Skip to contentSkip to conversation

mm: vmalloc: Refactor vm_area_alloc_pages() function

PATCH Applied Uladzislau Rezki (Sony) 3 participants 6 messages Aug 27, 2024 – Aug 29, 2024

Subsystems (2)
lore
Share & tools
lore

Uladzislau Rezki (Sony)

author maintainer
#0 lore raw

PATCH 1/1 mm: vmalloc: Refactor vm_area_alloc_pages() function

The aim is to simplify and making the vm_area_alloc_pages() function less confusing as it became more clogged nowadays:

- eliminate a "bulk_gfp" variable and do not overwrite a gfp flag for bulk allocator;
- drop __GFP_NOFAIL flag for high-order-page requests on upper layer. It becomes less spread between levels when it comes to __GFP_NOFAIL allocations;
- add a comment about a fallback path if high-order attempt is unsuccessful because for such cases __GFP_NOFAIL is dropped;
- fix a typo in a commit message.

---
 mm/vmalloc.c | 37 +++++++++++++++++--------------------
 1 file changed, 17 insertions(+), 20 deletions(-)
mm/vmalloc.c +17 -20
index 3f9b6bd707d2..57862865e808 100644
@@ -3531,8 +3531,6 @@ vm_area_alloc_pages(gfp_t gfp, int nid,
 		unsigned int order, unsigned int nr_pages, struct page **pages)
 {
 	unsigned int nr_allocated = 0;
-	gfp_t alloc_gfp = gfp;
-	bool nofail = gfp & __GFP_NOFAIL;
 	struct page *page;
 	int i;
 
@@ -3543,9 +3541,6 @@ vm_area_alloc_pages(gfp_t gfp, int nid,
 	 * more permissive.
 	 */
 	if (!order) {
-		/* bulk allocator doesn't support nofail req. officially */
-		gfp_t bulk_gfp = gfp & ~__GFP_NOFAIL;
-
 		while (nr_allocated < nr_pages) {
 			unsigned int nr, nr_pages_request;
 
@@ -3563,12 +3558,11 @@ vm_area_alloc_pages(gfp_t gfp, int nid,
 			 * but mempolicy wants to alloc memory by interleaving.
 			 */
 			if (IS_ENABLED(CONFIG_NUMA) && nid == NUMA_NO_NODE)
-				nr = alloc_pages_bulk_array_mempolicy_noprof(bulk_gfp,
+				nr = alloc_pages_bulk_array_mempolicy_noprof(gfp,
 							nr_pages_request,
 							pages + nr_allocated);
-
 			else
-				nr = alloc_pages_bulk_array_node_noprof(bulk_gfp, nid,
+				nr = alloc_pages_bulk_array_node_noprof(gfp, nid,
 							nr_pages_request,
 							pages + nr_allocated);
 
@@ -3582,30 +3576,24 @@ vm_area_alloc_pages(gfp_t gfp, int nid,
 			if (nr != nr_pages_request)
 				break;
 		}
-	} else if (gfp & __GFP_NOFAIL) {
-		/*
-		 * Higher order nofail allocations are really expensive and
-		 * potentially dangerous (pre-mature OOM, disruptive reclaim
-		 * and compaction etc.
-		 */
-		alloc_gfp &= ~__GFP_NOFAIL;
 	}
 
 	/* High-order pages or fallback path if "bulk" fails. */
 	while (nr_allocated < nr_pages) {
-		if (!nofail && fatal_signal_pending(current))
+		if (!(gfp & __GFP_NOFAIL) && fatal_signal_pending(current))
 			break;
 
 		if (nid == NUMA_NO_NODE)
-			page = alloc_pages_noprof(alloc_gfp, order);
+			page = alloc_pages_noprof(gfp, order);
 		else
-			page = alloc_pages_node_noprof(nid, alloc_gfp, order);
+			page = alloc_pages_node_noprof(nid, gfp, order);
+
 		if (unlikely(!page))
 			break;
 
