diff mbox series

[RESEND] mm/hugetlb: fix kernel NULL pointer dereference when migrating hugetlb folio

Message ID 20240709120433.4136700-1-linmiaohe@huawei.com (mailing list archive)
State New
Headers show
Series [RESEND] mm/hugetlb: fix kernel NULL pointer dereference when migrating hugetlb folio | expand

Commit Message

Miaohe Lin July 9, 2024, 12:04 p.m. UTC 
A kernel crash was observed when migrating hugetlb folio:

BUG: kernel NULL pointer dereference, address: 0000000000000008
PGD 0 P4D 0
Oops: Oops: 0002 [#1] PREEMPT SMP NOPTI
CPU: 0 PID: 3435 Comm: bash Not tainted 6.10.0-rc6-00450-g8578ca01f21f #66
RIP: 0010:__folio_undo_large_rmappable+0x70/0xb0
RSP: 0018:ffffb165c98a7b38 EFLAGS: 00000097
RAX: fffffbbc44528090 RBX: 0000000000000000 RCX: 0000000000000000
RDX: ffffa30e000a2800 RSI: 0000000000000246 RDI: ffffa3153ffffcc0
RBP: fffffbbc44528000 R08: 0000000000002371 R09: ffffffffbe4e5868
R10: 0000000000000001 R11: 0000000000000001 R12: ffffa3153ffffcc0
R13: fffffbbc44468000 R14: 0000000000000001 R15: 0000000000000001
FS:  00007f5b3a716740(0000) GS:ffffa3151fc00000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 0000000000000008 CR3: 000000010959a000 CR4: 00000000000006f0
Call Trace:
 <TASK>
 __folio_migrate_mapping+0x59e/0x950
 __migrate_folio.constprop.0+0x5f/0x120
 move_to_new_folio+0xfd/0x250
 migrate_pages+0x383/0xd70
 soft_offline_page+0x2ab/0x7f0
 soft_offline_page_store+0x52/0x90
 kernfs_fop_write_iter+0x12c/0x1d0
 vfs_write+0x380/0x540
 ksys_write+0x64/0xe0
 do_syscall_64+0xb9/0x1d0
 entry_SYSCALL_64_after_hwframe+0x77/0x7f
RIP: 0033:0x7f5b3a514887
RSP: 002b:00007ffe138fce68 EFLAGS: 00000246 ORIG_RAX: 0000000000000001
RAX: ffffffffffffffda RBX: 000000000000000c RCX: 00007f5b3a514887
RDX: 000000000000000c RSI: 0000556ab809ee10 RDI: 0000000000000001
RBP: 0000556ab809ee10 R08: 00007f5b3a5d1460 R09: 000000007fffffff
R10: 0000000000000000 R11: 0000000000000246 R12: 000000000000000c
R13: 00007f5b3a61b780 R14: 00007f5b3a617600 R15: 00007f5b3a616a00

It's because hugetlb folio is passed to __folio_undo_large_rmappable()
unexpectedly. large_rmappable flag is imperceptibly set to hugetlb folio
since commit f6a8dd98a2ce ("hugetlb: convert alloc_buddy_hugetlb_folio to
use a folio"). Then commit be9581ea8c05 ("mm: fix crashes from deferred
split racing folio migration") makes folio_migrate_mapping() call
folio_undo_large_rmappable() triggering the bug. Fix this issue by
clearing large_rmappable flag for hugetlb folios. They don't need that
flag set anyway.

Fixes: f6a8dd98a2ce ("hugetlb: convert alloc_buddy_hugetlb_folio to use a folio")
Fixes: be9581ea8c05 ("mm: fix crashes from deferred split racing folio migration")
Signed-off-by: Miaohe Lin <linmiaohe@huawei.com>
Cc: <stable@vger.kernel.org>
---
 mm/hugetlb.c | 3 +++
 1 file changed, 3 insertions(+)

Comments

Hugh Dickins July 10, 2024, 12:14 a.m. UTC | #1 
On Tue, 9 Jul 2024, Miaohe Lin wrote:

