Skip to contentSkip to conversation

mm/hugetlb_cgroup: call page_counter_set_max() outside VM_BUG_ON()

PATCH Applied Narek Jilavyan 3 participants 3 messages Aug 17

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 mm/hugetlb_cgroup: call page_counter_set_max() outside VM_BUG_ON()

hugetlb_cgroup_css_alloc() rounds the counter limit down to a multiple of the huge page size and then applies it inside an assertion:

	VM_BUG_ON(page_counter_set_max(fault, limit));
	VM_BUG_ON(page_counter_set_max(rsvd, limit));

With CONFIG_DEBUG_VM=n, VM_BUG_ON(cond) is BUILD_BUG_ON_INVALID(cond), i.e. ((void)(sizeof((__force long)(cond)))), whose operand is never evaluated. page_counter_set_max() is not a predicate - it performs xchg(&counter->max, nr_pages) - so on every non-debug kernel the limit is never applied and the counters keep page_counter_init()'s PAGE_COUNTER_MAX.

That is user-visible, because hugetlb_cgroup_read_u64_max() recomputes the same rounded value and uses equality as its "unlimited" sentinel. PAGE_COUNTER_MAX is LONG_MAX / PAGE_SIZE = 2251799813685247, which is odd, so round_down() really does change it and the two sides disagree. With CONFIG_DEBUG_VM=n:

	$ cat /sys/fs/cgroup/t/hugetlb.2MB.max
	9223372036854771712

and with this patch:

	$ cat /sys/fs/cgroup/t/hugetlb.2MB.max
	max

A debug option should not change cgroup output.

Call the function, then assert the result, as v6.12 did. Use VM_WARN_ON_ONCE() rather than restoring VM_BUG_ON(): the two are identical under CONFIG_DEBUG_VM=n, and checkpatch asks that new code not use BUG() variants.

Fixes 0e2759afcaf9 ("page_counter: track failcnt only for legacy cgroups")
---
 mm/hugetlb_cgroup.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)
mm/hugetlb_cgroup.c +5 -2
index e0083de1ca..ecb6e0b781 100644
@@ -97,6 +97,7 @@ static void hugetlb_cgroup_init(struct hugetlb_cgroup *h_cgroup,
 		struct page_counter *fault, *fault_parent = NULL;
 		struct page_counter *rsvd, *rsvd_parent = NULL;
 		unsigned long limit;
+		int ret;
 
 		if (parent_h_cgroup) {
 			fault_parent = hugetlb_cgroup_counter_from_cgroup(
@@ -118,8 +119,10 @@ static void hugetlb_cgroup_init(struct hugetlb_cgroup *h_cgroup,
 		limit = round_down(PAGE_COUNTER_MAX,
 				   pages_per_huge_page(&hstates[idx]));
 
-		VM_BUG_ON(page_counter_set_max(fault, limit));
-		VM_BUG_ON(page_counter_set_max(rsvd, limit));
+		ret = page_counter_set_max(fault, limit);
+		VM_WARN_ON_ONCE(ret);
+		ret = page_counter_set_max(rsvd, limit);
+		VM_WARN_ON_ONCE(ret);
 	}
 }
 
Signature · 1 line
2.43.0
maintainer Reviewed-by on v1 1/1
#1 lore raw
Quoted 33 lines · Narek Jilavyan

hugetlb_cgroup_css_alloc() rounds the counter limit down to a multiple of the huge page size and then applies it inside an assertion:

VM_BUG_ON(page_counter_set_max(fault, limit));
VM_BUG_ON(page_counter_set_max(rsvd, limit));

With CONFIG_DEBUG_VM=n, VM_BUG_ON(cond) is BUILD_BUG_ON_INVALID(cond), i.e. ((void)(sizeof((__force long)(cond)))), whose operand is never evaluated. page_counter_set_max() is not a predicate - it performs xchg(&counter->max, nr_pages) - so on every non-debug kernel the limit is never applied and the counters keep page_counter_init()'s PAGE_COUNTER_MAX.

That is user-visible, because hugetlb_cgroup_read_u64_max() recomputes the same rounded value and uses equality as its "unlimited" sentinel. PAGE_COUNTER_MAX is LONG_MAX / PAGE_SIZE = 2251799813685247, which is odd, so round_down() really does change it and the two sides disagree. With CONFIG_DEBUG_VM=n:

$ cat /sys/fs/cgroup/t/hugetlb.2MB.max
9223372036854771712

and with this patch:

$ cat /sys/fs/cgroup/t/hugetlb.2MB.max
max

A debug option should not change cgroup output.

Quoted 7 lines

Call the function, then assert the result, as v6.12 did. Use VM_WARN_ON_ONCE() rather than restoring VM_BUG_ON(): the two are identical under CONFIG_DEBUG_VM=n, and checkpatch asks that new code not use BUG() variants.

Fixes 0e2759afcaf9 ("page_counter: track failcnt only for legacy cgroups")

Thanks.

maintainer
#2 lore raw
In reply to Narek Jilavyan
Quoted 27 lines · On Mon, 17 Aug 2026 10:34:33 +0000 Narek Jilavyan <[email protected]> wrote:

hugetlb_cgroup_css_alloc() rounds the counter limit down to a multiple of the huge page size and then applies it inside an assertion:

	VM_BUG_ON(page_counter_set_max(fault, limit));
	VM_BUG_ON(page_counter_set_max(rsvd, limit));

With CONFIG_DEBUG_VM=n, VM_BUG_ON(cond) is BUILD_BUG_ON_INVALID(cond), i.e. ((void)(sizeof((__force long)(cond)))), whose operand is never evaluated. page_counter_set_max() is not a predicate - it performs xchg(&counter->max, nr_pages) - so on every non-debug kernel the limit is never applied and the counters keep page_counter_init()'s PAGE_COUNTER_MAX.

That is user-visible, because hugetlb_cgroup_read_u64_max() recomputes the same rounded value and uses equality as its "unlimited" sentinel. PAGE_COUNTER_MAX is LONG_MAX / PAGE_SIZE = 2251799813685247, which is odd, so round_down() really does change it and the two sides disagree. With CONFIG_DEBUG_VM=n:

	$ cat /sys/fs/cgroup/t/hugetlb.2MB.max
	9223372036854771712

and with this patch:

	$ cat /sys/fs/cgroup/t/hugetlb.2MB.max
	max
Quoted 6 lines

A debug option should not change cgroup output.

Call the function, then assert the result, as v6.12 did. Use VM_WARN_ON_ONCE() rather than restoring VM_BUG_ON(): the two are identical under CONFIG_DEBUG_VM=n, and checkpatch asks that new code not use BUG() variants.

Nice, thanks, I'll add cc:stable to this, to help ensure that users of earlier kernels get to enjoy it.