Skip to contentSkip to conversation

vmalloc enhancements

PATCH Superseded Roman Gushchin 4 participants 7 messages Dec 19, 2018 – Feb 11, 2019

Subsystems (4)
Memory management MM core Proc filesystem vmalloc

v1 of 4 Superseded Latest is v4 →

lore
Share & tools
lore

Roman Gushchin

author
#0 lore raw

cover vmalloc enhancements

The patchset contains few changes to the vmalloc code, which are leading to some performance gains and code simplification.

Also, it exports a number of pages, used by vmalloc(), in /proc/meminfo.

Patch (1) removes some redundancy on __vunmap(). Patch (2) separates memory allocation and data initialization
in alloc_vmap_area()
Patch (3) adds vmalloc counter to /proc/meminfo.

RFC->v1:
- removed bogus empty lines (suggested by Matthew Wilcox)
- made nr_vmalloc_pages static (suggested by Matthew Wilcox)
- dropped patch 3 from RFC patchset, will post later with some other changes
- dropped RFC

Roman Gushchin (3):
mm: refactor __vunmap() to avoid duplicated call to find_vm_area()
mm: separate memory allocation and actual work in alloc_vmap_area()
mm: show number of vmalloc pages in /proc/meminfo

 fs/proc/meminfo.c       |   2 +-
 include/linux/vmalloc.h |   2 +
 mm/vmalloc.c            | 107 ++++++++++++++++++++++++++--------------
 3 files changed, 73 insertions(+), 38 deletions(-)
Signature · 1 line
2.19.2
author
#1 lore raw

PATCH 1/3 mm: refactor __vunmap() to avoid duplicated call to find_vm_area()

__vunmap() calls find_vm_area() twice without an obvious reason:
first directly to get the area pointer, second indirectly by calling remove_vm_area(), which is again searching for the area.

To remove this redundancy, let's split remove_vm_area() into __remove_vm_area(struct vmap_area *), which performs the actual area removal, and remove_vm_area(const void *addr) wrapper, which can be used everywhere, where it has been used before.

On my test setup, I've got up to 12% speed up on vfree()'ing 1000000 of 4-pages vmalloc blocks.

Signed-off-by Roman Gushchin <[email protected]>
Acked-by Johannes Weiner <[email protected]>
Reviewed-by Matthew Wilcox <[email protected]>
---
 mm/vmalloc.c | 47 +++++++++++++++++++++++++++--------------------
 1 file changed, 27 insertions(+), 20 deletions(-)
mm/vmalloc.c +27 -20
index 871e41c55e23..7660e3ef4133 100644
@@ -1462,6 +1462,24 @@ struct vm_struct *find_vm_area(const void *addr)
 	return NULL;
 }
 
+static struct vm_struct *__remove_vm_area(struct vmap_area *va)
+{
+	struct vm_struct *vm = va->vm;
+
+	might_sleep();
+
+	spin_lock(&vmap_area_lock);
+	va->vm = NULL;
+	va->flags &= ~VM_VM_AREA;
+	va->flags |= VM_LAZY_FREE;
+	spin_unlock(&vmap_area_lock);
+
+	kasan_free_shadow(vm);
+	free_unmap_vmap_area(va);
+
+	return vm;
+}
+
 /**
  *	remove_vm_area  -  find and remove a continuous kernel virtual area
  *	@addr:		base address

@@ -1472,31 +1490,20 @@ struct vm_struct *find_vm_area(const void *addr)
  */
 struct vm_struct *remove_vm_area(const void *addr)
 {
+	struct vm_struct *vm = NULL;
 	struct vmap_area *va;
 
-	might_sleep();
-
 	va = find_vmap_area((unsigned long)addr);
-	if (va && va->flags & VM_VM_AREA) {
-		struct vm_struct *vm = va->vm;
-
-		spin_lock(&vmap_area_lock);
-		va->vm = NULL;
-		va->flags &= ~VM_VM_AREA;
-		va->flags |= VM_LAZY_FREE;
-		spin_unlock(&vmap_area_lock);
-
-		kasan_free_shadow(vm);
-		free_unmap_vmap_area(va);
+	if (va && va->flags & VM_VM_AREA)
+		vm = __remove_vm_area(va);
 
