public inbox for dev@dpdk.org
 help / color / mirror / Atom feed
* [PATCH] vhost: Fix the crash caused by accessing the released memory
@ 2024-06-17  3:35 zhaoxinxin
  0 siblings, 0 replies; 9+ messages in thread
From: zhaoxinxin @ 2024-06-17  3:35 UTC (permalink / raw)
  To: dev; +Cc: zhaoxinxin

The rte_vhost_driver_unregister() 、vhost_user_read_cb()、vhost_user_client_reconnect() can be called at the same time by 3 threads.
when memory of vsocket is freed in rte_vhost_driver_unregister(), then vhost_user_read_cb() maybe add vsocket to reconn_list, the invalid memory of vsocket is accessed in vhost_user_client_reconnect().
The core trace is:
    Program terminated with signal 11, Segmentation fault.
    The fix is to perform a delete operation again after releasing the memory

Signed-off-by: zhaoxinxin <15957197901@163.com>
---
 lib/vhost/socket.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/lib/vhost/socket.c b/lib/vhost/socket.c
index a75728a2e4..01946096c4 100644
--- a/lib/vhost/socket.c
+++ b/lib/vhost/socket.c
@@ -1121,6 +1121,8 @@ rte_vhost_driver_unregister(const char *path)
 		if (vsocket->is_server) {
 			close(vsocket->socket_fd);
 			unlink(path);
+		} else if (vsocket->reconnect) {
+			vhost_user_remove_reconnect(vsocket);
 		}
 
 		pthread_mutex_destroy(&vsocket->conn_mutex);
-- 
2.45.2


^ permalink raw reply related	[flat|nested] 9+ messages in thread

* [PATCH] vhost: Fix the crash caused by accessing the released memory
@ 2024-06-18  5:59 zhaoxinxin
  0 siblings, 0 replies; 9+ messages in thread
From: zhaoxinxin @ 2024-06-18  5:59 UTC (permalink / raw)
  To: dev; +Cc: zhaoxinxin

The rte_vhost_driver_unregister() vhost_user_read_cb() vhost_user_client_reconnect() can be called at the same time by 3 threads.
when memory of vsocket is freed in rte_vhost_driver_unregister(),
then vhost_user_read_cb() maybe add vsocket to reconn_list,
the invalid memory of vsocket is accessed in vhost_user_client_reconnect().

The core trace is:
    Program terminated with signal 11, Segmentation fault.
    The fix is to perform a delete operation again after releasing the memory

Signed-off-by: zhaoxinxin <15957197901@163.com>
---
 lib/vhost/socket.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/lib/vhost/socket.c b/lib/vhost/socket.c
index a75728a2e4..01946096c4 100644
--- a/lib/vhost/socket.c
+++ b/lib/vhost/socket.c
@@ -1121,6 +1121,8 @@ rte_vhost_driver_unregister(const char *path)
 		if (vsocket->is_server) {
 			close(vsocket->socket_fd);
 			unlink(path);
+		} else if (vsocket->reconnect) {
+			vhost_user_remove_reconnect(vsocket);
 		}
 
 		pthread_mutex_destroy(&vsocket->conn_mutex);
-- 
2.45.2


^ permalink raw reply related	[flat|nested] 9+ messages in thread

* [PATCH] vhost: Fix the crash caused by accessing the released memory
@ 2024-06-19  8:39 zhaoxinxin
  0 siblings, 0 replies; 9+ messages in thread
From: zhaoxinxin @ 2024-06-19  8:39 UTC (permalink / raw)
  To: dev; +Cc: zhaoxinxin

The rte_vhost_driver_unregister() vhost_user_read_cb()
vhost_user_client_reconnect() can be called at the same time by 3 threads.
when memory of vsocket is freed in rte_vhost_driver_unregister(),
then vhost_user_read_cb() maybe add vsocket to reconn_list,
the invalid memory of vsocket is accessed in vhost_user_client_reconnect().

