Re: [patch 08/11 -mmotm] oom: invoke oom killer for __GFP_NOFAIL

[David Rientjes]

On Mon, 11 May 2009, Minchan Kim wrote:

> I agree KOSAKI's opinion. 
> We already have a different flags. 
> 
>  * __GFP_REPEAT: Try hard to allocate the memory, but the allocation attempt
>  * _might_ fail.  This depends upon the particular VM implementation.
>  *
>  * __GFP_NOFAIL: The VM implementation _must_ retry infinitely: the caller
>  * cannot handle allocation failures.
> 
> When we use __GFP_NOFAIL, we always have to use it carefully.
> If you change the meaning of __GFP_NOFAIL, the intension of them who have been used it carefully  may be lost. It's my concern.
> 

You pointed out yourself that __GFP_NOFAIL allocations can fail by way of 
having alloc_pages() return NULL even without attempting to free memory by 
the oom killer for order > PAGE_ALLOC_COSTLY_ORDER.  The definition you 
posted above is unambiguous to me, it means we must retry infinitely.  And 
that's very stupid if we are going to neglect to free memory by killing a 
task and relying solely on reclaim which may not make any progress.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/