Re: [PATCH bpf v2] bpf: Fix deadlock when freeing cgroup storage

[Martin KaFai Lau <martin.lau@linux.dev>]

On 12/20/24 10:10 PM, Abel Wu wrote:
> The following commit
> bc235cdb423a ("bpf: Prevent deadlock from recursive bpf_task_storage_[get|delete]")
> first introduced deadlock prevention for fentry/fexit programs attaching
> on bpf_task_storage helpers. That commit also employed the logic in map
> free path in its v6 version.
> 
> Later bpf_cgrp_storage was first introduced in
> c4bcfb38a95e ("bpf: Implement cgroup storage available to non-cgroup-attached bpf progs")
> which faces the same issue as bpf_task_storage, instead of its busy
> counter, NULL was passed to bpf_local_storage_map_free() which opened
> a window to cause deadlock:
> 
> 	<TASK>
> 		(acquiring local_storage->lock)
> 	_raw_spin_lock_irqsave+0x3d/0x50
> 	bpf_local_storage_update+0xd1/0x460
> 	bpf_cgrp_storage_get+0x109/0x130
> 	bpf_prog_a4d4a370ba857314_cgrp_ptr+0x139/0x170
> 	? __bpf_prog_enter_recur+0x16/0x80
> 	bpf_trampoline_6442485186+0x43/0xa4
> 	cgroup_storage_ptr+0x9/0x20
> 		(holding local_storage->lock)
> 	bpf_selem_unlink_storage_nolock.constprop.0+0x135/0x160
> 	bpf_selem_unlink_storage+0x6f/0x110
> 	bpf_local_storage_map_free+0xa2/0x110
> 	bpf_map_free_deferred+0x5b/0x90
> 	process_one_work+0x17c/0x390
> 	worker_thread+0x251/0x360
> 	kthread+0xd2/0x100
> 	ret_from_fork+0x34/0x50
> 	ret_from_fork_asm+0x1a/0x30
> 	</TASK>
> 
> Progs:
>   - A: SEC("fentry/cgroup_storage_ptr")

The v1 thread has suggested using notrace in a few functions. I didn't see any 
counterarguments that wouldn't be sufficient.

imo, that should be a better option instead of having more unnecessary failures 
in all other normal use cases which will not be interested in tracing 
cgroup_storage_ptr().

pw-bot: cr

>     - cgid (BPF_MAP_TYPE_HASH)
> 	Record the id of the cgroup the current task belonging
> 	to in this hash map, using the address of the cgroup
> 	as the map key.
>     - cgrpa (BPF_MAP_TYPE_CGRP_STORAGE)
> 	If current task is a kworker, lookup the above hash
> 	map using function parameter @owner as the key to get
> 	its corresponding cgroup id which is then used to get
> 	a trusted pointer to the cgroup through
> 	bpf_cgroup_from_id(). This trusted pointer can then
> 	be passed to bpf_cgrp_storage_get() to finally trigger
> 	the deadlock issue.
>   - B: SEC("tp_btf/sys_enter")
>     - cgrpb (BPF_MAP_TYPE_CGRP_STORAGE)
> 	The only purpose of this prog is to fill Prog A's
> 	hash map by calling bpf_cgrp_storage_get() for as
> 	many userspace tasks as possible.
> 
> Steps to reproduce:
>   - Run A;
>   - while (true) { Run B; Destroy B; }
> 
> Fix this issue by passing its busy counter to the free procedure so
> it can be properly incremented before storage/smap locking.
> 
> Fixes: c4bcfb38a95e ("bpf: Implement cgroup storage available to non-cgroup-attached bpf progs")
> Signed-off-by: Abel Wu <wuyun.abel@bytedance.com>
> ---
>   kernel/bpf/bpf_cgrp_storage.c | 2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/kernel/bpf/bpf_cgrp_storage.c b/kernel/bpf/bpf_cgrp_storage.c
> index 20f05de92e9c..7996fcea3755 100644
> --- a/kernel/bpf/bpf_cgrp_storage.c
> +++ b/kernel/bpf/bpf_cgrp_storage.c
> @@ -154,7 +154,7 @@ static struct bpf_map *cgroup_storage_map_alloc(union bpf_attr *attr)
>   
>   static void cgroup_storage_map_free(struct bpf_map *map)
>   {
> -	bpf_local_storage_map_free(map, &cgroup_cache, NULL);
> +	bpf_local_storage_map_free(map, &cgroup_cache, &bpf_cgrp_storage_busy);
>   }
>   
>   /* *gfp_flags* is a hidden argument provided by the verifier */