[PATCH] hugetlb unmap_vmas() SMP && PREEMPT fix

Patch from Kevin Pedretti <pedretti@ieee.org>

The unmap_vmas() logic is designed to chew away at all the pages without
holding off preemption for too long.  But with CONFIG_SMP=y and
CONFIG_PREEMPT=y the number of pages which we batch up between rescheduling
opportunities is not a multiple of HPAGE_SIZE.  So unmap_vmas() ends up
calling unmap_hugepage_range() with a poorly aligned&sized region, and it
goes BUG.

Fix that up by ensuring that we always work across hugepage regions in
HPAGE_SIZE chunks.

include/linux/hugetlb.h

@@ -66,6 +66,7 @@ static inline int is_vm_hugetlb_page(struct vm_area_struct *vma)
 
 #ifndef HPAGE_MASK
 #define HPAGE_MASK	0		/* Keep the compiler happy */
+#define HPAGE_SIZE	0
 #endif
 
 #endif /* !CONFIG_HUGETLB_PAGE */

mm/memory.c

@@ -484,6 +484,16 @@ 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
@@ -514,7 +524,7 @@ 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;
+	unsigned long zap_bytes = zap_block_size(vma);
 	unsigned long tlb_start;	/* For tlb_finish_mmu */
 	int tlb_start_valid = 0;
 	int ret = 0;
@@ -562,7 +572,7 @@ 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;
+			zap_bytes = zap_block_size(vma);
 		}
 		if (vma->vm_next && vma->vm_next->vm_start < vma->vm_end)
 			printk("%s: VMA list is not sorted correctly!\n",