[Kernel Bug] KASAN: slab-use-after-free Read in __blkcg_rstat_flush

[李龙兴]

Dear Linux kernel developers and maintainers,

We would like to report a new kernel bug found by our tool. KASAN:
slab-use-after-free Read in __blkcg_rstat_flush. Details are as
follows.

Kernel commit: v6.18.2
Kernel config: see attachment
report: see attachment

We are currently analyzing the root cause and working on a
reproducible PoC. We will provide further updates in this thread as
soon as we have more information.

Best regards,
Longxing Li

==================================================================
BUG: KASAN: slab-use-after-free in
__blkcg_rstat_flush.isra.0+0x73c/0x800 block/blk-cgroup.c:1069
Read of size 8 at addr ffff88810a8ba830 by task pool_workqueue_/3

CPU: 1 UID: 0 PID: 3 Comm: pool_workqueue_ Not tainted 6.18.2 #1 PREEMPT(full)
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014
Call Trace:
<IRQ>
__dump_stack lib/dump_stack.c:94 [inline]
dump_stack_lvl+0x116/0x1f0 lib/dump_stack.c:120
print_address_description mm/kasan/report.c:378 [inline]
print_report+0xcd/0x630 mm/kasan/report.c:482
kasan_report+0xe0/0x110 mm/kasan/report.c:595
__blkcg_rstat_flush.isra.0+0x73c/0x800 block/blk-cgroup.c:1069
__blkg_release+0x1a6/0x2d0 block/blk-cgroup.c:179
rcu_do_batch kernel/rcu/tree.c:2605 [inline]
rcu_core+0x79c/0x1530 kernel/rcu/tree.c:2861
handle_softirqs+0x219/0x8e0 kernel/softirq.c:622
__do_softirq kernel/softirq.c:656 [inline]
invoke_softirq kernel/softirq.c:496 [inline]
__irq_exit_rcu+0x109/0x170 kernel/softirq.c:723
irq_exit_rcu+0x9/0x30 kernel/softirq.c:739
instr_sysvec_apic_timer_interrupt arch/x86/kernel/apic/apic.c:1052 [inline]
sysvec_apic_timer_interrupt+0xa4/0xc0 arch/x86/kernel/apic/apic.c:1052
</IRQ>
<TASK>
asm_sysvec_apic_timer_interrupt+0x1a/0x20 arch/x86/include/asm/idtentry.h:697
RIP: 0010:lockdep_unregister_key+0xe9/0x140 kernel/locking/lockdep.c:6616
Code: 48 89 ef e8 b9 e5 ff ff 48 83 2d e9 3d 04 14 01 89 c3 e8 da ee
ff ff 9c 58 f6 c4 02 75 52 41 f7 c4 00 02 00 00 74 01 fb 84 db <75> 1b
5b 5d 41 5c e9 4c 79 0a 00 8b 05 6a c5 c7 0e 31 db 85 c0 74
RSP: 0000:ffffc9000005fd88 EFLAGS: 00000246
RAX: 0000000000000046 RBX: 0000000000000000 RCX: 0000000000000001
RDX: 0000000000000000 RSI: ffffffff8d91cc66 RDI: ffffffff8bd07340
RBP: ffffffff97038a88 R08: 000000000002f4fa R09: ffffffff95a27414
R10: 0000000000000000 R11: 0000000000000001 R12: 0000000000000246
R13: ffff8880204ad400 R14: ffff8880204ad408 R15: ffff88802c98b001
wq_unregister_lockdep kernel/workqueue.c:4848 [inline]
pwq_release_workfn+0x5e6/0xa70 kernel/workqueue.c:5144
kthread_worker_fn+0x310/0xc50 kernel/kthread.c:1010
kthread+0x3c5/0x780 kernel/kthread.c:463
ret_from_fork+0x675/0x7d0 arch/x86/kernel/process.c:158
ret_from_fork_asm+0x1a/0x30 arch/x86/entry/entry_64.S:245
</TASK>

