Description of problem: Attempting to boot F34 Workstation on an Nvidia Jetson Nano. Version-Release number of selected component: gnome-shell-40.0~alpha.1.1-8.20210212git829a096ba.fc34 Additional info: reporter: libreport-2.14.0 backtrace_rating: 4 cgroup: 0::/user.slice/user-984.slice/session-c1.scope cmdline: /usr/bin/gnome-shell crash_function: nouveau_fence_signalled executable: /usr/bin/gnome-shell journald_cursor: s=d55a9ac4f3e740739bf362a2b77352f1;i=812;b=ded02adf973940a6b627881eed971702;m=46f6ccf;t=5bbb6e1795154;x=9bcd970d5d0e2ae6 kernel: 5.11.0-156.fc34.aarch64 rootdir: / runlevel: N 5 type: CCpp uid: 984 Truncated backtrace: Thread no. 1 (10 frames) #0 nouveau_fence_signalled at ../src/gallium/drivers/nouveau/nouveau_fence.c:170 #1 nvc0_draw_indirect at ../src/gallium/drivers/nouveau/nvc0/nvc0_vbo.c:819 #2 nvc0_draw_vbo at ../src/gallium/drivers/nouveau/nvc0/nvc0_vbo.c:1116 #3 tegra_draw_vbo at ../src/gallium/drivers/tegra/tegra_context.c:87 #4 _mesa_validated_drawrangeelements at ../src/mesa/main/draw.c:1118 #6 _mesa_DrawElements at ../src/mesa/main/draw.c:1272 #7 cogl_gl_framebuffer_draw_indexed_attributes at ../cogl/cogl/driver/gl/cogl-framebuffer-gl.c:401 #8 cogl_framebuffer_driver_draw_indexed_attributes at ../cogl/cogl/cogl-framebuffer-driver.c:137 #9 _cogl_framebuffer_draw_indexed_attributes at ../cogl/cogl/cogl-framebuffer.c:2517 #10 _cogl_journal_flush_modelview_and_entries at ../cogl/cogl/cogl-journal.c:311
Created attachment 1758285 [details] File: backtrace
Created attachment 1758286 [details] File: core_backtrace
Created attachment 1758287 [details] File: cpuinfo
Created attachment 1758288 [details] File: dso_list
Created attachment 1758289 [details] File: environ
Created attachment 1758290 [details] File: exploitable
Created attachment 1758291 [details] File: limits
Created attachment 1758292 [details] File: maps
Created attachment 1758293 [details] File: mountinfo
Created attachment 1758294 [details] File: open_fds
Created attachment 1758295 [details] File: proc_pid_status
Created attachment 1758296 [details] File: var_log_messages
Proposing as a beta blocker for F34, this affects a supported hardware platform (Jetson Nano).
After consulting #gnome-shell, the suggestion is to move this to mesa. "cogl should not be able to trigger crashes like that"
Discussed during the 2021-02-22 blocker review meeting: [0] The decision to classify this bug as an "AcceptedBlocker (Beta)" was made as it violates the following Basic criterion: "A system installed with a release-blocking desktop must boot to a log in screen where it is possible to log in to a working desktop using a user account created during installation or a 'first boot' utility" for a supported aarch64 platform. [0] https://meetbot.fedoraproject.org/fedora-blocker-review/2021-02-22/f34-blocker-review.2021-02-22-17.07.txt
What version of mesa is used here? I was under the impression we already fixed it upstream.
(In reply to Karol Herbst from comment #16) > What version of mesa is used here? I was under the impression we already > fixed it upstream. Looking at the bug report time against the builds I'm guessing it's mesa-21.0.0~rc4, we got rc5 on 22nd Feb.
*** Bug 1930978 has been marked as a duplicate of this bug. ***
No change with mesa-21.0.0~rc5-2.fc34
yeah, already on it. Just had to update my jetson nano setup. Anyway, I was able to reproduce the issue.
So while I already have a patch to fix the issue reported here, sadly I am hitting a different regression a bit later :/ Still debugging this one and will send out patches once I am done fixing it.
patches posted upstream: https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9425 And they should added to 21.0 automatically once they land. Question is if we want to backport those or just wait for the next mesa release?
We need to backport them, this is a release blocker for Beta which is due soon. We can't wait for a new release, and we can't really land an entire release through freeze just to fix this bug anyway. If you could please backport exactly and only the fix for this and do a new package build, that would be great. Thanks.
(In reply to Adam Williamson from comment #23) > We need to backport them, this is a release blocker for Beta which is due > soon. We can't wait for a new release, and we can't really land an entire > release through freeze just to fix this bug anyway. If you could please > backport exactly and only the fix for this and do a new package build, that > would be great. Thanks. maybe I wasn't clear on what I meant. I meant backporting in Fedora instead of waiting for mesa-21.0. For me the question was rather if I should simply make sure the fixes get added to mesa-21.0 or if I also have to change the packaging and add the patch there.
Yes, I'm talking about backporting for Fedora. I'm saying we need to do it :) Fedora 34 Beta is due to be signed off Thursday and has been frozen for some time now. This bug is blocking that release, but by policy we should not take an entire new release of mesa to fix it. The policy is that the specific fixes necessary to fix this bug should be backported to the Fedora 34 package.
