From: Hugh Dickins <hugh@veritas.com>
To: Aaro Koskinen <aaro.koskinen@nokia.com>
Cc: "linux-arm-kernel@lists.arm.linux.org.uk"
<linux-arm-kernel@lists.arm.linux.org.uk>,
"linux-mm@kvack.org" <linux-mm@kvack.org>,
"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>
Subject: Re: [RFC PATCH 1/2] mm: tlb: Add range to tlb_start_vma() and tlb_end_vma()
Date: Mon, 9 Mar 2009 18:20:31 +0000 (GMT) [thread overview]
Message-ID: <Pine.LNX.4.64.0903091810400.17134@blonde.anvils> (raw)
In-Reply-To: <49B54B2A.9090408@nokia.com>
On Mon, 9 Mar 2009, Aaro Koskinen wrote:
> Hugh Dickins wrote:
>
> > I don't see that you need to change the interface and other arches
> > at all. What prevents ARM from noting the first and last addresses
> > freed in its struct mmu_gather when tlb_remove_tlb_entry() is called
> > (see arch/um/include/asm/tlb.h for an example of that), then using
> > that in its tlb_end_vma() TLB flushing?
>
> This would probably work, thanks for pointing it out. I should have taken a
> better look of the full API, not just what was implemented in ARM.
>
> So, there's a new ARM-only patch draft below based on this idea, adding also
> linux-arm-kernel again.
This one is much better, thank you. I would think it more natural
to do the initialization of range_start and range_end in your
tlb_start_vma() - to complement tlb_end_vma() where you deal with
the final result - rather than in two places you have sited it;
but that's somewhat a matter of taste, your patch should work as is.
Hugh
>
> ---
>
> From: Aaro Koskinen <Aaro.Koskinen@nokia.com>
> Subject: [RFC PATCH] [ARM] Flush only the needed range when unmapping VMA
>
> Signed-off-by: Aaro Koskinen <Aaro.Koskinen@nokia.com>
> ---
> arch/arm/include/asm/tlb.h | 25 ++++++++++++++++++++++---
> 1 files changed, 22 insertions(+), 3 deletions(-)
>
> diff --git a/arch/arm/include/asm/tlb.h b/arch/arm/include/asm/tlb.h
> index 857f1df..2729fb9 100644
> --- a/arch/arm/include/asm/tlb.h
> +++ b/arch/arm/include/asm/tlb.h
> @@ -36,6 +36,8 @@
> struct mmu_gather {
> struct mm_struct *mm;
> unsigned int fullmm;
> + unsigned long range_start;
> + unsigned long range_end;
> };
>
> DECLARE_PER_CPU(struct mmu_gather, mmu_gathers);
> @@ -47,6 +49,8 @@ tlb_gather_mmu(struct mm_struct *mm, unsigned int
> full_mm_flush)
>
> tlb->mm = mm;
> tlb->fullmm = full_mm_flush;
> + tlb->range_start = TASK_SIZE;
> + tlb->range_end = 0;
>
> return tlb;
> }
> @@ -63,7 +67,19 @@ tlb_finish_mmu(struct mmu_gather *tlb, unsigned long start,
> unsigned long end)
> put_cpu_var(mmu_gathers);
> }
>
> -#define tlb_remove_tlb_entry(tlb,ptep,address) do { } while (0)
> +/*
> + * Memorize the range for the TLB flush.
> + */
> +static inline void
> +tlb_remove_tlb_entry(struct mmu_gather *tlb, pte_t *ptep, unsigned long addr)
> +{
> + if (!tlb->fullmm) {
> + if (addr < tlb->range_start)
> + tlb->range_start = addr;
> + if (addr + PAGE_SIZE > tlb->range_end)
> + tlb->range_end = addr + PAGE_SIZE;
> + }
> +}
>
> /*
> * In the case of tlb vma handling, we can optimise these away in the
> @@ -80,8 +96,11 @@ tlb_start_vma(struct mmu_gather *tlb, struct vm_area_struct
> *vma)
> static inline void
> tlb_end_vma(struct mmu_gather *tlb, struct vm_area_struct *vma)
> {
> - if (!tlb->fullmm)
> - flush_tlb_range(vma, vma->vm_start, vma->vm_end);
> + if (!tlb->fullmm && tlb->range_end > 0) {
> + flush_tlb_range(vma, tlb->range_start, tlb->range_end);
> + tlb->range_start = TASK_SIZE;
> + tlb->range_end = 0;
> + }
> }
>
> #define tlb_remove_page(tlb,page) free_page_and_swap_cache(page)
> --
> 1.5.4.3
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
next prev parent reply other threads:[~2009-03-09 18:20 UTC|newest]
Thread overview: 13+ messages / expand[flat|nested] mbox.gz Atom feed top
2009-03-09 12:56 [RFC PATCH 0/2] mm: tlb: unmap scalability Aaro Koskinen
2009-03-09 12:59 ` [RFC PATCH 1/2] mm: tlb: Add range to tlb_start_vma() and tlb_end_vma() Aaro Koskinen
2009-03-09 12:59 ` [RFC PATCH 2/2] ARM: tlb: Use range in " Aaro Koskinen
2009-03-09 14:16 ` [RFC PATCH 1/2] mm: tlb: Add range to " Hugh Dickins
2009-03-09 17:00 ` Aaro Koskinen
2009-03-09 18:20 ` Hugh Dickins [this message]
2009-03-10 13:01 ` [PATCH] [ARM] Flush only the needed range when unmapping a VMA Aaro Koskinen
2009-03-10 14:49 ` Hugh Dickins
2009-03-12 21:30 ` Russell King - ARM Linux
2009-03-13 11:36 ` Aaro Koskinen
2009-03-15 10:27 ` Russell King - ARM Linux
2009-03-16 12:58 ` Aaro Koskinen
2009-03-16 13:30 ` [PATCH v2] " Aaro Koskinen
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=Pine.LNX.4.64.0903091810400.17134@blonde.anvils \
--to=hugh@veritas.com \
--cc=aaro.koskinen@nokia.com \
--cc=linux-arm-kernel@lists.arm.linux.org.uk \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox