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.

Sign up for GitHub

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

Jump to bottom

Kernel panic switching netvm on Linux 5.16.5 #7257

Closed
pgerber opened this issue Feb 10, 2022 · 11 comments
Closed

Kernel panic switching netvm on Linux 5.16.5 #7257

pgerber opened this issue Feb 10, 2022 · 11 comments
Labels
C: kernel diagnosed Technical diagnosis has been performed (see issue comments). P: default Priority: default. Default priority for new issues, to be replaced given sufficient information. r4.0-dom0-stable r4.1-dom0-stable r4.2-host-stable T: bug Type: bug report. A problem or defect resulting in unintended behavior in something that exists.
Milestone
Release 4.0 updates

Comments

@pgerber
Copy link

pgerber commented Feb 10, 2022

How to file a helpful issue

Qubes OS release

$ cat /etc/qubes-release 
Qubes release 4.0 (R4.0)

Brief summary

Kernel panic trying to switch netvm on Linux kernel 5.16.5-1.fc25.qubes.x86_64 from kernel-latest-qubes-vm package.

Steps to reproduce

  1. start vm (qvm-start example)
  2. switch netvm (qvm-prefs example netvm example-netvm)

Expected behavior

VM is attached to newly selected netvm.

Actual behavior

Kernel crashes:

[ 2017.626976] BUG: kernel NULL pointer dereference, address: 0000000000000000
[ 2017.626997] #PF: supervisor read access in kernel mode
[ 2017.627009] #PF: error_code(0x0000) - not-present page
[ 2017.627022] PGD 0 P4D 0 
[ 2017.627030] Oops: 0000 [#1] PREEMPT SMP PTI
[ 2017.627041] CPU: 0 PID: 63 Comm: xenwatch Tainted: G           O      5.16.5-1.fc25.qubes.x86_64 #1
[ 2017.627062] RIP: 0010:free_netdev+0xb2/0x180
[ 2017.627077] Code: 8b 45 50 48 8b 08 49 39 c4 48 8d b8 a0 fe ff ff 48 8d 99 a0 fe ff ff 74 27 e8 3a fe ff ff e8 45 c4 ff ff 48 8d 83 60 01 00 00 <48> 8b 93 60 01 00 00 48 89 df 4c 39 e0 48 8d 9a a0 fe ff ff 75 d9
[ 2017.627117] RSP: 0018:ffffc90000357cf0 EFLAGS: 00010282
[ 2017.627129] RAX: 0000000000000000 RBX: fffffffffffffea0 RCX: 0000000000000002
[ 2017.627147] RDX: 0000000000000000 RSI: ffffc90000357c20 RDI: 00000000ffffffff
[ 2017.627164] RBP: ffff8880a73a3000 R08: 0000000000000001 R09: 0000000000000000
[ 2017.627182] R10: ffffffff82865480 R11: 0000000000000000 R12: ffff8880a73a3050
[ 2017.627199] R13: ffffffffc03ce180 R14: dead000000000100 R15: 0000000000000000
[ 2017.627217] FS:  0000000000000000(0000) GS:ffff8882e3200000(0000) knlGS:0000000000000000
[ 2017.627236] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 2017.627250] CR2: 0000000000000000 CR3: 00000000d59b2002 CR4: 00000000003706f0
[ 2017.627272] Call Trace:
[ 2017.627280]  <TASK>
[ 2017.627289]  xennet_remove+0x275/0x2f0 [xen_netfront]
[ 2017.627308]  xenbus_dev_remove+0x6e/0xf0
[ 2017.627320]  device_release_driver_internal+0xfa/0x1d0
[ 2017.627335]  bus_remove_device+0xd8/0x140
[ 2017.627349]  device_del+0x18d/0x3c0
[ 2017.627359]  ? _raw_spin_unlock+0x16/0x30
[ 2017.627370]  ? klist_iter_exit+0x14/0x20
[ 2017.627382]  ? xenbus_probe_devices+0x120/0x120
[ 2017.627395]  device_unregister+0x13/0x60
[ 2017.627406]  xenbus_dev_changed+0xce/0x210
[ 2017.627417]  xenwatch_thread+0xdd/0x1b0
[ 2017.627428]  ? remove_wait_queue+0x60/0x60
[ 2017.627438]  ? find_watch+0x40/0x40
[ 2017.627448]  kthread+0x14b/0x190
[ 2017.627459]  ? set_kthread_struct+0x40/0x40
[ 2017.627470]  ret_from_fork+0x22/0x30
[ 2017.627483]  </TASK>
[ 2017.627489] Modules linked in: nf_conntrack_netlink xt_addrtype br_netfilter bridge stp llc ip6t_REJECT nf_reject_ipv6 ip6_tables ipt_REJECT nf_reject_ipv4 xt_state xt_conntrack nft_counter nft_chain_nat xt_MASQUERADE nf_nat nf_conntrack nf_defrag_ipv6 nf_defrag_ipv4 nft_compat nf_tables nfnetlink intel_rapl_msr intel_rapl_common crct10dif_pclmul crc32_pclmul crc32c_intel ghash_clmulni_intel xen_netfront snd_pcm snd_timer snd soundcore pcspkr xen_netback u2mfn(O) xen_gntdev xen_gntalloc xen_blkback xen_evtchn parport_pc ppdev lp parport drm fuse sunrpc xenfs xen_privcmd bpf_preload ip_tables overlay xen_blkfront
[ 2017.627629] CR2: 0000000000000000
[ 2017.627641] ---[ end trace d83f7ae915afd12e ]---
[ 2017.627653] RIP: 0010:free_netdev+0xb2/0x180
[ 2017.627667] Code: 8b 45 50 48 8b 08 49 39 c4 48 8d b8 a0 fe ff ff 48 8d 99 a0 fe ff ff 74 27 e8 3a fe ff ff e8 45 c4 ff ff 48 8d 83 60 01 00 00 <48> 8b 93 60 01 00 00 48 89 df 4c 39 e0 48 8d 9a a0 fe ff ff 75 d9
[ 2017.627706] RSP: 0018:ffffc90000357cf0 EFLAGS: 00010282
[ 2017.627719] RAX: 0000000000000000 RBX: fffffffffffffea0 RCX: 0000000000000002
[ 2017.627736] RDX: 0000000000000000 RSI: ffffc90000357c20 RDI: 00000000ffffffff
[ 2017.627753] RBP: ffff8880a73a3000 R08: 0000000000000001 R09: 0000000000000000
[ 2017.627771] R10: ffffffff82865480 R11: 0000000000000000 R12: ffff8880a73a3050
[ 2017.627788] R13: ffffffffc03ce180 R14: dead000000000100 R15: 0000000000000000
[ 2017.627806] FS:  0000000000000000(0000) GS:ffff8882e3200000(0000) knlGS:0000000000000000
[ 2017.627823] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 2017.627839] CR2: 0000000000000000 CR3: 00000000d59b2002 CR4: 00000000003706f0
[ 2017.627856] Kernel panic - not syncing: Fatal exception
[ 2017.628402] Kernel Offset: disabled
@pgerber pgerber added P: default Priority: default. Default priority for new issues, to be replaced given sufficient information. T: bug Type: bug report. A problem or defect resulting in unintended behavior in something that exists. labels Feb 10, 2022
@andrewdavidwong andrewdavidwong added C: kernel needs diagnosis Requires technical diagnosis from developer. Replace with "diagnosed" or remove if otherwise closed. labels Feb 10, 2022
@andrewdavidwong andrewdavidwong added this to the Release 4.0 updates milestone Feb 10, 2022
This was referenced Feb 12, 2022
Closed
Closed
Closed
@marmarek
Copy link
Member

Patch posted: https://lore.kernel.org/xen-devel/20220220134202.2187485-1-marmarek@invisiblethingslab.com/T/#u

fengguang pushed a commit to 0day-ci/linux that referenced this issue Feb 20, 2022
@marmarek @intel-lab-lkp
xen/netfront: destroy queues before real_num_tx_queues is zeroed
b7fd9c4 
xennet_destroy_queues() relies on info->netdev->real_num_tx_queues to
delete queues. Since d7dac08
("net-sysfs: update the queue counts in the unregistration path"),
unregister_netdev() indirectly sets real_num_tx_queues to 0. Those two
facts together means, that xennet_destroy_queues() called from
xennet_remove() cannot do its job, because it's called after
unregister_netdev(). This results in kfree-ing queues that are still
linked in napi, which ultimately crashes:

    BUG: kernel NULL pointer dereference, address: 0000000000000000
    #PF: supervisor read access in kernel mode
    #PF: error_code(0x0000) - not-present page
    PGD 0 P4D 0
    Oops: 0000 [#1] PREEMPT SMP PTI
    CPU: 1 PID: 52 Comm: xenwatch Tainted: G        W         5.16.10-1.32.fc32.qubes.x86_64+ torvalds#226
    RIP: 0010:free_netdev+0xa3/0x1a0
    Code: ff 48 89 df e8 2e e9 00 00 48 8b 43 50 48 8b 08 48 8d b8 a0 fe ff ff 48 8d a9 a0 fe ff ff 49 39 c4 75 26 eb 47 e8 ed c1 66 ff <48> 8b 85 60 01 00 00 48 8d 95 60 01 00 00 48 89 ef 48 2d 60 01 00
    RSP: 0000:ffffc90000bcfd00 EFLAGS: 00010286
    RAX: 0000000000000000 RBX: ffff88800edad000 RCX: 0000000000000000
    RDX: 0000000000000001 RSI: ffffc90000bcfc30 RDI: 00000000ffffffff
    RBP: fffffffffffffea0 R08: 0000000000000000 R09: 0000000000000000
    R10: 0000000000000000 R11: 0000000000000001 R12: ffff88800edad050
    R13: ffff8880065f8f88 R14: 0000000000000000 R15: ffff8880066c6680
    FS:  0000000000000000(0000) GS:ffff8880f3300000(0000) knlGS:0000000000000000
    CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
    CR2: 0000000000000000 CR3: 00000000e998c006 CR4: 00000000003706e0
    Call Trace:
     <TASK>
     xennet_remove+0x13d/0x300 [xen_netfront]
     xenbus_dev_remove+0x6d/0xf0
     __device_release_driver+0x17a/0x240
     device_release_driver+0x24/0x30
     bus_remove_device+0xd8/0x140
     device_del+0x18b/0x410
     ? _raw_spin_unlock+0x16/0x30
     ? klist_iter_exit+0x14/0x20
     ? xenbus_dev_request_and_reply+0x80/0x80
     device_unregister+0x13/0x60
     xenbus_dev_changed+0x18e/0x1f0
     xenwatch_thread+0xc0/0x1a0
     ? do_wait_intr_irq+0xa0/0xa0
     kthread+0x16b/0x190
     ? set_kthread_struct+0x40/0x40
     ret_from_fork+0x22/0x30
     </TASK>

Fix this by calling xennet_destroy_queues() from xennet_close() too,
when real_num_tx_queues is still available. This ensures that queues are
destroyed when real_num_tx_queues is set to 0, regardless of how
unregister_netdev() was called.

