On Tue, 2 Apr 2024 at 14:20, Paolo Bonzini <pbonz...@redhat.com> wrote: > > The following changes since commit b9dbf6f9bf533564f6a4277d03906fcd32bb0245: > > Merge tag 'pull-tcg-20240329' of https://gitlab.com/rth7680/qemu into > staging (2024-03-30 14:54:57 +0000) > > are available in the Git repository at: > > https://gitlab.com/bonzini/qemu.git tags/for-upstream > > for you to fetch changes up to eac4af186f6db46fc90ec571a855bd6fa4cb7841: > > pc_q35: remove unnecessary m->alias assignment (2024-04-02 15:14:02 +0200) > > ---------------------------------------------------------------- > * lsi53c895a: fix assertion failure with invalid Block Move > * vga: fix assertion failure with 4- and 16-color modes > * remove unnecessary assignment > > ---------------------------------------------------------------- > Paolo Bonzini (7): > vga: merge conditionals on shift control register > vga: move computation of dirty memory region later > vga: adjust dirty memory region if pel panning is active > vga: do not treat horiz pel panning value of 8 as "enabled" > lsi53c895a: avoid out of bounds access to s->msg[] > lsi53c895a: detect invalid Block Move instruction > pc_q35: remove unnecessary m->alias assignment
This seems to break the avocado test tests/avocado/ppc_prep_40p.py:IbmPrep40pMachine.test_openbios_and_netbsd and it's consistent even with retrying the job: https://gitlab.com/qemu-project/qemu/-/jobs/6529626987 https://gitlab.com/qemu-project/qemu/-/jobs/6528696711 https://gitlab.com/qemu-project/qemu/-/jobs/6529196532 The debug log says: 14:23:32 DEBUG| Transitioning from 'Runstate.CONNECTING' to 'Runstate.RUNNING'. 14:23:32 DEBUG| Opening console file 14:23:32 DEBUG| Opening console socket 14:23:32 DEBUG| >> ============================================================= 14:23:32 DEBUG| >> OpenBIOS 1.1 [Mar 7 2023 22:21] 14:23:32 DEBUG| >> Configuration device id QEMU version 1 machine id 0 14:23:32 DEBUG| >> CPUs: 0 14:23:32 DEBUG| >> Memory: 128M 14:23:32 DEBUG| >> UUID: 00000000-0000-0000-0000-000000000000 14:23:32 DEBUG| >> CPU type PowerPC,604 14:23:32 DEBUG| milliseconds isn't unique. 14:23:32 DEBUG| Output device screen not found. 14:23:32 DEBUG| Output device screen not found. 14:23:32 DEBUG| Trying cd:,\\:tbxi... 14:23:32 DEBUG| Trying cd:,\ppc\bootinfo.txt... 14:23:32 DEBUG| Trying cd:,%BOOT... 14:23:32 DEBUG| No valid state has been set by load or init-program and then the test times out because it never sees the NetBSD console output it's waiting for. Successful job for a previous pullreq: https://gitlab.com/qemu-project/qemu/-/jobs/6527774374 Here the debug log says: 12:36:14 DEBUG| >> ============================================================= 12:36:14 DEBUG| >> OpenBIOS 1.1 [Mar 7 2023 22:21] 12:36:14 DEBUG| >> Configuration device id QEMU version 1 machine id 0 12:36:14 DEBUG| >> CPUs: 0 12:36:14 DEBUG| >> Memory: 128M 12:36:14 DEBUG| >> UUID: 00000000-0000-0000-0000-000000000000 12:36:14 DEBUG| >> CPU type PowerPC,604 12:36:14 DEBUG| milliseconds isn't unique. 12:36:14 DEBUG| Output device screen not found. 12:36:14 DEBUG| Output device screen not found. 12:36:14 DEBUG| Trying cd:,\\:tbxi... 12:36:14 DEBUG| >> Not a bootable ELF image 12:36:15 DEBUG| >> switching to new context: 12:36:15 DEBUG| >> NetBSD/prep BOOT, Revision 1.9 12:36:15 DEBUG| Shutting down VM appliance; timeout=30 12:36:15 DEBUG| Attempting graceful termination 12:36:15 DEBUG| Closing console file 12:36:15 DEBUG| Closing console socket 12:36:15 DEBUG| Politely asking QEMU to terminate This machine uses the lsi53c810 SCSI controller, and it's failing to load from the CDROM, so my guess is the problem is in one of the two SCSI patches. thanks -- PMM