2007-05-11 21:34:01

by Oliver Xymoron

[permalink] [raw]
Subject: SLUB under lguest on i386

Tried to boot using SLUB under lguest with 2.6.21-mm2. Got the
following:

...
[ 0.388000] NET: Registered protocol family 17
[ 0.388000] Using IPI Shortcut mode
[ 0.420000] EXT2-fs warning: mounting unchecked fs, running e2fsck
is recommended
[ 0.420000] VFS: Mounted root (ext2 filesystem).
[ 0.420000] Freeing unused kernel memory: 188k freed
[ 17.536000] BUG: spinlock lockup on CPU#0, init/1, c17de0c0
[ 17.536000] [<c0104584>] show_trace_log_lvl+0x1a/0x30
[ 17.536000] [<c0105196>] show_trace+0x12/0x14
[ 17.536000] [<c0105222>] dump_stack+0x15/0x17
[ 17.536000] [<c021339f>] _raw_spin_lock+0xbf/0xd8
[ 17.536000] [<c03cd41a>] _spin_lock+0x3c/0x42
[ 17.536000] [<c0154f0c>] vma_adjust+0x1a9/0x35f
[ 17.536000] [<c01554c3>] vma_merge+0x158/0x168
[ 17.536000] [<c0155b3b>] mmap_region+0x13b/0x405
[ 17.536000] [<c01560ec>] do_mmap_pgoff+0x211/0x269
[ 17.536000] [<c0107594>] sys_mmap2+0x64/0xb4
[ 17.536000] [<c010763d>] old_mmap+0x59/0x5f
[ 17.536000] [<c0103ed0>] syscall_call+0x7/0xb
[ 17.536000] =======================

And no sign of further progress. SLAB worked fine.

--
"Love the dolphins," she advised him. "Write by W.A.S.T.E.."


2007-05-12 00:26:42

by Christoph Lameter

[permalink] [raw]
Subject: Re: SLUB under lguest on i386

On Fri, 11 May 2007, Oliver Xymoron wrote:

> And no sign of further progress. SLAB worked fine.

Add slub_debug to the command line. Any changes or any additional
diagnostic output?


2007-05-12 18:28:56

by Jens Axboe

[permalink] [raw]
Subject: Re: SLUB under lguest on i386

On Fri, May 11 2007, Oliver Xymoron wrote:
> Tried to boot using SLUB under lguest with 2.6.21-mm2. Got the
> following:
>
> ...
> [ 0.388000] NET: Registered protocol family 17
> [ 0.388000] Using IPI Shortcut mode
> [ 0.420000] EXT2-fs warning: mounting unchecked fs, running e2fsck
> is recommended
> [ 0.420000] VFS: Mounted root (ext2 filesystem).
> [ 0.420000] Freeing unused kernel memory: 188k freed
> [ 17.536000] BUG: spinlock lockup on CPU#0, init/1, c17de0c0
> [ 17.536000] [<c0104584>] show_trace_log_lvl+0x1a/0x30
> [ 17.536000] [<c0105196>] show_trace+0x12/0x14
> [ 17.536000] [<c0105222>] dump_stack+0x15/0x17
> [ 17.536000] [<c021339f>] _raw_spin_lock+0xbf/0xd8
> [ 17.536000] [<c03cd41a>] _spin_lock+0x3c/0x42
> [ 17.536000] [<c0154f0c>] vma_adjust+0x1a9/0x35f
> [ 17.536000] [<c01554c3>] vma_merge+0x158/0x168
> [ 17.536000] [<c0155b3b>] mmap_region+0x13b/0x405
> [ 17.536000] [<c01560ec>] do_mmap_pgoff+0x211/0x269
> [ 17.536000] [<c0107594>] sys_mmap2+0x64/0xb4
> [ 17.536000] [<c010763d>] old_mmap+0x59/0x5f
> [ 17.536000] [<c0103ed0>] syscall_call+0x7/0xb
> [ 17.536000] =======================
>
> And no sign of further progress. SLAB worked fine.

Exact same thing happens on real hardware here.

--
Jens Axboe

2007-05-13 23:48:00

by Christoph Lameter

[permalink] [raw]
Subject: Re: SLUB under lguest on i386

On Sat, 12 May 2007, Jens Axboe wrote:

> On Fri, May 11 2007, Oliver Xymoron wrote:
> > Tried to boot using SLUB under lguest with 2.6.21-mm2. Got the
> > following:
> >
> > ...
> > [ 0.388000] NET: Registered protocol family 17
> > [ 0.388000] Using IPI Shortcut mode
> > [ 0.420000] EXT2-fs warning: mounting unchecked fs, running e2fsck
> > is recommended
> > [ 0.420000] VFS: Mounted root (ext2 filesystem).
> > [ 0.420000] Freeing unused kernel memory: 188k freed
> > [ 17.536000] BUG: spinlock lockup on CPU#0, init/1, c17de0c0
> > [ 17.536000] [<c0104584>] show_trace_log_lvl+0x1a/0x30
> > [ 17.536000] [<c0105196>] show_trace+0x12/0x14
> > [ 17.536000] [<c0105222>] dump_stack+0x15/0x17
> > [ 17.536000] [<c021339f>] _raw_spin_lock+0xbf/0xd8
> > [ 17.536000] [<c03cd41a>] _spin_lock+0x3c/0x42
> > [ 17.536000] [<c0154f0c>] vma_adjust+0x1a9/0x35f
> > [ 17.536000] [<c01554c3>] vma_merge+0x158/0x168
> > [ 17.536000] [<c0155b3b>] mmap_region+0x13b/0x405
> > [ 17.536000] [<c01560ec>] do_mmap_pgoff+0x211/0x269
> > [ 17.536000] [<c0107594>] sys_mmap2+0x64/0xb4
> > [ 17.536000] [<c010763d>] old_mmap+0x59/0x5f
> > [ 17.536000] [<c0103ed0>] syscall_call+0x7/0xb
> > [ 17.536000] =======================
> >
> > And no sign of further progress. SLAB worked fine.
>
> Exact same thing happens on real hardware here.

Anyone could try this with slub_debug on the kernel command line? Another
assumption that page size allocs are page size aligned?