 changes dma_to_virt() which breaks this driver.
Open code the old dma_to_virt() implementation to work around this.

Limit the use of __bus_to_virt() to cases where transfer_buffer_length
is set and transfer_buffer is not set. This is done to increase the
chance that this driver will also work on ARCH_BCM2835.

transfer_buffer should not be NULL if the length is set, but the
comment in the code indicates that there are situations where this
might happen. drivers/usb/isp1760/isp1760-hcd.c also has a similar
comment pointing to a possible: 'usb storage / SCSI bug'.

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>

dwc_otg: Fix crash when fiq_enable=0

dwc_otg: fiq_fsm: Make high-speed isochronous strided transfers work properly

Certain low-bandwidth high-speed USB devices (specialist audio devices,
compressed-frame webcams) have packet intervals > 1 microframe.

Stride these transfers in the FIQ by using the start-of-frame interrupt
to restart the channel at the right time.

dwc_otg: Force host mode to fix incorrect compute module boards

dwc_otg: Add ARCH_BCM2835 support

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>

dwc_otg: Simplify FIQ irq number code

Dropping ATAGS means we can simplify the FIQ irq number code.
Also add error checking on the returned irq number.

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>

dwc_otg: Remove duplicate gadget probe/unregister function

dwc_otg: Properly set the HFIR

Douglas Anderson reported:

According to the most up to date version of the dwc2 databook, the FRINT
field of the HFIR register should be programmed to:
* 125 us * (PHY clock freq for HS) - 1
* 1000 us * (PHY clock freq for FS/LS) - 1

This is opposed to older versions of the doc that claimed it should be:
* 125 us * (PHY clock freq for HS)
* 1000 us * (PHY clock freq for FS/LS)

and reported lower timing jitter on a USB analyser

dcw_otg: trim xfer length when buffer larger than allocated size is received

dwc_otg: Don't free qh align buffers in atomic context

dwc_otg: Enable the hack for Split Interrupt transactions by default

dwc_otg.fiq_fsm_mask=0xF has long been a suggestion for users with audio stutters or other USB bandwidth issues.
So far we are aware of many success stories but no failure caused by this setting.
Make it a default to learn more.

See: https://www.raspberrypi.org/forums/viewtopic.php?f=28&t=70437

Signed-off-by: popcornmix <popcornmix@gmail.com>

dwc_otg: Use kzalloc when suitable

dwc_otg: Pass struct device to dma_alloc*()

This makes it possible to get the bus address from Device Tree.

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>

dwc_otg: fix summarize urb->actual_length for isochronous transfers

Kernel does not copy input data of ISO transfers to userspace
if actual_length is set only in ISO transfers and not summarized
in urb->actual_length. Fixes #903

fiq_fsm: Use correct states when starting isoc OUT transfers

In fiq_fsm_start_next_periodic() if an isochronous OUT transfer
was selected, no regard was given as to whether this was a single-packet
transfer or a multi-packet staged transfer.

For single-packet transfers, this had the effect of repeatedly sending
OUT packets with bogus data and lengths.

Eventually if the channel was repeatedly enabled enough times, this
would lock up the OTG core and no further bus transfers would happen.

Set the FSM state up properly if we select a single-packet transfer.

Fixes #1842

dwc_otg: make nak_holdoff work as intended with empty queues

If URBs reading from non-periodic split endpoints were dequeued and
the last transfer from the endpoint was a NAK handshake, the resulting
qh->nak_frame value was stale which would result in unnecessarily long
polling intervals for the first subsequent transfer with a fresh URB.

Fixup qh->nak_frame in dwc_otg_hcd_urb_dequeue and also guard against
a case where a single URB is submitted to the endpoint, a NAK was
received on the transfer immediately prior to receiving data and the
device subsequently resubmits another URB past the qh->nak_frame interval.

Fixes #1709

dwc_otg: fix split transaction data toggle handling around dequeues

See #1709

Fix several issues regarding endpoint state when URBs are dequeued
- If the HCD is disconnected, flush FIQ-enabled channels properly
- Save the data toggle state for bulk endpoints if the last transfer
  from an endpoint where URBs were dequeued returned a data packet
- Reset hc->start_pkt_count properly in assign_and_init_hc()

dwc_otg: fix several potential crash sources

On root port disconnect events, the host driver state is cleared and
in-progress host channels are forcibly stopped. This doesn't play
well with the FIQ running in the background, so:
- Guard the disconnect callback with both the host spinlock and FIQ
  spinlock
- Move qtd dereference in dwc_otg_handle_hc_fsm() after the early-out
  so we don't dereference a qtd that has gone away
- Turn catch-all BUG()s in dwc_otg_handle_hc_fsm() into warnings.

dwc_otg: delete hcd->channel_lock

The lock serves no purpose as it is only held while the HCD spinlock
is already being held.

dwc_otg: remove unnecessary dma-mode channel halts on disconnect interrupt

Host channels are already halted in kill_urbs_in_qh_list() with the
subsequent interrupt processing behaving as if the URB was dequeued
via HCD callback.

There's no need to clobber the host channel registers a second time
as this exposes races between the driver and host channel resulting
in hcd->free_hc_list becoming corrupted.

dwcotg: Allow to build without FIQ on ARM64

Signed-off-by: popcornmix <popcornmix@gmail.com>

dwc_otg: make periodic scheduling behave properly for FS buses

If the root port is in full-speed mode, transfer times at 12mbit/s
would be calculated but matched against high-speed quotas.

Reinitialise hcd->frame_usecs[i] on each port enable event so that
full-speed bandwidth can be tracked sensibly.

Also, don't bother using the FIQ for transfers when in full-speed
mode - at the slower bus speed, interrupt frequency is reduced by
an order of magnitude.

Related issue: #2020

dwc_otg: fiq_fsm: Make isochronous compatibility checks work properly

Get rid of the spammy printk and local pointer mangling.
Also, there is a nominal benefit for using fiq_fsm for isochronous
transfers in FS mode (~1.1k IRQs per second vs 2.1k IRQs per second)
so remove the root port speed check.

dwc_otg: add module parameter int_ep_interval_min

Add a module parameter (defaulting to ignored) that clamps the polling rate
of high-speed Interrupt endpoints to a minimum microframe interval.

The parameter is modifiable at runtime as it is used when activating new
endpoints (such as on device connect).

dwc_otg: fiq_fsm: Add non-periodic TT exclusivity constraints

Certain hub types do not discriminate between pipe direction (IN or OUT)
when considering non-periodic transfers. Therefore these hubs get confused
if multiple transfers are issued in different directions with the same
device address and endpoint number.

Constrain queuing non-periodic split transactions so they are performed
serially in such cases.

Related: #2024

dwc_otg: Fixup change to DRIVER_ATTR interface

dwc_otg: Fix compilation warnings

Signed-off-by: Phil Elwell <phil@raspberrypi.org>

USB_DWCOTG: Disable building dwc_otg as a module (#2265)

When dwc_otg is built as a module, build will fail with the following
error:

ERROR: "DWC_TASK_HI_SCHEDULE" [drivers/usb/host/dwc_otg/dwc_otg.ko] undefined!
scripts/Makefile.modpost:91: recipe for target '__modpost' failed
make[1]: *** [__modpost] Error 1
Makefile:1199: recipe for target 'modules' failed
make: *** [modules] Error 2

Even if the error is solved by including the missing
DWC_TASK_HI_SCHEDULE function, the kernel will panic when loading
dwc_otg.

As a workaround, simply prevent user from building dwc_otg as a module
as the current kernel does not support it.

See: #2258

Signed-off-by: Malik Olivier Boussejra <malik@boussejra.com>

dwc_otg: New timer API

dwc_otg: Fix removed ACCESS_ONCE->READ_ONCE

dwc_otg: don't unconditionally force host mode in dwc_otg_cil_init()

Add the ability to disable force_host_mode for those that want to use
dwc_otg in both device and host modes.

dwc_otg: Fix a regression when dequeueing isochronous transfers

In 282bed9 (dwc_otg: make nak_holdoff work as intended with empty queues)
the dequeue mechanism was changed to leave FIQ-enabled transfers to run
to completion - to avoid leaving hub TT buffers with stale packets lying
around.

This broke FIQ-accelerated isochronous transfers, as this then meant that
dozens of transfers were performed after the dequeue function returned.

Restore the state machine fence for isochronous transfers.

fiq_fsm: rewind DMA pointer for OUT transactions that fail (#2288)

See: #2140

dwc_otg: add smp_mb() to prevent driver state corruption on boot

Occasional crashes have been seen where the FIQ code dereferences
invalid/random pointers immediately after being set up, leading to
panic on boot.

The crash occurs as the FIQ code races against hcd_init_fiq() and
the hcd_init_fiq() code races against the outstanding memory stores
from dwc_otg_hcd_init(). Use explicit barriers after touching
driver state.

usb: dwc_otg: fix memory corruption in dwc_otg driver

[Upstream commit 51b1b64]

The move from the staging tree to the main tree exposed a
longstanding memory corruption bug in the dwc2 driver. The
reordering of the driver initialization caused the dwc2 driver
to corrupt the initialization data of the sdhci driver on the
Raspberry Pi platform, which made the bug show up.

The error is in calling to_usb_device(hsotg->dev), since ->dev
is not a member of struct usb_device. The easiest fix is to
just remove the offending code, since it is not really needed.

Thanks to Stephen Warren for tracking down the cause of this.

Reported-by: Andre Heider <a.heider@gmail.com>
Tested-by: Stephen Warren <swarren@wwwdotorg.org>
Signed-off-by: Paul Zimmerman <paulz@synopsys.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[lukas: port from upstream dwc2 to out-of-tree dwc_otg driver]
Signed-off-by: Lukas Wunner <lukas@wunner.de>

usb: dwb_otg: Fix unreachable switch statement warning

This warning appears with GCC 7.3.0 from toolchains.bootlin.com:

../drivers/usb/host/dwc_otg/dwc_otg_fiq_fsm.c: In function ‘fiq_fsm_update_hs_isoc’:
../drivers/usb/host/dwc_otg/dwc_otg_fiq_fsm.c:595:61: warning: statement will never be executed [-Wswitch-unreachable]
   st->hctsiz_copy.b.xfersize = nrpackets * st->hcchar_copy.b.mps;
                                            ~~~~~~~~~~~~~~~~~^~~~

Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>

dwc_otg: fiq_fsm: fix incorrect DMA register offset calculation

Rationalise the offset and update all call sites.

Fixes #2408

popcornmix added a commit that referenced this issue Feb 12, 2019

Add dwc_otg driver
Signed-off-by: popcornmix <popcornmix@gmail.com>

usb: dwc: fix lockdep false positive

Signed-off-by: Kari Suvanto <karis79@gmail.com>

usb: dwc: fix inconsistent lock state

Signed-off-by: Kari Suvanto <karis79@gmail.com>

Add FIQ patch to dwc_otg driver. Enable with dwc_otg.fiq_fix_enable=1. Should give about 10% more ARM performance.
Thanks to Gordon and Costas

Avoid dynamic memory allocation for channel lock in USB driver. Thanks ddv2005.

Add NAK holdoff scheme. Enabled by default, disable with dwc_otg.nak_holdoff_enable=0. Thanks gsh

Make sure we wait for the reset to finish

dwc_otg: fix bug in dwc_otg_hcd.c resulting in silent kernel
	 memory corruption, escalating to OOPS under high USB load.

dwc_otg: Fix unsafe access of QTD during URB enqueue

In dwc_otg_hcd_urb_enqueue during qtd creation, it was possible that the
transaction could complete almost immediately after the qtd was assigned
to a host channel during URB enqueue, which meant the qtd pointer was no
longer valid having been completed and removed. Usually, this resulted in
an OOPS during URB submission. By predetermining whether transactions
need to be queued or not, this unsafe pointer access is avoided.

This bug was only evident on the Pi model A where a device was attached
that had no periodic endpoints (e.g. USB pendrive or some wlan devices).

dwc_otg: Fix incorrect URB allocation error handling

If the memory allocation for a dwc_otg_urb failed, the kernel would OOPS
because for some reason a member of the *unallocated* struct was set to
zero. Error handling changed to fail correctly.

dwc_otg: fix potential use-after-free case in interrupt handler

If a transaction had previously aborted, certain interrupts are
enabled to track error counts and reset where necessary. On IN
endpoints the host generates an ACK interrupt near-simultaneously
with completion of transfer. In the case where this transfer had
previously had an error, this results in a use-after-free on
the QTD memory space with a 1-byte length being overwritten to
0x00.

dwc_otg: add handling of SPLIT transaction data toggle errors

Previously a data toggle error on packets from a USB1.1 device behind
a TT would result in the Pi locking up as the driver never handled
the associated interrupt. Patch adds basic retry mechanism and
interrupt acknowledgement to cater for either a chance toggle error or
for devices that have a broken initial toggle state (FT8U232/FT232BM).

dwc_otg: implement tasklet for returning URBs to usbcore hcd layer

The dwc_otg driver interrupt handler for transfer completion will spend
a very long time with interrupts disabled when a URB is completed -
this is because usb_hcd_giveback_urb is called from within the handler
which for a USB device driver with complicated processing (e.g. webcam)
will take an exorbitant amount of time to complete. This results in
missed completion interrupts for other USB packets which lead to them
being dropped due to microframe overruns.

This patch splits returning the URB to the usb hcd layer into a
high-priority tasklet. This will have most benefit for isochronous IN
transfers but will also have incidental benefit where multiple periodic
devices are active at once.

dwc_otg: fix NAK holdoff and allow on split transactions only

This corrects a bug where if a single active non-periodic endpoint
had at least one transaction in its qh, on frnum == MAX_FRNUM the qh
would get skipped and never get queued again. This would result in
a silent device until error detection (automatic or otherwise) would
either reset the device or flush and requeue the URBs.

Additionally the NAK holdoff was enabled for all transactions - this
would potentially stall a HS endpoint for 1ms if a previous error state
enabled this interrupt and the next response was a NAK. Fix so that
only split transactions get held off.

dwc_otg: Call usb_hcd_unlink_urb_from_ep with lock held in completion handler

usb_hcd_unlink_urb_from_ep must be called with the HCD lock held.  Calling it
asynchronously in the tasklet was not safe (regression in
c4564d4).

This change unlinks it from the endpoint prior to queueing it for handling in
the tasklet, and also adds a check to ensure the urb is OK to be unlinked
before doing so.

NULL pointer dereference kernel oopses had been observed in usb_hcd_giveback_urb
when a USB device was unplugged/replugged during data transfer.  This effect
was reproduced using automated USB port power control, hundreds of replug
events were performed during active transfers to confirm that the problem was
eliminated.

USB fix using a FIQ to implement split transactions

This commit adds a FIQ implementaion that schedules
the split transactions using a FIQ so we don't get
held off by the interrupt latency of Linux

dwc_otg: fix device attributes and avoid kernel warnings on boot

dcw_otg: avoid logging function that can cause panics

See: raspberrypi/firmware#21
Thanks to cleverca22 for fix

dwc_otg: mask correct interrupts after transaction error recovery

The dwc_otg driver will unmask certain interrupts on a transaction
that previously halted in the error state in order to reset the
QTD error count. The various fine-grained interrupt handlers do not
consider that other interrupts besides themselves were unmasked.

By disabling the two other interrupts only ever enabled in DMA mode
for this purpose, we can avoid unnecessary function calls in the
IRQ handler. This will also prevent an unneccesary FIQ interrupt
from being generated if the FIQ is enabled.

dwc_otg: fiq: prevent FIQ thrash and incorrect state passing to IRQ

In the case of a transaction to a device that had previously aborted
due to an error, several interrupts are enabled to reset the error
count when a device responds. This has the side-effect of making the
FIQ thrash because the hardware will generate multiple instances of
a NAK on an IN bulk/interrupt endpoint and multiple instances of ACK
on an OUT bulk/interrupt endpoint. Make the FIQ mask and clear the
associated interrupts.

Additionally, on non-split transactions make sure that only unmasked
interrupts are cleared. This caused a hard-to-trigger but serious
race condition when you had the combination of an endpoint awaiting
error recovery and a transaction completed on an endpoint - due to
the sequencing and timing of interrupts generated by the dwc_otg core,
it was possible to confuse the IRQ handler.

Fix function tracing

dwc_otg: whitespace cleanup in dwc_otg_urb_enqueue

dwc_otg: prevent OOPSes during device disconnects

The dwc_otg_urb_enqueue function is thread-unsafe. In particular the
access of urb->hcpriv, usb_hcd_link_urb_to_ep, dwc_otg_urb->qtd and
friends does not occur within a critical section and so if a device
was unplugged during activity there was a high chance that the
usbcore hub_thread would try to disable the endpoint with partially-
formed entries in the URB queue. This would result in BUG() or null
pointer dereferences.

Fix so that access of urb->hcpriv, enqueuing to the hardware and
adding to usbcore endpoint URB lists is contained within a single
critical section.

dwc_otg: prevent BUG() in TT allocation if hub address is > 16

A fixed-size array is used to track TT allocation. This was
previously set to 16 which caused a crash because
dwc_otg_hcd_allocate_port would read past the end of the array.

This was hit if a hub was plugged in which enumerated as addr > 16,
due to previous device resets or unplugs.

Also add #ifdef FIQ_DEBUG around hcd->hub_port_alloc[], which grows
to a large size if 128 hub addresses are supported. This field is
for debug only for tracking which frame an allocate happened in.

dwc_otg: make channel halts with unknown state less damaging

If the IRQ received a channel halt interrupt through the FIQ
with no other bits set, the IRQ would not release the host
channel and never complete the URB.

Add catchall handling to treat as a transaction error and retry.

dwc_otg: fiq_split: use TTs with more granularity

This fixes certain issues with split transaction scheduling.

- Isochronous multi-packet OUT transactions now hog the TT until
  they are completed - this prevents hubs aborting transactions
  if they get a periodic start-split out-of-order
- Don't perform TT allocation on non-periodic endpoints - this
  allows simultaneous use of the TT's bulk/control and periodic
  transaction buffers

This commit will mainly affect USB audio playback.

dwc_otg: fix potential sleep while atomic during urb enqueue

Fixes a regression introduced with eb1b482. Kmalloc called from
dwc_otg_hcd_qtd_add / dwc_otg_hcd_qtd_create did not always have
the GPF_ATOMIC flag set. Force this flag when inside the larger
critical section.

dwc_otg: make fiq_split_enable imply fiq_fix_enable

Failing to set up the FIQ correctly would result in
"IRQ 32: nobody cared" errors in dmesg.

dwc_otg: prevent crashes on host port disconnects

Fix several issues resulting in crashes or inconsistent state
if a Model A root port was disconnected.

- Clean up queue heads properly in kill_urbs_in_qh_list by
  removing the empty QHs from the schedule lists
- Set the halt status properly to prevent IRQ handlers from
  using freed memory
- Add fiq_split related cleanup for saved registers
- Make microframe scheduling reclaim host channels if
  active during a disconnect
- Abort URBs with -ESHUTDOWN status response, informing
  device drivers so they respond in a more correct fashion
  and don't try to resubmit URBs
- Prevent IRQ handlers from attempting to handle channel
  interrupts if the associated URB was dequeued (and the
  driver state was cleared)

dwc_otg: prevent leaking URBs during enqueue

A dwc_otg_urb would get leaked if the HCD enqueue function
failed for any reason. Free the URB at the appropriate points.

dwc_otg: Enable NAK holdoff for control split transactions

Certain low-speed devices take a very long time to complete a
data or status stage of a control transaction, producing NAK
responses until they complete internal processing - the USB2.0
spec limit is up to 500mS. This causes the same type of interrupt
storm as seen with USB-serial dongles prior to c8edb23.

In certain circumstances, usually while booting, this interrupt
storm could cause SD card timeouts.

dwc_otg: Fix for occasional lockup on boot when doing a USB reset

dwc_otg: Don't issue traffic to LS devices in FS mode

Issuing low-speed packets when the root port is in full-speed mode
causes the root port to stop responding. Explicitly fail when
enqueuing URBs to a LS endpoint on a FS bus.

Fix ARM architecture issue with local_irq_restore()

If local_fiq_enable() is called before a local_irq_restore(flags) where
the flags variable has the F bit set, the FIQ will be erroneously disabled.

Fixup arch_local_irq_restore to avoid trampling the F bit in CPSR.

Also fix some of the hacks previously implemented for previous dwc_otg
incarnations.

dwc_otg: fiq_fsm: Base commit for driver rewrite

This commit removes the previous FIQ fixes entirely and adds fiq_fsm.

This rewrite features much more complete support for split transactions
and takes into account several OTG hardware bugs. High-speed
isochronous transactions are also capable of being performed by fiq_fsm.

All driver options have been removed and replaced with:
  - dwc_otg.fiq_enable (bool)
  - dwc_otg.fiq_fsm_enable (bool)
  - dwc_otg.fiq_fsm_mask (bitmask)
  - dwc_otg.nak_holdoff (unsigned int)

Defaults are specified such that fiq_fsm behaves similarly to the
previously implemented FIQ fixes.

fiq_fsm: Push error recovery into the FIQ when fiq_fsm is used

If the transfer associated with a QTD failed due to a bus error, the HCD
would retry the transfer up to 3 times (implementing the USB2.0
three-strikes retry in software).

Due to the masking mechanism used by fiq_fsm, it is only possible to pass
a single interrupt through to the HCD per-transfer.

In this instance host channels would fall off the radar because the error
reset would function, but the subsequent channel halt would be lost.

Push the error count reset into the FIQ handler.

fiq_fsm: Implement timeout mechanism

For full-speed endpoints with a large packet size, interrupt latency
runs the risk of the FIQ starting a transaction too late in a full-speed
frame. If the device is still transmitting data when EOF2 for the
downstream frame occurs, the hub will disable the port. This change is
not reflected in the hub status endpoint and the device becomes
unresponsive.

Prevent high-bandwidth transactions from being started too late in a
frame. The mechanism is not guaranteed: a combination of bit stuffing
and hub latency may still result in a device overrunning.

fiq_fsm: fix bounce buffer utilisation for Isochronous OUT

Multi-packet isochronous OUT transactions were subject to a few bounday
bugs. Fix them.

Audio playback is now much more robust: however, an issue stands with
devices that have adaptive sinks - ALSA plays samples too fast.

dwc_otg: Return full-speed frame numbers in HS mode

The frame counter increments on every *microframe* in high-speed mode.
Most device drivers expect this number to be in full-speed frames - this
caused considerable confusion to e.g. snd_usb_audio which uses the
frame counter to estimate the number of samples played.

fiq_fsm: save PID on completion of interrupt OUT transfers

Also add edge case handling for interrupt transports.

Note that for periodic split IN, data toggles are unimplemented in the
OTG host hardware - it unconditionally accepts any PID.

fiq_fsm: add missing case for fiq_fsm_tt_in_use()

Certain combinations of bitrate and endpoint activity could
result in a periodic transaction erroneously getting started
while the previous Isochronous OUT was still active.

fiq_fsm: clear hcintmsk for aborted transactions

Prevents the FIQ from erroneously handling interrupts
on a timed out channel.

fiq_fsm: enable by default

fiq_fsm: fix dequeues for non-periodic split transactions

If a dequeue happened between the SSPLIT and CSPLIT phases of the
transaction, the HCD would never receive an interrupt.

fiq_fsm: Disable by default

fiq_fsm: Handle HC babble errors

The HCTSIZ transfer size field raises a babble interrupt if
the counter wraps. Handle the resulting interrupt in this case.

dwc_otg: fix interrupt registration for fiq_enable=0

Additionally make the module parameter conditional for wherever
hcd->fiq_state is touched.

fiq_fsm: Enable by default

dwc_otg: Fix various issues with root port and transaction errors

Process the host port interrupts correctly (and don't trample them).
Root port hotplug now functional again.

Fix a few thinkos with the transaction error passthrough for fiq_fsm.

fiq_fsm: Implement hack for Split Interrupt transactions

Hubs aren't too picky about which endpoint we send Control type split
transactions to. By treating Interrupt transfers as Control, it is
possible to use the non-periodic queue in the OTG core as well as the
non-periodic FIFOs in the hub itself. This massively reduces the
microframe exclusivity/contention that periodic split transactions
otherwise have to enforce.

It goes without saying that this is a fairly egregious USB specification
violation, but it works.

Original idea by Hans Petter Selasky @ FreeBSD.org.

dwc_otg: FIQ support on SMP. Set up FIQ stack and handler on Core 0 only.

dwc_otg: introduce fiq_fsm_spin(un|)lock()

SMP safety for the FIQ relies on register read-modify write cycles being
completed in the correct order. Several places in the DWC code modify
registers also touched by the FIQ. Protect these by a bare-bones lock
mechanism.

This also makes it possible to run the FIQ and IRQ handlers on different
cores.

fiq_fsm: fix build on bcm2708 and bcm2709 platforms

dwc_otg: put some barriers back where they should be for UP

bcm2709/dwc_otg: Setup FIQ on core 1 if >1 core active

dwc_otg: fixup read-modify-write in critical paths

Be more careful about read-modify-write on registers that the FIQ
also touches.

Guard fiq_fsm_spin_lock with fiq_enable check

fiq_fsm: Falling out of the state machine isn't fatal

This edge case can be hit if the port is disabled while the FIQ is
in the middle of a transaction. Make the effects less severe.

Also get rid of the useless return value.

squash: dwc_otg: Allow to build without SMP

usb: core: make overcurrent messages more prominent

Hub overcurrent messages are more serious than "debug". Increase loglevel.

usb: dwc_otg: Don't use dma_to_virt()

Commit 6ce0d20 changes dma_to_virt() which breaks this driver.
Open code the old dma_to_virt() implementation to work around this.

Limit the use of __bus_to_virt() to cases where transfer_buffer_length
is set and transfer_buffer is not set. This is done to increase the
chance that this driver will also work on ARCH_BCM2835.

transfer_buffer should not be NULL if the length is set, but the
comment in the code indicates that there are situations where this
might happen. drivers/usb/isp1760/isp1760-hcd.c also has a similar
comment pointing to a possible: 'usb storage / SCSI bug'.

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>

dwc_otg: Fix crash when fiq_enable=0

dwc_otg: fiq_fsm: Make high-speed isochronous strided transfers work properly

Certain low-bandwidth high-speed USB devices (specialist audio devices,
compressed-frame webcams) have packet intervals > 1 microframe.

Stride these transfers in the FIQ by using the start-of-frame interrupt
to restart the channel at the right time.

dwc_otg: Force host mode to fix incorrect compute module boards

dwc_otg: Add ARCH_BCM2835 support

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>

dwc_otg: Simplify FIQ irq number code

Dropping ATAGS means we can simplify the FIQ irq number code.
Also add error checking on the returned irq number.

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>

dwc_otg: Remove duplicate gadget probe/unregister function

dwc_otg: Properly set the HFIR

Douglas Anderson reported:

According to the most up to date version of the dwc2 databook, the FRINT
field of the HFIR register should be programmed to:
* 125 us * (PHY clock freq for HS) - 1
* 1000 us * (PHY clock freq for FS/LS) - 1

This is opposed to older versions of the doc that claimed it should be:
* 125 us * (PHY clock freq for HS)
* 1000 us * (PHY clock freq for FS/LS)

and reported lower timing jitter on a USB analyser

dcw_otg: trim xfer length when buffer larger than allocated size is received

dwc_otg: Don't free qh align buffers in atomic context

dwc_otg: Enable the hack for Split Interrupt transactions by default

dwc_otg.fiq_fsm_mask=0xF has long been a suggestion for users with audio stutters or other USB bandwidth issues.
So far we are aware of many success stories but no failure caused by this setting.
Make it a default to learn more.

See: https://www.raspberrypi.org/forums/viewtopic.php?f=28&t=70437

Signed-off-by: popcornmix <popcornmix@gmail.com>

dwc_otg: Use kzalloc when suitable

dwc_otg: Pass struct device to dma_alloc*()

This makes it possible to get the bus address from Device Tree.

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>

dwc_otg: fix summarize urb->actual_length for isochronous transfers

Kernel does not copy input data of ISO transfers to userspace
if actual_length is set only in ISO transfers and not summarized
in urb->actual_length. Fixes #903

fiq_fsm: Use correct states when starting isoc OUT transfers

In fiq_fsm_start_next_periodic() if an isochronous OUT transfer
was selected, no regard was given as to whether this was a single-packet
transfer or a multi-packet staged transfer.

For single-packet transfers, this had the effect of repeatedly sending
OUT packets with bogus data and lengths.

Eventually if the channel was repeatedly enabled enough times, this
would lock up the OTG core and no further bus transfers would happen.

Set the FSM state up properly if we select a single-packet transfer.

Fixes #1842

dwc_otg: make nak_holdoff work as intended with empty queues

If URBs reading from non-periodic split endpoints were dequeued and
the last transfer from the endpoint was a NAK handshake, the resulting
qh->nak_frame value was stale which would result in unnecessarily long
polling intervals for the first subsequent transfer with a fresh URB.

Fixup qh->nak_frame in dwc_otg_hcd_urb_dequeue and also guard against
a case where a single URB is submitted to the endpoint, a NAK was
received on the transfer immediately prior to receiving data and the
device subsequently resubmits another URB past the qh->nak_frame interval.

Fixes #1709

dwc_otg: fix split transaction data toggle handling around dequeues

See #1709

Fix several issues regarding endpoint state when URBs are dequeued
- If the HCD is disconnected, flush FIQ-enabled channels properly
- Save the data toggle state for bulk endpoints if the last transfer
  from an endpoint where URBs were dequeued returned a data packet
- Reset hc->start_pkt_count properly in assign_and_init_hc()

dwc_otg: fix several potential crash sources

On root port disconnect events, the host driver state is cleared and
in-progress host channels are forcibly stopped. This doesn't play
well with the FIQ running in the background, so:
- Guard the disconnect callback with both the host spinlock and FIQ
  spinlock
- Move qtd dereference in dwc_otg_handle_hc_fsm() after the early-out
  so we don't dereference a qtd that has gone away
- Turn catch-all BUG()s in dwc_otg_handle_hc_fsm() into warnings.

dwc_otg: delete hcd->channel_lock

The lock serves no purpose as it is only held while the HCD spinlock
is already being held.

dwc_otg: remove unnecessary dma-mode channel halts on disconnect interrupt

Host channels are already halted in kill_urbs_in_qh_list() with the
subsequent interrupt processing behaving as if the URB was dequeued
via HCD callback.

There's no need to clobber the host channel registers a second time
as this exposes races between the driver and host channel resulting
in hcd->free_hc_list becoming corrupted.

dwcotg: Allow to build without FIQ on ARM64

Signed-off-by: popcornmix <popcornmix@gmail.com>

dwc_otg: make periodic scheduling behave properly for FS buses

If the root port is in full-speed mode, transfer times at 12mbit/s
would be calculated but matched against high-speed quotas.

Reinitialise hcd->frame_usecs[i] on each port enable event so that
full-speed bandwidth can be tracked sensibly.

