[PATCH] 9p/rdma: create RDMA CM ID in caller's net namespace

[PATCH] 9p/rdma: create RDMA CM ID in caller's net namespace

[Yizhou Zhao]

The 9p RDMA transport currently passes &init_net to rdma_create_id().
As a result, RDMA address resolution and connection setup are performed
in the initial network namespace, even when the mount is initiated from
a non-initial network namespace.

This differs from the socket-based 9p transports, which create sockets
in current->nsproxy->net_ns.  Use the caller's network namespace for the
RDMA CM ID as well, so that RDMA transport setup follows the same
namespace context as the process performing the mount.

This avoids surprising behaviour where a 9p RDMA mount from a container
or other non-initial network namespace may use the host namespace for
RDMA CM operations.

Fixes: fa20105e09e9 ("IB/cma: Add support for network namespaces")
Reported-by: Yizhou Zhao <zhaoyz24@mails.tsinghua.edu.cn>
Reported-by: Yuxiang Yang <yangyx22@mails.tsinghua.edu.cn>
Reported-by: Ao Wang <wangao@seu.edu.cn>
Reported-by: Xuewei Feng <fengxw06@126.com>
Reported-by: Qi Li <qli01@tsinghua.edu.cn>
Reported-by: Ke Xu <xuke@tsinghua.edu.cn>
Assisted-by: GLM:GLM-5.1
Signed-off-by: Yizhou Zhao <zhaoyz24@mails.tsinghua.edu.cn>
---
 net/9p/trans_rdma.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/net/9p/trans_rdma.c b/net/9p/trans_rdma.c
index aa5bd74..ea49b15 100644
--- a/net/9p/trans_rdma.c
+++ b/net/9p/trans_rdma.c
@@ -541,7 +541,7 @@ rdma_create_trans(struct p9_client *client, struct fs_context *fc)
 		return -ENOMEM;
 
 	/* Create the RDMA CM ID */
-	rdma->cm_id = rdma_create_id(&init_net, p9_cm_event_handler, client,
+	rdma->cm_id = rdma_create_id(current->nsproxy->net_ns, p9_cm_event_handler, client,
 				     RDMA_PS_TCP, IB_QPT_RC);
 	if (IS_ERR(rdma->cm_id))
 		goto error;
-- 
2.43.0

Re: [PATCH] 9p/rdma: create RDMA CM ID in caller's net namespace

[Dominique Martinet]

Yizhou Zhao wrote on Fri, May 29, 2026 at 03:08:15PM +0800:
> The 9p RDMA transport currently passes &init_net to rdma_create_id().
> As a result, RDMA address resolution and connection setup are performed
> in the initial network namespace, even when the mount is initiated from
> a non-initial network namespace.
> 
> This differs from the socket-based 9p transports, which create sockets
> in current->nsproxy->net_ns.  Use the caller's network namespace for the
> RDMA CM ID as well, so that RDMA transport setup follows the same
> namespace context as the process performing the mount.
> 
> This avoids surprising behaviour where a 9p RDMA mount from a container
> or other non-initial network namespace may use the host namespace for
> RDMA CM operations.
> 
> Fixes: fa20105e09e9 ("IB/cma: Add support for network namespaces")
> Reported-by: Yizhou Zhao <zhaoyz24@mails.tsinghua.edu.cn>
> Reported-by: Yuxiang Yang <yangyx22@mails.tsinghua.edu.cn>
> Reported-by: Ao Wang <wangao@seu.edu.cn>
> Reported-by: Xuewei Feng <fengxw06@126.com>
> Reported-by: Qi Li <qli01@tsinghua.edu.cn>
> Reported-by: Ke Xu <xuke@tsinghua.edu.cn>
> Assisted-by: GLM:GLM-5.1
> Signed-off-by: Yizhou Zhao <zhaoyz24@mails.tsinghua.edu.cn>
> ---
>  net/9p/trans_rdma.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/net/9p/trans_rdma.c b/net/9p/trans_rdma.c
> index aa5bd74..ea49b15 100644
> --- a/net/9p/trans_rdma.c
> +++ b/net/9p/trans_rdma.c
> @@ -541,7 +541,7 @@ rdma_create_trans(struct p9_client *client, struct fs_context *fc)
>  		return -ENOMEM;
>  
>  	/* Create the RDMA CM ID */
> -	rdma->cm_id = rdma_create_id(&init_net, p9_cm_event_handler, client,
> +	rdma->cm_id = rdma_create_id(current->nsproxy->net_ns, p9_cm_event_handler, client,

We apparently should use fc->net_ns[1] (likewise in other transports that
use current->nsproxy->net_ns)

[1] https://sashiko.dev/#/patchset/20260529070817.76798-1-zhaoyz24%40mails.tsinghua.edu.cn

If you have time please send a (separate) patch fixing trans_fd.c ?

-- 
Dominique Martinet | Asmadeus

Re: [PATCH] 9p/rdma: create RDMA CM ID in caller's net namespace

[Yizhou Zhao]

On Fri, May 29, 2026 at 03:08:15PM +0800, Dominique Martinet wrote:
> We apparently should use fc->net_ns[1] (likewise in other transports that
> use current->nsproxy->net_ns)
>
> [1] https://sashiko.dev/#/patchset/20260529070817.76798-1-zhaoyz24%40mails.tsinghua.edu.cn
>
> If you have time please send a (separate) patch fixing trans_fd.c ?

Thank you for the review!  You're right, fc->net_ns is the proper way.

Instead of sending a separate patch for trans_fd.c, We've folded the
trans_fd.c fix into the same patch and updated trans_rdma.c to use
fc->net_ns as well.  Please find v2 here:

  https://lore.kernel.org/v9fs/20260529090625.78761-1-zhaoyz24@mails.tsinghua.edu.cn/

Best regards,
Yizhou