[PATCH 0/3] simple mmap() cleanups

[PATCH 0/3] simple mmap() cleanups

[Oleg Nesterov]

Hello.

Hopefully the simple cleanups, but need the authoritative acks.

I simply can't understand the VM_GROWS* check in mmap_region(),
did I miss something subtle? I will appreciate any review.

Can't resist. "(prot&PROT_WRITE) && !(file->f_mode&FMODE_WRITE)"
asks for the whitespace cleanup, but this will break the 80-col
rule. Perhaps we should relax it...

Oleg.

 mm/mmap.c |   28 ++++++++++------------------
 1 files changed, 10 insertions(+), 18 deletions(-)

[PATCH 1/3] mm: shift VM_GROWS* check from mmap_region() to do_mmap_pgoff()

[Oleg Nesterov]

mmap() doesn't allow the non-anonymous mappings with VM_GROWS* bit set.
In particular this means that mmap_region()->vma_merge(file, vm_flags)
must always fail if vm_flags & VM_GROWS. So it does not make sense to
check VM_GROWS* after we already allocated the new vma, the only caller,
do_mmap_pgoff(), which can pass this flag can do the check itself.

And this looks a bit more correct, mmap_region() already unmapped the
old mapping at this stage. But if mmap() is going to fail, it should
avoid do_munmap() if possible.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
---
 mm/mmap.c |    9 +++------
 1 files changed, 3 insertions(+), 6 deletions(-)

diff --git a/mm/mmap.c b/mm/mmap.c
index f681e18..6b69352 100644
--- a/mm/mmap.c
+++ b/mm/mmap.c
@@ -1327,6 +1327,9 @@ unsigned long do_mmap_pgoff(struct file *file, unsigned long addr,
 		}
 	}
 
+	/* Only MAP_PRIVATE|MAP_ANONYMOUS can use MAP_GROWS */
+	if ((vm_flags & VM_MAYSHARE) && (vm_flags & (VM_GROWSDOWN|VM_GROWSUP)))
+		return -EINVAL;
 	/*
 	 * Set 'VM_NORESERVE' if we should not account for the
 	 * memory use of this mapping.
@@ -1542,11 +1545,7 @@ munmap_back:
 	vma->vm_pgoff = pgoff;
 	INIT_LIST_HEAD(&vma->anon_vma_chain);
 
-	error = -EINVAL;	/* when rejecting VM_GROWSDOWN|VM_GROWSUP */
-
 	if (file) {
-		if (vm_flags & (VM_GROWSDOWN|VM_GROWSUP))
-			goto free_vma;
 		if (vm_flags & VM_DENYWRITE) {
 			error = deny_write_access(file);
 			if (error)
@@ -1571,8 +1570,6 @@ munmap_back:
 		pgoff = vma->vm_pgoff;
 		vm_flags = vma->vm_flags;
 	} else if (vm_flags & VM_SHARED) {
-		if (unlikely(vm_flags & (VM_GROWSDOWN|VM_GROWSUP)))
-			goto free_vma;
 		error = shmem_zero_setup(vma);
 		if (error)
 			goto free_vma;
-- 
1.5.5.1

[PATCH 2/3] mm: do_mmap_pgoff: cleanup the usage of file_inode()

[Oleg Nesterov]

Simple cleanup. Move "struct inode *inode" variable into "if (file)"
block to simplify the code and avoid the unnecessary check.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
---
 mm/mmap.c |    5 ++---
 1 files changed, 2 insertions(+), 3 deletions(-)

diff --git a/mm/mmap.c b/mm/mmap.c
index 6b69352..759dce7 100644
--- a/mm/mmap.c
+++ b/mm/mmap.c
@@ -1202,7 +1202,6 @@ unsigned long do_mmap_pgoff(struct file *file, unsigned long addr,
 			unsigned long *populate)
 {
 	struct mm_struct * mm = current->mm;
-	struct inode *inode;
 	vm_flags_t vm_flags;
 
