Skip to content

i915 0000:00:02.0: GPU HANG: ecode 9:1:0x00000000, hang on rcs0

It's the 3rd time in two weeks that my display completly becomes frozen (remain the same, stops to be updated), but the laptop continues to work (webcam, radio playing, SSH works, etc.) :-) This issue is focused on the bug that I had today (January 28).

  • January 22 at 18:45:51

    • "janv. 22 18:45:55 apu kernel: i915 0000:00:02.0: GPU HANG: ecode 9:1:0x00000000, hang on rcs0"
    • Attached card0_error_jan22
    • "Kernel: 5.4.10-200.fc31.x86_64 x86_64", "Suspend count: 0"

  • January 23 at 16:32:22

    • "janv. 23 16:32:22 apu kernel: i915 0000:00:02.0: GPU HANG: ecode 9:1:0x84332723, in Xwayland [1685], hang on rcs0"
    • Attached card0_error_jan23
    • The error file contains a full "rcs0 command stream:"
    • "i915.mmio_debug=1"
    • "Active process (on ring rcs0): Xwayland [1685]"
    • "Kernel: 5.4.12-200.fc31.x86_64 x86_64", "Suspend count: 2"

  • January 28 at 15:29:42

    • "janv. 28 15:29:42 apu kernel: i915 0000:00:02.0: GPU HANG: ecode 9:1:0x00000000, hang on rcs0"
    • Attached card0_error_jan28
    • "Kernel: 5.4.13-201.fc31.x86_64 x86_64", "Suspend count: 11"

Note: the 3 error files contain "Reset count: 0".

I have a Lenovo P50 with Hybrid Graphics, Intel Corporation HD Graphics 530 (rev 06) + NVIDIA Corporation GM107GLM [Quadro M1000M] (rev a2). Also known as: Intel IGP (Intel HD Graphics 530) + NVIDIA GPU (NVIDIA Quadro M1000M).

The internal screen of the laptop is off (laptop closed), I have 2 external monitor connected by DisplayPort (2560x1440 pixels) and HDMI (1920x1080 pixels). The NVIDIA GPU is powered (/sys/bus/pci/drivers/nouveau/0000:01:00.0/power/runtime_status is "active"), but I don't know if it's used.

I'm using GNOME with Wayland on an up to date Fedora 31, Xwayland is running as well.

I even wrote an article Hybrid Graphics of this laptop while debugging graphical issues in Firefox :-) https://vstinner.github.io/debug-hybrid-graphics-issues-linux.html

/sys/class/drm/card0/error content attached as card0_error file.

journalctl logs since the laptop boot (yesterday, I rebooted after a kernel upgrade):

-- Logs begin at Wed 2019-09-25 23:09:57 CEST, end at Tue 2020-01-28 16:01:24 CET. --
janv. 27 11:06:27 apu kernel: microcode: microcode updated early to revision 0xd6, date = 2019-10-03
janv. 27 11:06:27 apu kernel: Linux version 5.4.13-201.fc31.x86_64 (mockbuild@bkernel03.phx2.fedoraproject.org) (gcc version 9.2.1 20190827 (Red Hat 9.2.1-1) (GCC)) #1 SMP Tue Jan 21 17:21:47 UTC 2020
janv. 27 11:06:27 apu kernel: Command line: BOOT_IMAGE=(hd0,msdos1)/vmlinuz-5.4.13-201.fc31.x86_64 root=UUID=19d7b043-9a99-4e23-9ef9-e4b6ace2c4df ro rootflags=subvol=root rhgb quiet xdg.force_integrated=0

(...)