Allocated by task 12789:
kasan_save_stack+0x33/0x60 mm/kasan/common.c:56
kasan_save_track+0x14/0x30 mm/kasan/common.c:77
poison_kmalloc_redzone mm/kasan/common.c:400 [inline]
__kasan_kmalloc+0xaa/0xb0 mm/kasan/common.c:417
kmalloc_node_noprof include/linux/slab.h:983 [inline]
blkg_alloc+0xb7/0xb60 block/blk-cgroup.c:305
blkg_create+0xb66/0x1400 block/blk-cgroup.c:391
blkg_lookup_create block/blk-cgroup.c:508 [inline]
blkg_tryget_closest block/blk-cgroup.c:2096 [inline]
bio_associate_blkg_from_css+0xcf3/0x13e0 block/blk-cgroup.c:2130
bio_associate_blkg block/blk-cgroup.c:2161 [inline]
bio_associate_blkg+0x10c/0x2a0 block/blk-cgroup.c:2147
bio_init+0x2dd/0x5e0 block/bio.c:266
bio_init_inline include/linux/bio.h:411 [inline]
bio_alloc_bioset+0x2c4/0x8d0 block/bio.c:584
bio_alloc include/linux/bio.h:372 [inline]
do_mpage_readpage+0xd39/0x18d0 fs/mpage.c:287
mpage_readahead+0x321/0x5a0 fs/mpage.c:371
read_pages+0x1c4/0xc70 mm/readahead.c:163
page_cache_ra_unbounded+0x4b9/0xa10 mm/readahead.c:302
do_page_cache_ra mm/readahead.c:332 [inline]
force_page_cache_ra+0x246/0x340 mm/readahead.c:361
page_cache_sync_ra+0x20c/0xbf0 mm/readahead.c:579
filemap_get_pages+0x717/0x1d40 mm/filemap.c:2638
filemap_read+0x3d4/0xe60 mm/filemap.c:2748
blkdev_read_iter+0x1ac/0x500 block/fops.c:859
new_sync_read fs/read_write.c:491 [inline]
vfs_read+0x8bf/0xcf0 fs/read_write.c:572
ksys_read+0x12a/0x250 fs/read_write.c:715
do_syscall_x64 arch/x86/entry/syscall_64.c:63 [inline]
do_syscall_64+0xcd/0xfa0 arch/x86/entry/syscall_64.c:94
entry_SYSCALL_64_after_hwframe+0x77/0x7f

Freed by task 9843:
kasan_save_stack+0x33/0x60 mm/kasan/common.c:56
kasan_save_track+0x14/0x30 mm/kasan/common.c:77
__kasan_save_free_info+0x3b/0x60 mm/kasan/generic.c:587
kasan_save_free_info mm/kasan/kasan.h:406 [inline]
poison_slab_object mm/kasan/common.c:252 [inline]
__kasan_slab_free+0x5f/0x80 mm/kasan/common.c:284
kasan_slab_free include/linux/kasan.h:234 [inline]
slab_free_hook mm/slub.c:2543 [inline]
slab_free mm/slub.c:6642 [inline]
kfree+0x2b8/0x6d0 mm/slub.c:6849
process_one_work+0x9cf/0x1b70 kernel/workqueue.c:3263
process_scheduled_works kernel/workqueue.c:3346 [inline]
worker_thread+0x6c8/0xf10 kernel/workqueue.c:3427
kthread+0x3c5/0x780 kernel/kthread.c:463
ret_from_fork+0x675/0x7d0 arch/x86/kernel/process.c:158
ret_from_fork_asm+0x1a/0x30 arch/x86/entry/entry_64.S:245

