Linux Kernel Module Cheat

Run one command, get a QEMU or gem5 Buildroot BusyBox virtual machine built from source with several minimal Linux kernel 4.15 module development example tutorials with GDB and KGDB step debugging and minimal educational hardware models. "Tested" in x86, ARM and MIPS guests, Ubuntu 17.10 host.

• 1. Getting started
  • 1.1. Module documentation
  • 1.2. Rebuild
  • 1.3. Clean the build
  • 1.4. Filesystem persistency
  • 1.5. Message control
  • 1.6. Text mode
  • 1.7. Automatic startup commands
  • 1.8. Kernel command line parameters
  • 1.9. Kernel command line parameters escaping
  • 1.10. What command was actually run?
  • 1.11. modprobe
  • 1.12. myinsmod
• 2. GDB step debugging
  • 2.1. Kernel module debugging
    • 2.1.1. Bypass lx-symbols
  • 2.2. Debug kernel early boot
  • 2.3. GDB call
• 3. KGDB
  • 3.1. KGDB kernel modules
  • 3.2. KDB
• 4. gdbserver
  • 4.1. gdbserver different archs
  • 4.2. gdbserver BusyBox
  • 4.3. gdbserver shared libraries
  • 4.4. Debug userland process without gdbserver
• 5. Architectures
  • 5.1. arm
  • 5.2. aarch64
  • 5.3. mips64
• 6. init
  • 6.1. Replace init
    • 6.1.1. The kernel panics despite poweroff
  • 6.2. Run command at the end of BusyBox init
  • 6.3. Path to init
  • 6.4. Init environment
  • 6.5. Disable networking
• 7. KVM
• 8. X11
  • 8.1. X11 ARM
  • 8.2. X11 MIPS
• 9. Count boot instructions
• 10. initrd
  • 10.1. initrd in desktop distros
  • 10.2. initramfs
  • 10.3. gem5 initrd
• 11. Linux kernel
  • 11.1. Linux kernel configuration
    • 11.1.1. Use your own kernel config
    • 11.1.2. Find the kernel config
    • 11.1.3. Force Linux kernel configuration update
  • 11.2. Find the kernel version
  • 11.3. Update the Linux kernel
  • 11.4. Downgrade the Linux kernel
  • 11.5. ftrace
  • 11.6. User mode Linux
• 12. QEMU
  • 12.1. Snapshot
  • 12.2. Educational hardware models
  • 12.3. 9P
    • 12.3.1. 9P gem5
    • 12.3.2. OverlayFS
  • 12.4. Guest host networking
    • 12.4.1. Host to guest
      • 12.4.1.1. ssh into guest
    • 12.4.2. Guest to host
  • 12.5. QEMU user mode
  • 12.6. Debug QEMU
  • 12.7. QEMU record and replay
  • 12.8. QEMU GUI is unresponsive
• 13. gem5
  • 13.1. gem5 getting started
  • 13.2. gem5 vs QEMU
    • 13.2.1. gem5 vs QEMU performance
  • 13.3. gem5 run benchmark
    • 13.3.1. gem5 system parameters
      • 13.3.1.1. gem5 number of cores
      • 13.3.1.2. gem5 cache size
      • 13.3.1.3. gem5 memory latency
      • 13.3.1.4. gem5 memory size
      • 13.3.1.5. gem5 disk and network latency
      • 13.3.1.6. gem5 clock frequency
    • 13.3.2. Enable compiler optimizations
    • 13.3.3. Interesting benchmarks
      • 13.3.3.1. BLAS
      • 13.3.3.2. PARSEC benchmark
        • 13.3.3.2.1. PARSEC benchmark without parsecmgmt
        • 13.3.3.2.2. BR2_TARGET_ROOTFS_EXT2_SIZE
        • 13.3.3.2.3. PARSEC benchmark with parsecmgmt
        • 13.3.3.2.4. PARSEC change the input size
        • 13.3.3.2.5. PARSEC uninstall
        • 13.3.3.2.6. PARSEC benchmark hacking
        • 13.3.3.2.7. Buildroot rebuild is slow when the root filesystem is large
  • 13.4. gem5 kernel command line parameters
  • 13.5. gem5 GDB step debugging
  • 13.6. gem5 checkpoint
    • 13.6.1. gem5 restore checkpoint with a different CPU
  • 13.7. Pass extra options to gem5
  • 13.8. Run multiple gem5 instances at once
  • 13.9. gem5 and QEMU with the same kernel configuration
    • 13.9.1. QEMU with gem5 kernel configuration
    • 13.9.2. gem5 with QEMU kernel configuration
  • 13.10. gem5 limitations
    • 13.10.1. gem5 x86_64 limitations
• 14. Insane action
  • 14.1. Run on host
  • 14.2. Hello host
• 15. Buildroot
  • 15.1. Change Buildroot options
  • 15.2. Find Buildroot options with make menuconfig
  • 15.3. ccache
• 16. Benchmark this repo
  • 16.1. Find which packages are making the build slow
  • 16.2. Benchmark machines
    • 16.2.1. P51
      • 16.2.1.1. P51 baseline benchmarks
    • 16.2.2. T430
  • 16.3. Benchmark Internets
    • 16.3.1. 38Mbps
• 17. Conversation
  • 17.1. kmod
    • 17.1.1. module-init-tools
    • 17.1.2. kmod modprobe
  • 17.2. Device tree
  • 17.3. Directory structure
  • 17.4. About
    • 17.4.1. Fairy tale
  • 17.5. Bibliography

1. Getting started

Reserve 12Gb of disk and run:

git clone https://github.com/cirosantilli/linux-kernel-module-cheat
cd linux-kernel-module-cheat
./configure && ./build && ./run

The first configure will take a while (30 minutes to 2 hours) to clone and build, see [benchmarking-this-repo] for more details.

If you don’t want to wait, you could also try to compile the examples and run them on your host computer as explained on at Run on host, but as explained on that section, that is dangerous, limited, and will likely not work.

After QEMU opens up, you can start playing with the kernel modules:

root
insmod /hello.ko
insmod /hello2.ko
rmmod hello
rmmod hello2

This should print to the screen:

hello init
hello2 init
hello cleanup
hello2 cleanup

which are printk messages from init and cleanup methods of those modules.

All available modules can be found in the kernel_module directory.

1.1. Module documentation

head kernel_module/modulename.c

Many of the modules have userland test scripts / executables with the same name as the module, e.g. form inside the guest:

/modulename.sh
/modulename.out

The sources of those tests will further clarify what the corresponding kernel modules does. To find them on the host, do a quick:

git ls-files | grep modulename

1.2. Rebuild

After making changes to a package, you must explicitly tell it to be rebuilt.

For example, you you modify the kernel modules, you must rebuild with:

./build -k

which is just an alias for:

./build -- kernel_module-reconfigure

where kernel_module is the name of out Buildroot package that contains the kernel modules.

Other important targets are:

./build -- linux-reconfigure host-qemu-reconfigure

which are aliased respectively to:

./build -l -q

We don’t rebuild by default because, even with make incremental rebuilds, the timestamp check takes a few annoying seconds.

1.3. Clean the build

You did something crazy, and nothing seems to work anymore?

All builds are stored under buildroot/,

The most coarse thing you can do is:

cd buildroot
git checkout -- .
git clean -xdf .

To only nuke one architecture, do:

rm -rf buildroot/output.x86_64~

Only nuke one one package:

rm -rf buildroot/output.x86_64~/build/host-qemu-custom
./build -q

This is sometimes necessary when changing the version of the submodules, and then builds fail. We should try to understand why and report bugs.

1.4. Filesystem persistency

The root filesystem is persistent across:

./run
date >f
# poweroff syncs by default without -n.
poweroff

then:

./run
cat f

The command:

sync

also saves the disk.

This is particularly useful to re-run shell commands from the history of a previous session with Ctrl + R.

However, when you do:

./build

the disk image gets overwritten by a fresh filesystem and you lose all changes.

Remember that if you forcibly turn QEMU off without sync or poweroff from inside the VM, e.g. by closing the QEMU window, disk changes may not be saved.

When booting from initrd however without a disk, persistency is lost.

1.5. Message control

We use printk a lot, and it shows on the QEMU terminal by default. If that annoys you (e.g. you want to see stdout separately), do:

dmesg -n 1

See also: https://superuser.com/questions/351387/how-to-stop-kernel-messages-from-flooding-my-console

You can scroll up a bit on the default TTY with:

Shift + PgUp

but I never managed to increase that buffer:

• https://askubuntu.com/questions/709697/how-to-increase-scrollback-lines-in-ubuntu14-04-2-server-edition

• https://unix.stackexchange.com/questions/346018/how-to-increase-the-scrollback-buffer-size-for-tty

The superior alternative is to use text mode or a telnet connection.

1.6. Text mode

Show serial console directly on the current terminal, without opening a QEMU window:

./run -n

To quit QEMU forcefully, just use Ctrl + C as usual.

This mode is very useful to:

• get full panic traces when you start making the kernel crash :-) See also: https://unix.stackexchange.com/questions/208260/how-to-scroll-up-after-a-kernel-panic

• copy and paste commands and stdout output to / from host

• have a large scroll buffer, and be able to search it, e.g. by using GNU screen on host

Limitations:

• TODO: Ctrl + C kills the emulator, and not sent to guest processes. See:

  • https://unix.stackexchange.com/questions/167165/how-to-pass-ctrl-c-in-qemu

  • cloudius-systems/osv#49

  It is also hard to enter the monitor for the same reason:

• http://stackoverflow.com/questions/14165158/how-to-switch-to-qemu-monitor-console-when-running-with-curses

• https://superuser.com/questions/488263/how-to-switch-to-the-qemu-control-panel-with-nographics

+ Our workaround is:

+

./qemumonitor

+ I think the problem was reversed in older QEMU versions: https://superuser.com/questions/1087859/how-to-quit-the-qemu-monitor-when-not-using-a-gui/1211516#1211516

+ This is however fortunate when running QEMU with GDB, as the Ctrl + C reaches GDB and breaks. * Very early kernel messages such as early console in extract_kernel only show on the GUI, since at such early stages, not even the serial has been setup.

1.7. Automatic startup commands

When debugging a module, it becomes tedious to wait for build and re-type:

root
/modulename.sh

every time.

To automate that, use the methods described at: init

1.8. Kernel command line parameters

Bootloaders can pass a string as input to the Linux kernel when it is booting to control its behaviour, much like the execve system call does to userland processes.

This allows us to control the behaviour of the kernel without rebuilding anything.

With QEMU, QEMU itself acts as the bootloader, and provides the -append option and we expose it through ./run -e, e.g.:

./run -e 'foo bar'

Then inside the host, you can check which options were given with:

cat /proc/cmdline

They are also printed at the beginning of the boot message:

dmesg | grep "Command line"

See also:

• https://unix.stackexchange.com/questions/48601/how-to-display-the-linux-kernel-command-line-parameters-given-for-the-current-bo

• https://askubuntu.com/questions/32654/how-do-i-find-the-boot-parameters-used-by-the-running-kernel

The arguments are documented in the kernel documentation: https://www.kernel.org/doc/html/v4.14/admin-guide/kernel-parameters.html

When dealing with real boards, extra command line options are provided on some magic bootloader configuration file, e.g.:

• GRUB configuration files: https://askubuntu.com/questions/19486/how-do-i-add-a-kernel-boot-parameter

• Raspberry pi /boot/cmdline.txt on a magic partition: https://raspberrypi.stackexchange.com/questions/14839/how-to-change-the-kernel-commandline-for-archlinuxarm-on-raspberry-pi-effectly

1.9. Kernel command line parameters escaping

Double quotes can be used to escape spaces as in opt="a b", but double quotes themselves cannot be escaped, e.g. opt"a\"b"

This even lead us to use base64 encoding with -E!

1.10. What command was actually run?

When asking for help on upstream repositories outside of this repository, you will need to provide the commands that you are running in detail without referencing our scripts.

For example, QEMU developers will only want to see the final QEMU command that you are running.

We make that easy by building commands as strings, and then echoing them before evaling.

So for example when you run:

./run -a arm

Stdout shows a line with the full command of type:

./buildroot/output.arm~/host/usr/bin/qemu-system-arm -m 128M -monitor telnet::45454,server,nowait -netdev user,hostfwd=tcp::45455-:45455,id=net0 -smp 1  -M versatilepb -append 'root=/dev/sda nokaslr norandmaps printk.devkmsg=on printk.time=y' -device rtl8139,netdev=net0 -dtb ./buildroot/output.arm~/images/versatile-pb.dtb -kernel ./buildroot/output.arm~/images/zImage -serial stdio    -drive file='./buildroot/output.arm~/images/rootfs.ext2.qcow2,if=scsi,format=qcow2'

This line is also saved to a file for convenience:

cat ./run.log

1.11. modprobe

If you are feeling fancy, you can also insert modules with:

modprobe dep2
lsmod
# dep and dep2

This method also deals with module dependencies, which we almost don’t use to make examples simpler:

• https://askubuntu.com/questions/20070/whats-the-difference-between-insmod-and-modprobe

• https://stackoverflow.com/questions/22891705/whats-the-difference-between-insmod-and-modprobe

Removal also removes required modules that have zero usage count:

modprobe -r dep2
lsmod
# Nothing.

but it can’t know if you actually insmodded them separately or not:

modprobe dep
modprobe dep2
modprobe -r dep2
# Nothing.

so it is a bit risky.

modprobe searches for modules under:

ls /lib/modules/*/extra/

Kernel modules built from the Linux mainline tree with CONFIG_SOME_MOD=m, are automatically available with modprobe, e.g.:

modprobe dummy-irq

1.12. myinsmod

https://stackoverflow.com/questions/5947286/how-to-load-linux-kernel-modules-from-c-code

If you are feeling raw, you can insert and remove modules with our own minimal module inserter and remover!

/myinsmod.out /hello.ko
/myrmmod.out hello

which teaches you how it is done from C code.

2. GDB step debugging

To GDB step debug the Linux kernel, first run:

./run -d

If you want to break immediately at a symbol, e.g. start_kernel of the boot sequence, run on another shell:

./rungdb start_kernel

or at a given line:

./rungdb init/main.c:1088

Now QEMU will stop there, and you can use the normal GDB commands:

l
n
c

To skip the boot, run just:

./rungdb

and when you want to break, do Ctrl + C from GDB.

To have some fun, you can first run inside QEMU:

/count.sh

which counts to infinity to stdout, and then in GDB:

Ctrl + C
break sys_write
continue
continue
continue

And you now control the counting from GDB.

See also:

• http://stackoverflow.com/questions/11408041/how-to-debug-the-linux-kernel-with-gdb-and-qemu/33203642#33203642

• http://stackoverflow.com/questions/4943857/linux-kernel-live-debugging-how-its-done-and-what-tools-are-used/42316607#42316607

O=0 is an impossible dream, O=2 being the default: https://stackoverflow.com/questions/29151235/how-to-de-optimize-the-linux-kernel-to-and-compile-it-with-o0 So get ready for some weird jumps, and <value optimized out> fun. Why, Linux, why.

2.1. Kernel module debugging

Loadable kernel modules are a bit trickier since the kernel can place them at different memory locations depending on load order.

So we cannot set the breakpoints before insmod.

However, the Linux kernel GDB scripts offer the lx-symbols command, which takes care of that beautifully for us:

./run -d
./rungdb

In QEMU:

insmod /timer.ko

In GDB, hit Ctrl + C, and note how it says:

scanning for modules in /home/ciro/bak/git/linux-kernel-module-cheat/buildroot/output.x86_64~/build/linux-custom
loading @0xffffffffc0000000: ../kernel_module-1.0//timer.ko

That’s lx-symbols working! Now simply:

b lkmc_timer_callback
c
c
c

and we now control the callback from GDB!

Just don’t forget to remove your breakpoints after rmmod, or they will point to stale memory locations.

TODO: why does break work_func for insmod kthread.ko not break the first time I insmod, but breaks the second time?

See also: http://stackoverflow.com/questions/28607538/how-to-debug-linux-kernel-modules-with-qemu/44095831#44095831

2.1.1. Bypass lx-symbols

Useless, but a good way to show how hardcore you are. From inside QEMU:

insmod /fops.ko
cat /proc/modules

This will give a line of form:

fops 2327 0 - Live 0xfffffffa00000000

And then tell GDB where the module was loaded with:

Ctrl + C
add-symbol-file ../kernel_module-1.0/fops.ko 0xfffffffa00000000

2.2. Debug kernel early boot

TODO: why can’t we break at early startup stuff such as:

./rungdb extract_kernel
./rungdb main

See also: https://stackoverflow.com/questions/2589845/what-are-the-first-operations-that-the-linux-kernel-executes-on-boot

2.3. GDB call

GDB can call functions as explained at: https://stackoverflow.com/questions/1354731/how-to-evaluate-functions-in-gdb

However this is failing for us:

• some symbols are not visible to call even though b sees them

• for those that are, call fails with an E14 error

E.g.: if we break on sys_write on /count.sh:

>>> call printk(0, "asdf")
Could not fetch register "orig_rax"; remote failure reply 'E14'
>>> b printk
Breakpoint 2 at 0xffffffff81091bca: file kernel/printk/printk.c, line 1824.
>>> call fdget_pos(fd)
No symbol "fdget_pos" in current context.
>>> b fdget_pos
Breakpoint 3 at 0xffffffff811615e3: fdget_pos. (9 locations)
>>>

even though fdget_pos is the first thing sys_write does:

581 SYSCALL_DEFINE3(write, unsigned int, fd, const char __user *, buf,
582         size_t, count)
583 {
584     struct fd f = fdget_pos(fd);

See also: cirosantilli#19

3. KGDB

KGDB is kernel dark magic that allows you to GDB the kernel on real hardware without any extra hardware support.

It is useless with QEMU since we already have full system visibility with -gdb, but this is a good way to learn it.

Cheaper than JTAG (free) and easier to setup (all you need is serial), but with less visibility as it depends on the kernel working, so e.g.: dies on panic, does not see boot sequence.

Usage:

./run -k
./rungdb -k

In GDB:

c

In QEMU:

/count.sh &
/kgdb.sh

In GDB:

b sys_write
c
c
c
c

And now you can count from GDB!

If you do: b sys_write immediately after ./rungdb -k, it fails with KGDB: BP remove failed: <address>. I think this is because it would break too early on the boot sequence, and KGDB is not yet ready.

See also:

• https://github.com/torvalds/linux/blob/v4.9/Documentation/DocBook/kgdb.tmpl

• https://stackoverflow.com/questions/22004616/qemu-kernel-debugging-with-kgdb/44197715#44197715

3.1. KGDB kernel modules

In QEMU:

/kgdb-mod.sh

In GDB:

lx-symbols ../kernel_module-1.0/
b fop_write
c
c
c

and you now control the count.

TODO: if I -ex lx-symbols to the gdb command, just like done for QEMU -gdb, the kernel oops. How to automate this step?

3.2. KDB

If you modify runqemu to use:

-append kgdboc=kbd

instead of kgdboc=ttyS0,115200, you enter a different debugging mode called KDB.

Usage: in QEMU:

[0]kdb> go

Boot finishes, then:

/kgdb.sh

And you are back in KDB. Now you can:

[0]kdb> help
[0]kdb> bp sys_write
[0]kdb> go

And you will break whenever sys_write is hit.

The other KDB commands allow you to instruction steps, view memory, registers and some higher level kernel runtime data.

But TODO I don’t think you can see where you are in the kernel source code and line step as from GDB, since the kernel source is not available on guest (ah, if only debugging information supported full source).

4. gdbserver

Step debug userland processes to understand how they are talking to the kernel.

In guest:

/gdbserver.sh /myinsmod.out /hello.ko

In host:

./rungdbserver kernel_module-1.0/user/myinsmod.out

You can find the executable with:

find buildroot/output.x86_64~/build -name myinsmod.out

TODO: automate the path finding:

• using the executable from under buildroot/output.x86_64~/target would be easier as the path is the same as in guest, but unfortunately those executables are stripped to make the guest smaller. BR2_STRIP_none=y should disable stripping, but make the image way larger.

• outputx86_64~/staging/ would be even better than target/ as the docs say that this directory contains binaries before they were stripped. However, only a few binaries are pre-installed there by default, and it seems to be a manual per package thing.

  E.g. pciutils has for lspci:

  define PCIUTILS_INSTALL_STAGING_CMDS
      $(TARGET_MAKE_ENV) $(MAKE1) -C $(@D) $(PCIUTILS_MAKE_OPTS) \
          PREFIX=$(STAGING_DIR)/usr SBINDIR=$(STAGING_DIR)/usr/bin \
          install install-lib install-pcilib
  endef

  and the docs describe the *_INSTALL_STAGING per package config, which is normally set for shared library packages.

  Feature request: https://bugs.busybox.net/show_bug.cgi?id=10386

An implementation overview can be found at: https://reverseengineering.stackexchange.com/questions/8829/cross-debugging-for-mips-elf-with-qemu-toolchain/16214#16214

4.1. gdbserver different archs

As usual, different archs work with:

./rungdbserver -a arm kernel_module-1.0/user/myinsmod.out

4.2. gdbserver BusyBox

BusyBox executables are all symlinks, so if you do on guest:

/gdbserver.sh ls

on host you need:

./rungdbserver busybox-1.26.2/busybox

4.3. gdbserver shared libraries

Our setup gives you the rare opportunity to step debug libc and other system libraries e.g. with:

b open
c

Or simply by stepping into calls:

s

This is made possible by the GDB command:

set sysroot ${buildroot_out_dir}/staging

which automatically finds unstripped shared libraries on the host for us.

See also: https://stackoverflow.com/questions/8611194/debugging-shared-libraries-with-gdbserver/45252113#45252113

4.4. Debug userland process without gdbserver

QEMU -gdb GDB breakpoints are set on virtual addresses, so you can in theory debug userland processes as well.

• https://stackoverflow.com/questions/26271901/is-it-possible-to-use-gdb-and-qemu-to-debug-linux-user-space-programs-and-kernel

• https://stackoverflow.com/questions/16273614/debug-init-on-qemu-using-gdb

The only use case I can see for this is to debug the init process (and have fun), otherwise, why wouldn’t you just use gdbserver? Known limitations of direct userland debugging:

• the kernel might switch context to another process, and you would enter "garbage"

• TODO step into shared libraries. If I attempt to load them explicitly:

  (gdb) sharedlibrary ../../staging/lib/libc.so.0
  No loaded shared libraries match the pattern `../../staging/lib/libc.so.0'.

  since GDB does not know that libc is loaded.

Custom init process:

• Shell 1:

  ./run -d -e 'init=/sleep_forever.out' -n

• Shell 2:

  ./rungdb-user kernel_module-1.0/user/sleep_forever.out main

BusyBox custom init process:

• Shell 1:

  ./run -d -e 'init=/bin/ls' -n

• Shell 2:

  ./rungdb-user -h busybox-1.26.2/busybox ls_main

This follows BusyBox' convention of calling the main for each executable as <exec>_main since the busybox executable has many "mains".

