Skip to contentSkip to conversation

migration/rdma: Fix qemu_rdma_cleanup null check

PATCH In review Dr. David Alan Gilbert (git) 3 participants 6 messages Feb 14, 2019 – Mar 5, 2019

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 migration/rdma: Fix qemu_rdma_cleanup null check

From: "Dr. David Alan Gilbert" <[email protected]>

If the migration fails before the channel is open (e.g. a bad address) we end up in the cleanup with rdma->channel==NULL.

Spotted by Coverity: CID 1398634

Fixes fbbaacab2758cb3f32a0
---
 migration/rdma.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
migration/rdma.c +3 -1
index 54a3c11540..9fa3b176eb 100644
@@ -2321,7 +2321,9 @@ static void qemu_rdma_cleanup(RDMAContext *rdma)
         rdma->connected = false;
     }
 
-    qemu_set_fd_handler(rdma->channel->fd, NULL, NULL, NULL);
+    if (rdma->channel) {
+        qemu_set_fd_handler(rdma->channel->fd, NULL, NULL, NULL);
+    }
     g_free(rdma->dest_blocks);
     rdma->dest_blocks = NULL;
 
Signature · 1 line
2.20.1
#1 lore raw
Quoted 17 lines · Dr. David Alan Gilbert (git)

From: "Dr. David Alan Gilbert" <[email protected]>

If the migration fails before the channel is open (e.g. a bad address) we end up in the cleanup with rdma->channel==NULL.

Spotted by Coverity: CID 1398634

Fixes fbbaacab2758cb3f32a0
---
 migration/rdma.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
migration/rdma.c +0 -0
index 54a3c11540..9fa3b176eb 100644

migration/rdma.c static void qemu_rdma_cleanup(RDMAContext *rdma) · line 2326

Quoted 7 lines
migration/rdma.c +3 -1
@@ -2321,7 +2321,9 @@ static void qemu_rdma_cleanup(RDMAContext *rdma)
         rdma->connected = false;
     }
 
-    qemu_set_fd_handler(rdma->channel->fd, NULL, NULL, NULL);
+    if (rdma->channel) {
+        qemu_set_fd_handler(rdma->channel->fd, NULL, NULL, NULL);
+    }

IIUC there's no strict ordering constraint on resetting the fd handler, then how about simply moving this line into the below "if (rdma->channel)" altogether?

Regards,

Signature · 1 line
Peter Xu
#2 lore raw
Quoted 21 lines · * Peter Xu ([email protected]) wrote:
Quoted 20 lines · Dr. David Alan Gilbert (git)

From: "Dr. David Alan Gilbert" <[email protected]>

If the migration fails before the channel is open (e.g. a bad address) we end up in the cleanup with rdma->channel==NULL.

Spotted by Coverity: CID 1398634

Fixes fbbaacab2758cb3f32a0
---
 migration/rdma.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
migration/rdma.c +0 -0
index 54a3c11540..9fa3b176eb 100644
@@ -2321,7 +2321,9 @@ static void qemu_rdma_cleanup(RDMAContext *rdma)
         rdma->connected = false;
     }
 
Quoted 8 lines

migration/rdma.c static void qemu_rdma_cleanup(RDMAContext *rdma) · line 2326

Quoted 4 lines
migration/rdma.c +3 -1
-    qemu_set_fd_handler(rdma->channel->fd, NULL, NULL, NULL);
+    if (rdma->channel) {
+        qemu_set_fd_handler(rdma->channel->fd, NULL, NULL, NULL);
+    }

IIUC there's no strict ordering constraint on resetting the fd handler, then how about simply moving this line into the below "if (rdma->channel)" altogether?

The logic around the closing of the return path makes that check later a bit messy; rdma->channel can get set to Null before the other check.

Dave

Quoted 4 lines

Regards,

Signature · 1 line
Peter Xu

--
Dr. David Alan Gilbert / [email protected] / Manchester, UK

Reviewed-by on v1 1/1
#4 lore raw
In reply to Dr. David Alan Gilbert (git)
Quoted 18 lines · Dr. David Alan Gilbert (git)

From: "Dr. David Alan Gilbert" <[email protected]>

If the migration fails before the channel is open (e.g. a bad address) we end up in the cleanup with rdma->channel==NULL.