 		/*
 		 * Higher order allocations must be able to be treated as
-		 * indepdenent small pages by callers (as they can with
+		 * independent small pages by callers (as they can with
 		 * small-page vmallocs). Some drivers do their own refcounting
 		 * on vmalloc_to_page() pages, some use page->mapping,
 		 * page->lru, etc.
@@ -3666,7 +3654,16 @@ static void *__vmalloc_area_node(struct vm_struct *area, gfp_t gfp_mask,
 	set_vm_area_page_order(area, page_shift - PAGE_SHIFT);
 	page_order = vm_area_page_order(area);
 
-	area->nr_pages = vm_area_alloc_pages(gfp_mask | __GFP_NOWARN,
+	/*
+	 * Higher order nofail allocations are really expensive and
+	 * potentially dangerous (pre-mature OOM, disruptive reclaim
+	 * and compaction etc.
+	 *
+	 * Please note, the __vmalloc_node_range_noprof() falls-back
+	 * to order-0 pages if high-order attempt is unsuccessful.
+	 */
+	area->nr_pages = vm_area_alloc_pages((page_order ?
+		gfp_mask & ~__GFP_NOFAIL : gfp_mask) | __GFP_NOWARN,
 		node, page_order, nr_small_pages, area->pages);
 
 	atomic_long_add(area->nr_pages, &nr_vmalloc_pages);
Signature · 1 line
2.39.2
Acked-by on v1 1/1
#1 lore raw
Quoted 12 lines · Uladzislau Rezki

The aim is to simplify and making the vm_area_alloc_pages() function less confusing as it became more clogged nowadays:

- eliminate a "bulk_gfp" variable and do not overwrite a gfp flag for bulk allocator;
- drop __GFP_NOFAIL flag for high-order-page requests on upper layer. It becomes less spread between levels when it comes to __GFP_NOFAIL allocations;
- add a comment about a fallback path if high-order attempt is unsuccessful because for such cases __GFP_NOFAIL is dropped;
- fix a typo in a commit message.

Quoted 1 line

Thanks!

Quoted 98 lines
---
 mm/vmalloc.c | 37 +++++++++++++++++--------------------
 1 file changed, 17 insertions(+), 20 deletions(-)
mm/vmalloc.c +17 -20
index 3f9b6bd707d2..57862865e808 100644
@@ -3531,8 +3531,6 @@ vm_area_alloc_pages(gfp_t gfp, int nid,
 		unsigned int order, unsigned int nr_pages, struct page **pages)
 {
 	unsigned int nr_allocated = 0;
-	gfp_t alloc_gfp = gfp;
-	bool nofail = gfp & __GFP_NOFAIL;
 	struct page *page;
 	int i;
 
@@ -3543,9 +3541,6 @@ vm_area_alloc_pages(gfp_t gfp, int nid,
 	 * more permissive.
 	 */
 	if (!order) {
-		/* bulk allocator doesn't support nofail req. officially */
-		gfp_t bulk_gfp = gfp & ~__GFP_NOFAIL;
-
 		while (nr_allocated < nr_pages) {
 			unsigned int nr, nr_pages_request;
 
@@ -3563,12 +3558,11 @@ vm_area_alloc_pages(gfp_t gfp, int nid,
 			 * but mempolicy wants to alloc memory by interleaving.
 			 */
 			if (IS_ENABLED(CONFIG_NUMA) && nid == NUMA_NO_NODE)
-				nr = alloc_pages_bulk_array_mempolicy_noprof(bulk_gfp,
+				nr = alloc_pages_bulk_array_mempolicy_noprof(gfp,
 							nr_pages_request,
 							pages + nr_allocated);
-
 			else
-				nr = alloc_pages_bulk_array_node_noprof(bulk_gfp, nid,
+				nr = alloc_pages_bulk_array_node_noprof(gfp, nid,
 							nr_pages_request,
 							pages + nr_allocated);
 
