On Wed, Sep 07, 2011 at 07:20:19PM +0200, Michael Schwendt wrote: > On Wed, 7 Sep 2011 11:00:57 +1000, PH (Peter) wrote: > > > sometimes a +1 after weeks in testing is the only or at least easy way to > > nudge a package into stable. > > > > e.g: https://admin.fedoraproject.org/updates/libXi-1.4.3-2.fc15 > > even with my +1 still not there, and this isn't the only package I've done > > this for. > > | Details > | Don't corrupt memory if the server sends unknown classes in > XIQueryDevice. > > # rpm -qi libXi|grep -A2 ^De > Description : > X.Org X11 libXi runtime library > > # rpm -qd libXi > /usr/share/doc/libXi-1.4.3/COPYING > > # rpm -ql libXi > /usr/lib64/libXi.so.6 > /usr/lib64/libXi.so.6.1.0 > /usr/share/doc/libXi-1.4.3 > /usr/share/doc/libXi-1.4.3/COPYING > > No bug ticket number either. > Your own +1 in bodhi is without a comment. > Assuming I wanted to test this, what would I do?
ok, fair call, should've created a bug for this. This bug is triggered with new features added in the next X server version (that's not even upstream). Corruption happens when the XIQueryDevice reply includes input classes that are not recognised by Xlib. You can't trigger this bug without having the right git trees, but you can verify that the fix doesn't break anything by running "xinput list" or pretty much any application that uses XI2. fwiw, the reason I pushed this in now is because I didn't want anyone to have that library still around when the X server patches hit rawhide. plus, there's the odd chance that proprietary extensions can trigger this, though I don't think any of these exist. Cheers, Peter -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
