mm: vmalloc: avoid calling __find_vmap_area() twice in __vunmap()

Currently the __vunmap() path calls __find_vmap_area() twice.  Once on
entry to check that the area exists, then inside the remove_vm_area()
function which also performs a new search for the VA.

In order to improvie it from a performance point of view we split
remove_vm_area() into two new parts:
  - find_unlink_vmap_area() that does a search and unlink from tree;
  - __remove_vm_area() that removes without searching.

In this case there is no any functional change for remove_vm_area()
whereas vm_remove_mappings(), where a second search happens, switches to
the __remove_vm_area() variant where the already detached VA is passed as
a parameter, so there is no need to find it again.

Performance wise, i use test_vmalloc.sh with 32 threads doing alloc
free on a 64-CPUs-x86_64-box:

perf without this patch:
-   31.41%     0.50%  vmalloc_test/10  [kernel.vmlinux]    [k] __vunmap
   - 30.92% __vunmap
      - 17.67% _raw_spin_lock
           native_queued_spin_lock_slowpath
      - 12.33% remove_vm_area
         - 11.79% free_vmap_area_noflush
            - 11.18% _raw_spin_lock
                 native_queued_spin_lock_slowpath
        0.76% free_unref_page

perf with this patch:
-   11.35%     0.13%  vmalloc_test/14  [kernel.vmlinux]    [k] __vunmap
   - 11.23% __vunmap
      - 8.28% find_unlink_vmap_area
         - 7.95% _raw_spin_lock
              7.44% native_queued_spin_lock_slowpath
      - 1.93% free_vmap_area_noflush
         - 0.56% _raw_spin_lock
              0.53% native_queued_spin_lock_slowpath
        0.60% __vunmap_range_noflush

__vunmap() consumes around ~20% less CPU cycles on this test.

Also, switch from find_vmap_area() to find_unlink_vmap_area() to prevent a
double access to the vmap_area_lock: one for finding area, second time is
for unlinking from a tree.

[urezki@gmail.com: switch to find_unlink_vmap_area() in vm_unmap_ram()]
Link: https://lkml.kernel.org/r/20221222190022.134380-2-urezki@gmail.com
Link: https://lkml.kernel.org/r/20221222190022.134380-1-urezki@gmail.com
Signed-off-by: Uladzislau Rezki (Sony) <urezki@gmail.com>
Reported-by: Roman Gushchin <roman.gushchin@linux.dev>
Reviewed-by: Lorenzo Stoakes <lstoakes@gmail.com>
Cc: Baoquan He <bhe@redhat.com>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Nicholas Piggin <npiggin@gmail.com>
Cc: Oleksiy Avramchenko <oleksiy.avramchenko@sony.com>
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

diff --git a/mm/vmalloc.c b/mm/vmalloc.c
index 10fe83c..476ccbf 100644 (file)
--- a/mm/vmalloc.c
+++ b/mm/vmalloc.c
@@ -1815,9 +1815,9 @@ static void drain_vmap_area_work(struct work_struct *work)
 }
 
 /*
- * Free a vmap area, caller ensuring that the area has been unmapped
- * and flush_cache_vunmap had been called for the correct range
- * previously.
+ * Free a vmap area, caller ensuring that the area has been unmapped,
+ * unlinked and flush_cache_vunmap had been called for the correct
+ * range previously.
  */
 static void free_vmap_area_noflush(struct vmap_area *va)
 {
@@ -1825,9 +1825,8 @@ static void free_vmap_area_noflush(struct vmap_area *va)
        unsigned long va_start = va->va_start;
        unsigned long nr_lazy;
 
-       spin_lock(&vmap_area_lock);
-       unlink_va(va, &vmap_area_root);
-       spin_unlock(&vmap_area_lock);
+       if (WARN_ON_ONCE(!list_empty(&va->list)))
+               return;
 
        nr_lazy = atomic_long_add_return((va->va_end - va->va_start) >>
                                PAGE_SHIFT, &vmap_lazy_nr);
@@ -1871,6 +1870,19 @@ struct vmap_area *find_vmap_area(unsigned long addr)
        return va;
 }
 
+static struct vmap_area *find_unlink_vmap_area(unsigned long addr)
+{
+       struct vmap_area *va;
+
+       spin_lock(&vmap_area_lock);
+       va = __find_vmap_area(addr, &vmap_area_root);
+       if (va)
+               unlink_va(va, &vmap_area_root);
+       spin_unlock(&vmap_area_lock);
+
+       return va;
+}
+
 /*** Per cpu kva allocator ***/
 
