[openib-general] oops in ib_at when running with CONFIG_DEBUG_PAGEALLOC

[Hal Rosenstock]

On Sun, 2005-05-29 at 06:23, Itamar Rabenstein wrote:
> I am running kdapltest -T Q with CONFIG_DEBUG_PAGEALLOC on 
> and i am geting this oops
>  
> x86 + Suse9.1 + k2.6.11 + svn rev 2507 + kdapl and ib_at from
> users/jlentini
>  
> when running without CONFIG_DEBUG_PAGEALLOC i dont see the oops
>  
>  
> May 29 11:38:24 mtvs42 kernel: Kdapltest Module Loaded - major 254
> May 29 11:38:51 mtvs42 kernel: ib_mthca 0000:04:00.0: Registering
> memory at 2c5a8000 (iova 2c5a8000) in PD 5; shift 12, npages 6.
> May 29 11:38:52 mtvs42 kernel: Unable to handle kernel paging request
> at virtual address eca6af74
> May 29 11:38:52 mtvs42 kernel:  printing eip:
> May 29 11:38:52 mtvs42 kernel: f8ad33bc
> May 29 11:38:52 mtvs42 kernel: *pde = 0047e067
> May 29 11:38:52 mtvs42 kernel: Oops: 0002 [#1]
> May 29 11:38:52 mtvs42 kernel: SMP DEBUG_PAGEALLOC
> May 29 11:38:52 mtvs42 kernel: Modules linked in: kdapltest
> ib_dat_provider ib_cm ib_at dat i
> b_umad ib_ipoib ib_sa edd joydev st sd_mod sr_mod scsi_mod ide_cd
> cdrom nvram usbserial parpo
> rt_pc lp parport autofs ipv6 thermal processor fan button battery ac
> e100 mii ib_mthca ib_mad
>  ib_core i2c_i801 i2c_core hw_random uhci_hcd usbcore evdev dm_mod
> ext3 jbd
> May 29 11:38:52 mtvs42 kernel: CPU:    0
> May 29 11:38:52 mtvs42 kernel: EIP:   
> 0060:[pg0+946295740/1069155328]    Not tainted VLI
> May 29 11:38:52 mtvs42 kernel: EIP:    0060:[<f8ad33bc>]    Not
> tainted VLI
> May 29 11:38:52 mtvs42 kernel: EFLAGS: 00010297   (2.6.11-gen2) 
> May 29 11:38:52 mtvs42 kernel: EIP is at req_free+0x4c/0x70 [ib_at]
> May 29 11:38:52 mtvs42 kernel: eax: 00000001   ebx: eca6af6c   ecx:
> c1594d40   edx: 00000640
> May 29 11:38:52 mtvs42 kernel: esi: ec6ae7f8   edi: eca6af90   ebp:
> eca6af6c   esp: f0409f5c
> May 29 11:38:52 mtvs42 kernel: ds: 007b   es: 007b   ss: 0068
> May 29 11:38:52 mtvs42 kernel: Process ib_at_wq/0 (pid: 9619,
> threadinfo=f0408000 task=ec8f7ad0)
> May 29 11:38:52 mtvs42 kernel: Stack: eca6af8c c012eea8 ec6ae820
> f8ad3470 00000286 f0408000 ec6ae808 ffffffff 
> May 29 11:38:52 mtvs42 kernel:        ffffffff 00000001 00000000
> c011bbd0 00010000 00000000 00000000 ef579f38 
> May 29 11:38:52 mtvs42 kernel:        ece8ed8c 00000000 ec8f7ad0
> c011bbd0 00100100 00200200 88497fc0 000f428d 
> May 29 11:38:52 mtvs42 kernel: Call Trace:
> May 29 11:38:52 mtvs42 kernel:  [worker_thread+424/560]
> worker_thread+0x1a8/0x230
> May 29 11:38:52 mtvs42 kernel:  [<c012eea8>] worker_thread+0x1a8/0x230
> May 29 11:38:52 mtvs42 kernel:  [pg0+946295920/1069155328]
> req_comp_work+0x0/0x20 [ib_at]
> May 29 11:38:52 mtvs42 kernel:  [<f8ad3470>] req_comp_work+0x0/0x20
> [ib_at]
> May 29 11:38:52 mtvs42 kernel:  [default_wake_function+0/16]
> default_wake_function+0x0/0x10
> May 29 11:38:52 mtvs42 kernel:  [<c011bbd0>]
> default_wake_function+0x0/0x10
> May 29 11:38:52 mtvs42 kernel:  [default_wake_function+0/16]
> default_wake_function+0x0/0x10
> May 29 11:38:52 mtvs42 kernel:  [<c011bbd0>]
> default_wake_function+0x0/0x10
> May 29 11:38:52 mtvs42 kernel:  [worker_thread+0/560]
> worker_thread+0x0/0x230
> May 29 11:38:52 mtvs42 kernel:  [<c012ed00>] worker_thread+0x0/0x230
> May 29 11:38:52 mtvs42 kernel:  [kthread+136/176] kthread+0x88/0xb0
> May 29 11:38:52 mtvs42 kernel:  [<c0132a28>] kthread+0x88/0xb0
> May 29 11:38:52 mtvs42 kernel:  [kthread+0/176] kthread+0x0/0xb0
> May 29 11:38:52 mtvs42 kernel:  [<c01329a0>] kthread+0x0/0xb0
> May 29 11:38:52 mtvs42 kernel:  [kernel_thread_helper+5/16]
> kernel_thread_helper+0x5/0x10
> May 29 11:38:52 mtvs42 kernel:  [<c01022f5>]
> kernel_thread_helper+0x5/0x10
> May 29 11:38:52 mtvs42 kernel: Code: ad f8 e8 28 ca 64 c7 83 c4 0c c7
> 43 08 00 00 00 00 c7 43
>  0c 00 00 00 00 c7 83 84 00 00 00 00 00 00 00 5b c3 89 d8 e8 34 03 00
> 00 <c7> 43 08 00 00 00 
> 00 c7 43 0c 00 00 00 00 c7 83 84 00 00 00 00 

I think this may be an earlier form of the other oops reported by Tom.

I think I now see what is causing this. I should have a patch for this
in a short while.

-- Hal