2022-10-20 18:30:53

by Saurabh Singh Sengar

Subject: [PATCH] mm/gup: fix gup_pud_range() for dax

From: John Starks <[email protected]>

For dax pud, pud_huge() returns true on x86. So the function works as long
as hugetlb is configured. However, dax doesn't depend on hugetlb.
Commit 414fd080d125 ("mm/gup: fix gup_pmd_range() for dax") fixed
devmap-backed huge PMDs, but missed devmap-backed huge PUDs. Fix this as
well.

Fixes: 414fd080d125 ("mm/gup: fix gup_pmd_range() for dax")
Signed-off-by: John Starks <[email protected]>
Signed-off-by: Saurabh Sengar <[email protected]>
---
mm/gup.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/mm/gup.c b/mm/gup.c
index 05068d3d2557..9e07aa54a4cd 100644
--- a/mm/gup.c
+++ b/mm/gup.c
@@ -2687,7 +2687,7 @@ static int gup_pud_range(p4d_t *p4dp, p4d_t p4d, unsigned long addr, unsigned lo
next = pud_addr_end(addr, end);
if (unlikely(!pud_present(pud)))
return 0;
- if (unlikely(pud_huge(pud))) {
+ if (unlikely(pud_huge(pud) || pud_devmap(pud))) {
if (!gup_huge_pud(pud, pudp, addr, next, flags,
pages, nr))
return 0;
--
2.25.1

2022-10-20 22:29:02

by Yu Zhao

Subject: Re: [PATCH] mm/gup: fix gup_pud_range() for dax

On Thu, Oct 20, 2022 at 12:14 PM Saurabh Sengar
<[email protected]> wrote:
>
> From: John Starks <[email protected]>
>
> For dax pud, pud_huge() returns true on x86. So the function works as long
> as hugetlb is configured. However, dax doesn't depend on hugetlb.
> Commit 414fd080d125 ("mm/gup: fix gup_pmd_range() for dax") fixed
> devmap-backed huge PMDs, but missed devmap-backed huge PUDs. Fix this as
> well.
>
> Fixes: 414fd080d125 ("mm/gup: fix gup_pmd_range() for dax")
> Signed-off-by: John Starks <[email protected]>
> Signed-off-by: Saurabh Sengar <[email protected]>
> ---
> mm/gup.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/mm/gup.c b/mm/gup.c
> index 05068d3d2557..9e07aa54a4cd 100644
> --- a/mm/gup.c
> +++ b/mm/gup.c
> @@ -2687,7 +2687,7 @@ static int gup_pud_range(p4d_t *p4dp, p4d_t p4d, unsigned long addr, unsigned lo
> next = pud_addr_end(addr, end);
> if (unlikely(!pud_present(pud)))
> return 0;
> - if (unlikely(pud_huge(pud))) {
> + if (unlikely(pud_huge(pud) || pud_devmap(pud))) {

Perhaps s/pud_huge/pud_leaf/ ?

> if (!gup_huge_pud(pud, pudp, addr, next, flags,
> pages, nr))
> return 0;
> --
> 2.25.1
>

2022-12-07 06:13:33

by Saurabh Singh Sengar

Subject: Re: [PATCH] mm/gup: fix gup_pud_range() for dax

On Thu, Oct 20, 2022 at 04:18:37PM -0600, Yu Zhao wrote:
> On Thu, Oct 20, 2022 at 12:14 PM Saurabh Sengar
> <[email protected]> wrote:
> >
> > From: John Starks <[email protected]>
> >
> > For dax pud, pud_huge() returns true on x86. So the function works as long
> > as hugetlb is configured. However, dax doesn't depend on hugetlb.
> > Commit 414fd080d125 ("mm/gup: fix gup_pmd_range() for dax") fixed
> > devmap-backed huge PMDs, but missed devmap-backed huge PUDs. Fix this as
> > well.
> >
> > Fixes: 414fd080d125 ("mm/gup: fix gup_pmd_range() for dax")
> > Signed-off-by: John Starks <[email protected]>
> > Signed-off-by: Saurabh Sengar <[email protected]>
> > ---
> > mm/gup.c | 2 +-
> > 1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/mm/gup.c b/mm/gup.c
> > index 05068d3d2557..9e07aa54a4cd 100644
> > --- a/mm/gup.c
> > +++ b/mm/gup.c
> > @@ -2687,7 +2687,7 @@ static int gup_pud_range(p4d_t *p4dp, p4d_t p4d, unsigned long addr, unsigned lo
> > next = pud_addr_end(addr, end);
> > if (unlikely(!pud_present(pud)))
> > return 0;
> > - if (unlikely(pud_huge(pud))) {
> > + if (unlikely(pud_huge(pud) || pud_devmap(pud))) {
>
> Perhaps s/pud_huge/pud_leaf/ ?
Looks good to me but I am not sure the wider impact of this, I will let maintainers
comment on it. Meanwhile I will send a V2 with more description of panic observed.
>
> > if (!gup_huge_pud(pud, pudp, addr, next, flags,
> > pages, nr))
> > return 0;
> > --
> > 2.25.1
> >