linux-mm.kvack.org archive mirror
 help / color / mirror / Atom feed
* [BUG -next] "mm: per-thread vma caching fix 5" breaks s390
@ 2014-03-18 12:41 Heiko Carstens
  2014-03-18 16:38 ` Davidlohr Bueso
  2014-03-18 23:06 ` Tony Luck
  0 siblings, 2 replies; 6+ messages in thread
From: Heiko Carstens @ 2014-03-18 12:41 UTC (permalink / raw)
  To: Andrew Morton, Davidlohr Bueso, Michel Lespinasse, Sasha Levin,
	Rik van Riel, Linus Torvalds
  Cc: linux-next, linux-kernel, linux-mm, Martin Schwidefsky

Hi Andrew,

your patch "mm-per-thread-vma-caching-fix-5" in linux-next (see below) breaks s390:

[   10.101173] kernel BUG at mm/vmacache.c:76!
[   10.101206] illegal operation: 0001 [#1] SMP DEBUG_PAGEALLOC
[   10.101210] Modules linked in:
[   10.101212] CPU: 3 PID: 2286 Comm: ifup-eth Not tainted 3.14.0-rc6-00193-g7f31667faba3 #20
[   10.101214] task: 000000003f65cb90 ti: 000000003db30000 task.ti: 000000003db30000
[   10.101220] Krnl PSW : 0704d00180000000 000000000025df40 (vma_interval_tree_augment_rotate+0x0/0x64)
[   10.101222]            R:0 T:1 IO:1 EX:1 Key:0 M:1 W:0 P:0 AS:3 CC:1 PM:0 EA:3
               Krnl GPRS: 0000000000000000 0000000000000018 000000003a42cfd0 00000000800fb000
[   10.101225]            0000000000000001 000000003f65cb90 0000000000000000 000000003dbacba8
[   10.101226]            0705100180000000 000000003dbacb00 000000003f65cb90 000000003dbacb00
[   10.101227]            000000003a42cfd0 00000000800fb000 0000000000269e54 000000003db33d80
[   10.101235] Krnl Code: 000000000025df32: e3b0c0400020        cg      %r11,64(%r12)
                          000000000025df38: a784ffd1            brc     8,25deda
                         #000000000025df3c: a7f40001            brc     15,25df3e
                         >000000000025df40: e31020180004        lg      %r1,24(%r2)
                          000000000025df46: e31030180024        stg     %r1,24(%r3)
                          000000000025df4c: e3302fb0ff04        lg      %r3,-80(%r2)
                          000000000025df52: e31020400004        lg      %r1,64(%r2)
                          000000000025df58: e3302fa8ff09        sg      %r3,-88(%r2)
[   10.101251] Call Trace:
[   10.101253] ([<000000003dbacb00>] 0x3dbacb00)
[   10.101256]  [<00000000007a62da>] do_protection_exception+0x12a/0x3b4
[   10.101258]  [<00000000007a4862>] pgm_check_handler+0x17a/0x17e
[   10.101259]  [<0000000080086806>] 0x80086806
[   10.101260] INFO: lockdep is turned off.
[   10.101261] Last Breaking-Event-Address:
[   10.101262]  [<000000000025df3c>] vmacache_find+0x80/0x84
[   10.101264]  
[   10.101265] Kernel panic - not syncing: Fatal exception: panic_on_oops

Given that this is just an addon patch to Davidlohr's "mm: per-thread
vma caching" patch I was wondering if something in there is architecture
specific.
But it doesn't look like that. So I'm wondering if this only breaks on
s390?