Also, don't bother using the FIQ for transfers when in full-speed
mode - at the slower bus speed, interrupt frequency is reduced by
an order of magnitude.

Related issue: #2020

dwc_otg: fiq_fsm: Make isochronous compatibility checks work properly

Get rid of the spammy printk and local pointer mangling.
Also, there is a nominal benefit for using fiq_fsm for isochronous
transfers in FS mode (~1.1k IRQs per second vs 2.1k IRQs per second)
so remove the root port speed check.

dwc_otg: add module parameter int_ep_interval_min

Add a module parameter (defaulting to ignored) that clamps the polling rate
of high-speed Interrupt endpoints to a minimum microframe interval.

The parameter is modifiable at runtime as it is used when activating new
endpoints (such as on device connect).

dwc_otg: fiq_fsm: Add non-periodic TT exclusivity constraints

Certain hub types do not discriminate between pipe direction (IN or OUT)
when considering non-periodic transfers. Therefore these hubs get confused
if multiple transfers are issued in different directions with the same
device address and endpoint number.

Constrain queuing non-periodic split transactions so they are performed
serially in such cases.

Related: #2024

dwc_otg: Fixup change to DRIVER_ATTR interface

dwc_otg: Fix compilation warnings

Signed-off-by: Phil Elwell <phil@raspberrypi.org>

USB_DWCOTG: Disable building dwc_otg as a module (#2265)

When dwc_otg is built as a module, build will fail with the following
error:

ERROR: "DWC_TASK_HI_SCHEDULE" [drivers/usb/host/dwc_otg/dwc_otg.ko] undefined!
scripts/Makefile.modpost:91: recipe for target '__modpost' failed
make[1]: *** [__modpost] Error 1
Makefile:1199: recipe for target 'modules' failed
make: *** [modules] Error 2

Even if the error is solved by including the missing
DWC_TASK_HI_SCHEDULE function, the kernel will panic when loading
dwc_otg.

As a workaround, simply prevent user from building dwc_otg as a module
as the current kernel does not support it.

See: #2258

Signed-off-by: Malik Olivier Boussejra <malik@boussejra.com>

dwc_otg: New timer API

dwc_otg: Fix removed ACCESS_ONCE->READ_ONCE

dwc_otg: don't unconditionally force host mode in dwc_otg_cil_init()

Add the ability to disable force_host_mode for those that want to use
dwc_otg in both device and host modes.

dwc_otg: Fix a regression when dequeueing isochronous transfers

In 282bed9 (dwc_otg: make nak_holdoff work as intended with empty queues)
the dequeue mechanism was changed to leave FIQ-enabled transfers to run
to completion - to avoid leaving hub TT buffers with stale packets lying
around.

This broke FIQ-accelerated isochronous transfers, as this then meant that
dozens of transfers were performed after the dequeue function returned.

Restore the state machine fence for isochronous transfers.

fiq_fsm: rewind DMA pointer for OUT transactions that fail (#2288)

See: #2140

dwc_otg: add smp_mb() to prevent driver state corruption on boot

Occasional crashes have been seen where the FIQ code dereferences
invalid/random pointers immediately after being set up, leading to
panic on boot.

The crash occurs as the FIQ code races against hcd_init_fiq() and
the hcd_init_fiq() code races against the outstanding memory stores
from dwc_otg_hcd_init(). Use explicit barriers after touching
driver state.

usb: dwc_otg: fix memory corruption in dwc_otg driver

[Upstream commit 51b1b64]

The move from the staging tree to the main tree exposed a
longstanding memory corruption bug in the dwc2 driver. The
reordering of the driver initialization caused the dwc2 driver
to corrupt the initialization data of the sdhci driver on the
Raspberry Pi platform, which made the bug show up.

The error is in calling to_usb_device(hsotg->dev), since ->dev
is not a member of struct usb_device. The easiest fix is to
just remove the offending code, since it is not really needed.

Thanks to Stephen Warren for tracking down the cause of this.

Reported-by: Andre Heider <a.heider@gmail.com>
Tested-by: Stephen Warren <swarren@wwwdotorg.org>
Signed-off-by: Paul Zimmerman <paulz@synopsys.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[lukas: port from upstream dwc2 to out-of-tree dwc_otg driver]
Signed-off-by: Lukas Wunner <lukas@wunner.de>

usb: dwb_otg: Fix unreachable switch statement warning

This warning appears with GCC 7.3.0 from toolchains.bootlin.com:

../drivers/usb/host/dwc_otg/dwc_otg_fiq_fsm.c: In function ‘fiq_fsm_update_hs_isoc’:
../drivers/usb/host/dwc_otg/dwc_otg_fiq_fsm.c:595:61: warning: statement will never be executed [-Wswitch-unreachable]
   st->hctsiz_copy.b.xfersize = nrpackets * st->hcchar_copy.b.mps;
                                            ~~~~~~~~~~~~~~~~~^~~~

Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>

dwc_otg: fiq_fsm: fix incorrect DMA register offset calculation

Rationalise the offset and update all call sites.

Fixes #2408

popcornmix added a commit that referenced this issue Feb 18, 2019

Add dwc_otg driver
Signed-off-by: popcornmix <popcornmix@gmail.com>

usb: dwc: fix lockdep false positive

Signed-off-by: Kari Suvanto <karis79@gmail.com>

usb: dwc: fix inconsistent lock state

Signed-off-by: Kari Suvanto <karis79@gmail.com>

Add FIQ patch to dwc_otg driver. Enable with dwc_otg.fiq_fix_enable=1. Should give about 10% more ARM performance.
Thanks to Gordon and Costas

Avoid dynamic memory allocation for channel lock in USB driver. Thanks ddv2005.

Add NAK holdoff scheme. Enabled by default, disable with dwc_otg.nak_holdoff_enable=0. Thanks gsh

Make sure we wait for the reset to finish

dwc_otg: fix bug in dwc_otg_hcd.c resulting in silent kernel
	 memory corruption, escalating to OOPS under high USB load.

dwc_otg: Fix unsafe access of QTD during URB enqueue

In dwc_otg_hcd_urb_enqueue during qtd creation, it was possible that the
transaction could complete almost immediately after the qtd was assigned
to a host channel during URB enqueue, which meant the qtd pointer was no
longer valid having been completed and removed. Usually, this resulted in
an OOPS during URB submission. By predetermining whether transactions
need to be queued or not, this unsafe pointer access is avoided.

This bug was only evident on the Pi model A where a device was attached
that had no periodic endpoints (e.g. USB pendrive or some wlan devices).

dwc_otg: Fix incorrect URB allocation error handling

If the memory allocation for a dwc_otg_urb failed, the kernel would OOPS
because for some reason a member of the *unallocated* struct was set to
zero. Error handling changed to fail correctly.

dwc_otg: fix potential use-after-free case in interrupt handler

If a transaction had previously aborted, certain interrupts are
enabled to track error counts and reset where necessary. On IN
endpoints the host generates an ACK interrupt near-simultaneously
with completion of transfer. In the case where this transfer had
previously had an error, this results in a use-after-free on
the QTD memory space with a 1-byte length being overwritten to
0x00.

dwc_otg: add handling of SPLIT transaction data toggle errors

Previously a data toggle error on packets from a USB1.1 device behind
a TT would result in the Pi locking up as the driver never handled
the associated interrupt. Patch adds basic retry mechanism and
interrupt acknowledgement to cater for either a chance toggle error or
for devices that have a broken initial toggle state (FT8U232/FT232BM).

dwc_otg: implement tasklet for returning URBs to usbcore hcd layer

The dwc_otg driver interrupt handler for transfer completion will spend
a very long time with interrupts disabled when a URB is completed -
this is because usb_hcd_giveback_urb is called from within the handler
which for a USB device driver with complicated processing (e.g. webcam)
will take an exorbitant amount of time to complete. This results in
missed completion interrupts for other USB packets which lead to them
being dropped due to microframe overruns.

This patch splits returning the URB to the usb hcd layer into a
high-priority tasklet. This will have most benefit for isochronous IN
transfers but will also have incidental benefit where multiple periodic
devices are active at once.

dwc_otg: fix NAK holdoff and allow on split transactions only

This corrects a bug where if a single active non-periodic endpoint
had at least one transaction in its qh, on frnum == MAX_FRNUM the qh
would get skipped and never get queued again. This would result in
a silent device until error detection (automatic or otherwise) would
either reset the device or flush and requeue the URBs.

Additionally the NAK holdoff was enabled for all transactions - this
would potentially stall a HS endpoint for 1ms if a previous error state
enabled this interrupt and the next response was a NAK. Fix so that
only split transactions get held off.

dwc_otg: Call usb_hcd_unlink_urb_from_ep with lock held in completion handler

usb_hcd_unlink_urb_from_ep must be called with the HCD lock held.  Calling it
asynchronously in the tasklet was not safe (regression in
c4564d4).

This change unlinks it from the endpoint prior to queueing it for handling in
the tasklet, and also adds a check to ensure the urb is OK to be unlinked
before doing so.

NULL pointer dereference kernel oopses had been observed in usb_hcd_giveback_urb
when a USB device was unplugged/replugged during data transfer.  This effect
was reproduced using automated USB port power control, hundreds of replug
events were performed during active transfers to confirm that the problem was
eliminated.

USB fix using a FIQ to implement split transactions

This commit adds a FIQ implementaion that schedules
the split transactions using a FIQ so we don't get
held off by the interrupt latency of Linux

dwc_otg: fix device attributes and avoid kernel warnings on boot

dcw_otg: avoid logging function that can cause panics

See: raspberrypi/firmware#21
Thanks to cleverca22 for fix

dwc_otg: mask correct interrupts after transaction error recovery

The dwc_otg driver will unmask certain interrupts on a transaction
that previously halted in the error state in order to reset the
QTD error count. The various fine-grained interrupt handlers do not
consider that other interrupts besides themselves were unmasked.

By disabling the two other interrupts only ever enabled in DMA mode
for this purpose, we can avoid unnecessary function calls in the
IRQ handler. This will also prevent an unneccesary FIQ interrupt
from being generated if the FIQ is enabled.

dwc_otg: fiq: prevent FIQ thrash and incorrect state passing to IRQ

In the case of a transaction to a device that had previously aborted
due to an error, several interrupts are enabled to reset the error
count when a device responds. This has the side-effect of making the
FIQ thrash because the hardware will generate multiple instances of
a NAK on an IN bulk/interrupt endpoint and multiple instances of ACK
on an OUT bulk/interrupt endpoint. Make the FIQ mask and clear the
associated interrupts.

Additionally, on non-split transactions make sure that only unmasked
interrupts are cleared. This caused a hard-to-trigger but serious
race condition when you had the combination of an endpoint awaiting
error recovery and a transaction completed on an endpoint - due to
the sequencing and timing of interrupts generated by the dwc_otg core,
it was possible to confuse the IRQ handler.

Fix function tracing

dwc_otg: whitespace cleanup in dwc_otg_urb_enqueue

dwc_otg: prevent OOPSes during device disconnects

The dwc_otg_urb_enqueue function is thread-unsafe. In particular the
access of urb->hcpriv, usb_hcd_link_urb_to_ep, dwc_otg_urb->qtd and
friends does not occur within a critical section and so if a device
was unplugged during activity there was a high chance that the
usbcore hub_thread would try to disable the endpoint with partially-
formed entries in the URB queue. This would result in BUG() or null
pointer dereferences.

Fix so that access of urb->hcpriv, enqueuing to the hardware and
adding to usbcore endpoint URB lists is contained within a single
critical section.

dwc_otg: prevent BUG() in TT allocation if hub address is > 16

A fixed-size array is used to track TT allocation. This was
previously set to 16 which caused a crash because
dwc_otg_hcd_allocate_port would read past the end of the array.

This was hit if a hub was plugged in which enumerated as addr > 16,
due to previous device resets or unplugs.

Also add #ifdef FIQ_DEBUG around hcd->hub_port_alloc[], which grows
to a large size if 128 hub addresses are supported. This field is
for debug only for tracking which frame an allocate happened in.

dwc_otg: make channel halts with unknown state less damaging

If the IRQ received a channel halt interrupt through the FIQ
with no other bits set, the IRQ would not release the host
channel and never complete the URB.

Add catchall handling to treat as a transaction error and retry.

dwc_otg: fiq_split: use TTs with more granularity

This fixes certain issues with split transaction scheduling.

- Isochronous multi-packet OUT transactions now hog the TT until
  they are completed - this prevents hubs aborting transactions
  if they get a periodic start-split out-of-order
- Don't perform TT allocation on non-periodic endpoints - this
  allows simultaneous use of the TT's bulk/control and periodic
  transaction buffers

This commit will mainly affect USB audio playback.

dwc_otg: fix potential sleep while atomic during urb enqueue

Fixes a regression introduced with eb1b482. Kmalloc called from
dwc_otg_hcd_qtd_add / dwc_otg_hcd_qtd_create did not always have
the GPF_ATOMIC flag set. Force this flag when inside the larger
critical section.

dwc_otg: make fiq_split_enable imply fiq_fix_enable

Failing to set up the FIQ correctly would result in
"IRQ 32: nobody cared" errors in dmesg.

dwc_otg: prevent crashes on host port disconnects

Fix several issues resulting in crashes or inconsistent state
if a Model A root port was disconnected.

- Clean up queue heads properly in kill_urbs_in_qh_list by
  removing the empty QHs from the schedule lists
- Set the halt status properly to prevent IRQ handlers from
  using freed memory
- Add fiq_split related cleanup for saved registers
- Make microframe scheduling reclaim host channels if
  active during a disconnect
- Abort URBs with -ESHUTDOWN status response, informing
  device drivers so they respond in a more correct fashion
  and don't try to resubmit URBs
- Prevent IRQ handlers from attempting to handle channel
  interrupts if the associated URB was dequeued (and the
  driver state was cleared)

dwc_otg: prevent leaking URBs during enqueue

A dwc_otg_urb would get leaked if the HCD enqueue function
failed for any reason. Free the URB at the appropriate points.

dwc_otg: Enable NAK holdoff for control split transactions

Certain low-speed devices take a very long time to complete a
data or status stage of a control transaction, producing NAK
responses until they complete internal processing - the USB2.0
spec limit is up to 500mS. This causes the same type of interrupt
storm as seen with USB-serial dongles prior to c8edb23.

In certain circumstances, usually while booting, this interrupt
storm could cause SD card timeouts.

dwc_otg: Fix for occasional lockup on boot when doing a USB reset

dwc_otg: Don't issue traffic to LS devices in FS mode

Issuing low-speed packets when the root port is in full-speed mode
causes the root port to stop responding. Explicitly fail when
enqueuing URBs to a LS endpoint on a FS bus.

Fix ARM architecture issue with local_irq_restore()

If local_fiq_enable() is called before a local_irq_restore(flags) where
the flags variable has the F bit set, the FIQ will be erroneously disabled.

Fixup arch_local_irq_restore to avoid trampling the F bit in CPSR.

Also fix some of the hacks previously implemented for previous dwc_otg
incarnations.

dwc_otg: fiq_fsm: Base commit for driver rewrite

This commit removes the previous FIQ fixes entirely and adds fiq_fsm.

This rewrite features much more complete support for split transactions
and takes into account several OTG hardware bugs. High-speed
isochronous transactions are also capable of being performed by fiq_fsm.

All driver options have been removed and replaced with:
  - dwc_otg.fiq_enable (bool)
  - dwc_otg.fiq_fsm_enable (bool)
  - dwc_otg.fiq_fsm_mask (bitmask)
  - dwc_otg.nak_holdoff (unsigned int)

Defaults are specified such that fiq_fsm behaves similarly to the
previously implemented FIQ fixes.

fiq_fsm: Push error recovery into the FIQ when fiq_fsm is used

If the transfer associated with a QTD failed due to a bus error, the HCD
would retry the transfer up to 3 times (implementing the USB2.0
three-strikes retry in software).

Due to the masking mechanism used by fiq_fsm, it is only possible to pass
a single interrupt through to the HCD per-transfer.

In this instance host channels would fall off the radar because the error
reset would function, but the subsequent channel halt would be lost.

Push the error count reset into the FIQ handler.

fiq_fsm: Implement timeout mechanism

For full-speed endpoints with a large packet size, interrupt latency
runs the risk of the FIQ starting a transaction too late in a full-speed
frame. If the device is still transmitting data when EOF2 for the
downstream frame occurs, the hub will disable the port. This change is
not reflected in the hub status endpoint and the device becomes
unresponsive.

Prevent high-bandwidth transactions from being started too late in a
frame. The mechanism is not guaranteed: a combination of bit stuffing
and hub latency may still result in a device overrunning.

fiq_fsm: fix bounce buffer utilisation for Isochronous OUT

Multi-packet isochronous OUT transactions were subject to a few bounday
bugs. Fix them.

Audio playback is now much more robust: however, an issue stands with
devices that have adaptive sinks - ALSA plays samples too fast.

dwc_otg: Return full-speed frame numbers in HS mode

The frame counter increments on every *microframe* in high-speed mode.
Most device drivers expect this number to be in full-speed frames - this
caused considerable confusion to e.g. snd_usb_audio which uses the
frame counter to estimate the number of samples played.

fiq_fsm: save PID on completion of interrupt OUT transfers

Also add edge case handling for interrupt transports.

Note that for periodic split IN, data toggles are unimplemented in the
OTG host hardware - it unconditionally accepts any PID.

fiq_fsm: add missing case for fiq_fsm_tt_in_use()

Certain combinations of bitrate and endpoint activity could
result in a periodic transaction erroneously getting started
while the previous Isochronous OUT was still active.

fiq_fsm: clear hcintmsk for aborted transactions

Prevents the FIQ from erroneously handling interrupts
on a timed out channel.

fiq_fsm: enable by default

fiq_fsm: fix dequeues for non-periodic split transactions

If a dequeue happened between the SSPLIT and CSPLIT phases of the
transaction, the HCD would never receive an interrupt.

fiq_fsm: Disable by default

fiq_fsm: Handle HC babble errors

The HCTSIZ transfer size field raises a babble interrupt if
the counter wraps. Handle the resulting interrupt in this case.

dwc_otg: fix interrupt registration for fiq_enable=0

Additionally make the module parameter conditional for wherever
hcd->fiq_state is touched.

fiq_fsm: Enable by default

dwc_otg: Fix various issues with root port and transaction errors

Process the host port interrupts correctly (and don't trample them).
Root port hotplug now functional again.

Fix a few thinkos with the transaction error passthrough for fiq_fsm.

fiq_fsm: Implement hack for Split Interrupt transactions

Hubs aren't too picky about which endpoint we send Control type split
transactions to. By treating Interrupt transfers as Control, it is
possible to use the non-periodic queue in the OTG core as well as the
non-periodic FIFOs in the hub itself. This massively reduces the
microframe exclusivity/contention that periodic split transactions
otherwise have to enforce.

It goes without saying that this is a fairly egregious USB specification
violation, but it works.

Original idea by Hans Petter Selasky @ FreeBSD.org.

dwc_otg: FIQ support on SMP. Set up FIQ stack and handler on Core 0 only.

dwc_otg: introduce fiq_fsm_spin(un|)lock()

SMP safety for the FIQ relies on register read-modify write cycles being
completed in the correct order. Several places in the DWC code modify
registers also touched by the FIQ. Protect these by a bare-bones lock
mechanism.

This also makes it possible to run the FIQ and IRQ handlers on different
cores.

fiq_fsm: fix build on bcm2708 and bcm2709 platforms

dwc_otg: put some barriers back where they should be for UP

bcm2709/dwc_otg: Setup FIQ on core 1 if >1 core active

dwc_otg: fixup read-modify-write in critical paths

Be more careful about read-modify-write on registers that the FIQ
also touches.

Guard fiq_fsm_spin_lock with fiq_enable check

fiq_fsm: Falling out of the state machine isn't fatal

This edge case can be hit if the port is disabled while the FIQ is
in the middle of a transaction. Make the effects less severe.

Also get rid of the useless return value.

squash: dwc_otg: Allow to build without SMP

usb: core: make overcurrent messages more prominent

Hub overcurrent messages are more serious than "debug". Increase loglevel.

usb: dwc_otg: Don't use dma_to_virt()

Commit 6ce0d20 changes dma_to_virt() which breaks this driver.
Open code the old dma_to_virt() implementation to work around this.

Limit the use of __bus_to_virt() to cases where transfer_buffer_length
is set and transfer_buffer is not set. This is done to increase the
chance that this driver will also work on ARCH_BCM2835.

transfer_buffer should not be NULL if the length is set, but the
comment in the code indicates that there are situations where this
might happen. drivers/usb/isp1760/isp1760-hcd.c also has a similar
comment pointing to a possible: 'usb storage / SCSI bug'.

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>

dwc_otg: Fix crash when fiq_enable=0

dwc_otg: fiq_fsm: Make high-speed isochronous strided transfers work properly

Certain low-bandwidth high-speed USB devices (specialist audio devices,
compressed-frame webcams) have packet intervals > 1 microframe.

Stride these transfers in the FIQ by using the start-of-frame interrupt
to restart the channel at the right time.

dwc_otg: Force host mode to fix incorrect compute module boards

dwc_otg: Add ARCH_BCM2835 support

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>

dwc_otg: Simplify FIQ irq number code

Dropping ATAGS means we can simplify the FIQ irq number code.
Also add error checking on the returned irq number.

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>

dwc_otg: Remove duplicate gadget probe/unregister function

dwc_otg: Properly set the HFIR

Douglas Anderson reported:

According to the most up to date version of the dwc2 databook, the FRINT
field of the HFIR register should be programmed to:
* 125 us * (PHY clock freq for HS) - 1
* 1000 us * (PHY clock freq for FS/LS) - 1

This is opposed to older versions of the doc that claimed it should be:
* 125 us * (PHY clock freq for HS)
* 1000 us * (PHY clock freq for FS/LS)

and reported lower timing jitter on a USB analyser

dcw_otg: trim xfer length when buffer larger than allocated size is received

dwc_otg: Don't free qh align buffers in atomic context

dwc_otg: Enable the hack for Split Interrupt transactions by default

dwc_otg.fiq_fsm_mask=0xF has long been a suggestion for users with audio stutters or other USB bandwidth issues.
So far we are aware of many success stories but no failure caused by this setting.
Make it a default to learn more.

See: https://www.raspberrypi.org/forums/viewtopic.php?f=28&t=70437

Signed-off-by: popcornmix <popcornmix@gmail.com>

dwc_otg: Use kzalloc when suitable

dwc_otg: Pass struct device to dma_alloc*()

This makes it possible to get the bus address from Device Tree.

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>

dwc_otg: fix summarize urb->actual_length for isochronous transfers

Kernel does not copy input data of ISO transfers to userspace
if actual_length is set only in ISO transfers and not summarized
in urb->actual_length. Fixes #903

fiq_fsm: Use correct states when starting isoc OUT transfers

In fiq_fsm_start_next_periodic() if an isochronous OUT transfer
was selected, no regard was given as to whether this was a single-packet
transfer or a multi-packet staged transfer.

For single-packet transfers, this had the effect of repeatedly sending
OUT packets with bogus data and lengths.

Eventually if the channel was repeatedly enabled enough times, this
would lock up the OTG core and no further bus transfers would happen.

Set the FSM state up properly if we select a single-packet transfer.

Fixes #1842

dwc_otg: make nak_holdoff work as intended with empty queues

If URBs reading from non-periodic split endpoints were dequeued and
the last transfer from the endpoint was a NAK handshake, the resulting
qh->nak_frame value was stale which would result in unnecessarily long
polling intervals for the first subsequent transfer with a fresh URB.

Fixup qh->nak_frame in dwc_otg_hcd_urb_dequeue and also guard against
a case where a single URB is submitted to the endpoint, a NAK was
received on the transfer immediately prior to receiving data and the
device subsequently resubmits another URB past the qh->nak_frame interval.

Fixes #1709

dwc_otg: fix split transaction data toggle handling around dequeues

See #1709

Fix several issues regarding endpoint state when URBs are dequeued
- If the HCD is disconnected, flush FIQ-enabled channels properly
- Save the data toggle state for bulk endpoints if the last transfer
  from an endpoint where URBs were dequeued returned a data packet
- Reset hc->start_pkt_count properly in assign_and_init_hc()

dwc_otg: fix several potential crash sources

On root port disconnect events, the host driver state is cleared and
in-progress host channels are forcibly stopped. This doesn't play
well with the FIQ running in the background, so:
- Guard the disconnect callback with both the host spinlock and FIQ
  spinlock
- Move qtd dereference in dwc_otg_handle_hc_fsm() after the early-out
  so we don't dereference a qtd that has gone away
- Turn catch-all BUG()s in dwc_otg_handle_hc_fsm() into warnings.

dwc_otg: delete hcd->channel_lock

The lock serves no purpose as it is only held while the HCD spinlock
is already being held.

dwc_otg: remove unnecessary dma-mode channel halts on disconnect interrupt

Host channels are already halted in kill_urbs_in_qh_list() with the
subsequent interrupt processing behaving as if the URB was dequeued
via HCD callback.

There's no need to clobber the host channel registers a second time
as this exposes races between the driver and host channel resulting
in hcd->free_hc_list becoming corrupted.

dwcotg: Allow to build without FIQ on ARM64

Signed-off-by: popcornmix <popcornmix@gmail.com>

dwc_otg: make periodic scheduling behave properly for FS buses

If the root port is in full-speed mode, transfer times at 12mbit/s
would be calculated but matched against high-speed quotas.

Reinitialise hcd->frame_usecs[i] on each port enable event so that
full-speed bandwidth can be tracked sensibly.

Also, don't bother using the FIQ for transfers when in full-speed
mode - at the slower bus speed, interrupt frequency is reduced by
an order of magnitude.

Related issue: #2020

dwc_otg: fiq_fsm: Make isochronous compatibility checks work properly

Get rid of the spammy printk and local pointer mangling.
Also, there is a nominal benefit for using fiq_fsm for isochronous
transfers in FS mode (~1.1k IRQs per second vs 2.1k IRQs per second)
so remove the root port speed check.

dwc_otg: add module parameter int_ep_interval_min

Add a module parameter (defaulting to ignored) that clamps the polling rate
of high-speed Interrupt endpoints to a minimum microframe interval.

The parameter is modifiable at runtime as it is used when activating new
endpoints (such as on device connect).

dwc_otg: fiq_fsm: Add non-periodic TT exclusivity constraints

Certain hub types do not discriminate between pipe direction (IN or OUT)
when considering non-periodic transfers. Therefore these hubs get confused
if multiple transfers are issued in different directions with the same
device address and endpoint number.

Constrain queuing non-periodic split transactions so they are performed
serially in such cases.

Related: #2024

dwc_otg: Fixup change to DRIVER_ATTR interface

dwc_otg: Fix compilation warnings

Signed-off-by: Phil Elwell <phil@raspberrypi.org>

USB_DWCOTG: Disable building dwc_otg as a module (#2265)

When dwc_otg is built as a module, build will fail with the following
error:

ERROR: "DWC_TASK_HI_SCHEDULE" [drivers/usb/host/dwc_otg/dwc_otg.ko] undefined!
scripts/Makefile.modpost:91: recipe for target '__modpost' failed
make[1]: *** [__modpost] Error 1
Makefile:1199: recipe for target 'modules' failed
make: *** [modules] Error 2

Even if the error is solved by including the missing
DWC_TASK_HI_SCHEDULE function, the kernel will panic when loading
dwc_otg.

As a workaround, simply prevent user from building dwc_otg as a module
as the current kernel does not support it.

See: #2258

Signed-off-by: Malik Olivier Boussejra <malik@boussejra.com>

dwc_otg: New timer API

dwc_otg: Fix removed ACCESS_ONCE->READ_ONCE

dwc_otg: don't unconditionally force host mode in dwc_otg_cil_init()

Add the ability to disable force_host_mode for those that want to use
dwc_otg in both device and host modes.

dwc_otg: Fix a regression when dequeueing isochronous transfers

In 282bed9 (dwc_otg: make nak_holdoff work as intended with empty queues)
the dequeue mechanism was changed to leave FIQ-enabled transfers to run
to completion - to avoid leaving hub TT buffers with stale packets lying
around.

This broke FIQ-accelerated isochronous transfers, as this then meant that
dozens of transfers were performed after the dequeue function returned.

Restore the state machine fence for isochronous transfers.

fiq_fsm: rewind DMA pointer for OUT transactions that fail (#2288)

See: #2140

dwc_otg: add smp_mb() to prevent driver state corruption on boot

Occasional crashes have been seen where the FIQ code dereferences
invalid/random pointers immediately after being set up, leading to
panic on boot.

The crash occurs as the FIQ code races against hcd_init_fiq() and
the hcd_init_fiq() code races against the outstanding memory stores
from dwc_otg_hcd_init(). Use explicit barriers after touching
driver state.

usb: dwc_otg: fix memory corruption in dwc_otg driver

[Upstream commit 51b1b64]

The move from the staging tree to the main tree exposed a
longstanding memory corruption bug in the dwc2 driver. The
reordering of the driver initialization caused the dwc2 driver
to corrupt the initialization data of the sdhci driver on the
Raspberry Pi platform, which made the bug show up.

The error is in calling to_usb_device(hsotg->dev), since ->dev
is not a member of struct usb_device. The easiest fix is to
just remove the offending code, since it is not really needed.

Thanks to Stephen Warren for tracking down the cause of this.

Reported-by: Andre Heider <a.heider@gmail.com>
Tested-by: Stephen Warren <swarren@wwwdotorg.org>
Signed-off-by: Paul Zimmerman <paulz@synopsys.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[lukas: port from upstream dwc2 to out-of-tree dwc_otg driver]
Signed-off-by: Lukas Wunner <lukas@wunner.de>

usb: dwb_otg: Fix unreachable switch statement warning

This warning appears with GCC 7.3.0 from toolchains.bootlin.com:

../drivers/usb/host/dwc_otg/dwc_otg_fiq_fsm.c: In function ‘fiq_fsm_update_hs_isoc’:
../drivers/usb/host/dwc_otg/dwc_otg_fiq_fsm.c:595:61: warning: statement will never be executed [-Wswitch-unreachable]
   st->hctsiz_copy.b.xfersize = nrpackets * st->hcchar_copy.b.mps;
                                            ~~~~~~~~~~~~~~~~~^~~~

Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>

dwc_otg: fiq_fsm: fix incorrect DMA register offset calculation

Rationalise the offset and update all call sites.

Fixes #2408

popcornmix added a commit that referenced this issue Feb 18, 2019

Add dwc_otg driver
Signed-off-by: popcornmix <popcornmix@gmail.com>

usb: dwc: fix lockdep false positive

Signed-off-by: Kari Suvanto <karis79@gmail.com>

usb: dwc: fix inconsistent lock state

Signed-off-by: Kari Suvanto <karis79@gmail.com>

Add FIQ patch to dwc_otg driver. Enable with dwc_otg.fiq_fix_enable=1. Should give about 10% more ARM performance.
Thanks to Gordon and Costas

Avoid dynamic memory allocation for channel lock in USB driver. Thanks ddv2005.

