lkml.org 
[lkml]   [2012]   [Mar]   [6]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [RFC PATCH] checkpatch: Warn on use of yield()
On 03/06/2012 04:45 PM, Peter Zijlstra wrote:
> On Mon, 2012-03-05 at 18:01 -0800, Joe Perches wrote:
>
>> +# check for use of yield()
>> + if ($line =~ /\byield\s*\(\s*\)/ {
>> + WARN("YIELD",
>> + "yield() is deprecated, consider cpu_relax()\n" . $herecurr);
>> + }
>
> Its not deprecated as such, its just a very dangerous and ill considered
> API.
>
> cpu_relax() is not a good substitute suggestion in that its still a busy
> wait and prone to much of the same problems.
>
> The case at hand was a life-lock due to expecting that yield() would run
> another process which it needed in order to complete. Yield() does not
> provide that guarantee.
>
> Looking at fs/ext4/mballoc.c, we have this gem:
>
>
> /*
> * Yield the CPU here so that we don't get soft lockup
> * in non preempt case.
> */
> yield();
>
> This is of course complete crap as well.. I suspect they want
> cond_resched() there. And:
>
> /* let others to free the space */
> yield();
>
> Like said, yield() doesn't guarantee anything like running anybody else,
> does it rely on that? Or is it optimistic?
>
> Another fun user:
>
> void tasklet_kill(struct tasklet_struct *t)
> {
> if (in_interrupt())
> printk("Attempt to kill tasklet from interrupt\n");
>
> while (test_and_set_bit(TASKLET_STATE_SCHED,&t->state)) {
> do {
> yield();
> } while (test_bit(TASKLET_STATE_SCHED,&t->state));
> }
> tasklet_unlock_wait(t);
> clear_bit(TASKLET_STATE_SCHED,&t->state);
> }
>
> The only reason that doesn't explode is because running tasklets is
> non-preemptible, However since they're non-preemptible they shouldn't
> run long and you might as well busy spin. If they can run long, yield()
> isn't your biggest problem.
>
> mm/memory_hotplug.c has two yield() calls in offline_pages() and I've no
> idea what they're trying to achieve.
>
> But really, yield() is basically _always_ the wrong thing. The right
> thing can be:
>
> cond_resched(); wait_event(); or something entirely different.
>
> So instead of suggesting an alternative, I would suggest thinking about
> the actual problem in order to avoid the non-thinking solutions the
> checkpatch brigade is so overly fond of :/
>
> Maybe something like:
>
> "yield() is dangerous and wrong, rework your code to not use it."
>
> That at least requires some sort of thinking and doesn't suggest blind
> substitution.
>

Can't we point people to some Documentation file that explains the
alternatives?


\
 
 \ /
  Last update: 2012-03-06 14:17    [W:0.096 / U:0.556 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site