 /*
@@ -2015,6 +2027,10 @@ static void free_vmap_block(struct vmap_block *vb)
        tmp = xa_erase(&vmap_blocks, addr_to_vb_idx(vb->va->va_start));
        BUG_ON(tmp != vb);
 
+       spin_lock(&vmap_area_lock);
+       unlink_va(vb->va, &vmap_area_root);
+       spin_unlock(&vmap_area_lock);
+
        free_vmap_area_noflush(vb->va);
        kfree_rcu(vb, rcu_head);
 }
@@ -2236,7 +2252,7 @@ void vm_unmap_ram(const void *mem, unsigned int count)
                return;
        }
 
-       va = find_vmap_area(addr);
+       va = find_unlink_vmap_area(addr);
        BUG_ON(!va);
        debug_check_no_locks_freed((void *)va->va_start,
                                    (va->va_end - va->va_start));
@@ -2591,6 +2607,20 @@ struct vm_struct *find_vm_area(const void *addr)
        return va->vm;
 }
 
+static struct vm_struct *__remove_vm_area(struct vmap_area *va)
+{
+       struct vm_struct *vm;
+
+       if (!va || !va->vm)
+               return NULL;
+
+       vm = va->vm;
+       kasan_free_module_shadow(vm);
+       free_unmap_vmap_area(va);
+
+       return vm;
+}
+
 /**
  * remove_vm_area - find and remove a continuous kernel virtual area
  * @addr:          base address
@@ -2603,26 +2633,10 @@ struct vm_struct *find_vm_area(const void *addr)
  */
 struct vm_struct *remove_vm_area(const void *addr)
 {
-       struct vmap_area *va;
-
        might_sleep();
 
-       spin_lock(&vmap_area_lock);
-       va = __find_vmap_area((unsigned long)addr, &vmap_area_root);
-       if (va && va->vm) {
-               struct vm_struct *vm = va->vm;
-
-               va->vm = NULL;
-               spin_unlock(&vmap_area_lock);
-
-               kasan_free_module_shadow(vm);
-               free_unmap_vmap_area(va);
-
-               return vm;
-       }
-
-       spin_unlock(&vmap_area_lock);
-       return NULL;
+       return __remove_vm_area(
+               find_unlink_vmap_area((unsigned long) addr));
 }
 
 static inline void set_area_direct_map(const struct vm_struct *area,
@@ -2636,16 +2650,17 @@ static inline void set_area_direct_map(const struct vm_struct *area,
                        set_direct_map(area->pages[i]);
 }
 
-/* Handle removing and resetting vm mappings related to the vm_struct. */
-static void vm_remove_mappings(struct vm_struct *area, int deallocate_pages)
+/* Handle removing and resetting vm mappings related to the VA's vm_struct. */
+static void va_remove_mappings(struct vmap_area *va, int deallocate_pages)
 {
+       struct vm_struct *area = va->vm;
        unsigned long start = ULONG_MAX, end = 0;
        unsigned int page_order = vm_area_page_order(area);
        int flush_reset = area->flags & VM_FLUSH_RESET_PERMS;
        int flush_dmap = 0;
        int i;
 
-       remove_vm_area(area->addr);
+       __remove_vm_area(va);
 
        /* If this is not VM_FLUSH_RESET_PERMS memory, no need for the below. */
        if (!flush_reset)
@@ -2690,6 +2705,7 @@ static void vm_remove_mappings(struct vm_struct *area, int deallocate_pages)
 static void __vunmap(const void *addr, int deallocate_pages)
 {
        struct vm_struct *area;
+       struct vmap_area *va;
 
        if (!addr)
                return;
@@ -2698,19 +2714,20 @@ static void __vunmap(const void *addr, int deallocate_pages)
                        addr))
                return;
 
-       area = find_vm_area(addr);
-       if (unlikely(!area)) {
+       va = find_unlink_vmap_area((unsigned long)addr);
+       if (unlikely(!va)) {
                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));
 
        kasan_poison_vmalloc(area->addr, get_vm_area_size(area));
 
-       vm_remove_mappings(area, deallocate_pages);
+       va_remove_mappings(va, deallocate_pages);
 
        if (deallocate_pages) {
                int i;