@@ -3582,30 +3576,24 @@ vm_area_alloc_pages(gfp_t gfp, int nid,
 			if (nr != nr_pages_request)
 				break;
 		}
-	} else if (gfp & __GFP_NOFAIL) {
-		/*
-		 * Higher order nofail allocations are really expensive and
-		 * potentially dangerous (pre-mature OOM, disruptive reclaim
-		 * and compaction etc.
-		 */
-		alloc_gfp &= ~__GFP_NOFAIL;
 	}
 
 	/* High-order pages or fallback path if "bulk" fails. */
 	while (nr_allocated < nr_pages) {
-		if (!nofail && fatal_signal_pending(current))
+		if (!(gfp & __GFP_NOFAIL) && fatal_signal_pending(current))
 			break;
 
 		if (nid == NUMA_NO_NODE)
-			page = alloc_pages_noprof(alloc_gfp, order);
+			page = alloc_pages_noprof(gfp, order);
 		else
-			page = alloc_pages_node_noprof(nid, alloc_gfp, order);
+			page = alloc_pages_node_noprof(nid, gfp, order);
+
 		if (unlikely(!page))
 			break;
 
 		/*
 		 * Higher order allocations must be able to be treated as
-		 * indepdenent small pages by callers (as they can with
+		 * independent small pages by callers (as they can with
 		 * small-page vmallocs). Some drivers do their own refcounting
 		 * on vmalloc_to_page() pages, some use page->mapping,
 		 * page->lru, etc.
@@ -3666,7 +3654,16 @@ static void *__vmalloc_area_node(struct vm_struct *area, gfp_t gfp_mask,
 	set_vm_area_page_order(area, page_shift - PAGE_SHIFT);
 	page_order = vm_area_page_order(area);
 
-	area->nr_pages = vm_area_alloc_pages(gfp_mask | __GFP_NOWARN,
+	/*
+	 * Higher order nofail allocations are really expensive and
+	 * potentially dangerous (pre-mature OOM, disruptive reclaim
+	 * and compaction etc.
+	 *
+	 * Please note, the __vmalloc_node_range_noprof() falls-back
+	 * to order-0 pages if high-order attempt is unsuccessful.
+	 */
+	area->nr_pages = vm_area_alloc_pages((page_order ?
+		gfp_mask & ~__GFP_NOFAIL : gfp_mask) | __GFP_NOWARN,
 		node, page_order, nr_small_pages, area->pages);
 
 	atomic_long_add(area->nr_pages, &nr_vmalloc_pages);
Signature · 1 line
2.39.2
Signature · 2 lines
Michal Hocko
SUSE Labs
Reviewed-by on v1 1/1
#2 lore raw
In reply to Uladzislau Rezki (Sony)
Quoted 92 lines · Uladzislau Rezki (Sony)

The aim is to simplify and making the vm_area_alloc_pages() function less confusing as it became more clogged nowadays:

- eliminate a "bulk_gfp" variable and do not overwrite a gfp flag for bulk allocator;
- drop __GFP_NOFAIL flag for high-order-page requests on upper layer. It becomes less spread between levels when it comes to __GFP_NOFAIL allocations;
- add a comment about a fallback path if high-order attempt is unsuccessful because for such cases __GFP_NOFAIL is dropped;
- fix a typo in a commit message.

---
 mm/vmalloc.c | 37 +++++++++++++++++--------------------
 1 file changed, 17 insertions(+), 20 deletions(-)
mm/vmalloc.c +7 -19
index 3f9b6bd707d2..57862865e808 100644
@@ -3531,8 +3531,6 @@ vm_area_alloc_pages(gfp_t gfp, int nid,
 		unsigned int order, unsigned int nr_pages, struct page **pages)
 {
 	unsigned int nr_allocated = 0;
-	gfp_t alloc_gfp = gfp;
-	bool nofail = gfp & __GFP_NOFAIL;
 	struct page *page;
 	int i;
 
@@ -3543,9 +3541,6 @@ vm_area_alloc_pages(gfp_t gfp, int nid,
 	 * more permissive.
 	 */
 	if (!order) {
-		/* bulk allocator doesn't support nofail req. officially */
-		gfp_t bulk_gfp = gfp & ~__GFP_NOFAIL;
-
 		while (nr_allocated < nr_pages) {
 			unsigned int nr, nr_pages_request;
 
@@ -3563,12 +3558,11 @@ vm_area_alloc_pages(gfp_t gfp, int nid,
 			 * but mempolicy wants to alloc memory by interleaving.
 			 */
 			if (IS_ENABLED(CONFIG_NUMA) && nid == NUMA_NO_NODE)
-				nr = alloc_pages_bulk_array_mempolicy_noprof(bulk_gfp,
+				nr = alloc_pages_bulk_array_mempolicy_noprof(gfp,
 							nr_pages_request,
 							pages + nr_allocated);
-
 			else
-				nr = alloc_pages_bulk_array_node_noprof(bulk_gfp, nid,
+				nr = alloc_pages_bulk_array_node_noprof(gfp, nid,
 							nr_pages_request,
 							pages + nr_allocated);
 
@@ -3582,30 +3576,24 @@ vm_area_alloc_pages(gfp_t gfp, int nid,
 			if (nr != nr_pages_request)
 				break;
 		}
-	} else if (gfp & __GFP_NOFAIL) {
-		/*
-		 * Higher order nofail allocations are really expensive and
-		 * potentially dangerous (pre-mature OOM, disruptive reclaim
-		 * and compaction etc.
-		 */
-		alloc_gfp &= ~__GFP_NOFAIL;
 	}
 
 	/* High-order pages or fallback path if "bulk" fails. */
 	while (nr_allocated < nr_pages) {
-		if (!nofail && fatal_signal_pending(current))
+		if (!(gfp & __GFP_NOFAIL) && fatal_signal_pending(current))
 			break;
 
 		if (nid == NUMA_NO_NODE)
-			page = alloc_pages_noprof(alloc_gfp, order);
+			page = alloc_pages_noprof(gfp, order);
 		else
-			page = alloc_pages_node_noprof(nid, alloc_gfp, order);
+			page = alloc_pages_node_noprof(nid, gfp, order);
+
 		if (unlikely(!page))
 			break;
 
 		/*
 		 * Higher order allocations must be able to be treated as
-		 * indepdenent small pages by callers (as they can with
+		 * independent small pages by callers (as they can with
 		 * small-page vmallocs). Some drivers do their own refcounting
 		 * on vmalloc_to_page() pages, some use page->mapping,
 		 * page->lru, etc.

mm/vmalloc.c static void *__vmalloc_area_node(struct vm_struct *area, gfp_t gfp_mask, · line 3658

Quoted 6 lines
mm/vmalloc.c +2 -1
@@ -3666,7 +3654,16 @@ static void *__vmalloc_area_node(struct vm_struct *area, gfp_t gfp_mask,
 	set_vm_area_page_order(area, page_shift - PAGE_SHIFT);
 	page_order = vm_area_page_order(area);
 
-	area->nr_pages = vm_area_alloc_pages(gfp_mask | __GFP_NOWARN,
+	/*
+	 * Higher order nofail allocations are really expensive and
           ~~~~~~~~~~~~

Seems we use both higher-order and high-order to describe the non 0-order pages in many places. I personally would take high-order, higher-order seems to be a little confusing because it's not explicit what is compared with and lower.

Surely this is not an issue to this patch, I see a lot of 'higher order' in kernel codes.

