Re: [2.6.30-rc1] RCU detected CPU 1 stall

[Hugh Dickins]

On Thu, 16 Apr 2009, Tetsuo Handa wrote:
> Indeed, we can't use "argv" put on stack memory if we use "no wait" mode.
> We need to ensure that memory region pointed by "argv" is valid until
> do_execve() finishes. However,
> | 
> | exit:
> | 	kfree(devpath);
> | 	kfree(env);
> | 	return retval;
> | }
> "env" can be overwritten as well as "argv".
> We need to ensure that memory region pointed by "env" is valid until
> do_execve() finishes, don't we?
> 
> I think we need to use UMH_WAIT_EXEC instead of UMH_NO_WAIT.

Yes, of course you're right, silly me.

It could be worked around using call_usermodehelper_setcleanup()
to arrange for the freeing there instead of here in this case;
but if CONFIG_UEVENT_HELPER_PATH="" is the better answer anyway,
let's not waste any more time on this, just revert the broken patch
(which is what simply changing UMH_NO_WAIT to UMH_WAIT_EXEC does).

I'd hoped to get confirmation that the mmap.c part of the patch
was all the extra needed; but never mind, that's clearly good.

Thanks a lot: I'll send in a revert request,
and the acct_stack_growth patch, later today.

Hugh
--
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/