linux-kernel - [PATCH v2] mm/vmalloc: lock contention optimization under multi-threading

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20240220090521.3316345-1-rulin.huang@intel.com>
Date: Tue, 20 Feb 2024 04:05:21 -0500
From: rulinhuang <rulin.huang@...el.com>
To: urezki@...il.com
Cc: akpm@...ux-foundation.org,
	colin.king@...el.com,
	hch@...radead.org,
	linux-kernel@...r.kernel.org,
	linux-mm@...ck.org,
	lstoakes@...il.com,
	rulin.huang@...el.com,
	tim.c.chen@...el.com,
	zhiguo.zhou@...el.com,
	wangyang.guo@...el.com,
	tianyou.li@...el.com
Subject: [PATCH v2] mm/vmalloc: lock contention optimization under multi-threading

When allocating a new memory area where the mapping address range is
known, it is observed that the vmap_area lock is acquired twice.
The first acquisition occurs in the alloc_vmap_area() function when
inserting the vm area into the vm mapping red-black tree. The second
acquisition occurs in the setup_vmalloc_vm() function when updating the
properties of the vm, such as flags and address, etc.
Combine these two operations together in alloc_vmap_area(), which
improves scalability when the vmap_area lock is contended. By doing so,
the need to acquire the lock twice can also be eliminated.
With the above change, tested on intel icelake platform(160 vcpu, kernel
v6.7), a 6% performance improvement and a 7% reduction in overall
spinlock hotspot are gained on
stress-ng/pthread(https://github.com/ColinIanKing/stress-ng), which is
the stress test of thread creations.

Reviewed-by: Chen Tim C <tim.c.chen@...el.com>
Reviewed-by: King Colin <colin.king@...el.com>
Signed-off-by: rulinhuang <rulin.huang@...el.com>
---
V1 -> V2: Avoided the partial initialization issue of vm and 
separated insert_vmap_area() from alloc_vmap_area()
---
 mm/vmalloc.c | 36 +++++++++++++++++++++---------------
 1 file changed, 21 insertions(+), 15 deletions(-)

diff --git a/mm/vmalloc.c b/mm/vmalloc.c
index d12a17fc0c171..768e45f2ed946 100644
--- a/mm/vmalloc.c
+++ b/mm/vmalloc.c
@@ -1630,17 +1630,18 @@ static struct vmap_area *alloc_vmap_area(unsigned long size,
 	va->vm = NULL;
 	va->flags = va_flags;
 
-	spin_lock(&vmap_area_lock);
-	insert_vmap_area(va, &vmap_area_root, &vmap_area_list);
-	spin_unlock(&vmap_area_lock);
-
 	BUG_ON(!IS_ALIGNED(va->va_start, align));
 	BUG_ON(va->va_start < vstart);
 	BUG_ON(va->va_end > vend);
 
 	ret = kasan_populate_vmalloc(addr, size);
 	if (ret) {
-		free_vmap_area(va);
+		/*
+		 * Insert/Merge it back to the free tree/list.
+		 */
+		spin_lock(&free_vmap_area_lock);
+		merge_or_add_vmap_area_augment(va, &free_vmap_area_root, &free_vmap_area_list);
+		spin_unlock(&free_vmap_area_lock);
 		return ERR_PTR(ret);
 	}
 
@@ -1669,6 +1670,13 @@ static struct vmap_area *alloc_vmap_area(unsigned long size,
 	return ERR_PTR(-EBUSY);
 }
 
+static inline void insert_vmap_area_with_lock(struct vmap_area *va)
+{
+	spin_lock(&vmap_area_lock);
+	insert_vmap_area(va, &vmap_area_root, &vmap_area_list);
+	spin_unlock(&vmap_area_lock);
+}
+
 int register_vmap_purge_notifier(struct notifier_block *nb)
 {
 	return blocking_notifier_chain_register(&vmap_notify_list, nb);
@@ -2045,6 +2053,8 @@ static void *new_vmap_block(unsigned int order, gfp_t gfp_mask)
 		return ERR_CAST(va);
 	}
 
+	insert_vmap_area_with_lock(va);
+
 	vaddr = vmap_block_vaddr(va->va_start, 0);
 	spin_lock_init(&vb->lock);
 	vb->va = va;
@@ -2398,6 +2408,8 @@ void *vm_map_ram(struct page **pages, unsigned int count, int node)
 		if (IS_ERR(va))
 			return NULL;
 
+		insert_vmap_area_with_lock(va);
+
 		addr = va->va_start;
 		mem = (void *)addr;
 	}
@@ -2538,7 +2550,7 @@ static void vmap_init_free_space(void)
 	}
 }
 
-static inline void setup_vmalloc_vm_locked(struct vm_struct *vm,
+static inline void setup_vmalloc_vm(struct vm_struct *vm,
 	struct vmap_area *va, unsigned long flags, const void *caller)
 {
 	vm->flags = flags;
@@ -2548,14 +2560,6 @@ static inline void setup_vmalloc_vm_locked(struct vm_struct *vm,
 	va->vm = vm;
 }
 
-static void setup_vmalloc_vm(struct vm_struct *vm, struct vmap_area *va,
-			      unsigned long flags, const void *caller)
-{
-	spin_lock(&vmap_area_lock);
-	setup_vmalloc_vm_locked(vm, va, flags, caller);
-	spin_unlock(&vmap_area_lock);
-}
-
 static void clear_vm_uninitialized_flag(struct vm_struct *vm)
 {
 	/*
@@ -2600,6 +2604,8 @@ static struct vm_struct *__get_vm_area_node(unsigned long size,
 
 	setup_vmalloc_vm(area, va, flags, caller);
 
+	insert_vmap_area_with_lock(va);
+
 	/*
 	 * Mark pages for non-VM_ALLOC mappings as accessible. Do it now as a
 	 * best-effort approach, as they can be mapped outside of vmalloc code.
@@ -4166,7 +4172,7 @@ struct vm_struct **pcpu_get_vm_areas(const unsigned long *offsets,
 	for (area = 0; area < nr_vms; area++) {
 		insert_vmap_area(vas[area], &vmap_area_root, &vmap_area_list);
 
-		setup_vmalloc_vm_locked(vms[area], vas[area], VM_ALLOC,
+		setup_vmalloc_vm(vms[area], vas[area], VM_ALLOC,
 				 pcpu_get_vm_areas);
 	}
 	spin_unlock(&vmap_area_lock);

base-commit: de927f6c0b07d9e698416c5b287c521b07694cac
-- 
2.43.0