On Tue, Mar 19, 2019 at 01:01:01PM -0700, John Hubbard wrote:
> On 3/19/19 7:06 AM, Kirill A. Shutemov wrote:
> > On Tue, Mar 19, 2019 at 09:47:24AM -0400, Jerome Glisse wrote:
> >> On Tue, Mar 19, 2019 at 03:04:17PM +0300, Kirill A. Shutemov wrote:
> >>> On Fri, Mar 08, 2019 at 01:36:33PM -0800, john.hubb...@gmail.com wrote:
> >>>> From: John Hubbard <jhubb...@nvidia.com>
> >>
> >> [...]
> >>
> >>>> diff --git a/mm/gup.c b/mm/gup.c
> >>>> index f84e22685aaa..37085b8163b1 100644
> >>>> --- a/mm/gup.c
> >>>> +++ b/mm/gup.c
> >>>> @@ -28,6 +28,88 @@ struct follow_page_context {
> >>>> unsigned int page_mask;
> >>>> };
> >>>>
> >>>> +typedef int (*set_dirty_func_t)(struct page *page);
> >>>> +
> >>>> +static void __put_user_pages_dirty(struct page **pages,
> >>>> + unsigned long npages,
> >>>> + set_dirty_func_t sdf)
> >>>> +{
> >>>> + unsigned long index;
> >>>> +
> >>>> + for (index = 0; index < npages; index++) {
> >>>> + struct page *page = compound_head(pages[index]);
> >>>> +
> >>>> + if (!PageDirty(page))
> >>>> + sdf(page);
> >>>
> >>> How is this safe? What prevents the page to be cleared under you?
> >>>
> >>> If it's safe to race clear_page_dirty*() it has to be stated explicitly
> >>> with a reason why. It's not very clear to me as it is.
> >>
> >> The PageDirty() optimization above is fine to race with clear the
> >> page flag as it means it is racing after a page_mkclean() and the
> >> GUP user is done with the page so page is about to be write back
> >> ie if (!PageDirty(page)) see the page as dirty and skip the sdf()
> >> call while a split second after TestClearPageDirty() happens then
> >> it means the racing clear is about to write back the page so all
> >> is fine (the page was dirty and it is being clear for write back).
> >>
> >> If it does call the sdf() while racing with write back then we
> >> just redirtied the page just like clear_page_dirty_for_io() would
> >> do if page_mkclean() failed so nothing harmful will come of that
> >> neither. Page stays dirty despite write back it just means that
> >> the page might be write back twice in a row.
> >
> > Fair enough. Should we get it into a comment here?
>
> How's this read to you? I reworded and slightly expanded Jerome's
> description:
>
> diff --git a/mm/gup.c b/mm/gup.c
> index d1df7b8ba973..86397ae23922 100644
> --- a/mm/gup.c
> +++ b/mm/gup.c
> @@ -61,6 +61,24 @@ static void __put_user_pages_dirty(struct page **pages,
> for (index = 0; index < npages; index++) {
> struct page *page = compound_head(pages[index]);
>
> + /*
> + * Checking PageDirty at this point may race with
> + * clear_page_dirty_for_io(), but that's OK. Two key cases:
> + *
> + * 1) This code sees the page as already dirty, so it skips
> + * the call to sdf(). That could happen because
> + * clear_page_dirty_for_io() called page_mkclean(),
> + * followed by set_page_dirty(). However, now the page is
> + * going to get written back, which meets the original
> + * intention of setting it dirty, so all is well:
> + * clear_page_dirty_for_io() goes on to call
> + * TestClearPageDirty(), and write the page back.
> + *
> + * 2) This code sees the page as clean, so it calls sdf().
> + * The page stays dirty, despite being written back, so it
> + * gets written back again in the next writeback cycle.
> + * This is harmless.
> + */
> if (!PageDirty(page))
> sdf(page);
Looks good to me.
Other nit: effectively the same type of callback called 'spd' in
set_page_dirty(). Should we rename 'sdf' to 'sdp' here too?
> >>>> +void put_user_pages(struct page **pages, unsigned long npages)
> >>>> +{
> >>>> + unsigned long index;
> >>>> +
> >>>> + for (index = 0; index < npages; index++)
> >>>> + put_user_page(pages[index]);
> >>>
> >>> I believe there's an room for improvement for compound pages.
> >>>
> >>> If there's multiple consequential pages in the array that belong to the
> >>> same compound page we can get away with a single atomic operation to
> >>> handle them all.
> >>
> >> Yes maybe just add a comment with that for now and leave this kind of
> >> optimization to latter ?
> >
> > Sounds good to me.
> >
>
> Here's a comment for that:
>
> @@ -127,6 +145,11 @@ void put_user_pages(struct page **pages, unsigned long
> npages)
> {
> unsigned long index;
>
> + /*
> + * TODO: this can be optimized for huge pages: if a series of pages is
> + * physically contiguous and part of the same compound page, then a
Comound pages are always physically contiguous. I initially ment that the
optimization makes sense if they are next to each other in 'pages' array.
> + * single operation to the head page should suffice.
> + */
> for (index = 0; index < npages; index++)
> put_user_page(pages[index]);
> }
>
>
> thanks,
> --
> John Hubbard
> NVIDIA
--
Kirill A. Shutemov
