[PATCH] fix the fix for unmap_vmas & hugepages

Patch from Kevin Pedretti <pedretti@ieee.org>

The previous fix for unmapping hugetlb regions could still produce incorrect
alignments if the munmap request covers multiple VMA's.

Fix it by always unmapped the entire hugepage VMA inside the inner loop.

diff --git a/mm/memory.c b/mm/memory.c
index 4dde60116530522eaa58e94c13bc7b344d2cff8e..a117bfb8da5435f1277df495ad6bcc269fa8ee52 100644 (file)
--- a/mm/memory.c
+++ b/mm/memory.c
@@ -484,16 +484,6 @@ void unmap_page_range(struct mmu_gather *tlb, struct vm_area_struct *vma,
 #define ZAP_BLOCK_SIZE (~(0UL))
 #endif
 
-/*
- * hugepage regions must be unmapped with HPAGE_SIZE granularity
- */
-static inline unsigned long zap_block_size(struct vm_area_struct *vma)
-{
-       if (is_vm_hugetlb_page(vma))
-               return HPAGE_SIZE;
-       return ZAP_BLOCK_SIZE;
-}
-
 /**
  * unmap_vmas - unmap a range of memory covered by a list of vma's
  * @tlbp: address of the caller's struct mmu_gather
@@ -524,7 +514,7 @@ int unmap_vmas(struct mmu_gather **tlbp, struct mm_struct *mm,
                struct vm_area_struct *vma, unsigned long start_addr,
                unsigned long end_addr, unsigned long *nr_accounted)
 {
-       unsigned long zap_bytes = zap_block_size(vma);
+       unsigned long zap_bytes = ZAP_BLOCK_SIZE;
        unsigned long tlb_start;        /* For tlb_finish_mmu */
        int tlb_start_valid = 0;
        int ret = 0;
@@ -554,7 +544,12 @@ int unmap_vmas(struct mmu_gather **tlbp, struct mm_struct *mm,
 
                ret++;
                while (start != end) {
-                       unsigned long block = min(zap_bytes, end - start);
+                       unsigned long block;
+
+                       if (is_vm_hugetlb_page(vma))
+                               block = end - start;
+                       else
+                               block = min(zap_bytes, end - start);
 
                        if (!tlb_start_valid) {
                                tlb_start = start;
@@ -564,7 +559,7 @@ int unmap_vmas(struct mmu_gather **tlbp, struct mm_struct *mm,
                        unmap_page_range(*tlbp, vma, start, start + block);
                        start += block;
                        zap_bytes -= block;
-                       if (zap_bytes != 0)
+                       if ((long)zap_bytes > 0)
                                continue;
                        if (need_resched()) {
                                tlb_finish_mmu(*tlbp, tlb_start, start);
@@ -572,7 +567,7 @@ int unmap_vmas(struct mmu_gather **tlbp, struct mm_struct *mm,
                                *tlbp = tlb_gather_mmu(mm, 0);
                                tlb_start_valid = 0;
                        }
-                       zap_bytes = zap_block_size(vma);
+                       zap_bytes = ZAP_BLOCK_SIZE;
                }
                if (vma->vm_next && vma->vm_next->vm_start < vma->vm_end)
                        printk("%s: VMA list is not sorted correctly!\n",