Add NAK holdoff scheme. Enabled by default, disable with dwc_otg.nak_holdoff_enable=0. Thanks gsh

Make sure we wait for the reset to finish

dwc_otg: fix bug in dwc_otg_hcd.c resulting in silent kernel
	 memory corruption, escalating to OOPS under high USB load.

dwc_otg: Fix unsafe access of QTD during URB enqueue

In dwc_otg_hcd_urb_enqueue during qtd creation, it was possible that the
transaction could complete almost immediately after the qtd was assigned
to a host channel during URB enqueue, which meant the qtd pointer was no
longer valid having been completed and removed. Usually, this resulted in
an OOPS during URB submission. By predetermining whether transactions
need to be queued or not, this unsafe pointer access is avoided.

This bug was only evident on the Pi model A where a device was attached
that had no periodic endpoints (e.g. USB pendrive or some wlan devices).

dwc_otg: Fix incorrect URB allocation error handling

If the memory allocation for a dwc_otg_urb failed, the kernel would OOPS
because for some reason a member of the *unallocated* struct was set to
zero. Error handling changed to fail correctly.

dwc_otg: fix potential use-after-free case in interrupt handler

If a transaction had previously aborted, certain interrupts are
enabled to track error counts and reset where necessary. On IN
endpoints the host generates an ACK interrupt near-simultaneously
with completion of transfer. In the case where this transfer had
previously had an error, this results in a use-after-free on
the QTD memory space with a 1-byte length being overwritten to
0x00.

dwc_otg: add handling of SPLIT transaction data toggle errors

Previously a data toggle error on packets from a USB1.1 device behind
a TT would result in the Pi locking up as the driver never handled
the associated interrupt. Patch adds basic retry mechanism and
interrupt acknowledgement to cater for either a chance toggle error or
for devices that have a broken initial toggle state (FT8U232/FT232BM).

dwc_otg: implement tasklet for returning URBs to usbcore hcd layer

The dwc_otg driver interrupt handler for transfer completion will spend
a very long time with interrupts disabled when a URB is completed -
this is because usb_hcd_giveback_urb is called from within the handler
which for a USB device driver with complicated processing (e.g. webcam)
will take an exorbitant amount of time to complete. This results in
missed completion interrupts for other USB packets which lead to them
being dropped due to microframe overruns.

This patch splits returning the URB to the usb hcd layer into a
high-priority tasklet. This will have most benefit for isochronous IN
transfers but will also have incidental benefit where multiple periodic
devices are active at once.

dwc_otg: fix NAK holdoff and allow on split transactions only

This corrects a bug where if a single active non-periodic endpoint
had at least one transaction in its qh, on frnum == MAX_FRNUM the qh
would get skipped and never get queued again. This would result in
a silent device until error detection (automatic or otherwise) would
either reset the device or flush and requeue the URBs.

Additionally the NAK holdoff was enabled for all transactions - this
would potentially stall a HS endpoint for 1ms if a previous error state
enabled this interrupt and the next response was a NAK. Fix so that
only split transactions get held off.

dwc_otg: Call usb_hcd_unlink_urb_from_ep with lock held in completion handler

usb_hcd_unlink_urb_from_ep must be called with the HCD lock held.  Calling it
asynchronously in the tasklet was not safe (regression in
c4564d4).

This change unlinks it from the endpoint prior to queueing it for handling in
the tasklet, and also adds a check to ensure the urb is OK to be unlinked
before doing so.

NULL pointer dereference kernel oopses had been observed in usb_hcd_giveback_urb
when a USB device was unplugged/replugged during data transfer.  This effect
was reproduced using automated USB port power control, hundreds of replug
events were performed during active transfers to confirm that the problem was
eliminated.

USB fix using a FIQ to implement split transactions

This commit adds a FIQ implementaion that schedules
the split transactions using a FIQ so we don't get
held off by the interrupt latency of Linux

dwc_otg: fix device attributes and avoid kernel warnings on boot

dcw_otg: avoid logging function that can cause panics

See: raspberrypi/firmware#21
Thanks to cleverca22 for fix

dwc_otg: mask correct interrupts after transaction error recovery

The dwc_otg driver will unmask certain interrupts on a transaction
that previously halted in the error state in order to reset the
QTD error count. The various fine-grained interrupt handlers do not
consider that other interrupts besides themselves were unmasked.

By disabling the two other interrupts only ever enabled in DMA mode
for this purpose, we can avoid unnecessary function calls in the
IRQ handler. This will also prevent an unneccesary FIQ interrupt
from being generated if the FIQ is enabled.

dwc_otg: fiq: prevent FIQ thrash and incorrect state passing to IRQ

In the case of a transaction to a device that had previously aborted
due to an error, several interrupts are enabled to reset the error
count when a device responds. This has the side-effect of making the
FIQ thrash because the hardware will generate multiple instances of
a NAK on an IN bulk/interrupt endpoint and multiple instances of ACK
on an OUT bulk/interrupt endpoint. Make the FIQ mask and clear the
associated interrupts.

Additionally, on non-split transactions make sure that only unmasked
interrupts are cleared. This caused a hard-to-trigger but serious
race condition when you had the combination of an endpoint awaiting
error recovery and a transaction completed on an endpoint - due to
the sequencing and timing of interrupts generated by the dwc_otg core,
it was possible to confuse the IRQ handler.

Fix function tracing

dwc_otg: whitespace cleanup in dwc_otg_urb_enqueue

dwc_otg: prevent OOPSes during device disconnects

The dwc_otg_urb_enqueue function is thread-unsafe. In particular the
access of urb->hcpriv, usb_hcd_link_urb_to_ep, dwc_otg_urb->qtd and
friends does not occur within a critical section and so if a device
was unplugged during activity there was a high chance that the
usbcore hub_thread would try to disable the endpoint with partially-
formed entries in the URB queue. This would result in BUG() or null
pointer dereferences.

Fix so that access of urb->hcpriv, enqueuing to the hardware and
adding to usbcore endpoint URB lists is contained within a single
critical section.

dwc_otg: prevent BUG() in TT allocation if hub address is > 16

A fixed-size array is used to track TT allocation. This was
previously set to 16 which caused a crash because
dwc_otg_hcd_allocate_port would read past the end of the array.

This was hit if a hub was plugged in which enumerated as addr > 16,
due to previous device resets or unplugs.

Also add #ifdef FIQ_DEBUG around hcd->hub_port_alloc[], which grows
to a large size if 128 hub addresses are supported. This field is
for debug only for tracking which frame an allocate happened in.

dwc_otg: make channel halts with unknown state less damaging

If the IRQ received a channel halt interrupt through the FIQ
with no other bits set, the IRQ would not release the host
channel and never complete the URB.

Add catchall handling to treat as a transaction error and retry.

dwc_otg: fiq_split: use TTs with more granularity

This fixes certain issues with split transaction scheduling.

- Isochronous multi-packet OUT transactions now hog the TT until
  they are completed - this prevents hubs aborting transactions
  if they get a periodic start-split out-of-order
- Don't perform TT allocation on non-periodic endpoints - this
  allows simultaneous use of the TT's bulk/control and periodic
  transaction buffers

This commit will mainly affect USB audio playback.

dwc_otg: fix potential sleep while atomic during urb enqueue

Fixes a regression introduced with eb1b482. Kmalloc called from
dwc_otg_hcd_qtd_add / dwc_otg_hcd_qtd_create did not always have
the GPF_ATOMIC flag set. Force this flag when inside the larger
critical section.

dwc_otg: make fiq_split_enable imply fiq_fix_enable

Failing to set up the FIQ correctly would result in
"IRQ 32: nobody cared" errors in dmesg.

dwc_otg: prevent crashes on host port disconnects

Fix several issues resulting in crashes or inconsistent state
if a Model A root port was disconnected.

- Clean up queue heads properly in kill_urbs_in_qh_list by
  removing the empty QHs from the schedule lists
- Set the halt status properly to prevent IRQ handlers from
  using freed memory
- Add fiq_split related cleanup for saved registers
- Make microframe scheduling reclaim host channels if
  active during a disconnect
- Abort URBs with -ESHUTDOWN status response, informing
  device drivers so they respond in a more correct fashion
  and don't try to resubmit URBs
- Prevent IRQ handlers from attempting to handle channel
  interrupts if the associated URB was dequeued (and the
  driver state was cleared)

dwc_otg: prevent leaking URBs during enqueue

A dwc_otg_urb would get leaked if the HCD enqueue function
failed for any reason. Free the URB at the appropriate points.

dwc_otg: Enable NAK holdoff for control split transactions

Certain low-speed devices take a very long time to complete a
data or status stage of a control transaction, producing NAK
responses until they complete internal processing - the USB2.0
spec limit is up to 500mS. This causes the same type of interrupt
storm as seen with USB-serial dongles prior to c8edb23.

In certain circumstances, usually while booting, this interrupt
storm could cause SD card timeouts.

dwc_otg: Fix for occasional lockup on boot when doing a USB reset

dwc_otg: Don't issue traffic to LS devices in FS mode

Issuing low-speed packets when the root port is in full-speed mode
causes the root port to stop responding. Explicitly fail when
enqueuing URBs to a LS endpoint on a FS bus.

Fix ARM architecture issue with local_irq_restore()

If local_fiq_enable() is called before a local_irq_restore(flags) where
the flags variable has the F bit set, the FIQ will be erroneously disabled.

Fixup arch_local_irq_restore to avoid trampling the F bit in CPSR.

Also fix some of the hacks previously implemented for previous dwc_otg
incarnations.

dwc_otg: fiq_fsm: Base commit for driver rewrite

This commit removes the previous FIQ fixes entirely and adds fiq_fsm.

This rewrite features much more complete support for split transactions
and takes into account several OTG hardware bugs. High-speed
isochronous transactions are also capable of being performed by fiq_fsm.

All driver options have been removed and replaced with:
  - dwc_otg.fiq_enable (bool)
  - dwc_otg.fiq_fsm_enable (bool)
  - dwc_otg.fiq_fsm_mask (bitmask)
  - dwc_otg.nak_holdoff (unsigned int)

Defaults are specified such that fiq_fsm behaves similarly to the
previously implemented FIQ fixes.

fiq_fsm: Push error recovery into the FIQ when fiq_fsm is used

If the transfer associated with a QTD failed due to a bus error, the HCD
would retry the transfer up to 3 times (implementing the USB2.0
three-strikes retry in software).

Due to the masking mechanism used by fiq_fsm, it is only possible to pass
a single interrupt through to the HCD per-transfer.

In this instance host channels would fall off the radar because the error
reset would function, but the subsequent channel halt would be lost.

Push the error count reset into the FIQ handler.

fiq_fsm: Implement timeout mechanism

For full-speed endpoints with a large packet size, interrupt latency
runs the risk of the FIQ starting a transaction too late in a full-speed
frame. If the device is still transmitting data when EOF2 for the
downstream frame occurs, the hub will disable the port. This change is
not reflected in the hub status endpoint and the device becomes
unresponsive.

Prevent high-bandwidth transactions from being started too late in a
frame. The mechanism is not guaranteed: a combination of bit stuffing
and hub latency may still result in a device overrunning.

fiq_fsm: fix bounce buffer utilisation for Isochronous OUT

Multi-packet isochronous OUT transactions were subject to a few bounday
bugs. Fix them.

Audio playback is now much more robust: however, an issue stands with
devices that have adaptive sinks - ALSA plays samples too fast.

dwc_otg: Return full-speed frame numbers in HS mode

The frame counter increments on every *microframe* in high-speed mode.
Most device drivers expect this number to be in full-speed frames - this
caused considerable confusion to e.g. snd_usb_audio which uses the
frame counter to estimate the number of samples played.

fiq_fsm: save PID on completion of interrupt OUT transfers

Also add edge case handling for interrupt transports.

Note that for periodic split IN, data toggles are unimplemented in the
OTG host hardware - it unconditionally accepts any PID.

fiq_fsm: add missing case for fiq_fsm_tt_in_use()

Certain combinations of bitrate and endpoint activity could
result in a periodic transaction erroneously getting started
while the previous Isochronous OUT was still active.

fiq_fsm: clear hcintmsk for aborted transactions

Prevents the FIQ from erroneously handling interrupts
on a timed out channel.

fiq_fsm: enable by default

fiq_fsm: fix dequeues for non-periodic split transactions

If a dequeue happened between the SSPLIT and CSPLIT phases of the
transaction, the HCD would never receive an interrupt.

fiq_fsm: Disable by default

fiq_fsm: Handle HC babble errors

The HCTSIZ transfer size field raises a babble interrupt if
the counter wraps. Handle the resulting interrupt in this case.

dwc_otg: fix interrupt registration for fiq_enable=0

Additionally make the module parameter conditional for wherever
hcd->fiq_state is touched.

fiq_fsm: Enable by default

dwc_otg: Fix various issues with root port and transaction errors

Process the host port interrupts correctly (and don't trample them).
Root port hotplug now functional again.

Fix a few thinkos with the transaction error passthrough for fiq_fsm.

fiq_fsm: Implement hack for Split Interrupt transactions

Hubs aren't too picky about which endpoint we send Control type split
transactions to. By treating Interrupt transfers as Control, it is
possible to use the non-periodic queue in the OTG core as well as the
non-periodic FIFOs in the hub itself. This massively reduces the
microframe exclusivity/contention that periodic split transactions
otherwise have to enforce.

It goes without saying that this is a fairly egregious USB specification
violation, but it works.

Original idea by Hans Petter Selasky @ FreeBSD.org.

dwc_otg: FIQ support on SMP. Set up FIQ stack and handler on Core 0 only.

dwc_otg: introduce fiq_fsm_spin(un|)lock()

SMP safety for the FIQ relies on register read-modify write cycles being
completed in the correct order. Several places in the DWC code modify
registers also touched by the FIQ. Protect these by a bare-bones lock
mechanism.

This also makes it possible to run the FIQ and IRQ handlers on different
cores.

fiq_fsm: fix build on bcm2708 and bcm2709 platforms

dwc_otg: put some barriers back where they should be for UP

bcm2709/dwc_otg: Setup FIQ on core 1 if >1 core active

dwc_otg: fixup read-modify-write in critical paths

Be more careful about read-modify-write on registers that the FIQ
also touches.

Guard fiq_fsm_spin_lock with fiq_enable check

fiq_fsm: Falling out of the state machine isn't fatal

This edge case can be hit if the port is disabled while the FIQ is
in the middle of a transaction. Make the effects less severe.

Also get rid of the useless return value.

squash: dwc_otg: Allow to build without SMP

usb: core: make overcurrent messages more prominent

Hub overcurrent messages are more serious than "debug". Increase loglevel.

usb: dwc_otg: Don't use dma_to_virt()

Commit 6ce0d20 changes dma_to_virt() which breaks this driver.
Open code the old dma_to_virt() implementation to work around this.

Limit the use of __bus_to_virt() to cases where transfer_buffer_length
is set and transfer_buffer is not set. This is done to increase the
chance that this driver will also work on ARCH_BCM2835.

transfer_buffer should not be NULL if the length is set, but the
comment in the code indicates that there are situations where this
might happen. drivers/usb/isp1760/isp1760-hcd.c also has a similar
comment pointing to a possible: 'usb storage / SCSI bug'.

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>

dwc_otg: Fix crash when fiq_enable=0

dwc_otg: fiq_fsm: Make high-speed isochronous strided transfers work properly

Certain low-bandwidth high-speed USB devices (specialist audio devices,
compressed-frame webcams) have packet intervals > 1 microframe.

Stride these transfers in the FIQ by using the start-of-frame interrupt
to restart the channel at the right time.

dwc_otg: Force host mode to fix incorrect compute module boards

dwc_otg: Add ARCH_BCM2835 support

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>

dwc_otg: Simplify FIQ irq number code

Dropping ATAGS means we can simplify the FIQ irq number code.
Also add error checking on the returned irq number.

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>

dwc_otg: Remove duplicate gadget probe/unregister function

dwc_otg: Properly set the HFIR

Douglas Anderson reported:

According to the most up to date version of the dwc2 databook, the FRINT
field of the HFIR register should be programmed to:
* 125 us * (PHY clock freq for HS) - 1
* 1000 us * (PHY clock freq for FS/LS) - 1

This is opposed to older versions of the doc that claimed it should be:
* 125 us * (PHY clock freq for HS)
* 1000 us * (PHY clock freq for FS/LS)

and reported lower timing jitter on a USB analyser

dcw_otg: trim xfer length when buffer larger than allocated size is received

dwc_otg: Don't free qh align buffers in atomic context

dwc_otg: Enable the hack for Split Interrupt transactions by default

dwc_otg.fiq_fsm_mask=0xF has long been a suggestion for users with audio stutters or other USB bandwidth issues.
So far we are aware of many success stories but no failure caused by this setting.
Make it a default to learn more.

See: https://www.raspberrypi.org/forums/viewtopic.php?f=28&t=70437

Signed-off-by: popcornmix <popcornmix@gmail.com>

dwc_otg: Use kzalloc when suitable

dwc_otg: Pass struct device to dma_alloc*()

This makes it possible to get the bus address from Device Tree.

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>

dwc_otg: fix summarize urb->actual_length for isochronous transfers

Kernel does not copy input data of ISO transfers to userspace
if actual_length is set only in ISO transfers and not summarized
in urb->actual_length. Fixes #903

fiq_fsm: Use correct states when starting isoc OUT transfers

In fiq_fsm_start_next_periodic() if an isochronous OUT transfer
was selected, no regard was given as to whether this was a single-packet
transfer or a multi-packet staged transfer.

For single-packet transfers, this had the effect of repeatedly sending
OUT packets with bogus data and lengths.

Eventually if the channel was repeatedly enabled enough times, this
would lock up the OTG core and no further bus transfers would happen.

Set the FSM state up properly if we select a single-packet transfer.

Fixes #1842

dwc_otg: make nak_holdoff work as intended with empty queues

If URBs reading from non-periodic split endpoints were dequeued and
the last transfer from the endpoint was a NAK handshake, the resulting
qh->nak_frame value was stale which would result in unnecessarily long
polling intervals for the first subsequent transfer with a fresh URB.

Fixup qh->nak_frame in dwc_otg_hcd_urb_dequeue and also guard against
a case where a single URB is submitted to the endpoint, a NAK was
received on the transfer immediately prior to receiving data and the
device subsequently resubmits another URB past the qh->nak_frame interval.

Fixes #1709

dwc_otg: fix split transaction data toggle handling around dequeues

See #1709

Fix several issues regarding endpoint state when URBs are dequeued
- If the HCD is disconnected, flush FIQ-enabled channels properly
- Save the data toggle state for bulk endpoints if the last transfer
  from an endpoint where URBs were dequeued returned a data packet
- Reset hc->start_pkt_count properly in assign_and_init_hc()

dwc_otg: fix several potential crash sources

On root port disconnect events, the host driver state is cleared and
in-progress host channels are forcibly stopped. This doesn't play
well with the FIQ running in the background, so:
- Guard the disconnect callback with both the host spinlock and FIQ
  spinlock
- Move qtd dereference in dwc_otg_handle_hc_fsm() after the early-out
  so we don't dereference a qtd that has gone away
- Turn catch-all BUG()s in dwc_otg_handle_hc_fsm() into warnings.

dwc_otg: delete hcd->channel_lock

The lock serves no purpose as it is only held while the HCD spinlock
is already being held.

dwc_otg: remove unnecessary dma-mode channel halts on disconnect interrupt

Host channels are already halted in kill_urbs_in_qh_list() with the
subsequent interrupt processing behaving as if the URB was dequeued
via HCD callback.

There's no need to clobber the host channel registers a second time
as this exposes races between the driver and host channel resulting
in hcd->free_hc_list becoming corrupted.

dwcotg: Allow to build without FIQ on ARM64

Signed-off-by: popcornmix <popcornmix@gmail.com>

dwc_otg: make periodic scheduling behave properly for FS buses

If the root port is in full-speed mode, transfer times at 12mbit/s
would be calculated but matched against high-speed quotas.

Reinitialise hcd->frame_usecs[i] on each port enable event so that
full-speed bandwidth can be tracked sensibly.

Also, don't bother using the FIQ for transfers when in full-speed
mode - at the slower bus speed, interrupt frequency is reduced by
an order of magnitude.

Related issue: #2020

dwc_otg: fiq_fsm: Make isochronous compatibility checks work properly

Get rid of the spammy printk and local pointer mangling.
Also, there is a nominal benefit for using fiq_fsm for isochronous
transfers in FS mode (~1.1k IRQs per second vs 2.1k IRQs per second)
so remove the root port speed check.

dwc_otg: add module parameter int_ep_interval_min

Add a module parameter (defaulting to ignored) that clamps the polling rate
of high-speed Interrupt endpoints to a minimum microframe interval.

The parameter is modifiable at runtime as it is used when activating new
endpoints (such as on device connect).

dwc_otg: fiq_fsm: Add non-periodic TT exclusivity constraints

Certain hub types do not discriminate between pipe direction (IN or OUT)
when considering non-periodic transfers. Therefore these hubs get confused
if multiple transfers are issued in different directions with the same
device address and endpoint number.

Constrain queuing non-periodic split transactions so they are performed
serially in such cases.

Related: #2024

dwc_otg: Fixup change to DRIVER_ATTR interface

dwc_otg: Fix compilation warnings

Signed-off-by: Phil Elwell <phil@raspberrypi.org>

USB_DWCOTG: Disable building dwc_otg as a module (#2265)

When dwc_otg is built as a module, build will fail with the following
error:

ERROR: "DWC_TASK_HI_SCHEDULE" [drivers/usb/host/dwc_otg/dwc_otg.ko] undefined!
scripts/Makefile.modpost:91: recipe for target '__modpost' failed
make[1]: *** [__modpost] Error 1
Makefile:1199: recipe for target 'modules' failed
make: *** [modules] Error 2

Even if the error is solved by including the missing
DWC_TASK_HI_SCHEDULE function, the kernel will panic when loading
dwc_otg.

As a workaround, simply prevent user from building dwc_otg as a module
as the current kernel does not support it.

See: #2258

Signed-off-by: Malik Olivier Boussejra <malik@boussejra.com>

dwc_otg: New timer API

dwc_otg: Fix removed ACCESS_ONCE->READ_ONCE

dwc_otg: don't unconditionally force host mode in dwc_otg_cil_init()

Add the ability to disable force_host_mode for those that want to use
dwc_otg in both device and host modes.

dwc_otg: Fix a regression when dequeueing isochronous transfers

In 282bed9 (dwc_otg: make nak_holdoff work as intended with empty queues)
the dequeue mechanism was changed to leave FIQ-enabled transfers to run
to completion - to avoid leaving hub TT buffers with stale packets lying
around.

This broke FIQ-accelerated isochronous transfers, as this then meant that
dozens of transfers were performed after the dequeue function returned.

Restore the state machine fence for isochronous transfers.

fiq_fsm: rewind DMA pointer for OUT transactions that fail (#2288)

See: #2140

dwc_otg: add smp_mb() to prevent driver state corruption on boot

Occasional crashes have been seen where the FIQ code dereferences
invalid/random pointers immediately after being set up, leading to
panic on boot.

The crash occurs as the FIQ code races against hcd_init_fiq() and
the hcd_init_fiq() code races against the outstanding memory stores
from dwc_otg_hcd_init(). Use explicit barriers after touching
driver state.

usb: dwc_otg: fix memory corruption in dwc_otg driver

[Upstream commit 51b1b64]

The move from the staging tree to the main tree exposed a
longstanding memory corruption bug in the dwc2 driver. The
reordering of the driver initialization caused the dwc2 driver
to corrupt the initialization data of the sdhci driver on the
Raspberry Pi platform, which made the bug show up.

The error is in calling to_usb_device(hsotg->dev), since ->dev
is not a member of struct usb_device. The easiest fix is to
just remove the offending code, since it is not really needed.

Thanks to Stephen Warren for tracking down the cause of this.

Reported-by: Andre Heider <a.heider@gmail.com>
Tested-by: Stephen Warren <swarren@wwwdotorg.org>
Signed-off-by: Paul Zimmerman <paulz@synopsys.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[lukas: port from upstream dwc2 to out-of-tree dwc_otg driver]
Signed-off-by: Lukas Wunner <lukas@wunner.de>

usb: dwb_otg: Fix unreachable switch statement warning

This warning appears with GCC 7.3.0 from toolchains.bootlin.com:

../drivers/usb/host/dwc_otg/dwc_otg_fiq_fsm.c: In function ‘fiq_fsm_update_hs_isoc’:
../drivers/usb/host/dwc_otg/dwc_otg_fiq_fsm.c:595:61: warning: statement will never be executed [-Wswitch-unreachable]
   st->hctsiz_copy.b.xfersize = nrpackets * st->hcchar_copy.b.mps;
                                            ~~~~~~~~~~~~~~~~~^~~~

Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>

dwc_otg: fiq_fsm: fix incorrect DMA register offset calculation

Rationalise the offset and update all call sites.

Fixes #2408

popcornmix added a commit that referenced this issue Feb 18, 2019

Add dwc_otg driver
Signed-off-by: popcornmix <popcornmix@gmail.com>

usb: dwc: fix lockdep false positive

Signed-off-by: Kari Suvanto <karis79@gmail.com>

usb: dwc: fix inconsistent lock state

Signed-off-by: Kari Suvanto <karis79@gmail.com>

Add FIQ patch to dwc_otg driver. Enable with dwc_otg.fiq_fix_enable=1. Should give about 10% more ARM performance.
Thanks to Gordon and Costas

Avoid dynamic memory allocation for channel lock in USB driver. Thanks ddv2005.

Add NAK holdoff scheme. Enabled by default, disable with dwc_otg.nak_holdoff_enable=0. Thanks gsh

Make sure we wait for the reset to finish

dwc_otg: fix bug in dwc_otg_hcd.c resulting in silent kernel
	 memory corruption, escalating to OOPS under high USB load.

dwc_otg: Fix unsafe access of QTD during URB enqueue

In dwc_otg_hcd_urb_enqueue during qtd creation, it was possible that the
transaction could complete almost immediately after the qtd was assigned
to a host channel during URB enqueue, which meant the qtd pointer was no
longer valid having been completed and removed. Usually, this resulted in
an OOPS during URB submission. By predetermining whether transactions
need to be queued or not, this unsafe pointer access is avoided.

This bug was only evident on the Pi model A where a device was attached
that had no periodic endpoints (e.g. USB pendrive or some wlan devices).

dwc_otg: Fix incorrect URB allocation error handling

If the memory allocation for a dwc_otg_urb failed, the kernel would OOPS
because for some reason a member of the *unallocated* struct was set to
zero. Error handling changed to fail correctly.

dwc_otg: fix potential use-after-free case in interrupt handler

If a transaction had previously aborted, certain interrupts are
enabled to track error counts and reset where necessary. On IN
endpoints the host generates an ACK interrupt near-simultaneously
with completion of transfer. In the case where this transfer had
previously had an error, this results in a use-after-free on
the QTD memory space with a 1-byte length being overwritten to
0x00.

dwc_otg: add handling of SPLIT transaction data toggle errors

Previously a data toggle error on packets from a USB1.1 device behind
a TT would result in the Pi locking up as the driver never handled
the associated interrupt. Patch adds basic retry mechanism and
interrupt acknowledgement to cater for either a chance toggle error or
for devices that have a broken initial toggle state (FT8U232/FT232BM).

dwc_otg: implement tasklet for returning URBs to usbcore hcd layer

The dwc_otg driver interrupt handler for transfer completion will spend
a very long time with interrupts disabled when a URB is completed -
this is because usb_hcd_giveback_urb is called from within the handler
which for a USB device driver with complicated processing (e.g. webcam)
will take an exorbitant amount of time to complete. This results in
missed completion interrupts for other USB packets which lead to them
being dropped due to microframe overruns.

This patch splits returning the URB to the usb hcd layer into a
high-priority tasklet. This will have most benefit for isochronous IN
transfers but will also have incidental benefit where multiple periodic
devices are active at once.

dwc_otg: fix NAK holdoff and allow on split transactions only

This corrects a bug where if a single active non-periodic endpoint
had at least one transaction in its qh, on frnum == MAX_FRNUM the qh
would get skipped and never get queued again. This would result in
a silent device until error detection (automatic or otherwise) would
either reset the device or flush and requeue the URBs.

Additionally the NAK holdoff was enabled for all transactions - this
would potentially stall a HS endpoint for 1ms if a previous error state
enabled this interrupt and the next response was a NAK. Fix so that
only split transactions get held off.

dwc_otg: Call usb_hcd_unlink_urb_from_ep with lock held in completion handler

usb_hcd_unlink_urb_from_ep must be called with the HCD lock held.  Calling it
asynchronously in the tasklet was not safe (regression in
c4564d4).

This change unlinks it from the endpoint prior to queueing it for handling in
the tasklet, and also adds a check to ensure the urb is OK to be unlinked
before doing so.

NULL pointer dereference kernel oopses had been observed in usb_hcd_giveback_urb
when a USB device was unplugged/replugged during data transfer.  This effect
was reproduced using automated USB port power control, hundreds of replug
events were performed during active transfers to confirm that the problem was
eliminated.

USB fix using a FIQ to implement split transactions

This commit adds a FIQ implementaion that schedules
the split transactions using a FIQ so we don't get
held off by the interrupt latency of Linux

dwc_otg: fix device attributes and avoid kernel warnings on boot

dcw_otg: avoid logging function that can cause panics

See: raspberrypi/firmware#21
Thanks to cleverca22 for fix

dwc_otg: mask correct interrupts after transaction error recovery

The dwc_otg driver will unmask certain interrupts on a transaction
that previously halted in the error state in order to reset the
QTD error count. The various fine-grained interrupt handlers do not
consider that other interrupts besides themselves were unmasked.

By disabling the two other interrupts only ever enabled in DMA mode
for this purpose, we can avoid unnecessary function calls in the
IRQ handler. This will also prevent an unneccesary FIQ interrupt
from being generated if the FIQ is enabled.

dwc_otg: fiq: prevent FIQ thrash and incorrect state passing to IRQ

In the case of a transaction to a device that had previously aborted
due to an error, several interrupts are enabled to reset the error
count when a device responds. This has the side-effect of making the
FIQ thrash because the hardware will generate multiple instances of
a NAK on an IN bulk/interrupt endpoint and multiple instances of ACK
on an OUT bulk/interrupt endpoint. Make the FIQ mask and clear the
associated interrupts.

Additionally, on non-split transactions make sure that only unmasked
interrupts are cleared. This caused a hard-to-trigger but serious
race condition when you had the combination of an endpoint awaiting
error recovery and a transaction completed on an endpoint - due to
the sequencing and timing of interrupts generated by the dwc_otg core,
it was possible to confuse the IRQ handler.