Originally reported at
QubesOS/qubes-issues#7257

Fixes: d7dac08 ("net-sysfs: update the queue counts in the unregistration path")
Cc: stable@vger.kernel.org # 5.16+
Signed-off-by: Marek Marczykowski-Górecki <marmarek@invisiblethingslab.com>
@andrewdavidwong andrewdavidwong added diagnosed Technical diagnosis has been performed (see issue comments). waiting for upstream This issue is waiting for something from an upstream project to arrive in Qubes. Remove when closed. and removed needs diagnosis Requires technical diagnosis from developer. Replace with "diagnosed" or remove if otherwise closed. labels Feb 20, 2022
@DemiMarie
Copy link

I get what looks like the same panic on Linux 5.10.96

@marmarek
Copy link
Member

I get what looks like the same panic on Linux 5.10.96

Indeed the bad commit was backported to 5.10.94 too

@DemiMarie
Copy link

I get what looks like the same panic on Linux 5.10.96

Indeed the bad commit was backported to 5.10.94 too

Time for an out-of-tree patch until there is an upstream fix?

ammarfaizi2 pushed a commit to ammarfaizi2/linux-block that referenced this issue Feb 25, 2022
@marmarek @davem330
xen/netfront: destroy queues before real_num_tx_queues is zeroed
dcf4ff7 
xennet_destroy_queues() relies on info->netdev->real_num_tx_queues to
delete queues. Since d7dac08
("net-sysfs: update the queue counts in the unregistration path"),
unregister_netdev() indirectly sets real_num_tx_queues to 0. Those two
facts together means, that xennet_destroy_queues() called from
xennet_remove() cannot do its job, because it's called after
unregister_netdev(). This results in kfree-ing queues that are still
linked in napi, which ultimately crashes:

    BUG: kernel NULL pointer dereference, address: 0000000000000000
    #PF: supervisor read access in kernel mode
    #PF: error_code(0x0000) - not-present page
    PGD 0 P4D 0
    Oops: 0000 [#1] PREEMPT SMP PTI
    CPU: 1 PID: 52 Comm: xenwatch Tainted: G        W         5.16.10-1.32.fc32.qubes.x86_64+ #226
    RIP: 0010:free_netdev+0xa3/0x1a0
    Code: ff 48 89 df e8 2e e9 00 00 48 8b 43 50 48 8b 08 48 8d b8 a0 fe ff ff 48 8d a9 a0 fe ff ff 49 39 c4 75 26 eb 47 e8 ed c1 66 ff <48> 8b 85 60 01 00 00 48 8d 95 60 01 00 00 48 89 ef 48 2d 60 01 00
    RSP: 0000:ffffc90000bcfd00 EFLAGS: 00010286
    RAX: 0000000000000000 RBX: ffff88800edad000 RCX: 0000000000000000
    RDX: 0000000000000001 RSI: ffffc90000bcfc30 RDI: 00000000ffffffff
    RBP: fffffffffffffea0 R08: 0000000000000000 R09: 0000000000000000
    R10: 0000000000000000 R11: 0000000000000001 R12: ffff88800edad050
    R13: ffff8880065f8f88 R14: 0000000000000000 R15: ffff8880066c6680
    FS:  0000000000000000(0000) GS:ffff8880f3300000(0000) knlGS:0000000000000000
    CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
    CR2: 0000000000000000 CR3: 00000000e998c006 CR4: 00000000003706e0
    Call Trace:
     <TASK>
     xennet_remove+0x13d/0x300 [xen_netfront]
     xenbus_dev_remove+0x6d/0xf0
     __device_release_driver+0x17a/0x240
     device_release_driver+0x24/0x30
     bus_remove_device+0xd8/0x140
     device_del+0x18b/0x410
     ? _raw_spin_unlock+0x16/0x30
     ? klist_iter_exit+0x14/0x20
     ? xenbus_dev_request_and_reply+0x80/0x80
     device_unregister+0x13/0x60
     xenbus_dev_changed+0x18e/0x1f0
     xenwatch_thread+0xc0/0x1a0
     ? do_wait_intr_irq+0xa0/0xa0
     kthread+0x16b/0x190
     ? set_kthread_struct+0x40/0x40
     ret_from_fork+0x22/0x30
     </TASK>

Fix this by calling xennet_destroy_queues() from xennet_uninit(),
when real_num_tx_queues is still available. This ensures that queues are
destroyed when real_num_tx_queues is set to 0, regardless of how
unregister_netdev() was called.

Originally reported at
QubesOS/qubes-issues#7257

Fixes: d7dac08 ("net-sysfs: update the queue counts in the unregistration path")
Cc: stable@vger.kernel.org
Signed-off-by: Marek Marczykowski-Górecki <marmarek@invisiblethingslab.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
@qubesos-bot
Copy link

Automated announcement from builder-github

The component linux-kernel-latest (including package kernel-latest-5.16.11-1.fc25.qubes) has been pushed to the r4.0 testing repository for dom0.
To test this update, please install it with the following command:

sudo qubes-dom0-update --enablerepo=qubes-dom0-current-testing

Changes included in this update

@qubesos-bot qubesos-bot mentioned this issue Feb 26, 2022
Closed
@qubesos-bot
Copy link

Automated announcement from builder-github

The component linux-kernel-latest (including package kernel-latest-5.16.11-1.fc32.qubes) has been pushed to the r4.1 testing repository for dom0.
To test this update, please install it with the following command:

sudo qubes-dom0-update --enablerepo=qubes-dom0-current-testing

Changes included in this update

@qubesos-bot qubesos-bot mentioned this issue Feb 26, 2022
Closed
@andrewdavidwong andrewdavidwong removed the waiting for upstream This issue is waiting for something from an upstream project to arrive in Qubes. Remove when closed. label Feb 26, 2022
@DemiMarie DemiMarie mentioned this issue Mar 3, 2022
Closed
@petaramesh
Copy link

Hi,

Confirming that installing the latest kernel updates from the testing repo as mentioned above solves issue mentioned in duplicate bug #7318 for me.

Thanks :)

@DemiMarie DemiMarie added the backport pending On closed issues, indicates fix released for newer version will be backported to older version. label Mar 3, 2022
ammarfaizi2 pushed a commit to ammarfaizi2/linux-block that referenced this issue Mar 5, 2022
@marmarek
xen/netfront: destroy queues before real_num_tx_queues is zeroed
ce4e6a5 
commit dcf4ff7 upstream.

xennet_destroy_queues() relies on info->netdev->real_num_tx_queues to
delete queues. Since d7dac08
("net-sysfs: update the queue counts in the unregistration path"),
unregister_netdev() indirectly sets real_num_tx_queues to 0. Those two
facts together means, that xennet_destroy_queues() called from
xennet_remove() cannot do its job, because it's called after
unregister_netdev(). This results in kfree-ing queues that are still
linked in napi, which ultimately crashes:

    BUG: kernel NULL pointer dereference, address: 0000000000000000
    #PF: supervisor read access in kernel mode
    #PF: error_code(0x0000) - not-present page
    PGD 0 P4D 0
    Oops: 0000 [#1] PREEMPT SMP PTI
    CPU: 1 PID: 52 Comm: xenwatch Tainted: G        W         5.16.10-1.32.fc32.qubes.x86_64+ #226
    RIP: 0010:free_netdev+0xa3/0x1a0
    Code: ff 48 89 df e8 2e e9 00 00 48 8b 43 50 48 8b 08 48 8d b8 a0 fe ff ff 48 8d a9 a0 fe ff ff 49 39 c4 75 26 eb 47 e8 ed c1 66 ff <48> 8b 85 60 01 00 00 48 8d 95 60 01 00 00 48 89 ef 48 2d 60 01 00
    RSP: 0000:ffffc90000bcfd00 EFLAGS: 00010286
    RAX: 0000000000000000 RBX: ffff88800edad000 RCX: 0000000000000000
    RDX: 0000000000000001 RSI: ffffc90000bcfc30 RDI: 00000000ffffffff
    RBP: fffffffffffffea0 R08: 0000000000000000 R09: 0000000000000000
    R10: 0000000000000000 R11: 0000000000000001 R12: ffff88800edad050
    R13: ffff8880065f8f88 R14: 0000000000000000 R15: ffff8880066c6680
    FS:  0000000000000000(0000) GS:ffff8880f3300000(0000) knlGS:0000000000000000
    CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
    CR2: 0000000000000000 CR3: 00000000e998c006 CR4: 00000000003706e0
    Call Trace:
     <TASK>
     xennet_remove+0x13d/0x300 [xen_netfront]
     xenbus_dev_remove+0x6d/0xf0
     __device_release_driver+0x17a/0x240
     device_release_driver+0x24/0x30
     bus_remove_device+0xd8/0x140
     device_del+0x18b/0x410
     ? _raw_spin_unlock+0x16/0x30
     ? klist_iter_exit+0x14/0x20
     ? xenbus_dev_request_and_reply+0x80/0x80
     device_unregister+0x13/0x60
     xenbus_dev_changed+0x18e/0x1f0
     xenwatch_thread+0xc0/0x1a0
     ? do_wait_intr_irq+0xa0/0xa0
     kthread+0x16b/0x190
     ? set_kthread_struct+0x40/0x40
     ret_from_fork+0x22/0x30
     </TASK>

Fix this by calling xennet_destroy_queues() from xennet_uninit(),
when real_num_tx_queues is still available. This ensures that queues are
destroyed when real_num_tx_queues is set to 0, regardless of how
unregister_netdev() was called.

Originally reported at
QubesOS/qubes-issues#7257

Fixes: d7dac08 ("net-sysfs: update the queue counts in the unregistration path")
Cc: stable@vger.kernel.org
Signed-off-by: Marek Marczykowski-Górecki <marmarek@invisiblethingslab.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
ammarfaizi2 pushed a commit to ammarfaizi2/linux-block that referenced this issue Mar 5, 2022
@marmarek
xen/netfront: destroy queues before real_num_tx_queues is zeroed
2824eef 
commit dcf4ff7 upstream.

xennet_destroy_queues() relies on info->netdev->real_num_tx_queues to
delete queues. Since d7dac08
("net-sysfs: update the queue counts in the unregistration path"),
unregister_netdev() indirectly sets real_num_tx_queues to 0. Those two
facts together means, that xennet_destroy_queues() called from
xennet_remove() cannot do its job, because it's called after
unregister_netdev(). This results in kfree-ing queues that are still
linked in napi, which ultimately crashes:

    BUG: kernel NULL pointer dereference, address: 0000000000000000
    #PF: supervisor read access in kernel mode
    #PF: error_code(0x0000) - not-present page
    PGD 0 P4D 0
    Oops: 0000 [#1] PREEMPT SMP PTI
    CPU: 1 PID: 52 Comm: xenwatch Tainted: G        W         5.16.10-1.32.fc32.qubes.x86_64+ #226
    RIP: 0010:free_netdev+0xa3/0x1a0
    Code: ff 48 89 df e8 2e e9 00 00 48 8b 43 50 48 8b 08 48 8d b8 a0 fe ff ff 48 8d a9 a0 fe ff ff 49 39 c4 75 26 eb 47 e8 ed c1 66 ff <48> 8b 85 60 01 00 00 48 8d 95 60 01 00 00 48 89 ef 48 2d 60 01 00
    RSP: 0000:ffffc90000bcfd00 EFLAGS: 00010286
    RAX: 0000000000000000 RBX: ffff88800edad000 RCX: 0000000000000000
    RDX: 0000000000000001 RSI: ffffc90000bcfc30 RDI: 00000000ffffffff
    RBP: fffffffffffffea0 R08: 0000000000000000 R09: 0000000000000000
    R10: 0000000000000000 R11: 0000000000000001 R12: ffff88800edad050
    R13: ffff8880065f8f88 R14: 0000000000000000 R15: ffff8880066c6680
    FS:  0000000000000000(0000) GS:ffff8880f3300000(0000) knlGS:0000000000000000
    CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
    CR2: 0000000000000000 CR3: 00000000e998c006 CR4: 00000000003706e0
    Call Trace:
     <TASK>
     xennet_remove+0x13d/0x300 [xen_netfront]
     xenbus_dev_remove+0x6d/0xf0
     __device_release_driver+0x17a/0x240
     device_release_driver+0x24/0x30
     bus_remove_device+0xd8/0x140
     device_del+0x18b/0x410
     ? _raw_spin_unlock+0x16/0x30
     ? klist_iter_exit+0x14/0x20
     ? xenbus_dev_request_and_reply+0x80/0x80
     device_unregister+0x13/0x60
     xenbus_dev_changed+0x18e/0x1f0
     xenwatch_thread+0xc0/0x1a0
     ? do_wait_intr_irq+0xa0/0xa0
     kthread+0x16b/0x190
     ? set_kthread_struct+0x40/0x40
     ret_from_fork+0x22/0x30
     </TASK>

Fix this by calling xennet_destroy_queues() from xennet_uninit(),
when real_num_tx_queues is still available. This ensures that queues are
destroyed when real_num_tx_queues is set to 0, regardless of how
unregister_netdev() was called.