Last potentially related work creation:
kasan_save_stack+0x33/0x60 mm/kasan/common.c:56
kasan_record_aux_stack+0xa7/0xc0 mm/kasan/generic.c:559
insert_work+0x36/0x230 kernel/workqueue.c:2186
__queue_work+0x97e/0x1160 kernel/workqueue.c:2341
queue_work_on+0x1a4/0x1f0 kernel/workqueue.c:2392
blkg_free block/blk-cgroup.c:152 [inline]
__blkg_release+0x113/0x2d0 block/blk-cgroup.c:183
rcu_do_batch kernel/rcu/tree.c:2605 [inline]
rcu_core+0x79c/0x1530 kernel/rcu/tree.c:2861
handle_softirqs+0x219/0x8e0 kernel/softirq.c:622
__do_softirq kernel/softirq.c:656 [inline]
invoke_softirq kernel/softirq.c:496 [inline]
__irq_exit_rcu+0x109/0x170 kernel/softirq.c:723
irq_exit_rcu+0x9/0x30 kernel/softirq.c:739
instr_sysvec_apic_timer_interrupt arch/x86/kernel/apic/apic.c:1052 [inline]
sysvec_apic_timer_interrupt+0xa4/0xc0 arch/x86/kernel/apic/apic.c:1052
asm_sysvec_apic_timer_interrupt+0x1a/0x20 arch/x86/include/asm/idtentry.h:697

Second to last potentially related work creation:
kasan_save_stack+0x33/0x60 mm/kasan/common.c:56
kasan_record_aux_stack+0xa7/0xc0 mm/kasan/generic.c:559
__call_rcu_common.constprop.0+0xa5/0xa10 kernel/rcu/tree.c:3123
percpu_ref_put_many.constprop.0+0x26c/0x2a0 include/linux/percpu-refcount.h:335
rcu_do_batch kernel/rcu/tree.c:2605 [inline]
rcu_core+0x79c/0x1530 kernel/rcu/tree.c:2861
handle_softirqs+0x219/0x8e0 kernel/softirq.c:622
do_softirq kernel/softirq.c:523 [inline]
do_softirq+0xb2/0xf0 kernel/softirq.c:510
__local_bh_enable_ip+0x100/0x120 kernel/softirq.c:450
spin_unlock_bh include/linux/spinlock.h:396 [inline]
nsim_dev_trap_report drivers/net/netdevsim/dev.c:835 [inline]
nsim_dev_trap_report_work+0x8b5/0xcf0 drivers/net/netdevsim/dev.c:866
process_one_work+0x9cf/0x1b70 kernel/workqueue.c:3263
process_scheduled_works kernel/workqueue.c:3346 [inline]
worker_thread+0x6c8/0xf10 kernel/workqueue.c:3427
kthread+0x3c5/0x780 kernel/kthread.c:463
ret_from_fork+0x675/0x7d0 arch/x86/kernel/process.c:158
ret_from_fork_asm+0x1a/0x30 arch/x86/entry/entry_64.S:245

The buggy address belongs to the object at ffff88810a8ba800
which belongs to the cache kmalloc-512 of size 512
The buggy address is located 48 bytes inside of
freed 512-byte region [ffff88810a8ba800, ffff88810a8baa00)