-		return vm;
-	}
-	return NULL;
+	return vm;
 }
 
 static void __vunmap(const void *addr, int deallocate_pages)
 {
 	struct vm_struct *area;
+	struct vmap_area *va;
 
 	if (!addr)
 		return;

@@ -1505,17 +1512,18 @@ static void __vunmap(const void *addr, int deallocate_pages)
 			addr))
 		return;
 
-	area = find_vmap_area((unsigned long)addr)->vm;
-	if (unlikely(!area)) {
+	va = find_vmap_area((unsigned long)addr);
+	if (unlikely(!va || !va->vm)) {
 		WARN(1, KERN_ERR "Trying to vfree() nonexistent vm area (%p)\n",
 				addr);
 		return;
 	}
 
+	area = va->vm;
 	debug_check_no_locks_freed(area->addr, get_vm_area_size(area));
 	debug_check_no_obj_freed(area->addr, get_vm_area_size(area));
 
-	remove_vm_area(addr);
+	__remove_vm_area(va);
 	if (deallocate_pages) {
 		int i;
 

@@ -1530,7 +1538,6 @@ static void __vunmap(const void *addr, int deallocate_pages)
 	}
 
 	kfree(area);
-	return;
 }
 
 static inline void __vfree_deferred(const void *addr)
Signature · 1 line
2.19.2
author
#2 lore raw

PATCH 2/3 mm: separate memory allocation and actual work in alloc_vmap_area()

In reply to Roman Gushchin

alloc_vmap_area() is allocating memory for the vmap_area, and performing the actual lookup of the vm area and vmap_area initialization.

This prevents us from using a pre-allocated memory for the map_area structure, which can be used in some cases to minimize the number of required memory allocations.

Let's keep the memory allocation part in alloc_vmap_area() and separate everything else into init_vmap_area().

Signed-off-by Roman Gushchin <[email protected]>
Acked-by Johannes Weiner <[email protected]>
Reviewed-by Matthew Wilcox <[email protected]>
---
 mm/vmalloc.c | 50 +++++++++++++++++++++++++++++++++-----------------
 1 file changed, 33 insertions(+), 17 deletions(-)
mm/vmalloc.c +33 -17
index 7660e3ef4133..042175d7d95f 100644
@@ -395,16 +395,10 @@ static void purge_vmap_area_lazy(void);
 
 static BLOCKING_NOTIFIER_HEAD(vmap_notify_list);
 
-/*
- * Allocate a region of KVA of the specified size and alignment, within the
- * vstart and vend.
- */
-static struct vmap_area *alloc_vmap_area(unsigned long size,
-				unsigned long align,
-				unsigned long vstart, unsigned long vend,
-				int node, gfp_t gfp_mask)
+static int init_vmap_area(struct vmap_area *va, unsigned long size,
+			  unsigned long align, unsigned long vstart,
+			  unsigned long vend, int node, gfp_t gfp_mask)
 {
-	struct vmap_area *va;
 	struct rb_node *n;
 	unsigned long addr;
 	int purged = 0;

@@ -416,11 +410,6 @@ static struct vmap_area *alloc_vmap_area(unsigned long size,
 
 	might_sleep();
 
-	va = kmalloc_node(sizeof(struct vmap_area),
-			gfp_mask & GFP_RECLAIM_MASK, node);
-	if (unlikely(!va))
-		return ERR_PTR(-ENOMEM);
-
 	/*
 	 * Only scan the relevant parts containing pointers to other objects
 	 * to avoid false negatives.

