RDMA/cma: Fix WARNING in res_to_rt
PATCH 1/1 RDMA/cma: Fix WARNING in res_to_rt
syzbot reported a WARN_ON(!res->dev) in res_to_rt() triggered via addr_handler() during asynchronous address resolution:
"
WARNING: drivers/infiniband/core/restrack.c:138 at res_to_rt+0x1c4/0x230 CPU#1: kworker/u8:4/59 Modules linked in: CPU: 1 UID: 0 PID: 59 Comm: kworker/u8:4 Not tainted syzkaller #0 PREEMPT(full) Hardware name: Google Compute Engine, BIOS Google 07/24/2026 Workqueue: ib_addr process_one_req RIP: 0010:res_to_rt+0x1c4/0x230 drivers/infiniband/core/restrack.c:138 RSP: 0018:ffffc9000201f850 EFLAGS: 00010293 RAX: ffffffff88d00ce5 RBX: ffff88807f0fd4f8 RCX: ffff88801e6e0000 RDX: 0000000000000000 RSI: ffffffff8fd996f0 RDI: 0000000000000003 RBP: 0000000000000000 R08: ffff88801e6e0000 R09: 000000000000000a R10: 0000000000000009 R11: 0000000000000000 R12: dffffc0000000000 R13: 1ffff1100fe1fa9f R14: 0000000000000000 R15: 0000000000000003 FS: 0000000000000000(0000) GS:ffff888125012000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 00001d559c3d2000 CR3: 0000000077c4c000 CR4: 00000000003526f0 Call Trace: <TASK> rdma_restrack_add+0x5a/0x8a0 drivers/infiniband/core/restrack.c:236 addr_handler+0x41a/0x5a0 drivers/infiniband/core/cma.c:3534 process_one_req+0x2eb/0x540 drivers/infiniband/core/addr.c:624 process_one_work kernel/workqueue.c:3375 [inline] process_scheduled_works+0xc4e/0x1630 kernel/workqueue.c:3458 worker_thread+0xa47/0xfb0 kernel/workqueue.c:3539 kthread+0x388/0x470 kernel/kthread.c:436 ret_from_fork+0x514/0xb70 arch/x86/kernel/process.c:158 ret_from_fork_asm+0x1a/0x30 arch/x86/entry/entry_64.S:245 </TASK>
"
The function addr_handler is invoked to process the resolved destination. If address resolution fails or encounters a state race before an RDMA device binding is established (that is, id_priv->id.device remains NULL), id_priv->res.dev is not initialized.
addr_handler() previously called rdma_restrack_add(&id_priv->res) unconditionally upon handling the resolution step. Passing an uninitialized restrack resource with a NULL dev pointer causes res_to_rt() to hit the WARN_ON(!res->dev) assertion.
Fix this by adding a check for id_priv->id.device in addr_handler() before registering the CMA ID with restrack.
--- drivers/infiniband/core/cma.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/infiniband/core/cma.c +2 -1
@@ -3531,7 +3531,8 @@ static void addr_handler(int status, struct sockaddr *src_addr,if (status)pr_debug_ratelimited ("RDMA CM: ADDR_ERROR: failed to acquire device. status %d\n" , status); -rdma_restrack_add (&id_priv->res); +if (id_priv->id.device) +rdma_restrack_add (&id_priv->res); }else if (status) {pr_debug_ratelimited ("RDMA CM: ADDR_ERROR: failed to resolve IP. status %d\n" , status); }
Signature · 1 line
2.43.0
drivers/infiniband/core/cma.c static void addr_handler(int status, struct sockaddr *src_addr, · line 3538
Quoted 10 lines · Zhu Yanjun
drivers/infiniband/core/cma.c +2
-1@@ -3531,7 +3531,8 @@ static void addr_handler(int status, struct sockaddr *src_addr,if (status)pr_debug_ratelimited ("RDMA CM: ADDR_ERROR: failed to acquire device. status %d\n" , status); -rdma_restrack_add (&id_priv->res); +if (id_priv->id.device) +rdma_restrack_add (&id_priv->res); }else if (status) {pr_debug_ratelimited ("RDMA CM: ADDR_ERROR: failed to resolve IP. status %d\n" , status); }
The full text is:
if (!status && !id_priv->cma_dev) {
status = cma_acquire_dev_by_src_ip(id_priv);
if (status)
pr_debug_ratelimited("RDMA CM: ADDR_ERROR: failed to acquire device. status %d\n",
status);
rdma_restrack_add(&id_priv->res);What it is trying to do is fill in cma_dev/id.device (they are linked).
'if (status)' means the resolution failed and the cm_id is not populated, so it should just be if/else instead of trying to check id.device
But this does seem to be the bug syzkaller is pointing at, but the commit message seems really strange.
If status is !0 then we know id.device is NULL and we cannot add this to restrack. Otherwise it must be !NULL and it must be safe to add it or we have a locking problem.
Jason