The core trace is:
    Program terminated with signal 11, Segmentation fault.
    The fix is to perform a delete operation again after releasing the memory

Signed-off-by: zhaoxinxin <15957197901@163.com>
---
 lib/vhost/socket.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/lib/vhost/socket.c b/lib/vhost/socket.c
index a75728a2e4..01946096c4 100644
--- a/lib/vhost/socket.c
+++ b/lib/vhost/socket.c
@@ -1121,6 +1121,8 @@ rte_vhost_driver_unregister(const char *path)
 		if (vsocket->is_server) {
 			close(vsocket->socket_fd);
 			unlink(path);
+		} else if (vsocket->reconnect) {
+			vhost_user_remove_reconnect(vsocket);
 		}
 
 		pthread_mutex_destroy(&vsocket->conn_mutex);
-- 
2.45.2


^ permalink raw reply related	[flat|nested] 9+ messages in thread

* [PATCH] vhost: Fix the crash caused by accessing the released memory
@ 2024-06-19 12:27 zhaoxinxin
  2024-06-24  9:20 ` Maxime Coquelin
  0 siblings, 1 reply; 9+ messages in thread
From: zhaoxinxin @ 2024-06-19 12:27 UTC (permalink / raw)
  To: dev; +Cc: zhaoxinxin

The rte_vhost_driver_unregister() vhost_user_read_cb()
vhost_user_client_reconnect() can be called at the same time by 3 threads.
when memory of vsocket is freed in rte_vhost_driver_unregister(),
then vhost_user_read_cb() maybe add vsocket to reconn_list,
the invalid memory of vsocket is accessed in vhost_user_client_reconnect().

The core trace is:
Program terminated with signal 11, Segmentation fault.
The fix is to perform a delete operation again after releasing the memory

Signed-off-by: zhaoxinxin <15957197901@163.com>
---
 lib/vhost/socket.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/lib/vhost/socket.c b/lib/vhost/socket.c
index a75728a2e4..01946096c4 100644
--- a/lib/vhost/socket.c
+++ b/lib/vhost/socket.c
@@ -1121,6 +1121,8 @@ rte_vhost_driver_unregister(const char *path)
 		if (vsocket->is_server) {
 			close(vsocket->socket_fd);
 			unlink(path);
+		} else if (vsocket->reconnect) {
+			vhost_user_remove_reconnect(vsocket);
 		}
 
 		pthread_mutex_destroy(&vsocket->conn_mutex);
-- 
2.45.2


^ permalink raw reply related	[flat|nested] 9+ messages in thread

* Re: [PATCH] vhost: Fix the crash caused by accessing the released memory
  2024-06-19 12:27 zhaoxinxin
@ 2024-06-24  9:20 ` Maxime Coquelin
  0 siblings, 0 replies; 9+ messages in thread
From: Maxime Coquelin @ 2024-06-24  9:20 UTC (permalink / raw)
  To: zhaoxinxin, dev

Hi,

On 6/19/24 14:27, zhaoxinxin wrote:
> The rte_vhost_driver_unregister() vhost_user_read_cb()
> vhost_user_client_reconnect() can be called at the same time by 3 threads.
> when memory of vsocket is freed in rte_vhost_driver_unregister(),
> then vhost_user_read_cb() maybe add vsocket to reconn_list,
> the invalid memory of vsocket is accessed in vhost_user_client_reconnect().

It is not clear to me why 3 threads are calling
rte_vhost_driver_unregister() at the same time, isn't it an application
issue?

> The core trace is:
> Program terminated with signal 11, Segmentation fault.
> The fix is to perform a delete operation again after releasing the memory
>

We need a Fixes tag and Cc stable@dpdk.org so that it is backported.

