From: Brian Gerst <brgerst@gmail.com>
To: Alexey Dobriyan <adobriyan@gmail.com>
Cc: "the arch/x86 maintainers" <x86@kernel.org>,
Linux Kernel Mailing List <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH] x86: branchless clear_page()
Date: Tue, 23 Aug 2016 09:43:25 -0400 [thread overview]
Message-ID: <CAMzpN2gsgEdk5SgsyzP8mwgtta4-nYjEui9Cw3CHkRXoxVVsZg@mail.gmail.com> (raw)
In-Reply-To: <20160822205229.GA2638@p183.telecom.by>
On Mon, Aug 22, 2016 at 4:52 PM, Alexey Dobriyan <adobriyan@gmail.com> wrote:
> Apply alternatives at the call site instead of function body.
> Save branch per clean page.
>
> Bonus: tell gcc to not flush whole shebang of registers,
> just RDI, RAX, RCX.
>
> Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
> ---
>
> arch/x86/include/asm/page_64.h | 16 +++++++++++++++-
> arch/x86/lib/clear_page_64.S | 18 ++++++------------
> 2 files changed, 21 insertions(+), 13 deletions(-)
>
> --- a/arch/x86/include/asm/page_64.h
> +++ b/arch/x86/include/asm/page_64.h
> @@ -4,6 +4,7 @@
> #include <asm/page_64_types.h>
>
> #ifndef __ASSEMBLY__
> +#include <asm/alternative.h>
>
> /* duplicated to the one in bootmem.h */
> extern unsigned long max_pfn;
> @@ -34,7 +35,20 @@ extern unsigned long __phys_addr_symbol(unsigned long);
> #define pfn_valid(pfn) ((pfn) < max_pfn)
> #endif
>
> -void clear_page(void *page);
> +void clear_page_mov(void *page);
> +void clear_page_rep_stosq(void *page);
> +void clear_page_rep_stosb(void *page);
> +static __always_inline void clear_page(void *page)
> +{
> + alternative_call_2(
> + clear_page_mov,
> + clear_page_rep_stosq, X86_FEATURE_REP_GOOD,
> + clear_page_rep_stosb, X86_FEATURE_ERMS,
> + "=D" (page),
> + "0" (page)
> + : "rax", "rcx", "memory"
> + );
> +}
> void copy_page(void *to, void *from);
>
> #endif /* !__ASSEMBLY__ */
> --- a/arch/x86/lib/clear_page_64.S
> +++ b/arch/x86/lib/clear_page_64.S
> @@ -1,6 +1,4 @@
> #include <linux/linkage.h>
> -#include <asm/cpufeatures.h>
> -#include <asm/alternative-asm.h>
>
> /*
> * Most CPUs support enhanced REP MOVSB/STOSB instructions. It is
> @@ -13,18 +11,14 @@
> * Zero a page.
> * %rdi - page
> */
> -ENTRY(clear_page)
> -
> - ALTERNATIVE_2 "jmp clear_page_orig", "", X86_FEATURE_REP_GOOD, \
> - "jmp clear_page_c_e", X86_FEATURE_ERMS
> -
> +ENTRY(clear_page_rep_stosq)
> movl $4096/8,%ecx
> xorl %eax,%eax
> rep stosq
> ret
> -ENDPROC(clear_page)
> +ENDPROC(clear_page_rep_stosq)
>
> -ENTRY(clear_page_orig)
> +ENTRY(clear_page_mov)
>
> xorl %eax,%eax
> movl $4096/64,%ecx
> @@ -44,11 +38,11 @@ ENTRY(clear_page_orig)
> jnz .Lloop
> nop
> ret
> -ENDPROC(clear_page_orig)
> +ENDPROC(clear_page_mov)
>
> -ENTRY(clear_page_c_e)
> +ENTRY(clear_page_rep_stosb)
> movl $4096,%ecx
> xorl %eax,%eax
> rep stosb
> ret
> -ENDPROC(clear_page_c_e)
> +ENDPROC(clear_page_rep_stosb)
I like this idea, but does it make sense to take it a step further and
inline the string instruction alternatives to avoid a call altogether?
Also, 32-bit should be converted to do the same thing as 64-bit.
--
Brian Gerst
next prev parent reply other threads:[~2016-08-23 13:43 UTC|newest]
Thread overview: 9+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-08-22 20:52 [PATCH] x86: branchless clear_page() Alexey Dobriyan
2016-08-22 20:54 ` Alexey Dobriyan
2016-08-22 21:06 ` Borislav Petkov
2016-08-23 10:46 ` Alexey Dobriyan
2016-08-23 13:43 ` Brian Gerst [this message]
2016-08-23 16:15 ` Alexey Dobriyan
2016-08-23 16:19 ` [PATCH 1/3] " Alexey Dobriyan
2016-08-23 16:25 ` [PATCH 2/3] x86: support REP MOVSB copy_page() Alexey Dobriyan
2016-08-23 16:28 ` [PATCH 3/3] x86: branchless copy_page() Alexey Dobriyan
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=CAMzpN2gsgEdk5SgsyzP8mwgtta4-nYjEui9Cw3CHkRXoxVVsZg@mail.gmail.com \
--to=brgerst@gmail.com \
--cc=adobriyan@gmail.com \
--cc=linux-kernel@vger.kernel.org \
--cc=x86@kernel.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;
as well as URLs for NNTP newsgroup(s).