[Nouveau] Kernel crash in 3.17-rc1 when loading nouveau on (non-POSTed) NV1A

[Bruno Prémont]

On Sun, 24 August 2014 Bruno Prémont <bonbons at linux-vserver.org> wrote:
> System was booted with PCI graphics card first in VGA text mode.
> 
> Loading nouveau from there on causes the following BUG, after what
> kernel produces trace over trace until overflowing its stack.
> (trace captured via netconsole)
> 
> [  154.323717] wmi: Mapper loaded
> [  154.735793] nouveau 0000:02:00.0: enabling device (0004 -> 0006)
> [  154.743189] ACPI: PCI Interrupt Link [LNK5] enabled at IRQ 16
> [  154.754844] nouveau  [  DEVICE][0000:02:00.0] BOOT0  : 0x01a000b1
> [  154.761111] nouveau  [  DEVICE][0000:02:00.0] Chipset: nForce (NV1A)
> [  154.767534] nouveau  [  DEVICE][0000:02:00.0] Family : NV10
> [  154.773918] nouveau  [   VBIOS][0000:02:00.0] checking PRAMIN for image...
> [  154.832093] nouveau  [   VBIOS][0000:02:00.0] ... appears to be valid
> [  154.838624] nouveau  [   VBIOS][0000:02:00.0] using image from PRAMIN
> [  154.845146] nouveau  [   VBIOS][0000:02:00.0] BMP version 5.14
> [  154.851300] nouveau  [   VBIOS][0000:02:00.0] version 03.1a.01.03.00
> [  154.857785] nouveau W[   VBIOS][0000:02:00.0] DCB contains no useful data
> [  154.866214] nouveau W[   VBIOS][0000:02:00.0] DCB contains no useful data
> [  154.874456] nouveau W[   VBIOS][0000:02:00.0] DCB contains no useful data
> [  154.881386] nouveau W[   VBIOS][0000:02:00.0] DCB contains no useful data
> [  154.888274] nouveau  [ DEVINIT][0000:02:00.0] adaptor not initialised
> [  154.894789] nouveau  [   VBIOS][0000:02:00.0] running init tables
> [  155.060171] nouveau W[  PTIMER][0000:02:00.0] unknown input clock freq
> [  155.066831] nouveau  [     PFB][0000:02:00.0] RAM type: stolen system memory
> [  155.073960] nouveau  [     PFB][0000:02:00.0] RAM size: 32 MiB
> [  155.079857] nouveau  [     PFB][0000:02:00.0]    ZCOMP: 0 tags
> [  155.090902] nouveau  [     CLK][0000:02:00.0] --:   
> [  155.096002] ------------[ cut here ]------------
> [  155.100004] kernel BUG at /usr/src/linux-git/drivers/gpu/drm/nouveau/core/core/event.c:42!
> 
> This is a BUG_ON(!spin_is_locked(&event->refs_lock))
> 
> Is that a valid check for CONFIG_SMP=n?
> As far as I know spin_locks are NOOP on UP configs... and in the recent
> past that kind of test has been complained about on lkml.

It was maybe even on dri-devel I saw it:
  https://lkml.org/lkml/2014/8/11/4

> [  155.100004] invalid opcode: 0000 [#1] 
> [  155.100004] Modules linked in: nouveau(+) wmi ttm drm_kms_helper nfsv3 nfs_acl nfs lockd sunrpc
> [  155.100004] CPU: 0 PID: 15 Comm: kworker/0:1 Not tainted 3.17.0-rc1-jupiter-00002-gec30df4 #6
> [  155.100004] Hardware name: NVIDIA Corporation. nFORCE-MCP/MS-6373, BIOS 6.00 PG 04/12/2002
> [  155.100004] Workqueue: events nouveau_pstate_work [nouveau]
> [  155.100004] task: dd451c70 ti: dd5c8000 task.ti: dd5c8000
> [  155.100004] EIP: 0060:[<dea19e13>] EFLAGS: 00010046 CPU: 0
> [  155.100004] EIP is at nvkm_event_get+0x3/0x10 [nouveau]
> [  155.100004] EAX: dcd4c484 EBX: 00000286 ECX: 00000000 EDX: 00000001
> [  155.100004] ESI: 00000000 EDI: ffffffff EBP: dd5c9ea8 ESP: dd5c9ea8
> [  155.100004]  DS: 007b ES: 007b FS: 0000 GS: 00e0 SS: 0068
> [  155.100004] CR0: 8005003b CR2: b77d9000 CR3: 1ce83000 CR4: 000007d0
> [  155.100004] Stack:
> [  155.100004]  dd5c9eb4 dea1cc92 dd694d34 dd5c9f04 dea2e3b6 00000000 00000005 deae3e69
> [  155.100004]  ffffffff 00000001 ffffffff ffffffff ffffffff 00000000 00000000 dd451c70
> [  155.100004]  c05bc820 ffffffff 00000000 dd694c60 dd694d34 00000000 dd5b8070 dd5c9f44
> [  155.100004] Call Trace:
> [  155.100004]  [<dea1cc92>] nvkm_notify_get+0x32/0x40 [nouveau]
> [  155.100004]  [<dea2e3b6>] nouveau_pstate_work+0x396/0x3a0 [nouveau]
> [  155.100004]  [<c1047fa7>] process_one_work+0x1d7/0x340
> [  155.100004]  [<c10485bf>] worker_thread+0x2af/0x380
> [  155.100004]  [<c1048310>] ? rescuer_thread+0x1d0/0x1d0
> [  155.100004]  [<c1048310>] ? rescuer_thread+0x1d0/0x1d0
> [  155.100004]  [<c104bfa4>] kthread+0xa4/0xb0
> [  155.100004]  [<c14b6c00>] ret_from_kernel_thread+0x20/0x30
> [  155.100004]  [<c104bf00>] ? flush_kthread_worker+0x70/0x70
> [  155.100004] Code: 71 a9 e2 90 8d 74 26 00 83 c4 0c 5b 5e 5f 5d c3 66 90 66 90 66 90 66 90 55 89 e5 0f 0b 8d 74 26 00 8d bc 27 00 00 00 00 55 89 e5 <0f> 0b 8d 74 
> [  155.100004] EIP: [<dea19e13>] nvkm_event_get+0x3/0x10 [nouveau] SS:ESP 0068:dd5c9ea8
> [  155.100004] ---[ end trace 6142147b1d3fed4d ]---
> 
> things go downhill from here on.