For this patch,

Quoted 14 lines
mm/vmalloc.c +8 -0
+	 * potentially dangerous (pre-mature OOM, disruptive reclaim
+	 * and compaction etc.
+	 *
+	 * Please note, the __vmalloc_node_range_noprof() falls-back
+	 * to order-0 pages if high-order attempt is unsuccessful.
+	 */
+	area->nr_pages = vm_area_alloc_pages((page_order ?
+		gfp_mask & ~__GFP_NOFAIL : gfp_mask) | __GFP_NOWARN,
 		node, page_order, nr_small_pages, area->pages);
 
 	atomic_long_add(area->nr_pages, &nr_vmalloc_pages);
Signature · 1 line
2.39.2
author maintainer
#3 lore raw
Quoted 100 lines · Baoquan He
Quoted 92 lines · Uladzislau Rezki (Sony)

The aim is to simplify and making the vm_area_alloc_pages() function less confusing as it became more clogged nowadays:

- eliminate a "bulk_gfp" variable and do not overwrite a gfp flag for bulk allocator;
- drop __GFP_NOFAIL flag for high-order-page requests on upper layer. It becomes less spread between levels when it comes to __GFP_NOFAIL allocations;
- add a comment about a fallback path if high-order attempt is unsuccessful because for such cases __GFP_NOFAIL is dropped;
- fix a typo in a commit message.

---
 mm/vmalloc.c | 37 +++++++++++++++++--------------------
 1 file changed, 17 insertions(+), 20 deletions(-)
mm/vmalloc.c +7 -19
index 3f9b6bd707d2..57862865e808 100644
@@ -3531,8 +3531,6 @@ vm_area_alloc_pages(gfp_t gfp, int nid,
 		unsigned int order, unsigned int nr_pages, struct page **pages)
 {
 	unsigned int nr_allocated = 0;
-	gfp_t alloc_gfp = gfp;
-	bool nofail = gfp & __GFP_NOFAIL;
 	struct page *page;
 	int i;
 
@@ -3543,9 +3541,6 @@ vm_area_alloc_pages(gfp_t gfp, int nid,
 	 * more permissive.
 	 */
 	if (!order) {
-		/* bulk allocator doesn't support nofail req. officially */
-		gfp_t bulk_gfp = gfp & ~__GFP_NOFAIL;
-
 		while (nr_allocated < nr_pages) {
 			unsigned int nr, nr_pages_request;
 
@@ -3563,12 +3558,11 @@ vm_area_alloc_pages(gfp_t gfp, int nid,
 			 * but mempolicy wants to alloc memory by interleaving.
 			 */
 			if (IS_ENABLED(CONFIG_NUMA) && nid == NUMA_NO_NODE)
-				nr = alloc_pages_bulk_array_mempolicy_noprof(bulk_gfp,
+				nr = alloc_pages_bulk_array_mempolicy_noprof(gfp,
 							nr_pages_request,
 							pages + nr_allocated);
-
 			else
-				nr = alloc_pages_bulk_array_node_noprof(bulk_gfp, nid,
+				nr = alloc_pages_bulk_array_node_noprof(gfp, nid,
 							nr_pages_request,
 							pages + nr_allocated);
 
@@ -3582,30 +3576,24 @@ vm_area_alloc_pages(gfp_t gfp, int nid,
 			if (nr != nr_pages_request)
 				break;
 		}
-	} else if (gfp & __GFP_NOFAIL) {
-		/*
-		 * Higher order nofail allocations are really expensive and
-		 * potentially dangerous (pre-mature OOM, disruptive reclaim
-		 * and compaction etc.
-		 */
-		alloc_gfp &= ~__GFP_NOFAIL;
 	}
 
 	/* High-order pages or fallback path if "bulk" fails. */
 	while (nr_allocated < nr_pages) {
-		if (!nofail && fatal_signal_pending(current))
+		if (!(gfp & __GFP_NOFAIL) && fatal_signal_pending(current))
 			break;
 
 		if (nid == NUMA_NO_NODE)
-			page = alloc_pages_noprof(alloc_gfp, order);
+			page = alloc_pages_noprof(gfp, order);
 		else
-			page = alloc_pages_node_noprof(nid, alloc_gfp, order);
+			page = alloc_pages_node_noprof(nid, gfp, order);
+
 		if (unlikely(!page))
 			break;
 