> A kernel crash was observed when migrating hugetlb folio:
> 
> BUG: kernel NULL pointer dereference, address: 0000000000000008
> PGD 0 P4D 0
> Oops: Oops: 0002 [#1] PREEMPT SMP NOPTI
> CPU: 0 PID: 3435 Comm: bash Not tainted 6.10.0-rc6-00450-g8578ca01f21f #66
> RIP: 0010:__folio_undo_large_rmappable+0x70/0xb0
> RSP: 0018:ffffb165c98a7b38 EFLAGS: 00000097
> RAX: fffffbbc44528090 RBX: 0000000000000000 RCX: 0000000000000000
> RDX: ffffa30e000a2800 RSI: 0000000000000246 RDI: ffffa3153ffffcc0
> RBP: fffffbbc44528000 R08: 0000000000002371 R09: ffffffffbe4e5868
> R10: 0000000000000001 R11: 0000000000000001 R12: ffffa3153ffffcc0
> R13: fffffbbc44468000 R14: 0000000000000001 R15: 0000000000000001
> FS:  00007f5b3a716740(0000) GS:ffffa3151fc00000(0000) knlGS:0000000000000000
> CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> CR2: 0000000000000008 CR3: 000000010959a000 CR4: 00000000000006f0
> Call Trace:
>  <TASK>
>  __folio_migrate_mapping+0x59e/0x950
>  __migrate_folio.constprop.0+0x5f/0x120
>  move_to_new_folio+0xfd/0x250
>  migrate_pages+0x383/0xd70
>  soft_offline_page+0x2ab/0x7f0
>  soft_offline_page_store+0x52/0x90
>  kernfs_fop_write_iter+0x12c/0x1d0
>  vfs_write+0x380/0x540
>  ksys_write+0x64/0xe0
>  do_syscall_64+0xb9/0x1d0
>  entry_SYSCALL_64_after_hwframe+0x77/0x7f
> RIP: 0033:0x7f5b3a514887
> RSP: 002b:00007ffe138fce68 EFLAGS: 00000246 ORIG_RAX: 0000000000000001
> RAX: ffffffffffffffda RBX: 000000000000000c RCX: 00007f5b3a514887
> RDX: 000000000000000c RSI: 0000556ab809ee10 RDI: 0000000000000001
> RBP: 0000556ab809ee10 R08: 00007f5b3a5d1460 R09: 000000007fffffff
> R10: 0000000000000000 R11: 0000000000000246 R12: 000000000000000c
> R13: 00007f5b3a61b780 R14: 00007f5b3a617600 R15: 00007f5b3a616a00
> 
> It's because hugetlb folio is passed to __folio_undo_large_rmappable()
> unexpectedly. large_rmappable flag is imperceptibly set to hugetlb folio
> since commit f6a8dd98a2ce ("hugetlb: convert alloc_buddy_hugetlb_folio to
> use a folio"). Then commit be9581ea8c05 ("mm: fix crashes from deferred
> split racing folio migration") makes folio_migrate_mapping() call
> folio_undo_large_rmappable() triggering the bug. Fix this issue by
> clearing large_rmappable flag for hugetlb folios. They don't need that
> flag set anyway.

Gosh, thanks a lot for catching this: it had not crossed my mind that
a folio which passes (folio_test_large and) folio_test_large_rmappable
might not be suitable for folio_undo_large_rmappable.

> 
> Fixes: f6a8dd98a2ce ("hugetlb: convert alloc_buddy_hugetlb_folio to use a folio")

That's in 6.10-rc, isn't it?

> Fixes: be9581ea8c05 ("mm: fix crashes from deferred split racing folio migration")

And that's in mm-hotfixes-stable intended for 6.10 final.

> Signed-off-by: Miaohe Lin <linmiaohe@huawei.com>
> Cc: <stable@vger.kernel.org>

So if all goes to plan, this shouldn't need the Cc stable.

I certainly deserve blame for not thinking of this possibility: but how
was it working before my commit, when the folio_undo_large_rmappable()
was being called from mem_cgroup_migrate()?  I think that was just as
liable to crash too.

I would like to hear definitively from Matthew, whether a hugetlb page
should or should not be reported as large_rmappable - is your patch here
just fixing a surprise, or in danger of adding another surprise somewhere?

