Re: [PATCH bpf] bpf/helpers: Use __GFP_HIGH instead of GFP_ATOMIC in __bpf_async_init()

[Shakeel Butt <shakeel.butt@linux.dev>]

On Fri, Sep 05, 2025 at 11:45:46PM +0000, Peilin Ye wrote:
> Currently, calling bpf_map_kmalloc_node() from __bpf_async_init() can
> cause various locking issues; see the following stack trace (edited for
> style) as one example:
> 
> ...
>  [10.011566]  do_raw_spin_lock.cold
>  [10.011570]  try_to_wake_up             (5) double-acquiring the same
>  [10.011575]  kick_pool                      rq_lock, causing a hardlockup
>  [10.011579]  __queue_work
>  [10.011582]  queue_work_on
>  [10.011585]  kernfs_notify
>  [10.011589]  cgroup_file_notify
>  [10.011593]  try_charge_memcg           (4) memcg accounting raises an
>  [10.011597]  obj_cgroup_charge_pages        MEMCG_MAX event
>  [10.011599]  obj_cgroup_charge_account
>  [10.011600]  __memcg_slab_post_alloc_hook
>  [10.011603]  __kmalloc_node_noprof
> ...
>  [10.011611]  bpf_map_kmalloc_node
>  [10.011612]  __bpf_async_init
>  [10.011615]  bpf_timer_init             (3) BPF calls bpf_timer_init()
>  [10.011617]  bpf_prog_xxxxxxxxxxxxxxxx_fcg_runnable
>  [10.011619]  bpf__sched_ext_ops_runnable
>  [10.011620]  enqueue_task_scx           (2) BPF runs with rq_lock held
>  [10.011622]  enqueue_task
>  [10.011626]  ttwu_do_activate
>  [10.011629]  sched_ttwu_pending         (1) grabs rq_lock
> ...
> 
> The above was reproduced on bpf-next (b338cf849ec8) by modifying
> ./tools/sched_ext/scx_flatcg.bpf.c to call bpf_timer_init() during
> ops.runnable(), and hacking [1] the memcg accounting code a bit to make
> a bpf_timer_init() call much more likely to raise an MEMCG_MAX event.
> 
> We have also run into other similar variants (both internally and on
> bpf-next), including double-acquiring cgroup_file_kn_lock, the same
> worker_pool::lock, etc.
> 
> As suggested by Shakeel, fix this by using __GFP_HIGH instead of
> GFP_ATOMIC in __bpf_async_init(), so that if try_charge_memcg() raises
> an MEMCG_MAX event, we call __memcg_memory_event() with
> @allow_spinning=false and skip calling cgroup_file_notify(), in order to
> avoid the locking issues described above.
> 
> Depends on mm patch "memcg: skip cgroup_file_notify if spinning is not
> allowed".  Tested with vmtest.sh (llvm-18, x86-64):
> 
>  $ ./test_progs -a '*timer*' -a '*wq*'
> ...
>  Summary: 7/12 PASSED, 0 SKIPPED, 0 FAILED
> 
> [1] Making bpf_timer_init() much more likely to raise an MEMCG_MAX event
> (gist-only, for brevity):
> 
> kernel/bpf/helpers.c:__bpf_async_init():
>  -        cb = bpf_map_kmalloc_node(map, size, GFP_ATOMIC, map->numa_node);
>  +        cb = bpf_map_kmalloc_node(map, size, GFP_ATOMIC | __GFP_HACK,
>  +                                  map->numa_node);
> 
> mm/memcontrol.c:try_charge_memcg():
>           if (!do_memsw_account() ||
>  -            page_counter_try_charge(&memcg->memsw, batch, &counter)) {
>  -                if (page_counter_try_charge(&memcg->memory, batch, &counter))
>  +            page_counter_try_charge_hack(&memcg->memsw, batch, &counter,
>  +                                         gfp_mask & __GFP_HACK)) {
>  +                if (page_counter_try_charge_hack(&memcg->memory, batch,
>  +                                                 &counter,
>  +                                                 gfp_mask & __GFP_HACK))
>                           goto done_restock;
> 
> mm/page_counter.c:page_counter_try_charge():
>  -bool page_counter_try_charge(struct page_counter *counter,
>  -                             unsigned long nr_pages,
>  -                             struct page_counter **fail)
>  +bool page_counter_try_charge_hack(struct page_counter *counter,
>  +                                  unsigned long nr_pages,
>  +                                  struct page_counter **fail, bool hack)
>  {
> ...
>  -                if (new > c->max) {
>  +                if (hack || new > c->max) {     // goto failed;
>                           atomic_long_sub(nr_pages, &c->usage);
> 
> Fixes: b00628b1c7d5 ("bpf: Introduce bpf timers.")
> Suggested-by: Shakeel Butt <shakeel.butt@linux.dev>
> Signed-off-by: Peilin Ye <yepeilin@google.com>
> ---
>  kernel/bpf/helpers.c | 10 ++++++++--
>  1 file changed, 8 insertions(+), 2 deletions(-)
> 
> diff --git a/kernel/bpf/helpers.c b/kernel/bpf/helpers.c
> index b9b0c5fe33f6..508b13c24778 100644
> --- a/kernel/bpf/helpers.c
> +++ b/kernel/bpf/helpers.c
> @@ -1274,8 +1274,14 @@ static int __bpf_async_init(struct bpf_async_kern *async, struct bpf_map *map, u
>  		goto out;
>  	}
>  
> -	/* allocate hrtimer via map_kmalloc to use memcg accounting */
> -	cb = bpf_map_kmalloc_node(map, size, GFP_ATOMIC, map->numa_node);
> +	/* Allocate via bpf_map_kmalloc_node() for memcg accounting. Use
> +	 * __GFP_HIGH instead of GFP_ATOMIC to avoid calling
> +	 * cgroup_file_notify() if an MEMCG_MAX event is raised by
> +	 * try_charge_memcg(). This prevents various locking issues, including
> +	 * double-acquiring locks that may already be held here (e.g.,
> +	 * cgroup_file_kn_lock, rq_lock).

Too much unnecessary information in the comment. Just mention that we
want nolock allocations and for that we need to remove __GFP_RECLAIM
flags until nolock allocation interfaces are available.