Re: linux-next: manual merge of the percpu tree with the tip tree

[Frederic Weisbecker <fweisbec@gmail.com>]

On Thu, Oct 09, 2014 at 03:50:18PM +1100, Stephen Rothwell wrote:
> Hi all,
> 
> Today's linux-next merge of the percpu tree got a conflict in
> kernel/irq_work.c between commit 76a33061b932 ("irq_work: Force raised
> irq work to run on irq work interrupt") from the tip tree and commit
> 22127e93c587 ("time: Replace __get_cpu_var uses") from the percpu tree.
> 
> I fixed it up (see below) and can carry the fix as necessary (no action
> is required).
> 
> -- 
> Cheers,
> Stephen Rothwell                    sfr@canb.auug.org.au
> 
> diff --cc kernel/irq_work.c
> index 385b85aded19,345d19edcdae..000000000000
> --- a/kernel/irq_work.c
> +++ b/kernel/irq_work.c
> @@@ -113,12 -113,10 +113,12 @@@ bool irq_work_needs_cpu(void
>   {
>   	struct llist_head *raised, *lazy;
>   
> - 	raised = &__get_cpu_var(raised_list);
> - 	lazy = &__get_cpu_var(lazy_list);
> + 	raised = this_cpu_ptr(&raised_list);
> + 	lazy = this_cpu_ptr(&lazy_list);

Ah thanks! The conflict is compile time rather than merge time, thanks
for spotting it!

Should we notify Linus about it? That's certainly something that should
be applied with the percpu tree.

>  -	if (llist_empty(raised) && llist_empty(lazy))
>  -		return false;
>  +
>  +	if (llist_empty(raised) || arch_irq_work_has_interrupt())
>  +		if (llist_empty(lazy))
>  +			return false;
>   
>   	/* All work should have been flushed before going offline */
>   	WARN_ON_ONCE(cpu_is_offline(smp_processor_id()));