[00/13] Clean up pmd_get_atomic() and i386-PAE

Message ID	20221022111403.531902164@infradead.org (mailing list archive)
Headers	show Return-Path: <owner-linux-mm@kvack.org> Message-ID: <20221022111403.531902164@infradead.org> User-Agent: quilt/0.66 Date: Sat, 22 Oct 2022 13:14:03 +0200 From: Peter Zijlstra <peterz@infradead.org> To: x86@kernel.org, willy@infradead.org, torvalds@linux-foundation.org, akpm@linux-foundation.org Cc: linux-kernel@vger.kernel.org, peterz@infradead.org, linux-mm@kvack.org, aarcange@redhat.com, kirill.shutemov@linux.intel.com, jroedel@suse.de, ubizjak@gmail.com Subject: [PATCH 00/13] Clean up pmd_get_atomic() and i386-PAE Sender: owner-linux-mm@kvack.org Precedence: bulk
Series	Clean up pmd_get_atomic() and i386-PAE \| expand [00/13] Clean up pmd_get_atomic() and i386-PAE [01/13] mm: Update ptep_get_lockless()s comment [02/13] x86/mm/pae: Make pmd_t similar to pte_t [03/13] sh/mm: Make pmd_t similar to pte_t [04/13] mm: Fix pmd_read_atomic() [05/13] mm: Rename GUP_GET_PTE_LOW_HIGH [06/13] mm: Rename pmd_read_atomic() [07/13] mm/gup: Fix the lockless PMD access [08/13] x86/mm/pae: Dont (ab)use atomic64 [09/13] x86/mm/pae: Use WRITE_ONCE() [10/13] x86/mm/pae: Be consistent with pXXp_get_and_clear() [11/13] x86_64: Remove pointless set_64bit() usage [12/13] x86/mm/pae: Get rid of set_64bit() [13/13] mm: Remove pointless barrier() after pmdp_get_lockless()

Message ID

20221022111403.531902164@infradead.org (mailing list archive)

Headers

Message-ID: <20221022111403.531902164@infradead.org>
User-Agent: quilt/0.66
Date: Sat, 22 Oct 2022 13:14:03 +0200
From: Peter Zijlstra <peterz@infradead.org>
To: x86@kernel.org,
 willy@infradead.org,
 torvalds@linux-foundation.org,
 akpm@linux-foundation.org
Cc: linux-kernel@vger.kernel.org,
 peterz@infradead.org,
 linux-mm@kvack.org,
 aarcange@redhat.com,
 kirill.shutemov@linux.intel.com,
 jroedel@suse.de,
 ubizjak@gmail.com
Subject: [PATCH 00/13] Clean up pmd_get_atomic() and i386-PAE
Sender: owner-linux-mm@kvack.org
Precedence: bulk

Series

Clean up pmd_get_atomic() and i386-PAE | expand

Message

Peter Zijlstra Oct. 22, 2022, 11:14 a.m. UTC

Hi,

At long *long* last a respin of the patches that clean up pmd_get_atomic() and
i386-PAE. I'd nearly forgotten why I did this, but the old posting gave clue
that patch #7 was the whole purpose of me doing these patches.

Having carried these patches for at least 2 years, they recently hit a rebase
bump against the mg-lru patches, which is what prompted this repost.

Linus' comment about try_cmpxchg64() (and Uros before him) made me redo those
patches (see patch #10) which resulted in pxx_xchg64(). This in turn led to
killing off set_64bit().

The robot doesn't hate on these patches and they boot in kvm (because who still
has i386 hardware).

Patches also available at:

  git://git.kernel.org/pub/scm/linux/kernel/git/peterz/queue.git x86/mm.pae

---
 arch/mips/Kconfig                           |   2 +-
 arch/sh/Kconfig                             |   2 +-
 arch/sh/include/asm/pgtable-3level.h        |  10 +-
 arch/um/include/asm/pgtable-3level.h        |   8 --
 arch/x86/Kconfig                            |   2 +-
 arch/x86/include/asm/cmpxchg_32.h           |  28 -----
 arch/x86/include/asm/cmpxchg_64.h           |   5 -
 arch/x86/include/asm/pgtable-3level.h       | 171 ++++++----------------------
 arch/x86/include/asm/pgtable-3level_types.h |   7 ++
 arch/x86/include/asm/pgtable_64_types.h     |   1 +
 arch/x86/include/asm/pgtable_types.h        |   4 +-
 drivers/iommu/intel/irq_remapping.c         |  10 +-
 include/linux/pgtable.h                     |  71 +++++++-----
 kernel/events/core.c                        |   2 +-
 mm/Kconfig                                  |   2 +-
 mm/gup.c                                    |   2 +-
 mm/hmm.c                                    |   3 +-
 mm/khugepaged.c                             |   2 +-
 mm/mapping_dirty_helpers.c                  |   2 +-
 mm/mprotect.c                               |   2 +-
 mm/userfaultfd.c                            |   2 +-
 mm/vmscan.c                                 |   5 +-
 22 files changed, 110 insertions(+), 233 deletions(-)

Comments

Linus Torvalds Oct. 22, 2022, 5:57 p.m. UTC | #1

On Sat, Oct 22, 2022 at 4:48 AM Peter Zijlstra <peterz@infradead.org> wrote:
>
> The robot doesn't hate on these patches and they boot in kvm (because who still
> has i386 hardware).

Well, I had a couple of comments, the only serious one being that odd
'__HAVE_ARCH_PMDP_GET' that I really didn't see anywhere else and
seemed actively wrong.

Other than that, it all looks good to me.

Thanks,
              Linus

Peter Zijlstra Oct. 29, 2022, 12:21 p.m. UTC | #2

On Sat, Oct 22, 2022 at 01:14:03PM +0200, Peter Zijlstra wrote:
> Hi,
> 
> At long *long* last a respin of the patches that clean up pmd_get_atomic() and
> i386-PAE. I'd nearly forgotten why I did this, but the old posting gave clue
> that patch #7 was the whole purpose of me doing these patches.
> 
> Having carried these patches for at least 2 years, they recently hit a rebase
> bump against the mg-lru patches, which is what prompted this repost.
> 
> Linus' comment about try_cmpxchg64() (and Uros before him) made me redo those
> patches (see patch #10) which resulted in pxx_xchg64(). This in turn led to
> killing off set_64bit().
> 
> The robot doesn't hate on these patches and they boot in kvm (because who still
> has i386 hardware).
> 
> Patches also available at:
> 
>   git://git.kernel.org/pub/scm/linux/kernel/git/peterz/queue.git x86/mm.pae

Aside from the whole discussions about TLB invalidates; I mean to commit
these patches to tip/x86/mm next week somewhere since they do improve
the current situation.

Holler if there's objections.