[PATCH] mm: uml pte atomicity
Hugh Dickins [Sun, 30 Oct 2005 01:16:38 +0000 (18:16 -0700)]
There's usually a good reason when a pte is examined without the lock; but it
makes me nervous when the pointer is dereferenced more than once.

Signed-off-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

arch/um/kernel/process_kern.c

index 0d73cee..34b54a3 100644 (file)
@@ -222,6 +222,7 @@ void *um_virt_to_phys(struct task_struct *task, unsigned long addr,
        pud_t *pud;
        pmd_t *pmd;
        pte_t *pte;
+       pte_t ptent;
 
        if(task->mm == NULL) 
                return(ERR_PTR(-EINVAL));
@@ -238,12 +239,13 @@ void *um_virt_to_phys(struct task_struct *task, unsigned long addr,
                return(ERR_PTR(-EINVAL));
 
        pte = pte_offset_kernel(pmd, addr);
-       if(!pte_present(*pte)) 
+       ptent = *pte;
+       if(!pte_present(ptent))
                return(ERR_PTR(-EINVAL));
 
        if(pte_out != NULL)
-               *pte_out = *pte;
-       return((void *) (pte_val(*pte) & PAGE_MASK) + (addr & ~PAGE_MASK));
+               *pte_out = ptent;
+       return((void *) (pte_val(ptent) & PAGE_MASK) + (addr & ~PAGE_MASK));
 }
 
 char *current_cmd(void)