Sasha reports various nasty trinity crashes when shmem_fault() tries to finish_wait(), we guess from rare cases when the wait_queue_head on shmem_fallocate()'s stack has already gone. Fix those by using TASK_UNINTERRUPIBLE instead of TASK_KILLABLE in prepare_to_wait(), that's much simpler and safer: TASK_KILLABLE was a nice aspiration, but not worth any more hassle. Reported-and-tested-by: Sasha Levin <sasha.levin@xxxxxxxxxx> Signed-off-by: Hugh Dickins <hughd@xxxxxxxxxx> --- Andrew, please fold this into shmem-fix-faulting-into-a-hole-not-taking-i_mutex.patch before sending the fixes on to Linus - thanks. mm/shmem.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) --- mmotm/mm/shmem.c 2014-07-22 16:35:49.683985586 -0700 +++ linux/mm/shmem.c 2014-07-22 16:36:35.459984108 -0700 @@ -1283,7 +1283,7 @@ static int shmem_fault(struct vm_area_st shmem_falloc_waitq = shmem_falloc->waitq; prepare_to_wait(shmem_falloc_waitq, &shmem_fault_wait, - TASK_KILLABLE); + TASK_UNINTERRUPTIBLE); spin_unlock(&inode->i_lock); schedule(); -- To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