commit 7f31667faba32a4cf1e20d042c17783168c95f1b
Author: Andrew Morton <akpm@linux-foundation.org>
Date:   Mon Mar 17 11:23:53 2014 +1100

    mm-per-thread-vma-caching-fix-5
    
    a sanity check
    
    Cc: Davidlohr Bueso <davidlohr@hp.com>
    Cc: Sasha Levin <sasha.levin@oracle.com>
    Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

diff --git a/mm/vmacache.c b/mm/vmacache.c
index add3162bf735..a265dd338228 100644
--- a/mm/vmacache.c
+++ b/mm/vmacache.c
@@ -72,8 +72,10 @@ struct vm_area_struct *vmacache_find(struct mm_struct *mm, unsigned long addr)
        for (i = 0; i < VMACACHE_SIZE; i++) {
                struct vm_area_struct *vma = current->vmacache[i];
 
-               if (vma && vma->vm_start <= addr && vma->vm_end > addr)
+               if (vma && vma->vm_start <= addr && vma->vm_end > addr) {
+                       BUG_ON(vma->vm_mm != mm);
                        return vma;
+               }
        }
 
        return NULL;

--
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>

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: [BUG -next] "mm: per-thread vma caching fix 5" breaks s390
  2014-03-18 12:41 [BUG -next] "mm: per-thread vma caching fix 5" breaks s390 Heiko Carstens
@ 2014-03-18 16:38 ` Davidlohr Bueso
  2014-03-18 23:06 ` Tony Luck
  1 sibling, 0 replies; 6+ messages in thread
From: Davidlohr Bueso @ 2014-03-18 16:38 UTC (permalink / raw)
  To: Heiko Carstens
  Cc: Andrew Morton, Michel Lespinasse, Sasha Levin, Rik van Riel,
	Linus Torvalds, linux-next, linux-kernel, linux-mm,
	Martin Schwidefsky

On Tue, 2014-03-18 at 13:41 +0100, Heiko Carstens wrote:
> Hi Andrew,
> 
> your patch "mm-per-thread-vma-caching-fix-5" in linux-next (see below) breaks s390:
> 
> [   10.101173] kernel BUG at mm/vmacache.c:76!
> [   10.101206] illegal operation: 0001 [#1] SMP DEBUG_PAGEALLOC
> [   10.101210] Modules linked in:
> [   10.101212] CPU: 3 PID: 2286 Comm: ifup-eth Not tainted 3.14.0-rc6-00193-g7f31667faba3 #20
> [   10.101214] task: 000000003f65cb90 ti: 000000003db30000 task.ti: 000000003db30000
> [   10.101220] Krnl PSW : 0704d00180000000 000000000025df40 (vma_interval_tree_augment_rotate+0x0/0x64)
> [   10.101222]            R:0 T:1 IO:1 EX:1 Key:0 M:1 W:0 P:0 AS:3 CC:1 PM:0 EA:3
>                Krnl GPRS: 0000000000000000 0000000000000018 000000003a42cfd0 00000000800fb000
> [   10.101225]            0000000000000001 000000003f65cb90 0000000000000000 000000003dbacba8
> [   10.101226]            0705100180000000 000000003dbacb00 000000003f65cb90 000000003dbacb00
> [   10.101227]            000000003a42cfd0 00000000800fb000 0000000000269e54 000000003db33d80
> [   10.101235] Krnl Code: 000000000025df32: e3b0c0400020        cg      %r11,64(%r12)
>                           000000000025df38: a784ffd1            brc     8,25deda
>                          #000000000025df3c: a7f40001            brc     15,25df3e
>                          >000000000025df40: e31020180004        lg      %r1,24(%r2)
>                           000000000025df46: e31030180024        stg     %r1,24(%r3)
>                           000000000025df4c: e3302fb0ff04        lg      %r3,-80(%r2)
>                           000000000025df52: e31020400004        lg      %r1,64(%r2)
>                           000000000025df58: e3302fa8ff09        sg      %r3,-88(%r2)
> [   10.101251] Call Trace:
> [   10.101253] ([<000000003dbacb00>] 0x3dbacb00)
> [   10.101256]  [<00000000007a62da>] do_protection_exception+0x12a/0x3b4
> [   10.101258]  [<00000000007a4862>] pgm_check_handler+0x17a/0x17e
> [   10.101259]  [<0000000080086806>] 0x80086806
> [   10.101260] INFO: lockdep is turned off.
> [   10.101261] Last Breaking-Event-Address:
> [   10.101262]  [<000000000025df3c>] vmacache_find+0x80/0x84
> [   10.101264]  
> [   10.101265] Kernel panic - not syncing: Fatal exception: panic_on_oops
> 
> Given that this is just an addon patch to Davidlohr's "mm: per-thread
> vma caching" patch I was wondering if something in there is architecture
> specific.
> But it doesn't look like that. So I'm wondering if this only breaks on
> s390?