Spotted by Coverity: CID 1398634

Fixes fbbaacab2758cb3f32a0
---
 migration/rdma.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
migration/rdma.c +0 -0
index 54a3c11540..9fa3b176eb 100644

migration/rdma.c static void qemu_rdma_cleanup(RDMAContext *rdma) · line 2329

Quoted 10 lines
migration/rdma.c +3 -1
@@ -2321,7 +2321,9 @@ static void qemu_rdma_cleanup(RDMAContext *rdma)
         rdma->connected = false;
     }
 
-    qemu_set_fd_handler(rdma->channel->fd, NULL, NULL, NULL);
+    if (rdma->channel) {
+        qemu_set_fd_handler(rdma->channel->fd, NULL, NULL, NULL);
+    }
     g_free(rdma->dest_blocks);
     rdma->dest_blocks = NULL;
 
Reviewed-by
#3 lore raw
In reply to Dr. David Alan Gilbert
Quoted 31 lines · Dr. David Alan Gilbert
Quoted 21 lines · * Peter Xu ([email protected]) wrote:
Quoted 20 lines · Dr. David Alan Gilbert (git)

From: "Dr. David Alan Gilbert" <[email protected]>

If the migration fails before the channel is open (e.g. a bad address) we end up in the cleanup with rdma->channel==NULL.

Spotted by Coverity: CID 1398634

Fixes fbbaacab2758cb3f32a0
---
 migration/rdma.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
migration/rdma.c +0 -0
index 54a3c11540..9fa3b176eb 100644
@@ -2321,7 +2321,9 @@ static void qemu_rdma_cleanup(RDMAContext *rdma)
         rdma->connected = false;
     }
 
Quoted 8 lines

migration/rdma.c static void qemu_rdma_cleanup(RDMAContext *rdma) · line 2326

Quoted 4 lines
migration/rdma.c +3 -1
-    qemu_set_fd_handler(rdma->channel->fd, NULL, NULL, NULL);
+    if (rdma->channel) {
+        qemu_set_fd_handler(rdma->channel->fd, NULL, NULL, NULL);
+    }

IIUC there's no strict ordering constraint on resetting the fd handler, then how about simply moving this line into the below "if (rdma->channel)" altogether?

Quoted 2 lines

The logic around the closing of the return path makes that check later a bit messy; rdma->channel can get set to Null before the other check.

Ah I see, it's the mess by sharing listen_id and channel on destination side... Maybe we can clean them up along the way? I gave it a shot:

    if (rdma->listen_id && rdma->is_return_path) {
        /*
         * The return path on the destination side, both listen_id and
         * channel are shared with the other context so we skip
         * freeing those but simply clear the pointers no matter what.
         * The main context will help us to clean these.
         */
        rdma->listen_id = NULL;
        rdma->channel = NULL;
    } else {
        /*
         * Either the source side, or the main context of the
         * destination side: we are responsible for listen_id/channel
         */
        if (rdma->listen_id) {
            rdma_destroy_id(rdma->listen_id);
            rdma->listen_id = NULL;
        }
        if (rdma->channel) {
            qemu_set_fd_handler(rdma->channel->fd, NULL, NULL, NULL);
            rdma_destroy_event_channel(rdma->channel);
            rdma->channel = NULL;
        }
    }

I slightly prefer to clean it up (if someone is still going to maintain the RDMA code... :), but either way is fine to me.

No matter what you prefer:

Thanks,

Signature · 1 line
Peter Xu
#5 lore raw
In reply to Dr. David Alan Gilbert (git)
Quoted 8 lines · * Dr. David Alan Gilbert (git) ([email protected]) wrote:

From: "Dr. David Alan Gilbert" <[email protected]>

If the migration fails before the channel is open (e.g. a bad address) we end up in the cleanup with rdma->channel==NULL.

Spotted by Coverity: CID 1398634

Fixes fbbaacab2758cb3f32a0

Queued

Quoted 23 lines
---
 migration/rdma.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
migration/rdma.c +3 -1
index 54a3c11540..9fa3b176eb 100644
@@ -2321,7 +2321,9 @@ static void qemu_rdma_cleanup(RDMAContext *rdma)
         rdma->connected = false;
     }
 
-    qemu_set_fd_handler(rdma->channel->fd, NULL, NULL, NULL);
+    if (rdma->channel) {
+        qemu_set_fd_handler(rdma->channel->fd, NULL, NULL, NULL);
+    }
     g_free(rdma->dest_blocks);
     rdma->dest_blocks = NULL;
 
Signature · 1 line
2.20.1

--
Dr. David Alan Gilbert / [email protected] / Manchester, UK