Fix function tracing

dwc_otg: whitespace cleanup in dwc_otg_urb_enqueue

dwc_otg: prevent OOPSes during device disconnects

The dwc_otg_urb_enqueue function is thread-unsafe. In particular the
access of urb->hcpriv, usb_hcd_link_urb_to_ep, dwc_otg_urb->qtd and
friends does not occur within a critical section and so if a device
was unplugged during activity there was a high chance that the
usbcore hub_thread would try to disable the endpoint with partially-
formed entries in the URB queue. This would result in BUG() or null
pointer dereferences.

Fix so that access of urb->hcpriv, enqueuing to the hardware and
adding to usbcore endpoint URB lists is contained within a single
critical section.

dwc_otg: prevent BUG() in TT allocation if hub address is > 16

A fixed-size array is used to track TT allocation. This was
previously set to 16 which caused a crash because
dwc_otg_hcd_allocate_port would read past the end of the array.

This was hit if a hub was plugged in which enumerated as addr > 16,
due to previous device resets or unplugs.

Also add #ifdef FIQ_DEBUG around hcd->hub_port_alloc[], which grows
to a large size if 128 hub addresses are supported. This field is
for debug only for tracking which frame an allocate happened in.

dwc_otg: make channel halts with unknown state less damaging

If the IRQ received a channel halt interrupt through the FIQ
with no other bits set, the IRQ would not release the host
channel and never complete the URB.

Add catchall handling to treat as a transaction error and retry.

dwc_otg: fiq_split: use TTs with more granularity

This fixes certain issues with split transaction scheduling.

- Isochronous multi-packet OUT transactions now hog the TT until
  they are completed - this prevents hubs aborting transactions
  if they get a periodic start-split out-of-order
- Don't perform TT allocation on non-periodic endpoints - this
  allows simultaneous use of the TT's bulk/control and periodic
  transaction buffers

This commit will mainly affect USB audio playback.

dwc_otg: fix potential sleep while atomic during urb enqueue

Fixes a regression introduced with eb1b482. Kmalloc called from
dwc_otg_hcd_qtd_add / dwc_otg_hcd_qtd_create did not always have
the GPF_ATOMIC flag set. Force this flag when inside the larger
critical section.

dwc_otg: make fiq_split_enable imply fiq_fix_enable

Failing to set up the FIQ correctly would result in
"IRQ 32: nobody cared" errors in dmesg.

dwc_otg: prevent crashes on host port disconnects

Fix several issues resulting in crashes or inconsistent state
if a Model A root port was disconnected.

- Clean up queue heads properly in kill_urbs_in_qh_list by
  removing the empty QHs from the schedule lists
- Set the halt status properly to prevent IRQ handlers from
  using freed memory
- Add fiq_split related cleanup for saved registers
- Make microframe scheduling reclaim host channels if
  active during a disconnect
- Abort URBs with -ESHUTDOWN status response, informing
  device drivers so they respond in a more correct fashion
  and don't try to resubmit URBs
- Prevent IRQ handlers from attempting to handle channel
  interrupts if the associated URB was dequeued (and the
  driver state was cleared)

dwc_otg: prevent leaking URBs during enqueue

A dwc_otg_urb would get leaked if the HCD enqueue function
failed for any reason. Free the URB at the appropriate points.

dwc_otg: Enable NAK holdoff for control split transactions

Certain low-speed devices take a very long time to complete a
data or status stage of a control transaction, producing NAK
responses until they complete internal processing - the USB2.0
spec limit is up to 500mS. This causes the same type of interrupt
storm as seen with USB-serial dongles prior to c8edb23.

In certain circumstances, usually while booting, this interrupt
storm could cause SD card timeouts.

dwc_otg: Fix for occasional lockup on boot when doing a USB reset

dwc_otg: Don't issue traffic to LS devices in FS mode

Issuing low-speed packets when the root port is in full-speed mode
causes the root port to stop responding. Explicitly fail when
enqueuing URBs to a LS endpoint on a FS bus.

Fix ARM architecture issue with local_irq_restore()

If local_fiq_enable() is called before a local_irq_restore(flags) where
the flags variable has the F bit set, the FIQ will be erroneously disabled.

Fixup arch_local_irq_restore to avoid trampling the F bit in CPSR.

Also fix some of the hacks previously implemented for previous dwc_otg
incarnations.

dwc_otg: fiq_fsm: Base commit for driver rewrite

This commit removes the previous FIQ fixes entirely and adds fiq_fsm.

This rewrite features much more complete support for split transactions
and takes into account several OTG hardware bugs. High-speed
isochronous transactions are also capable of being performed by fiq_fsm.

All driver options have been removed and replaced with:
  - dwc_otg.fiq_enable (bool)
  - dwc_otg.fiq_fsm_enable (bool)
  - dwc_otg.fiq_fsm_mask (bitmask)
  - dwc_otg.nak_holdoff (unsigned int)

Defaults are specified such that fiq_fsm behaves similarly to the
previously implemented FIQ fixes.

fiq_fsm: Push error recovery into the FIQ when fiq_fsm is used

If the transfer associated with a QTD failed due to a bus error, the HCD
would retry the transfer up to 3 times (implementing the USB2.0
three-strikes retry in software).

Due to the masking mechanism used by fiq_fsm, it is only possible to pass
a single interrupt through to the HCD per-transfer.

In this instance host channels would fall off the radar because the error
reset would function, but the subsequent channel halt would be lost.

Push the error count reset into the FIQ handler.

fiq_fsm: Implement timeout mechanism

For full-speed endpoints with a large packet size, interrupt latency
runs the risk of the FIQ starting a transaction too late in a full-speed
frame. If the device is still transmitting data when EOF2 for the
downstream frame occurs, the hub will disable the port. This change is
not reflected in the hub status endpoint and the device becomes
unresponsive.

Prevent high-bandwidth transactions from being started too late in a
frame. The mechanism is not guaranteed: a combination of bit stuffing
and hub latency may still result in a device overrunning.

fiq_fsm: fix bounce buffer utilisation for Isochronous OUT

Multi-packet isochronous OUT transactions were subject to a few bounday
bugs. Fix them.

Audio playback is now much more robust: however, an issue stands with
devices that have adaptive sinks - ALSA plays samples too fast.

dwc_otg: Return full-speed frame numbers in HS mode

The frame counter increments on every *microframe* in high-speed mode.
Most device drivers expect this number to be in full-speed frames - this
caused considerable confusion to e.g. snd_usb_audio which uses the
frame counter to estimate the number of samples played.

fiq_fsm: save PID on completion of interrupt OUT transfers

Also add edge case handling for interrupt transports.

Note that for periodic split IN, data toggles are unimplemented in the
OTG host hardware - it unconditionally accepts any PID.

fiq_fsm: add missing case for fiq_fsm_tt_in_use()

Certain combinations of bitrate and endpoint activity could
result in a periodic transaction erroneously getting started
while the previous Isochronous OUT was still active.

fiq_fsm: clear hcintmsk for aborted transactions

Prevents the FIQ from erroneously handling interrupts
on a timed out channel.

fiq_fsm: enable by default

fiq_fsm: fix dequeues for non-periodic split transactions

If a dequeue happened between the SSPLIT and CSPLIT phases of the
transaction, the HCD would never receive an interrupt.

fiq_fsm: Disable by default

fiq_fsm: Handle HC babble errors

The HCTSIZ transfer size field raises a babble interrupt if
the counter wraps. Handle the resulting interrupt in this case.

dwc_otg: fix interrupt registration for fiq_enable=0

Additionally make the module parameter conditional for wherever
hcd->fiq_state is touched.

fiq_fsm: Enable by default

dwc_otg: Fix various issues with root port and transaction errors

Process the host port interrupts correctly (and don't trample them).
Root port hotplug now functional again.

Fix a few thinkos with the transaction error passthrough for fiq_fsm.

fiq_fsm: Implement hack for Split Interrupt transactions

Hubs aren't too picky about which endpoint we send Control type split
transactions to. By treating Interrupt transfers as Control, it is
possible to use the non-periodic queue in the OTG core as well as the
non-periodic FIFOs in the hub itself. This massively reduces the
microframe exclusivity/contention that periodic split transactions
otherwise have to enforce.

It goes without saying that this is a fairly egregious USB specification
violation, but it works.

Original idea by Hans Petter Selasky @ FreeBSD.org.

dwc_otg: FIQ support on SMP. Set up FIQ stack and handler on Core 0 only.

dwc_otg: introduce fiq_fsm_spin(un|)lock()

SMP safety for the FIQ relies on register read-modify write cycles being
completed in the correct order. Several places in the DWC code modify
registers also touched by the FIQ. Protect these by a bare-bones lock
mechanism.

This also makes it possible to run the FIQ and IRQ handlers on different
cores.

fiq_fsm: fix build on bcm2708 and bcm2709 platforms

dwc_otg: put some barriers back where they should be for UP

bcm2709/dwc_otg: Setup FIQ on core 1 if >1 core active

dwc_otg: fixup read-modify-write in critical paths

Be more careful about read-modify-write on registers that the FIQ
also touches.

Guard fiq_fsm_spin_lock with fiq_enable check

fiq_fsm: Falling out of the state machine isn't fatal

This edge case can be hit if the port is disabled while the FIQ is
in the middle of a transaction. Make the effects less severe.

Also get rid of the useless return value.

squash: dwc_otg: Allow to build without SMP

usb: core: make overcurrent messages more prominent

Hub overcurrent messages are more serious than "debug". Increase loglevel.

usb: dwc_otg: Don't use dma_to_virt()

Commit 6ce0d20 changes dma_to_virt() which breaks this driver.
Open code the old dma_to_virt() implementation to work around this.

Limit the use of __bus_to_virt() to cases where transfer_buffer_length
is set and transfer_buffer is not set. This is done to increase the
chance that this driver will also work on ARCH_BCM2835.

transfer_buffer should not be NULL if the length is set, but the
comment in the code indicates that there are situations where this
might happen. drivers/usb/isp1760/isp1760-hcd.c also has a similar
comment pointing to a possible: 'usb storage / SCSI bug'.

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>

dwc_otg: Fix crash when fiq_enable=0

dwc_otg: fiq_fsm: Make high-speed isochronous strided transfers work properly

Certain low-bandwidth high-speed USB devices (specialist audio devices,
compressed-frame webcams) have packet intervals > 1 microframe.

Stride these transfers in the FIQ by using the start-of-frame interrupt
to restart the channel at the right time.

dwc_otg: Force host mode to fix incorrect compute module boards

dwc_otg: Add ARCH_BCM2835 support

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>

dwc_otg: Simplify FIQ irq number code

Dropping ATAGS means we can simplify the FIQ irq number code.
Also add error checking on the returned irq number.

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>

dwc_otg: Remove duplicate gadget probe/unregister function

dwc_otg: Properly set the HFIR

Douglas Anderson reported:

According to the most up to date version of the dwc2 databook, the FRINT
field of the HFIR register should be programmed to:
* 125 us * (PHY clock freq for HS) - 1
* 1000 us * (PHY clock freq for FS/LS) - 1

This is opposed to older versions of the doc that claimed it should be:
* 125 us * (PHY clock freq for HS)
* 1000 us * (PHY clock freq for FS/LS)

and reported lower timing jitter on a USB analyser

dcw_otg: trim xfer length when buffer larger than allocated size is received

dwc_otg: Don't free qh align buffers in atomic context

dwc_otg: Enable the hack for Split Interrupt transactions by default

dwc_otg.fiq_fsm_mask=0xF has long been a suggestion for users with audio stutters or other USB bandwidth issues.
So far we are aware of many success stories but no failure caused by this setting.
Make it a default to learn more.

See: https://www.raspberrypi.org/forums/viewtopic.php?f=28&t=70437

Signed-off-by: popcornmix <popcornmix@gmail.com>

dwc_otg: Use kzalloc when suitable

dwc_otg: Pass struct device to dma_alloc*()

This makes it possible to get the bus address from Device Tree.

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>

dwc_otg: fix summarize urb->actual_length for isochronous transfers

Kernel does not copy input data of ISO transfers to userspace
if actual_length is set only in ISO transfers and not summarized
in urb->actual_length. Fixes #903

fiq_fsm: Use correct states when starting isoc OUT transfers

In fiq_fsm_start_next_periodic() if an isochronous OUT transfer
was selected, no regard was given as to whether this was a single-packet
transfer or a multi-packet staged transfer.

For single-packet transfers, this had the effect of repeatedly sending
OUT packets with bogus data and lengths.

Eventually if the channel was repeatedly enabled enough times, this
would lock up the OTG core and no further bus transfers would happen.

Set the FSM state up properly if we select a single-packet transfer.

Fixes #1842

dwc_otg: make nak_holdoff work as intended with empty queues

If URBs reading from non-periodic split endpoints were dequeued and
the last transfer from the endpoint was a NAK handshake, the resulting
qh->nak_frame value was stale which would result in unnecessarily long
polling intervals for the first subsequent transfer with a fresh URB.

Fixup qh->nak_frame in dwc_otg_hcd_urb_dequeue and also guard against
a case where a single URB is submitted to the endpoint, a NAK was
received on the transfer immediately prior to receiving data and the
device subsequently resubmits another URB past the qh->nak_frame interval.

Fixes #1709

dwc_otg: fix split transaction data toggle handling around dequeues

See #1709

Fix several issues regarding endpoint state when URBs are dequeued
- If the HCD is disconnected, flush FIQ-enabled channels properly
- Save the data toggle state for bulk endpoints if the last transfer
  from an endpoint where URBs were dequeued returned a data packet
- Reset hc->start_pkt_count properly in assign_and_init_hc()

dwc_otg: fix several potential crash sources

On root port disconnect events, the host driver state is cleared and
in-progress host channels are forcibly stopped. This doesn't play
well with the FIQ running in the background, so:
- Guard the disconnect callback with both the host spinlock and FIQ
  spinlock
- Move qtd dereference in dwc_otg_handle_hc_fsm() after the early-out
  so we don't dereference a qtd that has gone away
- Turn catch-all BUG()s in dwc_otg_handle_hc_fsm() into warnings.

dwc_otg: delete hcd->channel_lock

The lock serves no purpose as it is only held while the HCD spinlock
is already being held.

dwc_otg: remove unnecessary dma-mode channel halts on disconnect interrupt

Host channels are already halted in kill_urbs_in_qh_list() with the
subsequent interrupt processing behaving as if the URB was dequeued
via HCD callback.

There's no need to clobber the host channel registers a second time
as this exposes races between the driver and host channel resulting
in hcd->free_hc_list becoming corrupted.

dwcotg: Allow to build without FIQ on ARM64

Signed-off-by: popcornmix <popcornmix@gmail.com>

dwc_otg: make periodic scheduling behave properly for FS buses

If the root port is in full-speed mode, transfer times at 12mbit/s
would be calculated but matched against high-speed quotas.

Reinitialise hcd->frame_usecs[i] on each port enable event so that
full-speed bandwidth can be tracked sensibly.

Also, don't bother using the FIQ for transfers when in full-speed
mode - at the slower bus speed, interrupt frequency is reduced by
an order of magnitude.

Related issue: #2020

dwc_otg: fiq_fsm: Make isochronous compatibility checks work properly

Get rid of the spammy printk and local pointer mangling.
Also, there is a nominal benefit for using fiq_fsm for isochronous
transfers in FS mode (~1.1k IRQs per second vs 2.1k IRQs per second)
so remove the root port speed check.

dwc_otg: add module parameter int_ep_interval_min

Add a module parameter (defaulting to ignored) that clamps the polling rate
of high-speed Interrupt endpoints to a minimum microframe interval.

The parameter is modifiable at runtime as it is used when activating new
endpoints (such as on device connect).

dwc_otg: fiq_fsm: Add non-periodic TT exclusivity constraints

Certain hub types do not discriminate between pipe direction (IN or OUT)
when considering non-periodic transfers. Therefore these hubs get confused
if multiple transfers are issued in different directions with the same
device address and endpoint number.

Constrain queuing non-periodic split transactions so they are performed
serially in such cases.

Related: #2024

dwc_otg: Fixup change to DRIVER_ATTR interface

dwc_otg: Fix compilation warnings

Signed-off-by: Phil Elwell <phil@raspberrypi.org>

USB_DWCOTG: Disable building dwc_otg as a module (#2265)

When dwc_otg is built as a module, build will fail with the following
error:

ERROR: "DWC_TASK_HI_SCHEDULE" [drivers/usb/host/dwc_otg/dwc_otg.ko] undefined!
scripts/Makefile.modpost:91: recipe for target '__modpost' failed
make[1]: *** [__modpost] Error 1
Makefile:1199: recipe for target 'modules' failed
make: *** [modules] Error 2

Even if the error is solved by including the missing
DWC_TASK_HI_SCHEDULE function, the kernel will panic when loading
dwc_otg.

As a workaround, simply prevent user from building dwc_otg as a module
as the current kernel does not support it.

See: #2258

Signed-off-by: Malik Olivier Boussejra <malik@boussejra.com>

dwc_otg: New timer API

dwc_otg: Fix removed ACCESS_ONCE->READ_ONCE

dwc_otg: don't unconditionally force host mode in dwc_otg_cil_init()

Add the ability to disable force_host_mode for those that want to use
dwc_otg in both device and host modes.

dwc_otg: Fix a regression when dequeueing isochronous transfers

In 282bed9 (dwc_otg: make nak_holdoff work as intended with empty queues)
the dequeue mechanism was changed to leave FIQ-enabled transfers to run
to completion - to avoid leaving hub TT buffers with stale packets lying
around.

This broke FIQ-accelerated isochronous transfers, as this then meant that
dozens of transfers were performed after the dequeue function returned.

Restore the state machine fence for isochronous transfers.

fiq_fsm: rewind DMA pointer for OUT transactions that fail (#2288)

See: #2140

dwc_otg: add smp_mb() to prevent driver state corruption on boot

Occasional crashes have been seen where the FIQ code dereferences
invalid/random pointers immediately after being set up, leading to
panic on boot.

The crash occurs as the FIQ code races against hcd_init_fiq() and
the hcd_init_fiq() code races against the outstanding memory stores
from dwc_otg_hcd_init(). Use explicit barriers after touching
driver state.

usb: dwc_otg: fix memory corruption in dwc_otg driver

[Upstream commit 51b1b64]

The move from the staging tree to the main tree exposed a
longstanding memory corruption bug in the dwc2 driver. The
reordering of the driver initialization caused the dwc2 driver
to corrupt the initialization data of the sdhci driver on the
Raspberry Pi platform, which made the bug show up.

The error is in calling to_usb_device(hsotg->dev), since ->dev
is not a member of struct usb_device. The easiest fix is to
just remove the offending code, since it is not really needed.

Thanks to Stephen Warren for tracking down the cause of this.

Reported-by: Andre Heider <a.heider@gmail.com>
Tested-by: Stephen Warren <swarren@wwwdotorg.org>
Signed-off-by: Paul Zimmerman <paulz@synopsys.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[lukas: port from upstream dwc2 to out-of-tree dwc_otg driver]
Signed-off-by: Lukas Wunner <lukas@wunner.de>

usb: dwb_otg: Fix unreachable switch statement warning

This warning appears with GCC 7.3.0 from toolchains.bootlin.com:

../drivers/usb/host/dwc_otg/dwc_otg_fiq_fsm.c: In function ‘fiq_fsm_update_hs_isoc’:
../drivers/usb/host/dwc_otg/dwc_otg_fiq_fsm.c:595:61: warning: statement will never be executed [-Wswitch-unreachable]
   st->hctsiz_copy.b.xfersize = nrpackets * st->hcchar_copy.b.mps;
                                            ~~~~~~~~~~~~~~~~~^~~~

Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>

dwc_otg: fiq_fsm: fix incorrect DMA register offset calculation

Rationalise the offset and update all call sites.

Fixes #2408

popcornmix added a commit that referenced this issue Feb 22, 2019

Add dwc_otg driver
Signed-off-by: popcornmix <popcornmix@gmail.com>

usb: dwc: fix lockdep false positive

Signed-off-by: Kari Suvanto <karis79@gmail.com>

usb: dwc: fix inconsistent lock state

Signed-off-by: Kari Suvanto <karis79@gmail.com>

Add FIQ patch to dwc_otg driver. Enable with dwc_otg.fiq_fix_enable=1. Should give about 10% more ARM performance.
Thanks to Gordon and Costas

Avoid dynamic memory allocation for channel lock in USB driver. Thanks ddv2005.

Add NAK holdoff scheme. Enabled by default, disable with dwc_otg.nak_holdoff_enable=0. Thanks gsh

Make sure we wait for the reset to finish

dwc_otg: fix bug in dwc_otg_hcd.c resulting in silent kernel
	 memory corruption, escalating to OOPS under high USB load.

dwc_otg: Fix unsafe access of QTD during URB enqueue

In dwc_otg_hcd_urb_enqueue during qtd creation, it was possible that the
transaction could complete almost immediately after the qtd was assigned
to a host channel during URB enqueue, which meant the qtd pointer was no
longer valid having been completed and removed. Usually, this resulted in
an OOPS during URB submission. By predetermining whether transactions
need to be queued or not, this unsafe pointer access is avoided.

This bug was only evident on the Pi model A where a device was attached
that had no periodic endpoints (e.g. USB pendrive or some wlan devices).

dwc_otg: Fix incorrect URB allocation error handling

If the memory allocation for a dwc_otg_urb failed, the kernel would OOPS
because for some reason a member of the *unallocated* struct was set to
zero. Error handling changed to fail correctly.

dwc_otg: fix potential use-after-free case in interrupt handler

If a transaction had previously aborted, certain interrupts are
enabled to track error counts and reset where necessary. On IN
endpoints the host generates an ACK interrupt near-simultaneously
with completion of transfer. In the case where this transfer had
previously had an error, this results in a use-after-free on
the QTD memory space with a 1-byte length being overwritten to
0x00.

dwc_otg: add handling of SPLIT transaction data toggle errors

Previously a data toggle error on packets from a USB1.1 device behind
a TT would result in the Pi locking up as the driver never handled
the associated interrupt. Patch adds basic retry mechanism and
interrupt acknowledgement to cater for either a chance toggle error or
for devices that have a broken initial toggle state (FT8U232/FT232BM).

dwc_otg: implement tasklet for returning URBs to usbcore hcd layer

The dwc_otg driver interrupt handler for transfer completion will spend
a very long time with interrupts disabled when a URB is completed -
this is because usb_hcd_giveback_urb is called from within the handler
which for a USB device driver with complicated processing (e.g. webcam)
will take an exorbitant amount of time to complete. This results in
missed completion interrupts for other USB packets which lead to them
being dropped due to microframe overruns.

This patch splits returning the URB to the usb hcd layer into a
high-priority tasklet. This will have most benefit for isochronous IN
transfers but will also have incidental benefit where multiple periodic
devices are active at once.

dwc_otg: fix NAK holdoff and allow on split transactions only

This corrects a bug where if a single active non-periodic endpoint
had at least one transaction in its qh, on frnum == MAX_FRNUM the qh
would get skipped and never get queued again. This would result in
a silent device until error detection (automatic or otherwise) would
either reset the device or flush and requeue the URBs.

Additionally the NAK holdoff was enabled for all transactions - this
would potentially stall a HS endpoint for 1ms if a previous error state
enabled this interrupt and the next response was a NAK. Fix so that
only split transactions get held off.

dwc_otg: Call usb_hcd_unlink_urb_from_ep with lock held in completion handler

usb_hcd_unlink_urb_from_ep must be called with the HCD lock held.  Calling it
asynchronously in the tasklet was not safe (regression in
c4564d4).

This change unlinks it from the endpoint prior to queueing it for handling in
the tasklet, and also adds a check to ensure the urb is OK to be unlinked
before doing so.

NULL pointer dereference kernel oopses had been observed in usb_hcd_giveback_urb
when a USB device was unplugged/replugged during data transfer.  This effect
was reproduced using automated USB port power control, hundreds of replug
events were performed during active transfers to confirm that the problem was
eliminated.

USB fix using a FIQ to implement split transactions

This commit adds a FIQ implementaion that schedules
the split transactions using a FIQ so we don't get
held off by the interrupt latency of Linux

dwc_otg: fix device attributes and avoid kernel warnings on boot

dcw_otg: avoid logging function that can cause panics

See: raspberrypi/firmware#21
Thanks to cleverca22 for fix

dwc_otg: mask correct interrupts after transaction error recovery

The dwc_otg driver will unmask certain interrupts on a transaction
that previously halted in the error state in order to reset the
QTD error count. The various fine-grained interrupt handlers do not
consider that other interrupts besides themselves were unmasked.

By disabling the two other interrupts only ever enabled in DMA mode
for this purpose, we can avoid unnecessary function calls in the
IRQ handler. This will also prevent an unneccesary FIQ interrupt
from being generated if the FIQ is enabled.

dwc_otg: fiq: prevent FIQ thrash and incorrect state passing to IRQ

In the case of a transaction to a device that had previously aborted
due to an error, several interrupts are enabled to reset the error
count when a device responds. This has the side-effect of making the
FIQ thrash because the hardware will generate multiple instances of
a NAK on an IN bulk/interrupt endpoint and multiple instances of ACK
on an OUT bulk/interrupt endpoint. Make the FIQ mask and clear the
associated interrupts.

Additionally, on non-split transactions make sure that only unmasked
interrupts are cleared. This caused a hard-to-trigger but serious
race condition when you had the combination of an endpoint awaiting
error recovery and a transaction completed on an endpoint - due to
the sequencing and timing of interrupts generated by the dwc_otg core,
it was possible to confuse the IRQ handler.

Fix function tracing

dwc_otg: whitespace cleanup in dwc_otg_urb_enqueue

dwc_otg: prevent OOPSes during device disconnects

The dwc_otg_urb_enqueue function is thread-unsafe. In particular the
access of urb->hcpriv, usb_hcd_link_urb_to_ep, dwc_otg_urb->qtd and
friends does not occur within a critical section and so if a device
was unplugged during activity there was a high chance that the
usbcore hub_thread would try to disable the endpoint with partially-
formed entries in the URB queue. This would result in BUG() or null
pointer dereferences.

Fix so that access of urb->hcpriv, enqueuing to the hardware and
adding to usbcore endpoint URB lists is contained within a single
critical section.

dwc_otg: prevent BUG() in TT allocation if hub address is > 16

A fixed-size array is used to track TT allocation. This was
previously set to 16 which caused a crash because
dwc_otg_hcd_allocate_port would read past the end of the array.

This was hit if a hub was plugged in which enumerated as addr > 16,
due to previous device resets or unplugs.

Also add #ifdef FIQ_DEBUG around hcd->hub_port_alloc[], which grows
to a large size if 128 hub addresses are supported. This field is
for debug only for tracking which frame an allocate happened in.

dwc_otg: make channel halts with unknown state less damaging

If the IRQ received a channel halt interrupt through the FIQ
with no other bits set, the IRQ would not release the host
channel and never complete the URB.

Add catchall handling to treat as a transaction error and retry.

dwc_otg: fiq_split: use TTs with more granularity

This fixes certain issues with split transaction scheduling.

- Isochronous multi-packet OUT transactions now hog the TT until
  they are completed - this prevents hubs aborting transactions
  if they get a periodic start-split out-of-order
- Don't perform TT allocation on non-periodic endpoints - this
  allows simultaneous use of the TT's bulk/control and periodic
  transaction buffers

This commit will mainly affect USB audio playback.

dwc_otg: fix potential sleep while atomic during urb enqueue

Fixes a regression introduced with eb1b482. Kmalloc called from
dwc_otg_hcd_qtd_add / dwc_otg_hcd_qtd_create did not always have
the GPF_ATOMIC flag set. Force this flag when inside the larger
critical section.

dwc_otg: make fiq_split_enable imply fiq_fix_enable

Failing to set up the FIQ correctly would result in
"IRQ 32: nobody cared" errors in dmesg.

dwc_otg: prevent crashes on host port disconnects

Fix several issues resulting in crashes or inconsistent state
if a Model A root port was disconnected.

- Clean up queue heads properly in kill_urbs_in_qh_list by
  removing the empty QHs from the schedule lists
- Set the halt status properly to prevent IRQ handlers from
  using freed memory
- Add fiq_split related cleanup for saved registers
- Make microframe scheduling reclaim host channels if
  active during a disconnect
- Abort URBs with -ESHUTDOWN status response, informing
  device drivers so they respond in a more correct fashion
  and don't try to resubmit URBs
- Prevent IRQ handlers from attempting to handle channel
  interrupts if the associated URB was dequeued (and the
  driver state was cleared)

dwc_otg: prevent leaking URBs during enqueue

A dwc_otg_urb would get leaked if the HCD enqueue function
failed for any reason. Free the URB at the appropriate points.

dwc_otg: Enable NAK holdoff for control split transactions

Certain low-speed devices take a very long time to complete a
data or status stage of a control transaction, producing NAK
responses until they complete internal processing - the USB2.0
spec limit is up to 500mS. This causes the same type of interrupt
storm as seen with USB-serial dongles prior to c8edb23.

In certain circumstances, usually while booting, this interrupt
storm could cause SD card timeouts.

dwc_otg: Fix for occasional lockup on boot when doing a USB reset

dwc_otg: Don't issue traffic to LS devices in FS mode

Issuing low-speed packets when the root port is in full-speed mode
causes the root port to stop responding. Explicitly fail when
enqueuing URBs to a LS endpoint on a FS bus.

Fix ARM architecture issue with local_irq_restore()

If local_fiq_enable() is called before a local_irq_restore(flags) where
the flags variable has the F bit set, the FIQ will be erroneously disabled.

Fixup arch_local_irq_restore to avoid trampling the F bit in CPSR.

Also fix some of the hacks previously implemented for previous dwc_otg
incarnations.

dwc_otg: fiq_fsm: Base commit for driver rewrite

This commit removes the previous FIQ fixes entirely and adds fiq_fsm.

This rewrite features much more complete support for split transactions
and takes into account several OTG hardware bugs. High-speed
isochronous transactions are also capable of being performed by fiq_fsm.

All driver options have been removed and replaced with:
  - dwc_otg.fiq_enable (bool)
  - dwc_otg.fiq_fsm_enable (bool)
  - dwc_otg.fiq_fsm_mask (bitmask)
  - dwc_otg.nak_holdoff (unsigned int)

Defaults are specified such that fiq_fsm behaves similarly to the
previously implemented FIQ fixes.

fiq_fsm: Push error recovery into the FIQ when fiq_fsm is used

If the transfer associated with a QTD failed due to a bus error, the HCD
would retry the transfer up to 3 times (implementing the USB2.0
three-strikes retry in software).

Due to the masking mechanism used by fiq_fsm, it is only possible to pass
a single interrupt through to the HCD per-transfer.

In this instance host channels would fall off the radar because the error
reset would function, but the subsequent channel halt would be lost.

Push the error count reset into the FIQ handler.

fiq_fsm: Implement timeout mechanism

For full-speed endpoints with a large packet size, interrupt latency
runs the risk of the FIQ starting a transaction too late in a full-speed
frame. If the device is still transmitting data when EOF2 for the
downstream frame occurs, the hub will disable the port. This change is
not reflected in the hub status endpoint and the device becomes
unresponsive.

