Re: [PATCH] platform/chrome: cros_ec_debugfs: cancel/schedule logging work only if supported

[Benson Leung]

Hi Guenter,

On Tue, Mar 19, 2019 at 09:32:36AM -0700, Guenter Roeck wrote:
> The following traceback was reported on ASUS C201, which does not support
> console logging.
> 
> ------------[ cut here ]------------
> WARNING: CPU: 2 PID: 361 at kernel/workqueue.c:3030 __flush_work+0x38/0x154
> Modules linked in: snd_soc_hdmi_codec cros_ec_debugfs cros_ec_sysfs uvcvideo dw_hdmi_cec dw_hdmi_i2s_audio videobuf2_vmalloc cfg80211 gpio_charger rk_crypto rfkill videobuf2_memops videobuf2_v4l2 des_generic videobuf2_common ofpart m25p80 spi_nor tpm_i2c_infineon sbs_battery mtd tpm joydev cros_ec_dev coreboot_table evdev mousedev ip_tables x_tables [last unloaded: brcmutil]
> CPU: 2 PID: 361 Comm: systemd-sleep Not tainted 5.1.0-rc1-1-ARCH+ #1
> Hardware name: Rockchip (Device Tree)
> [<c020e4b0>] (unwind_backtrace) from [<c020ac18>] (show_stack+0x10/0x14)
> [<c020ac18>] (show_stack) from [<c07a3e04>] (dump_stack+0x7c/0x9c)
> [<c07a3e04>] (dump_stack) from [<c0222748>] (__warn+0xd0/0xec)
> [<c0222748>] (__warn) from [<c022279c>] (warn_slowpath_null+0x38/0x44)
> [<c022279c>] (warn_slowpath_null) from [<c02365d0>] (__flush_work+0x38/0x154)
> [<c02365d0>] (__flush_work) from [<c023786c>] (__cancel_work_timer+0x114/0x1a4)
> [<c023786c>] (__cancel_work_timer) from [<bf33233c>] (cros_ec_debugfs_suspend+0x14/0x1c [cros_ec_debugfs])
> [<bf33233c>] (cros_ec_debugfs_suspend [cros_ec_debugfs]) from [<c056a888>] (dpm_run_callback+0x64/0xcc)
> [<c056a888>] (dpm_run_callback) from [<c056ad2c>] (__device_suspend+0x174/0x3a8)
> [<c056ad2c>] (__device_suspend) from [<c056b9e0>] (dpm_suspend+0x174/0x1e0)
> [<c056b9e0>] (dpm_suspend) from [<c026b3e0>] (suspend_devices_and_enter+0x6c/0x50c)
> [<c026b3e0>] (suspend_devices_and_enter) from [<c026ba8c>] (pm_suspend+0x20c/0x274)
> [<c026ba8c>] (pm_suspend) from [<c026a628>] (state_store+0x54/0x88)
> [<c026a628>] (state_store) from [<c03cd2d0>] (kernfs_fop_write+0x114/0x180)
> [<c03cd2d0>] (kernfs_fop_write) from [<c035d48c>] (__vfs_write+0x1c/0x154)
> [<c035d48c>] (__vfs_write) from [<c035f9e8>] (vfs_write+0xb8/0x198)
> [<c035f9e8>] (vfs_write) from [<c035fbc0>] (ksys_write+0x3c/0x74)
> [<c035fbc0>] (ksys_write) from [<c0201000>] (ret_fast_syscall+0x0/0x4c)
> Exception stack(0xe9365fa8 to 0xe9365ff0)
> 5fa0: 00000004 beef8b28 00000004 beef8b28 00000004 00000000
> 5fc0: 00000004 beef8b28 02319170 00000004 beef8b28 00000004 b6f3d900 beef8b74
> 5fe0: 0000006c beef8a98 b6c0adac b6c66534
> ---[ end trace f4ee5df14e8ea0ec ]---
> 
> If console logging is not supported, the work structure is never
> initialized, resulting in the traceback. Calling cancel/schedule functions
> conditionally fixes the problem.
> 
> While at it, also fix error handling in the probe function.
> 
> Reported-by: Urja Rannikko <urjaman@xxxxxxxxx>
> Cc: Urja Rannikko <urjaman@xxxxxxxxx>
> Fixes: 6fce0a2cf5a05 ("mfd / platform: cros_ec: Move debugfs attributes to its own driver")
> Signed-off-by: Guenter Roeck <linux@xxxxxxxxxxxx>


Looks good. Applied to 5.1-fixes. 

Thanks for the quick fix.
Benson


-- 
Benson Leung
Staff Software Engineer
Chrome OS Kernel
Google Inc.
bleung@xxxxxxxxxx
Chromium OS Project
bleung@xxxxxxxxxxxx

Attachment: signature.asc
Description: PGP signature