proc race (was: Re: HIGH MEMORY access)

From: Manfred Spraul (manfreds@colorfullife.com)
Date: Thu Apr 13 2000 - 12:18:45 EST

chris@muscat.com wrote:
>
> On yet another atempt I managed to
> get an oops out of it:
>

Great, found the problem:

>
> >>EIP; c0128c01 <access_process_vm+129/170> <=====

we are in access_process_vm(), kernel/ptrace.c:

        up(&tsk->mm->mmap_sem);

and tsk->mm is NULL.

> Trace; c016edeb <proc_pid_cmdline+37/3c>
> Trace; c016f22e <proc_info_read+176/2f8>

pro_info_read() calls task_lock(), thus the task structure is locked
into memory.

BUT:

do_exit() calls __exit_mm() before task_lock(), and __exit_mm() sets
tsk->mm to NULL before acquiring task_lock().

How should we fix that? With start_lazy_tlb() [kernel/exit.c], it's
possible to construct all kind of weird races. 

--
	Manfred

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/

This archive was generated by hypermail 2b29 : Sat Apr 15 2000 - 21:00:21 EST