Prevent high-bandwidth transactions from being started too late in a
frame. The mechanism is not guaranteed: a combination of bit stuffing
and hub latency may still result in a device overrunning.

fiq_fsm: fix bounce buffer utilisation for Isochronous OUT

Multi-packet isochronous OUT transactions were subject to a few bounday
bugs. Fix them.

Audio playback is now much more robust: however, an issue stands with
devices that have adaptive sinks - ALSA plays samples too fast.

dwc_otg: Return full-speed frame numbers in HS mode

The frame counter increments on every *microframe* in high-speed mode.
Most device drivers expect this number to be in full-speed frames - this
caused considerable confusion to e.g. snd_usb_audio which uses the
frame counter to estimate the number of samples played.

fiq_fsm: save PID on completion of interrupt OUT transfers

Also add edge case handling for interrupt transports.

Note that for periodic split IN, data toggles are unimplemented in the
OTG host hardware - it unconditionally accepts any PID.

fiq_fsm: add missing case for fiq_fsm_tt_in_use()

Certain combinations of bitrate and endpoint activity could
result in a periodic transaction erroneously getting started
while the previous Isochronous OUT was still active.

fiq_fsm: clear hcintmsk for aborted transactions

Prevents the FIQ from erroneously handling interrupts
on a timed out channel.

fiq_fsm: enable by default

fiq_fsm: fix dequeues for non-periodic split transactions

If a dequeue happened between the SSPLIT and CSPLIT phases of the
transaction, the HCD would never receive an interrupt.

fiq_fsm: Disable by default

fiq_fsm: Handle HC babble errors

The HCTSIZ transfer size field raises a babble interrupt if
the counter wraps. Handle the resulting interrupt in this case.

dwc_otg: fix interrupt registration for fiq_enable=0

Additionally make the module parameter conditional for wherever
hcd->fiq_state is touched.

fiq_fsm: Enable by default

dwc_otg: Fix various issues with root port and transaction errors

Process the host port interrupts correctly (and don't trample them).
Root port hotplug now functional again.

Fix a few thinkos with the transaction error passthrough for fiq_fsm.

fiq_fsm: Implement hack for Split Interrupt transactions

Hubs aren't too picky about which endpoint we send Control type split
transactions to. By treating Interrupt transfers as Control, it is
possible to use the non-periodic queue in the OTG core as well as the
non-periodic FIFOs in the hub itself. This massively reduces the
microframe exclusivity/contention that periodic split transactions
otherwise have to enforce.

It goes without saying that this is a fairly egregious USB specification
violation, but it works.

Original idea by Hans Petter Selasky @ FreeBSD.org.

dwc_otg: FIQ support on SMP. Set up FIQ stack and handler on Core 0 only.

dwc_otg: introduce fiq_fsm_spin(un|)lock()

SMP safety for the FIQ relies on register read-modify write cycles being
completed in the correct order. Several places in the DWC code modify
registers also touched by the FIQ. Protect these by a bare-bones lock
mechanism.

This also makes it possible to run the FIQ and IRQ handlers on different
cores.

fiq_fsm: fix build on bcm2708 and bcm2709 platforms

dwc_otg: put some barriers back where they should be for UP

bcm2709/dwc_otg: Setup FIQ on core 1 if >1 core active

dwc_otg: fixup read-modify-write in critical paths

Be more careful about read-modify-write on registers that the FIQ
also touches.

Guard fiq_fsm_spin_lock with fiq_enable check

fiq_fsm: Falling out of the state machine isn't fatal

This edge case can be hit if the port is disabled while the FIQ is
in the middle of a transaction. Make the effects less severe.

Also get rid of the useless return value.

squash: dwc_otg: Allow to build without SMP

usb: core: make overcurrent messages more prominent

Hub overcurrent messages are more serious than "debug". Increase loglevel.

usb: dwc_otg: Don't use dma_to_virt()

Commit 6ce0d20 changes dma_to_virt() which breaks this driver.
Open code the old dma_to_virt() implementation to work around this.

Limit the use of __bus_to_virt() to cases where transfer_buffer_length
is set and transfer_buffer is not set. This is done to increase the
chance that this driver will also work on ARCH_BCM2835.

transfer_buffer should not be NULL if the length is set, but the
comment in the code indicates that there are situations where this
might happen. drivers/usb/isp1760/isp1760-hcd.c also has a similar
comment pointing to a possible: 'usb storage / SCSI bug'.

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>

dwc_otg: Fix crash when fiq_enable=0

dwc_otg: fiq_fsm: Make high-speed isochronous strided transfers work properly

Certain low-bandwidth high-speed USB devices (specialist audio devices,
compressed-frame webcams) have packet intervals > 1 microframe.

Stride these transfers in the FIQ by using the start-of-frame interrupt
to restart the channel at the right time.

dwc_otg: Force host mode to fix incorrect compute module boards

dwc_otg: Add ARCH_BCM2835 support

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>

dwc_otg: Simplify FIQ irq number code

Dropping ATAGS means we can simplify the FIQ irq number code.
Also add error checking on the returned irq number.

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>

dwc_otg: Remove duplicate gadget probe/unregister function

dwc_otg: Properly set the HFIR

Douglas Anderson reported:

According to the most up to date version of the dwc2 databook, the FRINT
field of the HFIR register should be programmed to:
* 125 us * (PHY clock freq for HS) - 1
* 1000 us * (PHY clock freq for FS/LS) - 1

This is opposed to older versions of the doc that claimed it should be:
* 125 us * (PHY clock freq for HS)
* 1000 us * (PHY clock freq for FS/LS)

and reported lower timing jitter on a USB analyser

dcw_otg: trim xfer length when buffer larger than allocated size is received

dwc_otg: Don't free qh align buffers in atomic context

dwc_otg: Enable the hack for Split Interrupt transactions by default

dwc_otg.fiq_fsm_mask=0xF has long been a suggestion for users with audio stutters or other USB bandwidth issues.
So far we are aware of many success stories but no failure caused by this setting.
Make it a default to learn more.

See: https://www.raspberrypi.org/forums/viewtopic.php?f=28&t=70437

Signed-off-by: popcornmix <popcornmix@gmail.com>

dwc_otg: Use kzalloc when suitable

dwc_otg: Pass struct device to dma_alloc*()

This makes it possible to get the bus address from Device Tree.

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>

dwc_otg: fix summarize urb->actual_length for isochronous transfers

Kernel does not copy input data of ISO transfers to userspace
if actual_length is set only in ISO transfers and not summarized
in urb->actual_length. Fixes #903

fiq_fsm: Use correct states when starting isoc OUT transfers

In fiq_fsm_start_next_periodic() if an isochronous OUT transfer
was selected, no regard was given as to whether this was a single-packet
transfer or a multi-packet staged transfer.

For single-packet transfers, this had the effect of repeatedly sending
OUT packets with bogus data and lengths.

Eventually if the channel was repeatedly enabled enough times, this
would lock up the OTG core and no further bus transfers would happen.

Set the FSM state up properly if we select a single-packet transfer.

Fixes #1842

dwc_otg: make nak_holdoff work as intended with empty queues

If URBs reading from non-periodic split endpoints were dequeued and
the last transfer from the endpoint was a NAK handshake, the resulting
qh->nak_frame value was stale which would result in unnecessarily long
polling intervals for the first subsequent transfer with a fresh URB.

Fixup qh->nak_frame in dwc_otg_hcd_urb_dequeue and also guard against
a case where a single URB is submitted to the endpoint, a NAK was
received on the transfer immediately prior to receiving data and the
device subsequently resubmits another URB past the qh->nak_frame interval.

Fixes #1709

dwc_otg: fix split transaction data toggle handling around dequeues

See #1709

Fix several issues regarding endpoint state when URBs are dequeued
- If the HCD is disconnected, flush FIQ-enabled channels properly
- Save the data toggle state for bulk endpoints if the last transfer
  from an endpoint where URBs were dequeued returned a data packet
- Reset hc->start_pkt_count properly in assign_and_init_hc()

dwc_otg: fix several potential crash sources

On root port disconnect events, the host driver state is cleared and
in-progress host channels are forcibly stopped. This doesn't play
well with the FIQ running in the background, so:
- Guard the disconnect callback with both the host spinlock and FIQ
  spinlock
- Move qtd dereference in dwc_otg_handle_hc_fsm() after the early-out
  so we don't dereference a qtd that has gone away
- Turn catch-all BUG()s in dwc_otg_handle_hc_fsm() into warnings.

dwc_otg: delete hcd->channel_lock

The lock serves no purpose as it is only held while the HCD spinlock
is already being held.

dwc_otg: remove unnecessary dma-mode channel halts on disconnect interrupt

Host channels are already halted in kill_urbs_in_qh_list() with the
subsequent interrupt processing behaving as if the URB was dequeued
via HCD callback.

There's no need to clobber the host channel registers a second time
as this exposes races between the driver and host channel resulting
in hcd->free_hc_list becoming corrupted.

dwcotg: Allow to build without FIQ on ARM64

Signed-off-by: popcornmix <popcornmix@gmail.com>

dwc_otg: make periodic scheduling behave properly for FS buses

If the root port is in full-speed mode, transfer times at 12mbit/s
would be calculated but matched against high-speed quotas.

Reinitialise hcd->frame_usecs[i] on each port enable event so that
full-speed bandwidth can be tracked sensibly.

Also, don't bother using the FIQ for transfers when in full-speed
mode - at the slower bus speed, interrupt frequency is reduced by
an order of magnitude.

Related issue: #2020

dwc_otg: fiq_fsm: Make isochronous compatibility checks work properly

Get rid of the spammy printk and local pointer mangling.
Also, there is a nominal benefit for using fiq_fsm for isochronous
transfers in FS mode (~1.1k IRQs per second vs 2.1k IRQs per second)
so remove the root port speed check.

dwc_otg: add module parameter int_ep_interval_min

Add a module parameter (defaulting to ignored) that clamps the polling rate
of high-speed Interrupt endpoints to a minimum microframe interval.

The parameter is modifiable at runtime as it is used when activating new
endpoints (such as on device connect).

dwc_otg: fiq_fsm: Add non-periodic TT exclusivity constraints

Certain hub types do not discriminate between pipe direction (IN or OUT)
when considering non-periodic transfers. Therefore these hubs get confused
if multiple transfers are issued in different directions with the same
device address and endpoint number.

Constrain queuing non-periodic split transactions so they are performed
serially in such cases.

Related: #2024

dwc_otg: Fixup change to DRIVER_ATTR interface

dwc_otg: Fix compilation warnings

Signed-off-by: Phil Elwell <phil@raspberrypi.org>

USB_DWCOTG: Disable building dwc_otg as a module (#2265)

When dwc_otg is built as a module, build will fail with the following
error:

ERROR: "DWC_TASK_HI_SCHEDULE" [drivers/usb/host/dwc_otg/dwc_otg.ko] undefined!
scripts/Makefile.modpost:91: recipe for target '__modpost' failed
make[1]: *** [__modpost] Error 1
Makefile:1199: recipe for target 'modules' failed
make: *** [modules] Error 2

Even if the error is solved by including the missing
DWC_TASK_HI_SCHEDULE function, the kernel will panic when loading
dwc_otg.

As a workaround, simply prevent user from building dwc_otg as a module
as the current kernel does not support it.

See: #2258

Signed-off-by: Malik Olivier Boussejra <malik@boussejra.com>

dwc_otg: New timer API

dwc_otg: Fix removed ACCESS_ONCE->READ_ONCE

dwc_otg: don't unconditionally force host mode in dwc_otg_cil_init()

Add the ability to disable force_host_mode for those that want to use
dwc_otg in both device and host modes.

dwc_otg: Fix a regression when dequeueing isochronous transfers

In 282bed9 (dwc_otg: make nak_holdoff work as intended with empty queues)
the dequeue mechanism was changed to leave FIQ-enabled transfers to run
to completion - to avoid leaving hub TT buffers with stale packets lying
around.

This broke FIQ-accelerated isochronous transfers, as this then meant that
dozens of transfers were performed after the dequeue function returned.

Restore the state machine fence for isochronous transfers.

fiq_fsm: rewind DMA pointer for OUT transactions that fail (#2288)

See: #2140

dwc_otg: add smp_mb() to prevent driver state corruption on boot

Occasional crashes have been seen where the FIQ code dereferences
invalid/random pointers immediately after being set up, leading to
panic on boot.

The crash occurs as the FIQ code races against hcd_init_fiq() and
the hcd_init_fiq() code races against the outstanding memory stores
from dwc_otg_hcd_init(). Use explicit barriers after touching
driver state.

usb: dwc_otg: fix memory corruption in dwc_otg driver

[Upstream commit 51b1b64]

The move from the staging tree to the main tree exposed a
longstanding memory corruption bug in the dwc2 driver. The
reordering of the driver initialization caused the dwc2 driver
to corrupt the initialization data of the sdhci driver on the
Raspberry Pi platform, which made the bug show up.

The error is in calling to_usb_device(hsotg->dev), since ->dev
is not a member of struct usb_device. The easiest fix is to
just remove the offending code, since it is not really needed.

Thanks to Stephen Warren for tracking down the cause of this.

Reported-by: Andre Heider <a.heider@gmail.com>
Tested-by: Stephen Warren <swarren@wwwdotorg.org>
Signed-off-by: Paul Zimmerman <paulz@synopsys.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[lukas: port from upstream dwc2 to out-of-tree dwc_otg driver]
Signed-off-by: Lukas Wunner <lukas@wunner.de>

usb: dwb_otg: Fix unreachable switch statement warning

This warning appears with GCC 7.3.0 from toolchains.bootlin.com:

../drivers/usb/host/dwc_otg/dwc_otg_fiq_fsm.c: In function ‘fiq_fsm_update_hs_isoc’:
../drivers/usb/host/dwc_otg/dwc_otg_fiq_fsm.c:595:61: warning: statement will never be executed [-Wswitch-unreachable]
   st->hctsiz_copy.b.xfersize = nrpackets * st->hcchar_copy.b.mps;
                                            ~~~~~~~~~~~~~~~~~^~~~

Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>

dwc_otg: fiq_fsm: fix incorrect DMA register offset calculation

Rationalise the offset and update all call sites.

Fixes #2408

popcornmix added a commit that referenced this issue Feb 24, 2019

Add dwc_otg driver
Signed-off-by: popcornmix <popcornmix@gmail.com>

usb: dwc: fix lockdep false positive

Signed-off-by: Kari Suvanto <karis79@gmail.com>

usb: dwc: fix inconsistent lock state

Signed-off-by: Kari Suvanto <karis79@gmail.com>

Add FIQ patch to dwc_otg driver. Enable with dwc_otg.fiq_fix_enable=1. Should give about 10% more ARM performance.
Thanks to Gordon and Costas

Avoid dynamic memory allocation for channel lock in USB driver. Thanks ddv2005.

Add NAK holdoff scheme. Enabled by default, disable with dwc_otg.nak_holdoff_enable=0. Thanks gsh

Make sure we wait for the reset to finish

dwc_otg: fix bug in dwc_otg_hcd.c resulting in silent kernel
	 memory corruption, escalating to OOPS under high USB load.

dwc_otg: Fix unsafe access of QTD during URB enqueue

In dwc_otg_hcd_urb_enqueue during qtd creation, it was possible that the
transaction could complete almost immediately after the qtd was assigned
to a host channel during URB enqueue, which meant the qtd pointer was no
longer valid having been completed and removed. Usually, this resulted in
an OOPS during URB submission. By predetermining whether transactions
need to be queued or not, this unsafe pointer access is avoided.

This bug was only evident on the Pi model A where a device was attached
that had no periodic endpoints (e.g. USB pendrive or some wlan devices).

dwc_otg: Fix incorrect URB allocation error handling

If the memory allocation for a dwc_otg_urb failed, the kernel would OOPS
because for some reason a member of the *unallocated* struct was set to
zero. Error handling changed to fail correctly.

dwc_otg: fix potential use-after-free case in interrupt handler

If a transaction had previously aborted, certain interrupts are
enabled to track error counts and reset where necessary. On IN
endpoints the host generates an ACK interrupt near-simultaneously
with completion of transfer. In the case where this transfer had
previously had an error, this results in a use-after-free on
the QTD memory space with a 1-byte length being overwritten to
0x00.

dwc_otg: add handling of SPLIT transaction data toggle errors

Previously a data toggle error on packets from a USB1.1 device behind
a TT would result in the Pi locking up as the driver never handled
the associated interrupt. Patch adds basic retry mechanism and
interrupt acknowledgement to cater for either a chance toggle error or
for devices that have a broken initial toggle state (FT8U232/FT232BM).

dwc_otg: implement tasklet for returning URBs to usbcore hcd layer

The dwc_otg driver interrupt handler for transfer completion will spend
a very long time with interrupts disabled when a URB is completed -
this is because usb_hcd_giveback_urb is called from within the handler
which for a USB device driver with complicated processing (e.g. webcam)
will take an exorbitant amount of time to complete. This results in
missed completion interrupts for other USB packets which lead to them
being dropped due to microframe overruns.

This patch splits returning the URB to the usb hcd layer into a
high-priority tasklet. This will have most benefit for isochronous IN
transfers but will also have incidental benefit where multiple periodic
devices are active at once.

dwc_otg: fix NAK holdoff and allow on split transactions only

This corrects a bug where if a single active non-periodic endpoint
had at least one transaction in its qh, on frnum == MAX_FRNUM the qh
would get skipped and never get queued again. This would result in
a silent device until error detection (automatic or otherwise) would
either reset the device or flush and requeue the URBs.

Additionally the NAK holdoff was enabled for all transactions - this
would potentially stall a HS endpoint for 1ms if a previous error state
enabled this interrupt and the next response was a NAK. Fix so that
only split transactions get held off.

dwc_otg: Call usb_hcd_unlink_urb_from_ep with lock held in completion handler

usb_hcd_unlink_urb_from_ep must be called with the HCD lock held.  Calling it
asynchronously in the tasklet was not safe (regression in
c4564d4).

This change unlinks it from the endpoint prior to queueing it for handling in
the tasklet, and also adds a check to ensure the urb is OK to be unlinked
before doing so.

NULL pointer dereference kernel oopses had been observed in usb_hcd_giveback_urb
when a USB device was unplugged/replugged during data transfer.  This effect
was reproduced using automated USB port power control, hundreds of replug
events were performed during active transfers to confirm that the problem was
eliminated.

USB fix using a FIQ to implement split transactions

This commit adds a FIQ implementaion that schedules
the split transactions using a FIQ so we don't get
held off by the interrupt latency of Linux

dwc_otg: fix device attributes and avoid kernel warnings on boot

dcw_otg: avoid logging function that can cause panics

See: raspberrypi/firmware#21
Thanks to cleverca22 for fix

dwc_otg: mask correct interrupts after transaction error recovery

The dwc_otg driver will unmask certain interrupts on a transaction
that previously halted in the error state in order to reset the
QTD error count. The various fine-grained interrupt handlers do not
consider that other interrupts besides themselves were unmasked.

By disabling the two other interrupts only ever enabled in DMA mode
for this purpose, we can avoid unnecessary function calls in the
IRQ handler. This will also prevent an unneccesary FIQ interrupt
from being generated if the FIQ is enabled.

dwc_otg: fiq: prevent FIQ thrash and incorrect state passing to IRQ

In the case of a transaction to a device that had previously aborted
due to an error, several interrupts are enabled to reset the error
count when a device responds. This has the side-effect of making the
FIQ thrash because the hardware will generate multiple instances of
a NAK on an IN bulk/interrupt endpoint and multiple instances of ACK
on an OUT bulk/interrupt endpoint. Make the FIQ mask and clear the
associated interrupts.

Additionally, on non-split transactions make sure that only unmasked
interrupts are cleared. This caused a hard-to-trigger but serious
race condition when you had the combination of an endpoint awaiting
error recovery and a transaction completed on an endpoint - due to
the sequencing and timing of interrupts generated by the dwc_otg core,
it was possible to confuse the IRQ handler.

Fix function tracing

dwc_otg: whitespace cleanup in dwc_otg_urb_enqueue

dwc_otg: prevent OOPSes during device disconnects

The dwc_otg_urb_enqueue function is thread-unsafe. In particular the
access of urb->hcpriv, usb_hcd_link_urb_to_ep, dwc_otg_urb->qtd and
friends does not occur within a critical section and so if a device
was unplugged during activity there was a high chance that the
usbcore hub_thread would try to disable the endpoint with partially-
formed entries in the URB queue. This would result in BUG() or null
pointer dereferences.

Fix so that access of urb->hcpriv, enqueuing to the hardware and
adding to usbcore endpoint URB lists is contained within a single
critical section.

dwc_otg: prevent BUG() in TT allocation if hub address is > 16

A fixed-size array is used to track TT allocation. This was
previously set to 16 which caused a crash because
dwc_otg_hcd_allocate_port would read past the end of the array.

This was hit if a hub was plugged in which enumerated as addr > 16,
due to previous device resets or unplugs.

Also add #ifdef FIQ_DEBUG around hcd->hub_port_alloc[], which grows
to a large size if 128 hub addresses are supported. This field is
for debug only for tracking which frame an allocate happened in.

dwc_otg: make channel halts with unknown state less damaging

If the IRQ received a channel halt interrupt through the FIQ
with no other bits set, the IRQ would not release the host
channel and never complete the URB.

Add catchall handling to treat as a transaction error and retry.

dwc_otg: fiq_split: use TTs with more granularity

This fixes certain issues with split transaction scheduling.

- Isochronous multi-packet OUT transactions now hog the TT until
  they are completed - this prevents hubs aborting transactions
  if they get a periodic start-split out-of-order
- Don't perform TT allocation on non-periodic endpoints - this
  allows simultaneous use of the TT's bulk/control and periodic
  transaction buffers

This commit will mainly affect USB audio playback.

dwc_otg: fix potential sleep while atomic during urb enqueue

Fixes a regression introduced with eb1b482. Kmalloc called from
dwc_otg_hcd_qtd_add / dwc_otg_hcd_qtd_create did not always have
the GPF_ATOMIC flag set. Force this flag when inside the larger
critical section.

dwc_otg: make fiq_split_enable imply fiq_fix_enable

Failing to set up the FIQ correctly would result in
"IRQ 32: nobody cared" errors in dmesg.

dwc_otg: prevent crashes on host port disconnects

Fix several issues resulting in crashes or inconsistent state
if a Model A root port was disconnected.

- Clean up queue heads properly in kill_urbs_in_qh_list by
  removing the empty QHs from the schedule lists
- Set the halt status properly to prevent IRQ handlers from
  using freed memory
- Add fiq_split related cleanup for saved registers
- Make microframe scheduling reclaim host channels if
  active during a disconnect
- Abort URBs with -ESHUTDOWN status response, informing
  device drivers so they respond in a more correct fashion
  and don't try to resubmit URBs
- Prevent IRQ handlers from attempting to handle channel
  interrupts if the associated URB was dequeued (and the
  driver state was cleared)

dwc_otg: prevent leaking URBs during enqueue

A dwc_otg_urb would get leaked if the HCD enqueue function
failed for any reason. Free the URB at the appropriate points.

dwc_otg: Enable NAK holdoff for control split transactions

Certain low-speed devices take a very long time to complete a
data or status stage of a control transaction, producing NAK
responses until they complete internal processing - the USB2.0
spec limit is up to 500mS. This causes the same type of interrupt
storm as seen with USB-serial dongles prior to c8edb23.

In certain circumstances, usually while booting, this interrupt
storm could cause SD card timeouts.

dwc_otg: Fix for occasional lockup on boot when doing a USB reset

dwc_otg: Don't issue traffic to LS devices in FS mode

Issuing low-speed packets when the root port is in full-speed mode
causes the root port to stop responding. Explicitly fail when
enqueuing URBs to a LS endpoint on a FS bus.

Fix ARM architecture issue with local_irq_restore()

If local_fiq_enable() is called before a local_irq_restore(flags) where
the flags variable has the F bit set, the FIQ will be erroneously disabled.

Fixup arch_local_irq_restore to avoid trampling the F bit in CPSR.

Also fix some of the hacks previously implemented for previous dwc_otg
incarnations.

dwc_otg: fiq_fsm: Base commit for driver rewrite

This commit removes the previous FIQ fixes entirely and adds fiq_fsm.

This rewrite features much more complete support for split transactions
and takes into account several OTG hardware bugs. High-speed
isochronous transactions are also capable of being performed by fiq_fsm.

All driver options have been removed and replaced with:
  - dwc_otg.fiq_enable (bool)
  - dwc_otg.fiq_fsm_enable (bool)
  - dwc_otg.fiq_fsm_mask (bitmask)
  - dwc_otg.nak_holdoff (unsigned int)

Defaults are specified such that fiq_fsm behaves similarly to the
previously implemented FIQ fixes.

fiq_fsm: Push error recovery into the FIQ when fiq_fsm is used

If the transfer associated with a QTD failed due to a bus error, the HCD
would retry the transfer up to 3 times (implementing the USB2.0
three-strikes retry in software).

Due to the masking mechanism used by fiq_fsm, it is only possible to pass
a single interrupt through to the HCD per-transfer.

In this instance host channels would fall off the radar because the error
reset would function, but the subsequent channel halt would be lost.

Push the error count reset into the FIQ handler.

fiq_fsm: Implement timeout mechanism

For full-speed endpoints with a large packet size, interrupt latency
runs the risk of the FIQ starting a transaction too late in a full-speed
frame. If the device is still transmitting data when EOF2 for the
downstream frame occurs, the hub will disable the port. This change is
not reflected in the hub status endpoint and the device becomes
unresponsive.

Prevent high-bandwidth transactions from being started too late in a
frame. The mechanism is not guaranteed: a combination of bit stuffing
and hub latency may still result in a device overrunning.

fiq_fsm: fix bounce buffer utilisation for Isochronous OUT

Multi-packet isochronous OUT transactions were subject to a few bounday
bugs. Fix them.

Audio playback is now much more robust: however, an issue stands with
devices that have adaptive sinks - ALSA plays samples too fast.

dwc_otg: Return full-speed frame numbers in HS mode

The frame counter increments on every *microframe* in high-speed mode.
Most device drivers expect this number to be in full-speed frames - this
caused considerable confusion to e.g. snd_usb_audio which uses the
frame counter to estimate the number of samples played.

fiq_fsm: save PID on completion of interrupt OUT transfers

Also add edge case handling for interrupt transports.

Note that for periodic split IN, data toggles are unimplemented in the
OTG host hardware - it unconditionally accepts any PID.

fiq_fsm: add missing case for fiq_fsm_tt_in_use()

Certain combinations of bitrate and endpoint activity could
result in a periodic transaction erroneously getting started
while the previous Isochronous OUT was still active.

fiq_fsm: clear hcintmsk for aborted transactions

Prevents the FIQ from erroneously handling interrupts
on a timed out channel.

fiq_fsm: enable by default

fiq_fsm: fix dequeues for non-periodic split transactions

If a dequeue happened between the SSPLIT and CSPLIT phases of the
transaction, the HCD would never receive an interrupt.

fiq_fsm: Disable by default

fiq_fsm: Handle HC babble errors

The HCTSIZ transfer size field raises a babble interrupt if
the counter wraps. Handle the resulting interrupt in this case.

dwc_otg: fix interrupt registration for fiq_enable=0

Additionally make the module parameter conditional for wherever
hcd->fiq_state is touched.

fiq_fsm: Enable by default

dwc_otg: Fix various issues with root port and transaction errors

Process the host port interrupts correctly (and don't trample them).
Root port hotplug now functional again.

Fix a few thinkos with the transaction error passthrough for fiq_fsm.

fiq_fsm: Implement hack for Split Interrupt transactions

Hubs aren't too picky about which endpoint we send Control type split
transactions to. By treating Interrupt transfers as Control, it is
possible to use the non-periodic queue in the OTG core as well as the
non-periodic FIFOs in the hub itself. This massively reduces the
microframe exclusivity/contention that periodic split transactions
otherwise have to enforce.

It goes without saying that this is a fairly egregious USB specification
violation, but it works.

Original idea by Hans Petter Selasky @ FreeBSD.org.

dwc_otg: FIQ support on SMP. Set up FIQ stack and handler on Core 0 only.

dwc_otg: introduce fiq_fsm_spin(un|)lock()

SMP safety for the FIQ relies on register read-modify write cycles being
completed in the correct order. Several places in the DWC code modify
registers also touched by the FIQ. Protect these by a bare-bones lock
mechanism.

This also makes it possible to run the FIQ and IRQ handlers on different
cores.

fiq_fsm: fix build on bcm2708 and bcm2709 platforms

dwc_otg: put some barriers back where they should be for UP

bcm2709/dwc_otg: Setup FIQ on core 1 if >1 core active

dwc_otg: fixup read-modify-write in critical paths

Be more careful about read-modify-write on registers that the FIQ
also touches.

Guard fiq_fsm_spin_lock with fiq_enable check

fiq_fsm: Falling out of the state machine isn't fatal

This edge case can be hit if the port is disabled while the FIQ is
in the middle of a transaction. Make the effects less severe.

Also get rid of the useless return value.

squash: dwc_otg: Allow to build without SMP

usb: core: make overcurrent messages more prominent

Hub overcurrent messages are more serious than "debug". Increase loglevel.

usb: dwc_otg: Don't use dma_to_virt()

Commit 6ce0d20 changes dma_to_virt() which breaks this driver.
Open code the old dma_to_virt() implementation to work around this.

Limit the use of __bus_to_virt() to cases where transfer_buffer_length
is set and transfer_buffer is not set. This is done to increase the
chance that this driver will also work on ARCH_BCM2835.

transfer_buffer should not be NULL if the length is set, but the
comment in the code indicates that there are situations where this
might happen. drivers/usb/isp1760/isp1760-hcd.c also has a similar
comment pointing to a possible: 'usb storage / SCSI bug'.

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>

dwc_otg: Fix crash when fiq_enable=0

dwc_otg: fiq_fsm: Make high-speed isochronous strided transfers work properly

Certain low-bandwidth high-speed USB devices (specialist audio devices,
compressed-frame webcams) have packet intervals > 1 microframe.

Stride these transfers in the FIQ by using the start-of-frame interrupt
to restart the channel at the right time.

dwc_otg: Force host mode to fix incorrect compute module boards

dwc_otg: Add ARCH_BCM2835 support

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>

dwc_otg: Simplify FIQ irq number code

Dropping ATAGS means we can simplify the FIQ irq number code.
Also add error checking on the returned irq number.

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>

dwc_otg: Remove duplicate gadget probe/unregister function

dwc_otg: Properly set the HFIR

Douglas Anderson reported:

According to the most up to date version of the dwc2 databook, the FRINT
field of the HFIR register should be programmed to:
* 125 us * (PHY clock freq for HS) - 1
* 1000 us * (PHY clock freq for FS/LS) - 1

This is opposed to older versions of the doc that claimed it should be:
* 125 us * (PHY clock freq for HS)
* 1000 us * (PHY clock freq for FS/LS)

and reported lower timing jitter on a USB analyser

dcw_otg: trim xfer length when buffer larger than allocated size is received

dwc_otg: Don't free qh align buffers in atomic context

dwc_otg: Enable the hack for Split Interrupt transactions by default

dwc_otg.fiq_fsm_mask=0xF has long been a suggestion for users with audio stutters or other USB bandwidth issues.
So far we are aware of many success stories but no failure caused by this setting.
Make it a default to learn more.

