Re: [PATCHv7 00/14] x86: 5-level paging enabling for v4.13, Part 4
From: Ingo Molnar
Date: Thu Jun 22 2017 - 05:21:41 EST
* Kirill A. Shutemov <kirill@xxxxxxxxxxxxx> wrote:
> On Thu, Jun 22, 2017 at 11:04:22AM +0200, Ingo Molnar wrote:
> > * Kirill A. Shutemov <kirill@xxxxxxxxxxxxx> wrote:
> > > On Tue, Jun 06, 2017 at 02:31:19PM +0300, Kirill A. Shutemov wrote:
> > > > Please review and consider applying.
> > >
> > > Hi Ingo,
> > >
> > > I've noticed you haven't applied last two patches of the patchset.
> > >
> > > Is there any problem with them? Or what is you plan here?
> > As they change/extend the Linux ABI I still need to think about them some more.
> Okay, I see.
> Let me know if any action is required from my side.
Yeah, so I had a look, and the ABI principle of using the mmap() address hint to
trigger 57-bit address space allocations still looks mostly good to me, but please
split up this patch:
Subject: [PATCHv7 14/14] x86/mm: Allow to have userspace mappings above 47-bits
arch/x86/include/asm/elf.h | 4 ++--
arch/x86/include/asm/mpx.h | 9 +++++++++
arch/x86/include/asm/processor.h | 12 +++++++++---
arch/x86/kernel/sys_x86_64.c | 30 ++++++++++++++++++++++++++----
arch/x86/mm/hugetlbpage.c | 27 +++++++++++++++++++++++----
arch/x86/mm/mmap.c | 6 +++---
arch/x86/mm/mpx.c | 33 ++++++++++++++++++++++++++++++++-
7 files changed, 104 insertions(+), 17 deletions(-)
One patch should add the MPX quirk, another should add all the TASK_SIZE
variations, without actually changing the logic, etc. - while the final patch adds
the larger task size. Please try to split it into as many patches as possible -
I'd say 4-5 look ideal. All of this changes existing code paths and if things
break we'd like some small patch to be bisected to. The finer grained structure
also makes review easier.
Also, please rename the tasksize_*() functions to task_size_ (in yet another
patch) - it's absolutely silly that we have 'TASK_SIZE' in uppercase but
'tasksize' in lowercase ...