[PATCH 5.10.y] dma-buf: heaps: Fix off-by-one in CMA heap fault handler

[PATCH 5.10.y] dma-buf: heaps: Fix off-by-one in CMA heap fault handler

[T.J. Mercier]

commit ea5ff5d351b520524019f7ff7f9ce418de2dad87 upstream.

Until VM_DONTEXPAND was added in commit 1c1914d6e8c6 ("dma-buf: heaps:
Don't track CMA dma-buf pages under RssFile") it was possible to obtain
a mapping larger than the buffer size via mremap and bypass the overflow
check in dma_buf_mmap_internal. When using such a mapping to attempt to
fault past the end of the buffer, the CMA heap fault handler also checks
the fault offset against the buffer size, but gets the boundary wrong by
1. Fix the boundary check so that we don't read off the end of the pages
array and insert an arbitrary page in the mapping.

Reported-by: Xingyu Jin <xingyuj@google.com>
Fixes: a5d2d29e24be ("dma-buf: heaps: Move heap-helper logic into the cma_heap implementation")
Cc: stable@vger.kernel.org # Applicable >= 5.10. Needs adjustments only for 5.10.
Signed-off-by: T.J. Mercier <tjmercier@google.com>
Acked-by: John Stultz <jstultz@google.com>
Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240830192627.2546033-1-tjmercier@google.com
[TJ: Backport to 5.10. On this kernel the bug is located in
dma_heap_vm_fault which is used by both the CMA and system heaps.]
Signed-off-by: T.J. Mercier <tjmercier@google.com>
---
 drivers/dma-buf/heaps/heap-helpers.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/dma-buf/heaps/heap-helpers.c b/drivers/dma-buf/heaps/heap-helpers.c
index d0696cf937af..a852b5e8122f 100644
--- a/drivers/dma-buf/heaps/heap-helpers.c
+++ b/drivers/dma-buf/heaps/heap-helpers.c
@@ -161,7 +161,7 @@ static vm_fault_t dma_heap_vm_fault(struct vm_fault *vmf)
 	struct vm_area_struct *vma = vmf->vma;
 	struct heap_helper_buffer *buffer = vma->vm_private_data;
 
-	if (vmf->pgoff > buffer->pagecount)
+	if (vmf->pgoff >= buffer->pagecount)
 		return VM_FAULT_SIGBUS;
 
 	vmf->page = buffer->pages[vmf->pgoff];
-- 
2.46.0.662.g92d0881bb0-goog

Re: [PATCH 5.10.y] dma-buf: heaps: Fix off-by-one in CMA heap fault handler

[Greg KH]

On Mon, Sep 16, 2024 at 04:34:41AM +0000, T.J. Mercier wrote:
> commit ea5ff5d351b520524019f7ff7f9ce418de2dad87 upstream.
> 
> Until VM_DONTEXPAND was added in commit 1c1914d6e8c6 ("dma-buf: heaps:
> Don't track CMA dma-buf pages under RssFile") it was possible to obtain
> a mapping larger than the buffer size via mremap and bypass the overflow
> check in dma_buf_mmap_internal. When using such a mapping to attempt to
> fault past the end of the buffer, the CMA heap fault handler also checks
> the fault offset against the buffer size, but gets the boundary wrong by
> 1. Fix the boundary check so that we don't read off the end of the pages
> array and insert an arbitrary page in the mapping.
> 
> Reported-by: Xingyu Jin <xingyuj@google.com>
> Fixes: a5d2d29e24be ("dma-buf: heaps: Move heap-helper logic into the cma_heap implementation")

This commit is in 5.11, so why:

> Cc: stable@vger.kernel.org # Applicable >= 5.10. Needs adjustments only for 5.10.

does this say 5.10?

thanks,

greg k-h

Re: [PATCH 5.10.y] dma-buf: heaps: Fix off-by-one in CMA heap fault handler

[T.J. Mercier]

On Mon, Sep 16, 2024 at 12:02 AM Greg KH <gregkh@linuxfoundation.org> wrote:
>
> On Mon, Sep 16, 2024 at 04:34:41AM +0000, T.J. Mercier wrote:
> > commit ea5ff5d351b520524019f7ff7f9ce418de2dad87 upstream.
> >
> > Until VM_DONTEXPAND was added in commit 1c1914d6e8c6 ("dma-buf: heaps:
> > Don't track CMA dma-buf pages under RssFile") it was possible to obtain
> > a mapping larger than the buffer size via mremap and bypass the overflow
> > check in dma_buf_mmap_internal. When using such a mapping to attempt to
> > fault past the end of the buffer, the CMA heap fault handler also checks
> > the fault offset against the buffer size, but gets the boundary wrong by
> > 1. Fix the boundary check so that we don't read off the end of the pages
> > array and insert an arbitrary page in the mapping.
> >
> > Reported-by: Xingyu Jin <xingyuj@google.com>
> > Fixes: a5d2d29e24be ("dma-buf: heaps: Move heap-helper logic into the cma_heap implementation")
>
> This commit is in 5.11, so why:
>
> > Cc: stable@vger.kernel.org # Applicable >= 5.10. Needs adjustments only for 5.10.
>
> does this say 5.10?
>
> thanks,
>
> greg k-h

a5d2d29e24be ("dma-buf: heaps: Move heap-helper logic into the
cma_heap implementation") moved the code from this heap-helpers.c file
to cma_heap.c in 5.11, which is the only place it lives upstream now.
The bug still exists in the original location in this heap-helpers.c
file on 5.10.

