The last 2 or 3 kernel updates seem to have triggered a regression in the i915 (internal Intel GPU) driver. I get damaged screen updates, very noticeable tearing on videos and occasional freezes in the DE (Plasma under X11). A look at the journal shows:
Feb 19 16:33:16 Bree kernel: i915 0000:00:02.0: [drm] GPU HANG: ecode 7:1:cbd2d2d2, in chrome [8728] Feb 19 16:33:16 Bree kernel: GPU hangs can indicate a bug anywhere in the entire gfx stack, including userspace. Feb 19 16:33:16 Bree kernel: The GPU crash dump is required to analyze GPU hangs, so please always attach it. Feb 19 16:33:16 Bree kernel: GPU crash dump saved to /sys/class/drm/card0/error Feb 19 16:33:16 Bree kernel: i915 0000:00:02.0: [drm] chrome[8728] context reset due to GPU hang Feb 19 16:33:22 Bree kernel: i915 0000:00:02.0: [drm] GPU HANG: ecode 7:1:85ffbff8, in chrome [9071] Feb 19 16:33:22 Bree kernel: i915 0000:00:02.0: [drm] chrome[9071] context reset due to GPU hang Feb 19 16:33:59 Bree kernel: i915 0000:00:02.0: [drm] GPU HANG: ecode 7:1:7f7f7f7f, in chrome [9160] Feb 19 16:33:59 Bree kernel: i915 0000:00:02.0: [drm] chrome[9160] context reset due to GPU hang Feb 19 23:43:39 Bree kernel: i915 0000:00:02.0: [drm] GPU HANG: ecode 7:1:85fffffc, in Xorg [1182] Feb 19 23:43:39 Bree kernel: i915 0000:00:02.0: [drm] Xorg[1182] context reset due to GPU hang Feb 20 11:23:50 Bree kernel: i915 0000:00:02.0: [drm] GPU HANG: ecode 7:1:85fffff8, in plasmashell [17364] Feb 20 11:23:50 Bree kernel: i915 0000:00:02.0: [drm] plasmashell[17364] context reset due to GPU hang Feb 20 11:23:53 Bree kernel: i915 0000:00:02.0: [drm] GPU HANG: ecode 7:1:85fffdfc, in plasmashell [17364] Feb 20 11:23:53 Bree kernel: i915 0000:00:02.0: [drm] plasmashell[17364] context reset due to GPU hang
Unfortunately I'd rebooted before looking at the journal, so /sys/class/drm/card0/error is currently empty. I'll capture it next time around, but it the meantime has anyone else seen this?
poc
On 20/02/2021 19:41, Patrick O'Callaghan wrote:
The last 2 or 3 kernel updates seem to have triggered a regression in the i915 (internal Intel GPU) driver. I get damaged screen updates, very noticeable tearing on videos and occasional freezes in the DE (Plasma under X11). A look at the journal shows:
Feb 19 16:33:16 Bree kernel: i915 0000:00:02.0: [drm] GPU HANG: ecode 7:1:cbd2d2d2, in chrome [8728] Feb 19 16:33:16 Bree kernel: GPU hangs can indicate a bug anywhere in the entire gfx stack, including userspace. Feb 19 16:33:16 Bree kernel: The GPU crash dump is required to analyze GPU hangs, so please always attach it.
Unfortunately I'd rebooted before looking at the journal, so /sys/class/drm/card0/error is currently empty. I'll capture it next time around, but it the meantime has anyone else seen this?
Don't have this HW. But a quick search suggests adding "nomodeset" to the boot parameters.
On Sat, 2021-02-20 at 20:47 +0800, Ed Greshko wrote:
On 20/02/2021 19:41, Patrick O'Callaghan wrote:
The last 2 or 3 kernel updates seem to have triggered a regression in the i915 (internal Intel GPU) driver. I get damaged screen updates, very noticeable tearing on videos and occasional freezes in the DE (Plasma under X11). A look at the journal shows:
Feb 19 16:33:16 Bree kernel: i915 0000:00:02.0: [drm] GPU HANG: ecode 7:1:cbd2d2d2, in chrome [8728] Feb 19 16:33:16 Bree kernel: GPU hangs can indicate a bug anywhere in the entire gfx stack, including userspace. Feb 19 16:33:16 Bree kernel: The GPU crash dump is required to analyze GPU hangs, so please always attach it.
Unfortunately I'd rebooted before looking at the journal, so /sys/class/drm/card0/error is currently empty. I'll capture it next time around, but it the meantime has anyone else seen this?
Don't have this HW. But a quick search suggests adding "nomodeset" to the boot parameters.
As I understand it, that could have a big impact on graphics performance. I'll try it as a last resort, but currently I'm keeping a working kernel (kernel-core-5.10.12-200.fc33.x86_64) and testing each new one as it appears.
