[Cbe-oss-dev] [PATCH 6:7] spufs: implement deferred queuing for spu class 1 faults

[Kazunori Asayama]

Luke Browning wrote:
> Implement deferred queuing for spu class 1 faults

The problem still exists.

I tried your patches on both PS3 and QS20 as following:

  - PS3:  latest public ps3-linux tree + your patches
  - QS20: latest public spufs tree + your patches

then I got messages below from the kernel (sometimes silently hang).

However, I couldn't apply your patches to the public trees cleanly, 
so there might be significant difference between your internal tree 
and the public trees. At least, it looks like your internal tree 
doesn't contain the change in 
http://ozlabs.org/pipermail/cbe-oss-dev/2008-April/004661.html
which is included in the public trees.


Here are messages from the kernel:

  - PS3

    # make check-test_dma_page_fault.elf TEST_SPES=7

Badness at kernel/mutex.c:353
NIP: c00000000032c028 LR: c00000000032c00c CTR: 0000000000000000
REGS: c000000006cef7d0 TRAP: 0700   Not tainted  (2.6.26-rc2-21786-g565207b-dirty)
MSR: 8000000000028032 <EE,IR,DR>  CR: 24002422  XER: 20000000
TASK = c000000006303220[1818] 'test_dma_page_f' THREAD: c000000006cec000 CPU: 1
GPR00: 0000000000000000 c000000006cefa50 c000000000521850 0000000000000001 
GPR04: 2222222222222222 0000000000000000 c000000006cefaa8 0000000000000000 
GPR08: 0000000000000000 c000000000586ee0 0000000000000000 0000000000000001 
GPR12: 8000000000000032 c000000000551500 0000000010010380 0000000010004298 
GPR16: 0000000010004298 00000000f7fe98b0 00000000f7fe98b0 0000000000000008 
GPR20: 00000000f5645380 0000000000000000 000000000ff9ee38 00000000f564537c 
GPR24: c00000000669e000 0000000000000000 c00000000666d000 c00000000669e000 
GPR28: 0000000000000001 c0000000005522d8 c0000000004da8e0 c00000000669e8c0 
NIP [c00000000032c028] .mutex_trylock+0xdc/0x198
LR [c00000000032c00c] .mutex_trylock+0xc0/0x198
Call Trace:
[c000000006cefa50] [c00000000032c9ac] .mutex_lock+0x28/0x48 (unreliable)
[c000000006cefae0] [c000000000032d9c] .spu_activate+0x1a8/0x3f8
[c000000006cefbc0] [c000000000034c2c] .spufs_run_spu+0x4fc/0xa7c
[c000000006cefcc0] [c000000000030bb8] .do_spu_run+0xb0/0x17c
[c000000006cefd80] [c00000000002a36c] .sys_spu_run+0x80/0xc8
[c000000006cefe30] [c0000000000076b4] syscall_exit+0x0/0x40
Instruction dump:
4bffffc8 e81f0040 7fa0f800 41be0028 4be845c9 60000000 2fa30000 419e0018 
e93e8000 80090000 2f800000 409e0008 <0fe00000> 3800ffff 7c2004ac 7d20f828 

  - QS20

    # make check-test_dma_page_fault.elf TEST_SPES=17

Badness at kernel/mutex.c:353
NIP: c00000000035e358 LR: c00000000035e33c CTR: 0000000000000000
REGS: c00000001ef777b0 TRAP: 0700   Not tainted  (2.6.25-08839-g02539d7-dirty)
MSR: 9000000000029032 <EE,ME,IR,DR>  CR: 24002482  XER: 20000006
TASK = c00000001e13b840[1738] 'test_dma_page_f' THREAD: c00000001ef74000 CPU: 3
GPR00: 0000000000000000 c00000001ef77a30 c000000000564470 0000000000000001 
GPR04: 0000000000000000 0000000000000000 c00000001ef77a88 0000000000000400 
GPR08: c00000001e13b878 c0000000005f0240 0000000000000000 0000000000000001 
GPR12: 100000000000d032 c0000000005d3a80 00000000100d0000 0000000000000000 
GPR16: 00000000100e4558 00000000100600f4 00000000100600f0 0000000010001f30 
GPR20: 0000000000000000 0000000000000000 c0000000005d57b8 0000000000000000 
GPR24: c00000001e4db000 c00000001e67f000 c00000001e4db000 0000000000000001 
GPR28: 0000000000000001 c00000001e4db000 c000000000517dd8 c00000001e4db888 
NIP [c00000000035e358] .mutex_trylock+0xdc/0x198
LR [c00000000035e33c] .mutex_trylock+0xc0/0x198
Call Trace:
[c00000001ef77a30] [c00000000035ecd4] .mutex_lock+0x28/0x48 (unreliable)
[c00000001ef77ac0] [c000000000043108] .spu_activate+0x200/0x4b8
[c00000001ef77bc0] [c0000000000452f8] .spufs_run_spu+0x4fc/0xae8
[c00000001ef77cc0] [c000000000040cf8] .do_spu_run+0xb0/0x17c
[c00000001ef77d80] [c000000000039f48] .sys_spu_run+0x80/0xc8
[c00000001ef77e30] [c0000000000076b4] syscall_exit+0x0/0x40
Instruction dump:
4bffffc8 e81f0028 7fa0f800 41be0028 4be6e8e1 60000000 2fa30000 419e0018 
e93e8000 80090000 2f800000 409e0008 <0fe00000> 3800ffff 7c2004ac 7d20f828 

-- 
(ASAYAMA Kazunori
  (asayama at sm.sony.co.jp))
t