Originally reported at
QubesOS/qubes-issues#7257

Fixes: d7dac08 ("net-sysfs: update the queue counts in the unregistration path")
Cc: stable@vger.kernel.org
Signed-off-by: Marek Marczykowski-Górecki <marmarek@invisiblethingslab.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
DorianRudolph pushed a commit to DorianRudolph/linux that referenced this issue Mar 21, 2022
@marmarek @DorianRudolph
xen/netfront: destroy queues before real_num_tx_queues is zeroed
cf97f94 
xennet_destroy_queues() relies on info->netdev->real_num_tx_queues to
delete queues. Since d7dac08
("net-sysfs: update the queue counts in the unregistration path"),
unregister_netdev() indirectly sets real_num_tx_queues to 0. Those two
facts together means, that xennet_destroy_queues() called from
xennet_remove() cannot do its job, because it's called after
unregister_netdev(). This results in kfree-ing queues that are still
linked in napi, which ultimately crashes:

    BUG: kernel NULL pointer dereference, address: 0000000000000000
    #PF: supervisor read access in kernel mode
    #PF: error_code(0x0000) - not-present page
    PGD 0 P4D 0
    Oops: 0000 [#1] PREEMPT SMP PTI
    CPU: 1 PID: 52 Comm: xenwatch Tainted: G        W         5.16.10-1.32.fc32.qubes.x86_64+ torvalds#226
    RIP: 0010:free_netdev+0xa3/0x1a0
    Code: ff 48 89 df e8 2e e9 00 00 48 8b 43 50 48 8b 08 48 8d b8 a0 fe ff ff 48 8d a9 a0 fe ff ff 49 39 c4 75 26 eb 47 e8 ed c1 66 ff <48> 8b 85 60 01 00 00 48 8d 95 60 01 00 00 48 89 ef 48 2d 60 01 00
    RSP: 0000:ffffc90000bcfd00 EFLAGS: 00010286
    RAX: 0000000000000000 RBX: ffff88800edad000 RCX: 0000000000000000
    RDX: 0000000000000001 RSI: ffffc90000bcfc30 RDI: 00000000ffffffff
    RBP: fffffffffffffea0 R08: 0000000000000000 R09: 0000000000000000
    R10: 0000000000000000 R11: 0000000000000001 R12: ffff88800edad050
    R13: ffff8880065f8f88 R14: 0000000000000000 R15: ffff8880066c6680
    FS:  0000000000000000(0000) GS:ffff8880f3300000(0000) knlGS:0000000000000000
    CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
    CR2: 0000000000000000 CR3: 00000000e998c006 CR4: 00000000003706e0
    Call Trace:
     <TASK>
     xennet_remove+0x13d/0x300 [xen_netfront]
     xenbus_dev_remove+0x6d/0xf0
     __device_release_driver+0x17a/0x240
     device_release_driver+0x24/0x30
     bus_remove_device+0xd8/0x140
     device_del+0x18b/0x410
     ? _raw_spin_unlock+0x16/0x30
     ? klist_iter_exit+0x14/0x20
     ? xenbus_dev_request_and_reply+0x80/0x80
     device_unregister+0x13/0x60
     xenbus_dev_changed+0x18e/0x1f0
     xenwatch_thread+0xc0/0x1a0
     ? do_wait_intr_irq+0xa0/0xa0
     kthread+0x16b/0x190
     ? set_kthread_struct+0x40/0x40
     ret_from_fork+0x22/0x30
     </TASK>

Fix this by calling xennet_destroy_queues() from xennet_uninit(),
when real_num_tx_queues is still available. This ensures that queues are
destroyed when real_num_tx_queues is set to 0, regardless of how
unregister_netdev() was called.

Originally reported at
QubesOS/qubes-issues#7257

Fixes: d7dac08 ("net-sysfs: update the queue counts in the unregistration path")
Cc: stable@vger.kernel.org
Signed-off-by: Marek Marczykowski-Górecki <marmarek@invisiblethingslab.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
firebird11 pushed a commit to crdroidandroid/android_kernel_oneplus_sm8250 that referenced this issue Mar 22, 2022
@marmarek @firebird11
xen/netfront: destroy queues before real_num_tx_queues is zeroed
b0f1efe 
commit dcf4ff7a48e7598e6b10126cc02177abb8ae4f3f upstream.

xennet_destroy_queues() relies on info->netdev->real_num_tx_queues to
delete queues. Since d7dac083414eb5bb99a6d2ed53dc2c1b405224e5
("net-sysfs: update the queue counts in the unregistration path"),
unregister_netdev() indirectly sets real_num_tx_queues to 0. Those two
facts together means, that xennet_destroy_queues() called from
xennet_remove() cannot do its job, because it's called after
unregister_netdev(). This results in kfree-ing queues that are still
linked in napi, which ultimately crashes:

    BUG: kernel NULL pointer dereference, address: 0000000000000000
    #PF: supervisor read access in kernel mode
    #PF: error_code(0x0000) - not-present page
    PGD 0 P4D 0
    Oops: 0000 [#1] PREEMPT SMP PTI
    CPU: 1 PID: 52 Comm: xenwatch Tainted: G        W         5.16.10-1.32.fc32.qubes.x86_64+ #226
    RIP: 0010:free_netdev+0xa3/0x1a0
    Code: ff 48 89 df e8 2e e9 00 00 48 8b 43 50 48 8b 08 48 8d b8 a0 fe ff ff 48 8d a9 a0 fe ff ff 49 39 c4 75 26 eb 47 e8 ed c1 66 ff <48> 8b 85 60 01 00 00 48 8d 95 60 01 00 00 48 89 ef 48 2d 60 01 00
    RSP: 0000:ffffc90000bcfd00 EFLAGS: 00010286
    RAX: 0000000000000000 RBX: ffff88800edad000 RCX: 0000000000000000
    RDX: 0000000000000001 RSI: ffffc90000bcfc30 RDI: 00000000ffffffff
    RBP: fffffffffffffea0 R08: 0000000000000000 R09: 0000000000000000
    R10: 0000000000000000 R11: 0000000000000001 R12: ffff88800edad050
    R13: ffff8880065f8f88 R14: 0000000000000000 R15: ffff8880066c6680
    FS:  0000000000000000(0000) GS:ffff8880f3300000(0000) knlGS:0000000000000000
    CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
    CR2: 0000000000000000 CR3: 00000000e998c006 CR4: 00000000003706e0
    Call Trace:
     <TASK>
     xennet_remove+0x13d/0x300 [xen_netfront]
     xenbus_dev_remove+0x6d/0xf0
     __device_release_driver+0x17a/0x240
     device_release_driver+0x24/0x30
     bus_remove_device+0xd8/0x140
     device_del+0x18b/0x410
     ? _raw_spin_unlock+0x16/0x30
     ? klist_iter_exit+0x14/0x20
     ? xenbus_dev_request_and_reply+0x80/0x80
     device_unregister+0x13/0x60
     xenbus_dev_changed+0x18e/0x1f0
     xenwatch_thread+0xc0/0x1a0
     ? do_wait_intr_irq+0xa0/0xa0
     kthread+0x16b/0x190
     ? set_kthread_struct+0x40/0x40
     ret_from_fork+0x22/0x30
     </TASK>

Fix this by calling xennet_destroy_queues() from xennet_uninit(),
when real_num_tx_queues is still available. This ensures that queues are
destroyed when real_num_tx_queues is set to 0, regardless of how
unregister_netdev() was called.

Originally reported at
QubesOS/qubes-issues#7257

Fixes: d7dac083414eb5bb9 ("net-sysfs: update the queue counts in the unregistration path")
Cc: stable@vger.kernel.org
Signed-off-by: Marek Marczykowski-Górecki <marmarek@invisiblethingslab.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: firebird11 <hbgassel@gmail.com>
firebird11 pushed a commit to crdroidandroid/android_kernel_oneplus_sm8250 that referenced this issue Mar 22, 2022
@marmarek @firebird11
xen/netfront: destroy queues before real_num_tx_queues is zeroed
141e942 
commit dcf4ff7a48e7598e6b10126cc02177abb8ae4f3f upstream.

xennet_destroy_queues() relies on info->netdev->real_num_tx_queues to
delete queues. Since d7dac083414eb5bb99a6d2ed53dc2c1b405224e5
("net-sysfs: update the queue counts in the unregistration path"),
unregister_netdev() indirectly sets real_num_tx_queues to 0. Those two
facts together means, that xennet_destroy_queues() called from
xennet_remove() cannot do its job, because it's called after
unregister_netdev(). This results in kfree-ing queues that are still
linked in napi, which ultimately crashes:

    BUG: kernel NULL pointer dereference, address: 0000000000000000
    #PF: supervisor read access in kernel mode
    #PF: error_code(0x0000) - not-present page
    PGD 0 P4D 0
    Oops: 0000 [#1] PREEMPT SMP PTI
    CPU: 1 PID: 52 Comm: xenwatch Tainted: G        W         5.16.10-1.32.fc32.qubes.x86_64+ #226
    RIP: 0010:free_netdev+0xa3/0x1a0
    Code: ff 48 89 df e8 2e e9 00 00 48 8b 43 50 48 8b 08 48 8d b8 a0 fe ff ff 48 8d a9 a0 fe ff ff 49 39 c4 75 26 eb 47 e8 ed c1 66 ff <48> 8b 85 60 01 00 00 48 8d 95 60 01 00 00 48 89 ef 48 2d 60 01 00
    RSP: 0000:ffffc90000bcfd00 EFLAGS: 00010286
    RAX: 0000000000000000 RBX: ffff88800edad000 RCX: 0000000000000000
    RDX: 0000000000000001 RSI: ffffc90000bcfc30 RDI: 00000000ffffffff
    RBP: fffffffffffffea0 R08: 0000000000000000 R09: 0000000000000000
    R10: 0000000000000000 R11: 0000000000000001 R12: ffff88800edad050
    R13: ffff8880065f8f88 R14: 0000000000000000 R15: ffff8880066c6680
    FS:  0000000000000000(0000) GS:ffff8880f3300000(0000) knlGS:0000000000000000
    CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
    CR2: 0000000000000000 CR3: 00000000e998c006 CR4: 00000000003706e0
    Call Trace:
     <TASK>
     xennet_remove+0x13d/0x300 [xen_netfront]
     xenbus_dev_remove+0x6d/0xf0
     __device_release_driver+0x17a/0x240
     device_release_driver+0x24/0x30
     bus_remove_device+0xd8/0x140
     device_del+0x18b/0x410
     ? _raw_spin_unlock+0x16/0x30
     ? klist_iter_exit+0x14/0x20
     ? xenbus_dev_request_and_reply+0x80/0x80
     device_unregister+0x13/0x60
     xenbus_dev_changed+0x18e/0x1f0
     xenwatch_thread+0xc0/0x1a0
     ? do_wait_intr_irq+0xa0/0xa0
     kthread+0x16b/0x190
     ? set_kthread_struct+0x40/0x40
     ret_from_fork+0x22/0x30
     </TASK>

Fix this by calling xennet_destroy_queues() from xennet_uninit(),
when real_num_tx_queues is still available. This ensures that queues are
destroyed when real_num_tx_queues is set to 0, regardless of how
unregister_netdev() was called.

