GNOME Bugzilla – Bug 685072
memory: map(READ)/unmap clears the READONLY status
Last modified: 2012-09-28 23:37:33 UTC
Created attachment 225359 [details] [review] Patch to check the readonly flag on lock Currently, if one locks and unlocks a GstMiniObject in read-mode, the read-only status is cleared.. As read-only is only a read lock with a count of one.. After discussion with Wim, let's explicitly check the readonly flag.
ok, commit!
Created attachment 225360 [details] [review] miniobject: Make is_writable() only refer to the mini object, not it's content In the same vein, there was a confusion on what writable was for a mini-object. Let's try to clarify it
(In reply to comment #2) > Created an attachment (id=225360) [details] [review] > miniobject: Make is_writable() only refer to the mini object, not it's content > > In the same vein, there was a confusion on what writable was for a mini-object. > Let's try to clarify it This is wrong, is_writable() uses the exclusive counter when enabled and the refcount otherwise. This method should be called has_exclusive_access() or something.
Comment on attachment 225359 [details] [review] Patch to check the readonly flag on lock commit fd846a7c64faa263bcaff355e9a504b69afc6d73 Author: Olivier Crête <olivier.crete@collabora.com> Date: Fri Sep 28 15:17:27 2012 -0400 miniobject: Always reject WRITE locks on READONLY miniobjects Verify that mapping a read-only memory as read doesnt make it writable
Committed per Wim's command!