The buggy address belongs to the physical page:
page: refcount:0 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0x10a8b8
head: order:2 mapcount:0 entire_mapcount:0 nr_pages_mapped:0 pincount:0
anon flags: 0x57ff00000000040(head|node=1|zone=2|lastcpupid=0x7ff)
page_type: f5(slab)
raw: 057ff00000000040 ffff88801b041c80 0000000000000000 dead000000000001
raw: 0000000000000000 0000000000100010 00000000f5000000 0000000000000000
head: 057ff00000000040 ffff88801b041c80 0000000000000000 dead000000000001
head: 0000000000000000 0000000000100010 00000000f5000000 0000000000000000
head: 057ff00000000002 ffffea00042a2e01 00000000ffffffff 00000000ffffffff
head: ffffffffffffffff 0000000000000000 00000000ffffffff 0000000000000004
page dumped because: kasan: bad access detected
page_owner tracks the page as allocated
page last allocated via order 2, migratetype Unmovable, gfp_mask
0xd20c0(__GFP_IO|__GFP_FS|__GFP_NOWARN|__GFP_NORETRY|__GFP_COMP|__GFP_NOMEMALLOC),
pid 9875, tgid 9875 (syz-executor.5), ts 476252151009, free_ts
476208805854
set_page_owner include/linux/page_owner.h:32 [inline]
post_alloc_hook+0x1af/0x220 mm/page_alloc.c:1845
prep_new_page mm/page_alloc.c:1853 [inline]
get_page_from_freelist+0x10a3/0x3a30 mm/page_alloc.c:3879
__alloc_frozen_pages_noprof+0x25f/0x2470 mm/page_alloc.c:5178
alloc_pages_mpol+0x1fb/0x550 mm/mempolicy.c:2416
alloc_slab_page mm/slub.c:3059 [inline]
allocate_slab mm/slub.c:3232 [inline]
new_slab+0x24a/0x360 mm/slub.c:3286
___slab_alloc+0xd79/0x1a50 mm/slub.c:4655
__slab_alloc.constprop.0+0x63/0x110 mm/slub.c:4778
__slab_alloc_node mm/slub.c:4854 [inline]
slab_alloc_node mm/slub.c:5276 [inline]
__kmalloc_cache_noprof+0x477/0x780 mm/slub.c:5766
kmalloc_noprof include/linux/slab.h:957 [inline]
kzalloc_noprof include/linux/slab.h:1094 [inline]
binderfs_binder_device_create.isra.0+0x189/0xb10 drivers/android/binderfs.c:148
binderfs_fill_super+0x8d4/0x13a0 drivers/android/binderfs.c:720
vfs_get_super fs/super.c:1331 [inline]
get_tree_nodev+0xdd/0x190 fs/super.c:1350
vfs_get_tree+0x8e/0x340 fs/super.c:1758
fc_mount fs/namespace.c:1199 [inline]
do_new_mount_fc fs/namespace.c:3642 [inline]
do_new_mount fs/namespace.c:3718 [inline]
path_mount+0x7b9/0x23a0 fs/namespace.c:4028
do_mount fs/namespace.c:4041 [inline]
__do_sys_mount fs/namespace.c:4229 [inline]
__se_sys_mount fs/namespace.c:4206 [inline]
__x64_sys_mount+0x293/0x310 fs/namespace.c:4206
do_syscall_x64 arch/x86/entry/syscall_64.c:63 [inline]
do_syscall_64+0xcd/0xfa0 arch/x86/entry/syscall_64.c:94
entry_SYSCALL_64_after_hwframe+0x77/0x7f
page last free pid 12738 tgid 12736 stack trace:
reset_page_owner include/linux/page_owner.h:25 [inline]
free_pages_prepare mm/page_alloc.c:1394 [inline]
__free_frozen_pages+0x7ce/0x1160 mm/page_alloc.c:2901
stack_depot_save_flags+0x352/0x9c0 lib/stackdepot.c:727
kasan_save_stack+0x42/0x60 mm/kasan/common.c:57
kasan_save_track+0x14/0x30 mm/kasan/common.c:77
unpoison_slab_object mm/kasan/common.c:342 [inline]
__kasan_slab_alloc+0x89/0x90 mm/kasan/common.c:368
kasan_slab_alloc include/linux/kasan.h:252 [inline]
slab_post_alloc_hook mm/slub.c:4978 [inline]
slab_alloc_node mm/slub.c:5288 [inline]
kmem_cache_alloc_lru_noprof+0x254/0x6e0 mm/slub.c:5307
alloc_inode+0x64/0x240 fs/inode.c:346
new_inode+0x22/0x1c0 fs/inode.c:1145
debugfs_get_inode fs/debugfs/inode.c:72 [inline]
__debugfs_create_file+0x11c/0x6b0 fs/debugfs/inode.c:442
debugfs_create_file_full+0x41/0x60 fs/debugfs/inode.c:469
bdi_debug_register mm/backing-dev.c:239 [inline]
bdi_register_va+0x343/0x820 mm/backing-dev.c:1106
super_setup_bdi_name+0xff/0x250 fs/super.c:1817
btrfs_fill_super fs/btrfs/super.c:981 [inline]
btrfs_get_tree_super fs/btrfs/super.c:1951 [inline]
btrfs_get_tree_subvol fs/btrfs/super.c:2094 [inline]
btrfs_get_tree+0x12bd/0x2710 fs/btrfs/super.c:2128
vfs_get_tree+0x8e/0x340 fs/super.c:1758
fc_mount fs/namespace.c:1199 [inline]
do_new_mount_fc fs/namespace.c:3642 [inline]
do_new_mount fs/namespace.c:3718 [inline]
path_mount+0x7b9/0x23a0 fs/namespace.c:4028
do_mount fs/namespace.c:4041 [inline]
__do_sys_mount fs/namespace.c:4229 [inline]
__se_sys_mount fs/namespace.c:4206 [inline]
__x64_sys_mount+0x293/0x310 fs/namespace.c:4206

