Drivers: hv: don't leak memory in vmbus_establish_gpadl()
authorVitaly Kuznetsov <vkuznets@redhat.com>
Sat, 4 Jun 2016 00:09:24 +0000 (17:09 -0700)
committerWilly Tarreau <w@1wt.eu>
Tue, 20 Jun 2017 12:04:48 +0000 (14:04 +0200)
commit 7cc80c98070ccc7940fc28811c92cca0a681015d upstream.

In some cases create_gpadl_header() allocates submessages but we never
free them.

[sumits] Note for stable:
Upstream commit 4d63763296ab7865a98bc29cc7d77145815ef89f:
(Drivers: hv: get rid of redundant messagecount in create_gpadl_header())
changes the list usage to initialize list header in all cases; that patch
isn't added to stable, so the current patch is modified a little bit from
the upstream commit to check if the list is valid or not.

Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org>
Signed-off-by: Willy Tarreau <w@1wt.eu>
drivers/hv/channel.c

index 05e6a7d13d4e06e39f9cc6b52f347336f645bcf4..90d6885b5825cff545ece3847ae1ebcc83e200a9 100644 (file)
@@ -391,7 +391,7 @@ int vmbus_establish_gpadl(struct vmbus_channel *channel, void *kbuffer,
        struct vmbus_channel_gpadl_header *gpadlmsg;
        struct vmbus_channel_gpadl_body *gpadl_body;
        struct vmbus_channel_msginfo *msginfo = NULL;
-       struct vmbus_channel_msginfo *submsginfo;
+       struct vmbus_channel_msginfo *submsginfo, *tmp;
        u32 msgcount;
        struct list_head *curr;
        u32 next_gpadl_handle;
@@ -453,6 +453,13 @@ cleanup:
        list_del(&msginfo->msglistentry);
        spin_unlock_irqrestore(&vmbus_connection.channelmsg_lock, flags);
 
+       if (msgcount > 1) {
+               list_for_each_entry_safe(submsginfo, tmp, &msginfo->submsglist,
+                        msglistentry) {
+                       kfree(submsginfo);
+               }
+       }
+
        kfree(msginfo);
        return ret;
 }