Thread (2 messages) 2 messages, 2 authors, 5d ago

Re: [PATCH] RDMA/nldev: validate dynamic counter attribute length

From: yanjun.zhu <hidden>
Date: 2026-07-06 23:23:42
Also in: lkml

On 7/6/26 2:12 AM, Pengpeng Hou wrote:
quoted hunk ↗ jump to hunk
From: Pengpeng <redacted>

nldev_stat_set_counter_dynamic_doit() iterates nested hardware counter
attributes and reads each entry with nla_get_u32(). The nested container
proof does not by itself prove that each child attribute carries a
four-byte payload.

Reject hardware counter entries whose payload is shorter than a u32
before reading the index.

Signed-off-by: Pengpeng <redacted>
---
  drivers/infiniband/core/nldev.c | 5 +++++
  1 file changed, 5 insertions(+)
diff --git a/drivers/infiniband/core/nldev.c b/drivers/infiniband/core/nldev.c
index 02a0a9c0a4a6..40b323131c7b 100644
--- a/drivers/infiniband/core/nldev.c
+++ b/drivers/infiniband/core/nldev.c
@@ -2133,6 +2133,11 @@ static int nldev_stat_set_counter_dynamic_doit(struct nlattr *tb[],
  
  	nla_for_each_nested(entry_attr, tb[RDMA_NLDEV_ATTR_STAT_HWCOUNTERS],
  			    rem) {
+		if (nla_len(entry_attr) < sizeof(u32)) {
Maybe nla_len(entry_attr) > sizeof(u32) should also not be allowed?

As such, nla_len(entry_attr) != sizeof(u32) should be better?

Except the above, I am fine with this.

Reviewed-by: Zhu Yanjun <redacted>

Zhu Yanjun
+			ret = -EINVAL;
+			goto out;
+		}
+
  		index = nla_get_u32(entry_attr);
  		if ((index >= stats->num_counters) ||
  		    !(stats->descs[index].flags & IB_STAT_FLAG_OPTIONAL)) {
  
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help