Re: [PATCH 5.10.y] dma-buf: heaps: Fix off-by-one in CMA heap fault handler

[Greg KH]

On Mon, Sep 16, 2024 at 12:38:26AM -0700, T.J. Mercier wrote:
> On Mon, Sep 16, 2024 at 12:02 AM Greg KH <gregkh@linuxfoundation.org> wrote:
> >
> > On Mon, Sep 16, 2024 at 04:34:41AM +0000, T.J. Mercier wrote:
> > > commit ea5ff5d351b520524019f7ff7f9ce418de2dad87 upstream.
> > >
> > > Until VM_DONTEXPAND was added in commit 1c1914d6e8c6 ("dma-buf: heaps:
> > > Don't track CMA dma-buf pages under RssFile") it was possible to obtain
> > > a mapping larger than the buffer size via mremap and bypass the overflow
> > > check in dma_buf_mmap_internal. When using such a mapping to attempt to
> > > fault past the end of the buffer, the CMA heap fault handler also checks
> > > the fault offset against the buffer size, but gets the boundary wrong by
> > > 1. Fix the boundary check so that we don't read off the end of the pages
> > > array and insert an arbitrary page in the mapping.
> > >
> > > Reported-by: Xingyu Jin <xingyuj@google.com>
> > > Fixes: a5d2d29e24be ("dma-buf: heaps: Move heap-helper logic into the cma_heap implementation")
> >
> > This commit is in 5.11, so why:
> >
> > > Cc: stable@vger.kernel.org # Applicable >= 5.10. Needs adjustments only for 5.10.
> >
> > does this say 5.10?
> >
> > thanks,
> >
> > greg k-h
> 
> a5d2d29e24be ("dma-buf: heaps: Move heap-helper logic into the
> cma_heap implementation") moved the code from this heap-helpers.c file
> to cma_heap.c in 5.11, which is the only place it lives upstream now.
> The bug still exists in the original location in this heap-helpers.c
> file on 5.10.

Ah, then that was the wrong Fixes: tag :(

thanks, I'll go queue this up now.

greg k-h

Re: [PATCH 5.10.y] dma-buf: heaps: Fix off-by-one in CMA heap fault handler

[T.J. Mercier]

On Mon, Sep 16, 2024 at 12:44 AM Greg KH <gregkh@linuxfoundation.org> wrote:
>
> On Mon, Sep 16, 2024 at 12:38:26AM -0700, T.J. Mercier wrote:
> > On Mon, Sep 16, 2024 at 12:02 AM Greg KH <gregkh@linuxfoundation.org> wrote:
> > >
> > > On Mon, Sep 16, 2024 at 04:34:41AM +0000, T.J. Mercier wrote:
> > > > commit ea5ff5d351b520524019f7ff7f9ce418de2dad87 upstream.
> > > >
> > > > Until VM_DONTEXPAND was added in commit 1c1914d6e8c6 ("dma-buf: heaps:
> > > > Don't track CMA dma-buf pages under RssFile") it was possible to obtain
> > > > a mapping larger than the buffer size via mremap and bypass the overflow
> > > > check in dma_buf_mmap_internal. When using such a mapping to attempt to
> > > > fault past the end of the buffer, the CMA heap fault handler also checks
> > > > the fault offset against the buffer size, but gets the boundary wrong by
> > > > 1. Fix the boundary check so that we don't read off the end of the pages
> > > > array and insert an arbitrary page in the mapping.
> > > >
> > > > Reported-by: Xingyu Jin <xingyuj@google.com>
> > > > Fixes: a5d2d29e24be ("dma-buf: heaps: Move heap-helper logic into the cma_heap implementation")
> > >
> > > This commit is in 5.11, so why:
> > >
> > > > Cc: stable@vger.kernel.org # Applicable >= 5.10. Needs adjustments only for 5.10.
> > >
> > > does this say 5.10?
> > >
> > > thanks,
> > >
> > > greg k-h
> >
> > a5d2d29e24be ("dma-buf: heaps: Move heap-helper logic into the
> > cma_heap implementation") moved the code from this heap-helpers.c file
> > to cma_heap.c in 5.11, which is the only place it lives upstream now.
> > The bug still exists in the original location in this heap-helpers.c
> > file on 5.10.
>
> Ah, then that was the wrong Fixes: tag :(
>
> thanks, I'll go queue this up now.
>
> greg k-h

Ok sorry about that. Thanks.