Hugh

> ---
>  mm/hugetlb.c | 3 +++
>  1 file changed, 3 insertions(+)
> 
> diff --git a/mm/hugetlb.c b/mm/hugetlb.c
> index 6282dd9e37e3..45fd3bc75332 100644
> --- a/mm/hugetlb.c
> +++ b/mm/hugetlb.c
> @@ -2173,6 +2173,9 @@ static struct folio *alloc_buddy_hugetlb_folio(struct hstate *h,
>  		nid = numa_mem_id();
>  retry:
>  	folio = __folio_alloc(gfp_mask, order, nid, nmask);
> +	/* Ensure hugetlb folio won't have large_rmappable flag set. */
> +	if (folio)
> +		folio_clear_large_rmappable(folio);
>  
>  	if (folio && !folio_ref_freeze(folio, 1)) {
>  		folio_put(folio);
> -- 
> 2.33.0
Miaohe Lin July 10, 2024, 3:25 a.m. UTC | #2 
On 2024/7/10 8:14, Hugh Dickins wrote:
> On Tue, 9 Jul 2024, Miaohe Lin wrote:
> 
>> A kernel crash was observed when migrating hugetlb folio:
>>
>> BUG: kernel NULL pointer dereference, address: 0000000000000008
>> PGD 0 P4D 0
>> Oops: Oops: 0002 [#1] PREEMPT SMP NOPTI
>> CPU: 0 PID: 3435 Comm: bash Not tainted 6.10.0-rc6-00450-g8578ca01f21f #66
>> RIP: 0010:__folio_undo_large_rmappable+0x70/0xb0
>> RSP: 0018:ffffb165c98a7b38 EFLAGS: 00000097
>> RAX: fffffbbc44528090 RBX: 0000000000000000 RCX: 0000000000000000
>> RDX: ffffa30e000a2800 RSI: 0000000000000246 RDI: ffffa3153ffffcc0
>> RBP: fffffbbc44528000 R08: 0000000000002371 R09: ffffffffbe4e5868
>> R10: 0000000000000001 R11: 0000000000000001 R12: ffffa3153ffffcc0
>> R13: fffffbbc44468000 R14: 0000000000000001 R15: 0000000000000001
>> FS:  00007f5b3a716740(0000) GS:ffffa3151fc00000(0000) knlGS:0000000000000000
>> CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
>> CR2: 0000000000000008 CR3: 000000010959a000 CR4: 00000000000006f0
>> Call Trace:
>>  <TASK>
>>  __folio_migrate_mapping+0x59e/0x950
>>  __migrate_folio.constprop.0+0x5f/0x120
>>  move_to_new_folio+0xfd/0x250
>>  migrate_pages+0x383/0xd70
>>  soft_offline_page+0x2ab/0x7f0
>>  soft_offline_page_store+0x52/0x90
>>  kernfs_fop_write_iter+0x12c/0x1d0
>>  vfs_write+0x380/0x540
>>  ksys_write+0x64/0xe0
>>  do_syscall_64+0xb9/0x1d0
>>  entry_SYSCALL_64_after_hwframe+0x77/0x7f
>> RIP: 0033:0x7f5b3a514887
>> RSP: 002b:00007ffe138fce68 EFLAGS: 00000246 ORIG_RAX: 0000000000000001
>> RAX: ffffffffffffffda RBX: 000000000000000c RCX: 00007f5b3a514887
>> RDX: 000000000000000c RSI: 0000556ab809ee10 RDI: 0000000000000001
>> RBP: 0000556ab809ee10 R08: 00007f5b3a5d1460 R09: 000000007fffffff
>> R10: 0000000000000000 R11: 0000000000000246 R12: 000000000000000c
>> R13: 00007f5b3a61b780 R14: 00007f5b3a617600 R15: 00007f5b3a616a00
>>
>> It's because hugetlb folio is passed to __folio_undo_large_rmappable()
>> unexpectedly. large_rmappable flag is imperceptibly set to hugetlb folio
>> since commit f6a8dd98a2ce ("hugetlb: convert alloc_buddy_hugetlb_folio to
>> use a folio"). Then commit be9581ea8c05 ("mm: fix crashes from deferred
>> split racing folio migration") makes folio_migrate_mapping() call
>> folio_undo_large_rmappable() triggering the bug. Fix this issue by
>> clearing large_rmappable flag for hugetlb folios. They don't need that
>> flag set anyway.
> 
> Gosh, thanks a lot for catching this: it had not crossed my mind that
> a folio which passes (folio_test_large and) folio_test_large_rmappable
> might not be suitable for folio_undo_large_rmappable.
> 
>>
>> Fixes: f6a8dd98a2ce ("hugetlb: convert alloc_buddy_hugetlb_folio to use a folio")
> 
> That's in 6.10-rc, isn't it?
> 
>> Fixes: be9581ea8c05 ("mm: fix crashes from deferred split racing folio migration")
> 
> And that's in mm-hotfixes-stable intended for 6.10 final.
> 
>> Signed-off-by: Miaohe Lin <linmiaohe@huawei.com>
>> Cc: <stable@vger.kernel.org>
> 
> So if all goes to plan, this shouldn't need the Cc stable.

I think you are right. Cc stable should be removed. But this patch has been merged into
mm-hotfixes-stable branch, so might Andrew can kindly help modify this?

