[prev in list] [next in list] [prev in thread] [next in thread] 

List:       cairo
Subject:    [cairo] [PATCH V3 0/4] fix replay race condition.
From:       zhigang.gong () linux ! intel ! com
Date:       2012-07-11 9:06:11
Message-ID: 1341997575-4702-1-git-send-email-zhigang.gong () linux ! intel ! com
[Download RAW message or body]

From: Zhigang Gong <zhigang.gong@linux.intel.com>

Rebase the patchset to latest git master. Ping to comments.
Thx.

---

This version fixed most of the comments from Uli, including
remove those dead code and use mutex rather than normal lock
to reduce the duplicate code.

For the locking mechanism, this version remove the detction
of "self-copy" in previou version. As before enter the replaying
function, we already did one time flush on the target surface.
Just in case, put a assert there.

As to the dead lock issue on those platform without rwlock
implementation, I still haven't found a good idea. Behdad suggest
to use mutexes to implement rwlock, but I failed to find a way
to do that without introduce at least one extra platform dependent
API, such as thread_self. So I gave it up currently, and leave it
to someone who are interested to add native rwlock implementation
for those platforms.

For the v2 test case, just remove some space errors. any further
comments are welcome.
-- 
cairo mailing list
cairo@cairographics.org
http://lists.cairographics.org/mailman/listinfo/cairo
[prev in list] [next in list] [prev in thread] [next in thread]