 	*populate = 0;
@@ -1265,9 +1264,9 @@ unsigned long do_mmap_pgoff(struct file *file, unsigned long addr,
 			return -EAGAIN;
 	}
 
-	inode = file ? file_inode(file) : NULL;
-
 	if (file) {
+		struct inode *inode = file_inode(file);
+
 		switch (flags & MAP_TYPE) {
 		case MAP_SHARED:
 			if ((prot&PROT_WRITE) && !(file->f_mode&FMODE_WRITE))
-- 
1.5.5.1

[PATCH 3/3] mm: mmap_region: kill correct_wcount/inode, use allow_write_access()

[Oleg Nesterov]

correct_wcount and inode in mmap_region() just complicate the code.
This boolean was needed previously, when deny_write_access() was
called before vma_merge(), now we can simply check VM_DENYWRITE and
do allow_write_access() if it is set.

allow_write_access() checks file != NULL, so this is safe even if it
was possible to use VM_DENYWRITE && !file. Just we need to ensure we
use the same file which was deny_write_access()'ed, so the patch also
moves "file = vma->vm_file" down after allow_write_access().

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
---
 mm/mmap.c |   14 +++++---------
 1 files changed, 5 insertions(+), 9 deletions(-)

diff --git a/mm/mmap.c b/mm/mmap.c
index 759dce7..7be923a 100644
--- a/mm/mmap.c
+++ b/mm/mmap.c
@@ -1476,11 +1476,9 @@ unsigned long mmap_region(struct file *file, unsigned long addr,
 {
 	struct mm_struct *mm = current->mm;
 	struct vm_area_struct *vma, *prev;
-	int correct_wcount = 0;
 	int error;
 	struct rb_node **rb_link, *rb_parent;
 	unsigned long charged = 0;
-	struct inode *inode =  file ? file_inode(file) : NULL;
 
 	/* Check against address space limit. */
 	if (!may_expand_vm(mm, len >> PAGE_SHIFT)) {
@@ -1549,7 +1547,6 @@ munmap_back:
 			error = deny_write_access(file);
 			if (error)
 				goto free_vma;
-			correct_wcount = 1;
 		}
 		vma->vm_file = get_file(file);
 		error = file->f_op->mmap(file, vma);
@@ -1590,11 +1587,10 @@ munmap_back:
 	}
 
 	vma_link(mm, vma, prev, rb_link, rb_parent);
-	file = vma->vm_file;
-
 	/* Once vma denies write, undo our temporary denial count */
-	if (correct_wcount)
-		atomic_inc(&inode->i_writecount);
+	if (vm_flags & VM_DENYWRITE)
+		allow_write_access(file);
+	file = vma->vm_file;
 out:
 	perf_event_mmap(vma);
 
@@ -1613,8 +1609,8 @@ out:
 	return addr;
 
 unmap_and_free_vma:
-	if (correct_wcount)
-		atomic_inc(&inode->i_writecount);
+	if (vm_flags & VM_DENYWRITE)
+		allow_write_access(file);
 	vma->vm_file = NULL;
 	fput(file);
 
-- 
1.5.5.1

Re: [PATCH 1/3] mm: shift VM_GROWS* check from mmap_region() to do_mmap_pgoff()

[Andrew Morton]

On Sun, 14 Jul 2013 18:54:51 +0200 Oleg Nesterov <oleg@redhat.com> wrote:

