Re: [RFC] propagate gfp_t to page table alloc functions

[Minchan Kim <minchan@kernel.org>]

On 04/27/2012 07:43 PM, David Rientjes wrote:

> On Fri, 27 Apr 2012, Minchan Kim wrote:
> 
>>> Maybe a per-thread_info variant of gfp_allowed_mask?  So Andrew's 
>>> set_current_gfp() becomes set_current_gfp_allowed() that does
>>>
>>> 	void set_current_gfp_allowed(gfp_t gfp_mask)
>>> 	{
>>> 		current->gfp_allowed = gfp_mask & gfp_allowed_mask;
>>> 	}
>>>
>>> and then the page allocator does
>>>
>>> 	gfp_mask &= current->gfp_allowed;
>>>
>>> rather than how it currently does
>>>
>>> 	gfp_mask &= gfp_allowed_mask;
>>>
>>> and then the caller of set_current_gfp_allowed() cleans up with 
>>> set_current_gfp_allowed(__GFP_BITS_MASK).
>>
> 
> [trimmed the newsgroups from the reply, not sure what the point is?]
> 
>> Caller should restore old gfp_mask instead of __GFP_BITS_MASK in case of
>> nesting.And how do we care of atomic context?
>>
> 
> Eek, I'm hoping these aren't going to be nested but sure that seems 
> appropraite if they are.  (I'm also hoping these will only be either 
> __GFP_HIGH or __GFP_BITS_MASK and no other combinations.)
> 
> Forcing atomic context would just be set_current_gfp_allowed(__GFP_HIGH).


I mean it's not legal to access _current_ in atomic context so that
(gfp_mask &= current->gfp_allowed in page allocator) shouldn't.

-- 
Kind regards,
Minchan Kim

--
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/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>