(In reply to Adam Williamson from comment #25) > Yes, I'm talking about backporting for Fedora. I'm saying we need to do it > :) Fedora 34 Beta is due to be signed off Thursday and has been frozen for > some time now. This bug is blocking that release, but by policy we should > not take an entire new release of mesa to fix it. The policy is that the > specific fixes necessary to fix this bug should be backported to the Fedora > 34 package. oh, wasn't aware that we are this close. Yeah, makes sense then. I can open a PR on Pagure or is there a better way to include single patches? Patch upstream is located here btw: https://gitlab.freedesktop.org/mesa/mesa/-/commit/a84c8ddb19beef7580d6a4b8c417bb188517412a
Thanks. A PR on Pagure would work fine if you don't have commit privileges to the package, yep, or ajax or I can just do the package build, I'm a provenpackager so I can do that.
OK, I have a build running with both commits from the MR backported: https://koji.fedoraproject.org/koji/taskinfo?taskID=63368144
FEDORA-2021-255eb6ff52 has been submitted as an update to Fedora 34. https://bodhi.fedoraproject.org/updates/FEDORA-2021-255eb6ff52
mesa-21.0.0~rc5-3.fc34 fixes the crash reported in this bug. On reboot the onscreen pointer is not visible and once the mouse is moved: Mar 09 09:13:27 nano gnome-shell[1159]: Failed to post KMS update: drmModeAtomicCommit: Invalid argument Mar 09 09:13:27 nano gnome-shell[1159]: Page flip discarded: drmModeAtomicCommit: Invalid argument Mar 09 09:25:24 nano gnome-session-f[1626]: Negative content width -7 (allocation 1, extents 4x4) while allocating gadget (node headerbar, owner GtkHeaderBar) Mar 09 09:25:24 nano gnome-session-f[1626]: gtk_widget_size_allocate(): attempt to allocate widget with width -34 and height 18 Mar 09 09:25:24 nano gnome-session-f[1626]: Negative content width -23 (allocation 1, extents 12x12) while allocating gadget (node label, owner GtkLabel) Mar 09 09:25:24 nano gnome-session[1626]: *** BUG *** Mar 09 09:25:24 nano gnome-session[1626]: In pixman_region32_init_rect: Invalid rectangle passed Mar 09 09:25:24 nano gnome-session[1626]: Set a breakpoint on '_pixman_log_error' to debug Then a black screen with flashing cursor.
Created attachment 1762008 [details] journalctl with mesa-21.0.0~rc5-3.fc34
(In reply to Paul Whalen from comment #30) > mesa-21.0.0~rc5-3.fc34 fixes the crash reported in this bug. > > On reboot the onscreen pointer is not visible and once the mouse is moved: > > Mar 09 09:13:27 nano gnome-shell[1159]: Failed to post KMS update: > drmModeAtomicCommit: Invalid argument > Mar 09 09:13:27 nano gnome-shell[1159]: Page flip discarded: > drmModeAtomicCommit: Invalid argument > > Mar 09 09:25:24 nano gnome-session-f[1626]: Negative content width -7 > (allocation 1, extents 4x4) while allocating gadget (node headerbar, owner > GtkHeaderBar) > Mar 09 09:25:24 nano gnome-session-f[1626]: gtk_widget_size_allocate(): > attempt to allocate widget with width -34 and height 18 > Mar 09 09:25:24 nano gnome-session-f[1626]: Negative content width -23 > (allocation 1, extents 12x12) while allocating gadget (node label, owner > GtkLabel) > Mar 09 09:25:24 nano gnome-session[1626]: *** BUG *** > Mar 09 09:25:24 nano gnome-session[1626]: In pixman_region32_init_rect: > Invalid rectangle passed > Mar 09 09:25:24 nano gnome-session[1626]: Set a breakpoint on > '_pixman_log_error' to debug > > Then a black screen with flashing cursor. glad to hear the crash is fixed! But I think it would be best to file a new bug for that and from the looks of it it's not really clear if that's a tegradrm/kernel bug or a bug inside gnome. In any case might make sense to file it against gnome-shell or something.
(In reply to Karol Herbst from comment #32) > > glad to hear the crash is fixed! > > But I think it would be best to file a new bug for that and from the looks > of it it's not really clear if that's a tegradrm/kernel bug or a bug inside > gnome. In any case might make sense to file it against gnome-shell or > something. Will do, thank you for working on this and the fix!
Opened BZ#1936991
FEDORA-2021-255eb6ff52 has been pushed to the Fedora 34 testing repository. Soon you'll be able to install the update with the following command: `sudo dnf upgrade --enablerepo=updates-testing --advisory=FEDORA-2021-255eb6ff52` You can provide feedback for this update here: https://bodhi.fedoraproject.org/updates/FEDORA-2021-255eb6ff52 See also https://fedoraproject.org/wiki/QA:Updates_Testing for more information on how to test updates.
FEDORA-2021-255eb6ff52 has been pushed to the Fedora 34 stable repository. If problem still persists, please make note of it in this bug report.