> mmap() doesn't allow the non-anonymous mappings with VM_GROWS* bit set.
> In particular this means that mmap_region()->vma_merge(file, vm_flags)
> must always fail if vm_flags & VM_GROWS. So it does not make sense to
> check VM_GROWS* after we already allocated the new vma, the only caller,
> do_mmap_pgoff(), which can pass this flag can do the check itself.
> 
> And this looks a bit more correct, mmap_region() already unmapped the
> old mapping at this stage. But if mmap() is going to fail, it should
> avoid do_munmap() if possible.
> 
> ...
>
> --- a/mm/mmap.c
> +++ b/mm/mmap.c
> @@ -1327,6 +1327,9 @@ unsigned long do_mmap_pgoff(struct file *file, unsigned long addr,
>  		}
>  	}
>  
> +	/* Only MAP_PRIVATE|MAP_ANONYMOUS can use MAP_GROWS */
> +	if ((vm_flags & VM_MAYSHARE) && (vm_flags & (VM_GROWSDOWN|VM_GROWSUP)))

That is of course vm_flags&(VM_MAYSHARE|VM_GROWSDOWN|VM_GROWSUP), but
that perhaps is less clear.

> +		return -EINVAL;

I had to stare for a while but yes, the change looks OK to me.

Re: [PATCH 1/3] mm: shift VM_GROWS* check from mmap_region() to do_mmap_pgoff()

[Hugh Dickins]

On Tue, 16 Jul 2013, Andrew Morton wrote:
> On Sun, 14 Jul 2013 18:54:51 +0200 Oleg Nesterov <oleg@redhat.com> wrote:
> 
> > mmap() doesn't allow the non-anonymous mappings with VM_GROWS* bit set.
> > In particular this means that mmap_region()->vma_merge(file, vm_flags)
> > must always fail if vm_flags & VM_GROWS.

I didn't understand that sentence: if file is non-NULL perhaps?

> > So it does not make sense to
> > check VM_GROWS* after we already allocated the new vma, the only caller,
> > do_mmap_pgoff(), which can pass this flag can do the check itself.
> > 
> > And this looks a bit more correct, mmap_region() already unmapped the
> > old mapping at this stage. But if mmap() is going to fail, it should
> > avoid do_munmap() if possible.

I agree with the sentiment, but the patch looks wrong to me.

> > 
> > ...
> >
> > --- a/mm/mmap.c
> > +++ b/mm/mmap.c
> > @@ -1327,6 +1327,9 @@ unsigned long do_mmap_pgoff(struct file *file, unsigned long addr,
> >  		}
> >  	}
> >  
> > +	/* Only MAP_PRIVATE|MAP_ANONYMOUS can use MAP_GROWS */
> > +	if ((vm_flags & VM_MAYSHARE) && (vm_flags & (VM_GROWSDOWN|VM_GROWSUP)))
> 
> That is of course vm_flags&(VM_MAYSHARE|VM_GROWSDOWN|VM_GROWSUP),

Seems very plausible, but I believe you're wrong on that!

> but that perhaps is less clear.
> 
> > +		return -EINVAL;
> 
> I had to stare for a while but yes, the change looks OK to me.

It did need staring, yes, but it looks NOK to me: this change permits
mmap(addr, len, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_GROWSDOWN, fd, off)
where fd is for a real file: we never allowed MAP_GROWSDOWN on private
(or shared) mappings of real files before, and I think we should not now.

Hugh

[PATCH v2 0/1] mm: shift VM_GROWS* check from mmap_region() to do_mmap_pgoff()

[Oleg Nesterov]

Andrew,

This patch replaces the buggy

	mm-shift-vm_grows-check-from-mmap_region-to-do_mmap_pgoff.patch

I do not send the *-fix.patch because I'd like to update the changelog.

However it needs the explicit ack from Hugh.

On 07/19, Hugh Dickins wrote:
>
> On Tue, 16 Jul 2013, Andrew Morton wrote:
> > On Sun, 14 Jul 2013 18:54:51 +0200 Oleg Nesterov <oleg@redhat.com> wrote:
> >
> > > mmap() doesn't allow the non-anonymous mappings with VM_GROWS* bit set.
> > > In particular this means that mmap_region()->vma_merge(file, vm_flags)
> > > must always fail if vm_flags & VM_GROWS.
>
> I didn't understand that sentence: if file is non-NULL perhaps?