Originally reported at
QubesOS/qubes-issues#7257

Fixes: d7dac083414eb5bb9 ("net-sysfs: update the queue counts in the unregistration path")
Cc: stable@vger.kernel.org
Signed-off-by: Marek Marczykowski-Górecki <marmarek@invisiblethingslab.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: firebird11 <hbgassel@gmail.com>
AlexGhiti pushed a commit to AlexGhiti/riscv-linux that referenced this issue Apr 1, 2022
@marmarek @piso77
xen/netfront: destroy queues before real_num_tx_queues is zeroed
1d0ea12 
BugLink: https://bugs.launchpad.net/bugs/1964361

commit dcf4ff7 upstream.

xennet_destroy_queues() relies on info->netdev->real_num_tx_queues to
delete queues. Since d7dac08
("net-sysfs: update the queue counts in the unregistration path"),
unregister_netdev() indirectly sets real_num_tx_queues to 0. Those two
facts together means, that xennet_destroy_queues() called from
xennet_remove() cannot do its job, because it's called after
unregister_netdev(). This results in kfree-ing queues that are still
linked in napi, which ultimately crashes:

    BUG: kernel NULL pointer dereference, address: 0000000000000000
    #PF: supervisor read access in kernel mode
    #PF: error_code(0x0000) - not-present page
    PGD 0 P4D 0
    Oops: 0000 [riscvarchive#1] PREEMPT SMP PTI
    CPU: 1 PID: 52 Comm: xenwatch Tainted: G        W         5.16.10-1.32.fc32.qubes.x86_64+ #226
    RIP: 0010:free_netdev+0xa3/0x1a0
    Code: ff 48 89 df e8 2e e9 00 00 48 8b 43 50 48 8b 08 48 8d b8 a0 fe ff ff 48 8d a9 a0 fe ff ff 49 39 c4 75 26 eb 47 e8 ed c1 66 ff <48> 8b 85 60 01 00 00 48 8d 95 60 01 00 00 48 89 ef 48 2d 60 01 00
    RSP: 0000:ffffc90000bcfd00 EFLAGS: 00010286
    RAX: 0000000000000000 RBX: ffff88800edad000 RCX: 0000000000000000
    RDX: 0000000000000001 RSI: ffffc90000bcfc30 RDI: 00000000ffffffff
    RBP: fffffffffffffea0 R08: 0000000000000000 R09: 0000000000000000
    R10: 0000000000000000 R11: 0000000000000001 R12: ffff88800edad050
    R13: ffff8880065f8f88 R14: 0000000000000000 R15: ffff8880066c6680
    FS:  0000000000000000(0000) GS:ffff8880f3300000(0000) knlGS:0000000000000000
    CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
    CR2: 0000000000000000 CR3: 00000000e998c006 CR4: 00000000003706e0
    Call Trace:
     <TASK>
     xennet_remove+0x13d/0x300 [xen_netfront]
     xenbus_dev_remove+0x6d/0xf0
     __device_release_driver+0x17a/0x240
     device_release_driver+0x24/0x30
     bus_remove_device+0xd8/0x140
     device_del+0x18b/0x410
     ? _raw_spin_unlock+0x16/0x30
     ? klist_iter_exit+0x14/0x20
     ? xenbus_dev_request_and_reply+0x80/0x80
     device_unregister+0x13/0x60
     xenbus_dev_changed+0x18e/0x1f0
     xenwatch_thread+0xc0/0x1a0
     ? do_wait_intr_irq+0xa0/0xa0
     kthread+0x16b/0x190
     ? set_kthread_struct+0x40/0x40
     ret_from_fork+0x22/0x30
     </TASK>

Fix this by calling xennet_destroy_queues() from xennet_uninit(),
when real_num_tx_queues is still available. This ensures that queues are
destroyed when real_num_tx_queues is set to 0, regardless of how
unregister_netdev() was called.

Originally reported at
QubesOS/qubes-issues#7257

Fixes: d7dac08 ("net-sysfs: update the queue counts in the unregistration path")
Cc: stable@vger.kernel.org
Signed-off-by: Marek Marczykowski-Górecki <marmarek@invisiblethingslab.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Paolo Pisati <paolo.pisati@canonical.com>
DozNaka pushed a commit to DozNaka/KawaKernel-A217X that referenced this issue Apr 11, 2022
@marmarek @DozNaka
xen/netfront: destroy queues before real_num_tx_queues is zeroed
cdc0298 
commit dcf4ff7a48e7598e6b10126cc02177abb8ae4f3f upstream.

xennet_destroy_queues() relies on info->netdev->real_num_tx_queues to
delete queues. Since d7dac083414eb5bb99a6d2ed53dc2c1b405224e5
("net-sysfs: update the queue counts in the unregistration path"),
unregister_netdev() indirectly sets real_num_tx_queues to 0. Those two
facts together means, that xennet_destroy_queues() called from
xennet_remove() cannot do its job, because it's called after
unregister_netdev(). This results in kfree-ing queues that are still
linked in napi, which ultimately crashes:

    BUG: kernel NULL pointer dereference, address: 0000000000000000
    #PF: supervisor read access in kernel mode
    #PF: error_code(0x0000) - not-present page
    PGD 0 P4D 0
    Oops: 0000 [#1] PREEMPT SMP PTI
    CPU: 1 PID: 52 Comm: xenwatch Tainted: G        W         5.16.10-1.32.fc32.qubes.x86_64+ #226
    RIP: 0010:free_netdev+0xa3/0x1a0
    Code: ff 48 89 df e8 2e e9 00 00 48 8b 43 50 48 8b 08 48 8d b8 a0 fe ff ff 48 8d a9 a0 fe ff ff 49 39 c4 75 26 eb 47 e8 ed c1 66 ff <48> 8b 85 60 01 00 00 48 8d 95 60 01 00 00 48 89 ef 48 2d 60 01 00
    RSP: 0000:ffffc90000bcfd00 EFLAGS: 00010286
    RAX: 0000000000000000 RBX: ffff88800edad000 RCX: 0000000000000000
    RDX: 0000000000000001 RSI: ffffc90000bcfc30 RDI: 00000000ffffffff
    RBP: fffffffffffffea0 R08: 0000000000000000 R09: 0000000000000000
    R10: 0000000000000000 R11: 0000000000000001 R12: ffff88800edad050
    R13: ffff8880065f8f88 R14: 0000000000000000 R15: ffff8880066c6680
    FS:  0000000000000000(0000) GS:ffff8880f3300000(0000) knlGS:0000000000000000
    CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
    CR2: 0000000000000000 CR3: 00000000e998c006 CR4: 00000000003706e0
    Call Trace:
     <TASK>
     xennet_remove+0x13d/0x300 [xen_netfront]
     xenbus_dev_remove+0x6d/0xf0
     __device_release_driver+0x17a/0x240
     device_release_driver+0x24/0x30
     bus_remove_device+0xd8/0x140
     device_del+0x18b/0x410
     ? _raw_spin_unlock+0x16/0x30
     ? klist_iter_exit+0x14/0x20
     ? xenbus_dev_request_and_reply+0x80/0x80
     device_unregister+0x13/0x60
     xenbus_dev_changed+0x18e/0x1f0
     xenwatch_thread+0xc0/0x1a0
     ? do_wait_intr_irq+0xa0/0xa0
     kthread+0x16b/0x190
     ? set_kthread_struct+0x40/0x40
     ret_from_fork+0x22/0x30
     </TASK>

Fix this by calling xennet_destroy_queues() from xennet_uninit(),
when real_num_tx_queues is still available. This ensures that queues are
destroyed when real_num_tx_queues is set to 0, regardless of how
unregister_netdev() was called.

Originally reported at
QubesOS/qubes-issues#7257

Fixes: d7dac083414eb5bb9 ("net-sysfs: update the queue counts in the unregistration path")
Cc: stable@vger.kernel.org
Signed-off-by: Marek Marczykowski-Górecki <marmarek@invisiblethingslab.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
borkra pushed a commit to LairdCP/wb-kernel that referenced this issue Apr 20, 2022
@marmarek @borkra
xen/netfront: destroy queues before real_num_tx_queues is zeroed
edc1ae2 
commit dcf4ff7 upstream.

xennet_destroy_queues() relies on info->netdev->real_num_tx_queues to
delete queues. Since d7dac08
("net-sysfs: update the queue counts in the unregistration path"),
unregister_netdev() indirectly sets real_num_tx_queues to 0. Those two
facts together means, that xennet_destroy_queues() called from
xennet_remove() cannot do its job, because it's called after
unregister_netdev(). This results in kfree-ing queues that are still
linked in napi, which ultimately crashes:

    BUG: kernel NULL pointer dereference, address: 0000000000000000
    #PF: supervisor read access in kernel mode
    #PF: error_code(0x0000) - not-present page
    PGD 0 P4D 0
    Oops: 0000 [#1] PREEMPT SMP PTI
    CPU: 1 PID: 52 Comm: xenwatch Tainted: G        W         5.16.10-1.32.fc32.qubes.x86_64+ #226
    RIP: 0010:free_netdev+0xa3/0x1a0
    Code: ff 48 89 df e8 2e e9 00 00 48 8b 43 50 48 8b 08 48 8d b8 a0 fe ff ff 48 8d a9 a0 fe ff ff 49 39 c4 75 26 eb 47 e8 ed c1 66 ff <48> 8b 85 60 01 00 00 48 8d 95 60 01 00 00 48 89 ef 48 2d 60 01 00
    RSP: 0000:ffffc90000bcfd00 EFLAGS: 00010286
    RAX: 0000000000000000 RBX: ffff88800edad000 RCX: 0000000000000000
    RDX: 0000000000000001 RSI: ffffc90000bcfc30 RDI: 00000000ffffffff
    RBP: fffffffffffffea0 R08: 0000000000000000 R09: 0000000000000000
    R10: 0000000000000000 R11: 0000000000000001 R12: ffff88800edad050
    R13: ffff8880065f8f88 R14: 0000000000000000 R15: ffff8880066c6680
    FS:  0000000000000000(0000) GS:ffff8880f3300000(0000) knlGS:0000000000000000
    CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
    CR2: 0000000000000000 CR3: 00000000e998c006 CR4: 00000000003706e0
    Call Trace:
     <TASK>
     xennet_remove+0x13d/0x300 [xen_netfront]
     xenbus_dev_remove+0x6d/0xf0
     __device_release_driver+0x17a/0x240
     device_release_driver+0x24/0x30
     bus_remove_device+0xd8/0x140
     device_del+0x18b/0x410
     ? _raw_spin_unlock+0x16/0x30
     ? klist_iter_exit+0x14/0x20
     ? xenbus_dev_request_and_reply+0x80/0x80
     device_unregister+0x13/0x60
     xenbus_dev_changed+0x18e/0x1f0
     xenwatch_thread+0xc0/0x1a0
     ? do_wait_intr_irq+0xa0/0xa0
     kthread+0x16b/0x190
     ? set_kthread_struct+0x40/0x40
     ret_from_fork+0x22/0x30
     </TASK>

Fix this by calling xennet_destroy_queues() from xennet_uninit(),
when real_num_tx_queues is still available. This ensures that queues are
destroyed when real_num_tx_queues is set to 0, regardless of how
unregister_netdev() was called.

