mm/mmap.c: simplify the failure return working flow
authorChen Gang <gang.chen.5i5j@gmail.com>
Tue, 8 Sep 2015 22:03:38 +0000 (15:03 -0700)
committerLinus Torvalds <torvalds@linux-foundation.org>
Tue, 8 Sep 2015 22:35:28 +0000 (15:35 -0700)
__split_vma() doesn't need out_err label, neither need initializing err.

copy_vma() can return NULL directly when kmem_cache_alloc() fails.

Signed-off-by: Chen Gang <gang.chen.5i5j@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
mm/mmap.c

index 52a2373..7a3b123 100644 (file)
--- a/mm/mmap.c
+++ b/mm/mmap.c
@@ -2455,7 +2455,7 @@ static int __split_vma(struct mm_struct *mm, struct vm_area_struct *vma,
              unsigned long addr, int new_below)
 {
        struct vm_area_struct *new;
-       int err = -ENOMEM;
+       int err;
 
        if (is_vm_hugetlb_page(vma) && (addr &
                                        ~(huge_page_mask(hstate_vma(vma)))))
@@ -2463,7 +2463,7 @@ static int __split_vma(struct mm_struct *mm, struct vm_area_struct *vma,
 
        new = kmem_cache_alloc(vm_area_cachep, GFP_KERNEL);
        if (!new)
-               goto out_err;
+               return -ENOMEM;
 
        /* most fields are the same, copy all, and then fixup */
        *new = *vma;
@@ -2511,7 +2511,6 @@ static int __split_vma(struct mm_struct *mm, struct vm_area_struct *vma,
        mpol_put(vma_policy(new));
  out_free_vma:
        kmem_cache_free(vm_area_cachep, new);
- out_err:
        return err;
 }
 
@@ -2952,30 +2951,31 @@ struct vm_area_struct *copy_vma(struct vm_area_struct **vmap,
                *need_rmap_locks = (new_vma->vm_pgoff <= vma->vm_pgoff);
        } else {
                new_vma = kmem_cache_alloc(vm_area_cachep, GFP_KERNEL);
-               if (new_vma) {
-                       *new_vma = *vma;
-                       new_vma->vm_start = addr;
-                       new_vma->vm_end = addr + len;
-                       new_vma->vm_pgoff = pgoff;
-                       if (vma_dup_policy(vma, new_vma))
-                               goto out_free_vma;
-                       INIT_LIST_HEAD(&new_vma->anon_vma_chain);
-                       if (anon_vma_clone(new_vma, vma))
-                               goto out_free_mempol;
-                       if (new_vma->vm_file)
-                               get_file(new_vma->vm_file);
-                       if (new_vma->vm_ops && new_vma->vm_ops->open)
-                               new_vma->vm_ops->open(new_vma);
-                       vma_link(mm, new_vma, prev, rb_link, rb_parent);
-                       *need_rmap_locks = false;
-               }
+               if (!new_vma)
+                       goto out;
+               *new_vma = *vma;
+               new_vma->vm_start = addr;
+               new_vma->vm_end = addr + len;
+               new_vma->vm_pgoff = pgoff;
+               if (vma_dup_policy(vma, new_vma))
+                       goto out_free_vma;
+               INIT_LIST_HEAD(&new_vma->anon_vma_chain);
+               if (anon_vma_clone(new_vma, vma))
+                       goto out_free_mempol;
+               if (new_vma->vm_file)
+                       get_file(new_vma->vm_file);
+               if (new_vma->vm_ops && new_vma->vm_ops->open)
+                       new_vma->vm_ops->open(new_vma);
+               vma_link(mm, new_vma, prev, rb_link, rb_parent);
+               *need_rmap_locks = false;
        }
        return new_vma;
 
- out_free_mempol:
+out_free_mempol:
        mpol_put(vma_policy(new_vma));
- out_free_vma:
+out_free_vma:
        kmem_cache_free(vm_area_cachep, new_vma);
+out:
        return NULL;
 }