See: https://www.raspberrypi.org/forums/viewtopic.php?f=28&t=70437

Signed-off-by: popcornmix <popcornmix@gmail.com>

dwc_otg: Use kzalloc when suitable

dwc_otg: Pass struct device to dma_alloc*()

This makes it possible to get the bus address from Device Tree.

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>

dwc_otg: fix summarize urb->actual_length for isochronous transfers

Kernel does not copy input data of ISO transfers to userspace
if actual_length is set only in ISO transfers and not summarized
in urb->actual_length. Fixes #903

fiq_fsm: Use correct states when starting isoc OUT transfers

In fiq_fsm_start_next_periodic() if an isochronous OUT transfer
was selected, no regard was given as to whether this was a single-packet
transfer or a multi-packet staged transfer.

For single-packet transfers, this had the effect of repeatedly sending
OUT packets with bogus data and lengths.

Eventually if the channel was repeatedly enabled enough times, this
would lock up the OTG core and no further bus transfers would happen.

Set the FSM state up properly if we select a single-packet transfer.

Fixes #1842

dwc_otg: make nak_holdoff work as intended with empty queues

If URBs reading from non-periodic split endpoints were dequeued and
the last transfer from the endpoint was a NAK handshake, the resulting
qh->nak_frame value was stale which would result in unnecessarily long
polling intervals for the first subsequent transfer with a fresh URB.

Fixup qh->nak_frame in dwc_otg_hcd_urb_dequeue and also guard against
a case where a single URB is submitted to the endpoint, a NAK was
received on the transfer immediately prior to receiving data and the
device subsequently resubmits another URB past the qh->nak_frame interval.

Fixes #1709

dwc_otg: fix split transaction data toggle handling around dequeues

See #1709

Fix several issues regarding endpoint state when URBs are dequeued
- If the HCD is disconnected, flush FIQ-enabled channels properly
- Save the data toggle state for bulk endpoints if the last transfer
  from an endpoint where URBs were dequeued returned a data packet
- Reset hc->start_pkt_count properly in assign_and_init_hc()

dwc_otg: fix several potential crash sources

On root port disconnect events, the host driver state is cleared and
in-progress host channels are forcibly stopped. This doesn't play
well with the FIQ running in the background, so:
- Guard the disconnect callback with both the host spinlock and FIQ
  spinlock
- Move qtd dereference in dwc_otg_handle_hc_fsm() after the early-out
  so we don't dereference a qtd that has gone away
- Turn catch-all BUG()s in dwc_otg_handle_hc_fsm() into warnings.

dwc_otg: delete hcd->channel_lock

The lock serves no purpose as it is only held while the HCD spinlock
is already being held.

dwc_otg: remove unnecessary dma-mode channel halts on disconnect interrupt

Host channels are already halted in kill_urbs_in_qh_list() with the
subsequent interrupt processing behaving as if the URB was dequeued
via HCD callback.

There's no need to clobber the host channel registers a second time
as this exposes races between the driver and host channel resulting
in hcd->free_hc_list becoming corrupted.

dwcotg: Allow to build without FIQ on ARM64

Signed-off-by: popcornmix <popcornmix@gmail.com>

dwc_otg: make periodic scheduling behave properly for FS buses

If the root port is in full-speed mode, transfer times at 12mbit/s
would be calculated but matched against high-speed quotas.

Reinitialise hcd->frame_usecs[i] on each port enable event so that
full-speed bandwidth can be tracked sensibly.

Also, don't bother using the FIQ for transfers when in full-speed
mode - at the slower bus speed, interrupt frequency is reduced by
an order of magnitude.

Related issue: #2020

dwc_otg: fiq_fsm: Make isochronous compatibility checks work properly

Get rid of the spammy printk and local pointer mangling.
Also, there is a nominal benefit for using fiq_fsm for isochronous
transfers in FS mode (~1.1k IRQs per second vs 2.1k IRQs per second)
so remove the root port speed check.

dwc_otg: add module parameter int_ep_interval_min

Add a module parameter (defaulting to ignored) that clamps the polling rate
of high-speed Interrupt endpoints to a minimum microframe interval.

The parameter is modifiable at runtime as it is used when activating new
endpoints (such as on device connect).

dwc_otg: fiq_fsm: Add non-periodic TT exclusivity constraints

Certain hub types do not discriminate between pipe direction (IN or OUT)
when considering non-periodic transfers. Therefore these hubs get confused
if multiple transfers are issued in different directions with the same
device address and endpoint number.

Constrain queuing non-periodic split transactions so they are performed
serially in such cases.

Related: #2024

dwc_otg: Fixup change to DRIVER_ATTR interface

dwc_otg: Fix compilation warnings

Signed-off-by: Phil Elwell <phil@raspberrypi.org>

USB_DWCOTG: Disable building dwc_otg as a module (#2265)

When dwc_otg is built as a module, build will fail with the following
error:

ERROR: "DWC_TASK_HI_SCHEDULE" [drivers/usb/host/dwc_otg/dwc_otg.ko] undefined!
scripts/Makefile.modpost:91: recipe for target '__modpost' failed
make[1]: *** [__modpost] Error 1
Makefile:1199: recipe for target 'modules' failed
make: *** [modules] Error 2

Even if the error is solved by including the missing
DWC_TASK_HI_SCHEDULE function, the kernel will panic when loading
dwc_otg.

As a workaround, simply prevent user from building dwc_otg as a module
as the current kernel does not support it.

See: #2258

Signed-off-by: Malik Olivier Boussejra <malik@boussejra.com>

dwc_otg: New timer API

dwc_otg: Fix removed ACCESS_ONCE->READ_ONCE

dwc_otg: don't unconditionally force host mode in dwc_otg_cil_init()

Add the ability to disable force_host_mode for those that want to use
dwc_otg in both device and host modes.

dwc_otg: Fix a regression when dequeueing isochronous transfers

In 282bed9 (dwc_otg: make nak_holdoff work as intended with empty queues)
the dequeue mechanism was changed to leave FIQ-enabled transfers to run
to completion - to avoid leaving hub TT buffers with stale packets lying
around.

This broke FIQ-accelerated isochronous transfers, as this then meant that
dozens of transfers were performed after the dequeue function returned.

Restore the state machine fence for isochronous transfers.

fiq_fsm: rewind DMA pointer for OUT transactions that fail (#2288)

See: #2140

dwc_otg: add smp_mb() to prevent driver state corruption on boot

Occasional crashes have been seen where the FIQ code dereferences
invalid/random pointers immediately after being set up, leading to
panic on boot.

The crash occurs as the FIQ code races against hcd_init_fiq() and
the hcd_init_fiq() code races against the outstanding memory stores
from dwc_otg_hcd_init(). Use explicit barriers after touching
driver state.

usb: dwc_otg: fix memory corruption in dwc_otg driver

[Upstream commit 51b1b64]

The move from the staging tree to the main tree exposed a
longstanding memory corruption bug in the dwc2 driver. The
reordering of the driver initialization caused the dwc2 driver
to corrupt the initialization data of the sdhci driver on the
Raspberry Pi platform, which made the bug show up.

The error is in calling to_usb_device(hsotg->dev), since ->dev
is not a member of struct usb_device. The easiest fix is to
just remove the offending code, since it is not really needed.

Thanks to Stephen Warren for tracking down the cause of this.

Reported-by: Andre Heider <a.heider@gmail.com>
Tested-by: Stephen Warren <swarren@wwwdotorg.org>
Signed-off-by: Paul Zimmerman <paulz@synopsys.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[lukas: port from upstream dwc2 to out-of-tree dwc_otg driver]
Signed-off-by: Lukas Wunner <lukas@wunner.de>

usb: dwb_otg: Fix unreachable switch statement warning

This warning appears with GCC 7.3.0 from toolchains.bootlin.com:

../drivers/usb/host/dwc_otg/dwc_otg_fiq_fsm.c: In function ‘fiq_fsm_update_hs_isoc’:
../drivers/usb/host/dwc_otg/dwc_otg_fiq_fsm.c:595:61: warning: statement will never be executed [-Wswitch-unreachable]
   st->hctsiz_copy.b.xfersize = nrpackets * st->hcchar_copy.b.mps;
                                            ~~~~~~~~~~~~~~~~~^~~~

Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>

dwc_otg: fiq_fsm: fix incorrect DMA register offset calculation

Rationalise the offset and update all call sites.

Fixes #2408

popcornmix added a commit that referenced this issue Feb 26, 2019

Add dwc_otg driver
Signed-off-by: popcornmix <popcornmix@gmail.com>

usb: dwc: fix lockdep false positive

Signed-off-by: Kari Suvanto <karis79@gmail.com>

usb: dwc: fix inconsistent lock state

Signed-off-by: Kari Suvanto <karis79@gmail.com>

Add FIQ patch to dwc_otg driver. Enable with dwc_otg.fiq_fix_enable=1. Should give about 10% more ARM performance.
Thanks to Gordon and Costas

Avoid dynamic memory allocation for channel lock in USB driver. Thanks ddv2005.

Add NAK holdoff scheme. Enabled by default, disable with dwc_otg.nak_holdoff_enable=0. Thanks gsh

Make sure we wait for the reset to finish

dwc_otg: fix bug in dwc_otg_hcd.c resulting in silent kernel
	 memory corruption, escalating to OOPS under high USB load.

dwc_otg: Fix unsafe access of QTD during URB enqueue

In dwc_otg_hcd_urb_enqueue during qtd creation, it was possible that the
transaction could complete almost immediately after the qtd was assigned
to a host channel during URB enqueue, which meant the qtd pointer was no
longer valid having been completed and removed. Usually, this resulted in
an OOPS during URB submission. By predetermining whether transactions
need to be queued or not, this unsafe pointer access is avoided.

This bug was only evident on the Pi model A where a device was attached
that had no periodic endpoints (e.g. USB pendrive or some wlan devices).

dwc_otg: Fix incorrect URB allocation error handling

If the memory allocation for a dwc_otg_urb failed, the kernel would OOPS
because for some reason a member of the *unallocated* struct was set to
zero. Error handling changed to fail correctly.

dwc_otg: fix potential use-after-free case in interrupt handler

If a transaction had previously aborted, certain interrupts are
enabled to track error counts and reset where necessary. On IN
endpoints the host generates an ACK interrupt near-simultaneously
with completion of transfer. In the case where this transfer had
previously had an error, this results in a use-after-free on
the QTD memory space with a 1-byte length being overwritten to
0x00.

dwc_otg: add handling of SPLIT transaction data toggle errors

Previously a data toggle error on packets from a USB1.1 device behind
a TT would result in the Pi locking up as the driver never handled
the associated interrupt. Patch adds basic retry mechanism and
interrupt acknowledgement to cater for either a chance toggle error or
for devices that have a broken initial toggle state (FT8U232/FT232BM).

dwc_otg: implement tasklet for returning URBs to usbcore hcd layer

The dwc_otg driver interrupt handler for transfer completion will spend
a very long time with interrupts disabled when a URB is completed -
this is because usb_hcd_giveback_urb is called from within the handler
which for a USB device driver with complicated processing (e.g. webcam)
will take an exorbitant amount of time to complete. This results in
missed completion interrupts for other USB packets which lead to them
being dropped due to microframe overruns.

This patch splits returning the URB to the usb hcd layer into a
high-priority tasklet. This will have most benefit for isochronous IN
transfers but will also have incidental benefit where multiple periodic
devices are active at once.

dwc_otg: fix NAK holdoff and allow on split transactions only

This corrects a bug where if a single active non-periodic endpoint
had at least one transaction in its qh, on frnum == MAX_FRNUM the qh
would get skipped and never get queued again. This would result in
a silent device until error detection (automatic or otherwise) would
either reset the device or flush and requeue the URBs.

Additionally the NAK holdoff was enabled for all transactions - this
would potentially stall a HS endpoint for 1ms if a previous error state
enabled this interrupt and the next response was a NAK. Fix so that
only split transactions get held off.

dwc_otg: Call usb_hcd_unlink_urb_from_ep with lock held in completion handler

usb_hcd_unlink_urb_from_ep must be called with the HCD lock held.  Calling it
asynchronously in the tasklet was not safe (regression in
c4564d4).

This change unlinks it from the endpoint prior to queueing it for handling in
the tasklet, and also adds a check to ensure the urb is OK to be unlinked
before doing so.

NULL pointer dereference kernel oopses had been observed in usb_hcd_giveback_urb
when a USB device was unplugged/replugged during data transfer.  This effect
was reproduced using automated USB port power control, hundreds of replug
events were performed during active transfers to confirm that the problem was
eliminated.

USB fix using a FIQ to implement split transactions

This commit adds a FIQ implementaion that schedules
the split transactions using a FIQ so we don't get
held off by the interrupt latency of Linux

dwc_otg: fix device attributes and avoid kernel warnings on boot

dcw_otg: avoid logging function that can cause panics

See: raspberrypi/firmware#21
Thanks to cleverca22 for fix

dwc_otg: mask correct interrupts after transaction error recovery

The dwc_otg driver will unmask certain interrupts on a transaction
that previously halted in the error state in order to reset the
QTD error count. The various fine-grained interrupt handlers do not
consider that other interrupts besides themselves were unmasked.

By disabling the two other interrupts only ever enabled in DMA mode
for this purpose, we can avoid unnecessary function calls in the
IRQ handler. This will also prevent an unneccesary FIQ interrupt
from being generated if the FIQ is enabled.

dwc_otg: fiq: prevent FIQ thrash and incorrect state passing to IRQ

In the case of a transaction to a device that had previously aborted
due to an error, several interrupts are enabled to reset the error
count when a device responds. This has the side-effect of making the
FIQ thrash because the hardware will generate multiple instances of
a NAK on an IN bulk/interrupt endpoint and multiple instances of ACK
on an OUT bulk/interrupt endpoint. Make the FIQ mask and clear the
associated interrupts.

Additionally, on non-split transactions make sure that only unmasked
interrupts are cleared. This caused a hard-to-trigger but serious
race condition when you had the combination of an endpoint awaiting
error recovery and a transaction completed on an endpoint - due to
the sequencing and timing of interrupts generated by the dwc_otg core,
it was possible to confuse the IRQ handler.

Fix function tracing

dwc_otg: whitespace cleanup in dwc_otg_urb_enqueue

dwc_otg: prevent OOPSes during device disconnects

The dwc_otg_urb_enqueue function is thread-unsafe. In particular the
access of urb->hcpriv, usb_hcd_link_urb_to_ep, dwc_otg_urb->qtd and
friends does not occur within a critical section and so if a device
was unplugged during activity there was a high chance that the
usbcore hub_thread would try to disable the endpoint with partially-
formed entries in the URB queue. This would result in BUG() or null
pointer dereferences.

Fix so that access of urb->hcpriv, enqueuing to the hardware and
adding to usbcore endpoint URB lists is contained within a single
critical section.

dwc_otg: prevent BUG() in TT allocation if hub address is > 16

A fixed-size array is used to track TT allocation. This was
previously set to 16 which caused a crash because
dwc_otg_hcd_allocate_port would read past the end of the array.

This was hit if a hub was plugged in which enumerated as addr > 16,
due to previous device resets or unplugs.

Also add #ifdef FIQ_DEBUG around hcd->hub_port_alloc[], which grows
to a large size if 128 hub addresses are supported. This field is
for debug only for tracking which frame an allocate happened in.

dwc_otg: make channel halts with unknown state less damaging

If the IRQ received a channel halt interrupt through the FIQ
with no other bits set, the IRQ would not release the host
channel and never complete the URB.

Add catchall handling to treat as a transaction error and retry.

dwc_otg: fiq_split: use TTs with more granularity

This fixes certain issues with split transaction scheduling.

- Isochronous multi-packet OUT transactions now hog the TT until
  they are completed - this prevents hubs aborting transactions
  if they get a periodic start-split out-of-order
- Don't perform TT allocation on non-periodic endpoints - this
  allows simultaneous use of the TT's bulk/control and periodic
  transaction buffers

This commit will mainly affect USB audio playback.

dwc_otg: fix potential sleep while atomic during urb enqueue

Fixes a regression introduced with eb1b482. Kmalloc called from
dwc_otg_hcd_qtd_add / dwc_otg_hcd_qtd_create did not always have
the GPF_ATOMIC flag set. Force this flag when inside the larger
critical section.

dwc_otg: make fiq_split_enable imply fiq_fix_enable

Failing to set up the FIQ correctly would result in
"IRQ 32: nobody cared" errors in dmesg.

dwc_otg: prevent crashes on host port disconnects

Fix several issues resulting in crashes or inconsistent state
if a Model A root port was disconnected.

- Clean up queue heads properly in kill_urbs_in_qh_list by
  removing the empty QHs from the schedule lists
- Set the halt status properly to prevent IRQ handlers from
  using freed memory
- Add fiq_split related cleanup for saved registers
- Make microframe scheduling reclaim host channels if
  active during a disconnect
- Abort URBs with -ESHUTDOWN status response, informing
  device drivers so they respond in a more correct fashion
  and don't try to resubmit URBs
- Prevent IRQ handlers from attempting to handle channel
  interrupts if the associated URB was dequeued (and the
  driver state was cleared)

dwc_otg: prevent leaking URBs during enqueue

A dwc_otg_urb would get leaked if the HCD enqueue function
failed for any reason. Free the URB at the appropriate points.

dwc_otg: Enable NAK holdoff for control split transactions

Certain low-speed devices take a very long time to complete a
data or status stage of a control transaction, producing NAK
responses until they complete internal processing - the USB2.0
spec limit is up to 500mS. This causes the same type of interrupt
storm as seen with USB-serial dongles prior to c8edb23.

In certain circumstances, usually while booting, this interrupt
storm could cause SD card timeouts.

dwc_otg: Fix for occasional lockup on boot when doing a USB reset

dwc_otg: Don't issue traffic to LS devices in FS mode

Issuing low-speed packets when the root port is in full-speed mode
causes the root port to stop responding. Explicitly fail when
enqueuing URBs to a LS endpoint on a FS bus.

Fix ARM architecture issue with local_irq_restore()

If local_fiq_enable() is called before a local_irq_restore(flags) where
the flags variable has the F bit set, the FIQ will be erroneously disabled.

Fixup arch_local_irq_restore to avoid trampling the F bit in CPSR.

Also fix some of the hacks previously implemented for previous dwc_otg
incarnations.

dwc_otg: fiq_fsm: Base commit for driver rewrite

This commit removes the previous FIQ fixes entirely and adds fiq_fsm.

This rewrite features much more complete support for split transactions
and takes into account several OTG hardware bugs. High-speed
isochronous transactions are also capable of being performed by fiq_fsm.

All driver options have been removed and replaced with:
  - dwc_otg.fiq_enable (bool)
  - dwc_otg.fiq_fsm_enable (bool)
  - dwc_otg.fiq_fsm_mask (bitmask)
  - dwc_otg.nak_holdoff (unsigned int)

Defaults are specified such that fiq_fsm behaves similarly to the
previously implemented FIQ fixes.

fiq_fsm: Push error recovery into the FIQ when fiq_fsm is used

If the transfer associated with a QTD failed due to a bus error, the HCD
would retry the transfer up to 3 times (implementing the USB2.0
three-strikes retry in software).

Due to the masking mechanism used by fiq_fsm, it is only possible to pass
a single interrupt through to the HCD per-transfer.

In this instance host channels would fall off the radar because the error
reset would function, but the subsequent channel halt would be lost.

Push the error count reset into the FIQ handler.

fiq_fsm: Implement timeout mechanism

For full-speed endpoints with a large packet size, interrupt latency
runs the risk of the FIQ starting a transaction too late in a full-speed
frame. If the device is still transmitting data when EOF2 for the
downstream frame occurs, the hub will disable the port. This change is
not reflected in the hub status endpoint and the device becomes
unresponsive.

Prevent high-bandwidth transactions from being started too late in a
frame. The mechanism is not guaranteed: a combination of bit stuffing
and hub latency may still result in a device overrunning.

fiq_fsm: fix bounce buffer utilisation for Isochronous OUT

Multi-packet isochronous OUT transactions were subject to a few bounday
bugs. Fix them.

Audio playback is now much more robust: however, an issue stands with
devices that have adaptive sinks - ALSA plays samples too fast.

dwc_otg: Return full-speed frame numbers in HS mode

The frame counter increments on every *microframe* in high-speed mode.
Most device drivers expect this number to be in full-speed frames - this
caused considerable confusion to e.g. snd_usb_audio which uses the
frame counter to estimate the number of samples played.

fiq_fsm: save PID on completion of interrupt OUT transfers

Also add edge case handling for interrupt transports.

Note that for periodic split IN, data toggles are unimplemented in the
OTG host hardware - it unconditionally accepts any PID.

fiq_fsm: add missing case for fiq_fsm_tt_in_use()

Certain combinations of bitrate and endpoint activity could
result in a periodic transaction erroneously getting started
while the previous Isochronous OUT was still active.

fiq_fsm: clear hcintmsk for aborted transactions

Prevents the FIQ from erroneously handling interrupts
on a timed out channel.

fiq_fsm: enable by default

fiq_fsm: fix dequeues for non-periodic split transactions

If a dequeue happened between the SSPLIT and CSPLIT phases of the
transaction, the HCD would never receive an interrupt.

fiq_fsm: Disable by default

fiq_fsm: Handle HC babble errors

The HCTSIZ transfer size field raises a babble interrupt if
the counter wraps. Handle the resulting interrupt in this case.

dwc_otg: fix interrupt registration for fiq_enable=0

Additionally make the module parameter conditional for wherever
hcd->fiq_state is touched.

fiq_fsm: Enable by default

dwc_otg: Fix various issues with root port and transaction errors

Process the host port interrupts correctly (and don't trample them).
Root port hotplug now functional again.

Fix a few thinkos with the transaction error passthrough for fiq_fsm.

fiq_fsm: Implement hack for Split Interrupt transactions

Hubs aren't too picky about which endpoint we send Control type split
transactions to. By treating Interrupt transfers as Control, it is
possible to use the non-periodic queue in the OTG core as well as the
non-periodic FIFOs in the hub itself. This massively reduces the
microframe exclusivity/contention that periodic split transactions
otherwise have to enforce.

It goes without saying that this is a fairly egregious USB specification
violation, but it works.

Original idea by Hans Petter Selasky @ FreeBSD.org.

dwc_otg: FIQ support on SMP. Set up FIQ stack and handler on Core 0 only.

dwc_otg: introduce fiq_fsm_spin(un|)lock()

SMP safety for the FIQ relies on register read-modify write cycles being
completed in the correct order. Several places in the DWC code modify
registers also touched by the FIQ. Protect these by a bare-bones lock
mechanism.

This also makes it possible to run the FIQ and IRQ handlers on different
cores.

fiq_fsm: fix build on bcm2708 and bcm2709 platforms

dwc_otg: put some barriers back where they should be for UP

bcm2709/dwc_otg: Setup FIQ on core 1 if >1 core active

dwc_otg: fixup read-modify-write in critical paths

Be more careful about read-modify-write on registers that the FIQ
also touches.

Guard fiq_fsm_spin_lock with fiq_enable check

fiq_fsm: Falling out of the state machine isn't fatal

This edge case can be hit if the port is disabled while the FIQ is
in the middle of a transaction. Make the effects less severe.

Also get rid of the useless return value.

squash: dwc_otg: Allow to build without SMP

usb: core: make overcurrent messages more prominent

Hub overcurrent messages are more serious than "debug". Increase loglevel.

usb: dwc_otg: Don't use dma_to_virt()

Commit 6ce0d20 changes dma_to_virt() which breaks this driver.
Open code the old dma_to_virt() implementation to work around this.

Limit the use of __bus_to_virt() to cases where transfer_buffer_length
is set and transfer_buffer is not set. This is done to increase the
chance that this driver will also work on ARCH_BCM2835.

transfer_buffer should not be NULL if the length is set, but the
comment in the code indicates that there are situations where this
might happen. drivers/usb/isp1760/isp1760-hcd.c also has a similar
comment pointing to a possible: 'usb storage / SCSI bug'.

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>

dwc_otg: Fix crash when fiq_enable=0

dwc_otg: fiq_fsm: Make high-speed isochronous strided transfers work properly

Certain low-bandwidth high-speed USB devices (specialist audio devices,
compressed-frame webcams) have packet intervals > 1 microframe.

Stride these transfers in the FIQ by using the start-of-frame interrupt
to restart the channel at the right time.

dwc_otg: Force host mode to fix incorrect compute module boards

dwc_otg: Add ARCH_BCM2835 support

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>

dwc_otg: Simplify FIQ irq number code

Dropping ATAGS means we can simplify the FIQ irq number code.
Also add error checking on the returned irq number.

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>

dwc_otg: Remove duplicate gadget probe/unregister function

dwc_otg: Properly set the HFIR

Douglas Anderson reported:

According to the most up to date version of the dwc2 databook, the FRINT
field of the HFIR register should be programmed to:
* 125 us * (PHY clock freq for HS) - 1
* 1000 us * (PHY clock freq for FS/LS) - 1

This is opposed to older versions of the doc that claimed it should be:
* 125 us * (PHY clock freq for HS)
* 1000 us * (PHY clock freq for FS/LS)

and reported lower timing jitter on a USB analyser

dcw_otg: trim xfer length when buffer larger than allocated size is received

dwc_otg: Don't free qh align buffers in atomic context

dwc_otg: Enable the hack for Split Interrupt transactions by default

dwc_otg.fiq_fsm_mask=0xF has long been a suggestion for users with audio stutters or other USB bandwidth issues.
So far we are aware of many success stories but no failure caused by this setting.
Make it a default to learn more.

See: https://www.raspberrypi.org/forums/viewtopic.php?f=28&t=70437

Signed-off-by: popcornmix <popcornmix@gmail.com>

dwc_otg: Use kzalloc when suitable

dwc_otg: Pass struct device to dma_alloc*()

This makes it possible to get the bus address from Device Tree.

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>

dwc_otg: fix summarize urb->actual_length for isochronous transfers

Kernel does not copy input data of ISO transfers to userspace
if actual_length is set only in ISO transfers and not summarized
in urb->actual_length. Fixes #903

fiq_fsm: Use correct states when starting isoc OUT transfers

In fiq_fsm_start_next_periodic() if an isochronous OUT transfer
was selected, no regard was given as to whether this was a single-packet
transfer or a multi-packet staged transfer.

For single-packet transfers, this had the effect of repeatedly sending
OUT packets with bogus data and lengths.

Eventually if the channel was repeatedly enabled enough times, this
would lock up the OTG core and no further bus transfers would happen.

Set the FSM state up properly if we select a single-packet transfer.

Fixes #1842

dwc_otg: make nak_holdoff work as intended with empty queues

If URBs reading from non-periodic split endpoints were dequeued and
the last transfer from the endpoint was a NAK handshake, the resulting
qh->nak_frame value was stale which would result in unnecessarily long
polling intervals for the first subsequent transfer with a fresh URB.

Fixup qh->nak_frame in dwc_otg_hcd_urb_dequeue and also guard against
a case where a single URB is submitted to the endpoint, a NAK was
received on the transfer immediately prior to receiving data and the
device subsequently resubmits another URB past the qh->nak_frame interval.

Fixes #1709

dwc_otg: fix split transaction data toggle handling around dequeues

See #1709

Fix several issues regarding endpoint state when URBs are dequeued
- If the HCD is disconnected, flush FIQ-enabled channels properly
- Save the data toggle state for bulk endpoints if the last transfer
  from an endpoint where URBs were dequeued returned a data packet
- Reset hc->start_pkt_count properly in assign_and_init_hc()

dwc_otg: fix several potential crash sources

On root port disconnect events, the host driver state is cleared and
in-progress host channels are forcibly stopped. This doesn't play
well with the FIQ running in the background, so:
- Guard the disconnect callback with both the host spinlock and FIQ
  spinlock
- Move qtd dereference in dwc_otg_handle_hc_fsm() after the early-out
  so we don't dereference a qtd that has gone away
- Turn catch-all BUG()s in dwc_otg_handle_hc_fsm() into warnings.

dwc_otg: delete hcd->channel_lock

The lock serves no purpose as it is only held while the HCD spinlock
is already being held.

dwc_otg: remove unnecessary dma-mode channel halts on disconnect interrupt

Host channels are already halted in kill_urbs_in_qh_list() with the
subsequent interrupt processing behaving as if the URB was dequeued
via HCD callback.

There's no need to clobber the host channel registers a second time
as this exposes races between the driver and host channel resulting
in hcd->free_hc_list becoming corrupted.

dwcotg: Allow to build without FIQ on ARM64

Signed-off-by: popcornmix <popcornmix@gmail.com>

dwc_otg: make periodic scheduling behave properly for FS buses

If the root port is in full-speed mode, transfer times at 12mbit/s
would be calculated but matched against high-speed quotas.

Reinitialise hcd->frame_usecs[i] on each port enable event so that
full-speed bandwidth can be tracked sensibly.

Also, don't bother using the FIQ for transfers when in full-speed
mode - at the slower bus speed, interrupt frequency is reduced by
an order of magnitude.

Related issue: #2020

dwc_otg: fiq_fsm: Make isochronous compatibility checks work properly

Get rid of the spammy printk and local pointer mangling.
Also, there is a nominal benefit for using fiq_fsm for isochronous
transfers in FS mode (~1.1k IRQs per second vs 2.1k IRQs per second)
so remove the root port speed check.

dwc_otg: add module parameter int_ep_interval_min

Add a module parameter (defaulting to ignored) that clamps the polling rate
of high-speed Interrupt endpoints to a minimum microframe interval.

The parameter is modifiable at runtime as it is used when activating new
endpoints (such as on device connect).

dwc_otg: fiq_fsm: Add non-periodic TT exclusivity constraints

Certain hub types do not discriminate between pipe direction (IN or OUT)
when considering non-periodic transfers. Therefore these hubs get confused
if multiple transfers are issued in different directions with the same
device address and endpoint number.

Constrain queuing non-periodic split transactions so they are performed
serially in such cases.

Related: #2024

dwc_otg: Fixup change to DRIVER_ATTR interface

dwc_otg: Fix compilation warnings

Signed-off-by: Phil Elwell <phil@raspberrypi.org>

USB_DWCOTG: Disable building dwc_otg as a module (#2265)

When dwc_otg is built as a module, build will fail with the following
error:

ERROR: "DWC_TASK_HI_SCHEDULE" [drivers/usb/host/dwc_otg/dwc_otg.ko] undefined!
scripts/Makefile.modpost:91: recipe for target '__modpost' failed
make[1]: *** [__modpost] Error 1
Makefile:1199: recipe for target 'modules' failed
make: *** [modules] Error 2

Even if the error is solved by including the missing
DWC_TASK_HI_SCHEDULE function, the kernel will panic when loading
dwc_otg.