> Signed-off-by: zhaoxinxin <15957197901@163.com>

The format is Firstname Lastname <email>


> ---
>   lib/vhost/socket.c | 2 ++
>   1 file changed, 2 insertions(+)
> 
> diff --git a/lib/vhost/socket.c b/lib/vhost/socket.c
> index a75728a2e4..01946096c4 100644
> --- a/lib/vhost/socket.c
> +++ b/lib/vhost/socket.c
> @@ -1121,6 +1121,8 @@ rte_vhost_driver_unregister(const char *path)
>   		if (vsocket->is_server) {
>   			close(vsocket->socket_fd);
>   			unlink(path);
> +		} else if (vsocket->reconnect) {
> +			vhost_user_remove_reconnect(vsocket);
>   		}
>   
>   		pthread_mutex_destroy(&vsocket->conn_mutex);


^ permalink raw reply	[flat|nested] 9+ messages in thread

* [PATCH] vhost: Fix the crash caused by accessing the released memory
@ 2024-06-25  2:13 zhaoxinxin
  2024-06-25 13:51 ` Patrick Robb
  0 siblings, 1 reply; 9+ messages in thread
From: zhaoxinxin @ 2024-06-25  2:13 UTC (permalink / raw)
  To: dev; +Cc: zhaoxinxin, stable

The rte_vhost_driver_unregister() vhost_user_read_cb()
vhost_user_client_reconnect() can be called at the same time by 3 threads.
when memory of vsocket is freed in rte_vhost_driver_unregister(),
then vhost_user_read_cb() maybe add vsocket to reconn_list,
the invalid memory of vsocket is accessed in vhost_user_client_reconnect().
It's a bug for vhost as client.

E.g., vhostuser port is created as client.
Thread 1 calls rte_vhost_driver_unregister() to remove
the vsocket of reconn from the reconn list.
Then “vhost-events” thread calls vhost_user_read_cb() to
add the vsocket of reconn back to the reconn list.
At this time, after thread 1 releases the vsocket memory,
the socket of vhostuser reconnects successfully,
"vhost_reconn" thread will access the released memory.

The core trace is:
Program terminated with signal 11, Segmentation fault.
The fix is to perform a delete operation again after releasing the memory

Fixes: 451dc0f ("vhost: fix crash on port deletion")
Cc: stable@dpdk.org

Signed-off-by: Xinxin Zhao <15957197901@163.com>
---
 lib/vhost/socket.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/lib/vhost/socket.c b/lib/vhost/socket.c
index a75728a2e4..01946096c4 100644
--- a/lib/vhost/socket.c
+++ b/lib/vhost/socket.c
@@ -1121,6 +1121,8 @@ rte_vhost_driver_unregister(const char *path)
 		if (vsocket->is_server) {
 			close(vsocket->socket_fd);
 			unlink(path);
+		} else if (vsocket->reconnect) {
+			vhost_user_remove_reconnect(vsocket);
 		}
 
 		pthread_mutex_destroy(&vsocket->conn_mutex);
-- 
2.45.2


^ permalink raw reply related	[flat|nested] 9+ messages in thread

* [PATCH] vhost: Fix the crash caused by accessing the released memory
@ 2024-06-25  9:31 Xinxin Zhao
  2026-03-31 22:23 ` Stephen Hemminger
  0 siblings, 1 reply; 9+ messages in thread
From: Xinxin Zhao @ 2024-06-25  9:31 UTC (permalink / raw)
  To: dev; +Cc: Xinxin Zhao, stable

The rte_vhost_driver_unregister() vhost_user_read_cb()
vhost_user_client_reconnect() can be called at the same time by 3 threads.
when memory of vsocket is freed in rte_vhost_driver_unregister(),
then vhost_user_read_cb() maybe add vsocket to reconn_list,
the invalid memory of vsocket is accessed in vhost_user_client_reconnect().
It's a bug for vhost as client.

E.g., vhostuser port is created as client.
Thread 1 calls rte_vhost_driver_unregister() to remove
the vsocket of reconn from the reconn list.
Then “vhost-events” thread calls vhost_user_read_cb() to
add the vsocket of reconn back to the reconn list.
At this time, after thread 1 releases the vsocket memory,
the socket of vhostuser reconnects successfully,
"vhost_reconn" thread will access the released memory.