Originally reported at
QubesOS/qubes-issues#7257

Fixes: d7dac08 ("net-sysfs: update the queue counts in the unregistration path")
Cc: stable@vger.kernel.org
Signed-off-by: Marek Marczykowski-Górecki <marmarek@invisiblethingslab.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
@marmarek marmarek removed the backport pending On closed issues, indicates fix released for newer version will be backported to older version. label May 4, 2022
it-is-a-robot pushed a commit to openeuler-mirror/kernel that referenced this issue May 29, 2022
@marmarek
xen/netfront: destroy queues before real_num_tx_queues is zeroed
02ef4c9 
stable inclusion
from stable-v5.10.104
commit a1753d5c29a6fb9a8966dcf04cb4f3b71e303ae8
bugzilla: https://gitee.com/openeuler/kernel/issues/I56XAC

Reference: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id=a1753d5c29a6fb9a8966dcf04cb4f3b71e303ae8

--------------------------------

commit dcf4ff7 upstream.

xennet_destroy_queues() relies on info->netdev->real_num_tx_queues to
delete queues. Since d7dac08
("net-sysfs: update the queue counts in the unregistration path"),
unregister_netdev() indirectly sets real_num_tx_queues to 0. Those two
facts together means, that xennet_destroy_queues() called from
xennet_remove() cannot do its job, because it's called after
unregister_netdev(). This results in kfree-ing queues that are still
linked in napi, which ultimately crashes:

    BUG: kernel NULL pointer dereference, address: 0000000000000000
    #PF: supervisor read access in kernel mode
    #PF: error_code(0x0000) - not-present page
    PGD 0 P4D 0
    Oops: 0000 [#1] PREEMPT SMP PTI
    CPU: 1 PID: 52 Comm: xenwatch Tainted: G        W         5.16.10-1.32.fc32.qubes.x86_64+ #226
    RIP: 0010:free_netdev+0xa3/0x1a0
    Code: ff 48 89 df e8 2e e9 00 00 48 8b 43 50 48 8b 08 48 8d b8 a0 fe ff ff 48 8d a9 a0 fe ff ff 49 39 c4 75 26 eb 47 e8 ed c1 66 ff <48> 8b 85 60 01 00 00 48 8d 95 60 01 00 00 48 89 ef 48 2d 60 01 00
    RSP: 0000:ffffc90000bcfd00 EFLAGS: 00010286
    RAX: 0000000000000000 RBX: ffff88800edad000 RCX: 0000000000000000
    RDX: 0000000000000001 RSI: ffffc90000bcfc30 RDI: 00000000ffffffff
    RBP: fffffffffffffea0 R08: 0000000000000000 R09: 0000000000000000
    R10: 0000000000000000 R11: 0000000000000001 R12: ffff88800edad050
    R13: ffff8880065f8f88 R14: 0000000000000000 R15: ffff8880066c6680
    FS:  0000000000000000(0000) GS:ffff8880f3300000(0000) knlGS:0000000000000000
    CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
    CR2: 0000000000000000 CR3: 00000000e998c006 CR4: 00000000003706e0
    Call Trace:
     <TASK>
     xennet_remove+0x13d/0x300 [xen_netfront]
     xenbus_dev_remove+0x6d/0xf0
     __device_release_driver+0x17a/0x240
     device_release_driver+0x24/0x30
     bus_remove_device+0xd8/0x140
     device_del+0x18b/0x410
     ? _raw_spin_unlock+0x16/0x30
     ? klist_iter_exit+0x14/0x20
     ? xenbus_dev_request_and_reply+0x80/0x80
     device_unregister+0x13/0x60
     xenbus_dev_changed+0x18e/0x1f0
     xenwatch_thread+0xc0/0x1a0
     ? do_wait_intr_irq+0xa0/0xa0
     kthread+0x16b/0x190
     ? set_kthread_struct+0x40/0x40
     ret_from_fork+0x22/0x30
     </TASK>

Fix this by calling xennet_destroy_queues() from xennet_uninit(),
when real_num_tx_queues is still available. This ensures that queues are
destroyed when real_num_tx_queues is set to 0, regardless of how
unregister_netdev() was called.

Originally reported at
QubesOS/qubes-issues#7257

Fixes: d7dac08 ("net-sysfs: update the queue counts in the unregistration path")
Cc: stable@vger.kernel.org
Signed-off-by: Marek Marczykowski-Górecki <marmarek@invisiblethingslab.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Yu Liao <liaoyu15@huawei.com>
Reviewed-by: Wei Li <liwei391@huawei.com>
Signed-off-by: Zheng Zengkai <zhengzengkai@huawei.com>
it-is-a-robot pushed a commit to openeuler-mirror/kernel that referenced this issue May 29, 2022
@marmarek
xen/netfront: destroy queues before real_num_tx_queues is zeroed
5af73b9 
stable inclusion
from stable-v5.10.104
commit a1753d5c29a6fb9a8966dcf04cb4f3b71e303ae8
bugzilla: https://gitee.com/openeuler/kernel/issues/I56XAC

Reference: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id=a1753d5c29a6fb9a8966dcf04cb4f3b71e303ae8

--------------------------------

commit dcf4ff7 upstream.

xennet_destroy_queues() relies on info->netdev->real_num_tx_queues to
delete queues. Since d7dac08
("net-sysfs: update the queue counts in the unregistration path"),
unregister_netdev() indirectly sets real_num_tx_queues to 0. Those two
facts together means, that xennet_destroy_queues() called from
xennet_remove() cannot do its job, because it's called after
unregister_netdev(). This results in kfree-ing queues that are still
linked in napi, which ultimately crashes:

    BUG: kernel NULL pointer dereference, address: 0000000000000000
    #PF: supervisor read access in kernel mode
    #PF: error_code(0x0000) - not-present page
    PGD 0 P4D 0
    Oops: 0000 [#1] PREEMPT SMP PTI
    CPU: 1 PID: 52 Comm: xenwatch Tainted: G        W         5.16.10-1.32.fc32.qubes.x86_64+ #226
    RIP: 0010:free_netdev+0xa3/0x1a0
    Code: ff 48 89 df e8 2e e9 00 00 48 8b 43 50 48 8b 08 48 8d b8 a0 fe ff ff 48 8d a9 a0 fe ff ff 49 39 c4 75 26 eb 47 e8 ed c1 66 ff <48> 8b 85 60 01 00 00 48 8d 95 60 01 00 00 48 89 ef 48 2d 60 01 00
    RSP: 0000:ffffc90000bcfd00 EFLAGS: 00010286
    RAX: 0000000000000000 RBX: ffff88800edad000 RCX: 0000000000000000
    RDX: 0000000000000001 RSI: ffffc90000bcfc30 RDI: 00000000ffffffff
    RBP: fffffffffffffea0 R08: 0000000000000000 R09: 0000000000000000
    R10: 0000000000000000 R11: 0000000000000001 R12: ffff88800edad050
    R13: ffff8880065f8f88 R14: 0000000000000000 R15: ffff8880066c6680
    FS:  0000000000000000(0000) GS:ffff8880f3300000(0000) knlGS:0000000000000000
    CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
    CR2: 0000000000000000 CR3: 00000000e998c006 CR4: 00000000003706e0
    Call Trace:
     <TASK>
     xennet_remove+0x13d/0x300 [xen_netfront]
     xenbus_dev_remove+0x6d/0xf0
     __device_release_driver+0x17a/0x240
     device_release_driver+0x24/0x30
     bus_remove_device+0xd8/0x140
     device_del+0x18b/0x410
     ? _raw_spin_unlock+0x16/0x30
     ? klist_iter_exit+0x14/0x20
     ? xenbus_dev_request_and_reply+0x80/0x80
     device_unregister+0x13/0x60
     xenbus_dev_changed+0x18e/0x1f0
     xenwatch_thread+0xc0/0x1a0
     ? do_wait_intr_irq+0xa0/0xa0
     kthread+0x16b/0x190
     ? set_kthread_struct+0x40/0x40
     ret_from_fork+0x22/0x30
     </TASK>

Fix this by calling xennet_destroy_queues() from xennet_uninit(),
when real_num_tx_queues is still available. This ensures that queues are
destroyed when real_num_tx_queues is set to 0, regardless of how
unregister_netdev() was called.

Originally reported at
QubesOS/qubes-issues#7257

Fixes: d7dac08 ("net-sysfs: update the queue counts in the unregistration path")
Cc: stable@vger.kernel.org
Signed-off-by: Marek Marczykowski-Górecki <marmarek@invisiblethingslab.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Yu Liao <liaoyu15@huawei.com>
Reviewed-by: Wei Li <liwei391@huawei.com>
Signed-off-by: Zheng Zengkai <zhengzengkai@huawei.com>
delphix-devops-bot pushed a commit to delphix/linux-kernel-generic that referenced this issue Jun 10, 2022
@marmarek @smb49
xen/netfront: destroy queues before real_num_tx_queues is zeroed
d1f1d19 
BugLink: https://bugs.launchpad.net/bugs/1969239

commit dcf4ff7a48e7598e6b10126cc02177abb8ae4f3f upstream.

xennet_destroy_queues() relies on info->netdev->real_num_tx_queues to
delete queues. Since d7dac083414eb5bb99a6d2ed53dc2c1b405224e5
("net-sysfs: update the queue counts in the unregistration path"),
unregister_netdev() indirectly sets real_num_tx_queues to 0. Those two
facts together means, that xennet_destroy_queues() called from
xennet_remove() cannot do its job, because it's called after
unregister_netdev(). This results in kfree-ing queues that are still
linked in napi, which ultimately crashes:

    BUG: kernel NULL pointer dereference, address: 0000000000000000
    #PF: supervisor read access in kernel mode
    #PF: error_code(0x0000) - not-present page
    PGD 0 P4D 0
    Oops: 0000 [#1] PREEMPT SMP PTI
    CPU: 1 PID: 52 Comm: xenwatch Tainted: G        W         5.16.10-1.32.fc32.qubes.x86_64+ #226
    RIP: 0010:free_netdev+0xa3/0x1a0
    Code: ff 48 89 df e8 2e e9 00 00 48 8b 43 50 48 8b 08 48 8d b8 a0 fe ff ff 48 8d a9 a0 fe ff ff 49 39 c4 75 26 eb 47 e8 ed c1 66 ff <48> 8b 85 60 01 00 00 48 8d 95 60 01 00 00 48 89 ef 48 2d 60 01 00
    RSP: 0000:ffffc90000bcfd00 EFLAGS: 00010286
    RAX: 0000000000000000 RBX: ffff88800edad000 RCX: 0000000000000000
    RDX: 0000000000000001 RSI: ffffc90000bcfc30 RDI: 00000000ffffffff
    RBP: fffffffffffffea0 R08: 0000000000000000 R09: 0000000000000000
    R10: 0000000000000000 R11: 0000000000000001 R12: ffff88800edad050
    R13: ffff8880065f8f88 R14: 0000000000000000 R15: ffff8880066c6680
    FS:  0000000000000000(0000) GS:ffff8880f3300000(0000) knlGS:0000000000000000
    CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
    CR2: 0000000000000000 CR3: 00000000e998c006 CR4: 00000000003706e0
    Call Trace:
     <TASK>
     xennet_remove+0x13d/0x300 [xen_netfront]
     xenbus_dev_remove+0x6d/0xf0
     __device_release_driver+0x17a/0x240
     device_release_driver+0x24/0x30
     bus_remove_device+0xd8/0x140
     device_del+0x18b/0x410
     ? _raw_spin_unlock+0x16/0x30
     ? klist_iter_exit+0x14/0x20
     ? xenbus_dev_request_and_reply+0x80/0x80
     device_unregister+0x13/0x60
     xenbus_dev_changed+0x18e/0x1f0
     xenwatch_thread+0xc0/0x1a0
     ? do_wait_intr_irq+0xa0/0xa0
     kthread+0x16b/0x190
     ? set_kthread_struct+0x40/0x40
     ret_from_fork+0x22/0x30
     </TASK>

Fix this by calling xennet_destroy_queues() from xennet_uninit(),
when real_num_tx_queues is still available. This ensures that queues are
destroyed when real_num_tx_queues is set to 0, regardless of how
unregister_netdev() was called.

