On Thu, 13 Sep 2012 10:40:57 +0800
Bob Liu <lliubbo@xxxxxxxxx> wrote:
> The addr parameter may not page aligned eg. when it's come from
> vfb_mmap():vma->vm_start in video driver.
>
> This patch fix the check in remap_pfn_range() else some driver like v4l2 will
> fail in this function while calling mmap() on nommu arch like blackfin and st.
>
> Reported-by: Bhupesh SHARMA <bhupesh.sharma@xxxxxx>
> Reported-by: Scott Jiang <scott.jiang.linux@xxxxxxxxx>
> Signed-off-by: Bob Liu <lliubbo@xxxxxxxxx>
> ---
> mm/nommu.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/mm/nommu.c b/mm/nommu.c
> index d4b0c10..5d6068b 100644
> --- a/mm/nommu.c
> +++ b/mm/nommu.c
> @@ -1819,7 +1819,7 @@ struct page *follow_page(struct vm_area_struct *vma, unsigned long address,
> int remap_pfn_range(struct vm_area_struct *vma, unsigned long addr,
> unsigned long pfn, unsigned long size, pgprot_t prot)
> {
> - if (addr != (pfn << PAGE_SHIFT))
> + if ((addr & PAGE_MASK) != (pfn << PAGE_SHIFT))
> return -EINVAL;
>
> vma->vm_flags |= VM_IO | VM_RESERVED | VM_PFNMAP;
hm, what is the right thing to do here?
Yes, the MMU version of remap_pfn_range() does permit non-page-aligned
`addr' (at least, if the userspace maaping is a non-COW one). But I
suspect that was an implementation accident - it is a nonsensical thing
to do, isn't it? The MMU cannot map a bunch of kernel pages onto a
non-page-aligned userspace address.
So I'm thinking that we should declare ((addr & ~PAGE_MASK) != 0) to be
a caller bug, and fix up this regrettably unidentified v4l driver?
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@xxxxxxxxx. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>