Re: [PATCH bpf] bpf/helpers: Skip memcg accounting in __bpf_async_init()

[Peilin Ye <yepeilin@xxxxxxxxxx>]

On Fri, Sep 05, 2025 at 10:31:07AM -0700, Shakeel Butt wrote:
> On Fri, Sep 05, 2025 at 08:18:25AM -0700, Alexei Starovoitov wrote:
> > On Thu, Sep 4, 2025 at 11:20 PM Peilin Ye <yepeilin@xxxxxxxxxx> wrote:
> > > As pointed out by Kumar, we can use bpf_mem_alloc() and friends for
> > > bpf_hrtimer and bpf_work, to skip memcg accounting.
> > 
> > This is a short term workaround that we shouldn't take.
> > Long term bpf_mem_alloc() will use kmalloc_nolock() and
> > memcg accounting that was already made to work from any context
> > except that the path of memcg_memory_event() wasn't converted.
> > 
> > Shakeel,
> > 
> > Any suggestions how memcg_memory_event()->cgroup_file_notify()
> > can be fixed?
> > Can we just trylock and skip the event?
> 
> Will !gfpflags_allow_spinning(gfp_mask) be able to detect such call
> chains? If yes, then we can change memcg_memory_event() to skip calls to
> cgroup_file_notify() if spinning is not allowed.

I tried the below diff, but unfortunately __bpf_async_init() calls
bpf_map_kmalloc_node() with GFP_ATOMIC, so gfpflags_allow_spinning()
would return true.  I'll try the trylock-and-skip approach.

Thanks,
Peilin Ye

--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -2350,11 +2350,12 @@ static int try_charge_memcg(struct mem_cgroup *memcg, gfp_t gfp_mask,
        if (unlikely(task_in_memcg_oom(current)))
                goto nomem;
 
-       if (!gfpflags_allow_blocking(gfp_mask))
+       if (!gfpflags_allow_blocking(gfp_mask)) {
                goto nomem;
-
-       memcg_memory_event(mem_over_limit, MEMCG_MAX);
-       raised_max_event = true;
+       } else if (gfpflags_allow_spinning(gfp_mask)) {
+               memcg_memory_event(mem_over_limit, MEMCG_MAX);
+               raised_max_event = true;
+       }
 
        psi_memstall_enter(&pflags);
        nr_reclaimed = try_to_free_mem_cgroup_pages(mem_over_limit, nr_pages,
@@ -2419,7 +2420,7 @@ static int try_charge_memcg(struct mem_cgroup *memcg, gfp_t gfp_mask,
         * If the allocation has to be enforced, don't forget to raise
         * a MEMCG_MAX event.
         */
-       if (!raised_max_event)
+       if (!raised_max_event && gfpflags_allow_spinning(gfp_mask))
                memcg_memory_event(mem_over_limit, MEMCG_MAX);
 
        /*