Originally reported at
QubesOS/qubes-issues#7257

Fixes: d7dac083414eb5bb9 ("net-sysfs: update the queue counts in the unregistration path")
Cc: stable@vger.kernel.org
Signed-off-by: Marek Marczykowski-Górecki <marmarek@invisiblethingslab.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Kamal Mostafa <kamal@canonical.com>
Signed-off-by: Stefan Bader <stefan.bader@canonical.com>
hellobbn pushed a commit to hellobbn/android_kernel_sony_sm8250 that referenced this issue Jul 21, 2022
@marmarek @hellobbn
xen/netfront: destroy queues before real_num_tx_queues is zeroed
116aef3 
commit dcf4ff7a48e7598e6b10126cc02177abb8ae4f3f upstream.

xennet_destroy_queues() relies on info->netdev->real_num_tx_queues to
delete queues. Since d7dac083414eb5bb99a6d2ed53dc2c1b405224e5
("net-sysfs: update the queue counts in the unregistration path"),
unregister_netdev() indirectly sets real_num_tx_queues to 0. Those two
facts together means, that xennet_destroy_queues() called from
xennet_remove() cannot do its job, because it's called after
unregister_netdev(). This results in kfree-ing queues that are still
linked in napi, which ultimately crashes:

    BUG: kernel NULL pointer dereference, address: 0000000000000000
    #PF: supervisor read access in kernel mode
    #PF: error_code(0x0000) - not-present page
    PGD 0 P4D 0
    Oops: 0000 [#1] PREEMPT SMP PTI
    CPU: 1 PID: 52 Comm: xenwatch Tainted: G        W         5.16.10-1.32.fc32.qubes.x86_64+ #226
    RIP: 0010:free_netdev+0xa3/0x1a0
    Code: ff 48 89 df e8 2e e9 00 00 48 8b 43 50 48 8b 08 48 8d b8 a0 fe ff ff 48 8d a9 a0 fe ff ff 49 39 c4 75 26 eb 47 e8 ed c1 66 ff <48> 8b 85 60 01 00 00 48 8d 95 60 01 00 00 48 89 ef 48 2d 60 01 00
    RSP: 0000:ffffc90000bcfd00 EFLAGS: 00010286
    RAX: 0000000000000000 RBX: ffff88800edad000 RCX: 0000000000000000
    RDX: 0000000000000001 RSI: ffffc90000bcfc30 RDI: 00000000ffffffff
    RBP: fffffffffffffea0 R08: 0000000000000000 R09: 0000000000000000
    R10: 0000000000000000 R11: 0000000000000001 R12: ffff88800edad050
    R13: ffff8880065f8f88 R14: 0000000000000000 R15: ffff8880066c6680
    FS:  0000000000000000(0000) GS:ffff8880f3300000(0000) knlGS:0000000000000000
    CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
    CR2: 0000000000000000 CR3: 00000000e998c006 CR4: 00000000003706e0
    Call Trace:
     <TASK>
     xennet_remove+0x13d/0x300 [xen_netfront]
     xenbus_dev_remove+0x6d/0xf0
     __device_release_driver+0x17a/0x240
     device_release_driver+0x24/0x30
     bus_remove_device+0xd8/0x140
     device_del+0x18b/0x410
     ? _raw_spin_unlock+0x16/0x30
     ? klist_iter_exit+0x14/0x20
     ? xenbus_dev_request_and_reply+0x80/0x80
     device_unregister+0x13/0x60
     xenbus_dev_changed+0x18e/0x1f0
     xenwatch_thread+0xc0/0x1a0
     ? do_wait_intr_irq+0xa0/0xa0
     kthread+0x16b/0x190
     ? set_kthread_struct+0x40/0x40
     ret_from_fork+0x22/0x30
     </TASK>

Fix this by calling xennet_destroy_queues() from xennet_uninit(),
when real_num_tx_queues is still available. This ensures that queues are
destroyed when real_num_tx_queues is set to 0, regardless of how
unregister_netdev() was called.

Originally reported at
QubesOS/qubes-issues#7257

Fixes: d7dac083414eb5bb9 ("net-sysfs: update the queue counts in the unregistration path")
Cc: stable@vger.kernel.org
Signed-off-by: Marek Marczykowski-Górecki <marmarek@invisiblethingslab.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
xzeldon pushed a commit to xzeldon/IllusionX_sm8250 that referenced this issue Sep 12, 2022
@marmarek @xzeldon
xen/netfront: destroy queues before real_num_tx_queues is zeroed
1754954 
commit dcf4ff7a48e7598e6b10126cc02177abb8ae4f3f upstream.

xennet_destroy_queues() relies on info->netdev->real_num_tx_queues to
delete queues. Since d7dac083414eb5bb99a6d2ed53dc2c1b405224e5
("net-sysfs: update the queue counts in the unregistration path"),
unregister_netdev() indirectly sets real_num_tx_queues to 0. Those two
facts together means, that xennet_destroy_queues() called from
xennet_remove() cannot do its job, because it's called after
unregister_netdev(). This results in kfree-ing queues that are still
linked in napi, which ultimately crashes:

    BUG: kernel NULL pointer dereference, address: 0000000000000000
    #PF: supervisor read access in kernel mode
    #PF: error_code(0x0000) - not-present page
    PGD 0 P4D 0
    Oops: 0000 [#1] PREEMPT SMP PTI
    CPU: 1 PID: 52 Comm: xenwatch Tainted: G        W         5.16.10-1.32.fc32.qubes.x86_64+ #226
    RIP: 0010:free_netdev+0xa3/0x1a0
    Code: ff 48 89 df e8 2e e9 00 00 48 8b 43 50 48 8b 08 48 8d b8 a0 fe ff ff 48 8d a9 a0 fe ff ff 49 39 c4 75 26 eb 47 e8 ed c1 66 ff <48> 8b 85 60 01 00 00 48 8d 95 60 01 00 00 48 89 ef 48 2d 60 01 00
    RSP: 0000:ffffc90000bcfd00 EFLAGS: 00010286
    RAX: 0000000000000000 RBX: ffff88800edad000 RCX: 0000000000000000
    RDX: 0000000000000001 RSI: ffffc90000bcfc30 RDI: 00000000ffffffff
    RBP: fffffffffffffea0 R08: 0000000000000000 R09: 0000000000000000
    R10: 0000000000000000 R11: 0000000000000001 R12: ffff88800edad050
    R13: ffff8880065f8f88 R14: 0000000000000000 R15: ffff8880066c6680
    FS:  0000000000000000(0000) GS:ffff8880f3300000(0000) knlGS:0000000000000000
    CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
    CR2: 0000000000000000 CR3: 00000000e998c006 CR4: 00000000003706e0
    Call Trace:
     <TASK>
     xennet_remove+0x13d/0x300 [xen_netfront]
     xenbus_dev_remove+0x6d/0xf0
     __device_release_driver+0x17a/0x240
     device_release_driver+0x24/0x30
     bus_remove_device+0xd8/0x140
     device_del+0x18b/0x410
     ? _raw_spin_unlock+0x16/0x30
     ? klist_iter_exit+0x14/0x20
     ? xenbus_dev_request_and_reply+0x80/0x80
     device_unregister+0x13/0x60
     xenbus_dev_changed+0x18e/0x1f0
     xenwatch_thread+0xc0/0x1a0
     ? do_wait_intr_irq+0xa0/0xa0
     kthread+0x16b/0x190
     ? set_kthread_struct+0x40/0x40
     ret_from_fork+0x22/0x30
     </TASK>

Fix this by calling xennet_destroy_queues() from xennet_uninit(),
when real_num_tx_queues is still available. This ensures that queues are
destroyed when real_num_tx_queues is set to 0, regardless of how
unregister_netdev() was called.

Originally reported at
QubesOS/qubes-issues#7257

Fixes: d7dac083414eb5bb9 ("net-sysfs: update the queue counts in the unregistration path")
Cc: stable@vger.kernel.org
Signed-off-by: Marek Marczykowski-Górecki <marmarek@invisiblethingslab.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Blackmanx pushed a commit to Blackmanx/bigshot_kernel_realme_sm8250 that referenced this issue Dec 21, 2022
@marmarek @Blackmanx
xen/netfront: destroy queues before real_num_tx_queues is zeroed
1210382 
commit dcf4ff7a48e7598e6b10126cc02177abb8ae4f3f upstream.

xennet_destroy_queues() relies on info->netdev->real_num_tx_queues to
delete queues. Since d7dac083414eb5bb99a6d2ed53dc2c1b405224e5
("net-sysfs: update the queue counts in the unregistration path"),
unregister_netdev() indirectly sets real_num_tx_queues to 0. Those two
facts together means, that xennet_destroy_queues() called from
xennet_remove() cannot do its job, because it's called after
unregister_netdev(). This results in kfree-ing queues that are still
linked in napi, which ultimately crashes:

    BUG: kernel NULL pointer dereference, address: 0000000000000000
    #PF: supervisor read access in kernel mode
    #PF: error_code(0x0000) - not-present page
    PGD 0 P4D 0
    Oops: 0000 [#1] PREEMPT SMP PTI
    CPU: 1 PID: 52 Comm: xenwatch Tainted: G        W         5.16.10-1.32.fc32.qubes.x86_64+ #226
    RIP: 0010:free_netdev+0xa3/0x1a0
    Code: ff 48 89 df e8 2e e9 00 00 48 8b 43 50 48 8b 08 48 8d b8 a0 fe ff ff 48 8d a9 a0 fe ff ff 49 39 c4 75 26 eb 47 e8 ed c1 66 ff <48> 8b 85 60 01 00 00 48 8d 95 60 01 00 00 48 89 ef 48 2d 60 01 00
    RSP: 0000:ffffc90000bcfd00 EFLAGS: 00010286
    RAX: 0000000000000000 RBX: ffff88800edad000 RCX: 0000000000000000
    RDX: 0000000000000001 RSI: ffffc90000bcfc30 RDI: 00000000ffffffff
    RBP: fffffffffffffea0 R08: 0000000000000000 R09: 0000000000000000
    R10: 0000000000000000 R11: 0000000000000001 R12: ffff88800edad050
    R13: ffff8880065f8f88 R14: 0000000000000000 R15: ffff8880066c6680
    FS:  0000000000000000(0000) GS:ffff8880f3300000(0000) knlGS:0000000000000000
    CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
    CR2: 0000000000000000 CR3: 00000000e998c006 CR4: 00000000003706e0
    Call Trace:
     <TASK>
     xennet_remove+0x13d/0x300 [xen_netfront]
     xenbus_dev_remove+0x6d/0xf0
     __device_release_driver+0x17a/0x240
     device_release_driver+0x24/0x30
     bus_remove_device+0xd8/0x140
     device_del+0x18b/0x410
     ? _raw_spin_unlock+0x16/0x30
     ? klist_iter_exit+0x14/0x20
     ? xenbus_dev_request_and_reply+0x80/0x80
     device_unregister+0x13/0x60
     xenbus_dev_changed+0x18e/0x1f0
     xenwatch_thread+0xc0/0x1a0
     ? do_wait_intr_irq+0xa0/0xa0
     kthread+0x16b/0x190
     ? set_kthread_struct+0x40/0x40
     ret_from_fork+0x22/0x30
     </TASK>

Fix this by calling xennet_destroy_queues() from xennet_uninit(),
when real_num_tx_queues is still available. This ensures that queues are
destroyed when real_num_tx_queues is set to 0, regardless of how
unregister_netdev() was called.