No, there isn't anything arch specific. Please note that there are a few
other patches in -mm that fix the actual issue that triggers that
BUG_ON(), so you'll want to try those.

--
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>

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: [BUG -next] "mm: per-thread vma caching fix 5" breaks s390
  2014-03-18 12:41 [BUG -next] "mm: per-thread vma caching fix 5" breaks s390 Heiko Carstens
  2014-03-18 16:38 ` Davidlohr Bueso
@ 2014-03-18 23:06 ` Tony Luck
  2014-03-18 23:10   ` Andrew Morton
  1 sibling, 1 reply; 6+ messages in thread
From: Tony Luck @ 2014-03-18 23:06 UTC (permalink / raw)
  To: Heiko Carstens
  Cc: Andrew Morton, Davidlohr Bueso, Michel Lespinasse, Sasha Levin,
	Rik van Riel, Linus Torvalds, linux-next,
	Linux Kernel Mailing List, linux-mm, Martin Schwidefsky

On Tue, Mar 18, 2014 at 5:41 AM, Heiko Carstens
<heiko.carstens@de.ibm.com> wrote:
> Given that this is just an addon patch to Davidlohr's "mm: per-thread
> vma caching" patch I was wondering if something in there is architecture
> specific.
> But it doesn't look like that. So I'm wondering if this only breaks on
> s390?

I'm seeing this same BUG_ON() on ia64 (when trying out next-20140318)

Starting HAkernel BUG at mm/vmacache.c:76!
L daemon7[?25lps[3259]: bugcheck! 0 [1]
Modules linked in: mptctl

CPU: 0 PID: 3259 Comm: ps Not tainted 3.14.0-rc7-zx1-smp-next-20140318 #1
Hardware name: hp server rx2620                   , BIOS 03.17
                                                   03/31/2005
task: e000000001070000 ti: e000000001070c80 task.ti: e000000001070c80
psr : 0000101008526038 ifs : 8000000000000309 ip  :
[<a00000010019a930>]    Not tainted (3.14.0-rc7-zx1-smp-next-20140318)
ip is at vmacache_find+0x1d0/0x1e0
unat: 0000000000000000 pfs : 0000000000000309 rsc : 0000000000000003
rnat: 0009804c0270033f bsps: a00000010153e470 pr  : 0000000019a99565
ldrs: 0000000000000000 ccv : 00000010e0bb89b1 fpsr: 0009804c0270033f
csd : 0000000000000000 ssd : 0000000000000000
b0  : a00000010019a930 b6  : a0000001006cff20 b7  : a0000001006d1a40
f6  : 1003e000000167080aa40 f7  : 1003e0000000000000514
f8  : 1003e000000167080a52c f9  : 1003e0000000000000001
f10 : 1003e501ac672552bd930 f11 : 1003e0000000000000004
r1  : a00000010176aa00 r2  : a00000010153e468 r3  : a00000010156b5e0
r8  : 000000000000001f r9  : 0000000000001b14 r10 : ffffffffffffffff
r11 : 0000000000000000 r12 : e000000001077df0 r13 : e000000001070000
r14 : a00000010153e470 r15 : a00000010153e470 r16 : 000000001b140d8a
r17 : 0000000000000000 r18 : 0000000000007fff r19 : 0000000000000182
r20 : 0000000000000003 r21 : 0000000000000000 r22 : 0000000000000182
r23 : a00000010130a9a8 r24 : a0000001006cff20 r25 : a00000010130a9a8
r26 : a0000001006cff20 r27 : a000000101544658 r28 : a0000001015e2508
r29 : a00000010130a998 r30 : a0000001006cfec0 r31 : a0000001015e23e0

