RDMA/rtrs: guard against null kobj name

[PATCH] RDMA/rtrs: guard against null kobj name

Posted by Ryan Mehri 1 month ago

In the client, if `init_path()` errors, the callee tries to clean up
with `rtrs_clt_close_conns()`. However, this can lead to calling the
event tracing code with `clt_path->kobj->name` being `NULL` and thus
causing a null pointer dereference when trying to copy from it. 

This just adds a guard to check that the name is not `NULL` before
copying from it. The server appears to have a similar pattern.

Fixes: 5a93929d9f9a1 ("RDMA/rtrs-clt: Add event tracing support")
Fixes: c16762b7bf54d ("RDMA/rtrs-srv: Add event tracing support")
Reported-by: syzbot+1695193198994f4e7fed@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=1695193198994f4e7fed
Signed-off-by: Ryan Mehri <ryan.mehri1@gmail.com>
---
 drivers/infiniband/ulp/rtrs/rtrs-clt-trace.h | 2 +-
 drivers/infiniband/ulp/rtrs/rtrs-srv-trace.h | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/infiniband/ulp/rtrs/rtrs-clt-trace.h b/drivers/infiniband/ulp/rtrs/rtrs-clt-trace.h
index 7738e2676855..29e23404bb7b 100644
--- a/drivers/infiniband/ulp/rtrs/rtrs-clt-trace.h
+++ b/drivers/infiniband/ulp/rtrs/rtrs-clt-trace.h
@@ -55,7 +55,7 @@ DECLARE_EVENT_CLASS(rtrs_clt_conn_class,
 		__entry->max_reconnect_attempts = clt->max_reconnect_attempts;
 		__entry->fail_cnt = clt_path->stats->reconnects.fail_cnt;
 		__entry->success_cnt = clt_path->stats->reconnects.successful_cnt;
-		memcpy(__entry->sessname, kobject_name(&clt_path->kobj), NAME_MAX);
+		strscpy(__entry->sessname, kobject_name(&clt_path->kobj) ?: "", NAME_MAX);
 	),
 
 	TP_printk("RTRS-CLT: sess='%s' state=%s attempts='%d' max-attempts='%d' fail='%d' success='%d'",
diff --git a/drivers/infiniband/ulp/rtrs/rtrs-srv-trace.h b/drivers/infiniband/ulp/rtrs/rtrs-srv-trace.h
index 587d3e033081..a7d7b971e6c8 100644
--- a/drivers/infiniband/ulp/rtrs/rtrs-srv-trace.h
+++ b/drivers/infiniband/ulp/rtrs/rtrs-srv-trace.h
@@ -61,7 +61,7 @@ TRACE_EVENT(send_io_resp_imm,
 		__entry->msg_id = id->msg_id;
 		__entry->wr_cnt = atomic_read(&con->c.wr_cnt);
 		__entry->signal_interval = s->signal_interval;
-		memcpy(__entry->sessname, kobject_name(&srv_path->kobj), NAME_MAX);
+		strscpy(__entry->sessname, kobject_name(&srv_path->kobj) ?: "", NAME_MAX);
 	),
 
 	TP_printk("sess='%s' state='%s' dir=%s err='%d' inval='%d' glob-inval='%d' msgid='%u' wrcnt='%d' sig-interval='%u'",
-- 
2.55.0

Re: [PATCH] RDMA/rtrs: guard against null kobj name

Posted by Leon Romanovsky 3 weeks, 4 days ago

On Sat, 22 Aug 2026 23:33:43 -0400, Ryan Mehri wrote:
> In the client, if `init_path()` errors, the callee tries to clean up
> with `rtrs_clt_close_conns()`. However, this can lead to calling the
> event tracing code with `clt_path->kobj->name` being `NULL` and thus
> causing a null pointer dereference when trying to copy from it.
> 
> This just adds a guard to check that the name is not `NULL` before
> copying from it. The server appears to have a similar pattern.
> 
> [...]

Applied, thanks!