I filed a BZ at https://bugzilla.redhat.com/show_bug.cgi?id=1931065 which I see now has some comments.
poc
The last 2 or 3 kernel updates seem to have triggered a regression in the i915 (internal Intel GPU) driver. I get damaged screen updates, very noticeable tearing on videos and occasional freezes in the DE (Plasma under X11). A look at the journal shows:
simular here on an Intel HD Graphics 530 (GT2) Gen 9 (Gnome, X11), mostly a immediate freeze during web browsing. CRTL+F3 and dmesg shows what's up. note: I'm mostly running lastest homebrewed vanilla kernels (with GPU firmware loaded), currently 5.11. so I guess you will see this with fedora kernel 5.11 too (if not somehow patched away), cause this GPU hangs have a history. IMO, they were completely away around 5.9 (?) and came back with 5.10.x (?) moons ago I tested with/without GPU FW on Fedora kernels: no differences
anyway, I'll test what @ED Greshko suggested.
maybe for further reading: https://cgit.freedesktop.org/drm-intel
On Sat, 2021-02-20 at 11:41 +0000, Patrick O'Callaghan wrote:
The last 2 or 3 kernel updates seem to have triggered a regression in the i915 (internal Intel GPU) driver. I get damaged screen updates, very noticeable tearing on videos and occasional freezes in the DE
I may have a similar problem. See my thread "Strange Artifacts in the Panel"; the problem seems to affect the whole KDE Plasma subsystem, but not other functions.
On Mon, 2021-02-22 at 08:28 -0800, Jonathan Ryshpan wrote:
On Sat, 2021-02-20 at 11:41 +0000, Patrick O'Callaghan wrote:
The last 2 or 3 kernel updates seem to have triggered a regression in the i915 (internal Intel GPU) driver. I get damaged screen updates, very noticeable tearing on videos and occasional freezes in the DE
I may have a similar problem. See my thread "Strange Artifacts in the Panel"; the problem seems to affect the whole KDE Plasma subsystem, but not other functions.
See the BZ report at https://bugzilla.redhat.com/show_bug.cgi?id=1931065 A modified kernel seems to be working so far.
poc
On Mon, 2021-02-22 at 16:55 +0000, Patrick O'Callaghan wrote:
On Mon, 2021-02-22 at 08:28 -0800, Jonathan Ryshpan wrote:
On Sat, 2021-02-20 at 11:41 +0000, Patrick O'Callaghan wrote:
The last 2 or 3 kernel updates seem to have triggered a regression in the i915 (internal Intel GPU) driver. I get damaged screen updates, very noticeable tearing on videos and occasional freezes in the DE
I may have a similar problem. See my thread "Strange Artifacts in the Panel"; the problem seems to affect the whole KDE Plasma subsystem, but not other functions.
See the BZ report at https://bugzilla.redhat.com/show_bug.cgi?id=1931065 A modified kernel seems to be working so far.
It works for me too. Now I have to decide whether to install my new CPU, scheduled to arrive this week, or return it.
On Tue, 2021-02-23 at 01:49 -0800, Jonathan Ryshpan wrote:
On Mon, 2021-02-22 at 16:55 +0000, Patrick O'Callaghan wrote:
On Mon, 2021-02-22 at 08:28 -0800, Jonathan Ryshpan wrote:
On Sat, 2021-02-20 at 11:41 +0000, Patrick O'Callaghan wrote:
The last 2 or 3 kernel updates seem to have triggered a regression in the i915 (internal Intel GPU) driver. I get damaged screen updates, very noticeable tearing on videos and occasional freezes in the DE
I may have a similar problem. See my thread "Strange Artifacts in the Panel"; the problem seems to affect the whole KDE Plasma subsystem, but not other functions.
See the BZ report at https://bugzilla.redhat.com/show_bug.cgi?id=1931065 A modified kernel seems to be working so far.
It works for me too. Now I have to decide whether to install my new CPU, scheduled to arrive this week, or return it.
Your call of course, but if this issue is the only reason for changing CPUs I would be inclined not to proceed. Assuming the fix survives further testing, I would expect it to become part of the mainline kernel fairly soon, hopefully within the window where you have to decide.
poc
Patrick O'Callaghan wrote:
The last 2 or 3 kernel updates seem to have triggered a regression in the i915 (internal Intel GPU) driver. I get damaged screen updates, very noticeable tearing on videos and occasional freezes in the DE (Plasma under X11).
Same here, I had to rollback to using kernel-5.9.16. Haven't tried any 5.10's since 5.10.13 or so (all up to there was pretty busted on my hardware).
-- Rex
-
Ed Greshko -
elder sixpack13 -
Jonathan Ryshpan -
Patrick O'Callaghan -
Rex Dieter