From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7.5) Gecko/20041107 Firefox/1.0 Description of problem: kernel panic in ptrace Version-Release number of selected component (if applicable): How reproducible: Couldn't Reproduce Steps to Reproduce: 1. run gdb 2. 3. Actual Results: kernel panic Additional info: This happens a few times per week under heavy debug load.
Created attachment 114247 [details] kernel dump
Created attachment 114248 [details] Proposed fix. Yeah, looks obvious. We don't do the lazy fpu saving on SMP machines, and we object violently to being asked to save the FPU for a non-current task.
Hm. Actually I'm unconvinced the above is anything more than a workaround. The BUG() will only trigger if MSR_FP is enabled for the child, and that really shouldn't be happening.
*** Bug 157433 has been marked as a duplicate of this bug. ***
Fixing kernel version and assigning.
Created attachment 114293 [details] Upstream fix. This is the real fix. The problem we weren't correctly waiting for the inferior to be taken off its CPU. Its register set really was in some other CPU, not in the place we were looking for it, and that's why the FPU was still marked as enabled.
*** Bug 157831 has been marked as a duplicate of this bug. ***