Hi, On Mon, Aug 01, 2016 at 12:58:58PM +0200, Christophe Fergeau wrote: > Hey, > > On Sat, Jul 30, 2016 at 12:26:30AM +0200, Victor Toso wrote: > > This patch avoids a race condition. The race happens when the flush > > callback is in idle when we receive a completed transfer status for > > the same file-transfer which will be marked as completed and removed > > from FileTransferOperations hash table. > > Can you expand a bit on the race? Where is the "completed transfer > status" that we receive coming from? From the agent? Or do you just > mean that we have flushed the queue, so we get the expected "completed > transfer" message from the agent, but the flush callback did not get a > chance to run yet because it was scheduled to run in an idle while the > "completed transfer" message is dealt with immediatly? Only agent can send the 'complete transfer' message but in case of error, we mark the xfer-task as completed too and both situations could trigger the race but I only saw the first case; We can see it easily when we are transferring a lot of files at once, the message that we receive from agent is faster then flush callback in idle. I'll try to expand it in v2, thanks! > > Christophe > _______________________________________________ > Spice-devel mailing list > Spice-devel@xxxxxxxxxxxxxxxxxxxxx > https://lists.freedesktop.org/mailman/listinfo/spice-devel _______________________________________________ Spice-devel mailing list Spice-devel@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/spice-devel