The core trace is:
Program terminated with signal 11, Segmentation fault.
The fix is to perform a delete operation again after releasing the memory

Fixes: 451dc0f ("vhost: fix crash on port deletion")
Cc: stable@dpdk.org

Signed-off-by: Xinxin Zhao <15957197901@163.com>
---
 lib/vhost/socket.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/lib/vhost/socket.c b/lib/vhost/socket.c
index a75728a2e4..01946096c4 100644
--- a/lib/vhost/socket.c
+++ b/lib/vhost/socket.c
@@ -1121,6 +1121,8 @@ rte_vhost_driver_unregister(const char *path)
 		if (vsocket->is_server) {
 			close(vsocket->socket_fd);
 			unlink(path);
+		} else if (vsocket->reconnect) {
+			vhost_user_remove_reconnect(vsocket);
 		}
 
 		pthread_mutex_destroy(&vsocket->conn_mutex);
-- 
2.45.2


^ permalink raw reply related	[flat|nested] 9+ messages in thread

* Re: [PATCH] vhost: Fix the crash caused by accessing the released memory
  2024-06-25  2:13 zhaoxinxin
@ 2024-06-25 13:51 ` Patrick Robb
  0 siblings, 0 replies; 9+ messages in thread
From: Patrick Robb @ 2024-06-25 13:51 UTC (permalink / raw)
  To: zhaoxinxin; +Cc: dev, stable

rerunning the failing performance test in CI testing as it was a false failure.

On Tue, Jun 25, 2024 at 2:39 AM zhaoxinxin <15957197901@163.com> wrote:
>
> The rte_vhost_driver_unregister() vhost_user_read_cb()
> vhost_user_client_reconnect() can be called at the same time by 3 threads.
> when memory of vsocket is freed in rte_vhost_driver_unregister(),
> then vhost_user_read_cb() maybe add vsocket to reconn_list,
> the invalid memory of vsocket is accessed in vhost_user_client_reconnect().
> It's a bug for vhost as client.
>
> E.g., vhostuser port is created as client.
> Thread 1 calls rte_vhost_driver_unregister() to remove
> the vsocket of reconn from the reconn list.
> Then “vhost-events” thread calls vhost_user_read_cb() to
> add the vsocket of reconn back to the reconn list.
> At this time, after thread 1 releases the vsocket memory,
> the socket of vhostuser reconnects successfully,
> "vhost_reconn" thread will access the released memory.
>
> The core trace is:
> Program terminated with signal 11, Segmentation fault.
> The fix is to perform a delete operation again after releasing the memory
>
> Fixes: 451dc0f ("vhost: fix crash on port deletion")
> Cc: stable@dpdk.org
>
> Signed-off-by: Xinxin Zhao <15957197901@163.com>
> ---
>  lib/vhost/socket.c | 2 ++
>  1 file changed, 2 insertions(+)
>
> diff --git a/lib/vhost/socket.c b/lib/vhost/socket.c
> index a75728a2e4..01946096c4 100644
> --- a/lib/vhost/socket.c
> +++ b/lib/vhost/socket.c
> @@ -1121,6 +1121,8 @@ rte_vhost_driver_unregister(const char *path)
>                 if (vsocket->is_server) {
>                         close(vsocket->socket_fd);
>                         unlink(path);
> +               } else if (vsocket->reconnect) {
> +                       vhost_user_remove_reconnect(vsocket);
>                 }
>
>                 pthread_mutex_destroy(&vsocket->conn_mutex);
> --
> 2.45.2
>

^ permalink raw reply	[flat|nested] 9+ messages in thread

* Re: [PATCH] vhost: Fix the crash caused by accessing the released memory
  2024-06-25  9:31 [PATCH] vhost: Fix the crash caused by accessing the released memory Xinxin Zhao
@ 2026-03-31 22:23 ` Stephen Hemminger
  0 siblings, 0 replies; 9+ messages in thread
From: Stephen Hemminger @ 2026-03-31 22:23 UTC (permalink / raw)
  To: Xinxin Zhao; +Cc: dev, stable

