Hi, While testing delay accounting patches in 2.6.18-rc1, we experienced random panics in slab allocation/free routines. Since we have not seen these panics in earlier version we went back and tested the same set of patches in 2.6.17, and they behaved as expected. We ran the same test _without_ delay accounting in 2.6.18-rc1, and saw random panics. Which when tried in 2.6.17, the test did not fail. 2.6.17 kernel ran for more than 20 hrs without any issues. Since the bugs were seen in slab routines, I replaced mm/slab.c in 2.6.18-rc1 with the same file from 2.6.17(with a small change). I did not see any crashes. By adding one patch at a time to 2.6.17's mm/slab.c, I found that the removal patch is the cause of the panic. -------------- [PATCH] lockdep: annotate SLAB code author Ingo Molnar Mon, 3 Jul 2006 07:25:28 +0000 (00:25 -0700) committer Linus Torvalds Mon, 3 Jul 2006 22:27:09 +0000 (15:27 -0700) http://www.kernel.org/git/?p=linux/kernel/git/torvalds/linux-2.6.git;a=blobdiff;h=85c2e03098a7fcf60f9c2f250b0af5a72a0330cf;hp=3936af3445427b6fa71e48dea318ce302b2b0900;hb=2b2d5493e10051694ae3a57ea6a153e3cb4d4488;f=mm/slab.c -------------- To confirm that, i just removed the above patch from 2.6.18-rc1 and the tests ran successfully for more than 2 hrs. (is still running) Attached are the config file, panic stack and mkthreads.c file used to reproduce the problem. Running "while true;do ./mkthreads 1000 100; done" leads the 2.6.18-rc1 kernel to panic in 30-60 minutes. regards, chandra -- ---------------------------------------------------------------------- Chandra Seetharaman | Be careful what you choose.... - sekharan@us.ibm.com | .......you may get it. ----------------------------------------------------------------------