2020-11-06 11:50:04

by Christian König

Subject: [PATCH 1/2] mm: mmap: fix fput in error path v2

Patch "495c10cc1c0c CHROMIUM: dma-buf: restore args..."
adds a workaround for a bug in mmap_region.

As the comment states ->mmap() callback can change
vma->vm_file and so we might call fput() on the wrong file.

Revert the workaround and proper fix this in mmap_region.

v2: drop the extra if in dma_buf_mmap as well

Signed-off-by: Christian König <[email protected]>
Reviewed-by: Jason Gunthorpe <[email protected]>
---
drivers/dma-buf/dma-buf.c | 20 +++-----------------
mm/mmap.c | 2 +-
2 files changed, 4 insertions(+), 18 deletions(-)

diff --git a/drivers/dma-buf/dma-buf.c b/drivers/dma-buf/dma-buf.c
index 0eb80c1ecdab..282bd8b84170 100644
--- a/drivers/dma-buf/dma-buf.c
+++ b/drivers/dma-buf/dma-buf.c
@@ -1166,9 +1166,6 @@ EXPORT_SYMBOL_GPL(dma_buf_end_cpu_access);
int dma_buf_mmap(struct dma_buf *dmabuf, struct vm_area_struct *vma,
unsigned long pgoff)
{
- struct file *oldfile;
- int ret;
-
if (WARN_ON(!dmabuf || !vma))
return -EINVAL;

@@ -1186,22 +1183,11 @@ int dma_buf_mmap(struct dma_buf *dmabuf, struct vm_area_struct *vma,
return -EINVAL;

/* readjust the vma */
- get_file(dmabuf->file);
- oldfile = vma->vm_file;
- vma->vm_file = dmabuf->file;
+ fput(vma->vm_file);
+ vma->vm_file = get_file(dmabuf->file);
vma->vm_pgoff = pgoff;

- ret = dmabuf->ops->mmap(dmabuf, vma);
- if (ret) {
- /* restore old parameters on failure */
- vma->vm_file = oldfile;
- fput(dmabuf->file);
- } else {
- if (oldfile)
- fput(oldfile);
- }
- return ret;
-
+ return dmabuf->ops->mmap(dmabuf, vma);
}
EXPORT_SYMBOL_GPL(dma_buf_mmap);

diff --git a/mm/mmap.c b/mm/mmap.c
index d91ecb00d38c..30a4e8412a58 100644
--- a/mm/mmap.c
+++ b/mm/mmap.c
@@ -1899,8 +1899,8 @@ unsigned long mmap_region(struct file *file, unsigned long addr,
return addr;

unmap_and_free_vma:
+ fput(vma->vm_file);
vma->vm_file = NULL;
- fput(file);

/* Undo any partial mapping done by a device driver. */
unmap_region(mm, vma, prev, vma->vm_start, vma->vm_end);
--
2.25.1

2020-11-06 22:50:15

by Andrew Morton

Subject: Re: [PATCH 1/2] mm: mmap: fix fput in error path v2

On Fri, 6 Nov 2020 12:48:05 +0100 "Christian K?nig" <[email protected]> wrote:

> Patch "495c10cc1c0c CHROMIUM: dma-buf: restore args..."
> adds a workaround for a bug in mmap_region.
>
> As the comment states ->mmap() callback can change
> vma->vm_file and so we might call fput() on the wrong file.
>
> Revert the workaround and proper fix this in mmap_region.
>

Seems correct, best I can tell. Presumably all ->mmap() instances will
correctly fput() to original file* if they're rewriting vma->vm_file.

2020-11-18 11:00:28

by Christian König

Subject: Re: [PATCH 1/2] mm: mmap: fix fput in error path v2

Am 06.11.20 um 23:48 schrieb Andrew Morton:
> On Fri, 6 Nov 2020 12:48:05 +0100 "Christian König" <[email protected]> wrote:
>
>> Patch "495c10cc1c0c CHROMIUM: dma-buf: restore args..."
>> adds a workaround for a bug in mmap_region.
>>
>> As the comment states ->mmap() callback can change
>> vma->vm_file and so we might call fput() on the wrong file.
>>
>> Revert the workaround and proper fix this in mmap_region.
>>
> Seems correct, best I can tell. Presumably all ->mmap() instances will
> correctly fput() to original file* if they're rewriting vma->vm_file.

Yes, exactly.

Patch #2 provides a helper to make sure that everybody gets the
get_file()/fput() correctly while updating vma->vm_file.

Can I add your acked-by to the patches and push them upstream through
drm-misc-next?

Thanks,
Christian.

2020-11-18 22:33:15

by Andrew Morton

Subject: Re: [PATCH 1/2] mm: mmap: fix fput in error path v2

On Wed, 18 Nov 2020 11:57:44 +0100 Christian K?nig <[email protected]> wrote:

> Am 06.11.20 um 23:48 schrieb Andrew Morton:
> > On Fri, 6 Nov 2020 12:48:05 +0100 "Christian K?nig" <[email protected]> wrote:
> >
> >> Patch "495c10cc1c0c CHROMIUM: dma-buf: restore args..."
> >> adds a workaround for a bug in mmap_region.
> >>
> >> As the comment states ->mmap() callback can change
> >> vma->vm_file and so we might call fput() on the wrong file.
> >>
> >> Revert the workaround and proper fix this in mmap_region.
> >>
> > Seems correct, best I can tell. Presumably all ->mmap() instances will
> > correctly fput() to original file* if they're rewriting vma->vm_file.
>
> Yes, exactly.
>
> Patch #2 provides a helper to make sure that everybody gets the
> get_file()/fput() correctly while updating vma->vm_file.
>
> Can I add your acked-by to the patches and push them upstream through
> drm-misc-next?

Please go ahead.

Acked-by: Andrew Morton <[email protected]>