Memory state around the buggy address:
ffff88810a8ba700: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
ffff88810a8ba780: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
>ffff88810a8ba800: fa fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
^
ffff88810a8ba880: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
ffff88810a8ba900: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
==================================================================
----------------
Code disassembly (best guess):
0: 48 89 ef mov %rbp,%rdi
3: e8 b9 e5 ff ff call 0xffffe5c1
8: 48 83 2d e9 3d 04 14 subq $0x1,0x14043de9(%rip) # 0x14043df9
f: 01
10: 89 c3 mov %eax,%ebx
12: e8 da ee ff ff call 0xffffeef1
17: 9c pushf
18: 58 pop %rax
19: f6 c4 02 test $0x2,%ah
1c: 75 52 jne 0x70
1e: 41 f7 c4 00 02 00 00 test $0x200,%r12d
25: 74 01 je 0x28
27: fb sti
28: 84 db test %bl,%bl
* 2a: 75 1b jne 0x47 <-- trapping instruction
2c: 5b pop %rbx
2d: 5d pop %rbp
2e: 41 5c pop %r12
30: e9 4c 79 0a 00 jmp 0xa7981
35: 8b 05 6a c5 c7 0e mov 0xec7c56a(%rip),%eax # 0xec7c5a5
3b: 31 db xor %ebx,%ebx
3d: 85 c0 test %eax,%eax
3f: 74 .byte 0x74

https://drive.google.com/file/d/1ZjRNEOf0XFYtFl5UZ7dvpJpzmPhje8it/view?usp=drive_link

https://drive.google.com/file/d/1f695Y7Vob314irZfdduD4wnPJ51VzXrU/view?usp=drive_link

[Ming Lei]

Hello,

Can you try the following patch?


diff --git a/block/blk-cgroup.c b/block/blk-cgroup.c
index 3cffb68ba5d8..dc0cccfdca68 100644
--- a/block/blk-cgroup.c
+++ b/block/blk-cgroup.c
@@ -160,6 +160,20 @@ static void blkg_free(struct blkcg_gq *blkg)
schedule_work(&blkg->free_work);
}

+/*
+ * RCU callback to free blkg after an additional grace period.
+ * This ensures any concurrent __blkcg_rstat_flush() that might have
+ * removed our iostat entries via llist_del_all() has completed.
+ */
+static void __blkg_release_free_rcu(struct rcu_head *rcu)
+{
+ struct blkcg_gq *blkg = container_of(rcu, struct blkcg_gq, rcu_head);
+
+ /* release the blkcg and parent blkg refs this blkg has been holding */
+ css_put(&blkg->blkcg->css);
+ blkg_free(blkg);
+}
+
static void __blkg_release(struct rcu_head *rcu)
{
struct blkcg_gq *blkg = container_of(rcu, struct blkcg_gq, rcu_head);
@@ -178,9 +192,14 @@ static void __blkg_release(struct rcu_head *rcu)
for_each_possible_cpu(cpu)
__blkcg_rstat_flush(blkcg, cpu);

