Re: linux-next: manual merge of the akpm tree with Linus' tree

From: Mel Gorman
Date: Wed Aug 22 2012 - 05:03:54 EST


On Wed, Aug 22, 2012 at 03:59:41PM +1000, Stephen Rothwell wrote:
> Hi Andrew,
>
> Today's linux-next merge of the akpm tree got a conflict in
> mm/page_alloc.c between commit c67fe3752abe ("mm: compaction: Abort async
> compaction if locks are contended or taking too long") from Linus' tree
> and commit "mm: remove __GFP_NO_KSWAPD" from the akpm tree.
>
> I fixed it up (see below) and can carry the fix as necessary.

Hi Stephen,

I think this will be a temporary problem. In the patch series that contained
"mm: compaction: Abort async compaction if locks are contended or taking
too long" there is a replacement patch for "mm: remove __GFP_NO_KSWAPD"
that handles the conflict. According to the mm-commits mailing list the
replacement patch has been picked up by Andrew but I see the old
versions in linux-next/akpm. It should get resolved the next time
Andrews tree gets pushed out.

> diff --cc mm/page_alloc.c
> index 5b3cc33,cefac39..0000000
> --- a/mm/page_alloc.c
> +++ b/mm/page_alloc.c
> @@@ -2436,11 -2429,10 +2435,10 @@@ rebalance
> /*
> * If compaction is deferred for high-order allocations, it is because
> * sync compaction recently failed. In this is the case and the caller
> - * has requested the system not be heavily disrupted, fail the
> - * allocation now instead of entering direct reclaim
> + * requested a movable allocation that does not heavily disrupt the
> + * system then fail the allocation instead of entering direct reclaim.
> */
> - if ((deferred_compaction || contended_compaction) &&
> - (gfp_mask & __GFP_NO_KSWAPD))
> - if (deferred_compaction)
> ++ if (deferred_compaction || contended_compaction)
> goto nopage;
>
> /* Try direct reclaim and then allocating */

It is not obvious at all but the correct resolution is actually the
following. Andrew should already have the right version.

@@ -2437,7 +2436,7 @@ rebalance:
* system then fail the allocation instead of entering direct reclaim.
*/
if ((deferred_compaction || contended_compaction) &&
- (gfp_mask & __GFP_NO_KSWAPD))
+ (gfp_mask & (__GFP_MOVABLE|__GFP_REPEAT)) == __GFP_MOVABLE)
goto nopage;

/* Try direct reclaim and then allocating */

--
Mel Gorman
SUSE Labs
--
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/