 		/*
 		 * Higher order allocations must be able to be treated as
-		 * indepdenent small pages by callers (as they can with
+		 * independent small pages by callers (as they can with
 		 * small-page vmallocs). Some drivers do their own refcounting
 		 * on vmalloc_to_page() pages, some use page->mapping,
 		 * page->lru, etc.

mm/vmalloc.c static void *__vmalloc_area_node(struct vm_struct *area, gfp_t gfp_mask, · line 3658

Quoted 6 lines
mm/vmalloc.c +2 -1
@@ -3666,7 +3654,16 @@ static void *__vmalloc_area_node(struct vm_struct *area, gfp_t gfp_mask,
 	set_vm_area_page_order(area, page_shift - PAGE_SHIFT);
 	page_order = vm_area_page_order(area);
 
-	area->nr_pages = vm_area_alloc_pages(gfp_mask | __GFP_NOWARN,
+	/*
+	 * Higher order nofail allocations are really expensive and
           ~~~~~~~~~~~~
Quoted 8 lines

Seems we use both higher-order and high-order to describe the non 0-order pages in many places. I personally would take high-order, higher-order seems to be a little confusing because it's not explicit what is compared with and lower.

Surely this is not an issue to this patch, I see a lot of 'higher order' in kernel codes.

I agree. It sounds like hard to figure out the difference between both. Are you willing send the patch? If not, i can send it out :)

Quoted 4 lines

For this patch,

Thanks!

--
Uladzislau Rezki

#4 lore raw
Quoted 106 lines · Uladzislau Rezki
Quoted 105 lines · Baoquan He
Quoted 92 lines · Uladzislau Rezki (Sony)

The aim is to simplify and making the vm_area_alloc_pages() function less confusing as it became more clogged nowadays:

- eliminate a "bulk_gfp" variable and do not overwrite a gfp flag for bulk allocator;
- drop __GFP_NOFAIL flag for high-order-page requests on upper layer. It becomes less spread between levels when it comes to __GFP_NOFAIL allocations;
- add a comment about a fallback path if high-order attempt is unsuccessful because for such cases __GFP_NOFAIL is dropped;
- fix a typo in a commit message.

---
 mm/vmalloc.c | 37 +++++++++++++++++--------------------
 1 file changed, 17 insertions(+), 20 deletions(-)
mm/vmalloc.c +7 -19
index 3f9b6bd707d2..57862865e808 100644
@@ -3531,8 +3531,6 @@ vm_area_alloc_pages(gfp_t gfp, int nid,
 		unsigned int order, unsigned int nr_pages, struct page **pages)
 {
 	unsigned int nr_allocated = 0;
-	gfp_t alloc_gfp = gfp;
-	bool nofail = gfp & __GFP_NOFAIL;
 	struct page *page;
 	int i;
 
@@ -3543,9 +3541,6 @@ vm_area_alloc_pages(gfp_t gfp, int nid,
 	 * more permissive.
 	 */
 	if (!order) {
-		/* bulk allocator doesn't support nofail req. officially */
-		gfp_t bulk_gfp = gfp & ~__GFP_NOFAIL;
-
 		while (nr_allocated < nr_pages) {
 			unsigned int nr, nr_pages_request;
 
@@ -3563,12 +3558,11 @@ vm_area_alloc_pages(gfp_t gfp, int nid,
 			 * but mempolicy wants to alloc memory by interleaving.
 			 */
 			if (IS_ENABLED(CONFIG_NUMA) && nid == NUMA_NO_NODE)
-				nr = alloc_pages_bulk_array_mempolicy_noprof(bulk_gfp,
+				nr = alloc_pages_bulk_array_mempolicy_noprof(gfp,
 							nr_pages_request,
 							pages + nr_allocated);
-
 			else
-				nr = alloc_pages_bulk_array_node_noprof(bulk_gfp, nid,
+				nr = alloc_pages_bulk_array_node_noprof(gfp, nid,
 							nr_pages_request,
 							pages + nr_allocated);
 