On Tue, 25 Jun 2024 17:31:49 +0800
Xinxin Zhao <15957197901@163.com> wrote:

> The rte_vhost_driver_unregister() vhost_user_read_cb()
> vhost_user_client_reconnect() can be called at the same time by 3 threads.
> when memory of vsocket is freed in rte_vhost_driver_unregister(),
> then vhost_user_read_cb() maybe add vsocket to reconn_list,
> the invalid memory of vsocket is accessed in vhost_user_client_reconnect().
> It's a bug for vhost as client.
> 
> E.g., vhostuser port is created as client.
> Thread 1 calls rte_vhost_driver_unregister() to remove
> the vsocket of reconn from the reconn list.
> Then “vhost-events” thread calls vhost_user_read_cb() to
> add the vsocket of reconn back to the reconn list.
> At this time, after thread 1 releases the vsocket memory,
> the socket of vhostuser reconnects successfully,
> "vhost_reconn" thread will access the released memory.
> 
> The core trace is:
> Program terminated with signal 11, Segmentation fault.
> The fix is to perform a delete operation again after releasing the memory
> 
> Fixes: 451dc0f ("vhost: fix crash on port deletion")
> Cc: stable@dpdk.org
> 
> Signed-off-by: Xinxin Zhao <15957197901@163.com>
> ---

This patch looks like something that should get fixed.
But the current patch has some commit message issues.
AI review had some observations 

Review of [PATCH] vhost: Fix the crash caused by accessing the released memory

The use-after-free bug is real and the fix is correctly placed — the
reconnect entry is removed before vsocket is freed, and by this point
fdset_try_del has ensured vhost_user_read_cb can no longer re-add it.

Warning: Commit message says "perform a delete operation again after
  releasing the memory" but the code removes the entry BEFORE freeing.
  The description contradicts the (correct) code.

Warning: Fixes tag uses 7-char hash, DPDK convention is 12:
    Fixes: 451dc0f2d943 ("vhost: fix crash on port deletion")

Warning: Subject should be lowercase after prefix and more concise:
    vhost: fix use-after-free on client reconnect during unregister

Suggest this:

vhost: fix use-after-free on client reconnect during unregister

When vhost operates in client mode with reconnect enabled, three
threads can race: rte_vhost_driver_unregister(), vhost_user_read_cb(),
and vhost_user_client_reconnect(). If vhost_user_read_cb() adds the
vsocket back to the reconnect list after unregister has already
removed it, the reconnect thread will later dereference the freed
vsocket memory.

Fix by calling vhost_user_remove_reconnect() for client sockets
with reconnect enabled during unregister. This is safe because by
this point fdset_try_del has completed for all connection fds,
preventing vhost_user_read_cb from creating new reconnect entries.

Fixes: 451dc0f2d943 ("vhost: fix crash on port deletion")
Cc: stable@dpdk.org

Signed-off-by: Xinxin Zhao <15957197901@163.com>

^ permalink raw reply	[flat|nested] 9+ messages in thread

end of thread, other threads:[~2026-03-31 22:23 UTC | newest]

Thread overview: 9+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2024-06-25  9:31 [PATCH] vhost: Fix the crash caused by accessing the released memory Xinxin Zhao
2026-03-31 22:23 ` Stephen Hemminger
  -- strict thread matches above, loose matches on Subject: below --
2024-06-25  2:13 zhaoxinxin
2024-06-25 13:51 ` Patrick Robb
2024-06-19 12:27 zhaoxinxin
2024-06-24  9:20 ` Maxime Coquelin
2024-06-19  8:39 zhaoxinxin
2024-06-18  5:59 zhaoxinxin
2024-06-17  3:35 zhaoxinxin

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox