[PATCH] mm acct accounting fix
Kirill Korotaev [Tue, 17 May 2005 04:53:18 +0000 (21:53 -0700)]
This patch fixes mm->total_vm and mm->locked_vm acctounting in case when
move_page_tables() fails inside move_vma().

Signed-Off-By: Kirill Korotaev <dev@sw.ru>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

mm/mremap.c

index 0dd7ace..ec7238a 100644 (file)
@@ -224,6 +224,12 @@ static unsigned long move_vma(struct vm_area_struct *vma,
                        split = 1;
        }
 
+       /*
+        * if we failed to move page tables we still do total_vm increment
+        * since do_munmap() will decrement it by old_len == new_len
+        */
+       mm->total_vm += new_len >> PAGE_SHIFT;
+
        if (do_munmap(mm, old_addr, old_len) < 0) {
                /* OOM: unable to split vma, just get accounts right */
                vm_unacct_memory(excess >> PAGE_SHIFT);
@@ -237,7 +243,6 @@ static unsigned long move_vma(struct vm_area_struct *vma,
                        vma->vm_next->vm_flags |= VM_ACCOUNT;
        }
 
-       mm->total_vm += new_len >> PAGE_SHIFT;
        __vm_stat_account(mm, vma->vm_flags, vma->vm_file, new_len>>PAGE_SHIFT);
        if (vm_flags & VM_LOCKED) {
                mm->locked_vm += new_len >> PAGE_SHIFT;