BusyBox default init process:

• Shell 1:

  ./run -d -n

• Shell 2:

  ./rungdb-user -h busybox-1.26.2/busybox init_main

This cannot be debugged in another way without modifying the source, or /sbin/init exits early with:

"must be run as PID 1"

Non-init process:

• Shell 1

  ./run -d -n

• Shell 2

  ./rungdb-user kernel_module-1.0/user/sleep_forever.out
  Ctrl + C
  b main
  continue

• Shell 1

  /sleep_forever.out

This is of least reliable setup as there might be other processes that use the given virtual address.

5. Architectures

The portability of the kernel and toolchains is amazing: change an option and most things magically work on completely different hardware.

To use arm instead of x86 for example:

./build -a arm
./run -a arm

Debug:

./run -a arm -d
# On another terminal.
./rungdb -a arm

Known quirks of the supported architectures are documented in this section.

5.1. arm

TODOs:

• only managed to run in the terminal interface (but weirdly a blank QEMU window is still opened)

• GDB not connecting to KGDB. Possibly linked to -serial stdio. See also: https://stackoverflow.com/questions/14155577/how-to-use-kgdb-on-arm

• /poweroff.out does not exit QEMU nor gem5, the terminal just hangs: https://stackoverflow.com/questions/31990487/how-to-cleanly-exit-qemu-after-executing-bare-metal-program-without-user-interve

  A blunt resolution for QEMU is on host:

  pkill qemu

  On gem5, it is possible to use the m5 instrumentation from guest as a good workaround:

  m5 exit

• GDB step debugging of kernel modules broke at some point. This happens at 6420c31986e064c81561da8f2be0bd33483af598 on kernel v4.15, 6b0f89a8b43e8d33d3a3a436ed827f962da3008a v4.14 and 5ad68edd000685c016c45e344470f2c1867b8e39 v4.12 and also if kernel 4.9.6 is checked out. So maybe it was never working in the first place, but we never noticed?

  Just after GDB connects, we get the following message from the kernel GDB Python scripts:

loading vmlinux
Traceback (most recent call last):
  File "/home/ciro/bak/git/linux-kernel-module-cheat/buildroot/output.arm~/build/linux-custom/scripts/gdb/linux/symbols.py", line 163, in invoke
    self.load_all_symbols()
  File "/home/ciro/bak/git/linux-kernel-module-cheat/buildroot/output.arm~/build/linux-custom/scripts/gdb/linux/symbols.py", line 150, in load_all_symbols
    [self.load_module_symbols(module) for module in module_list]
  File "/home/ciro/bak/git/linux-kernel-module-cheat/buildroot/output.arm~/build/linux-custom/scripts/gdb/linux/symbols.py", line 110, in load_module_symbols
    module_name = module['name'].string()
  File "/home/ciro/bak/git/linux-kernel-module-cheat/buildroot/output.arm~/host/lib/python2.7/encodings/utf_8.py", line 16, in decode
    return codecs.utf_8_decode(input, errors, True)
UnicodeDecodeError: 'utf8' codec can't decode byte 0x9f in position 2: invalid start byte
Error occurred in Python command: 'utf8' codec can't decode byte 0x9f in position 2: invalid start byte

+ and then after inserting the module, symbols are not found, presumably because lx-symbols never ran.

5.2. aarch64

As usual, we use Buildroot’s recommended QEMU setup QEMU aarch64 setup:

• https://github.com/buildroot/buildroot/blob/2017.08/board/qemu/aarch64-virt/readme.txt

• https://github.com/buildroot/buildroot/blob/2017.08/configs/qemu_aarch64_virt_defconfig

This makes aarch64 a bit different from arm:

• uses -M virt. https://wiki.qemu.org/Documentation/Platforms/ARM explains:

  Most of the machines QEMU supports have annoying limitations (small amount of RAM, no PCI or other hard disk, etc) which are there because that’s what the real hardware is like. If you don’t care about reproducing the idiosyncrasies of a particular bit of hardware, the best choice today is the "virt" machine.

  -M virt has some limitations, e.g. I could not pass -drive if=scsi as for arm, and so Snapshot fails.

So, as long as you keep those points in mind, our -a aarch64 offers an interesting different setup to play with.

TODOs:

• GDB step debugging appears to be stuck on an infinite loop:

  no module object found for ''

5.3. mips64

Keep in mind that MIPS has the worst support compared to our other architectures due to the smaller community. Patches welcome as usual.

TODOs:

• networking is not working. See also:

  • https://stackoverflow.com/questions/21496449/networking-is-not-working-on-qemu-guest-malta-mips

  • https://unix.stackexchange.com/questions/208266/setting-up-qemu-and-mipsel-networking-trouble

  • https://unix.stackexchange.com/questions/354127/qemu-mips-and-debian

• GDB step debugging does not work properly, does not find start_kernel

6. init

When the Linux kernel finishes booting, it runs an executable as the first and only userland process.

This init process is then responsible for setting up the entire userland (or destroying everything when you want to have fun).

This typically means reading some configuration files (e.g. /etc/initrc) and forking a bunch of userland executables based on those files.

systemd provides a "popular" init implementation for desktop distros as of 2017.

BusyBox provides its own minimalistic init implementation which Buildroot, and therefore this repo, uses by default.

6.1. Replace init

To have more control over the system, you can replace BusyBox’s init with your own.

The following method replaces init and evals a command from the Kernel command line parameters:

./run -E 'echo "asdf qwer";insmod /hello.ko;/poweroff.out' -n

It is basically a shortcut for:

./run -e 'init=/eval.sh - lkmc_eval="insmod /hello.ko;/poweroff.out"' -n

although -E is smarter:

• allows quoting and newlines by using base64 encoding, see: Kernel command line parameters escaping

• automatically chooses between init= and rcinit= for you, see: Path to init

so you should almost always use it, unless you are really counting each cycle ;-)

If the script is large, you can add it to a gitignored file and pass that to -E as in:

echo '
insmod /hello.ko
/poweroff.out
' > gitignore.sh
./run -E "$(cat gitignore.sh)" -n

or add it to a file to the root filesystem guest and rebuild:

echo '#!/bin/sh
insmod /hello.ko
/poweroff.out
' > rootfs_overlay/gitignore.sh
chmod +x rootfs_overlay/gitignore.sh
./build
./run -e 'init=/gitignore.sh' -n

Remember that if your init returns, the kernel will panic, there are just two non-panic possibilities:

• run forever in a loop or long sleep

• poweroff the machine

6.1.1. The kernel panics despite poweroff

Just using BusyBox' poweroff at the end of the init does not work and the kernel panics:

./run -E poweroff

because BusyBox' poweroff tries to do some fancy stuff like killing init, likely to allow userland to shutdown nicely.

But this fails when we are init itself!

poweroff works more brutally and effectively if you add -f:

./run -E 'poweroff -f'

but why not just use your super simple and effective /poweroff.out and be done with it?

6.2. Run command at the end of BusyBox init

If you rely on something that BusyBox' init set up for you like networking, you caould do:

./run -e '- lkmc_eval="insmod /hello.ko;wget -S google.com;poweroff.out;"'

The lkmc_eval option gets evaled by our default S98 startup script if present.

Alternatively, add them to a new init.d entry to run at the end o the BusyBox init:

cp rootfs_overlay/etc/init.d/S98 rootfs_overlay/etc/init.d/S99.gitignore
vim rootfs_overlay/etc/init.d/S99.gitignore
./build
./run

and they will be run automatically before the login prompt.

Scripts under /etc/init.d are run by /etc/init.d/rcS, which gets called by the line ::sysinit:/etc/init.d/rcS in /etc/inittab.

6.3. Path to init

The init is selected at:

• initrd or initramfs system: /init, a custom one can be set with the rdinit= kernel command line parameter

• otherwise: default is /sbin/init, followed by some other paths, a custom one can be set with init=

More details: https://unix.stackexchange.com/questions/30414/what-can-make-passing-init-path-to-program-to-the-kernel-not-start-program-as-i/430614#430614

6.4. Init environment

Documented at https://www.kernel.org/doc/html/v4.14/admin-guide/kernel-parameters.html:

The kernel parses parameters from the kernel command line up to "-"; if it doesn’t recognize a parameter and it doesn’t contain a '.', the parameter gets passed to init: parameters with '=' go into init’s environment, others are passed as command line arguments to init. Everything after "-" is passed as an argument to init.

And you can try it out with:

./run -e 'init=/init_env_poweroff.sh - asdf=qwer zxcv' -n

Also note how the annoying dash - also gets passed as a parameter to init, which makes it impossible to use this method for most executables.

Finally, the docs are lying, arguments with dots that come after - are still treated specially (of the form subsystem.somevalue) and disappear:

./run -e 'init=/init_env_poweroff.sh - /poweroff.out' -n

6.5. Disable networking

The default BusyBox init scripts enable networking, and there is a 15 second timeout in case your network is down or if your kernel / emulator setup does not support it.

To disable networking, use:

./build -p -n

To restore it, run:

./build -- initscripts-reconfigure

7. KVM

You can make QEMU or gem5 run faster by passing enabling KVM with:

./run -K

but it was broken in gem5 with pending patches: https://www.mail-archive.com/gem5-users@gem5.org/msg15046.html

KVM uses the KVM Linux kernel feature of the host to run most instructions natively.

We don’t enable KVM by default because:

• only works if the architecture of the guest equals that of the host.

  We have only tested / supported it on x86, but it is rumoured that QEMU and gem5 also have ARM KVM support if you are running an ARM desktop for some weird reason :-)

• limits visibility, since more things are running natively:

  • can’t use GDB

  • can’t do instruction tracing

• kernel boots are already fast enough without -enable-kvm

The main use case for -enable-kvm in this repository is to test if something that takes a long time to run is functionally correct.

For example, when porting a benchmark to Buildroot, you can first use QEMU’s KVM to test that benchmarks is producing the correct results, before analysing them more deeply in gem5, which runs much slower.

8. X11

Only tested successfully in x86_64.

Build:

./build -b br2_x11
./run

We don’t build X11 by default because it takes a considerable amount of time (~20%), and is not expected to be used by most users: you need to pass the -x flag to enable it.

Inside QEMU:

startx

More details: https://unix.stackexchange.com/questions/70931/how-to-install-x11-on-my-own-linux-buildroot-system/306116#306116

Not sure how well that graphics stack represents real systems, but if it does it would be a good way to understand how it works.

8.1. X11 ARM

On ARM, startx hangs at a message:

vgaarb: this pci device is not a vga device

and nothing shows on the screen, and:

grep EE /var/log/Xorg.0.log

says:

(EE) Failed to load module "modesetting" (module does not exist, 0)

A friend told me this but I haven’t tried it yet:

• xf86-video-modesetting is likely the missing ingredient, but it does not seem possible to activate it from Buildroot currently without patching things.

• xf86-video-fbdev should work as well, but we need to make sure fbdev is enabled, and maybe add some line to the Xorg.conf

8.2. X11 MIPS

Haven’t tried it, doubt it will work out of the box! :-)

Maybe: https://stackoverflow.com/questions/47857023/booting-a-graphical-mips-qemu-machine

9. Count boot instructions

• https://www.quora.com/How-many-instructions-does-a-typical-Linux-kernel-boot-take

• https://github.com/cirosantilli/chat/issues/31

• https://rwmj.wordpress.com/2016/03/17/tracing-qemu-guest-execution/

• qemu/docs/tracing.txt and qemu/docs/replay.txt

• https://stackoverflow.com/questions/39149446/how-to-use-qemus-simple-trace-backend/46497873#46497873

Results:

Commit	Arch	Simulator	Instruction count
7228f75ac74c896417fb8c5ba3d375a14ed4d36b	arm	QEMU	680k
7228f75ac74c896417fb8c5ba3d375a14ed4d36b	arm	gem5 AtomicSimpleCPU	160M
7228f75ac74c896417fb8c5ba3d375a14ed4d36b	arm	gem5 HPI	155M
7228f75ac74c896417fb8c5ba3d375a14ed4d36b	x86_64	QEMU	3M
7228f75ac74c896417fb8c5ba3d375a14ed4d36b	x86_64	gem5 AtomicSimpleCPU	528M

QEMU:

./count-boot-instructions

sample output:

1778927 trace.txt
26264 trace-boot.txt

where:

• 1778927: is the total number of boot instructions

• 26264: is the number of bootloader instructions

gem5:

./run -a arm -g -E 'm5 exit'
# Or:
# ./run -a arm -g -E 'm5 exit' -- --cpu-type=HPI --caches
grep sim_insts m5out/stats.txt

Notes:

• -n is a good idea to reduce the chances that you send unwanted non-deterministic mouse or keyboard clicks to the VM.

