Netdev List
 help / color / mirror / Atom feed
* [PATCH net] net: kcm: Hold RCU read lock while running BPF parser
@ 2026-08-13  3:51 Junseo Lim
  2026-08-17 20:50 ` patchwork-bot+netdevbpf
  0 siblings, 1 reply; 2+ messages in thread
From: Junseo Lim @ 2026-08-13  3:51 UTC (permalink / raw)
  To: David S. Miller, Eric Dumazet, Jakub Kicinski, Paolo Abeni
  Cc: Simon Horman, netdev, bpf, Sechang Lim, Tom Herbert

kcm_parse_func_strparser() calls bpf_prog_run_pin_on_cpu() which
prevents CPU migration, but does not establish an RCU read-side
critical section. Consequently, BPF map operations can trigger
WARN_ON_ONCE(!bpf_rcu_lock_held()) when called from the KCM strparser
program.

Hold the RCU read lock while running the program.

Fixes: 9b73896a81dc ("kcm: Use stream parser")
Reported-by: Sechang Lim <rhkrqnwk98@gmail.com>
Signed-off-by: Junseo Lim <zirajs7@gmail.com>
---
This issue was found by a custom fuzzer developed by
Sechang Lim <rhkrqnwk98@gmail.com>.

The report could not be reproduced locally.  However, the trace shows
the BPF parser being invoked from strp_work without an RCU read-side
critical section.

Below is an excerpt of the warning:

    WARNING: kernel/bpf/hashtab.c:1547 at htab_lru_map_delete_elem+0x604/0x700, CPU#0: kworker/u4:2/28
    CPU: 0 UID: 0 PID: 28 Comm: kworker/u4:2 Not tainted 7.2.0-rc4-dirty #3 PREEMPT(full)
    Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.16.3-0-ga6ed6b701f0a-prebuilt.qemu.org 04/01/2014
    Workqueue: kstrp strp_work
    RIP: 0010:htab_lru_map_delete_elem+0x604/0x700
    Code: 8f d9 03 48 3b 44 24 30 0f 85 b0 00 00 00 4c 89 e0 48 83 c4 38 5b 41 5c 41 5d 41 5e 41 5f 5d c3 cc cc cc cc cc e8 ec a0 dd ff <0f> 0b e9 35 fa ff ff 44 89 f9 80 e1 07 80 c1 03 38 c1 0f 8c 54 fa
    RSP: 0018:ffffc900001cf678 EFLAGS: 00010293
    RAX: ffffffff8e8910f4 RBX: ffff888102bf8800 RCX: ffff888100ce8000
    RDX: 0000000000000000 RSI: ffffffff91df4082 RDI: ffffffff91597ac0
    RBP: ffffc900001cf7f8 R08: 0000000000000000 R09: 0000000000000000
    R10: ffff888106cf6180 R11: ffffffffc020540c R12: ffffc900001cf778
    R13: 1ffff1102019d07f R14: ffff888106cf6140 R15: ffffc90020a65000
    FS:  0000000000000000(0000) GS:0000000000000000(0000) knlGS:0000000000000000
    CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
    CR2: 0000001b34521ff8 CR3: 0000000021a28001 CR4: 0000000000370ef0
    Call Trace:
     <TASK>
     bpf_prog_5dda175565b852e0+0x140/0x153
     ? __copy_skb_header+0xba/0x480
     ? __asan_memcpy+0x40/0x70
     ? __skb_clone+0x61/0x8f0
     bpf_prog_run_pin_on_cpu+0xf4/0x330
     kcm_parse_func_strparser+0x60/0xb0
     __strp_recv+0x5d5/0x1880
     __tcp_read_sock+0x181/0x8c0
     ? __pfx_strp_recv+0x10/0x10
     ? __pfx_tcp_read_sock+0x10/0x10
     strp_work+0x20f/0x3c0
     ? __pfx_strp_work+0x10/0x10
     ? lock_acquire+0xf5/0x250
     ? process_scheduled_works+0x9ce/0x13c0
     process_scheduled_works+0xa3f/0x13c0
     ? __pfx_process_scheduled_works+0x10/0x10
     ? assign_work+0x366/0x530
     worker_thread+0x93c/0xe70
     kthread+0x34b/0x460
     ? __pfx_worker_thread+0x10/0x10
     ? __pfx_kthread+0x10/0x10
     ret_from_fork+0x348/0x700
     ? __pfx_ret_from_fork+0x10/0x10
     ? native_load_tls+0xd/0x40
     ? __switch_to+0x916/0xc30
     ? __pfx_kthread+0x10/0x10
     ret_from_fork_asm+0x19/0x30
     </TASK>

 
 net/kcm/kcmsock.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/net/kcm/kcmsock.c b/net/kcm/kcmsock.c
index d469abcd989b..71af69d442f2 100644
--- a/net/kcm/kcmsock.c
+++ b/net/kcm/kcmsock.c
@@ -5,6 +5,7 @@
  * Copyright (c) 2016 Tom Herbert <tom@herbertland.com>
  */
 
+#include <linux/rcupdate.h>
 #include <linux/bpf.h>
 #include <linux/errno.h>
 #include <linux/errqueue.h>
@@ -391,7 +392,9 @@ static int kcm_parse_func_strparser(struct strparser *strp, struct sk_buff *skb)
 	struct bpf_prog *prog = psock->bpf_prog;
 	int res;
 
+	rcu_read_lock();
 	res = bpf_prog_run_pin_on_cpu(prog, skb);
+	rcu_read_unlock();
 	return res;
 }
 
-- 
2.55.0


^ permalink raw reply related	[flat|nested] 2+ messages in thread

* Re: [PATCH net] net: kcm: Hold RCU read lock while running BPF parser
  2026-08-13  3:51 [PATCH net] net: kcm: Hold RCU read lock while running BPF parser Junseo Lim
@ 2026-08-17 20:50 ` patchwork-bot+netdevbpf
  0 siblings, 0 replies; 2+ messages in thread
From: patchwork-bot+netdevbpf @ 2026-08-17 20:50 UTC (permalink / raw)
  To: Junseo Lim
  Cc: davem, edumazet, kuba, pabeni, horms, netdev, bpf, rhkrqnwk98,
	tom

Hello:

This patch was applied to netdev/net.git (main)
by Jakub Kicinski <kuba@kernel.org>:

On Thu, 13 Aug 2026 12:51:36 +0900 you wrote:
> kcm_parse_func_strparser() calls bpf_prog_run_pin_on_cpu() which
> prevents CPU migration, but does not establish an RCU read-side
> critical section. Consequently, BPF map operations can trigger
> WARN_ON_ONCE(!bpf_rcu_lock_held()) when called from the KCM strparser
> program.
> 
> Hold the RCU read lock while running the program.
> 
> [...]

Here is the summary with links:
  - [net] net: kcm: Hold RCU read lock while running BPF parser
    https://git.kernel.org/netdev/net/c/b0346dd64e49

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



^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2026-08-17 20:50 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2026-08-13  3:51 [PATCH net] net: kcm: Hold RCU read lock while running BPF parser Junseo Lim
2026-08-17 20:50 ` patchwork-bot+netdevbpf

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox