Re: [PATCH 2/2] i2c: dev: don't allow user-space to deadlock the kernel

[Bartosz Golaszewski]

On Sun, Dec 11, 2022 at 5:32 PM kernel test robot <yujie.liu@xxxxxxxxx> wrote:
>
> Greeting,
>
> FYI, we noticed WARNING:at_drivers/base/core.c:#device_release due to commit (built with gcc-11):
>
> commit: 22f82cbe7636f59cc21f6837c500d704e3b412a5 ("[PATCH 2/2] i2c: dev: don't allow user-space to deadlock the kernel")
> url: https://github.com/intel-lab-lkp/linux/commits/Bartosz-Golaszewski/i2c-fortify-the-subsystem-against-user-space-induced-deadlocks/20221209-022450
> base: https://git.kernel.org/cgit/linux/kernel/git/wsa/linux.git i2c/for-next
> patch link: https://lore.kernel.org/all/20221208182142.250084-3-brgl@xxxxxxxx/
> patch subject: [PATCH 2/2] i2c: dev: don't allow user-space to deadlock the kernel
>
> in testcase: xfstests
> version: xfstests-x86_64-5a5e419-1_20221128
> with following parameters:
>
>         disk: 4HDD
>         fs: xfs
>         test: xfs-group-07
>
> test-description: xfstests is a regression test suite for xfs and other files ystems.
> test-url: git://git.kernel.org/pub/scm/fs/xfs/xfstests-dev.git
>
> on test machine: 4 threads 1 sockets Intel(R) Core(TM) i3-3220 CPU @ 3.30GHz (Ivy Bridge) with 8G memory
>
> caused below changes (please refer to attached dmesg/kmsg for entire log/backtrace):
>
>
> kern  :warn  : [   35.378637] ------------[ cut here ]------------
> kern  :err   : [   35.388198] Device 'i2c-7' does not have a release() function, it is broken and must be fixed. See Documentation/core-api/kobject.rst.
> kern :warn : [   35.401052] WARNING: CPU: 1 PID: 195 at drivers/base/core.c:2332 device_release (drivers/base/core.c:2332)
> kern  :warn  : [   35.410178] Modules linked in: x86_pkg_temp_thermal intel_powerclamp i915(+) sd_mod t10_pi crc64_rocksoft_generic crc64_rocksoft ipmi_devintf crc64 coretemp ipmi_msghandler sg kvm_intel drm_buddy intel_gtt kvm irqbypass crct10dif_pclmul drm_display_helper crc32_pclmul ttm crc32c_intel ghash_clmulni_intel drm_kms_helper sha512_ssse3 syscopyarea ahci mei_me rapl intel_cstate libahci sysfillrect wmi_bmof sysimgblt i2c_i801 mei intel_uncore libata i2c_smbus lpc_ich fb_sys_fops video wmi drm fuse ip_tables
> kern  :warn  : [   35.455372] CPU: 1 PID: 195 Comm: systemd-udevd Not tainted 6.1.0-rc8-00063-g22f82cbe7636 #1
> kern  :warn  : [   35.464544] Hardware name: Hewlett-Packard HP Pro 3340 MT/17A1, BIOS 8.07 01/24/2013
> kern :warn : [   35.473036] RIP: 0010:device_release (drivers/base/core.c:2332)
> kern :warn : [ 35.478405] Code: 48 8d 7d 50 48 89 fa 48 c1 ea 03 80 3c 02 00 0f 85 89 00 00 00 48 8b 75 50 48 85 f6 74 13 48 c7 c7 60 f0 ce 83 e8 6c 1c f2 00 <0f> 0b e9 0f ff ff ff 48 b8 00 00 00 00 00 fc ff df 48 89 ea 48 c1
> All code

Ah yes, the .release() callback must stay and while the adapter
resources are handled by the bus driver, we still need to free the idr
at device release time.

I'll fix that in v2 but I also want to wait for some comments.

Bart