Call Trace:
 [<a000000100015460>] show_stack+0x80/0xa0
                                sp=e0000000010779b0 bsp=e0000000010712e0
 [<a000000100015ac0>] show_regs+0x640/0x920
                                sp=e000000001077b80 bsp=e000000001071288
 [<a0000001000424c0>] die+0x1a0/0x2e0
                                sp=e000000001077b90 bsp=e000000001071248
 [<a000000100042650>] die_if_kernel+0x50/0x80
                                sp=e000000001077b90 bsp=e000000001071218
 [<a000000100f3df10>] ia64_bad_break+0x3d0/0x6e0
                                sp=e000000001077b90 bsp=e0000000010711e8
 [<a00000010000c740>] ia64_native_leave_kernel+0x0/0x270
                                sp=e000000001077c20 bsp=e0000000010711e8
 [<a00000010019a930>] vmacache_find+0x1d0/0x1e0
                                sp=e000000001077df0 bsp=e0000000010711a0
 [<a0000001001b4cb0>] find_vma+0x30/0x140
                                sp=e000000001077df0 bsp=e000000001071170
 [<a0000001001b4df0>] find_extend_vma+0x30/0x140
                                sp=e000000001077df0 bsp=e000000001071138
 [<a0000001001a9080>] __get_user_pages+0x120/0xd60
                                sp=e000000001077df0 bsp=e000000001071010
 [<a0000001001a9dd0>] get_user_pages+0x70/0xa0
                                sp=e000000001077e10 bsp=e000000001070fb0
 [<a0000001001a9eb0>] __access_remote_vm+0xb0/0x360
                                sp=e000000001077e10 bsp=e000000001070f18
 [<a0000001001aa1c0>] access_process_vm+0x60/0xa0
                                sp=e000000001077e20 bsp=e000000001070ed0
 [<a00000010018a390>] get_cmdline+0xb0/0x240
                                sp=e000000001077e20 bsp=e000000001070e88
 [<a0000001002ca2f0>] proc_pid_cmdline+0x30/0x60
                                sp=e000000001077e20 bsp=e000000001070e60
 [<a0000001002c7880>] proc_info_read+0x120/0x200
                                sp=e000000001077e20 bsp=e000000001070e10
 [<a000000100201da0>] vfs_read+0x140/0x340
                                sp=e000000001077e20 bsp=e000000001070dc0
 [<a000000100202180>] SyS_read+0xa0/0x120
                                sp=e000000001077e20 bsp=e000000001070d40
 [<a00000010000c5c0>] ia64_ret_from_syscall+0x0/0x20
                                sp=e000000001077e30 bsp=e000000001070d40
 [<a000000000040720>] ia64_ivt+0xffffffff00040720/0x400
                                sp=e000000001078000 bsp=e000000001070d40
Disabling lock debugging due to kernel taint

--
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>

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: [BUG -next] "mm: per-thread vma caching fix 5" breaks s390
  2014-03-18 23:06 ` Tony Luck
@ 2014-03-18 23:10   ` Andrew Morton
  2014-03-18 23:27     ` Stephen Rothwell
  0 siblings, 1 reply; 6+ messages in thread
From: Andrew Morton @ 2014-03-18 23:10 UTC (permalink / raw)
  To: Tony Luck
  Cc: Heiko Carstens, Davidlohr Bueso, Michel Lespinasse, Sasha Levin,
	Rik van Riel, Linus Torvalds, linux-next,
	Linux Kernel Mailing List, linux-mm, Martin Schwidefsky

On Tue, 18 Mar 2014 16:06:59 -0700 Tony Luck <tony.luck@gmail.com> wrote:

