Skip to contentSkip to conversation

userfaultfd: mcopy_atomic: update cases returning -ENOENT

PATCH In review Mike Rapoport 2 participants 2 messages Feb 7, 2017

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 userfaultfd: mcopy_atomic: update cases returning -ENOENT

Hello Andrew,

The patch below is an incremental fixup for concerns Andrea raised at [1]. Please let me know if you prefer me to update the original patch and resend.

[1] http://www.spinics.net/lists/linux-mm/msg121267.html

--
Sincerely yours,
Mike.

Quoted 1 line

From 8acff65ecee8ca4cc892d35b45125c34568d1c93 Mon Sep 17 00:00:00 2001

From: Mike Rapoport <[email protected]>
Date: Tue, 7 Feb 2017 11:50:17 +0200
Subject: [PATCH] userfaultfd: mcopy_atomic: update cases returning -ENOENT

As Andrea commented in [1], if the VMA covering the address was unmapped, we may end up with a VMA a way above the faulting address. In this case we would like to return -ENOENT to allow uffd monitor detection of VMA removal.

[1] http://www.spinics.net/lists/linux-mm/msg121267.html

---
 mm/userfaultfd.c | 14 ++++++++------
 1 file changed, 8 insertions(+), 6 deletions(-)
mm/userfaultfd.c +8 -6
index b861cf9..cf43456 100644
@@ -206,6 +206,10 @@ static __always_inline ssize_t __mcopy_atomic_hugetlb(struct mm_struct *dst_mm,
 		if (!dst_vma->vm_userfaultfd_ctx.ctx)
 			goto out_unlock;
 
+		if (dst_start < dst_vma->vm_start ||
+		    dst_start + len > dst_vma->vm_end)
+			goto out_unlock;
+
 		err = -EINVAL;
 		if (vma_hpagesize != vma_kernel_pagesize(dst_vma))
 			goto out_unlock;
@@ -216,9 +220,6 @@ static __always_inline ssize_t __mcopy_atomic_hugetlb(struct mm_struct *dst_mm,
 		 */
 		if (dst_vma->vm_flags & VM_SHARED)
 			goto out_unlock;
-		if (dst_start < dst_vma->vm_start ||
-		    dst_start + len > dst_vma->vm_end)
-			goto out_unlock;
 	}
 
 	if (WARN_ON(dst_addr & (vma_hpagesize - 1) ||
@@ -385,13 +386,14 @@ static __always_inline ssize_t __mcopy_atomic(struct mm_struct *dst_mm,
 	if (!dst_vma->vm_userfaultfd_ctx.ctx)
 		goto out_unlock;
 
-	err = -EINVAL;
-	if (!vma_is_shmem(dst_vma) && dst_vma->vm_flags & VM_SHARED)
-		goto out_unlock;
 	if (dst_start < dst_vma->vm_start ||
 	    dst_start + len > dst_vma->vm_end)
 		goto out_unlock;
 
+	err = -EINVAL;
+	if (!vma_is_shmem(dst_vma) && dst_vma->vm_flags & VM_SHARED)
+		goto out_unlock;
+
 	/*
 	 * If this is a HUGETLB vma, pass off to appropriate routine
 	 */
Signature · 1 line
1.9.1
Reviewed-by on v1 1/1
#1 lore raw
Quoted 17 lines · Mike Rapoport

Hello Andrew,

The patch below is an incremental fixup for concerns Andrea raised at [1]. Please let me know if you prefer me to update the original patch and resend.

[1] http://www.spinics.net/lists/linux-mm/msg121267.html

--
Sincerely yours,
Mike.

From 8acff65ecee8ca4cc892d35b45125c34568d1c93 Mon Sep 17 00:00:00 2001
From: Mike Rapoport <[email protected]>
Date: Tue, 7 Feb 2017 11:50:17 +0200
Subject: [PATCH] userfaultfd: mcopy_atomic: update cases returning -ENOENT

Quoted 8 lines

As Andrea commented in [1], if the VMA covering the address was unmapped, we may end up with a VMA a way above the faulting address. In this case we would like to return -ENOENT to allow uffd monitor detection of VMA removal.

[1] http://www.spinics.net/lists/linux-mm/msg121267.html

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>