Re: kmap_atomic and preemption

[Vineet Gupta <Vineet.Gupta1@synopsys.com>]

On Wednesday 04 May 2016 07:23 PM, Thomas Petazzoni wrote:
> Hello,
>
> On Wed, 4 May 2016 15:47:29 +0200, Peter Zijlstra wrote:
>
>> static inline void *kmap_atomic(struct page *page)
>> {
>> 	preempt_disable();
>> 	pagefault_disable();
>> 	if (!PageHighMem(page))
>> 		return page_address(page);
>>
>> 	return __kmap_atomic(page);
>> }
> This is essentially what has been done on ARM in commit
> 9ff0bb5ba60638a688a46e93df8c5009896672eb, showing a pretty significant
> improvement in network workloads.

ARC already has that semantically - only not inline ! I really want to avoid 2
needless LD-ADD-ST for the disabling of preemption and page fault for the low mem
pages by returning early !

static inline void *kmap_atomic(struct page *page)
{
        if (!PageHighMem(page))
		return page_address(page);

	preempt_disable();
	pagefault_disable();
	
	return __kmap_atomic(page);

}

>
> Best regards,
>
> Thomas


--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>