- /* release the blkcg and parent blkg refs this blkg has been holding */
- css_put(&blkg->blkcg->css);
- blkg_free(blkg);
+ /*
+ * Defer freeing via another call_rcu() to ensure any concurrent
+ * __blkcg_rstat_flush() (under rcu_read_lock) that might have removed
+ * our iostat entries via llist_del_all() has completed its iteration.
+ * The second grace period guarantees those RCU read-side critical
+ * sections have finished.
+ */
+ call_rcu(&blkg->rcu_head, __blkg_release_free_rcu);
}

/*


thanks,
Ming

[Michal Koutný]

Hello.

On Tue, Feb 03, 2026 at 11:03:01AM +0800, Ming Lei <ming...@redhat.com> wrote:
> Can you try the following patch?

I think it'd work thanks to the rcu_read_lock() in
__blkcg_rstat_flush(). However, the chaining of RCU callbacks makes
predictability of the release path less deterministic and may be
unnecessary.

What about this:

index 3cffb68ba5d87..e2f51e3bf04ef 100644
--- a/tmp/b.c
+++ b/tmp/a.c
@@ -1081,6 +1081,7 @@ static void __blkcg_rstat_flush(struct blkcg *blkcg, int cpu)
smp_mb();

WRITE_ONCE(bisc->lqueued, false);
+ blkg_put(blkg);
if (bisc == &blkg->iostat)
goto propagate_up; /* propagate up to parent only */

@@ -2220,8 +2221,10 @@ void blk_cgroup_bio_start(struct bio *bio)
if (!READ_ONCE(bis->lqueued)) {
struct llist_head *lhead = this_cpu_ptr(blkcg->lhead);

+ blkg_get(bio->bi_blkg);
llist_add(&bis->lnode, lhead);
WRITE_ONCE(bis->lqueued, true);
+
}

u64_stats_update_end_irqrestore(&bis->sync, flags);



(If only I remembered whether a reference taken from blkcg->lhead causes
reference cycle...)

Michal

[Ming Lei]

On Tue, Feb 03, 2026 at 11:54:34AM +0100, Michal Koutný wrote:
> Hello.
>
> On Tue, Feb 03, 2026 at 11:03:01AM +0800, Ming Lei <ming...@redhat.com> wrote:
> > Can you try the following patch?
>
> I think it'd work thanks to the rcu_read_lock() in
> __blkcg_rstat_flush(). However, the chaining of RCU callbacks makes
> predictability of the release path less deterministic and may be
> unnecessary.

RCU supports this way, here is just 2-stage RCU chain, and everything
is deterministic.

>
> What about this:
>
> index 3cffb68ba5d87..e2f51e3bf04ef 100644
> --- a/tmp/b.c
> +++ b/tmp/a.c
> @@ -1081,6 +1081,7 @@ static void __blkcg_rstat_flush(struct blkcg *blkcg, int cpu)
> smp_mb();
>
> WRITE_ONCE(bisc->lqueued, false);
> + blkg_put(blkg);
> if (bisc == &blkg->iostat)
> goto propagate_up; /* propagate up to parent only */
>
> @@ -2220,8 +2221,10 @@ void blk_cgroup_bio_start(struct bio *bio)
> if (!READ_ONCE(bis->lqueued)) {
> struct llist_head *lhead = this_cpu_ptr(blkcg->lhead);
>
> + blkg_get(bio->bi_blkg);
> llist_add(&bis->lnode, lhead);
> WRITE_ONCE(bis->lqueued, true);
> +

I thought about this way, but ->lqueued is lockless, and in theory the `blkg_iostat_set`
can be added again after WRITE_ONCE(bisc->lqueued, false) happens, so this way looks
fragile.


Thanks,
Ming

[Michal Koutný]

On Tue, Feb 03, 2026 at 07:11:11PM +0800, Ming Lei <ming...@redhat.com> wrote:
> RCU supports this way, here is just 2-stage RCU chain, and everything
> is deterministic.

The time when RCU callback runs is noisy, moreover when chained after
each other.
(I don't mean it doesn't work but it's debugging/testing nuisance. And
it also looks awkward.)

> I thought about this way, but ->lqueued is lockless, and in theory the `blkg_iostat_set`
> can be added again after WRITE_ONCE(bisc->lqueued, false) happens, so this way looks
> fragile.

Right, I brushed up on the cycles from the commit 20cb1c2fb7568
("blk-cgroup: Flush stats before releasing blkcg_gq") and it'd be a step
back.

Does anything prevent doing the each-cpu flush in blkg_release() (before
__blkg_release())?

Regards,
Michal

[Ming Lei]

On Tue, Feb 03, 2026 at 01:53:40PM +0100, Michal Koutný wrote:
> On Tue, Feb 03, 2026 at 07:11:11PM +0800, Ming Lei <ming...@redhat.com> wrote:
> > RCU supports this way, here is just 2-stage RCU chain, and everything
> > is deterministic.
>
> The time when RCU callback runs is noisy, moreover when chained after
> each other.
> (I don't mean it doesn't work but it's debugging/testing nuisance. And
> it also looks awkward.)

IMO it is one correct & simple fix for this complicated race.

>
> > I thought about this way, but ->lqueued is lockless, and in theory the `blkg_iostat_set`
> > can be added again after WRITE_ONCE(bisc->lqueued, false) happens, so this way looks
> > fragile.
>
> Right, I brushed up on the cycles from the commit 20cb1c2fb7568
> ("blk-cgroup: Flush stats before releasing blkcg_gq") and it'd be a step
> back.
>
> Does anything prevent doing the each-cpu flush in blkg_release() (before
> __blkg_release())?

I can't parse your question, here blkg_release() simply needs to flush
all stats. Why do you talk about preventing new flush? why is it related
with this UAF?


Thanks,
Ming

[Michal Koutný]

On Tue, Feb 03, 2026 at 10:08:07PM +0800, Ming Lei <ming...@redhat.com> wrote:
> I can't parse your question, here blkg_release() simply needs to flush
> all stats. Why do you talk about preventing new flush? why is it related
> with this UAF?

What prevents this fix:

--- a/block/blk-cgroup.c
+++ b/block/blk-cgroup.c
@@ -169,14 +169,6 @@ static void __blkg_release(struct rcu_head *rcu)
#ifdef CONFIG_BLK_CGROUP_PUNT_BIO
WARN_ON(!bio_list_empty(&blkg->async_bios));
#endif
- /*
- * Flush all the non-empty percpu lockless lists before releasing
- * us, given these stat belongs to us.
- *
- * blkg_stat_lock is for serializing blkg stat update
- */
- for_each_possible_cpu(cpu)
- __blkcg_rstat_flush(blkcg, cpu);

/* release the blkcg and parent blkg refs this blkg has been holding */

css_put(&blkg->blkcg->css);
@@ -195,6 +187,15 @@ static void blkg_release(struct percpu_ref *ref)
{
struct blkcg_gq *blkg = container_of(ref, struct blkcg_gq, refcnt);

+ /*
+ * Flush all the non-empty percpu lockless lists before releasing
+ * us, given these stat belongs to us.
+ *
+ * blkg_stat_lock is for serializing blkg stat update
+ */
+ for_each_possible_cpu(cpu)
+ __blkcg_rstat_flush(blkcg, cpu);
+
call_rcu(&blkg->rcu_head, __blkg_release);
}

[Ming Lei]

This one looks more clever, can you send one formal patch?


Thanks,
Ming

[Jens Axboe]

I'll drop the previous one for now - I didn't really mind it,
but if we can fix it without, then yeah agree that's better.

--
Jens Axboe