• -e 'init=/poweroff.out' is crucial as it reduces the instruction count from 40 million to 20 million, so half of the instructions were due to userland programs instead of the boot sequence.

  Without it, the bulk of the time seems to be spent in setting up the network with ifup that gets called from /etc/init.d/S40network from the default Buildroot BusyBox setup.

  And it becomes even worse if you try to -net none as recommended in the 2.7 replay.txt docs, because then ifup waits for 15 seconds before giving up as per /etc/network/interfaces line wait-delay 15.

• 0x1000000 is the address where QEMU puts the Linux kernel at with -kernel in x86.

  It can be found from:

  readelf -e buildroot/output.x86_64~/build/linux-*/vmlinux | grep Entry

  TODO confirm further. If I try to break there with:

  ./rungdb *0x1000000

  but I have no corresponding source line. Also note that this line is not actually the first line, since the kernel messages such as early console in extract_kernel have already shown on screen at that point. This does not break at all:

  ./rungdb extract_kernel

  It only appears once on every log I’ve seen so far, checked with grep 0x1000000 trace.txt

  Then when we count the instructions that run before the kernel entry point, there is only about 100k instructions, which is insignificant compared to the kernel boot itself.

• We can also discount the instructions after init runs by using readelf to get the initial address of init. One easy way to do that now is to just run:

  ./rungdb-user kernel_module-1.0/user/poweroff.out main

  And get that from the traces, e.g. if the address is 4003a0, then we search:

  grep -n 4003a0 trace.txt

  I have observed a single match for that instruction, so it must be the init, and there were only 20k instructions after it, so the impact is negligible.

• gem5 simulates memory latencies. So I think that the CPU loops idle while waiting for memory, and counts will be higher.

This works because we have already done the following with QEMU:

• ./configure --enable-trace-backends=simple. This logs in a binary format to the trace file.

  It makes 3x execution faster than the default trace backend which logs human readable data to stdout.

  This also alters the actual execution, and reduces the instruction count by 10M TODO understand exactly why, possibly due to the All QSes seen thing.

• patch QEMU source to remove the disable from exec_tb in the trace-events. See also: https://rwmj.wordpress.com/2016/03/17/tracing-qemu-guest-execution/

Possible improvements:

• to disable networking. Is replacing init enough?

  • https://superuser.com/questions/181254/how-do-you-boot-linux-with-networking-disabled

  • https://superuser.com/questions/684005/how-does-one-permanently-disable-gnu-linux-networking/1255015#1255015

    CONFIG_NET=n did not significantly reduce instruction, so maybe replacing init is enough.

• logging with the default backend log greatly slows down the CPU, and in particular leads to this during kernel boot:

  All QSes seen, last rcu_sched kthread activity 5252 (4294901421-4294896169), jiffies_till_next_fqs=1, root ->qsmask 0x0
  swapper/0       R  running task        0     1      0 0x00000008
   ffff880007c03ef8 ffffffff8107aa5d ffff880007c16b40 ffffffff81a3b100
   ffff880007c03f60 ffffffff810a41d1 0000000000000000 0000000007c03f20
   fffffffffffffedc 0000000000000004 fffffffffffffedc ffffffff00000000
  Call Trace:
   <IRQ>  [<ffffffff8107aa5d>] sched_show_task+0xcd/0x130
   [<ffffffff810a41d1>] rcu_check_callbacks+0x871/0x880
   [<ffffffff810a799f>] update_process_times+0x2f/0x60

  in which the boot appears to hang for a considerable time.

• Confirm that the kernel enters at 0x1000000, or where it enters. Once we have this, we can exclude what comes before in the BIOS.

10. initrd

The kernel can boot from an CPIO file, which is a directory serialization format much like tar: https://superuser.com/questions/343915/tar-vs-cpio-what-is-the-difference

The bootloader, which for us is QEMU itself, is then configured to put that CPIO into memory, and tell the kernel that it is there.

With this setup, you don’t even need to give a root filesystem to the kernel, it just does everything in memory in a ramfs.

To enable initrd instead of the default ext2 disk image, do:

./build -i
./run -i

Notice how it boots fine, even though this leads to not giving QEMU the -drive option, as can be verified with:

cat ./run.log

Also as expected, there is no filesystem persistency, since we are doing everything in memory:

date >f
poweroff
cat f
# can't open 'f': No such file or directory

which can be good for automated tests, as it ensures that you are using a pristine unmodified system image every time.

One downside of this method is that it has to put the entire filesystem into memory, and could lead to a panic:

end Kernel panic - not syncing: Out of memory and no killable processes...

This can be solved by increasing the memory with:

./run -im 256M

The main ingredients to get initrd working are:

• BR2_TARGET_ROOTFS_CPIO=y: make Buildroot generate output/images/rootfs.cpio in addition to the other images.

  It is also possible to compress that image with other options.

• qemu -initrd: make QEMU put the image into memory and tell the kernel about it.

• CONFIG_BLK_DEV_INITRD=y: Compile the kernel with initrd support, see also: https://unix.stackexchange.com/questions/67462/linux-kernel-is-not-finding-the-initrd-correctly/424496#424496

  Buildroot forces that option when BR2_TARGET_ROOTFS_CPIO=y is given

https://unix.stackexchange.com/questions/89923/how-does-linux-load-the-initrd-image asks how the mechanism works in more detail.

10.1. initrd in desktop distros

Most modern desktop distributions have an initrd in their root disk to do early setup.

The rationale for this is described at: https://en.wikipedia.org/wiki/Initial_ramdisk

One obvious use case is having an encrypted root filesystem: you keep the initrd in an unencrypted partition, and then setup decryption from there.

I think GRUB then knows read common disk formats, and then loads that initrd to memory with a /boot/grub/grub.cfg directive of type:

initrd /initrd.img-4.4.0-108-generic

Related: https://stackoverflow.com/questions/6405083/initrd-and-booting-the-linux-kernel

10.2. initramfs

initramfs is just like initrd, but you also glue the image directly to the kernel image itself.

So the only argument that QEMU needs is the -kernel, no -drive not even -initrd! Pretty cool.

Try it out with:

touch kernel_config_fragment
./build -I && ./run -I

The touch should only be used the first time you move to / from a different root filesystem method (ext2 or cpio) to initramfs to overcome: Force Linux kernel configuration update. Further builds should be done simply as:

./build -I && ./run -I

It is interesting to see how this increases the size of the kernel image if you do a:

ls -lh buildroot/output.x86_64~/images/bzImage

before and after using initramfs, since the .cpio is now glued to the kernel image.

In the background, it uses BR2_TARGET_ROOTFS_INITRAMFS, and this makes the kernel config option CONFIG_INITRAMFS_SOURCE point to the CPIO that will be embedded in the kernel image.

http://nairobi-embedded.org/initramfs_tutorial.html shows a full manual setup.

10.3. gem5 initrd

TODO we were not able to get it working yet: https://stackoverflow.com/questions/49261801/how-to-boot-the-linux-kernel-with-initrd-or-initramfs-with-gem5

11. Linux kernel

11.1. Linux kernel configuration

11.1.1. Use your own kernel config

By default, we use a .config that is a mixture of:

• Buildroot’s minimal per machine .config, which has the minimal options needed to boot

• our kernel_config_fragment which enables options we want to play with

If you want to just use your own exact .config instead, do:

touch myconfig
./build -K myconfig

The touch is only need the first time you move to a new kernel configuration: Force Linux kernel configuration update.

Beware that Buildroot can sed override some of the configurations we make no matter what, e.g. it forces CONFIG_BLK_DEV_INITRD=y when BR2_TARGET_ROOTFS_CPIO is on, so you might want to double check as explained at Find the kernel config. TODO check if there is a way to prevent that patching and maybe patch Buildroot for it, it is too fuzzy. People should be able to just build with whatever .config they want.

11.1.2. Find the kernel config

Build configuration can be observed in guest with:

/conf.sh

or on host:

cat buildroot/output.*~/build/linux-custom/.config

11.1.3. Force Linux kernel configuration update

https://stackoverflow.com/questions/49260466/why-when-i-change-br2-linux-kernel-custom-config-file-and-run-make-linux-reconfi

To save rebuild time, Buildroot does many things based on timestamps.

This means that some operations end up not updating the Linux kernel .config, and we just require you to do it manually with:

touch kernel_config_fragment

When this works, you should see Buildroot run the kernel rebuild commands:

The reason we don’t do that automatically all the time, is that this would defeat the purpose of the timestamp, and add some overhead even when you are not modifying the config at all.

Maybe we could think of smarter methods to overcome this problem more automatically, e.g. take hashes and check them whenever a suspicious operations happens.

But for now we just require you to do the touch manually yourself, and document when this is needed.

11.2. Find the kernel version

We try to use the latest possible kernel major release version.

In QEMU:

cat /proc/version

or in the source:

cd linux
git log | grep -E '    Linux [0-9]+\.' | head

11.3. Update the Linux kernel

# Last point before out patches.
last_mainline_revision=v4.14
next_mainline_revision=v4.15
cd linux

# Create a branch before the rebase.
git branch "lkmc-${last_mainline_revision}"
git remote set-url origin git@github.com:cirosantilli/linux.git
git push

git remote add up git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git
git fetch up
git rebase --onto "$next_mainline_revision" "$last_mainline_revision"
./build -l
# Manually fix our kernel modules if necessary.

cd ..
git branch "buildroot-2017.08-linux-${last_mainline_revision}"
git add .
git commit -m "Linux ${next_mainline_revision}"
git push

and update the README!

During update all you kernel modules may break since the kernel API is not stable.

They are usually trivial breaks of things moving around headers or to sub-structs.

The userland, however, should simply not break, as Linus enforces strict backwards compatibility of userland interfaces.

This backwards compatibility is just awesome, it makes getting and running the latest master painless.

This also makes this repo the perfect setup to develop the Linux kernel.

11.4. Downgrade the Linux kernel

The kernel is not forward compatible, however, so downgrading the Linux kernel requires downgrading the userland too to the latest Buildroot branch that supports it.

The default Linux kernel version is bumped in Buildroot with commit messages of type:

linux: bump default to version 4.9.6

So you can try:

git log --grep 'linux: bump default to version'

Those commits change BR2_LINUX_KERNEL_LATEST_VERSION in /linux/Config.in.

You should then look up if there is a branch that supports that kernel. Staying on branches is a good idea as they will get backports, in particular ones that fix the build as newer host versions come out.