> 
> I certainly deserve blame for not thinking of this possibility: but how
> was it working before my commit, when the folio_undo_large_rmappable()
> was being called from mem_cgroup_migrate()?  I think that was just as
> liable to crash too.

I reproduced the crash today with commit be9581ea8c05 ("mm: fix crashes from deferred split racing folio migration") reverted:

BUG: kernel NULL pointer dereference, address: 0000000000000008
#PF: supervisor write access in kernel mode
#PF: error_code(0x0002) - not-present page
PGD 0 P4D 0
Oops: 0002 [#1] PREEMPT SMP NOPTI
CPU: 11 PID: 1481 Comm: bash Not tainted 6.9.0-rc4-00265-gf6a8dd98a2ce-dirty #76
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.14.0-0-g155821a1990b-prebuilt.qemu.org 04/01/2014
RIP: 0010:folio_undo_large_rmappable+0xa0/0xe0
RSP: 0018:ffffa4104950fbd0 EFLAGS: 00000097
RAX: ffff9e06001c8800 RBX: ffffc8f6614f0090 RCX: 0000000000000001
RDX: 0000000000000000 RSI: 0000000000000286 RDI: ffff9e060b23db98
RBP: ffffc8f6614f0000 R08: 0000000000002453 R09: ffffffffaead2a48
R10: 0000000000000000 R11: 0000000000000001 R12: ffff9e060b23db98
R13: 0000000000000000 R14: 0000000000000000 R15: ffffa4104950fcc8
FS:  00007f6821eb0740(0000) GS:ffff9e0ddfcc0000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 0000000000000008 CR3: 0000000884502000 CR4: 00000000000006f0
Call Trace:
 <TASK>
 mem_cgroup_migrate+0x186/0x1d0
 migrate_folio_extra+0x5c/0x90
 move_to_new_folio+0xff/0x250
 migrate_pages+0x702/0xd20
 soft_offline_page+0x29b/0x7a0
 soft_offline_page_store+0x52/0x90
 kernfs_fop_write_iter+0x12c/0x1d0
 vfs_write+0x387/0x550
 ksys_write+0x64/0xe0
 do_syscall_64+0xc2/0x1d0
 entry_SYSCALL_64_after_hwframe+0x77/0x7f
RIP: 0033:0x7f6821d14887
RSP: 002b:00007ffeecdb19a8 EFLAGS: 00000246 ORIG_RAX: 0000000000000001
RAX: ffffffffffffffda RBX: 000000000000000c RCX: 00007f6821d14887
RDX: 000000000000000c RSI: 0000563cfd799e10 RDI: 0000000000000001
RBP: 0000563cfd799e10 R08: 00007f6821dd1460 R09: 000000007fffffff
R10: 0000000000000000 R11: 0000000000000246 R12: 000000000000000c
R13: 00007f6821e1b780 R14: 00007f6821e17600 R15: 00007f6821e16a00

This requires memory_hugetlb_accounting is enabled on cgroup2. Or folio_memcg of hugetlb folio will be NULL
and thus simply return before calling folio_undo_large_rmappable() in mem_cgroup_migrate().
memory_hugetlb_accounting isnot enable in my test env, so I didn't trigger this bug earlier.

So Fixes: be9581ea8c05 ("mm: fix crashes from deferred split racing folio migration") tag might also be removed?

> 
> I would like to hear definitively from Matthew, whether a hugetlb page
> should or should not be reported as large_rmappable - is your patch here
> just fixing a surprise, or in danger of adding another surprise somewhere?

IIUC, large_rmappable is only used for thp. See below code:

static inline bool is_transparent_hugepage(const struct folio *folio)
{
	if (!folio_test_large(folio))
		return false;

	return is_huge_zero_folio(folio) ||
		folio_test_large_rmappable(folio);
}

But I might be miss something.
Thanks.
.
diff mbox series

Patch

diff --git a/mm/hugetlb.c b/mm/hugetlb.c
index 6282dd9e37e3..45fd3bc75332 100644
--- a/mm/hugetlb.c
+++ b/mm/hugetlb.c
@@ -2173,6 +2173,9 @@  static struct folio *alloc_buddy_hugetlb_folio(struct hstate *h,
 		nid = numa_mem_id();
 retry:
 	folio = __folio_alloc(gfp_mask, order, nid, nmask);
+	/* Ensure hugetlb folio won't have large_rmappable flag set. */
+	if (folio)
+		folio_clear_large_rmappable(folio);
 
 	if (folio && !folio_ref_freeze(folio, 1)) {
 		folio_put(folio);