> On Tue, Mar 18, 2014 at 5:41 AM, Heiko Carstens
> <heiko.carstens@de.ibm.com> wrote:
> > Given that this is just an addon patch to Davidlohr's "mm: per-thread
> > vma caching" patch I was wondering if something in there is architecture
> > specific.
> > But it doesn't look like that. So I'm wondering if this only breaks on
> > s390?
> 
> I'm seeing this same BUG_ON() on ia64 (when trying out next-20140318)

-next is missing
http://ozlabs.org/~akpm/mmots/broken-out/mm-per-thread-vma-caching-fix-6.patch.
Am presently trying to cook up an mmotm for tomorrow's -next.  

--
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>

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: [BUG -next] "mm: per-thread vma caching fix 5" breaks s390
  2014-03-18 23:10   ` Andrew Morton
@ 2014-03-18 23:27     ` Stephen Rothwell
  2014-03-21 20:39       ` Tony Luck
  0 siblings, 1 reply; 6+ messages in thread
From: Stephen Rothwell @ 2014-03-18 23:27 UTC (permalink / raw)
  To: Andrew Morton
  Cc: Tony Luck, Heiko Carstens, Davidlohr Bueso, Michel Lespinasse,
	Sasha Levin, Rik van Riel, Linus Torvalds, linux-next,
	Linux Kernel Mailing List, linux-mm, Martin Schwidefsky

[-- Attachment #1: Type: text/plain, Size: 1016 bytes --]

Hi Andrew,

On Tue, 18 Mar 2014 16:10:50 -0700 Andrew Morton <akpm@linux-foundation.org> wrote:
>
> On Tue, 18 Mar 2014 16:06:59 -0700 Tony Luck <tony.luck@gmail.com> wrote:
> 
> > On Tue, Mar 18, 2014 at 5:41 AM, Heiko Carstens
> > <heiko.carstens@de.ibm.com> wrote:
> > > Given that this is just an addon patch to Davidlohr's "mm: per-thread
> > > vma caching" patch I was wondering if something in there is architecture
> > > specific.
> > > But it doesn't look like that. So I'm wondering if this only breaks on
> > > s390?
> > 
> > I'm seeing this same BUG_ON() on ia64 (when trying out next-20140318)
> 
> -next is missing
> http://ozlabs.org/~akpm/mmots/broken-out/mm-per-thread-vma-caching-fix-6.patch.
> Am presently trying to cook up an mmotm for tomorrow's -next.  

I could apply just that patch to today's linux-next if you like.
Otherwise if you get a new mmotm out in the next ~6-8 hours, I will use
that.
-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au

[-- Attachment #2: Type: application/pgp-signature, Size: 836 bytes --]

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: [BUG -next] "mm: per-thread vma caching fix 5" breaks s390
  2014-03-18 23:27     ` Stephen Rothwell
@ 2014-03-21 20:39       ` Tony Luck
  0 siblings, 0 replies; 6+ messages in thread
From: Tony Luck @ 2014-03-21 20:39 UTC (permalink / raw)
  To: Stephen Rothwell
  Cc: Andrew Morton, Heiko Carstens, Davidlohr Bueso,
	Michel Lespinasse, Sasha Levin, Rik van Riel, Linus Torvalds,
	linux-next, Linux Kernel Mailing List, linux-mm,
	Martin Schwidefsky

Problem is no longer present in next-20140321.

-Tony

--
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>

^ permalink raw reply	[flat|nested] 6+ messages in thread

end of thread, other threads:[~2014-03-21 20:39 UTC | newest]

Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2014-03-18 12:41 [BUG -next] "mm: per-thread vma caching fix 5" breaks s390 Heiko Carstens
2014-03-18 16:38 ` Davidlohr Bueso
2014-03-18 23:06 ` Tony Luck
2014-03-18 23:10   ` Andrew Morton
2014-03-18 23:27     ` Stephen Rothwell
2014-03-21 20:39       ` Tony Luck

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox