[Bug 1103426] New: Tumbleweed installation freezes when using QXL driver
http://bugzilla.opensuse.org/show_bug.cgi?id=1103426 Bug ID: 1103426 Summary: Tumbleweed installation freezes when using QXL driver Classification: openSUSE Product: openSUSE Tumbleweed Version: Current Hardware: x86-64 OS: Other Status: NEW Severity: Normal Priority: P5 - None Component: Virtualization:Tools Assignee: virt-bugs@suse.de Reporter: iforster@suse.com QA Contact: qa-bugs@suse.de CC: fabian@ritter-vogt.de, rbrown@suse.de Found By: --- Blocker: --- It happened several times to me that the installation of Tumbleweed just froze at various stages of the installation when using virt-manager together with the default QXL driver. Symptoms: The UI will just freeze completely. Switching to other terminals is not possible either. However when starting an SSH daemon it's still possible to connect to the machine. When doing so no relevant output can be seen in dmesg. When attaching to the Xorg process via strace nothing gets printed at all, getting the stack trace (which won't change) of X shows the following: 1:install:~ # cat /proc/`pidof Xorg`/task/*/stack [<0>] ttm_eu_reserve_buffers+0x1d9/0x2c0 [ttm] [<0>] qxl_release_reserve_list+0x53/0x130 [qxl] [<0>] qxl_cursor_atomic_update+0x17e/0x2a0 [qxl] [<0>] drm_atomic_helper_commit_planes+0xae/0x210 [drm_kms_helper] [<0>] drm_atomic_helper_commit_tail+0x26/0x60 [drm_kms_helper] [<0>] commit_tail+0x58/0x70 [drm_kms_helper] [<0>] drm_atomic_helper_commit+0x103/0x110 [drm_kms_helper] [<0>] drm_atomic_helper_update_plane+0xf0/0x110 [drm_kms_helper] [<0>] __setplane_internal+0x134/0x260 [drm] [<0>] drm_mode_cursor_universal+0xdf/0x1c0 [drm] [<0>] drm_mode_cursor_common+0x167/0x1d0 [drm] [<0>] drm_ioctl_kernel+0x5b/0xb0 [drm] [<0>] drm_ioctl+0x194/0x350 [drm] [<0>] do_vfs_ioctl+0x90/0x5d0 [<0>] ksys_ioctl+0x60/0x90 [<0>] __x64_sys_ioctl+0x16/0x20 [<0>] do_syscall_64+0x60/0x110 [<0>] entry_SYSCALL_64_after_hwframe+0x49/0xbe [<0>] 0xffffffffffffffff -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1103426 http://bugzilla.opensuse.org/show_bug.cgi?id=1103426#c1 --- Comment #1 from Ignaz Forster <iforster@suse.com> --- I forgot: This specific backtrace happened with kernel 4.17.7-1-default. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1103426 http://bugzilla.opensuse.org/show_bug.cgi?id=1103426#c5 Neil Rickert <nwr10cst-oslnx@yahoo.com> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |nwr10cst-oslnx@yahoo.com --- Comment #5 from Neil Rickert <nwr10cst-oslnx@yahoo.com> --- *** Bug 1106494 has been marked as a duplicate of this bug. *** -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1103426 http://bugzilla.opensuse.org/show_bug.cgi?id=1103426#c6 Richard Brown <rbrown@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |brogers@suse.com, | |rbrown@suse.com Flags| |needinfo?(brogers@suse.com) --- Comment #6 from Richard Brown <rbrown@suse.com> --- Why don't we just make virtio the default instead of QXL? -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1103426 http://bugzilla.opensuse.org/show_bug.cgi?id=1103426#c8 --- Comment #8 from Neil Rickert <nwr10cst-oslnx@yahoo.com> ---
but we would still need to solve this bug.
I agree with that. Since this problem only occurs intermittently, it is hard to test. I would like some guidance on what sort of information I should collect (and attach) when I next run into the problem. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1103426 http://bugzilla.opensuse.org/show_bug.cgi?id=1103426#c9 Chris Ward <tjcw@physics.org> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |tjcw@physics.org --- Comment #9 from Chris Ward <tjcw@physics.org> --- If you use boot parameter 'textmode=1' for the install, you can perform a text mode install. For me this completed normally and the resulting tumbleweed virtual system seems to be running normally (under leap 15.0). -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1103426 http://bugzilla.opensuse.org/show_bug.cgi?id=1103426#c11 --- Comment #11 from Neil Rickert <nwr10cst-oslnx@yahoo.com> --- Responding to c#10
Can you test with latest TW which runs 4.18.7-1.5 ?
The difficulty is that these freezes are intermittent, so they are hard to test. My most recent freeze was on Saturday or Sunday. I guess I should keep better records. However, I was running a 4.18 kernel at that time. Can you suggest what information I should collect when there is a freeze. Keep in mind that I am able to login with ssh and get a root command line. A little guidance on what to look for would be helpful here. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1103426 http://bugzilla.opensuse.org/show_bug.cgi?id=1103426#c12 --- Comment #12 from Chris Ward <tjcw@physics.org> --- The installation kernel from Tumbleweed 20180917 is 4.18.5-1 . Does this have the fix ? -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1103426 http://bugzilla.opensuse.org/show_bug.cgi?id=1103426#c29 --- Comment #29 from Chris Ward <tjcw@cantab.net> --- On further testing, the VM installation still hangs even with the modified iso. So it's still for consideration, what the difference is between an installed system and the installation system. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1103426 http://bugzilla.opensuse.org/show_bug.cgi?id=1103426#c40 --- Comment #40 from Neil Rickert <nwr10cst-oslnx@yahoo.com> --- Not much activity on this bug lately. Today, I tried an install of Ubuntu-19.04. And the install froze part way through. I then reconfigured the install to use "virtio" instead of "QXL". This time the install went through without problems. For the original install, the visible display froze about halfway through the "copying" phase. However, it seems that the actual install went well beyond that, and possibly even completed. When I did a forced stop and reboot, the installed machine actually booted. So the install got past where grub is install and got past where the user is added. I did not try to login, so I don't know how far it got beyond the login screen. I stopped at that point, and rebooted to reinstall. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1103426 http://bugzilla.opensuse.org/show_bug.cgi?id=1103426#c41 --- Comment #41 from Neil Rickert <nwr10cst-oslnx@yahoo.com> --- This problem seems to be solved in Leap 15.1. I have done several installs, using the default QXL driver. And I have not seen any freezes. -- You are receiving this mail because: You are on the CC list for the bug.
participants (1)
-
bugzilla_noreply＠novell.com