11.5. ftrace

Trace a single function:

cd /sys/kernel/debug/tracing/

# Stop tracing.
echo 0 > tracing_on

# Clear previous trace.
echo '' > trace

# List the available tracers, and pick one.
cat available_tracers
echo function > current_tracer

# List all functions that can be traced
# cat available_filter_functions
# Choose one.
echo __kmalloc >set_ftrace_filter
# Confirm that only __kmalloc is enabled.
cat enabled_functions

echo 1 > tracing_on

# Latest events.
head trace

# Observe trace continuously, and drain seen events out.
cat trace_pipe &

Sample output:

# tracer: function
#
# entries-in-buffer/entries-written: 97/97   #P:1
#
#                              _-----=> irqs-off
#                             / _----=> need-resched
#                            | / _---=> hardirq/softirq
#                            || / _--=> preempt-depth
#                            ||| /     delay
#           TASK-PID   CPU#  ||||    TIMESTAMP  FUNCTION
#              | |       |   ||||       |         |
            head-228   [000] ....   825.534637: __kmalloc <-load_elf_phdrs
            head-228   [000] ....   825.534692: __kmalloc <-load_elf_binary
            head-228   [000] ....   825.534815: __kmalloc <-load_elf_phdrs
            head-228   [000] ....   825.550917: __kmalloc <-__seq_open_private
            head-228   [000] ....   825.550953: __kmalloc <-tracing_open
            head-229   [000] ....   826.756585: __kmalloc <-load_elf_phdrs
            head-229   [000] ....   826.756627: __kmalloc <-load_elf_binary
            head-229   [000] ....   826.756719: __kmalloc <-load_elf_phdrs
            head-229   [000] ....   826.773796: __kmalloc <-__seq_open_private
            head-229   [000] ....   826.773835: __kmalloc <-tracing_open
            head-230   [000] ....   827.174988: __kmalloc <-load_elf_phdrs
            head-230   [000] ....   827.175046: __kmalloc <-load_elf_binary
            head-230   [000] ....   827.175171: __kmalloc <-load_elf_phdrs

Trace all possible functions, and draw a call graph:

echo 1 > max_graph_depth
echo 1 > events/enable
echo function_graph > current_tracer

Sample output:

# CPU  DURATION                  FUNCTION CALLS
# |     |   |                     |   |   |   |
 0)   2.173 us    |                  } /* ntp_tick_length */
 0)               |                  timekeeping_update() {
 0)   4.176 us    |                    ntp_get_next_leap();
 0)   5.016 us    |                    update_vsyscall();
 0)               |                    raw_notifier_call_chain() {
 0)   2.241 us    |                      notifier_call_chain();
 0) + 19.879 us   |                    }
 0)   3.144 us    |                    update_fast_timekeeper();
 0)   2.738 us    |                    update_fast_timekeeper();
 0) ! 117.147 us  |                  }
 0)               |                  _raw_spin_unlock_irqrestore() {
 0)   4.045 us    |                    _raw_write_unlock_irqrestore();
 0) + 22.066 us   |                  }
 0) ! 265.278 us  |                } /* update_wall_time */

TODO: what do + and ! mean?

Each enable under the events/ tree enables a certain set of functions, the higher the enable more functions are enabled.

11.6. User mode Linux

I once got UML running on a minimal Buildroot setup at: https://unix.stackexchange.com/questions/73203/how-to-create-rootfs-for-user-mode-linux-on-fedora-18/372207#372207

But in part because it is dying, I didn’t spend much effort to integrate it into this repo, although it would be a good fit in principle, since it is essentially a virtualization method.

Maybe some brave should will send a pull request one day.

12. QEMU

Some QEMU specific features to play with and limitations to cry over.

12.1. Snapshot

https://stackoverflow.com/questions/40227651/does-qemu-emulator-have-checkpoint-function/48724371#48724371

QEMU allows us to take snapshots at any time through the monitor.

You can then restore CPU, memory and disk state back at any time.

qcow2 filesystems must be used for that to work.

To test it out, login into the VM with and run:

/count.sh

On another shell, take a snapshot:

echo 'savevm my_snap_id' | ./qemumonitor

The counting continues.

Restore the snapshot:

echo 'loadvm  my_snap_id' | ./qemumonitor

and the counting goes back to where we saved. This shows that CPU and memory states were reverted.

We can also verify that the disk state is also reversed. Guest:

echo 0 >f

Monitor:

echo 'savevm my_snap_id' | ./qemumonitor

Guest:

echo 1 >f

Monitor:

echo 'loadvm my_snap_id' | ./qemumonitor

Guest:

cat f

And the output is 0.

Our setup does not allow for snapshotting while using initrd.

12.2. Educational hardware models

We have added and interacted with a few educational hardware models in QEMU.

This is useful to learn:

• how to create new hardware models for QEMU. Overview: https://stackoverflow.com/questions/28315265/how-to-add-a-new-device-in-qemu-source-code

• how the Linux kernel interacts with hardware

To get started, have a look at the "Hardware device drivers" section under kernel_module/README.adoc, and try to run those modules, and then grep the QEMU source code.

12.3. 9P

This protocol allows sharing a mountable filesystem between guest and host.

With networking, it’s boring, we can just use any of the old tools like sshfs and NFS.

https://superuser.com/questions/628169/how-to-share-a-directory-with-the-host-without-networking-in-qemu

One advantage of this method over NFS is that can run without sudo on host, or having to pass host credentials on guest for sshfs.

TODO performance compared to NFS.

As usual, we have already set everything up for you. On host:

cd 9p
uname -a > host

Guest:

cd /mnt/9p
cat host
uname -a > guest

Host:

cat guest

The main ingredients for this are:

• 9P settings in our kernel_config_fragment

• 9p entry on our rootfs_overlay/etc/fstab

  Alternatively, you could also mount your own with:

  mkdir /mnt/my9p
  mount -t 9p -o trans=virtio,version=9p2000.L host0 /mnt/my9p

• Launch QEMU with -virtfs as in your run script

  When we tried:

  security_model=mapped

  writes from guest failed due to user mismatch problems: https://serverfault.com/questions/342801/read-write-access-for-passthrough-9p-filesystems-with-libvirt-qemu

The feature is documented at: https://wiki.qemu.org/Documentation/9psetup

12.3.1. 9P gem5

Seems possible! Lets do it:

• http://gem5.org/wiki/images/b/b8/Summit2017_wa_devlib.pdf

• http://gem5.org/WA-gem5

12.3.2. OverlayFS

It would be uber awesome if we could overlay a 9p filesystem on top of the root.

That would allow us to have a second Buildroot target/ directory, and without any extra configs, keep the root filesystem image small, which implies:

• less host disk usage, no need to copy the entire target/ to the image again

• faster rebuild turnaround:

  • no need to regenerate the root filesystem at all and reboot

  • overcomes the check_bin_arch problem: Buildroot rebuild is slow when the root filesystem is large

• no need to worry about BR2_TARGET_ROOTFS_EXT2_SIZE

But TODO we didn’t get it working yet:

• https://stackoverflow.com/questions/41119656/how-can-i-overlayfs-the-root-filesystem-on-linux

• https://unix.stackexchange.com/questions/316018/how-to-use-overlayfs-to-protect-the-root-filesystem

Test with the script:

/overlayfs.sh

It shows that files from the upper/ does not show on the root.

Furthermore, if you try to mount the root elsewhere to prepare for a chroot:

/overlayfs.sh / /overlay
# chroot /overlay

it does not work well either because sub filesystems like /proc do not show on the mount:

ls /overlay/proc

A less good alternative is to set LD_LIBRARY_PATH on the 9p mount and run executables directly from the mount.

Even mor awesome than chroot be to pivot_root, but I couldn’t get that working either:

• https://stackoverflow.com/questions/28015688/pivot-root-device-or-resource-busy

• https://unix.stackexchange.com/questions/179788/pivot-root-device-or-resource-busy

12.4. Guest host networking

12.4.1. Host to guest

Guest, BusyBox nc enabled with CONFIG_NC=y:

nc -l -p 45455

Host, nc from the netcat-openbsd package:

echo asdf | nc localhost 45455

Then asdf appears on the guest.

Only this specific port works by default since we have forwarded it on the QEMU command line.

We us this exact procedure to connect to gdbserver.

12.4.1.1. ssh into guest

https://unix.stackexchange.com/questions/124681/how-to-ssh-from-host-to-guest-using-qemu/307557#307557

Uses OpenSSH’s sshd.

Not enabled by default due to the build / runtime overhead, but it was tested and worked at the time of this commit.

12.4.2. Guest to host

TODO. There is guestfwd, which sounds analogous to hostwfd used in the other sense, but I was not able to get it working, e.g.:

-netdev user,hostfwd=tcp::45455-:45455,guestfwd=tcp::45456-,id=net0 \

gives:

Could not open guest forwarding device 'guestfwd.tcp.45456'

Related:

• https://serverfault.com/questions/769874/how-to-forward-a-port-from-guest-to-host-in-qemu-kvm

12.5. QEMU user mode

This has nothing to do with the Linux kernel, but it is cool:

sudo apt-get install qemu-user
./build -a arm
cd buildroot/output.arm~/target
qemu-arm -L . bin/ls

This uses QEMU’s user-mode emulation mode that allows us to run cross-compiled userland programs directly on the host.

The reason this is cool, is that ls is not statically compiled, but since we have the Buildroot image, we are still able to find the shared linker and the shared library at the given path.

In other words, much cooler than:

arm-linux-gnueabi-gcc -o hello -static hello.c
qemu-arm hello

It is also possible to compile QEMU user mode from source with BR2_PACKAGE_HOST_QEMU_LINUX_USER_MODE=y, but then your compilation will likely fail with:

package/qemu/qemu.mk:110: *** "Refusing to build qemu-user: target Linux version newer than host's.".  Stop.

since we are using a bleeding edge kernel, which is a sanity check in the Buildroot QEMU package.

Anyways, this warns us that the userland emulation will likely not be reliable, which is good to know. TODO: where is it documented the host kernel must be as new as the target one?

GDB step debugging is also possible with:

qemu-arm -g 1234 -L . bin/ls
../host/usr/bin/arm-buildroot-linux-uclibcgnueabi-gdb -ex 'target remote localhost:1234'

TODO: find source. Lazy now.

12.6. Debug QEMU

When you start interacting with QEMU hardware, it is useful to see what is going on inside of QEMU itself.

This is of course trivial since QEMU is just an userland program on the host, but we make it a bit easier with:

./run -D

Then you could:

b edu_mmio_read
c

And in QEMU:

/pci.sh

Just make sure that you never click inside the QEMU window when doing that, otherwise you mouse gets captured forever, and the only solution I can find is to go to a TTY with Ctrl + Alt + F1 and kill QEMU.

You can still send key presses to QEMU however even without the mouse capture, just either click on the title bar, or alt tab to give it focus.

12.7. QEMU record and replay

QEMU supports deterministic record and replay by saving external inputs, which would be awesome to understand the kernel, as you would be able to examine a single run as many times as you would like.

Unfortunately it is not working in the current QEMU: https://stackoverflow.com/questions/46970215/how-to-use-qemus-deterministic-record-and-replay-feature-for-a-linux-kernel-boo

Alternatively, mozilla/rr claims it is able to run QEMU: but using it would require you to step through QEMU code itself. Likely doable, but do you really want to?

12.8. QEMU GUI is unresponsive

Sometimes in Ubuntu 14.04, after the QEMU SDL GUI starts, it does not get updated after keyboard strokes, and there are artifacts like disappearing text.

We have not managed to track this problem down yet, but the following workaround always works:

Ctrl + Shift + U
Ctrl + C
root

This started happening when we switched to building QEMU through Buildroot, and has not been observed on later Ubuntu.

Using text mode is another workaround if you don’t need GUI features.