@@ -512,7 +501,7 @@ static struct vmap_area *alloc_vmap_area(unsigned long size,
 	BUG_ON(va->va_start < vstart);
 	BUG_ON(va->va_end > vend);
 
-	return va;
+	return 0;
 
 overflow:
 	spin_unlock(&vmap_area_lock);

@@ -534,8 +523,35 @@ static struct vmap_area *alloc_vmap_area(unsigned long size,
 	if (!(gfp_mask & __GFP_NOWARN) && printk_ratelimit())
 		pr_warn("vmap allocation for size %lu failed: use vmalloc=<size> to increase size\n",
 			size);
-	kfree(va);
-	return ERR_PTR(-EBUSY);
+
+	return -EBUSY;
+}
+
+/*
+ * Allocate a region of KVA of the specified size and alignment, within the
+ * vstart and vend.
+ */
+static struct vmap_area *alloc_vmap_area(unsigned long size,
+					 unsigned long align,
+					 unsigned long vstart,
+					 unsigned long vend,
+					 int node, gfp_t gfp_mask)
+{
+	struct vmap_area *va;
+	int ret;
+
+	va = kmalloc_node(sizeof(struct vmap_area),
+			gfp_mask & GFP_RECLAIM_MASK, node);
+	if (unlikely(!va))
+		return ERR_PTR(-ENOMEM);
+
+	ret = init_vmap_area(va, size, align, vstart, vend, node, gfp_mask);
+	if (ret) {
+		kfree(va);
+		return ERR_PTR(ret);
+	}
+
+	return va;
 }
 
 int register_vmap_purge_notifier(struct notifier_block *nb)
Signature · 1 line
2.19.2
author
#3 lore raw

PATCH 3/3 mm: show number of vmalloc pages in /proc/meminfo

In reply to Roman Gushchin

Vmalloc() is getting more and more used these days (kernel stacks, bpf and percpu allocator are new top users), and the total % of memory consumed by vmalloc() can be pretty significant and changes dynamically.

/proc/meminfo is the best place to display this information:
its top goal is to show top consumers of the memory.

Since the VmallocUsed field in /proc/meminfo is not in use for quite a long time (it has been defined to 0 by the commit a5ad88ce8c7f ("mm: get rid of 'vmalloc_info' from /proc/meminfo")), let's reuse it for showing the actual physical memory consumption of vmalloc().

Signed-off-by Roman Gushchin <[email protected]>
Acked-by Johannes Weiner <[email protected]>
---
 fs/proc/meminfo.c       |  2 +-
 include/linux/vmalloc.h |  2 ++
 mm/vmalloc.c            | 10 ++++++++++
 3 files changed, 13 insertions(+), 1 deletion(-)
fs/proc/meminfo.c +1 -1
index 568d90e17c17..465ea0153b2a 100644
@@ -120,7 +120,7 @@ static int meminfo_proc_show(struct seq_file *m, void *v)
 	show_val_kb(m, "Committed_AS:   ", committed);
 	seq_printf(m, "VmallocTotal:   %8lu kB\n",
 		   (unsigned long)VMALLOC_TOTAL >> 10);
-	show_val_kb(m, "VmallocUsed:    ", 0ul);
+	show_val_kb(m, "VmallocUsed:    ", vmalloc_nr_pages());
 	show_val_kb(m, "VmallocChunk:   ", 0ul);
 	show_val_kb(m, "Percpu:         ", pcpu_nr_pages());
include/linux/vmalloc.h +2 -0
index 398e9c95cd61..0b497408272b 100644
@@ -63,10 +63,12 @@ extern void vm_unmap_aliases(void);
 
 #ifdef CONFIG_MMU
 extern void __init vmalloc_init(void);
+extern unsigned long vmalloc_nr_pages(void);
 #else
 static inline void vmalloc_init(void)
 {
 }
+static inline unsigned long vmalloc_nr_pages(void) { return 0; }
 #endif
 
 extern void *vmalloc(unsigned long size);
mm/vmalloc.c +10 -0
index 042175d7d95f..efca916940e9 100644
@@ -340,6 +340,13 @@ static unsigned long cached_align;
 
 static unsigned long vmap_area_pcpu_hole;
 