Originally reported at
QubesOS/qubes-issues#7257

Fixes: d7dac083414eb5bb9 ("net-sysfs: update the queue counts in the unregistration path")
Cc: stable@vger.kernel.org
Signed-off-by: Marek Marczykowski-Górecki <marmarek@invisiblethingslab.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
bggRGjQaUbCoE pushed a commit to bggRGjQaUbCoE/android_kernel_samsung_sm8250-mohammad92 that referenced this issue Apr 6, 2023
@marmarek @bggRGjQaUbCoE
xen/netfront: destroy queues before real_num_tx_queues is zeroed
457cae6 
commit dcf4ff7a48e7598e6b10126cc02177abb8ae4f3f upstream.

xennet_destroy_queues() relies on info->netdev->real_num_tx_queues to
delete queues. Since d7dac083414eb5bb99a6d2ed53dc2c1b405224e5
("net-sysfs: update the queue counts in the unregistration path"),
unregister_netdev() indirectly sets real_num_tx_queues to 0. Those two
facts together means, that xennet_destroy_queues() called from
xennet_remove() cannot do its job, because it's called after
unregister_netdev(). This results in kfree-ing queues that are still
linked in napi, which ultimately crashes:

    BUG: kernel NULL pointer dereference, address: 0000000000000000
    #PF: supervisor read access in kernel mode
    #PF: error_code(0x0000) - not-present page
    PGD 0 P4D 0
    Oops: 0000 [#1] PREEMPT SMP PTI
    CPU: 1 PID: 52 Comm: xenwatch Tainted: G        W         5.16.10-1.32.fc32.qubes.x86_64+ #226
    RIP: 0010:free_netdev+0xa3/0x1a0
    Code: ff 48 89 df e8 2e e9 00 00 48 8b 43 50 48 8b 08 48 8d b8 a0 fe ff ff 48 8d a9 a0 fe ff ff 49 39 c4 75 26 eb 47 e8 ed c1 66 ff <48> 8b 85 60 01 00 00 48 8d 95 60 01 00 00 48 89 ef 48 2d 60 01 00
    RSP: 0000:ffffc90000bcfd00 EFLAGS: 00010286
    RAX: 0000000000000000 RBX: ffff88800edad000 RCX: 0000000000000000
    RDX: 0000000000000001 RSI: ffffc90000bcfc30 RDI: 00000000ffffffff
    RBP: fffffffffffffea0 R08: 0000000000000000 R09: 0000000000000000
    R10: 0000000000000000 R11: 0000000000000001 R12: ffff88800edad050
    R13: ffff8880065f8f88 R14: 0000000000000000 R15: ffff8880066c6680
    FS:  0000000000000000(0000) GS:ffff8880f3300000(0000) knlGS:0000000000000000
    CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
    CR2: 0000000000000000 CR3: 00000000e998c006 CR4: 00000000003706e0
    Call Trace:
     <TASK>
     xennet_remove+0x13d/0x300 [xen_netfront]
     xenbus_dev_remove+0x6d/0xf0
     __device_release_driver+0x17a/0x240
     device_release_driver+0x24/0x30
     bus_remove_device+0xd8/0x140
     device_del+0x18b/0x410
     ? _raw_spin_unlock+0x16/0x30
     ? klist_iter_exit+0x14/0x20
     ? xenbus_dev_request_and_reply+0x80/0x80
     device_unregister+0x13/0x60
     xenbus_dev_changed+0x18e/0x1f0
     xenwatch_thread+0xc0/0x1a0
     ? do_wait_intr_irq+0xa0/0xa0
     kthread+0x16b/0x190
     ? set_kthread_struct+0x40/0x40
     ret_from_fork+0x22/0x30
     </TASK>

Fix this by calling xennet_destroy_queues() from xennet_uninit(),
when real_num_tx_queues is still available. This ensures that queues are
destroyed when real_num_tx_queues is set to 0, regardless of how
unregister_netdev() was called.

Originally reported at
QubesOS/qubes-issues#7257

Fixes: d7dac083414eb5bb9 ("net-sysfs: update the queue counts in the unregistration path")
Cc: stable@vger.kernel.org
Signed-off-by: Marek Marczykowski-Górecki <marmarek@invisiblethingslab.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
@qubesos-bot
Copy link

Automated announcement from builder-github

The component linux-kernel (including package kernel-6.1.26-1.qubes.fc32) has been pushed to the r4.1 testing repository for dom0.
To test this update, please install it with the following command:

sudo qubes-dom0-update --enablerepo=qubes-dom0-current-testing

Changes included in this update

@qubesos-bot qubesos-bot mentioned this issue Apr 28, 2023
Closed
@qubesos-bot qubesos-bot mentioned this issue May 12, 2023
Closed
@qubesos-bot qubesos-bot mentioned this issue Jun 25, 2023
Closed
@qubesos-bot
Copy link

Automated announcement from builder-github

The component linux-kernel (including package kernel-6.1.35-1.qubes.fc32) has been pushed to the r4.1 stable repository for dom0.
To install this update, please use the standard update command:

sudo qubes-dom0-update

Or update dom0 via Qubes Manager.

Changes included in this update

intersectRaven pushed a commit to intersectRaven/rk356x-kernel that referenced this issue Sep 16, 2023
@marmarek @intersectRaven
xen/netfront: destroy queues before real_num_tx_queues is zeroed
7af6371 
commit dcf4ff7a48e7598e6b10126cc02177abb8ae4f3f upstream.

xennet_destroy_queues() relies on info->netdev->real_num_tx_queues to
delete queues. Since d7dac083414eb5bb99a6d2ed53dc2c1b405224e5
("net-sysfs: update the queue counts in the unregistration path"),
unregister_netdev() indirectly sets real_num_tx_queues to 0. Those two
facts together means, that xennet_destroy_queues() called from
xennet_remove() cannot do its job, because it's called after
unregister_netdev(). This results in kfree-ing queues that are still
linked in napi, which ultimately crashes:

    BUG: kernel NULL pointer dereference, address: 0000000000000000
    #PF: supervisor read access in kernel mode
    #PF: error_code(0x0000) - not-present page
    PGD 0 P4D 0
    Oops: 0000 [#1] PREEMPT SMP PTI
    CPU: 1 PID: 52 Comm: xenwatch Tainted: G        W         5.16.10-1.32.fc32.qubes.x86_64+ #226
    RIP: 0010:free_netdev+0xa3/0x1a0
    Code: ff 48 89 df e8 2e e9 00 00 48 8b 43 50 48 8b 08 48 8d b8 a0 fe ff ff 48 8d a9 a0 fe ff ff 49 39 c4 75 26 eb 47 e8 ed c1 66 ff <48> 8b 85 60 01 00 00 48 8d 95 60 01 00 00 48 89 ef 48 2d 60 01 00
    RSP: 0000:ffffc90000bcfd00 EFLAGS: 00010286
    RAX: 0000000000000000 RBX: ffff88800edad000 RCX: 0000000000000000
    RDX: 0000000000000001 RSI: ffffc90000bcfc30 RDI: 00000000ffffffff
    RBP: fffffffffffffea0 R08: 0000000000000000 R09: 0000000000000000
    R10: 0000000000000000 R11: 0000000000000001 R12: ffff88800edad050
    R13: ffff8880065f8f88 R14: 0000000000000000 R15: ffff8880066c6680
    FS:  0000000000000000(0000) GS:ffff8880f3300000(0000) knlGS:0000000000000000
    CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
    CR2: 0000000000000000 CR3: 00000000e998c006 CR4: 00000000003706e0
    Call Trace:
     <TASK>
     xennet_remove+0x13d/0x300 [xen_netfront]
     xenbus_dev_remove+0x6d/0xf0
     __device_release_driver+0x17a/0x240
     device_release_driver+0x24/0x30
     bus_remove_device+0xd8/0x140
     device_del+0x18b/0x410
     ? _raw_spin_unlock+0x16/0x30
     ? klist_iter_exit+0x14/0x20
     ? xenbus_dev_request_and_reply+0x80/0x80
     device_unregister+0x13/0x60
     xenbus_dev_changed+0x18e/0x1f0
     xenwatch_thread+0xc0/0x1a0
     ? do_wait_intr_irq+0xa0/0xa0
     kthread+0x16b/0x190
     ? set_kthread_struct+0x40/0x40
     ret_from_fork+0x22/0x30
     </TASK>

Fix this by calling xennet_destroy_queues() from xennet_uninit(),
when real_num_tx_queues is still available. This ensures that queues are
destroyed when real_num_tx_queues is set to 0, regardless of how
unregister_netdev() was called.

Originally reported at
QubesOS/qubes-issues#7257

Fixes: d7dac083414eb5bb9 ("net-sysfs: update the queue counts in the unregistration path")
Cc: stable@vger.kernel.org
Signed-off-by: Marek Marczykowski-Górecki <marmarek@invisiblethingslab.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
backslashxx pushed a commit to backslashxx/android_karnol_ximi_fog that referenced this issue Feb 14, 2024
@marmarek @backslashxx
xen/netfront: destroy queues before real_num_tx_queues is zeroed
305d03b 
commit dcf4ff7a48e7598e6b10126cc02177abb8ae4f3f upstream.

xennet_destroy_queues() relies on info->netdev->real_num_tx_queues to
delete queues. Since d7dac083414eb5bb99a6d2ed53dc2c1b405224e5
("net-sysfs: update the queue counts in the unregistration path"),
unregister_netdev() indirectly sets real_num_tx_queues to 0. Those two
facts together means, that xennet_destroy_queues() called from
xennet_remove() cannot do its job, because it's called after
unregister_netdev(). This results in kfree-ing queues that are still
linked in napi, which ultimately crashes:

    BUG: kernel NULL pointer dereference, address: 0000000000000000
    #PF: supervisor read access in kernel mode
    #PF: error_code(0x0000) - not-present page
    PGD 0 P4D 0
    Oops: 0000 [#1] PREEMPT SMP PTI
    CPU: 1 PID: 52 Comm: xenwatch Tainted: G        W         5.16.10-1.32.fc32.qubes.x86_64+ #226
    RIP: 0010:free_netdev+0xa3/0x1a0
    Code: ff 48 89 df e8 2e e9 00 00 48 8b 43 50 48 8b 08 48 8d b8 a0 fe ff ff 48 8d a9 a0 fe ff ff 49 39 c4 75 26 eb 47 e8 ed c1 66 ff <48> 8b 85 60 01 00 00 48 8d 95 60 01 00 00 48 89 ef 48 2d 60 01 00
    RSP: 0000:ffffc90000bcfd00 EFLAGS: 00010286
    RAX: 0000000000000000 RBX: ffff88800edad000 RCX: 0000000000000000
    RDX: 0000000000000001 RSI: ffffc90000bcfc30 RDI: 00000000ffffffff
    RBP: fffffffffffffea0 R08: 0000000000000000 R09: 0000000000000000
    R10: 0000000000000000 R11: 0000000000000001 R12: ffff88800edad050
    R13: ffff8880065f8f88 R14: 0000000000000000 R15: ffff8880066c6680
    FS:  0000000000000000(0000) GS:ffff8880f3300000(0000) knlGS:0000000000000000
    CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
    CR2: 0000000000000000 CR3: 00000000e998c006 CR4: 00000000003706e0
    Call Trace:
     <TASK>
     xennet_remove+0x13d/0x300 [xen_netfront]
     xenbus_dev_remove+0x6d/0xf0
     __device_release_driver+0x17a/0x240
     device_release_driver+0x24/0x30
     bus_remove_device+0xd8/0x140
     device_del+0x18b/0x410
     ? _raw_spin_unlock+0x16/0x30
     ? klist_iter_exit+0x14/0x20
     ? xenbus_dev_request_and_reply+0x80/0x80
     device_unregister+0x13/0x60
     xenbus_dev_changed+0x18e/0x1f0
     xenwatch_thread+0xc0/0x1a0
     ? do_wait_intr_irq+0xa0/0xa0
     kthread+0x16b/0x190
     ? set_kthread_struct+0x40/0x40
     ret_from_fork+0x22/0x30
     </TASK>

Fix this by calling xennet_destroy_queues() from xennet_uninit(),
when real_num_tx_queues is still available. This ensures that queues are
destroyed when real_num_tx_queues is set to 0, regardless of how
unregister_netdev() was called.

