Am Freitag, 9. November 2007 03:26:52 schrieb Dan Kegel:
Data online at http://kegel.com/wine/valgrind/20071108/
For some reason, today I had to do 'wineserver -k' about five times (yesterday and the day before only once). And one of those times was during a d3d9 test which took over / hung the screen; not even ctl-alt-backspace helped, I actually had to reboot (I didn't try ssh'ing in). Running the tests was like pulling teeth.
That said, here are the changes in the list of bad tests;
- means test didn't show any valgrind errors today but did yesterday,
- means it did show valgrind errors today but not yesterday.
Do you have time to do a regression test for the patch that broke the d3d9 test? The most suspicious patch is b4b10d8bc2d6c7a9191898d3a03420bdcae9c6f2 . Finding the patch might give some hints to the X server crashes on Mesa drivers.
The d3d9:visual failures are gone, but I think this is because the test didn't run to completion. This patch fixes a memleak, but a few other memleaks are still left.