http://bugs.winehq.org/show_bug.cgi?id=9916
--- Comment #84 from Lionel Debroux lionel_debroux@yahoo.fr 2008-10-07 04:17:32 --- Created an attachment (id=16519) --> (http://bugs.winehq.org/attachment.cgi?id=16519) Trace of tests on wine-1.1.5-450-ga113178 with warn+heap
I hadn't run the Wine tests since about six months. Yesterday evening, I gave it a try, on the same computer as in March: Asus F3Jv-AS022P with Core 2 Duo @ 2 GHz, 2 GB RAM, Nvidia GeForce 7600 Go, running NVidia proprietary drivers v.100.14.19 under SimplyMEPIS 7 (KDE 3.5.x).
Wine is compiled with -O0 to save on compilation time and generated debug info size. I was in a place without Internet access, so I took the occasion to force wired and WiFi network interfaces down and /etc/resolv.conf empty, like I had done at times since last October.
On the positive side: * there are more successful resolution changes of DirectDraw/Direct3D tests than back in march.
On the negative side: 19 tests failures (though this number doesn't mean too much because some of the failures, e.g. url, are consequences of no network connectivity), 5 crashes, 1 hang: * gdiplus graphics and region tests: both try to read 0xaaaaaaaa, which means they're accessing memory after it was freed. * user32 dde test: read access to 0x00130000. * wininet http test: read access to 0x00000000. * winmm mci test: read access to 0x00130000. * winmm wave test: repeatably hangs for several minutes for me, both with and without warn+heap.
The wininet http test is the only test that crashes both with and without warn+heap.
I hadn't time to do it yesterday evening, it was late, but tonight, I'll focus on providing more detailed traces (trace+wininet; trace+ddeml,trace+msg; etc. ) for the tests that crash.