On 2023-03-30 15:42, Andrew Morton wrote:
On Thu, 30 Mar 2023 09:38:22 -0400 Mathieu Desnoyers <mathieu.desnoyers@xxxxxxxxxxxx> wrote:
commit f1a7941243c1 ("mm: convert mm's rss stats into percpu_counter")
introduces a memory leak by missing a call to destroy_context() when a
percpu_counter fails to allocate.
Before introducing the per-cpu counter allocations, init_new_context()
was the last call that could fail in mm_init(), and thus there was no
need to ever invoke destroy_context() in the error paths. Adding the
following percpu counter allocations adds error paths after
init_new_context(), which means its associated destroy_context() needs
to be called when percpu counters fail to allocate.
...
--- a/kernel/fork.c
+++ b/kernel/fork.c
@@ -1171,6 +1171,7 @@ static struct mm_struct *mm_init(struct mm_struct *mm, struct task_struct *p,
fail_pcpu:
while (i > 0)
percpu_counter_destroy(&mm->rss_stat[--i]);
+ destroy_context(mm);
fail_nocontext:
mm_free_pgd(mm);
fail_nopgd:
Is there really a leak? I wasn't able to find a version of
init_new_context() which performs allocation.
AFAIU, at least on powerpc:
arch/powerpc/mm/book3s64/mmu_context.c: init_new_context() calls
radix__init_new_context() or hash__init_new_context() which
leak IDs through ida_alloc_range.
Thanks,
Mathieu
--
Mathieu Desnoyers
EfficiOS Inc.
https://www.efficios.com