Originally reported at
QubesOS/qubes-issues#7257

Fixes: d7dac083414eb5bb9 ("net-sysfs: update the queue counts in the unregistration path")
Cc: stable@vger.kernel.org
Signed-off-by: Marek Marczykowski-Górecki <marmarek@invisiblethingslab.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
backslashxx pushed a commit to backslashxx/android_karnol_ximi_fog that referenced this issue Feb 20, 2024
@marmarek @backslashxx
xen/netfront: destroy queues before real_num_tx_queues is zeroed
8ec2e75 
commit dcf4ff7a48e7598e6b10126cc02177abb8ae4f3f upstream.

xennet_destroy_queues() relies on info->netdev->real_num_tx_queues to
delete queues. Since d7dac083414eb5bb99a6d2ed53dc2c1b405224e5
("net-sysfs: update the queue counts in the unregistration path"),
unregister_netdev() indirectly sets real_num_tx_queues to 0. Those two
facts together means, that xennet_destroy_queues() called from
xennet_remove() cannot do its job, because it's called after
unregister_netdev(). This results in kfree-ing queues that are still
linked in napi, which ultimately crashes:

    BUG: kernel NULL pointer dereference, address: 0000000000000000
    #PF: supervisor read access in kernel mode
    #PF: error_code(0x0000) - not-present page
    PGD 0 P4D 0
    Oops: 0000 [#1] PREEMPT SMP PTI
    CPU: 1 PID: 52 Comm: xenwatch Tainted: G        W         5.16.10-1.32.fc32.qubes.x86_64+ #226
    RIP: 0010:free_netdev+0xa3/0x1a0
    Code: ff 48 89 df e8 2e e9 00 00 48 8b 43 50 48 8b 08 48 8d b8 a0 fe ff ff 48 8d a9 a0 fe ff ff 49 39 c4 75 26 eb 47 e8 ed c1 66 ff <48> 8b 85 60 01 00 00 48 8d 95 60 01 00 00 48 89 ef 48 2d 60 01 00
    RSP: 0000:ffffc90000bcfd00 EFLAGS: 00010286
    RAX: 0000000000000000 RBX: ffff88800edad000 RCX: 0000000000000000
    RDX: 0000000000000001 RSI: ffffc90000bcfc30 RDI: 00000000ffffffff
    RBP: fffffffffffffea0 R08: 0000000000000000 R09: 0000000000000000
    R10: 0000000000000000 R11: 0000000000000001 R12: ffff88800edad050
    R13: ffff8880065f8f88 R14: 0000000000000000 R15: ffff8880066c6680
    FS:  0000000000000000(0000) GS:ffff8880f3300000(0000) knlGS:0000000000000000
    CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
    CR2: 0000000000000000 CR3: 00000000e998c006 CR4: 00000000003706e0
    Call Trace:
     <TASK>
     xennet_remove+0x13d/0x300 [xen_netfront]
     xenbus_dev_remove+0x6d/0xf0
     __device_release_driver+0x17a/0x240
     device_release_driver+0x24/0x30
     bus_remove_device+0xd8/0x140
     device_del+0x18b/0x410
     ? _raw_spin_unlock+0x16/0x30
     ? klist_iter_exit+0x14/0x20
     ? xenbus_dev_request_and_reply+0x80/0x80
     device_unregister+0x13/0x60
     xenbus_dev_changed+0x18e/0x1f0
     xenwatch_thread+0xc0/0x1a0
     ? do_wait_intr_irq+0xa0/0xa0
     kthread+0x16b/0x190
     ? set_kthread_struct+0x40/0x40
     ret_from_fork+0x22/0x30
     </TASK>

Fix this by calling xennet_destroy_queues() from xennet_uninit(),
when real_num_tx_queues is still available. This ensures that queues are
destroyed when real_num_tx_queues is set to 0, regardless of how
unregister_netdev() was called.

Originally reported at
QubesOS/qubes-issues#7257

Fixes: d7dac083414eb5bb9 ("net-sysfs: update the queue counts in the unregistration path")
Cc: stable@vger.kernel.org
Signed-off-by: Marek Marczykowski-Górecki <marmarek@invisiblethingslab.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Dheeraj3031A pushed a commit to Dheeraj3031A/kernel_oplus_RMX3461 that referenced this issue Mar 13, 2024
@marmarek @Dheeraj3031A
xen/netfront: destroy queues before real_num_tx_queues is zeroed
7cc70c0 
commit dcf4ff7a48e7598e6b10126cc02177abb8ae4f3f upstream.

xennet_destroy_queues() relies on info->netdev->real_num_tx_queues to
delete queues. Since d7dac083414eb5bb99a6d2ed53dc2c1b405224e5
("net-sysfs: update the queue counts in the unregistration path"),
unregister_netdev() indirectly sets real_num_tx_queues to 0. Those two
facts together means, that xennet_destroy_queues() called from
xennet_remove() cannot do its job, because it's called after
unregister_netdev(). This results in kfree-ing queues that are still
linked in napi, which ultimately crashes:

    BUG: kernel NULL pointer dereference, address: 0000000000000000
    #PF: supervisor read access in kernel mode
    #PF: error_code(0x0000) - not-present page
    PGD 0 P4D 0
    Oops: 0000 [#1] PREEMPT SMP PTI
    CPU: 1 PID: 52 Comm: xenwatch Tainted: G        W         5.16.10-1.32.fc32.qubes.x86_64+ #226
    RIP: 0010:free_netdev+0xa3/0x1a0
    Code: ff 48 89 df e8 2e e9 00 00 48 8b 43 50 48 8b 08 48 8d b8 a0 fe ff ff 48 8d a9 a0 fe ff ff 49 39 c4 75 26 eb 47 e8 ed c1 66 ff <48> 8b 85 60 01 00 00 48 8d 95 60 01 00 00 48 89 ef 48 2d 60 01 00
    RSP: 0000:ffffc90000bcfd00 EFLAGS: 00010286
    RAX: 0000000000000000 RBX: ffff88800edad000 RCX: 0000000000000000
    RDX: 0000000000000001 RSI: ffffc90000bcfc30 RDI: 00000000ffffffff
    RBP: fffffffffffffea0 R08: 0000000000000000 R09: 0000000000000000
    R10: 0000000000000000 R11: 0000000000000001 R12: ffff88800edad050
    R13: ffff8880065f8f88 R14: 0000000000000000 R15: ffff8880066c6680
    FS:  0000000000000000(0000) GS:ffff8880f3300000(0000) knlGS:0000000000000000
    CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
    CR2: 0000000000000000 CR3: 00000000e998c006 CR4: 00000000003706e0
    Call Trace:
     <TASK>
     xennet_remove+0x13d/0x300 [xen_netfront]
     xenbus_dev_remove+0x6d/0xf0
     __device_release_driver+0x17a/0x240
     device_release_driver+0x24/0x30
     bus_remove_device+0xd8/0x140
     device_del+0x18b/0x410
     ? _raw_spin_unlock+0x16/0x30
     ? klist_iter_exit+0x14/0x20
     ? xenbus_dev_request_and_reply+0x80/0x80
     device_unregister+0x13/0x60
     xenbus_dev_changed+0x18e/0x1f0
     xenwatch_thread+0xc0/0x1a0
     ? do_wait_intr_irq+0xa0/0xa0
     kthread+0x16b/0x190
     ? set_kthread_struct+0x40/0x40
     ret_from_fork+0x22/0x30
     </TASK>

Fix this by calling xennet_destroy_queues() from xennet_uninit(),
when real_num_tx_queues is still available. This ensures that queues are
destroyed when real_num_tx_queues is set to 0, regardless of how
unregister_netdev() was called.

Originally reported at
QubesOS/qubes-issues#7257

Fixes: d7dac083414eb5bb9 ("net-sysfs: update the queue counts in the unregistration path")
Cc: stable@vger.kernel.org
Signed-off-by: Marek Marczykowski-Górecki <marmarek@invisiblethingslab.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
nokidevz pushed a commit to nokidevz/android_kernel_xiaomi_mt6781 that referenced this issue Jun 26, 2024
@marmarek @nokidevz
xen/netfront: destroy queues before real_num_tx_queues is zeroed
4181843 
commit dcf4ff7a48e7598e6b10126cc02177abb8ae4f3f upstream.

xennet_destroy_queues() relies on info->netdev->real_num_tx_queues to
delete queues. Since d7dac083414eb5bb99a6d2ed53dc2c1b405224e5
("net-sysfs: update the queue counts in the unregistration path"),
unregister_netdev() indirectly sets real_num_tx_queues to 0. Those two
facts together means, that xennet_destroy_queues() called from
xennet_remove() cannot do its job, because it's called after
unregister_netdev(). This results in kfree-ing queues that are still
linked in napi, which ultimately crashes:

    BUG: kernel NULL pointer dereference, address: 0000000000000000
    #PF: supervisor read access in kernel mode
    #PF: error_code(0x0000) - not-present page
    PGD 0 P4D 0
    Oops: 0000 [#1] PREEMPT SMP PTI
    CPU: 1 PID: 52 Comm: xenwatch Tainted: G        W         5.16.10-1.32.fc32.qubes.x86_64+ #226
    RIP: 0010:free_netdev+0xa3/0x1a0
    Code: ff 48 89 df e8 2e e9 00 00 48 8b 43 50 48 8b 08 48 8d b8 a0 fe ff ff 48 8d a9 a0 fe ff ff 49 39 c4 75 26 eb 47 e8 ed c1 66 ff <48> 8b 85 60 01 00 00 48 8d 95 60 01 00 00 48 89 ef 48 2d 60 01 00
    RSP: 0000:ffffc90000bcfd00 EFLAGS: 00010286
    RAX: 0000000000000000 RBX: ffff88800edad000 RCX: 0000000000000000
    RDX: 0000000000000001 RSI: ffffc90000bcfc30 RDI: 00000000ffffffff
    RBP: fffffffffffffea0 R08: 0000000000000000 R09: 0000000000000000
    R10: 0000000000000000 R11: 0000000000000001 R12: ffff88800edad050
    R13: ffff8880065f8f88 R14: 0000000000000000 R15: ffff8880066c6680
    FS:  0000000000000000(0000) GS:ffff8880f3300000(0000) knlGS:0000000000000000
    CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
    CR2: 0000000000000000 CR3: 00000000e998c006 CR4: 00000000003706e0
    Call Trace:
     <TASK>
     xennet_remove+0x13d/0x300 [xen_netfront]
     xenbus_dev_remove+0x6d/0xf0
     __device_release_driver+0x17a/0x240
     device_release_driver+0x24/0x30
     bus_remove_device+0xd8/0x140
     device_del+0x18b/0x410
     ? _raw_spin_unlock+0x16/0x30
     ? klist_iter_exit+0x14/0x20
     ? xenbus_dev_request_and_reply+0x80/0x80
     device_unregister+0x13/0x60
     xenbus_dev_changed+0x18e/0x1f0
     xenwatch_thread+0xc0/0x1a0
     ? do_wait_intr_irq+0xa0/0xa0
     kthread+0x16b/0x190
     ? set_kthread_struct+0x40/0x40
     ret_from_fork+0x22/0x30
     </TASK>

Fix this by calling xennet_destroy_queues() from xennet_uninit(),
when real_num_tx_queues is still available. This ensures that queues are
destroyed when real_num_tx_queues is set to 0, regardless of how
unregister_netdev() was called.

Originally reported at
QubesOS/qubes-issues#7257

Fixes: d7dac083414eb5bb9 ("net-sysfs: update the queue counts in the unregistration path")
Cc: stable@vger.kernel.org
Signed-off-by: Marek Marczykowski-Górecki <marmarek@invisiblethingslab.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
C: kernel diagnosed Technical diagnosis has been performed (see issue comments). P: default Priority: default. Default priority for new issues, to be replaced given sufficient information. r4.0-dom0-stable r4.1-dom0-stable r4.2-host-stable T: bug Type: bug report. A problem or defect resulting in unintended behavior in something that exists.
Projects
None yet
Milestone
Release 4.0 updates
Development

No branches or pull requests
6 participants
@marmarek @DemiMarie @petaramesh @andrewdavidwong @pgerber @qubesos-bot