13. gem5

13.1. gem5 getting started

gem5 is a system simulator, much like QEMU: http://gem5.org/

For the most part, just add the -g option to the QEMU commands and everything should magically work:

./configure -gq && ./build -a arm -g
./run -a arm -g

On another shell:

./gem5-shell

A full rebuild is currently needed even if you already have QEMU working unfortunately, see: gem5 and QEMU with the same kernel configuration

Tested architectures:

• arm

• aarch64

• x86_64

Like QEMU, gem5 also has a syscall emulation mode (SE), but in this tutorial we focus on the full system emulation mode (FS). For a comparison see: https://stackoverflow.com/questions/48986597/when-should-you-use-full-system-fs-vs-syscall-emulation-se-with-userland-program

13.2. gem5 vs QEMU

• advantages of gem5:

  • simulates a generic more realistic pipelined and optionally out of order CPU cycle by cycle, including a realistic DRAM memory access model with latencies, caches and page table manipulations. This allows us to:

    • do much more realistic performance benchmarking with it, which makes absolutely no sense in QEMU, which is purely functional

    • make certain functional cache observations that are not possible in QEMU, e.g.:

      • use Linux kernel APIs that flush memory like DMA, which are crucial for driver development. In QEMU, the driver would still work even if we forget to flush caches.

      • TODO spectre / meltdown

        It is not of course truly cycle accurate, as that

  • would require exposing proprietary information of the CPU designs: https://stackoverflow.com/questions/17454955/can-you-check-performance-of-a-program-running-with-qemu-simulator/33580850#33580850

  • would make the simulation even slower TODO confirm, by how much

    but the approximation is reasonable.

    It is used mostly for microarchitecture research purposes: when you are making a new chip technology, you don’t really need to specialize enormously to an existing microarchitecture, but rather develop something that will work with a wide range of future architectures.

  • runs are deterministic by default, unlike QEMU which has a special [record-and-replay] mode, that requires first playing the content once and then replaying

• disadvantage of gem5: slower than QEMU, see: gem5 vs QEMU performance

  This implies that the user base is much smaller, since no Android devs.

  Instead, we have only chip makers, who keep everything that really works closed, and researchers, who can’t version track or document code properly >:-) And this implies that:

  • the documentation is more scarce

  • it takes longer to support new hardware features

  Well, not that AOSP is that much better anyways.

• not sure: gem5 has BSD license while QEMU has GPL

  This suits chip makers that want to distribute forks with secret IP to their customers.

  On the other hand, the chip makers tend to upstream less, and the project becomes more crappy in average :-)

13.2.1. gem5 vs QEMU performance

We have benchmarked a Linux kernel boot at commit da79d6c6cde0fbe5473ce868c9be4771160a003b with the commands:

# Try to manually hit Ctrl + C as soon as system shutdown message appears.
time ./run -a arm -e 'init=/poweroff.out'
time ./run -a arm -e 'm5 exit' -g
time ./run -a arm -e 'm5 exit' -g -- --caches --cpu-type=HPI
time ./run -a x86_64 -e 'init=/poweroff.out'
time ./run -a x86_64 -e 'init=/poweroff.out' -- -enable-kvm
time ./run -a x86_64 -e 'init=/poweroff.out' -g

and the results were:

Emulator	Time	N times slower than QEMU
QEMU ARM	6 seconds	1
gem5 ARM AtomicSimpleCPU	1 minute 40 seconds	17
gem5 ARM HPI	10 minutes	100
QEMU X86_64	4 seconds	1
QEMU X86_64 KVM	2 seconds	0.5
gem5 X86_64	5 minutes 30 seconds	82

tested on the P51.

13.3. gem5 run benchmark

OK, this is why we used gem5 in the first place, performance measurements!

https://stackoverflow.com/questions/48944587/how-to-count-the-number-of-cpu-clock-cycles-between-the-start-and-end-of-a-bench

Let’s benchmark Dhrystone which Buildroot provides:

./gem5-bench dhrystone 1000
./gem5-bench -r dhrystone 1000

These commands output the approximate number of CPU cycles it took Dhrystone to run.

It works like this:

• the first command boots linux with the default simplified AtomicSimpleCPU, and generates a checkpoint after the kernel boots and before running the benchmark

• the second command restores the checkpoint with the more detailed HPI CPU model, and runs the benchmark. We don’t boot with it because that is much slower.

ARM employees have just been modifying benchmarking code with instrumentation directly: https://github.com/arm-university/arm-gem5-rsk/blob/aa3b51b175a0f3b6e75c9c856092ae0c8f2a7cdc/parsec_patches/xcompile-patch.diff#L230

A few imperfections of our benchmarking method are:

• when we do m5 resetstats and m5 exit, there is some time passed before the exec system call returns and the actual benchmark starts and ends

• the benchmark outputs to stdout, which means so extra cycles in addition to the actual computation. But TODO: how to get the output to check that it is correct without such IO cycles?

Those problems should be insignificant if the benchmark runs for long enough however.

TODO: even if we don’t switch to the detailed CPU model, the cycle counts on the original run and the one with checkpoint restore differ slightly. Why? Multiple checkpoint restores give the same results as expected however:

./run -a arm -E 'm5 checkpoint;m5 resetstats;dhrystone 1000;m5 exit' -g
./run -a arm -g -- -r 1

Now you can play a fun little game with your friends:

• pick a computational problem

• make a program that solves the computation problem, and outputs output to stdout

• write the code that runs the correct computation in the smallest number of cycles possible

To find out why your program is slow, a good first step is to have a look at the statistics for the run:

cat m5out/stats.txt

Whenever we run m5 dumpstats or m5 exit, a section with the following format is added to that file:

---------- Begin Simulation Statistics ----------
[the stats]
---------- End Simulation Statistics   ----------

13.3.1. gem5 system parameters

Besides optimizing a program for a given CPU setup, chip developers can also do the inverse, and optimize the chip for a given benchmark!

The rabbit hole is likely deep, but let’s scratch a bit of the surface.

13.3.1.1. gem5 number of cores

./run -a arm -c 2 -g

Check with:

cat /proc/cpuinfo
getconf _NPROCESSORS_CONF

13.3.1.2. gem5 cache size

A quick ./run -g -- -h leads us to the options:

--caches
--l1d_size=1024
--l1i_size=1024
--l2cache
--l2_size=1024
--l3_size=1024

But keep in mind that it only affects benchmark performance of the most detailed CPU types:

arch	CPU type	caches used
X86	AtomicSimpleCPU	no
X86	DerivO3CPU	?*
ARM	AtomicSimpleCPU	no
ARM	HPI	yes

*: couldn’t test because of:

• https://stackoverflow.com/questions/49011096/how-to-switch-cpu-models-in-gem5-after-restoring-a-checkpoint-and-then-observe-t

• gem5/gem5#16

This has been verified with:

m5 resetstats && dhrystone 10000 && m5 dumpstats

at commit da79d6c6cde0fbe5473ce868c9be4771160a003b with the following gem5 commands cycle counts:

# 11M
./run -a arm -g
./run -a arm -g -- --caches --l2cache

# 175M
./run -a arm -g -- --caches --l1d_size=1024   --l1i_size=1024   --l2cache --l2_size=1024   --l3_size=1024   --cpu-type=HPI

# 16M
./run -a arm -g -- --caches --l1d_size=1024MB --l1i_size=1024MB --l2cache --l2_size=1024MB --l3_size=1024MB --cpu-type=HPI

# 20M
./run -a x86_64 -g -- --caches --l1d_size=1024 --l2cache --l2_size=1024 --l3_size=1024
./run -a x86_64 -g -- --caches --l1d_size=1024MB --l2cache --l2_size=1024MB --l3_size=1024MB

Cache sizes can in theory be checked with the methods described at: https://superuser.com/questions/55776/finding-l2-cache-size-in-linux:

getconf -a | grep CACHE
lscpu
cat /sys/devices/system/cpu/cpu0/cache/index2/size

but for some reason the Linux kernel is not seeing the cache sizes:

• https://stackoverflow.com/questions/49008792/why-doesnt-the-linux-kernel-see-the-cache-sizes-in-the-gem5-emulator-in-full-sy

• http://gem5-users.gem5.narkive.com/4xVBlf3c/verify-cache-configuration

Behaviour breakdown:

• arm QEMU and gem5 (both AtomicSimpleCPU or HPI), x86 gem5: /sys files don’t exist, and getconf and lscpu value empty

• x86 QEMU: /sys files exist, but getconf and lscpu values still empty

13.3.1.3. gem5 memory latency

TODO These look promising:

--list-mem-types
--mem-type=MEM_TYPE
--mem-channels=MEM_CHANNELS
--mem-ranks=MEM_RANKS
--mem-size=MEM_SIZE

TODO: now to verify this with the Linux kernel? Besides raw performance benchmarks.

13.3.1.4. gem5 memory size

./run -a arm -m 512M

and verify inside the guest with:

free -m

13.3.1.5. gem5 disk and network latency

TODO These look promising:

--ethernet-linkspeed
--ethernet-linkdelay

and also: gem5-dist: https://publish.illinois.edu/icsl-pdgem5/

13.3.1.6. gem5 clock frequency

Clock frequency: TODO how does it affect performance in benchmarks?

./run -a arm -g -- --cpu-clock 10000000

Check with:

m5 resetstats && sleep 10 && m5 dumpstats

and then:

grep numCycles m5out/stats.txt

TODO: why doesn’t this exist:

ls /sys/devices/system/cpu/cpu0/cpufreq

13.3.2. Enable compiler optimizations

If you are benchmarking compiled programs instead of hand written assembly, remember that we configure Buildroot to disable optimizations by default with:

BR2_OPTIMIZE_0=y

to improve the debugging experience.

You will likely want to change that to:

BR2_OPTIMIZE_3=y

and do a full rebuild.

TODO is it possible to compile a single package with optimizations enabled? In any case, this wouldn’t be very representative, since calls to an unoptimized libc will also have an impact on performance. Kernel-wise it should be fine though, since the kernel requires O=2.

13.3.3. Interesting benchmarks

Buildroot built-in libraries, mostly under Libraries > Other:

• Armadillo C++: linear algebra

• fftw: Fourier transform

• Eigen: linear algebra

• Flann

• GSL: various

• liblinear

• libspacialindex

• libtommath

• qhull

There are not yet enabled, but it should be easy to so:

• enable them in br2 and rebuild

• create a test program that uses each library under kernel_module/user

External open source benchmarks. We will try to create Buildroot packages for them, add them to this repo, and potentially upstream:

• http://parsec.cs.princeton.edu/ Mentioned on docs: http://gem5.org/PARSEC_benchmarks

• http://www.m5sim.org/Splash_benchmarks

13.3.3.1. BLAS

Buildroot supports it, which makes everything just trivial:

printf 'BR2_PACKAGE_OPENBLAS=y\n' >> br2.gitignore
printf 'BR2_TARGET_ROOTFS_EXT2_SIZE="128M"\n' >> br2.gitignore
./build -a arm -g -b br2.gitignore -- kernel_module-reconfigure

and then inside the guest run our test program:

/openblas.out

For x86, you also need:

printf 'BR2_PACKAGE_OPENBLAS_TARGET="NEHALEM"\n' >> br2.gitignore

to overcome this bug: https://bugs.busybox.net/show_bug.cgi?id=10856

sgemm_kernel.o: No such file or directory

13.3.3.2. PARSEC benchmark

We have ported parts of the PARSEC benchmark for cross compilation at: https://github.com/cirosantilli/parsec-benchmark See the documentation on that repo to find out which benchmarks have been ported. Furthermore, some of the benchmarks were are segfaulting, see parsec-benchmark/test.sh

There are two ways to run PARSEC with this repo:

• without pasecmgmt, most likely what you want

• with pasecmgmt

13.3.3.2.1. PARSEC benchmark without parsecmgmt

configure -gpq && ./build -a arm -g -b br2_parsec
./run -a arm -g

Once inside the guest, launch one of the test input sized benchmarks manually as in:

cd /parsec/ext/splash2x/apps/fmm/run
../inst/arm-linux.gcc/bin/fmm 1 < input_1

To find out how to run many of the benchmarks, you can either:

• have a look at: parsec-benchmark/test.sh

• do a search on the build stdout on your terminal for a line of type:

  Running /parsec/ext/splash2x/apps/fmm/inst/arm-linux.gcc/bin/fmm 1 < input_1:

  Yes, we do run the benchmarks on host just to unpack / generate inputs…​ and they almost always fail to run since they were build for the guest instead of host. Hopefully, since we don’t want to wait for them to finish anyways.