As a workaround, simply prevent user from building dwc_otg as a module
as the current kernel does not support it.

See: #2258

Signed-off-by: Malik Olivier Boussejra <malik@boussejra.com>

dwc_otg: New timer API

dwc_otg: Fix removed ACCESS_ONCE->READ_ONCE

dwc_otg: don't unconditionally force host mode in dwc_otg_cil_init()

Add the ability to disable force_host_mode for those that want to use
dwc_otg in both device and host modes.

dwc_otg: Fix a regression when dequeueing isochronous transfers

In 282bed9 (dwc_otg: make nak_holdoff work as intended with empty queues)
the dequeue mechanism was changed to leave FIQ-enabled transfers to run
to completion - to avoid leaving hub TT buffers with stale packets lying
around.

This broke FIQ-accelerated isochronous transfers, as this then meant that
dozens of transfers were performed after the dequeue function returned.

Restore the state machine fence for isochronous transfers.

fiq_fsm: rewind DMA pointer for OUT transactions that fail (#2288)

See: #2140

dwc_otg: add smp_mb() to prevent driver state corruption on boot

Occasional crashes have been seen where the FIQ code dereferences
invalid/random pointers immediately after being set up, leading to
panic on boot.

The crash occurs as the FIQ code races against hcd_init_fiq() and
the hcd_init_fiq() code races against the outstanding memory stores
from dwc_otg_hcd_init(). Use explicit barriers after touching
driver state.

usb: dwc_otg: fix memory corruption in dwc_otg driver

[Upstream commit 51b1b64]

The move from the staging tree to the main tree exposed a
longstanding memory corruption bug in the dwc2 driver. The
reordering of the driver initialization caused the dwc2 driver
to corrupt the initialization data of the sdhci driver on the
Raspberry Pi platform, which made the bug show up.

The error is in calling to_usb_device(hsotg->dev), since ->dev
is not a member of struct usb_device. The easiest fix is to
just remove the offending code, since it is not really needed.

Thanks to Stephen Warren for tracking down the cause of this.

Reported-by: Andre Heider <a.heider@gmail.com>
Tested-by: Stephen Warren <swarren@wwwdotorg.org>
Signed-off-by: Paul Zimmerman <paulz@synopsys.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[lukas: port from upstream dwc2 to out-of-tree dwc_otg driver]
Signed-off-by: Lukas Wunner <lukas@wunner.de>

usb: dwb_otg: Fix unreachable switch statement warning

This warning appears with GCC 7.3.0 from toolchains.bootlin.com:

../drivers/usb/host/dwc_otg/dwc_otg_fiq_fsm.c: In function ‘fiq_fsm_update_hs_isoc’:
../drivers/usb/host/dwc_otg/dwc_otg_fiq_fsm.c:595:61: warning: statement will never be executed [-Wswitch-unreachable]
   st->hctsiz_copy.b.xfersize = nrpackets * st->hcchar_copy.b.mps;
                                            ~~~~~~~~~~~~~~~~~^~~~

Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>

dwc_otg: fiq_fsm: fix incorrect DMA register offset calculation

Rationalise the offset and update all call sites.

Fixes #2408

popcornmix added a commit that referenced this issue Mar 6, 2019

Add dwc_otg driver
Signed-off-by: popcornmix <popcornmix@gmail.com>

usb: dwc: fix lockdep false positive

Signed-off-by: Kari Suvanto <karis79@gmail.com>

usb: dwc: fix inconsistent lock state

Signed-off-by: Kari Suvanto <karis79@gmail.com>

Add FIQ patch to dwc_otg driver. Enable with dwc_otg.fiq_fix_enable=1. Should give about 10% more ARM performance.
Thanks to Gordon and Costas

Avoid dynamic memory allocation for channel lock in USB driver. Thanks ddv2005.

Add NAK holdoff scheme. Enabled by default, disable with dwc_otg.nak_holdoff_enable=0. Thanks gsh

Make sure we wait for the reset to finish

dwc_otg: fix bug in dwc_otg_hcd.c resulting in silent kernel
	 memory corruption, escalating to OOPS under high USB load.

dwc_otg: Fix unsafe access of QTD during URB enqueue

In dwc_otg_hcd_urb_enqueue during qtd creation, it was possible that the
transaction could complete almost immediately after the qtd was assigned
to a host channel during URB enqueue, which meant the qtd pointer was no
longer valid having been completed and removed. Usually, this resulted in
an OOPS during URB submission. By predetermining whether transactions
need to be queued or not, this unsafe pointer access is avoided.

This bug was only evident on the Pi model A where a device was attached
that had no periodic endpoints (e.g. USB pendrive or some wlan devices).

dwc_otg: Fix incorrect URB allocation error handling

If the memory allocation for a dwc_otg_urb failed, the kernel would OOPS
because for some reason a member of the *unallocated* struct was set to
zero. Error handling changed to fail correctly.

dwc_otg: fix potential use-after-free case in interrupt handler

If a transaction had previously aborted, certain interrupts are
enabled to track error counts and reset where necessary. On IN
endpoints the host generates an ACK interrupt near-simultaneously
with completion of transfer. In the case where this transfer had
previously had an error, this results in a use-after-free on
the QTD memory space with a 1-byte length being overwritten to
0x00.

dwc_otg: add handling of SPLIT transaction data toggle errors

Previously a data toggle error on packets from a USB1.1 device behind
a TT would result in the Pi locking up as the driver never handled
the associated interrupt. Patch adds basic retry mechanism and
interrupt acknowledgement to cater for either a chance toggle error or
for devices that have a broken initial toggle state (FT8U232/FT232BM).

dwc_otg: implement tasklet for returning URBs to usbcore hcd layer

The dwc_otg driver interrupt handler for transfer completion will spend
a very long time with interrupts disabled when a URB is completed -
this is because usb_hcd_giveback_urb is called from within the handler
which for a USB device driver with complicated processing (e.g. webcam)
will take an exorbitant amount of time to complete. This results in
missed completion interrupts for other USB packets which lead to them
being dropped due to microframe overruns.

This patch splits returning the URB to the usb hcd layer into a
high-priority tasklet. This will have most benefit for isochronous IN
transfers but will also have incidental benefit where multiple periodic
devices are active at once.

dwc_otg: fix NAK holdoff and allow on split transactions only

This corrects a bug where if a single active non-periodic endpoint
had at least one transaction in its qh, on frnum == MAX_FRNUM the qh
would get skipped and never get queued again. This would result in
a silent device until error detection (automatic or otherwise) would
either reset the device or flush and requeue the URBs.

Additionally the NAK holdoff was enabled for all transactions - this
would potentially stall a HS endpoint for 1ms if a previous error state
enabled this interrupt and the next response was a NAK. Fix so that
only split transactions get held off.

dwc_otg: Call usb_hcd_unlink_urb_from_ep with lock held in completion handler

usb_hcd_unlink_urb_from_ep must be called with the HCD lock held.  Calling it
asynchronously in the tasklet was not safe (regression in
c4564d4).

This change unlinks it from the endpoint prior to queueing it for handling in
the tasklet, and also adds a check to ensure the urb is OK to be unlinked
before doing so.

NULL pointer dereference kernel oopses had been observed in usb_hcd_giveback_urb
when a USB device was unplugged/replugged during data transfer.  This effect
was reproduced using automated USB port power control, hundreds of replug
events were performed during active transfers to confirm that the problem was
eliminated.

USB fix using a FIQ to implement split transactions

This commit adds a FIQ implementaion that schedules
the split transactions using a FIQ so we don't get
held off by the interrupt latency of Linux

dwc_otg: fix device attributes and avoid kernel warnings on boot

dcw_otg: avoid logging function that can cause panics

See: raspberrypi/firmware#21
Thanks to cleverca22 for fix

dwc_otg: mask correct interrupts after transaction error recovery

The dwc_otg driver will unmask certain interrupts on a transaction
that previously halted in the error state in order to reset the
QTD error count. The various fine-grained interrupt handlers do not
consider that other interrupts besides themselves were unmasked.

By disabling the two other interrupts only ever enabled in DMA mode
for this purpose, we can avoid unnecessary function calls in the
IRQ handler. This will also prevent an unneccesary FIQ interrupt
from being generated if the FIQ is enabled.

dwc_otg: fiq: prevent FIQ thrash and incorrect state passing to IRQ

In the case of a transaction to a device that had previously aborted
due to an error, several interrupts are enabled to reset the error
count when a device responds. This has the side-effect of making the
FIQ thrash because the hardware will generate multiple instances of
a NAK on an IN bulk/interrupt endpoint and multiple instances of ACK
on an OUT bulk/interrupt endpoint. Make the FIQ mask and clear the
associated interrupts.

Additionally, on non-split transactions make sure that only unmasked
interrupts are cleared. This caused a hard-to-trigger but serious
race condition when you had the combination of an endpoint awaiting
error recovery and a transaction completed on an endpoint - due to
the sequencing and timing of interrupts generated by the dwc_otg core,
it was possible to confuse the IRQ handler.

Fix function tracing

dwc_otg: whitespace cleanup in dwc_otg_urb_enqueue

dwc_otg: prevent OOPSes during device disconnects

The dwc_otg_urb_enqueue function is thread-unsafe. In particular the
access of urb->hcpriv, usb_hcd_link_urb_to_ep, dwc_otg_urb->qtd and
friends does not occur within a critical section and so if a device
was unplugged during activity there was a high chance that the
usbcore hub_thread would try to disable the endpoint with partially-
formed entries in the URB queue. This would result in BUG() or null
pointer dereferences.

Fix so that access of urb->hcpriv, enqueuing to the hardware and
adding to usbcore endpoint URB lists is contained within a single
critical section.

dwc_otg: prevent BUG() in TT allocation if hub address is > 16

A fixed-size array is used to track TT allocation. This was
previously set to 16 which caused a crash because
dwc_otg_hcd_allocate_port would read past the end of the array.

This was hit if a hub was plugged in which enumerated as addr > 16,
due to previous device resets or unplugs.

Also add #ifdef FIQ_DEBUG around hcd->hub_port_alloc[], which grows
to a large size if 128 hub addresses are supported. This field is
for debug only for tracking which frame an allocate happened in.

dwc_otg: make channel halts with unknown state less damaging

If the IRQ received a channel halt interrupt through the FIQ
with no other bits set, the IRQ would not release the host
channel and never complete the URB.

Add catchall handling to treat as a transaction error and retry.

dwc_otg: fiq_split: use TTs with more granularity

This fixes certain issues with split transaction scheduling.

- Isochronous multi-packet OUT transactions now hog the TT until
  they are completed - this prevents hubs aborting transactions
  if they get a periodic start-split out-of-order
- Don't perform TT allocation on non-periodic endpoints - this
  allows simultaneous use of the TT's bulk/control and periodic
  transaction buffers

This commit will mainly affect USB audio playback.

dwc_otg: fix potential sleep while atomic during urb enqueue

Fixes a regression introduced with eb1b482. Kmalloc called from
dwc_otg_hcd_qtd_add / dwc_otg_hcd_qtd_create did not always have
the GPF_ATOMIC flag set. Force this flag when inside the larger
critical section.

dwc_otg: make fiq_split_enable imply fiq_fix_enable

Failing to set up the FIQ correctly would result in
"IRQ 32: nobody cared" errors in dmesg.

dwc_otg: prevent crashes on host port disconnects

Fix several issues resulting in crashes or inconsistent state
if a Model A root port was disconnected.

- Clean up queue heads properly in kill_urbs_in_qh_list by
  removing the empty QHs from the schedule lists
- Set the halt status properly to prevent IRQ handlers from
  using freed memory
- Add fiq_split related cleanup for saved registers
- Make microframe scheduling reclaim host channels if
  active during a disconnect
- Abort URBs with -ESHUTDOWN status response, informing
  device drivers so they respond in a more correct fashion
  and don't try to resubmit URBs
- Prevent IRQ handlers from attempting to handle channel
  interrupts if the associated URB was dequeued (and the
  driver state was cleared)

dwc_otg: prevent leaking URBs during enqueue

A dwc_otg_urb would get leaked if the HCD enqueue function
failed for any reason. Free the URB at the appropriate points.

dwc_otg: Enable NAK holdoff for control split transactions

Certain low-speed devices take a very long time to complete a
data or status stage of a control transaction, producing NAK
responses until they complete internal processing - the USB2.0
spec limit is up to 500mS. This causes the same type of interrupt
storm as seen with USB-serial dongles prior to c8edb23.

In certain circumstances, usually while booting, this interrupt
storm could cause SD card timeouts.

dwc_otg: Fix for occasional lockup on boot when doing a USB reset

dwc_otg: Don't issue traffic to LS devices in FS mode

Issuing low-speed packets when the root port is in full-speed mode
causes the root port to stop responding. Explicitly fail when
enqueuing URBs to a LS endpoint on a FS bus.

Fix ARM architecture issue with local_irq_restore()

If local_fiq_enable() is called before a local_irq_restore(flags) where
the flags variable has the F bit set, the FIQ will be erroneously disabled.

Fixup arch_local_irq_restore to avoid trampling the F bit in CPSR.

Also fix some of the hacks previously implemented for previous dwc_otg
incarnations.

dwc_otg: fiq_fsm: Base commit for driver rewrite

This commit removes the previous FIQ fixes entirely and adds fiq_fsm.

This rewrite features much more complete support for split transactions
and takes into account several OTG hardware bugs. High-speed
isochronous transactions are also capable of being performed by fiq_fsm.

All driver options have been removed and replaced with:
  - dwc_otg.fiq_enable (bool)
  - dwc_otg.fiq_fsm_enable (bool)
  - dwc_otg.fiq_fsm_mask (bitmask)
  - dwc_otg.nak_holdoff (unsigned int)

Defaults are specified such that fiq_fsm behaves similarly to the
previously implemented FIQ fixes.

fiq_fsm: Push error recovery into the FIQ when fiq_fsm is used

If the transfer associated with a QTD failed due to a bus error, the HCD
would retry the transfer up to 3 times (implementing the USB2.0
three-strikes retry in software).

Due to the masking mechanism used by fiq_fsm, it is only possible to pass
a single interrupt through to the HCD per-transfer.

In this instance host channels would fall off the radar because the error
reset would function, but the subsequent channel halt would be lost.

Push the error count reset into the FIQ handler.

fiq_fsm: Implement timeout mechanism

For full-speed endpoints with a large packet size, interrupt latency
runs the risk of the FIQ starting a transaction too late in a full-speed
frame. If the device is still transmitting data when EOF2 for the
downstream frame occurs, the hub will disable the port. This change is
not reflected in the hub status endpoint and the device becomes
unresponsive.

Prevent high-bandwidth transactions from being started too late in a
frame. The mechanism is not guaranteed: a combination of bit stuffing
and hub latency may still result in a device overrunning.

fiq_fsm: fix bounce buffer utilisation for Isochronous OUT

Multi-packet isochronous OUT transactions were subject to a few bounday
bugs. Fix them.

Audio playback is now much more robust: however, an issue stands with
devices that have adaptive sinks - ALSA plays samples too fast.

dwc_otg: Return full-speed frame numbers in HS mode

The frame counter increments on every *microframe* in high-speed mode.
Most device drivers expect this number to be in full-speed frames - this
caused considerable confusion to e.g. snd_usb_audio which uses the
frame counter to estimate the number of samples played.

fiq_fsm: save PID on completion of interrupt OUT transfers

Also add edge case handling for interrupt transports.

Note that for periodic split IN, data toggles are unimplemented in the
OTG host hardware - it unconditionally accepts any PID.

fiq_fsm: add missing case for fiq_fsm_tt_in_use()

Certain combinations of bitrate and endpoint activity could
result in a periodic transaction erroneously getting started
while the previous Isochronous OUT was still active.

fiq_fsm: clear hcintmsk for aborted transactions

Prevents the FIQ from erroneously handling interrupts
on a timed out channel.

fiq_fsm: enable by default

fiq_fsm: fix dequeues for non-periodic split transactions

If a dequeue happened between the SSPLIT and CSPLIT phases of the
transaction, the HCD would never receive an interrupt.

fiq_fsm: Disable by default

fiq_fsm: Handle HC babble errors

The HCTSIZ transfer size field raises a babble interrupt if
the counter wraps. Handle the resulting interrupt in this case.

dwc_otg: fix interrupt registration for fiq_enable=0

Additionally make the module parameter conditional for wherever
hcd->fiq_state is touched.

fiq_fsm: Enable by default

dwc_otg: Fix various issues with root port and transaction errors

Process the host port interrupts correctly (and don't trample them).
Root port hotplug now functional again.

Fix a few thinkos with the transaction error passthrough for fiq_fsm.

fiq_fsm: Implement hack for Split Interrupt transactions

Hubs aren't too picky about which endpoint we send Control type split
transactions to. By treating Interrupt transfers as Control, it is
possible to use the non-periodic queue in the OTG core as well as the
non-periodic FIFOs in the hub itself. This massively reduces the
microframe exclusivity/contention that periodic split transactions
otherwise have to enforce.

It goes without saying that this is a fairly egregious USB specification
violation, but it works.

Original idea by Hans Petter Selasky @ FreeBSD.org.

dwc_otg: FIQ support on SMP. Set up FIQ stack and handler on Core 0 only.

dwc_otg: introduce fiq_fsm_spin(un|)lock()

SMP safety for the FIQ relies on register read-modify write cycles being
completed in the correct order. Several places in the DWC code modify
registers also touched by the FIQ. Protect these by a bare-bones lock
mechanism.

This also makes it possible to run the FIQ and IRQ handlers on different
cores.

fiq_fsm: fix build on bcm2708 and bcm2709 platforms

dwc_otg: put some barriers back where they should be for UP

bcm2709/dwc_otg: Setup FIQ on core 1 if >1 core active

dwc_otg: fixup read-modify-write in critical paths

Be more careful about read-modify-write on registers that the FIQ
also touches.

Guard fiq_fsm_spin_lock with fiq_enable check

fiq_fsm: Falling out of the state machine isn't fatal

This edge case can be hit if the port is disabled while the FIQ is
in the middle of a transaction. Make the effects less severe.

Also get rid of the useless return value.

squash: dwc_otg: Allow to build without SMP

usb: core: make overcurrent messages more prominent

Hub overcurrent messages are more serious than "debug". Increase loglevel.

usb: dwc_otg: Don't use dma_to_virt()

Commit 6ce0d20 changes dma_to_virt() which breaks this driver.
Open code the old dma_to_virt() implementation to work around this.

Limit the use of __bus_to_virt() to cases where transfer_buffer_length
is set and transfer_buffer is not set. This is done to increase the
chance that this driver will also work on ARCH_BCM2835.

transfer_buffer should not be NULL if the length is set, but the
comment in the code indicates that there are situations where this
might happen. drivers/usb/isp1760/isp1760-hcd.c also has a similar
comment pointing to a possible: 'usb storage / SCSI bug'.

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>

dwc_otg: Fix crash when fiq_enable=0

dwc_otg: fiq_fsm: Make high-speed isochronous strided transfers work properly

Certain low-bandwidth high-speed USB devices (specialist audio devices,
compressed-frame webcams) have packet intervals > 1 microframe.

Stride these transfers in the FIQ by using the start-of-frame interrupt
to restart the channel at the right time.

dwc_otg: Force host mode to fix incorrect compute module boards

dwc_otg: Add ARCH_BCM2835 support

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>

dwc_otg: Simplify FIQ irq number code

Dropping ATAGS means we can simplify the FIQ irq number code.
Also add error checking on the returned irq number.

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>

dwc_otg: Remove duplicate gadget probe/unregister function

dwc_otg: Properly set the HFIR

Douglas Anderson reported:

According to the most up to date version of the dwc2 databook, the FRINT
field of the HFIR register should be programmed to:
* 125 us * (PHY clock freq for HS) - 1
* 1000 us * (PHY clock freq for FS/LS) - 1

This is opposed to older versions of the doc that claimed it should be:
* 125 us * (PHY clock freq for HS)
* 1000 us * (PHY clock freq for FS/LS)

and reported lower timing jitter on a USB analyser

dcw_otg: trim xfer length when buffer larger than allocated size is received

dwc_otg: Don't free qh align buffers in atomic context

dwc_otg: Enable the hack for Split Interrupt transactions by default

dwc_otg.fiq_fsm_mask=0xF has long been a suggestion for users with audio stutters or other USB bandwidth issues.
So far we are aware of many success stories but no failure caused by this setting.
Make it a default to learn more.

See: https://www.raspberrypi.org/forums/viewtopic.php?f=28&t=70437

Signed-off-by: popcornmix <popcornmix@gmail.com>

dwc_otg: Use kzalloc when suitable

dwc_otg: Pass struct device to dma_alloc*()

This makes it possible to get the bus address from Device Tree.

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>

dwc_otg: fix summarize urb->actual_length for isochronous transfers

Kernel does not copy input data of ISO transfers to userspace
if actual_length is set only in ISO transfers and not summarized
in urb->actual_length. Fixes #903

fiq_fsm: Use correct states when starting isoc OUT transfers

In fiq_fsm_start_next_periodic() if an isochronous OUT transfer
was selected, no regard was given as to whether this was a single-packet
transfer or a multi-packet staged transfer.

For single-packet transfers, this had the effect of repeatedly sending
OUT packets with bogus data and lengths.

Eventually if the channel was repeatedly enabled enough times, this
would lock up the OTG core and no further bus transfers would happen.

Set the FSM state up properly if we select a single-packet transfer.

Fixes #1842

dwc_otg: make nak_holdoff work as intended with empty queues

If URBs reading from non-periodic split endpoints were dequeued and
the last transfer from the endpoint was a NAK handshake, the resulting
qh->nak_frame value was stale which would result in unnecessarily long
polling intervals for the first subsequent transfer with a fresh URB.

Fixup qh->nak_frame in dwc_otg_hcd_urb_dequeue and also guard against
a case where a single URB is submitted to the endpoint, a NAK was
received on the transfer immediately prior to receiving data and the
device subsequently resubmits another URB past the qh->nak_frame interval.

Fixes #1709

dwc_otg: fix split transaction data toggle handling around dequeues

See #1709

Fix several issues regarding endpoint state when URBs are dequeued
- If the HCD is disconnected, flush FIQ-enabled channels properly
- Save the data toggle state for bulk endpoints if the last transfer
  from an endpoint where URBs were dequeued returned a data packet
- Reset hc->start_pkt_count properly in assign_and_init_hc()

dwc_otg: fix several potential crash sources

On root port disconnect events, the host driver state is cleared and
in-progress host channels are forcibly stopped. This doesn't play
well with the FIQ running in the background, so:
- Guard the disconnect callback with both the host spinlock and FIQ
  spinlock
- Move qtd dereference in dwc_otg_handle_hc_fsm() after the early-out
  so we don't dereference a qtd that has gone away
- Turn catch-all BUG()s in dwc_otg_handle_hc_fsm() into warnings.

dwc_otg: delete hcd->channel_lock

The lock serves no purpose as it is only held while the HCD spinlock
is already being held.

dwc_otg: remove unnecessary dma-mode channel halts on disconnect interrupt

Host channels are already halted in kill_urbs_in_qh_list() with the
subsequent interrupt processing behaving as if the URB was dequeued
via HCD callback.

There's no need to clobber the host channel registers a second time
as this exposes races between the driver and host channel resulting
in hcd->free_hc_list becoming corrupted.

dwcotg: Allow to build without FIQ on ARM64

Signed-off-by: popcornmix <popcornmix@gmail.com>

dwc_otg: make periodic scheduling behave properly for FS buses

If the root port is in full-speed mode, transfer times at 12mbit/s
would be calculated but matched against high-speed quotas.

Reinitialise hcd->frame_usecs[i] on each port enable event so that
full-speed bandwidth can be tracked sensibly.

Also, don't bother using the FIQ for transfers when in full-speed
mode - at the slower bus speed, interrupt frequency is reduced by
an order of magnitude.

Related issue: #2020

dwc_otg: fiq_fsm: Make isochronous compatibility checks work properly

Get rid of the spammy printk and local pointer mangling.
Also, there is a nominal benefit for using fiq_fsm for isochronous
transfers in FS mode (~1.1k IRQs per second vs 2.1k IRQs per second)
so remove the root port speed check.

dwc_otg: add module parameter int_ep_interval_min

Add a module parameter (defaulting to ignored) that clamps the polling rate
of high-speed Interrupt endpoints to a minimum microframe interval.

The parameter is modifiable at runtime as it is used when activating new
endpoints (such as on device connect).

dwc_otg: fiq_fsm: Add non-periodic TT exclusivity constraints

Certain hub types do not discriminate between pipe direction (IN or OUT)
when considering non-periodic transfers. Therefore these hubs get confused
if multiple transfers are issued in different directions with the same
device address and endpoint number.

Constrain queuing non-periodic split transactions so they are performed
serially in such cases.

Related: #2024

dwc_otg: Fixup change to DRIVER_ATTR interface

dwc_otg: Fix compilation warnings

Signed-off-by: Phil Elwell <phil@raspberrypi.org>

USB_DWCOTG: Disable building dwc_otg as a module (#2265)

When dwc_otg is built as a module, build will fail with the following
error:

ERROR: "DWC_TASK_HI_SCHEDULE" [drivers/usb/host/dwc_otg/dwc_otg.ko] undefined!
scripts/Makefile.modpost:91: recipe for target '__modpost' failed
make[1]: *** [__modpost] Error 1
Makefile:1199: recipe for target 'modules' failed
make: *** [modules] Error 2

Even if the error is solved by including the missing
DWC_TASK_HI_SCHEDULE function, the kernel will panic when loading
dwc_otg.

As a workaround, simply prevent user from building dwc_otg as a module
as the current kernel does not support it.

See: #2258

Signed-off-by: Malik Olivier Boussejra <malik@boussejra.com>

dwc_otg: New timer API

dwc_otg: Fix removed ACCESS_ONCE->READ_ONCE

dwc_otg: don't unconditionally force host mode in dwc_otg_cil_init()

Add the ability to disable force_host_mode for those that want to use
dwc_otg in both device and host modes.

dwc_otg: Fix a regression when dequeueing isochronous transfers

In 282bed9 (dwc_otg: make nak_holdoff work as intended with empty queues)
the dequeue mechanism was changed to leave FIQ-enabled transfers to run
to completion - to avoid leaving hub TT buffers with stale packets lying
around.

This broke FIQ-accelerated isochronous transfers, as this then meant that
dozens of transfers were performed after the dequeue function returned.

Restore the state machine fence for isochronous transfers.

fiq_fsm: rewind DMA pointer for OUT transactions that fail (#2288)

See: #2140

dwc_otg: add smp_mb() to prevent driver state corruption on boot

Occasional crashes have been seen where the FIQ code dereferences
invalid/random pointers immediately after being set up, leading to
panic on boot.

The crash occurs as the FIQ code races against hcd_init_fiq() and
the hcd_init_fiq() code races against the outstanding memory stores
from dwc_otg_hcd_init(). Use explicit barriers after touching
driver state.

usb: dwc_otg: fix memory corruption in dwc_otg driver

[Upstream commit 51b1b64]

The move from the staging tree to the main tree exposed a
longstanding memory corruption bug in the dwc2 driver. The
reordering of the driver initialization caused the dwc2 driver
to corrupt the initialization data of the sdhci driver on the
Raspberry Pi platform, which made the bug show up.

The error is in calling to_usb_device(hsotg->dev), since ->dev
is not a member of struct usb_device. The easiest fix is to
just remove the offending code, since it is not really needed.

Thanks to Stephen Warren for tracking down the cause of this.

Reported-by: Andre Heider <a.heider@gmail.com>
Tested-by: Stephen Warren <swarren@wwwdotorg.org>
Signed-off-by: Paul Zimmerman <paulz@synopsys.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[lukas: port from upstream dwc2 to out-of-tree dwc_otg driver]
Signed-off-by: Lukas Wunner <lukas@wunner.de>

usb: dwb_otg: Fix unreachable switch statement warning

This warning appears with GCC 7.3.0 from toolchains.bootlin.com:

../drivers/usb/host/dwc_otg/dwc_otg_fiq_fsm.c: In function ‘fiq_fsm_update_hs_isoc’:
../drivers/usb/host/dwc_otg/dwc_otg_fiq_fsm.c:595:61: warning: statement will never be executed [-Wswitch-unreachable]
   st->hctsiz_copy.b.xfersize = nrpackets * st->hcchar_copy.b.mps;
                                            ~~~~~~~~~~~~~~~~~^~~~

Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>

dwc_otg: fiq_fsm: fix incorrect DMA register offset calculation

Rationalise the offset and update all call sites.

Fixes #2408

popcornmix added a commit that referenced this issue Mar 6, 2019

Add dwc_otg driver
Signed-off-by: popcornmix <popcornmix@gmail.com>

usb: dwc: fix lockdep false positive

Signed-off-by: Kari Suvanto <karis79@gmail.com>

usb: dwc: fix inconsistent lock state

Signed-off-by: Kari Suvanto <karis79@gmail.com>

Add FIQ patch to dwc_otg driver. Enable with dwc_otg.fiq_fix_enable=1. Should give about 10% more ARM performance.
Thanks to Gordon and Costas

Avoid dynamic memory allocation for channel lock in USB driver. Thanks ddv2005.

Add NAK holdoff scheme. Enabled by default, disable with dwc_otg.nak_holdoff_enable=0. Thanks gsh

Make sure we wait for the reset to finish

dwc_otg: fix bug in dwc_otg_hcd.c resulting in silent kernel
	 memory corruption, escalating to OOPS under high USB load.

dwc_otg: Fix unsafe access of QTD during URB enqueue

In dwc_otg_hcd_urb_enqueue during qtd creation, it was possible that the
transaction could complete almost immediately after the qtd was assigned
to a host channel during URB enqueue, which meant the qtd pointer was no
longer valid having been completed and removed. Usually, this resulted in
an OOPS during URB submission. By predetermining whether transactions
need to be queued or not, this unsafe pointer access is avoided.

This bug was only evident on the Pi model A where a device was attached
that had no periodic endpoints (e.g. USB pendrive or some wlan devices).

dwc_otg: Fix incorrect URB allocation error handling

If the memory allocation for a dwc_otg_urb failed, the kernel would OOPS
because for some reason a member of the *unallocated* struct was set to
zero. Error handling changed to fail correctly.

dwc_otg: fix potential use-after-free case in interrupt handler

If a transaction had previously aborted, certain interrupts are
enabled to track error counts and reset where necessary. On IN
endpoints the host generates an ACK interrupt near-simultaneously
with completion of transfer. In the case where this transfer had
previously had an error, this results in a use-after-free on
the QTD memory space with a 1-byte length being overwritten to
0x00.

dwc_otg: add handling of SPLIT transaction data toggle errors

Previously a data toggle error on packets from a USB1.1 device behind
a TT would result in the Pi locking up as the driver never handled
the associated interrupt. Patch adds basic retry mechanism and
interrupt acknowledgement to cater for either a chance toggle error or
for devices that have a broken initial toggle state (FT8U232/FT232BM).

dwc_otg: implement tasklet for returning URBs to usbcore hcd layer

The dwc_otg driver interrupt handler for transfer completion will spend
a very long time with interrupts disabled when a URB is completed -
this is because usb_hcd_giveback_urb is called from within the handler
which for a USB device driver with complicated processing (e.g. webcam)
will take an exorbitant amount of time to complete. This results in
missed completion interrupts for other USB packets which lead to them
being dropped due to microframe overruns.

