Messages in this thread | | | Date | Thu, 5 Sep 2002 01:45:02 -0700 | From | William Lee Irwin III <> | Subject | Re: [BUG] __write_lock_failed() oops |
| |
On Thu, Sep 05, 2002 at 01:42:52AM -0700, Andrew Morton wrote: >> That's all the assembly hacks in the rwlock code not having proper >> stack frames. You may have to ksymoops it. >> At a guess: use-after-free bug against an address_space. You may >> be able to catch it with slab poisoning.
On Thu, Sep 05, 2002 at 01:32:40AM -0700, William Lee Irwin III wrote: > (gdb) p/x $eax > $25 = 0xc0331ca0 > (gdb) p &tasklist_lock > $27 = (rwlock_t *) 0xc0331ca0
The NMI oopser is going here as well (nmi_watchdog=2 for extra safety) so I suspect the tasklist_lock semantics are behaving badly. But it's not easily reproducible enough to test a quick attempt at a fix if it can't be recognized a priori.
This is literally so difficult to reproduce it hasn't been seen in 2 releases. kgdb is still going and dhowells is helping me fish stuff off the stack.
Cheers, Bill - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
| |