"booke: Fix problem with _tlbil_va being interrupted" causes BUG()

[Kumar Gala]

On Dec 17, 2008, at 3:34 PM, Sebastian Andrzej Siewior wrote:

> I just pulled from Linus and the Kernel I booted gave me this:
> | Freeing unused kernel memory: 152k init
> | BUG: sleeping function called from invalid context at /home/ 
> bigeasy/git/linux-2.6-powerpc/fs/exec.c:417
> | in_atomic(): 0, irqs_disabled(): 1, pid: 1, name: swapper
> | Call Trace:
> | [ef831dd0] [c0007764] show_stack+0x48/0x154 (unreliable)
> | [ef831e00] [c002b4d0] __might_sleep+0x118/0x11c
> | [ef831e10] [c009bcfc] copy_strings+0x70/0x324
> | [ef831e70] [c009bfd0] copy_strings_kernel+0x20/0x38
> | [ef831e90] [c009d3b8] do_execve+0x15c/0x1ec
> | [ef831ec0] [c00084ac] sys_execve+0x58/0x84
> | [ef831ee0] [c000f0e8] ret_from_syscall+0x0/0x3c
> | [ef831fa0] [c0001d68] init_post+0x54/0x174
> | [ef831fc0] [c038a1c0] kernel_init+0xdc/0x104
> | [ef831ff0] [c000ee68] kernel_thread+0x4c/0x68
> | ------------[ cut here ]------------
>
> It looks like userspace starts here. A bisect leads to
>
> |commit e5e774d8833de1a0037be2384efccadf16935675
> |Author: Kumar Gala <galak at kernel.crashing.org>
> |Date:   Sat Dec 13 17:01:37 2008 -0600
> |
> |    powerpc/fsl-booke: Fix problem with _tlbil_va being interrupted
> |
> |    An example calling sequence which we did see:
> |
> |    copy_user_highpage -> kmap_atomic -> flush_tlb_page -> _tlbil_va
> |
> |    We got interrupted after setting up the MAS registers before the
> |    tlbwe and the interrupt handler that caused the interrupt also  
> did
> |    a kmap_atomic (ide code) and thus on returning from the interrupt
> |    the MAS registers no longer contained the proper values.
> |
> |    Since we dont save/restore MAS registers for normal interrupts we
> |    need to disable interrupts in _tlbil_va to ensure atomicity.
> |
> |    Signed-off-by: Kumar Gala <galak at kernel.crashing.org>
> |
> |:040000 040000 1f298a8930bd2704f4481307f3d22589943bb5c0  
> 7e3847dab748c332c8533111de32d23731ec778b M      arch
>
> as the guilty commit.
> This box has highmem.
>
> Sebastian

Dave Liu noticed something similar.. Feel free to send me a brown  
paper bag.

Fixed in my 'merge' branch and just waiting for Paul or Linus to pick  
it up.

http://ozlabs.org/pipermail/linuxppc-dev/2008-December/066352.html

- k