janv. 28 15:29:34 apu gnome-shell[2084]: [101893:0128/152934:INFO:shell_content_browser_plugin.cc(53)] Allowing plugin: bjnplugin : /home/vstinner/.config/Blue Jeans/desktop/>

janv. 28 15:29:42 apu kernel:

    i915 0000:00:02.0: GPU HANG: ecode 9:1:0x00000000, hang on rcs0
    GPU hangs can indicate a bug anywhere in the entire gfx stack, including userspace.
    Please file a _new_ bug report on bugs.freedesktop.org against DRI -> DRM/Intel
    drm/i915 developers can then reassign to the right component if it's not a kernel issue.
    The GPU[card0_error](/uploads/d8aea4ed47bc180de407e4f5ed668f3b/card0_error) crash dump is required to analyze GPU hangs, so please always attach it.
    GPU crash dump saved to /sys/class/drm/card0/error
    i915 0000:00:02.0: Resetting rcs0 for hang on rcs0
    [drm:gen8_reset_engines [i915]] *ERROR* rcs0 reset request timed out: {request: 00000001, RESET_CTL: 00000001}
    i915 0000:00:02.0: Resetting chip for hang on rcs0
    [drm:gen8_reset_engines [i915]] *ERROR* rcs0 reset request timed out: {request: 00000001, RESET_CTL: 00000001}
    [drm:gen8_reset_engines [i915]] *ERROR* rcs0 reset request timed out: {request: 00000001, RESET_CTL: 00000001}

janv. 28 15:29:42 apu gnome-shell[2084]: [0128/152942:INFO:shell_main_delegate.cc(178)] Trying to add Blue Jeans Resources to flash players trust directory
janv. 28 15:29:42 apu gnome-shell[2084]: [0128/152942:INFO:shell_main_delegate.cc(181)] The app data path is /home/vstinner/.config
janv. 28 15:29:42 apu gnome-shell[2084]: [0128/152942:INFO:shell_main_delegate.cc(221)] bluejeans.flashtrust.cfg already exists in ./home/vstinner/.macromedia/Flash_Player/#S>
janv. 28 15:29:42 apu gnome-shell[2084]: [101893:0128/152942:INFO:shell_content_browser_plugin.cc(53)] Allowing plugin: bjnplugin : /home/vstinner/.config/Blue Jeans/desktop/>
janv. 28 15:29:42 apu gnome-shell[2084]: [101893:0128/152942:INFO:shell_content_browser_plugin.cc(53)] Allowing plugin: bjnplugin : /home/vstinner/.config/Blue Jeans/desktop/>
janv. 28 15:35:44 apu audit[107003]: CRYPTO_KEY_USER pid=107003 uid=0 auid=4294967295 ses=4294967295 subj=system_u:system_r:sshd_t:s0-s0:c0.c1023 msg='op=destroy kind=server >
janv. 28 15:35:44 apu audit[107002]: CRYPTO_SESSION pid=107002 uid=0 auid=4294967295 ses=4294967295 subj=system_u:system_r:sshd_t:s0-s0:c0.c1023 msg='op=start direction=from->

If I recall correctly, I was using the following applications:

  • Bluejeans client (bluejeans-1.37.22-1.x86_64)
  • Firefox (firefox-72.0.1-2.fc31.x86_64)
  • Gnome Terminal (gnome-terminal-3.34.2-1.fc31.x86_64)

I was typing in Gnome Terminal or Firefox, I don't recall. Maybe I used CTRL+ALT to switch between two applications.

Versions:

  • uname -r: 5.4.13-201.fc31.x86_64 (kernel-5.4.13-201.fc31.x86_64)
  • libwayland-client-1.17.0-2.fc31.x86_64
  • libdrm-2.4.100-1.fc31.x86_64
  • xorg-x11-drv-intel-2.99.917-43.20180618.fc31.x86_64
  • xorg-x11-server-Xwayland-1.20.6-1.fc31.x86_64

Sometimes, the GPU hangs for 5 seconds, but restart after a hardware reset (logged hopefully). Sadly, sometimes, it hangs forever.

I don't know if it's related, but at January 22, I got another kernel bug which might come from the Intel driver as well:

A few similar i915 ecode 9 errors:

See also: https://fedoraproject.org/wiki/How_to_debug_Wayland_problems

To upload designs, you'll need to enable LFS and have an admin enable hashed storage. More information