Sorry if you just wanted Valve to look at these. However, from what the mdmp is saying (and if I'm understanding it correctly) is the engine jumped to a null location. Since there are absolutely no null checks before a jump occurs (sorry if I'm using incorrect terminology, as well), the server will crash. This is what happened in CSS_crash_srcds.exe_20110104132753_1.dmp, TF2_crash_srcds.exe_20110101130045_1.dmp, and TF2_crash_srcds.exe_20110102154253_1.dmp (only ones I've looked at). Again, this is only valid if I'm understanding the dump correctly, it's trying to jump to 0x0000000.
I filed a bug on AM a while ago, foolishly thinking it was an issue with SDKTools. To keep this brief, it wasn't. For reference, here's the bug that I filed: https://bugs.alliedmods.net/show_bug.cgi?id=4767 There's a SourcePawn test case there that will crash SRCDS (CS:S) near flawlessly every time. I hope this helps, crashes absolutely suck. I can reproduce at least one server freeze on Linux (101% CPU is just delightful), if anyone wants to seriously look into this. I also hope I wasn't incorrect, and sent this potentially misinformation filled mail out. Kyle. On Tue, Jan 4, 2011 at 8:15 AM, Dominik Friedrichs <[email protected]> wrote: > Hey Valve! > > How about fixing these engine.dll crashes that happen almost daily on my > servers: > > http://forlix.org/_res/host/dumps/ > > Sadly there is another crash with CSS that hangs the server and doesnt > generate a dump, but fixing the above would still be great. > > > _______________________________________________ > To unsubscribe, edit your list preferences, or view the list archives, > please visit: > http://list.valvesoftware.com/mailman/listinfo/hlds > _______________________________________________ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds
