Qubes-vm-settings (qubesmanager) crashes Xorg when Advanced tab selected

I recently installed Qubes on a system with an ASUS Z490-V motherboard, using the integrated Intel UHD 630 graphics.

The system seems otherwise stable – I’m typing into a web browser it’s running right this moment – but every time I try to click on the “Advanced” tab – for any qube – in qubesmanager, the X server immediately segfaults.

Some of these segfaults were in the SNA driver, so I tried reconfiguring X to avoid it (notably, Option "AccelMethod" "uxa" isn’t supported at all with this chipset, so blt and Option "Accel" "off" were the only available changes to be made); however, it still happens even with acceleration turned off entirely.

Some examples of what this crash looks like with SNA enabled:

Feb 20 13:29:46 dom0 systemd-coredump[19333]: Process 17738 (Xorg) of user 0 dumped core.
                                              
                                              Stack trace of thread 17738:
                                              #0  0x00007af53186d8df raise (libc.so.6)
                                              #1  0x00007af53186f4da abort (libc.so.6)
                                              #2  0x00007af5318b11f0 __libc_message (libc.so.6)
                                              #3  0x00007af5318ba85a _int_free (libc.so.6)
                                              #4  0x00007af5318be28c __libc_free (libc.so.6)
                                              #5  0x00007af52d66656f sna_destroy_pixmap (intel_drv.so)
                                              #6  0x000000000051fa34 damageDestroyPixmap (Xorg)
                                              #7  0x00000000004d7fe5 XvDestroyPixmap (Xorg)
                                              #8  0x00000000004d6b3c ShmDestroyPixmap (Xorg)
                                              #9  0x000000000045bda2 doFreeResource (Xorg)
                                              #10 0x000000000045c968 FreeResource (Xorg)
                                              #11 0x00000000004325ee ProcFreePixmap (Xorg)
                                              #12 0x00000000004370f5 Dispatch (Xorg)
                                              #13 0x000000000043b078 dix_main (Xorg)
                                              #14 0x00007af531858431 __libc_start_main (libc.so.6)
                                              #15 0x0000000000424d5a _start (Xorg)

…or…

                                              Stack trace of thread 19356:
                                              #0  0x0000710c2d0c28df raise (libc.so.6)
                                              #1  0x0000710c2d0c44da abort (libc.so.6)
                                              #2  0x0000710c2d10cd38 __malloc_assert (libc.so.6)
                                              #3  0x0000710c2d111b7a _int_malloc (libc.so.6)
                                              #4  0x0000710c2d112ee4 malloc (libc.so.6)
                                              #5  0x0000710c2e4310cb _pixman_image_allocate (libpixman-1.so.0)
                                              #6  0x0000710c2e3f9910 create_bits_image_internal (libpixman-1.so.0)
                                              #7  0x0000710c28f91d4c image_from_pict_internal (intel_drv.so)
                                              #8  0x0000710c28ec6f0e sna_composite_fb (intel_drv.so)
                                              #9  0x0000710c28ec7647 sna_composite (intel_drv.so)
                                              #10 0x0000000000520ce6 damageComposite (Xorg)
                                              #11 0x0000000000516243 ProcRenderComposite (Xorg)
                                              #12 0x00000000004370f5 Dispatch (Xorg)
                                              #13 0x000000000043b078 dix_main (Xorg)
                                              #14 0x0000710c2d0ad431 __libc_start_main (libc.so.6)
                                              #15 0x0000000000424d5a _start (Xorg)

Are these settings reachable from the CLI? I’m trying to set up a Windows guest with an unused graphics card enabled with PCI passthrough; some of the requisite settings live in that tab.

Many thanks,
– Charles

I can’t really help here, but perhaps this link could be useful: Qubes gpu passthrough.

Thanks!

I found the settings I need (via qvm-prefs and qvm-features); any further passthrough issues I’ll be opening a completely separate thread for, so this one can be used strictly for the reproducible dom0 X server crash.

2 Likes

If nobody is able to provide an answer it may be better to report this as a formal bug so it catches the developer’s eyes.