@@ -3582,30 +3576,24 @@ vm_area_alloc_pages(gfp_t gfp, int nid,
 			if (nr != nr_pages_request)
 				break;
 		}
-	} else if (gfp & __GFP_NOFAIL) {
-		/*
-		 * Higher order nofail allocations are really expensive and
-		 * potentially dangerous (pre-mature OOM, disruptive reclaim
-		 * and compaction etc.
-		 */
-		alloc_gfp &= ~__GFP_NOFAIL;
 	}
 
 	/* High-order pages or fallback path if "bulk" fails. */
 	while (nr_allocated < nr_pages) {
-		if (!nofail && fatal_signal_pending(current))
+		if (!(gfp & __GFP_NOFAIL) && fatal_signal_pending(current))
 			break;
 
 		if (nid == NUMA_NO_NODE)
-			page = alloc_pages_noprof(alloc_gfp, order);
+			page = alloc_pages_noprof(gfp, order);
 		else
-			page = alloc_pages_node_noprof(nid, alloc_gfp, order);
+			page = alloc_pages_node_noprof(nid, gfp, order);
+
 		if (unlikely(!page))
 			break;
 
 		/*
 		 * Higher order allocations must be able to be treated as
-		 * indepdenent small pages by callers (as they can with
+		 * independent small pages by callers (as they can with
 		 * small-page vmallocs). Some drivers do their own refcounting
 		 * on vmalloc_to_page() pages, some use page->mapping,
 		 * page->lru, etc.

mm/vmalloc.c static void *__vmalloc_area_node(struct vm_struct *area, gfp_t gfp_mask, · line 3658

Quoted 6 lines
mm/vmalloc.c +2 -1
@@ -3666,7 +3654,16 @@ static void *__vmalloc_area_node(struct vm_struct *area, gfp_t gfp_mask,
 	set_vm_area_page_order(area, page_shift - PAGE_SHIFT);
 	page_order = vm_area_page_order(area);
 
-	area->nr_pages = vm_area_alloc_pages(gfp_mask | __GFP_NOWARN,
+	/*
+	 * Higher order nofail allocations are really expensive and
           ~~~~~~~~~~~~

Seems we use both higher-order and high-order to describe the non 0-order pages in many places. I personally would take high-order, higher-order seems to be a little confusing because it's not explicit what is compared with and lower.

Quoted 5 lines
Quoted 3 lines

Surely this is not an issue to this patch, I see a lot of 'higher order' in kernel codes.

I agree. It sounds like hard to figure out the difference between both. Are you willing send the patch? If not, i can send it out :)

I am fine, please go ahead.

author maintainer
#5 lore raw
Quoted 107 lines · Baoquan He
Quoted 106 lines · Uladzislau Rezki
Quoted 105 lines · Baoquan He
Quoted 92 lines · Uladzislau Rezki (Sony)

The aim is to simplify and making the vm_area_alloc_pages() function less confusing as it became more clogged nowadays:

- eliminate a "bulk_gfp" variable and do not overwrite a gfp flag for bulk allocator;
- drop __GFP_NOFAIL flag for high-order-page requests on upper layer. It becomes less spread between levels when it comes to __GFP_NOFAIL allocations;
- add a comment about a fallback path if high-order attempt is unsuccessful because for such cases __GFP_NOFAIL is dropped;
- fix a typo in a commit message.