• have a quick peak at the package sources, usually src/run.sh and parsec/*.runconf.

PARSEC simply wasn’t designed with non native machines in mind.

Running a benchmark of a different size requires a rebuild wit:

./build \
  -a arm \
  -c 'BR2_PACKAGE_PARSEC_BENCHMARK_INPUT_SIZE="simsmall"' \
  -c BR2_TARGET_ROOTFS_EXT2_SIZE="500M" \
  -g \
  -b br2_parsec \
  -- parsec-benchmark-reconfigure \
;

and then try running the benchmarks as before.

The rebuild is required because some of the input sizes

Separating input sizes also allows to create smaller images when only running the smaller benchmarks.

13.3.3.2.2. BR2_TARGET_ROOTFS_EXT2_SIZE

When adding new large package to the Buildroot root filesystem, it may fail with the message:

Maybe you need to increase the filesystem size (BR2_TARGET_ROOTFS_EXT2_SIZE)

The solution is to simply add:

./build -c BR2_TARGET_ROOTFS_EXT2_SIZE="500M"

where 500M is "large enough".

Note that dots cannot be used as in 1.5G, so just use Megs as in 1500M instead.

Unfortunately, TODO we don’t have a perfect way to find the right value for BR2_TARGET_ROOTFS_EXT2_SIZE. One good heuristic is:

du -hsx buildroot/output.arm-gem5~/target/parsec

https://stackoverflow.com/questions/49211241/is-there-a-way-to-automatically-detect-the-minimum-required-br2-target-rootfs-ex

One way to overcome this problem is to mount benchmarks from host instead of adding them to the root filesystem, e.g. with: 9P

13.3.3.2.3. PARSEC benchmark with parsecmgmt

Most users won’t want to use this method because:

• running the parsecmgmt Bash scripts takes forever before it ever starts running the actual benchmarks on gem5

  Running on QEMU is feasible, but not the main use case, since QEMU cannot be used for performance measurements

• it requires putting the full .tar inputs on the guest, which makes the image twice as large (1x for the .tar, 1x for the unpacked input files)

It would be awesome if it were possible to use this method, since this is what Parsec supports officially, and so:

• you don’t have to dig into what raw command to run

• there is an easy way to run all the benchmarks in one go to test them out

• you can just run any of the benchmarks that you want

but it simply is not feasible in gem5 because it takes too long.

If you still want to run this, try it out with:

./build -a arm \
  -c BR2_TARGET_ROOTFS_EXT2_SIZE="3G" \
  -g
  -b br2_parsec
  -- parsec-benchmark-reconfigure \
;

And then you can run it just as you would on the host:

cd /parsec/
bash
. env.sh
parsecmgmt -a run -p splash2x.fmm -i test

13.3.3.2.4. PARSEC change the input size

One limitation is that only one input size is available on the guest for a given build.

To change that, edit br2_parsec to contain for example:

BR2_PACKAGE_PARSEC_BENCHMARK_INPUT_SIZE=simsmall

and then rebuild with:

./build -a arm -g -b br2_parsec -- parsec-benchmark-reconfigure

This limitation exists because parsecmgmt generates the input files just before running via the Bash scripts, but we can’t run parsecmgmt on gem5 as it is too slow!

One option would be to do that inside the guest with QEMU, but this would required a full rebuild due to gem5 and QEMU with the same kernel configuration.

Also, we can’t generate all input sizes at once, because many of them have the same name and would overwrite one another…​ Parsec clearly needs a redesign for embedded, maybe we will patch it later.

13.3.3.2.5. PARSEC uninstall

If you want to remove PARSEC later, Buildroot doesn’t provide an automated package removal mechanism as documented at: https://github.com/buildroot/buildroot/blob/2017.08/docs/manual/rebuilding-packages.txt#L90, but the following procedure should be satisfactory:

rm -rf \
  ./buildroot/dl/parsec-* \
  ./buildroot/output.arm-gem5~/build/parsec-* \
  ./buildroot/output.arm-gem5~/build/packages-file-list.txt \
  ./buildroot/output.arm-gem5~/images/rootfs.* \
  ./buildroot/output.arm-gem5~/target/parsec-* \
;
./build -a arm -g

13.3.3.2.6. PARSEC benchmark hacking

If you end up going inside parsec-benchmark/parsec-benchmark to hack up the benchmark (you will!), these tips will be helpful.

Buildroot was not designed to deal with large images, and currently cross rebuilds are a bit slow, due to some image generation and validation steps.

A few workarounds are:

• develop in host first as much as you can. Our PARSEC fork supports it.

  If you do this, don’t forget to do a:

  cd parsec-benchmark/parsec-benchmark
  git clean -xdf .

  before going for the cross compile build.

• patch Buildroot to work well, and keep cross compiling all the way. This should be totally viable, and we should do it.

  Don’t forget to explicitly rebuild PARSEC with:

  ./build -a arm -g -b br2_parsec parsec-benchmark-reconfigure

  You may also want to test if your patches are still functionally correct inside of QEMU first, which is a faster emulator.

• sell your soul, and compile natively inside the guest. We won’t do this, not only because it is evil, but also because Buildroot explicitly does not support it: https://buildroot.org/downloads/manual/manual.html#faq-no-compiler-on-target ARM employees have been known to do this: https://github.com/arm-university/arm-gem5-rsk/blob/aa3b51b175a0f3b6e75c9c856092ae0c8f2a7cdc/parsec_patches/qemu-patch.diff

13.3.3.2.7. Buildroot rebuild is slow when the root filesystem is large

Buildroot is not designed for large root filesystem images, and the rebuild becomes very slow when we add a large package to it.

This is due mainly to the pkg-generic GLOBAL_INSTRUMENTATION_HOOKS sanitation which go over the entire tree doing complex operations…​ I no like, in particular check_bin_arch and check_host_rpath, which get stuck for a long time on the message:

>>>   Sanitizing RPATH in target tree

The pause is followed by:

buildroot/output.arm~/build/parsec-benchmark-custom/.stamp_target_installed

so which shows that the whole delay is inside our install itself.

I put an echo f in check_bin_arch, and it just loops forever, does not stop on a particular package.

13.4. gem5 kernel command line parameters

Analogous to QEMU:

./run -a arm -e 'init=/poweroff.out' -g

Internals: when we give --command-line= to gem5, it overrides default command lines, including some mandatory ones which are required to boot properly.

Our run script hardcodes the require options in the default --command-line and appends extra options given by -e.

To find the default options in the first place, we removed --command-line and ran:

./run -a arm -g

and then looked at the line of the Linux kernel that starts with:

Kernel command line:

13.5. gem5 GDB step debugging

Analogous to QEMU, on the first shell:

./run -a arm -d -g

On the second shell:

./rungdb -a arm -g

This makes the VM stop, so from inside GDB:

continue

On a third shell:

./gem5-shell

And we now see the boot messages, and then get a shell. Now try the /continue.sh procedure described for QEMU.

TODO: how to stop at start_kernel? gem5 listens for GDB by default, and therefore does not wait for a GDB connection to start like QEMU does. So when GDB connects we might have already passed start_kernel. Maybe --debug-break=0 can be used?

13.6. gem5 checkpoint

Analogous to QEMU’s Snapshot, but better since it can be started from inside the guest, so we can easily checkpoint after a specific guest event, e.g. just before init is done.

Documentation: http://gem5.org/Checkpoints

./run -a arm -g

In the guest, wait for the boot to end and run:

m5 checkpoint

To restore the checkpoint, kill the VM and run:

./run -a arm -g -- -r 1

Let’s create a second checkpoint to see how it works, in guest:

date >f
m5 checkpoint

Kill the VM, and try it out:

./run -a arm -g -- -r 2

and now in the guest:

cat f

contains the date. The file f wouldn’t exist had we used the first checkpoint with -r 1.

If you automate things with Kernel command line parameters as in:

./run -a arm -E 'm5 checkpoint;m5 resetstats;dhrystone 1000;m5 exit' -g

Then there is no need to pass the kernel command line again to gem5 for replay:

./run -a arm -g -- -r 1

since boot has already happened, and the parameters are already in the RAM of the snapshot.

Our scripts "namespace" with the checkpoint by architecture with --checkpoint-dir, so if you make two checkpoints:

• one in x86

• the other in arm

Then you would still restore both of them with -- -r 1.

This makes it easier to remember which checkpoint is which, especially since there appears to be no runtime way to set the checkpoint names.

Internals:

• the checkpoints are stored under m5out/cpts/$arch/cpt.$todo_whatisthis

• m5 is a guest utility present inside the gem5 tree which we cross-compiled and installed into the guest

13.6.1. gem5 restore checkpoint with a different CPU

gem5 can switch to a different CPU model when restoring a checkpoint.

A common combo is to boot Linux with a fast CPU, make a checkpoint and then replay the benchmark of interest with a slower CPU.

An illustrative interactive run:

./run -a arm -g

In guest:

m5 checkpoint

And then restore the checkpoint with a different CPU:

./run -a arm -g -- --caches -r 1 --restore-with-cpu=HPI

13.7. Pass extra options to gem5

Pass options to the fs.py script:

• get help:

  ./run -g -- -h

• boot with the more detailed and slow HPI CPU model:

  ./run -a arm -g -- --caches --cpu-type=HPI

Pass options to the gem5 executable itself:

• get help:

  ./run -G '-h' -g

13.8. Run multiple gem5 instances at once

gem5 just assigns new ports if some ports are occupied, so we can do:

./run -g
# Same as ./gem5-shell 0
./gem5-shell

And a second instance:

./run -g
./gem5-shell 1

TODO Now we just need to network them up to have some more fun! See dist-gem5: http://publish.illinois.edu/icsl-pdgem5/

13.9. gem5 and QEMU with the same kernel configuration

We would like to be able to run both gem5 and QEMU with the same kernel build to avoid duplication, but TODO we haven’t been able to get that working yet.

This documents our failed attempts so far.

As a result, we currently have to create two full buildroot/output* directories, which means two full GCC builds.

13.9.1. QEMU with gem5 kernel configuration

To test this, hack up run to use the buildroot/output.arm-gem5~ directory, and then run:

./run -a arm

Now QEMU hangs at:

audio: Could not init `oss' audio driver

and the display shows:

Guest has not initialized the display (yet).

13.9.2. gem5 with QEMU kernel configuration

Test it out with:

./run -a arm -g

TODO hangs at:

**** REAL SIMULATION ****
warn: Existing EnergyCtrl, but no enabled DVFSHandler found.
info: Entering event queue @ 0.  Starting simulation...
1614868500: system.terminal: attach terminal 0

and the telnet at:

$ ./gem5-shell
Trying 127.0.0.1...
Connected to localhost.
Escape character is '^]'.
==== m5 slave terminal: Terminal 0 ====

I have also tried to copy the exact same kernel command line options used by QEMU, but nothing changed.