+static atomic_long_t nr_vmalloc_pages;
+
+unsigned long vmalloc_nr_pages(void)
+{
+	return atomic_long_read(&nr_vmalloc_pages);
+}
+
 static struct vmap_area *__find_vmap_area(unsigned long addr)
 {
 	struct rb_node *n = vmap_area_root.rb_node;

@@ -1549,6 +1556,7 @@ static void __vunmap(const void *addr, int deallocate_pages)
 			BUG_ON(!page);
 			__free_pages(page, 0);
 		}
+		atomic_long_sub(area->nr_pages, &nr_vmalloc_pages);
 
 		kvfree(area->pages);
 	}

@@ -1717,12 +1725,14 @@ static void *__vmalloc_area_node(struct vm_struct *area, gfp_t gfp_mask,
 		if (unlikely(!page)) {
 			/* Successfully allocated i pages, free them in __vunmap() */
 			area->nr_pages = i;
+			atomic_long_add(area->nr_pages, &nr_vmalloc_pages);
 			goto fail;
 		}
 		area->pages[i] = page;
 		if (gfpflags_allow_blocking(gfp_mask|highmem_mask))
 			cond_resched();
 	}
+	atomic_long_add(area->nr_pages, &nr_vmalloc_pages);
 
 	if (map_vm_area(area, prot, pages))
 		goto fail;
Signature · 1 line
2.19.2
#4 lore raw
In reply to Roman Gushchin

These slipped through the cracks. +CC Andrew directly. Andrew, if it's not too late yet, could you consider them for 5.1?

Quoted 31 lines · Roman Gushchin

The patchset contains few changes to the vmalloc code, which are leading to some performance gains and code simplification.

Also, it exports a number of pages, used by vmalloc(), in /proc/meminfo.

Patch (1) removes some redundancy on __vunmap(). Patch (2) separates memory allocation and data initialization
in alloc_vmap_area()
Patch (3) adds vmalloc counter to /proc/meminfo.

RFC->v1:
- removed bogus empty lines (suggested by Matthew Wilcox)
- made nr_vmalloc_pages static (suggested by Matthew Wilcox)
- dropped patch 3 from RFC patchset, will post later with some other changes
- dropped RFC

Roman Gushchin (3):
mm: refactor __vunmap() to avoid duplicated call to find_vm_area()
mm: separate memory allocation and actual work in alloc_vmap_area()
mm: show number of vmalloc pages in /proc/meminfo

 fs/proc/meminfo.c       |   2 +-
 include/linux/vmalloc.h |   2 +
 mm/vmalloc.c            | 107 ++++++++++++++++++++++++++--------------
 3 files changed, 73 insertions(+), 38 deletions(-)
Signature · 1 line
2.19.2
maintainer
#5 lore raw
Quoted 10 lines · On Mon, 11 Feb 2019 14:08:22 -0500 Johannes Weiner <[email protected]> wrote:
Quoted 5 lines · Roman Gushchin

The patchset contains few changes to the vmalloc code, which are leading to some performance gains and code simplification.

Also, it exports a number of pages, used by vmalloc(), in /proc/meminfo.

These slipped through the cracks. +CC Andrew directly. Andrew, if it's not too late yet, could you consider them for 5.1?

There's been some activity in vmalloc.c lately and these have bitrotted. They'll need a redo, please.

#6 lore raw
Quoted 9 lines · Andrew Morton
Quoted 7 lines · On Mon, 11 Feb 2019 14:08:22 -0500 Johannes Weiner <[email protected]> wrote:
Quoted 5 lines · Roman Gushchin

The patchset contains few changes to the vmalloc code, which are leading to some performance gains and code simplification.

Also, it exports a number of pages, used by vmalloc(), in /proc/meminfo.

Quoted 7 lines
Quoted 3 lines

These slipped through the cracks. +CC Andrew directly. Andrew, if it's not too late yet, could you consider them for 5.1?

There's been some activity in vmalloc.c lately and these have bitrotted. They'll need a redo, please.

Will do. Thanks!