---
 mm/vmalloc.c | 37 +++++++++++++++++--------------------
 1 file changed, 17 insertions(+), 20 deletions(-)
mm/vmalloc.c +7 -19
index 3f9b6bd707d2..57862865e808 100644
@@ -3531,8 +3531,6 @@ vm_area_alloc_pages(gfp_t gfp, int nid,
 		unsigned int order, unsigned int nr_pages, struct page **pages)
 {
 	unsigned int nr_allocated = 0;
-	gfp_t alloc_gfp = gfp;
-	bool nofail = gfp & __GFP_NOFAIL;
 	struct page *page;
 	int i;
 
@@ -3543,9 +3541,6 @@ vm_area_alloc_pages(gfp_t gfp, int nid,
 	 * more permissive.
 	 */
 	if (!order) {
-		/* bulk allocator doesn't support nofail req. officially */
-		gfp_t bulk_gfp = gfp & ~__GFP_NOFAIL;
-
 		while (nr_allocated < nr_pages) {
 			unsigned int nr, nr_pages_request;
 
@@ -3563,12 +3558,11 @@ vm_area_alloc_pages(gfp_t gfp, int nid,
 			 * but mempolicy wants to alloc memory by interleaving.
 			 */
 			if (IS_ENABLED(CONFIG_NUMA) && nid == NUMA_NO_NODE)
-				nr = alloc_pages_bulk_array_mempolicy_noprof(bulk_gfp,
+				nr = alloc_pages_bulk_array_mempolicy_noprof(gfp,
 							nr_pages_request,
 							pages + nr_allocated);
-
 			else
-				nr = alloc_pages_bulk_array_node_noprof(bulk_gfp, nid,
+				nr = alloc_pages_bulk_array_node_noprof(gfp, nid,
 							nr_pages_request,
 							pages + nr_allocated);
 
@@ -3582,30 +3576,24 @@ vm_area_alloc_pages(gfp_t gfp, int nid,
 			if (nr != nr_pages_request)
 				break;
 		}
-	} else if (gfp & __GFP_NOFAIL) {
-		/*
-		 * Higher order nofail allocations are really expensive and
-		 * potentially dangerous (pre-mature OOM, disruptive reclaim
-		 * and compaction etc.
-		 */
-		alloc_gfp &= ~__GFP_NOFAIL;
 	}
 
 	/* High-order pages or fallback path if "bulk" fails. */
 	while (nr_allocated < nr_pages) {
-		if (!nofail && fatal_signal_pending(current))
+		if (!(gfp & __GFP_NOFAIL) && fatal_signal_pending(current))
 			break;
 
 		if (nid == NUMA_NO_NODE)
-			page = alloc_pages_noprof(alloc_gfp, order);
+			page = alloc_pages_noprof(gfp, order);
 		else
-			page = alloc_pages_node_noprof(nid, alloc_gfp, order);
+			page = alloc_pages_node_noprof(nid, gfp, order);
+
 		if (unlikely(!page))
 			break;
 
 		/*
 		 * Higher order allocations must be able to be treated as
-		 * indepdenent small pages by callers (as they can with
+		 * independent small pages by callers (as they can with
 		 * small-page vmallocs). Some drivers do their own refcounting
 		 * on vmalloc_to_page() pages, some use page->mapping,
 		 * page->lru, etc.

mm/vmalloc.c static void *__vmalloc_area_node(struct vm_struct *area, gfp_t gfp_mask, · line 3658

Quoted 6 lines
mm/vmalloc.c +2 -1
@@ -3666,7 +3654,16 @@ static void *__vmalloc_area_node(struct vm_struct *area, gfp_t gfp_mask,
 	set_vm_area_page_order(area, page_shift - PAGE_SHIFT);
 	page_order = vm_area_page_order(area);
 
-	area->nr_pages = vm_area_alloc_pages(gfp_mask | __GFP_NOWARN,
+	/*
+	 * Higher order nofail allocations are really expensive and
           ~~~~~~~~~~~~

Seems we use both higher-order and high-order to describe the non 0-order pages in many places. I personally would take high-order, higher-order seems to be a little confusing because it's not explicit what is compared with and lower.

Quoted 8 lines
Quoted 5 lines
Quoted 3 lines

Surely this is not an issue to this patch, I see a lot of 'higher order' in kernel codes.

I agree. It sounds like hard to figure out the difference between both. Are you willing send the patch? If not, i can send it out :)

I am fine, please go ahead.

Good! I will fix it.

--
Uladzislau Rezki