CK4 [00/15] Sl[auo]b: Common kmalloc caches V4

CK4 [00/15] Sl[auo]b: Common kmalloc caches V4

[Christoph Lameter]

V3->V4:
 - Further fixes of issues pointed out by Joonsoo and Glauber.

V2-V3:
- Further cleanup and reordering as suggested by Glauber

V1-V2:
- Clean up numerous things as suggested by Glauber.
- Add two more patches that extract more kmalloc stuff
  into common files.

This patchset cleans up the bootstrap of the allocators
and creates a common functions to handle the kmalloc
array. The results are more common data structures and
functions that will simplify further work
on having common functions for all allocators.

This patchset is against Pekka's slab/next tree as of today.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: CK4 [00/15] Sl[auo]b: Common kmalloc caches V4

[Glauber Costa]

On 10/24/2012 07:05 PM, Christoph Lameter wrote:
> V3->V4:
>  - Further fixes of issues pointed out by Joonsoo and Glauber.
> 
> V2-V3:
> - Further cleanup and reordering as suggested by Glauber
> 
> V1-V2:
> - Clean up numerous things as suggested by Glauber.
> - Add two more patches that extract more kmalloc stuff
>   into common files.
> 
> This patchset cleans up the bootstrap of the allocators
> and creates a common functions to handle the kmalloc
> array. The results are more common data structures and
> functions that will simplify further work
> on having common functions for all allocators.
> 
> This patchset is against Pekka's slab/next tree as of today.

It looks good to me.


--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: CK4 [00/15] Sl[auo]b: Common kmalloc caches V4

[Ezequiel Garcia]

Hi Christoph,

On Wed, Oct 24, 2012 at 12:05 PM, Christoph Lameter <cl@linux.com> wrote:
> V3->V4:
>  - Further fixes of issues pointed out by Joonsoo and Glauber.
>
> V2-V3:
> - Further cleanup and reordering as suggested by Glauber
>
> V1-V2:
> - Clean up numerous things as suggested by Glauber.
> - Add two more patches that extract more kmalloc stuff
>   into common files.
>
> This patchset cleans up the bootstrap of the allocators
> and creates a common functions to handle the kmalloc
> array. The results are more common data structures and
> functions that will simplify further work
> on having common functions for all allocators.
>
> This patchset is against Pekka's slab/next tree as of today.
>

While testing this patchset, I found a BUG.

All I did was "sudo mount -a" to mount my development partitions.

[   25.366266] BUG: unable to handle kernel paging request at ffffffc0
[   25.366419] IP: [<c10d93b2>] slab_unmergeable+0x12/0x30
[   25.366497] *pde = 016f5067 *pte = 00000000
[   25.366601] Oops: 0000 [#1] SMP
[   25.366703] Modules linked in: radeon snd_usb_audio snd_usbmidi_lib
snd_rawmidi snd_hda_intel snd_hda_codec snd_hwdep snd_pcm ttm
snd_timer snd pcspkr drm_kms_helper soundcore snd_page_alloc
cfbcopyarea cfbimgblt cfbfillrect evdev
[   25.367373] Pid: 1428, comm: mount Not tainted
3.7.0-rc1-athlon-full-preempt-gentoo-69954-g12ae62c #14 Gigabyte
Technology Co., Ltd. GA-MA74GMT-S2/GA-MA74GMT-S2
[   25.367451] EIP: 0060:[<c10d93b2>] EFLAGS: 00010246 CPU: 2
[   25.367498] EIP is at slab_unmergeable+0x12/0x30
[   25.367543] EAX: ffffffbc EBX: 00030d00 ECX: 00000000 EDX: 00000001
[   25.367588] ESI: 00000098 EDI: ffffffbc EBP: f47ddd00 ESP: f47ddd00
[   25.367635]  DS: 007b ES: 007b FS: 00d8 GS: 0033 SS: 0068
[   25.367680] CR0: 8005003b CR2: ffffffc0 CR3: 34742000 CR4: 000007d0
[   25.367726] DR0: 00000000 DR1: 00000000 DR2: 00000000 DR3: 00000000
[   25.367772] DR6: ffff0ff0 DR7: 00000400
[   25.367817] Process mount (pid: 1428, ti=f47dc000 task=f3998000
task.ti=f47dc000)
[   25.367877] Stack:
[   25.367922]  f47ddd24 c10dcad7 00000094 fffffff8 00000097 c15b924f
00000002 c15b924f
[   25.368214]  00000094 f47ddd4c c10c1ed0 00030d00 00000000 22222222
22222222 00000000
[   25.368506]  00000002 f4419770 0000000c f47ddd90 c117e64f 00020000
00000000 f34079a8
[   25.368798] Call Trace:
[   25.368843]  [<c10dcad7>] __kmem_cache_alias+0x97/0x130
[   25.368891]  [<c10c1ed0>] kmem_cache_create+0x40/0x1c0
[   25.368938]  [<c117e64f>] ext4_mb_init+0x2ef/0x520
[   25.368986]  [<c148434d>] ? _raw_spin_unlock+0x1d/0x20
[   25.369033]  [<c116cf2e>] ext4_fill_super+0x2c2e/0x3310
[   25.369081]  [<c1075df5>] ? mark_held_locks+0x85/0xe0
[   25.369128]  [<c1482329>] ? mutex_lock_nested+0x229/0x2d0
[   25.369175]  [<c110e06e>] ? sb_set_blocksize+0x1e/0x70
[   25.369222]  [<c10e3205>] mount_bdev+0x165/0x190
[   25.369269]  [<c10c2450>] ? slab_account_alloc+0xd0/0x1a0
[   25.369315]  [<c10dd171>] ? __kmalloc_track_caller+0xc1/0x160
[   25.369362]  [<c11650da>] ext4_mount+0x1a/0x20
[   25.369408]  [<c116a300>] ? ext4_calculate_overhead+0x460/0x460
[   25.369455]  [<c10e3c7c>] mount_fs+0x1c/0xc0
[   25.369501]  [<c10c1aba>] ? __alloc_percpu+0xa/0x10
[   25.369549]  [<c10fa13f>] ? alloc_vfsmnt+0x9f/0x140
[   25.369596]  [<c10fa259>] vfs_kern_mount+0x49/0xe0
[   25.369642]  [<c10faa47>] do_kern_mount+0x37/0xf0
[   25.369688]  [<c10fc1f4>] do_mount+0x3b4/0x700
[   25.369735]  [<c10bd7c9>] ? strndup_user+0x49/0x70
[   25.369782]  [<c10fc5a6>] sys_mount+0x66/0xa0
[   25.369828]  [<c1484d7a>] sysenter_do_call+0x12/0x32
[   25.369873] Code: 34 c7 43 2c 00 00 00 00 c7 43 30 00 00 00 00 89
43 34 89 43 38 5b 5d c3 90 8b 0d 90 70 c3 c1 ba 01 00 00 00 55 89 e5
85 c9 75 10 <f7> 40 04 00 0c a9 00 75 07 8b 48 30 85 c9 74 06 89 d0 5d
c3 66
[   25.371837] EIP: [<c10d93b2>] slab_unmergeable+0x12/0x30 SS:ESP 0068:f47ddd00
[   25.371942] CR2: 00000000ffffffc0
[   25.371987] ---[ end trace 278dfa9b282c605e ]---

Hope it helps,

    Ezequiel

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: CK4 [00/15] Sl[auo]b: Common kmalloc caches V4

[Christoph Lameter]

On Thu, 1 Nov 2012, Ezequiel Garcia wrote:

> While testing this patchset, I found a BUG.
>
> All I did was "sudo mount -a" to mount my development partitions.
>
> [   25.366266] BUG: unable to handle kernel paging request at ffffffc0
> [   25.366419] IP: [<c10d93b2>] slab_unmergeable+0x12/0x30

Arg. More sysfs trouble I guess. Sysfs is the cause for a lot of slub
fragility. Sigh.

Can you rerun this with "slub_debug" as a kernel option?

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: CK4 [00/15] Sl[auo]b: Common kmalloc caches V4

[Ezequiel Garcia]

On Thu, Nov 1, 2012 at 5:24 PM, Christoph Lameter <cl@linux.com> wrote:
> On Thu, 1 Nov 2012, Ezequiel Garcia wrote:
>
>> While testing this patchset, I found a BUG.
>>
>> All I did was "sudo mount -a" to mount my development partitions.
>>
>> [   25.366266] BUG: unable to handle kernel paging request at ffffffc0
>> [   25.366419] IP: [<c10d93b2>] slab_unmergeable+0x12/0x30
>
> Arg. More sysfs trouble I guess. Sysfs is the cause for a lot of slub
> fragility. Sigh.
>
> Can you rerun this with "slub_debug" as a kernel option?

I will.

Also I will test *without* a few patches I was playing around with...
I should have done that before reporting :/
Until then, please consider this noise, just in case.

    Ezequiel

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: CK4 [00/15] Sl[auo]b: Common kmalloc caches V4

[Christoph Lameter]

On Thu, 1 Nov 2012, Ezequiel Garcia wrote:

> On Thu, Nov 1, 2012 at 5:24 PM, Christoph Lameter <cl@linux.com> wrote:
> > On Thu, 1 Nov 2012, Ezequiel Garcia wrote:
> >
> >> While testing this patchset, I found a BUG.
> >>
> >> All I did was "sudo mount -a" to mount my development partitions.
> >>
> >> [   25.366266] BUG: unable to handle kernel paging request at ffffffc0
> >> [   25.366419] IP: [<c10d93b2>] slab_unmergeable+0x12/0x30
> >
> > Arg. More sysfs trouble I guess. Sysfs is the cause for a lot of slub
> > fragility. Sigh.
> >
> > Can you rerun this with "slub_debug" as a kernel option?
>
> I will.
>
> Also I will test *without* a few patches I was playing around with...
> I should have done that before reporting :/
> Until then, please consider this noise, just in case.

Well found it. The create common boot functions patch (#2) did a list add
of a structure that was moved later during slub bootstrap. Fix will come
wiht V5.


--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: CK4 [00/15] Sl[auo]b: Common kmalloc caches V4

[Ezequiel Garcia]

On Thu, Nov 1, 2012 at 6:03 PM, Christoph Lameter <cl@linux.com> wrote:
> On Thu, 1 Nov 2012, Ezequiel Garcia wrote:
>
>> On Thu, Nov 1, 2012 at 5:24 PM, Christoph Lameter <cl@linux.com> wrote:
>> > On Thu, 1 Nov 2012, Ezequiel Garcia wrote:
>> >
>> >> While testing this patchset, I found a BUG.
>> >>
>> >> All I did was "sudo mount -a" to mount my development partitions.
>> >>
>> >> [   25.366266] BUG: unable to handle kernel paging request at ffffffc0
>> >> [   25.366419] IP: [<c10d93b2>] slab_unmergeable+0x12/0x30
>> >
>> > Arg. More sysfs trouble I guess. Sysfs is the cause for a lot of slub
>> > fragility. Sigh.
>> >
>> > Can you rerun this with "slub_debug" as a kernel option?
>>
>> I will.
>>
>> Also I will test *without* a few patches I was playing around with...
>> I should have done that before reporting :/
>> Until then, please consider this noise, just in case.
>
> Well found it. The create common boot functions patch (#2) did a list add
> of a structure that was moved later during slub bootstrap. Fix will come
> wiht V5.
>
>

Okey, great. Feel free to drop me a patch to test a fix.

    Ezequiel

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: CK4 [00/15] Sl[auo]b: Common kmalloc caches V4

[Christoph Lameter]

On Thu, 1 Nov 2012, Ezequiel Garcia wrote:

> Okey, great. Feel free to drop me a patch to test a fix.

CK5 coming in a few secs.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>