Yes, this looks confusing, sorry.

I meant, vma_merge() must fail if "vm_flags & VM_GROWS" is set incorrectly.
is_mergeable_vma() compares both vm_file and vm_flags.

Even if file == NULL (at this stage), "VM_SHARED | VM_GROWS" is not correct
too, and vma_merge() can't succeed.

And, to clarify, I only mentioned this because I tried to convince myself
that this change (if correct) can't make any difference except "avoid the
not-really-correct do_munmap".

> > > So it does not make sense to
> > > check VM_GROWS* after we already allocated the new vma, the only caller,
> > > do_mmap_pgoff(), which can pass this flag can do the check itself.
> > >
> > > And this looks a bit more correct, mmap_region() already unmapped the
> > > old mapping at this stage. But if mmap() is going to fail, it should
> > > avoid do_munmap() if possible.
>
> I agree with the sentiment, but the patch looks wrong to me.

Heh. You are right of course.

> It did need staring, yes, but it looks NOK to me: this change permits
> mmap(addr, len, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_GROWSDOWN, fd, off)
> where fd is for a real file:

And note that "Only MAP_PRIVATE|MAP_ANONYMOUS can use MAP_GROWS" even
tries to document that "MAP_PRIVATE && file" is not allowed too.

I have no idea how I managed to forget that MAP_PRIVATE never sets
VM_MAYSHARE.

Thanks a lot Hugh.

Oleg.

[PATCH v2 1/1] mm: shift VM_GROWS* check from mmap_region() to do_mmap_pgoff()

[Oleg Nesterov]

mmap() doesn't allow the non-anonymous mappings with VM_GROWS* bit set.
In particular this means that mmap_region()->vma_merge(file, vm_flags)
must always fail if "vm_flags & VM_GROWS" is set incorrectly.

So it does not make sense to check VM_GROWS* after we already allocated
the new vma, the only caller, do_mmap_pgoff(), which can pass this flag
can do the check itself.

And this looks a bit more correct, mmap_region() already unmapped the
old mapping at this stage. But if mmap() is going to fail, it should
avoid do_munmap() if possible.

Note: we check VM_GROWS at the end to ensure that do_mmap_pgoff() won't
return EINVAL in the case when it currently returns another error code.

Many thanks to Hugh who nacked the buggy v1.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
---
 mm/mmap.c |   10 ++++------
 1 files changed, 4 insertions(+), 6 deletions(-)

diff --git a/mm/mmap.c b/mm/mmap.c
index fbad7b0..92d9f54 100644
--- a/mm/mmap.c
+++ b/mm/mmap.c
@@ -1302,6 +1302,8 @@ unsigned long do_mmap_pgoff(struct file *file, unsigned long addr,
 
 			if (!file->f_op || !file->f_op->mmap)
 				return -ENODEV;
+			if (vm_flags & (VM_GROWSDOWN|VM_GROWSUP))
+				return -EINVAL;
 			break;
 
 		default:
@@ -1310,6 +1312,8 @@ unsigned long do_mmap_pgoff(struct file *file, unsigned long addr,
 	} else {
 		switch (flags & MAP_TYPE) {
 		case MAP_SHARED:
+			if (vm_flags & (VM_GROWSDOWN|VM_GROWSUP))
+				return -EINVAL;
 			/*
 			 * Ignore pgoff.
 			 */
@@ -1544,11 +1548,7 @@ munmap_back:
 	vma->vm_pgoff = pgoff;
 	INIT_LIST_HEAD(&vma->anon_vma_chain);
 
