The patch titled Subject: mm, pagealloc: micro-optimisation: save two branches on hot page allocation path has been removed from the -mm tree. Its filename was mm-micro-optimisation-save-two-branches-on-hot-page-allocation-path.patch This patch was dropped because it was merged into mainline or a subsystem tree ------------------------------------------------------ From: Mateusz Nosek <mateusznosek0@xxxxxxxxx> Subject: mm, pagealloc: micro-optimisation: save two branches on hot page allocation path This patch makes ALLOC_KSWAPD equal to __GFP_KSWAPD_RECLAIM (cast to int). Thanks to that code like: if (gfp_mask & __GFP_KSWAPD_RECLAIM) alloc_flags |= ALLOC_KSWAPD; can be changed to: alloc_flags |= (__force int) (gfp_mask &__GFP_KSWAPD_RECLAIM); Thanks to this one branch less is generated in the assembly. In case of ALLOC_KSWAPD flag two branches are saved, first one in code that always executes in the beginning of page allocation and the second one in loop in page allocator slowpath. Link: http://lkml.kernel.org/r/20200304162118.14784-1-mateusznosek0@xxxxxxxxx Signed-off-by: Mateusz Nosek <mateusznosek0@xxxxxxxxx> Acked-by: Vlastimil Babka <vbabka@xxxxxxx> Acked-by: Mel Gorman <mgorman@xxxxxxxxxxxxxxxxxxx> Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> --- mm/internal.h | 2 +- mm/page_alloc.c | 22 ++++++++++++++-------- 2 files changed, 15 insertions(+), 9 deletions(-) --- a/mm/internal.h~mm-micro-optimisation-save-two-branches-on-hot-page-allocation-path +++ a/mm/internal.h @@ -555,7 +555,7 @@ unsigned long reclaim_clean_pages_from_l #else #define ALLOC_NOFRAGMENT 0x0 #endif -#define ALLOC_KSWAPD 0x200 /* allow waking of kswapd */ +#define ALLOC_KSWAPD 0x800 /* allow waking of kswapd, __GFP_KSWAPD_RECLAIM set */ enum ttu_flags; struct tlbflush_unmap_batch; --- a/mm/page_alloc.c~mm-micro-optimisation-save-two-branches-on-hot-page-allocation-path +++ a/mm/page_alloc.c @@ -3536,10 +3536,13 @@ static bool zone_allows_reclaim(struct z static inline unsigned int alloc_flags_nofragment(struct zone *zone, gfp_t gfp_mask) { - unsigned int alloc_flags = 0; + unsigned int alloc_flags; - if (gfp_mask & __GFP_KSWAPD_RECLAIM) - alloc_flags |= ALLOC_KSWAPD; + /* + * __GFP_KSWAPD_RECLAIM is assumed to be the same as ALLOC_KSWAPD + * to save a branch. + */ + alloc_flags = (__force int) (gfp_mask & __GFP_KSWAPD_RECLAIM); #ifdef CONFIG_ZONE_DMA32 if (!zone) @@ -4175,8 +4178,13 @@ gfp_to_alloc_flags(gfp_t gfp_mask) { unsigned int alloc_flags = ALLOC_WMARK_MIN | ALLOC_CPUSET; - /* __GFP_HIGH is assumed to be the same as ALLOC_HIGH to save a branch. */ + /* + * __GFP_HIGH is assumed to be the same as ALLOC_HIGH + * and __GFP_KSWAPD_RECLAIM is assumed to be the same as ALLOC_KSWAPD + * to save two branches. + */ BUILD_BUG_ON(__GFP_HIGH != (__force gfp_t) ALLOC_HIGH); + BUILD_BUG_ON(__GFP_KSWAPD_RECLAIM != (__force gfp_t) ALLOC_KSWAPD); /* * The caller may dip into page reserves a bit more if the caller @@ -4184,7 +4192,8 @@ gfp_to_alloc_flags(gfp_t gfp_mask) * policy or is asking for __GFP_HIGH memory. GFP_ATOMIC requests will * set both ALLOC_HARDER (__GFP_ATOMIC) and ALLOC_HIGH (__GFP_HIGH). */ - alloc_flags |= (__force int) (gfp_mask & __GFP_HIGH); + alloc_flags |= (__force int) + (gfp_mask & (__GFP_HIGH | __GFP_KSWAPD_RECLAIM)); if (gfp_mask & __GFP_ATOMIC) { /* @@ -4201,9 +4210,6 @@ gfp_to_alloc_flags(gfp_t gfp_mask) } else if (unlikely(rt_task(current)) && !in_interrupt()) alloc_flags |= ALLOC_HARDER; - if (gfp_mask & __GFP_KSWAPD_RECLAIM) - alloc_flags |= ALLOC_KSWAPD; - #ifdef CONFIG_CMA if (gfpflags_to_migratetype(gfp_mask) == MIGRATE_MOVABLE) alloc_flags |= ALLOC_CMA; _ Patches currently in -mm which might be from mateusznosek0@xxxxxxxxx are mm-shmemc-clean-code-by-removing-unnecessary-assignment.patch mm-mm_initc-clean-code-use-build_bug_on-when-comparing-compile-time-constant.patch mm-dmapoolc-micro-optimisation-remove-unnecessary-branch.patch