diff options
| author | Dan Carpenter <dan.carpenter@linaro.org> | 2023-05-03 17:11:35 +0200 |
|---|---|---|
| committer | Juergen Gross <jgross@suse.com> | 2023-05-24 17:25:43 +0200 |
| commit | 8fafac202d18230bb9926bda48e563fd2cce2a4f (patch) | |
| tree | 7260271387adf7aab269fe3c833b889945ae57e5 /drivers/xen | |
| parent | xen/blkfront: Only check REQ_FUA for writes (diff) | |
| download | linux-8fafac202d18230bb9926bda48e563fd2cce2a4f.tar.xz linux-8fafac202d18230bb9926bda48e563fd2cce2a4f.zip | |
xen/pvcalls-back: fix double frees with pvcalls_new_active_socket()
In the pvcalls_new_active_socket() function, most error paths call
pvcalls_back_release_active(fedata->dev, fedata, map) which calls
sock_release() on "sock". The bug is that the caller also frees sock.
Fix this by making every error path in pvcalls_new_active_socket()
release the sock, and don't free it in the caller.
Fixes: 5db4d286a8ef ("xen/pvcalls: implement connect command")
Signed-off-by: Dan Carpenter <dan.carpenter@linaro.org>
Reviewed-by: Juergen Gross <jgross@suse.com>
Link: https://lore.kernel.org/r/e5f98dc2-0305-491f-a860-71bbd1398a2f@kili.mountain
Signed-off-by: Juergen Gross <jgross@suse.com>
Diffstat (limited to 'drivers/xen')
| -rw-r--r-- | drivers/xen/pvcalls-back.c | 9 |
1 files changed, 4 insertions, 5 deletions
diff --git a/drivers/xen/pvcalls-back.c b/drivers/xen/pvcalls-back.c index 1f5219e12cc3..7beaf2c41fbb 100644 --- a/drivers/xen/pvcalls-back.c +++ b/drivers/xen/pvcalls-back.c @@ -325,8 +325,10 @@ static struct sock_mapping *pvcalls_new_active_socket( void *page; map = kzalloc(sizeof(*map), GFP_KERNEL); - if (map == NULL) + if (map == NULL) { + sock_release(sock); return NULL; + } map->fedata = fedata; map->sock = sock; @@ -418,10 +420,8 @@ static int pvcalls_back_connect(struct xenbus_device *dev, req->u.connect.ref, req->u.connect.evtchn, sock); - if (!map) { + if (!map) ret = -EFAULT; - sock_release(sock); - } out: rsp = RING_GET_RESPONSE(&fedata->ring, fedata->ring.rsp_prod_pvt++); @@ -561,7 +561,6 @@ static void __pvcalls_back_accept(struct work_struct *work) sock); if (!map) { ret = -EFAULT; - sock_release(sock); goto out_error; } |