[PATCH bpf v5] bpf: Fix invalid mem access when update_effective_progs fails in __cgroup_bpf_detach

[PATCH bpf v5] bpf: Fix invalid mem access when update_effective_progs fails in __cgroup_bpf_detach

[Pu Lehui]

From: Pu Lehui <pulehui@huawei.com>

Syzkaller triggers an invalid memory access issue following fault
injection in update_effective_progs. The issue can be described as
follows:

__cgroup_bpf_detach
  update_effective_progs
    compute_effective_progs
      bpf_prog_array_alloc <-- fault inject
  purge_effective_progs
    /* change to dummy_bpf_prog */
    array->items[index] = &dummy_bpf_prog.prog

---softirq start---
__do_softirq
  ...
    __cgroup_bpf_run_filter_skb
      __bpf_prog_run_save_cb
        bpf_prog_run
          stats = this_cpu_ptr(prog->stats)
          /* invalid memory access */
          flags = u64_stats_update_begin_irqsave(&stats->syncp)
---softirq end---

  static_branch_dec(&cgroup_bpf_enabled_key[atype])

The reason is that fault injection caused update_effective_progs to fail
and then changed the original prog into dummy_bpf_prog.prog in
purge_effective_progs. Then a softirq came, and accessing the members of
dummy_bpf_prog.prog in the softirq triggers invalid mem access.

To fix it, we can skip updating stats when stats is NULL.

Fixes: 492ecee892c2 ("bpf: enable program stats")
Signed-off-by: Pu Lehui <pulehui@huawei.com>
---
v5:
- Add NULL check in bpf_prog_inc_misses_counter (AI)
- update Fixes tag

v4:
- roll back to the skip solution, but skip only updating stats.

v3:
- add static for the per-cpu variable.

v2:
- Use static per-cpu variables to initialize the stats of
  dummy_bpf_prog.prog suggested by Eduard.

 include/linux/filter.h | 12 +++++++-----
 kernel/bpf/syscall.c   |  3 +++
 2 files changed, 10 insertions(+), 5 deletions(-)

diff --git a/include/linux/filter.h b/include/linux/filter.h
index 973233b82dc1..569de3b14279 100644
--- a/include/linux/filter.h
+++ b/include/linux/filter.h
@@ -712,11 +712,13 @@ static __always_inline u32 __bpf_prog_run(const struct bpf_prog *prog,
 		ret = dfunc(ctx, prog->insnsi, prog->bpf_func);
 
 		duration = sched_clock() - start;
-		stats = this_cpu_ptr(prog->stats);
-		flags = u64_stats_update_begin_irqsave(&stats->syncp);
-		u64_stats_inc(&stats->cnt);
-		u64_stats_add(&stats->nsecs, duration);
-		u64_stats_update_end_irqrestore(&stats->syncp, flags);
+		if (likely(prog->stats)) {
+			stats = this_cpu_ptr(prog->stats);
+			flags = u64_stats_update_begin_irqsave(&stats->syncp);
+			u64_stats_inc(&stats->cnt);
+			u64_stats_add(&stats->nsecs, duration);
+			u64_stats_update_end_irqrestore(&stats->syncp, flags);
+		}
 	} else {
 		ret = dfunc(ctx, prog->insnsi, prog->bpf_func);
 	}
diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c
index 8a129746bd6c..15f9afdbfc27 100644
--- a/kernel/bpf/syscall.c
+++ b/kernel/bpf/syscall.c
@@ -2462,6 +2462,9 @@ void notrace bpf_prog_inc_misses_counter(struct bpf_prog *prog)
 	struct bpf_prog_stats *stats;
 	unsigned int flags;
 
+	if (unlikely(!prog->stats))
+		return;
+
 	stats = this_cpu_ptr(prog->stats);
 	flags = u64_stats_update_begin_irqsave(&stats->syncp);
 	u64_stats_inc(&stats->misses);
-- 
2.34.1

Re: [PATCH bpf v5] bpf: Fix invalid mem access when update_effective_progs fails in __cgroup_bpf_detach

[patchwork-bot+netdevbpf]

Hello:

This patch was applied to bpf/bpf-next.git (master)
by Alexei Starovoitov <ast@kernel.org>:

On Sat, 15 Nov 2025 10:23:43 +0000 you wrote:
> From: Pu Lehui <pulehui@huawei.com>
> 
> Syzkaller triggers an invalid memory access issue following fault
> injection in update_effective_progs. The issue can be described as
> follows:
> 
> __cgroup_bpf_detach
>   update_effective_progs
>     compute_effective_progs
>       bpf_prog_array_alloc <-- fault inject
>   purge_effective_progs
>     /* change to dummy_bpf_prog */
>     array->items[index] = &dummy_bpf_prog.prog
> 
> [...]

Here is the summary with links:
  - [bpf,v5] bpf: Fix invalid mem access when update_effective_progs fails in __cgroup_bpf_detach
    https://git.kernel.org/bpf/bpf-next/c/7dc211c1159d

You are awesome, thank you!
-- 
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html