Created attachment 17862 [details] [review] [PATCH 1/9] Make sure a pending call timeout isn't assumed. From: Scott James Remnant <scott@netsplit.com>
Created attachment 17863 [details] [review] [PATCH 2/9] Allow a pending call to block forever
Some discussion context for these patches: http://lists.freedesktop.org/archives/dbus/2008-July/010125.html
Created attachment 17864 [details] [review] [PATCH 3/9] Don't allocate DBusTimeout for pending call when passed INT_MAX
Created attachment 17865 [details] [review] [PATCH 4/9] Update documentation now that INT_MAX means no timeout.
Created attachment 17866 [details] [review] [PATCH 5/9] Fix issue where timeouts can overflow.
Created attachment 17867 [details] [review] [PATCH 6/9] Remove 6 hour timeout restriction.
Created attachment 17868 [details] [review] [PATCH 7/9] Explicitly check for zero time fields.
Created attachment 17869 [details] [review] [PATCH 8/9] Expire list timeout may be negative for no expiry.
Created attachment 17870 [details] [review] [PATCH 9/9] Change default reply timeout.
To state the obvious, a lot of changes here. Do you have any thoughts on extending the test suite to cover some of this? I guess a tricky part about that would be simulating timeout values while not actually making the developer runnig "make check" wait hours =)
For archaeology purposes, links to mailing list discussion: http://lists.freedesktop.org/archives/dbus/2008-July/010125.html It sounded like Havoc agreed on these. David, what do you think at a high level?
(In reply to comment #11) > For archaeology purposes, links to mailing list discussion: > > http://lists.freedesktop.org/archives/dbus/2008-July/010125.html > > It sounded like Havoc agreed on these. David, what do you think at a high > level? > Sorry for the lag - yes, I think using INT_MAX to mean infinite timeout would be sane... David
Created attachment 25769 [details] [review] [PATCH 01/11] Make sure a pending call timeout isn't assumed.
Created attachment 25770 [details] [review] [PATCH 02/11] Allow a pending call to block forever
Created attachment 25771 [details] [review] [PATCH 03/11] Don't allocate DBusTimeout for pending call when passed INT_MAX
Created attachment 25772 [details] [review] [PATCH 04/11] Update documentation now that INT_MAX means no timeout.
Created attachment 25773 [details] [review] [PATCH 05/11] Fix issue where timeouts can overflow.
Created attachment 25774 [details] [review] [PATCH 06/11] Remove 6 hour timeout restriction.
Created attachment 25775 [details] [review] [PATCH 07/11] Explicitly check for zero time fields.
Created attachment 25776 [details] [review] [PATCH 08/11] Expire list timeout may be negative for no expiry.
Created attachment 25777 [details] [review] [PATCH 09/11] Change default reply timeout.
Created attachment 25778 [details] [review] [PATCH 10/11] Unrestrict session bus timeout.
Created attachment 25779 [details] [review] [PATCH 11/11] Add tests for pending call timeouts
Refreshed the patch set. The only change was to the first patch, which had a pretty obvious think-o. I also added a 10th patch to unrestrict the reply timeout on the session bus (the conf was intended to raise it, but with these patches it lowers it) and at Colin's suggestion, I've added some pending call tests that at least follow the code path.
This all looks sane to me, I think it's fine to commit.
Pushed to GIT.
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.