mbox series

[0/2] Fix two above-47bit hint address vs. THP bugs

Message ID 20191220142548.7118-1-kirill.shutemov@linux.intel.com (mailing list archive)
Headers show
Series Fix two above-47bit hint address vs. THP bugs | expand

Message

Kirill A. Shutemov Dec. 20, 2019, 2:25 p.m. UTC 
The two get_unmapped_area() implementations have to be fixed to provide
THP-friendly mappings if above-47bit hint address is specified.

Kirill A. Shutemov (2):
  thp: Fix conflict of above-47bit hint address and PMD alignment
  thp, shmem: Fix conflict of above-47bit hint address and PMD alignment

 mm/huge_memory.c | 38 ++++++++++++++++++++++++--------------
 mm/shmem.c       |  7 ++++---
 2 files changed, 28 insertions(+), 17 deletions(-)

Comments

Andrew Morton Dec. 23, 2019, 10:25 p.m. UTC | #1 
On Fri, 20 Dec 2019 17:25:46 +0300 "Kirill A. Shutemov" <kirill@shutemov.name> wrote:

> The two get_unmapped_area() implementations have to be fixed to provide
> THP-friendly mappings if above-47bit hint address is specified.
> 

Do we need a cc:stable for these?
Kirill A. Shutemov Dec. 23, 2019, 11:16 p.m. UTC | #2 
On Mon, Dec 23, 2019 at 02:25:52PM -0800, Andrew Morton wrote:
> On Fri, 20 Dec 2019 17:25:46 +0300 "Kirill A. Shutemov" <kirill@shutemov.name> wrote:
> 
> > The two get_unmapped_area() implementations have to be fixed to provide
> > THP-friendly mappings if above-47bit hint address is specified.
> > 
> 
> Do we need a cc:stable for these?

Yes, please. Otherwise users that use above-47bit hint would regress on
number of PMD-mapped THPs.

But it's not urgent. It's okay to cook it up in linux-next for a while.