This patch splits returning the URB to the usb hcd layer into a
high-priority tasklet. This will have most benefit for isochronous IN
transfers but will also have incidental benefit where multiple periodic
devices are active at once.

dwc_otg: fix NAK holdoff and allow on split transactions only

This corrects a bug where if a single active non-periodic endpoint
had at least one transaction in its qh, on frnum == MAX_FRNUM the qh
would get skipped and never get queued again. This would result in
a silent device until error detection (automatic or otherwise) would
either reset the device or flush and requeue the URBs.

Additionally the NAK holdoff was enabled for all transactions - this
would potentially stall a HS endpoint for 1ms if a previous error state
enabled this interrupt and the next response was a NAK. Fix so that
only split transactions get held off.

dwc_otg: Call usb_hcd_unlink_urb_from_ep with lock held in completion handler

usb_hcd_unlink_urb_from_ep must be called with the HCD lock held.  Calling it
asynchronously in the tasklet was not safe (regression in
c4564d4).

This change unlinks it from the endpoint prior to queueing it for handling in
the tasklet, and also adds a check to ensure the urb is OK to be unlinked
before doing so.

NULL pointer dereference kernel oopses had been observed in usb_hcd_giveback_urb
when a USB device was unplugged/replugged during data transfer.  This effect
was reproduced using automated USB port power control, hundreds of replug
events were performed during active transfers to confirm that the problem was
eliminated.

USB fix using a FIQ to implement split transactions

This commit adds a FIQ implementaion that schedules
the split transactions using a FIQ so we don't get
held off by the interrupt latency of Linux

dwc_otg: fix device attributes and avoid kernel warnings on boot

dcw_otg: avoid logging function that can cause panics

See: raspberrypi/firmware#21
Thanks to cleverca22 for fix

dwc_otg: mask correct interrupts after transaction error recovery

The dwc_otg driver will unmask certain interrupts on a transaction
that previously halted in the error state in order to reset the
QTD error count. The various fine-grained interrupt handlers do not
consider that other interrupts besides themselves were unmasked.

By disabling the two other interrupts only ever enabled in DMA mode
for this purpose, we can avoid unnecessary function calls in the
IRQ handler. This will also prevent an unneccesary FIQ interrupt
from being generated if the FIQ is enabled.

dwc_otg: fiq: prevent FIQ thrash and incorrect state passing to IRQ

In the case of a transaction to a device that had previously aborted
due to an error, several interrupts are enabled to reset the error
count when a device responds. This has the side-effect of making the
FIQ thrash because the hardware will generate multiple instances of
a NAK on an IN bulk/interrupt endpoint and multiple instances of ACK
on an OUT bulk/interrupt endpoint. Make the FIQ mask and clear the
associated interrupts.

Additionally, on non-split transactions make sure that only unmasked
interrupts are cleared. This caused a hard-to-trigger but serious
race condition when you had the combination of an endpoint awaiting
error recovery and a transaction completed on an endpoint - due to
the sequencing and timing of interrupts generated by the dwc_otg core,
it was possible to confuse the IRQ handler.

Fix function tracing

dwc_otg: whitespace cleanup in dwc_otg_urb_enqueue

dwc_otg: prevent OOPSes during device disconnects

The dwc_otg_urb_enqueue function is thread-unsafe. In particular the
access of urb->hcpriv, usb_hcd_link_urb_to_ep, dwc_otg_urb->qtd and
friends does not occur within a critical section and so if a device
was unplugged during activity there was a high chance that the
usbcore hub_thread would try to disable the endpoint with partially-
formed entries in the URB queue. This would result in BUG() or null
pointer dereferences.

Fix so that access of urb->hcpriv, enqueuing to the hardware and
adding to usbcore endpoint URB lists is contained within a single
critical section.

dwc_otg: prevent BUG() in TT allocation if hub address is > 16

A fixed-size array is used to track TT allocation. This was
previously set to 16 which caused a crash because
dwc_otg_hcd_allocate_port would read past the end of the array.

This was hit if a hub was plugged in which enumerated as addr > 16,
due to previous device resets or unplugs.

Also add #ifdef FIQ_DEBUG around hcd->hub_port_alloc[], which grows
to a large size if 128 hub addresses are supported. This field is
for debug only for tracking which frame an allocate happened in.

dwc_otg: make channel halts with unknown state less damaging

If the IRQ received a channel halt interrupt through the FIQ
with no other bits set, the IRQ would not release the host
channel and never complete the URB.

Add catchall handling to treat as a transaction error and retry.

dwc_otg: fiq_split: use TTs with more granularity

This fixes certain issues with split transaction scheduling.

- Isochronous multi-packet OUT transactions now hog the TT until
  they are completed - this prevents hubs aborting transactions
  if they get a periodic start-split out-of-order
- Don't perform TT allocation on non-periodic endpoints - this
  allows simultaneous use of the TT's bulk/control and periodic
  transaction buffers

This commit will mainly affect USB audio playback.

dwc_otg: fix potential sleep while atomic during urb enqueue

Fixes a regression introduced with eb1b482. Kmalloc called from
dwc_otg_hcd_qtd_add / dwc_otg_hcd_qtd_create did not always have
the GPF_ATOMIC flag set. Force this flag when inside the larger
critical section.

dwc_otg: make fiq_split_enable imply fiq_fix_enable

Failing to set up the FIQ correctly would result in
"IRQ 32: nobody cared" errors in dmesg.

dwc_otg: prevent crashes on host port disconnects

Fix several issues resulting in crashes or inconsistent state
if a Model A root port was disconnected.

- Clean up queue heads properly in kill_urbs_in_qh_list by
  removing the empty QHs from the schedule lists
- Set the halt status properly to prevent IRQ handlers from
  using freed memory
- Add fiq_split related cleanup for saved registers
- Make microframe scheduling reclaim host channels if
  active during a disconnect
- Abort URBs with -ESHUTDOWN status response, informing
  device drivers so they respond in a more correct fashion
  and don't try to resubmit URBs
- Prevent IRQ handlers from attempting to handle channel
  interrupts if the associated URB was dequeued (and the
  driver state was cleared)

dwc_otg: prevent leaking URBs during enqueue

A dwc_otg_urb would get leaked if the HCD enqueue function
failed for any reason. Free the URB at the appropriate points.

dwc_otg: Enable NAK holdoff for control split transactions

Certain low-speed devices take a very long time to complete a
data or status stage of a control transaction, producing NAK
responses until they complete internal processing - the USB2.0
spec limit is up to 500mS. This causes the same type of interrupt
storm as seen with USB-serial dongles prior to c8edb23.

In certain circumstances, usually while booting, this interrupt
storm could cause SD card timeouts.

dwc_otg: Fix for occasional lockup on boot when doing a USB reset

dwc_otg: Don't issue traffic to LS devices in FS mode

Issuing low-speed packets when the root port is in full-speed mode
causes the root port to stop responding. Explicitly fail when
enqueuing URBs to a LS endpoint on a FS bus.

Fix ARM architecture issue with local_irq_restore()

If local_fiq_enable() is called before a local_irq_restore(flags) where
the flags variable has the F bit set, the FIQ will be erroneously disabled.

Fixup arch_local_irq_restore to avoid trampling the F bit in CPSR.

Also fix some of the hacks previously implemented for previous dwc_otg
incarnations.

dwc_otg: fiq_fsm: Base commit for driver rewrite

This commit removes the previous FIQ fixes entirely and adds fiq_fsm.

This rewrite features much more complete support for split transactions
and takes into account several OTG hardware bugs. High-speed
isochronous transactions are also capable of being performed by fiq_fsm.

All driver options have been removed and replaced with:
  - dwc_otg.fiq_enable (bool)
  - dwc_otg.fiq_fsm_enable (bool)
  - dwc_otg.fiq_fsm_mask (bitmask)
  - dwc_otg.nak_holdoff (unsigned int)

Defaults are specified such that fiq_fsm behaves similarly to the
previously implemented FIQ fixes.

fiq_fsm: Push error recovery into the FIQ when fiq_fsm is used

If the transfer associated with a QTD failed due to a bus error, the HCD
would retry the transfer up to 3 times (implementing the USB2.0
three-strikes retry in software).

Due to the masking mechanism used by fiq_fsm, it is only possible to pass
a single interrupt through to the HCD per-transfer.

In this instance host channels would fall off the radar because the error
reset would function, but the subsequent channel halt would be lost.

Push the error count reset into the FIQ handler.

fiq_fsm: Implement timeout mechanism

For full-speed endpoints with a large packet size, interrupt latency
runs the risk of the FIQ starting a transaction too late in a full-speed
frame. If the device is still transmitting data when EOF2 for the
downstream frame occurs, the hub will disable the port. This change is
not reflected in the hub status endpoint and the device becomes
unresponsive.

Prevent high-bandwidth transactions from being started too late in a
frame. The mechanism is not guaranteed: a combination of bit stuffing
and hub latency may still result in a device overrunning.

fiq_fsm: fix bounce buffer utilisation for Isochronous OUT

Multi-packet isochronous OUT transactions were subject to a few bounday
bugs. Fix them.

Audio playback is now much more robust: however, an issue stands with
devices that have adaptive sinks - ALSA plays samples too fast.

dwc_otg: Return full-speed frame numbers in HS mode

The frame counter increments on every *microframe* in high-speed mode.
Most device drivers expect this number to be in full-speed frames - this
caused considerable confusion to e.g. snd_usb_audio which uses the
frame counter to estimate the number of samples played.

fiq_fsm: save PID on completion of interrupt OUT transfers

Also add edge case handling for interrupt transports.

Note that for periodic split IN, data toggles are unimplemented in the
OTG host hardware - it unconditionally accepts any PID.

fiq_fsm: add missing case for fiq_fsm_tt_in_use()

Certain combinations of bitrate and endpoint activity could
result in a periodic transaction erroneously getting started
while the previous Isochronous OUT was still active.

fiq_fsm: clear hcintmsk for aborted transactions

Prevents the FIQ from erroneously handling interrupts
on a timed out channel.

fiq_fsm: enable by default

fiq_fsm: fix dequeues for non-periodic split transactions

If a dequeue happened between the SSPLIT and CSPLIT phases of the
transaction, the HCD would never receive an interrupt.

fiq_fsm: Disable by default

fiq_fsm: Handle HC babble errors

The HCTSIZ transfer size field raises a babble interrupt if
the counter wraps. Handle the resulting interrupt in this case.

dwc_otg: fix interrupt registration for fiq_enable=0

Additionally make the module parameter conditional for wherever
hcd->fiq_state is touched.

fiq_fsm: Enable by default

dwc_otg: Fix various issues with root port and transaction errors

Process the host port interrupts correctly (and don't trample them).
Root port hotplug now functional again.

Fix a few thinkos with the transaction error passthrough for fiq_fsm.

fiq_fsm: Implement hack for Split Interrupt transactions

Hubs aren't too picky about which endpoint we send Control type split
transactions to. By treating Interrupt transfers as Control, it is
possible to use the non-periodic queue in the OTG core as well as the
non-periodic FIFOs in the hub itself. This massively reduces the
microframe exclusivity/contention that periodic split transactions
otherwise have to enforce.

It goes without saying that this is a fairly egregious USB specification
violation, but it works.

Original idea by Hans Petter Selasky @ FreeBSD.org.

dwc_otg: FIQ support on SMP. Set up FIQ stack and handler on Core 0 only.

dwc_otg: introduce fiq_fsm_spin(un|)lock()

SMP safety for the FIQ relies on register read-modify write cycles being
completed in the correct order. Several places in the DWC code modify
registers also touched by the FIQ. Protect these by a bare-bones lock
mechanism.

This also makes it possible to run the FIQ and IRQ handlers on different
cores.

fiq_fsm: fix build on bcm2708 and bcm2709 platforms

dwc_otg: put some barriers back where they should be for UP

bcm2709/dwc_otg: Setup FIQ on core 1 if >1 core active

dwc_otg: fixup read-modify-write in critical paths

Be more careful about read-modify-write on registers that the FIQ
also touches.

Guard fiq_fsm_spin_lock with fiq_enable check

fiq_fsm: Falling out of the state machine isn't fatal

This edge case can be hit if the port is disabled while the FIQ is
in the middle of a transaction. Make the effects less severe.

Also get rid of the useless return value.

squash: dwc_otg: Allow to build without SMP

usb: core: make overcurrent messages more prominent

Hub overcurrent messages are more serious than "debug". Increase loglevel.

usb: dwc_otg: Don't use dma_to_virt()

Commit 6ce0d20 changes dma_to_virt() which breaks this driver.
Open code the old dma_to_virt() implementation to work around this.

Limit the use of __bus_to_virt() to cases where transfer_buffer_length
is set and transfer_buffer is not set. This is done to increase the
chance that this driver will also work on ARCH_BCM2835.

transfer_buffer should not be NULL if the length is set, but the
comment in the code indicates that there are situations where this
might happen. drivers/usb/isp1760/isp1760-hcd.c also has a similar
comment pointing to a possible: 'usb storage / SCSI bug'.

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>

dwc_otg: Fix crash when fiq_enable=0

dwc_otg: fiq_fsm: Make high-speed isochronous strided transfers work properly

Certain low-bandwidth high-speed USB devices (specialist audio devices,
compressed-frame webcams) have packet intervals > 1 microframe.

Stride these transfers in the FIQ by using the start-of-frame interrupt
to restart the channel at the right time.

dwc_otg: Force host mode to fix incorrect compute module boards

dwc_otg: Add ARCH_BCM2835 support

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>

dwc_otg: Simplify FIQ irq number code

Dropping ATAGS means we can simplify the FIQ irq number code.
Also add error checking on the returned irq number.

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>

dwc_otg: Remove duplicate gadget probe/unregister function

dwc_otg: Properly set the HFIR

Douglas Anderson reported:

According to the most up to date version of the dwc2 databook, the FRINT
field of the HFIR register should be programmed to:
* 125 us * (PHY clock freq for HS) - 1
* 1000 us * (PHY clock freq for FS/LS) - 1

This is opposed to older versions of the doc that claimed it should be:
* 125 us * (PHY clock freq for HS)
* 1000 us * (PHY clock freq for FS/LS)

and reported lower timing jitter on a USB analyser

dcw_otg: trim xfer length when buffer larger than allocated size is received

dwc_otg: Don't free qh align buffers in atomic context

dwc_otg: Enable the hack for Split Interrupt transactions by default

dwc_otg.fiq_fsm_mask=0xF has long been a suggestion for users with audio stutters or other USB bandwidth issues.
So far we are aware of many success stories but no failure caused by this setting.
Make it a default to learn more.

See: https://www.raspberrypi.org/forums/viewtopic.php?f=28&t=70437

Signed-off-by: popcornmix <popcornmix@gmail.com>

dwc_otg: Use kzalloc when suitable

dwc_otg: Pass struct device to dma_alloc*()

This makes it possible to get the bus address from Device Tree.

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>

dwc_otg: fix summarize urb->actual_length for isochronous transfers

Kernel does not copy input data of ISO transfers to userspace
if actual_length is set only in ISO transfers and not summarized
in urb->actual_length. Fixes #903

fiq_fsm: Use correct states when starting isoc OUT transfers

In fiq_fsm_start_next_periodic() if an isochronous OUT transfer
was selected, no regard was given as to whether this was a single-packet
transfer or a multi-packet staged transfer.

For single-packet transfers, this had the effect of repeatedly sending
OUT packets with bogus data and lengths.

Eventually if the channel was repeatedly enabled enough times, this
would lock up the OTG core and no further bus transfers would happen.

Set the FSM state up properly if we select a single-packet transfer.

Fixes #1842

dwc_otg: make nak_holdoff work as intended with empty queues

If URBs reading from non-periodic split endpoints were dequeued and
the last transfer from the endpoint was a NAK handshake, the resulting
qh->nak_frame value was stale which would result in unnecessarily long
polling intervals for the first subsequent transfer with a fresh URB.

Fixup qh->nak_frame in dwc_otg_hcd_urb_dequeue and also guard against
a case where a single URB is submitted to the endpoint, a NAK was
received on the transfer immediately prior to receiving data and the
device subsequently resubmits another URB past the qh->nak_frame interval.

Fixes #1709

dwc_otg: fix split transaction data toggle handling around dequeues

See #1709

Fix several issues regarding endpoint state when URBs are dequeued
- If the HCD is disconnected, flush FIQ-enabled channels properly
- Save the data toggle state for bulk endpoints if the last transfer
  from an endpoint where URBs were dequeued returned a data packet
- Reset hc->start_pkt_count properly in assign_and_init_hc()

dwc_otg: fix several potential crash sources

On root port disconnect events, the host driver state is cleared and
in-progress host channels are forcibly stopped. This doesn't play
well with the FIQ running in the background, so:
- Guard the disconnect callback with both the host spinlock and FIQ
  spinlock
- Move qtd dereference in dwc_otg_handle_hc_fsm() after the early-out
  so we don't dereference a qtd that has gone away
- Turn catch-all BUG()s in dwc_otg_handle_hc_fsm() into warnings.

dwc_otg: delete hcd->channel_lock

The lock serves no purpose as it is only held while the HCD spinlock
is already being held.

dwc_otg: remove unnecessary dma-mode channel halts on disconnect interrupt

Host channels are already halted in kill_urbs_in_qh_list() with the
subsequent interrupt processing behaving as if the URB was dequeued
via HCD callback.

There's no need to clobber the host channel registers a second time
as this exposes races between the driver and host channel resulting
in hcd->free_hc_list becoming corrupted.

dwcotg: Allow to build without FIQ on ARM64

Signed-off-by: popcornmix <popcornmix@gmail.com>

dwc_otg: make periodic scheduling behave properly for FS buses

If the root port is in full-speed mode, transfer times at 12mbit/s
would be calculated but matched against high-speed quotas.

Reinitialise hcd->frame_usecs[i] on each port enable event so that
full-speed bandwidth can be tracked sensibly.

Also, don't bother using the FIQ for transfers when in full-speed
mode - at the slower bus speed, interrupt frequency is reduced by
an order of magnitude.

Related issue: #2020

dwc_otg: fiq_fsm: Make isochronous compatibility checks work properly

Get rid of the spammy printk and local pointer mangling.
Also, there is a nominal benefit for using fiq_fsm for isochronous
transfers in FS mode (~1.1k IRQs per second vs 2.1k IRQs per second)
so remove the root port speed check.

dwc_otg: add module parameter int_ep_interval_min

Add a module parameter (defaulting to ignored) that clamps the polling rate
of high-speed Interrupt endpoints to a minimum microframe interval.

The parameter is modifiable at runtime as it is used when activating new
endpoints (such as on device connect).

dwc_otg: fiq_fsm: Add non-periodic TT exclusivity constraints

Certain hub types do not discriminate between pipe direction (IN or OUT)
when considering non-periodic transfers. Therefore these hubs get confused
if multiple transfers are issued in different directions with the same
device address and endpoint number.

Constrain queuing non-periodic split transactions so they are performed
serially in such cases.

Related: #2024

dwc_otg: Fixup change to DRIVER_ATTR interface

dwc_otg: Fix compilation warnings

Signed-off-by: Phil Elwell <phil@raspberrypi.org>

USB_DWCOTG: Disable building dwc_otg as a module (#2265)

When dwc_otg is built as a module, build will fail with the following
error:

ERROR: "DWC_TASK_HI_SCHEDULE" [drivers/usb/host/dwc_otg/dwc_otg.ko] undefined!
scripts/Makefile.modpost:91: recipe for target '__modpost' failed
make[1]: *** [__modpost] Error 1
Makefile:1199: recipe for target 'modules' failed
make: *** [modules] Error 2

Even if the error is solved by including the missing
DWC_TASK_HI_SCHEDULE function, the kernel will panic when loading
dwc_otg.

As a workaround, simply prevent user from building dwc_otg as a module
as the current kernel does not support it.

See: #2258

Signed-off-by: Malik Olivier Boussejra <malik@boussejra.com>

dwc_otg: New timer API

dwc_otg: Fix removed ACCESS_ONCE->READ_ONCE

dwc_otg: don't unconditionally force host mode in dwc_otg_cil_init()

Add the ability to disable force_host_mode for those that want to use
dwc_otg in both device and host modes.

dwc_otg: Fix a regression when dequeueing isochronous transfers

In 282bed9 (dwc_otg: make nak_holdoff work as intended with empty queues)
the dequeue mechanism was changed to leave FIQ-enabled transfers to run
to completion - to avoid leaving hub TT buffers with stale packets lying
around.

This broke FIQ-accelerated isochronous transfers, as this then meant that
dozens of transfers were performed after the dequeue function returned.

Restore the state machine fence for isochronous transfers.

fiq_fsm: rewind DMA pointer for OUT transactions that fail (#2288)

See: #2140

dwc_otg: add smp_mb() to prevent driver state corruption on boot

Occasional crashes have been seen where the FIQ code dereferences
invalid/random pointers immediately after being set up, leading to
panic on boot.

The crash occurs as the FIQ code races against hcd_init_fiq() and
the hcd_init_fiq() code races against the outstanding memory stores
from dwc_otg_hcd_init(). Use explicit barriers after touching
driver state.

usb: dwc_otg: fix memory corruption in dwc_otg driver

[Upstream commit 51b1b64]

The move from the staging tree to the main tree exposed a
longstanding memory corruption bug in the dwc2 driver. The
reordering of the driver initialization caused the dwc2 driver
to corrupt the initialization data of the sdhci driver on the
Raspberry Pi platform, which made the bug show up.

The error is in calling to_usb_device(hsotg->dev), since ->dev
is not a member of struct usb_device. The easiest fix is to
just remove the offending code, since it is not really needed.

Thanks to Stephen Warren for tracking down the cause of this.

Reported-by: Andre Heider <a.heider@gmail.com>
Tested-by: Stephen Warren <swarren@wwwdotorg.org>
Signed-off-by: Paul Zimmerman <paulz@synopsys.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[lukas: port from upstream dwc2 to out-of-tree dwc_otg driver]
Signed-off-by: Lukas Wunner <lukas@wunner.de>

usb: dwb_otg: Fix unreachable switch statement warning

This warning appears with GCC 7.3.0 from toolchains.bootlin.com:

../drivers/usb/host/dwc_otg/dwc_otg_fiq_fsm.c: In function ‘fiq_fsm_update_hs_isoc’:
../drivers/usb/host/dwc_otg/dwc_otg_fiq_fsm.c:595:61: warning: statement will never be executed [-Wswitch-unreachable]
   st->hctsiz_copy.b.xfersize = nrpackets * st->hcchar_copy.b.mps;
                                            ~~~~~~~~~~~~~~~~~^~~~

Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>

dwc_otg: fiq_fsm: fix incorrect DMA register offset calculation

Rationalise the offset and update all call sites.

Fixes #2408

popcornmix added a commit that referenced this issue Mar 6, 2019

Add dwc_otg driver
Signed-off-by: popcornmix <popcornmix@gmail.com>

usb: dwc: fix lockdep false positive

Signed-off-by: Kari Suvanto <karis79@gmail.com>

usb: dwc: fix inconsistent lock state

Signed-off-by: Kari Suvanto <karis79@gmail.com>

Add FIQ patch to dwc_otg driver. Enable with dwc_otg.fiq_fix_enable=1. Should give about 10% more ARM performance.
Thanks to Gordon and Costas

Avoid dynamic memory allocation for channel lock in USB driver. Thanks ddv2005.

Add NAK holdoff scheme. Enabled by default, disable with dwc_otg.nak_holdoff_enable=0. Thanks gsh

Make sure we wait for the reset to finish

dwc_otg: fix bug in dwc_otg_hcd.c resulting in silent kernel
	 memory corruption, escalating to OOPS under high USB load.

dwc_otg: Fix unsafe access of QTD during URB enqueue

In dwc_otg_hcd_urb_enqueue during qtd creation, it was possible that the
transaction could complete almost immediately after the qtd was assigned
to a host channel during URB enqueue, which meant the qtd pointer was no
longer valid having been completed and removed. Usually, this resulted in
an OOPS during URB submission. By predetermining whether transactions
need to be queued or not, this unsafe pointer access is avoided.

This bug was only evident on the Pi model A where a device was attached
that had no periodic endpoints (e.g. USB pendrive or some wlan devices).

dwc_otg: Fix incorrect URB allocation error handling

If the memory allocation for a dwc_otg_urb failed, the kernel would OOPS
because for some reason a member of the *unallocated* struct was set to
zero. Error handling changed to fail correctly.

dwc_otg: fix potential use-after-free case in interrupt handler

If a transaction had previously aborted, certain interrupts are
enabled to track error counts and reset where necessary. On IN
endpoints the host generates an ACK interrupt near-simultaneously
with completion of transfer. In the case where this transfer had
previously had an error, this results in a use-after-free on
the QTD memory space with a 1-byte length being overwritten to
0x00.

dwc_otg: add handling of SPLIT transaction data toggle errors

Previously a data toggle error on packets from a USB1.1 device behind
a TT would result in the Pi locking up as the driver never handled
the associated interrupt. Patch adds basic retry mechanism and
interrupt acknowledgement to cater for either a chance toggle error or
for devices that have a broken initial toggle state (FT8U232/FT232BM).

dwc_otg: implement tasklet for returning URBs to usbcore hcd layer

The dwc_otg driver interrupt handler for transfer completion will spend
a very long time with interrupts disabled when a URB is completed -
this is because usb_hcd_giveback_urb is called from within the handler
which for a USB device driver with complicated processing (e.g. webcam)
will take an exorbitant amount of time to complete. This results in
missed completion interrupts for other USB packets which lead to them
being dropped due to microframe overruns.

This patch splits returning the URB to the usb hcd layer into a
high-priority tasklet. This will have most benefit for isochronous IN
transfers but will also have incidental benefit where multiple periodic
devices are active at once.

dwc_otg: fix NAK holdoff and allow on split transactions only

This corrects a bug where if a single active non-periodic endpoint
had at least one transaction in its qh, on frnum == MAX_FRNUM the qh
would get skipped and never get queued again. This would result in
a silent device until error detection (automatic or otherwise) would
either reset the device or flush and requeue the URBs.

Additionally the NAK holdoff was enabled for all transactions - this
would potentially stall a HS endpoint for 1ms if a previous error state
enabled this interrupt and the next response was a NAK. Fix so that
only split transactions get held off.

dwc_otg: Call usb_hcd_unlink_urb_from_ep with lock held in completion handler

usb_hcd_unlink_urb_from_ep must be called with the HCD lock held.  Calling it
asynchronously in the tasklet was not safe (regression in
c4564d4).

This change unlinks it from the endpoint prior to queueing it for handling in
the tasklet, and also adds a check to ensure the urb is OK to be unlinked
before doing so.

NULL pointer dereference kernel oopses had been observed in usb_hcd_giveback_urb
when a USB device was unplugged/replugged during data transfer.  This effect
was reproduced using automated USB port power control, hundreds of replug
events were performed during active transfers to confirm that the problem was
eliminated.

USB fix using a FIQ to implement split transactions

This commit adds a FIQ implementaion that schedules
the split transactions using a FIQ so we don't get
held off by the interrupt latency of Linux

dwc_otg: fix device attributes and avoid kernel warnings on boot

dcw_otg: avoid logging function that can cause panics

See: raspberrypi/firmware#21
Thanks to cleverca22 for fix

dwc_otg: mask correct interrupts after transaction error recovery

The dwc_otg driver will unmask certain interrupts on a transaction
that previously halted in the error state in order to reset the
QTD error count. The various fine-grained interrupt handlers do not
consider that other interrupts besides themselves were unmasked.

By disabling the two other interrupts only ever enabled in DMA mode
for this purpose, we can avoid unnecessary function calls in the
IRQ handler. This will also prevent an unneccesary FIQ interrupt
from being generated if the FIQ is enabled.

dwc_otg: fiq: prevent FIQ thrash and incorrect state passing to IRQ

In the case of a transaction to a device that had previously aborted
due to an error, several interrupts are enabled to reset the error
count when a device responds. This has the side-effect of making the
FIQ thrash because the hardware will generate multiple instances of
a NAK on an IN bulk/interrupt endpoint and multiple instances of ACK
on an OUT bulk/interrupt endpoint. Make the FIQ mask and clear the
associated interrupts.

Additionally, on non-split transactions make sure that only unmasked
interrupts are cleared. This caused a hard-to-trigger but serious
race condition when you had the combination of an endpoint awaiting
error recovery and a transaction completed on an endpoint - due to
the sequencing and timing of interrupts generated by the dwc_otg core,
it was possible to confuse the IRQ handler.

Fix function tracing

dwc_otg: whitespace cleanup in dwc_otg_urb_enqueue

dwc_otg: prevent OOPSes during device disconnects

The dwc_otg_urb_enqueue function is thread-unsafe. In particular the
access of urb->hcpriv, usb_hcd_link_urb_to_ep, dwc_otg_urb->qtd and
friends does not occur within a critical section and so if a device
was unplugged during activity there was a high chance that the
usbcore hub_thread would try to disable the endpoint with partially-
formed entries in the URB queue. This would result in BUG() or null
pointer dereferences.

Fix so that access of urb->hcpriv, enqueuing to the hardware and
adding to usbcore endpoint URB lists is contained within a single
critical section.

dwc_otg: prevent BUG() in TT allocation if hub address is > 16

A fixed-size array is used to track TT allocation. This was
previously set to 16 which caused a crash because
dwc_otg_hcd_allocate_port would read past the end of the array.

This was hit if a hub was plugged in which enumerated as addr > 16,
due to previous device resets or unplugs.

Also add #ifdef FIQ_DEBUG around hcd->hub_port_alloc[], which grows
to a large size if 128 hub addresses are supported. This field is
for debug only for tracking which frame an allocate happened in.

dwc_otg: make channel halts with unknown state less damaging

If the IRQ received a channel halt interrupt through the FIQ
with no other bits set, the IRQ would not release the host
channel and never complete the URB.

Add catchall handling to treat as a transaction error and retry.

dwc_otg: fiq_split: use TTs with more granularity

This fixes certain issues with split transaction scheduling.

- Isochronous multi-packet OUT transactions now hog the TT until
  they are completed - this prevents hubs aborting transactions
  if they get a periodic start-split out-of-order
- Don't perform TT allocation on non-periodic endpoints - this
  allows simultaneous use of the TT's bulk/control and periodic
  transaction buffers

This commit will mainly affect USB audio playback.

dwc_otg: fix potential sleep while atomic during urb enqueue

Fixes a regression introduced with