Re: How is Code in do_sys_settimeofday() safe in case of SMP andNest Kernel Path?

From: Nick Piggin
Date: Sat Sep 30 2006 - 10:38:44 EST


Dong Feng wrote:
2006/9/30, Christoph Lameter <clameter@xxxxxxx>:

On Fri, 29 Sep 2006, Dong Feng wrote:


> If the comments mean the subsequent code is SMP-safe and can prevent
> nest-kernel-path, how does it achieves that?

It relies on locking outside of do_sys_settimeofday(). Seems that this
indicates locking is to be performed by the arch before calling
do_sys_settimeofday. Looks suspicious to me. Check that this function is
always called with the same lock.


Yes, that is the question. The whole invocation path is
sys_settimeofday() -> do_sys_settimeofday()

I do not find a lock embracing do_sys_settimeofday().

Moreover, seems neither write operations nor read operations on sys_tz
is protected by any locks, in sys_gettimeofday() and
sys_settimeofday() respectively.

Did you get to the bottom of this yet? It looks like you're right,
and I suggest a seqlock might be a good option.

You should write a patch and send it to Mister Morton.

--
SUSE Labs, Novell Inc.
Send instant messages to your online friends http://au.messenger.yahoo.com -
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/