Re: [PATCH 3/3] 2.6.0 aacraid driver update

Re: [PATCH 3/3] 2.6.0 aacraid driver update

[Matthew Wilcox]

On Mon, Oct 06, 2003 at 05:48:58PM -0400, Jeff Garzik wrote:
> >+		/*
> >+		 *	Yield the processor in case we are slow 
> >+		 */
> >+		set_current_state(TASK_UNINTERRUPTIBLE);
> >+		schedule_timeout(1);
> 
> hmmm... why not simply call yield() here instead?  I think yield() is 
> closer to the intent you wish to achieve...

Gods, no.  I believe it is always a bug for drivers to call yield()
in 2.6.  What is probably meant here is cond_resched().  I'd support
deleting the EXPORT_SYMBOL(yield) line and fixing the breakage afterwards
as it causes lots of very subtle breakage ("Under certain circumstances,
Linux just stops doing anything for 5 seconds").

-- 
"It's not Hollywood.  War is real, war is primarily not about defeat or
victory, it is about death.  I've seen thousands and thousands of dead bodies.
Do you think I want to have an academic debate on this subject?" -- Robert Fisk

Re: [PATCH 3/3] 2.6.0 aacraid driver update

[Jeff Garzik]

Matthew Wilcox wrote:
> On Mon, Oct 06, 2003 at 05:48:58PM -0400, Jeff Garzik wrote:
> 
>>>+		/*
>>>+		 *	Yield the processor in case we are slow 
>>>+		 */
>>>+		set_current_state(TASK_UNINTERRUPTIBLE);
>>>+		schedule_timeout(1);
>>
>>hmmm... why not simply call yield() here instead?  I think yield() is 
>>closer to the intent you wish to achieve...
> 
> 
> Gods, no.  I believe it is always a bug for drivers to call yield()
> in 2.6.  What is probably meant here is cond_resched().  I'd support
> deleting the EXPORT_SYMBOL(yield) line and fixing the breakage afterwards
> as it causes lots of very subtle breakage ("Under certain circumstances,
> Linux just stops doing anything for 5 seconds").


Yes, you're right, and thank you for the correction.  I was thinking

	if (need_resched)
		schedule();

which I incorrectly translated to yield() when searching my brain for 
the 2.6 equivalent.

	Jeff