Re: [PATCH 14/19] vfio, mm: pin_longterm_pages (FOLL_PIN) and put_user_page() conversion

[John Hubbard <jhubbard@nvidia.com>]

On 10/30/19 3:49 PM, John Hubbard wrote:
> This also fixes one or two likely bugs.

Well, actually just one...

> 
> 1. Change vfio from get_user_pages(FOLL_LONGTERM), to
> pin_longterm_pages(), which sets both FOLL_LONGTERM and FOLL_PIN.
> 
> Note that this is a change in behavior, because the
> get_user_pages_remote() call was not setting FOLL_LONGTERM, but the
> new pin_user_pages_remote() call that replaces it, *is* setting
> FOLL_LONGTERM. It is important to set FOLL_LONGTERM, because the
> DMA case requires it. Please see the FOLL_PIN documentation in
> include/linux/mm.h, and Documentation/pin_user_pages.rst for details.

Correction: the above comment is stale and wrong. I wrote it before 
getting further into the details, and the patch doesn't do this. 

Instead, it keeps exactly the old behavior: pin_longterm_pages_remote()
is careful to avoid setting FOLL_LONGTERM. Instead of setting that flag,
it drops in a "TODO" comment nearby. :)

I'll update the commit description in the next version of the series.


thanks,

John Hubbard
NVIDIA

> 
> 2. Because all FOLL_PIN-acquired pages must be released via
> put_user_page(), also convert the put_page() call over to
> put_user_pages().
> 
> Note that this effectively changes the code's behavior in
> vfio_iommu_type1.c: put_pfn(): it now ultimately calls
> set_page_dirty_lock(), instead of set_page_dirty(). This is
> probably more accurate.
> 
> As Christoph Hellwig put it, "set_page_dirty() is only safe if we are
> dealing with a file backed page where we have reference on the inode it
> hangs off." [1]
> 
> [1] https://lore.kernel.org/r/20190723153640.GB720@lst.de
> 
> Cc: Alex Williamson <alex.williamson@redhat.com>
> Signed-off-by: John Hubbard <jhubbard@nvidia.com>
> ---
>  drivers/vfio/vfio_iommu_type1.c | 15 +++++++--------
>  1 file changed, 7 insertions(+), 8 deletions(-)
> 
> diff --git a/drivers/vfio/vfio_iommu_type1.c b/drivers/vfio/vfio_iommu_type1.c
> index d864277ea16f..795e13f3ef08 100644
> --- a/drivers/vfio/vfio_iommu_type1.c
> +++ b/drivers/vfio/vfio_iommu_type1.c
> @@ -327,9 +327,8 @@ static int put_pfn(unsigned long pfn, int prot)
>  {
>  	if (!is_invalid_reserved_pfn(pfn)) {
>  		struct page *page = pfn_to_page(pfn);
> -		if (prot & IOMMU_WRITE)
> -			SetPageDirty(page);
> -		put_page(page);
> +
> +		put_user_pages_dirty_lock(&page, 1, prot & IOMMU_WRITE);
>  		return 1;
>  	}
>  	return 0;
> @@ -349,11 +348,11 @@ static int vaddr_get_pfn(struct mm_struct *mm, unsigned long vaddr,
>  
>  	down_read(&mm->mmap_sem);
>  	if (mm == current->mm) {
> -		ret = get_user_pages(vaddr, 1, flags | FOLL_LONGTERM, page,
> -				     vmas);
> +		ret = pin_longterm_pages(vaddr, 1, flags, page, vmas);
>  	} else {
> -		ret = get_user_pages_remote(NULL, mm, vaddr, 1, flags, page,
> -					    vmas, NULL);
> +		ret = pin_longterm_pages_remote(NULL, mm, vaddr, 1,
> +						flags, page, vmas,
> +						NULL);
>  		/*
>  		 * The lifetime of a vaddr_get_pfn() page pin is
>  		 * userspace-controlled. In the fs-dax case this could
> @@ -363,7 +362,7 @@ static int vaddr_get_pfn(struct mm_struct *mm, unsigned long vaddr,
>  		 */
>  		if (ret > 0 && vma_is_fsdax(vmas[0])) {
>  			ret = -EOPNOTSUPP;
> -			put_page(page[0]);
> +			put_user_page(page[0]);
>  		}
>  	}
>  	up_read(&mm->mmap_sem);
>