[1/1] RDMA/rtrs: guard against null kobj name
      https://git.kernel.org/rdma/rdma/c/99c24a8968ebef

Best regards,
-- 
Leon Romanovsky <leon@kernel.org>

Re: [PATCH] RDMA/rtrs: guard against null kobj name

Posted by Jinpu Wang 1 month ago

On Sun, Aug 23, 2026 at 5:43 AM Ryan Mehri <ryan.mehri1@gmail.com> wrote:
>
> In the client, if `init_path()` errors, the callee tries to clean up
> with `rtrs_clt_close_conns()`. However, this can lead to calling the
> event tracing code with `clt_path->kobj->name` being `NULL` and thus
> causing a null pointer dereference when trying to copy from it.
>
> This just adds a guard to check that the name is not `NULL` before
> copying from it. The server appears to have a similar pattern.
>
> Fixes: 5a93929d9f9a1 ("RDMA/rtrs-clt: Add event tracing support")
> Fixes: c16762b7bf54d ("RDMA/rtrs-srv: Add event tracing support")
> Reported-by: syzbot+1695193198994f4e7fed@syzkaller.appspotmail.com
> Closes: https://syzkaller.appspot.com/bug?extid=1695193198994f4e7fed
> Signed-off-by: Ryan Mehri <ryan.mehri1@gmail.com>
lgtm, thx!
Reviewed-by: Jack Wang <jinpu.wang@cloud.ionos.com>

> ---
>  drivers/infiniband/ulp/rtrs/rtrs-clt-trace.h | 2 +-
>  drivers/infiniband/ulp/rtrs/rtrs-srv-trace.h | 2 +-
>  2 files changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/infiniband/ulp/rtrs/rtrs-clt-trace.h b/drivers/infiniband/ulp/rtrs/rtrs-clt-trace.h
> index 7738e2676855..29e23404bb7b 100644
> --- a/drivers/infiniband/ulp/rtrs/rtrs-clt-trace.h
> +++ b/drivers/infiniband/ulp/rtrs/rtrs-clt-trace.h
> @@ -55,7 +55,7 @@ DECLARE_EVENT_CLASS(rtrs_clt_conn_class,
>                 __entry->max_reconnect_attempts = clt->max_reconnect_attempts;
>                 __entry->fail_cnt = clt_path->stats->reconnects.fail_cnt;
>                 __entry->success_cnt = clt_path->stats->reconnects.successful_cnt;
> -               memcpy(__entry->sessname, kobject_name(&clt_path->kobj), NAME_MAX);
> +               strscpy(__entry->sessname, kobject_name(&clt_path->kobj) ?: "", NAME_MAX);
>         ),
>
>         TP_printk("RTRS-CLT: sess='%s' state=%s attempts='%d' max-attempts='%d' fail='%d' success='%d'",
> diff --git a/drivers/infiniband/ulp/rtrs/rtrs-srv-trace.h b/drivers/infiniband/ulp/rtrs/rtrs-srv-trace.h
> index 587d3e033081..a7d7b971e6c8 100644
> --- a/drivers/infiniband/ulp/rtrs/rtrs-srv-trace.h
> +++ b/drivers/infiniband/ulp/rtrs/rtrs-srv-trace.h
> @@ -61,7 +61,7 @@ TRACE_EVENT(send_io_resp_imm,
>                 __entry->msg_id = id->msg_id;
>                 __entry->wr_cnt = atomic_read(&con->c.wr_cnt);
>                 __entry->signal_interval = s->signal_interval;
> -               memcpy(__entry->sessname, kobject_name(&srv_path->kobj), NAME_MAX);
> +               strscpy(__entry->sessname, kobject_name(&srv_path->kobj) ?: "", NAME_MAX);
>         ),
>
>         TP_printk("sess='%s' state='%s' dir=%s err='%d' inval='%d' glob-inval='%d' msgid='%u' wrcnt='%d' sig-interval='%u'",
> --
> 2.55.0
>