13.10. gem5 limitations

• networking not working. We currently just disable it from inittab by default to prevent waiting at startup

• gdbserver: https://stackoverflow.com/questions/48941494/how-to-do-port-forwarding-from-guest-to-host-in-gem5

13.10.1. gem5 x86_64 limitations

• gdb debugging not working. First we need to remove the initial set arch i386:x86-64:intel required for QEMU. Then it cannot find the symbols.

14. Insane action

14.1. Run on host

This method runs the kernel modules directly on your host computer without a VM, and saves you the compilation time and disk usage of the virtual machine method.

It has however severe limitations, and you will soon see that the compilation time and disk usage are well worth it:

• can’t control which kernel version and build options to use. So some of the modules will likely not compile because of kernel API changes, since the Linux kernel does not have a stable kernel module API.

• bugs can easily break you system. E.g.:

  • segfaults can trivially lead to a kernel crash, and require a reboot

  • your disk could get erased. Yes, this can also happen with sudo from userland. But you should not use sudo when developing newbie programs. And for the kernel you don’t have the choice not to use sudo

  • even more subtle system corruption such as not being able to rmmod

• can’t control which hardware is used, notably the CPU architecture

• can’t step debug it with GDB easily

Still interested?

cd kernel_module
./make-host.sh

If the compilation of any of the C files fails because of kernel or toolchain differences that we don’t control on the host, just rename it to remove the .c extension and try again:

mv broken.c broken.c~
./build_host

Once you manage to compile, and have come to terms with the fact that this may blow up your host, try it out with:

sudo insmod hello.ko

# Our module is there.
sudo lsmod | grep hello

# Last message should be: hello init
dmest -T

sudo rmmod hello

# Last message should be: hello exit
dmesg -T

# Not present anymore
sudo lsmod | grep hello

Once you are done with this method, you must clean up the in-tree build objects before you decide to do the right thing and move on to the superior ./build Buildroot method:

cd "kernel_module"
./make-host.sh clean

otherwise they will cause problems.

14.2. Hello host

Minimal host build system sanity check example.

cd hello_host
make
insmod hello.ko
dmesg
rmmod hello.ko
dmesg

15. Buildroot

15.1. Change Buildroot options

We provide the following mechanisms:

• ./build -b br2.gitignore: append the file br2.gitignore to a single build. Must be passed every time you run ./build.

  cp br2.gitignore.example br2.gitignore

• ./build -c 'BR2_SOM_OPTION="myval"': append a single option to a single build.

15.2. Find Buildroot options with make menuconfig

make menuconfig is a convenient way to find Buildroot configurations:

cd buildroot/output.x86_64~
make menuconfig

Hit / and search for the settings.

Save and quit.

diff -u .config.olg .config

Then copy and paste the diff additions to br2 to make them permanent.

15.3. ccache

We have enabled ccached builds by default.

BR2_CCACHE_USE_BASEDIR=n is used, which means that:

• absolute paths are used and GDB can find source files

• but builds are not reused across separated LKMC directories

ccache can considerably speed up builds when you:

• are switching between multiple configurations for a given package to bisect something out, as mentioned at: Use your own kernel config

• clean the build because things stopped working. We store the cache outside of this repository, so you can nuke away without fear

The default ccache environment variables are honored if you have them set, which we recommend you do. E.g., in your .bashrc:

export CCACHE_DIR=~/.ccache
export CCACHE_MAXSIZE="20G"

The choice basically comes down to:

• should I store my cache on my HD or SSD?

• how big is my build, and how many build configurations do I need to keep around at a time?

If you don’t set it, the default is to use ~/.buildroot-ccache with 5G, which is a bit small for us.

I find it very relaxing to watch ccache at work with:

watch -n1 'make -C buildroot/output.x86_64~/ ccache-stats'

or if you have it installed on host and the environment variables exported simply with:

watch -n1 'ccache -s'

while a build is going on in another terminal and my cooler is humming. Especially when the hit count goes up ;-) The joys of system programming.

16. Benchmark this repo

In this section document how fast the build and clone are, and how to investigate them.

Send a pull request if you try it out on something significantly different.

16.1. Find which packages are making the build slow

cd buildroot/output.x86_64~
make graph-build graph-depends
xdg-open graphs/build.pie-packages.pdf
xdg-open graphs/graph-depends.pdf

Our philosophy is:

• if something adds little to the build time, build it in by default

• otherwise, make it optional

The build biggest time hog is always GCC, and it does not look like we can use a precompiled one: https://stackoverflow.com/questions/10833672/buildroot-environment-with-host-toolchain

16.2. Benchmark machines

The build times are calculated after doing make source, which downloads the sources, and basically benchmarks the Internet.

16.2.1. P51

Lenovo ThinkPad P51 laptop:

• 2500 USD in 2018 (high end)

• Intel Core i7-7820HQ Processor (8MB Cache, up to 3.90GHz) (4 cores 8 threads)

• 32GB(16+16) DDR4 2400MHz SODIMM

• 512GB SSD PCIe TLC OPAL2

• Ubuntu 17.10

Build time at 2c12b21b304178a81c9912817b782ead0286d282: 28 minutes, 15 with full ccache hits. Breakdown: 19% GCC, 13% Linux kernel, 7% uclibc, 6% host-python, 5% host-qemu, 5% host-gdb, 2% host-binutils

Single file change on ./build kernel_module-reconfigure: 7 seconds.

16.2.1.1. P51 baseline benchmarks

This is the minimal build we could expect to get away with.

On the upstream Buildroot repo at 7d43534625ac06ae01987113e912ffaf1aec2302 we run:

make qemu_x86_64_defconfig
printf 'BR2_CCACHE=y\n' >>.config
make olddefconfig
time make BR2_JLEVEL="$(nproc)"

Time: 11 minutes, 7 with full ccache hits. Breakdown: 47% GCC, 15% Linux kernel, 9% uclibc, 5% host-binutils. Conclusions:

• we have bloated our kernel build 3x with all those delicious features :-)

• GCC time increased 1.5x by our bloat, but its percentage of the total was greatly reduced, due to new packages being introduced.

  make graph-depends shows that most new dependencies come from QEMU and GDB, which we can’t get rid of anyways.

A quick look at the system monitor reveals that the build switches between times when:

• CPUs are at a max, memory is fine. So we must be CPU / memory speed bound. I bet that this happens during heavy compilation.

• CPUs are not at a max, and memory is fine. So we are likely disk bound. I bet that this happens during configuration steps.

This is consistent with the fact that ccache reduces the build time only partially, since ccache should only overcome the CPU bound compilation steps, but not the disk bound ones.

The instructions counts varied very little between the baseline and LKMC, so runtime overhead is not a big deal apparently.

16.2.2. T430

Build time: 2 hours.

TODO specs, SHA.

16.3. Benchmark Internets

16.3.1. 38Mbps

2c12b21b304178a81c9912817b782ead0286d282:

• shallow clone of all submodules: 4 minutes.

• make source: 2 minutes

Google M-lab speed test: 36.4Mbps

17. Conversation

17.1. kmod

https://git.kernel.org/pub/scm/utils/kernel/kmod/kmod.git

Multi-call executable that implements: lsmod, insmod, rmmod, and other tools on desktop distros such as Ubuntu 16.04, where e.g.:

ls -l /bin/lsmod

gives:

lrwxrwxrwx 1 root root 4 Jul 25 15:35 /bin/lsmod -> kmod

and:

dpkg -l | grep -Ei

contains:

ii  kmod                                        22-1ubuntu5                                         amd64        tools for managing Linux kernel modules

BusyBox also implements its own version of those executables. There are some differences.

Buildroot also has a kmod package, but we are not using it since BusyBox' version is good enough so far.

This page will only describe features that differ from kmod to the BusyBox implementation.

17.1.1. module-init-tools

Name of a predecessor set of tools.

17.1.2. kmod modprobe

kmod’s modprobe can also load modules under different names to avoid conflicts, e.g.:

sudo modprobe vmhgfs -o vm_hgfs

17.2. Device tree

platform_device.c together with its kernel and QEMU forks contains a minimal runnable example.

Good format descriptions:

• https://www.raspberrypi.org/documentation/configuration/device-tree.md

Minimal example

/dts-v1/;

/ {
    a;
};

Check correctness with:

dtc a.dts

Separate nodes are simply merged by node path, e.g.:

/dts-v1/;

/ {
    a;
};

/ {
    b;
};

then dtc a.dts gives:

/dts-v1/;

/ {
        a;
        b;
};

17.3. Directory structure

buildroot_patches/README.adoc

global_patch_dir/README.adoc

kernel_module/README.adoc

kernel_module/user/README.adoc

rootfs_overlay/README.adoc

CONTRIBUTING.adoc

17.4. About

This project is for people who want to learn and modify low level system components:

• Linux kernel and Linux kernel modules

• full systems emulators like QEMU and gem5

• C standard libraries. This could also be put on a submodule if people show interest.

• Buildroot. We use and therefore document, a large part of its feature set.

Philosophy:

• automate as much as possible to make things more reproducible

• do everything from source to make things understandable and hackable

This project should be called "Linux kernel playground", like: https://github.com/Fuzion24/AndroidKernelExploitationPlayground maybe I’ll rename it some day. Would semi conflict with: http://copr-fe.cloud.fedoraproject.org/coprs/jwboyer/kernel-playground/ though.

17.4.1. Fairy tale

Once upon a time, there was a boy called Linus.

Linus made a super fun toy, and since he was not very humble, decided to call it Linux.

Linux was an awesome toy, but it had one big problem: it was very difficult to learn how to play with it!

As a result, only some weird kids who were very bored ended up playing with Linux, and everyone thought those kids were very cool, in their own weird way.

One day, a mysterious new kid called Ciro tried to play with Linux, and like many before him, got very frustrated, and gave up.

A few years later, Ciro had grown up a bit, and by chance came across a very cool toy made by the boy Petazzoni and his gang: it was called Buildroot.

Ciro noticed that if you used Buildroot together with Linux, Linux suddenly became very fun to play with!

So Ciro decided to explain to as many kids as possible how to use Buildroot to play with Linux.

And so everyone was happy. Except some of the old weird kernel hackers who wanted to keep their mystique, but so be it.

THE END

17.5. Bibliography

Runnable stuff:

• https://lwn.net/Kernel/LDD3/ the best book, but outdated. Updated source: https://github.com/martinezjavier/ldd3 But examples non-minimal and take too much brain power to understand.

• https://github.com/satoru-takeuchi/elkdat manual build process without Buildroot, very few and simple kernel modules

• https://github.com/tinyclub/linux-lab Buildroot based, no kernel modules?

• https://github.com/agelastic/eudyptula

• https://github.com/linux-kernel-labs Yocto based, source inside a kernel fork subdir: https://github.com/linux-kernel-labs/linux/tree/f08b9e4238dfc612a9d019e3705bd906930057fc/tools/labs which the author would like to upstream https://www.reddit.com/r/programming/comments/79w2q9/linux_device_driver_labs_the_linux_kernel/dp6of43/

• Android AOSP: https://stackoverflow.com/questions/1809774/how-to-compile-the-android-aosp-kernel-and-test-it-with-the-android-emulator/48310014#48310014 AOSP is basically a uber bloated Buildroot (2 hours build vs 30 minutes), Android is Linux based, and QEMU is the emulator backend. These instructions might work for debugging the kernel: https://github.com/Fuzion24/AndroidKernelExploitationPlayground

Theory:

• http://nairobi-embedded.org you will fall here a lot when the hard Google queries start popping. They have covered everything we do here basically, but with a more manual approach, while this repo automates everything.

• https://balau82.wordpress.com awesome low level resource

• https://lwn.net

• http://www.makelinux.net

Awesome lists:

• https://github.com/gurugio/lowlevelprogramming-university

• https://github.com/uhub/awesome-c