-	error = -EINVAL;	/* when rejecting VM_GROWSDOWN|VM_GROWSUP */
-
 	if (file) {
-		if (vm_flags & (VM_GROWSDOWN|VM_GROWSUP))
-			goto free_vma;
 		if (vm_flags & VM_DENYWRITE) {
 			error = deny_write_access(file);
 			if (error)
@@ -1573,8 +1573,6 @@ munmap_back:
 		pgoff = vma->vm_pgoff;
 		vm_flags = vma->vm_flags;
 	} else if (vm_flags & VM_SHARED) {
-		if (unlikely(vm_flags & (VM_GROWSDOWN|VM_GROWSUP)))
-			goto free_vma;
 		error = shmem_zero_setup(vma);
 		if (error)
 			goto free_vma;
-- 
1.5.5.1

Re: [PATCH v2 1/1] mm: shift VM_GROWS* check from mmap_region() to do_mmap_pgoff()

[Hugh Dickins]

On Sat, 20 Jul 2013, Oleg Nesterov wrote:

> mmap() doesn't allow the non-anonymous mappings with VM_GROWS* bit set.
> In particular this means that mmap_region()->vma_merge(file, vm_flags)
> must always fail if "vm_flags & VM_GROWS" is set incorrectly.
> 
> So it does not make sense to check VM_GROWS* after we already allocated
> the new vma, the only caller, do_mmap_pgoff(), which can pass this flag
> can do the check itself.
> 
> And this looks a bit more correct, mmap_region() already unmapped the
> old mapping at this stage. But if mmap() is going to fail, it should
> avoid do_munmap() if possible.
> 
> Note: we check VM_GROWS at the end to ensure that do_mmap_pgoff() won't
> return EINVAL in the case when it currently returns another error code.
> 
> Many thanks to Hugh who nacked the buggy v1.

Aw shucks, no need for that!

> 
> Signed-off-by: Oleg Nesterov <oleg@redhat.com>

Looks fine to me now, thanks Oleg; as do your other two.

Acked-by: Hugh Dickins <hughd@google.com>

> ---
>  mm/mmap.c |   10 ++++------
>  1 files changed, 4 insertions(+), 6 deletions(-)
> 
> diff --git a/mm/mmap.c b/mm/mmap.c
> index fbad7b0..92d9f54 100644
> --- a/mm/mmap.c
> +++ b/mm/mmap.c
> @@ -1302,6 +1302,8 @@ unsigned long do_mmap_pgoff(struct file *file, unsigned long addr,
>  
>  			if (!file->f_op || !file->f_op->mmap)
>  				return -ENODEV;
> +			if (vm_flags & (VM_GROWSDOWN|VM_GROWSUP))
> +				return -EINVAL;
>  			break;
>  
>  		default:
> @@ -1310,6 +1312,8 @@ unsigned long do_mmap_pgoff(struct file *file, unsigned long addr,
>  	} else {
>  		switch (flags & MAP_TYPE) {
>  		case MAP_SHARED:
> +			if (vm_flags & (VM_GROWSDOWN|VM_GROWSUP))
> +				return -EINVAL;
>  			/*
>  			 * Ignore pgoff.
>  			 */
> @@ -1544,11 +1548,7 @@ munmap_back:
>  	vma->vm_pgoff = pgoff;
>  	INIT_LIST_HEAD(&vma->anon_vma_chain);
>  
> -	error = -EINVAL;	/* when rejecting VM_GROWSDOWN|VM_GROWSUP */
> -
>  	if (file) {
> -		if (vm_flags & (VM_GROWSDOWN|VM_GROWSUP))
> -			goto free_vma;
>  		if (vm_flags & VM_DENYWRITE) {
>  			error = deny_write_access(file);
>  			if (error)
> @@ -1573,8 +1573,6 @@ munmap_back:
>  		pgoff = vma->vm_pgoff;
>  		vm_flags = vma->vm_flags;
>  	} else if (vm_flags & VM_SHARED) {
> -		if (unlikely(vm_flags & (VM_GROWSDOWN|VM_GROWSUP)))
> -			goto free_vma;
>  		error = shmem_zero_setup(vma);
>  		if (error)
>  			goto free_vma;
> -- 
> 1.5.5.1