On Wed, Nov 08, 2017 at 07:25:21AM -0500, Frediano Ziglio wrote: > > > > On Wed, Nov 08, 2017 at 10:31:38AM +0100, Christophe Fergeau wrote: > > > Hey, > > > > > > On Wed, Nov 08, 2017 at 08:20:12AM +0000, Frediano Ziglio wrote: > > > > Missing some names cause some debugging messages to be > > > > generated and some of our tests to fail. > > > > > > Something I wanted to check before sending this patch is if > > > the commit adding human-readable channel did not cause a small > > > regression when using spice-server to create a channel whose type is > > > not known. This can happen in somehow valid situations, if > > > spice-protocol has a new channel type, which code tries to use, when it > > > hasn't been added to spice-server yet. I'd even add a test case for that > > > actually. > > > > Hmm, ok, while it would be nicer not to have a warning in these cases, I > > would say it's acceptable for now as this is unlikely to happen in > > practice. > > > > Acked-by: Christophe Fergeau <cfergeau@xxxxxxxxxx> > > > > Christophe > > > > Is not clear the problem. > > Say that we add a channel in the protocol, the server won't use > any channel with the new type. The only condition will have to > try the new channel is receiving a request from the client but > the client should not do such request as the new channel won't > be in the main channel list. So this should happen only if client > is broken of malicious. In this case reds_handle_other_links > is called but reds_find_channel will return NULL and connection > will be terminated. Ah right, I was thinking code could create its RedChannel subclass and hit that, but this is only possible from the testsuite as the needed headers aren't installed. Christophe
Attachment:
signature.asc
Description: PGP signature
_______________________________________________ Spice-devel mailing list Spice-devel@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/spice-devel