arch/x86/mm/ioremap unification grew by 10x

arch/x86/mm/ioremap unification grew by 10x

[Matt Mackall]

In 2.6.24 defconfig, my build stats show ioremap_32.o was 1.8k. In
2.6.25-rc1, the unified ioremap.o is 20.8k.

-- 
Mathematics is the supreme nostalgia of our time.

Re: arch/x86/mm/ioremap unification grew by 10x

[Sam Ravnborg]

On Fri, Feb 15, 2008 at 02:25:54PM -0600, Matt Mackall wrote:
> In 2.6.24 defconfig, my build stats show ioremap_32.o was 1.8k. In
> 2.6.25-rc1, the unified ioremap.o is 20.8k.

Just an observation - 17 commits touches said file after
the unification (at least in latest -linus).

	Sam

Re: arch/x86/mm/ioremap unification grew by 10x

[Matt Mackall]

On Fri, 2008-02-15 at 21:32 +0100, Sam Ravnborg wrote:
> On Fri, Feb 15, 2008 at 02:25:54PM -0600, Matt Mackall wrote:
> > In 2.6.24 defconfig, my build stats show ioremap_32.o was 1.8k. In
> > 2.6.25-rc1, the unified ioremap.o is 20.8k.
> 
> Just an observation - 17 commits touches said file after
> the unification (at least in latest -linus).

Correction: those numbers should be halved. So we're going from .9k to
10.4k.

-- 
Mathematics is the supreme nostalgia of our time.

Re: arch/x86/mm/ioremap unification grew by 10x

[Matt Mackall]

On Fri, 2008-02-15 at 15:21 -0600, Matt Mackall wrote:
> On Fri, 2008-02-15 at 21:32 +0100, Sam Ravnborg wrote:
> > On Fri, Feb 15, 2008 at 02:25:54PM -0600, Matt Mackall wrote:
> > > In 2.6.24 defconfig, my build stats show ioremap_32.o was 1.8k. In
> > > 2.6.25-rc1, the unified ioremap.o is 20.8k.
> > 
> > Just an observation - 17 commits touches said file after
> > the unification (at least in latest -linus).
> 
> Correction: those numbers should be halved. So we're going from .9k to
> 10.4k.

And here's most of the cause:

000002b8 00000124 T early_ioremap
00001000 00001000 t bm_pte
00002000 00000004 T early_ioremap_debug

static __initdata pte_t bm_pte[PAGE_SIZE/sizeof(pte_t)]
				__attribute__((aligned(PAGE_SIZE)));

Double ouch. First, this isn't in BSS. Second, even though it's
initdata, the alignment slop won't get recovered.

Don't we have a special section for page-aligned crap so it doesn't
waste most of two pages?

-- 
Mathematics is the supreme nostalgia of our time.

Re: arch/x86/mm/ioremap unification grew by 10x

[Ian Campbell]

[-- Attachment #1: Type: text/plain, Size: 3092 bytes --]


On Fri, 2008-02-15 at 17:17 -0600, Matt Mackall wrote:
> On Fri, 2008-02-15 at 15:21 -0600, Matt Mackall wrote:
> > On Fri, 2008-02-15 at 21:32 +0100, Sam Ravnborg wrote:
> > > On Fri, Feb 15, 2008 at 02:25:54PM -0600, Matt Mackall wrote:
> > > > In 2.6.24 defconfig, my build stats show ioremap_32.o was 1.8k. In
> > > > 2.6.25-rc1, the unified ioremap.o is 20.8k.
> > > 
> > > Just an observation - 17 commits touches said file after
> > > the unification (at least in latest -linus).
> > 
> > Correction: those numbers should be halved. So we're going from .9k to
> > 10.4k.
> 
> And here's most of the cause:
> 
> 000002b8 00000124 T early_ioremap
> 00001000 00001000 t bm_pte
> 00002000 00000004 T early_ioremap_debug
> 
> static __initdata pte_t bm_pte[PAGE_SIZE/sizeof(pte_t)]
> 				__attribute__((aligned(PAGE_SIZE)));
> 
> Double ouch. First, this isn't in BSS. Second, even though it's
> initdata, the alignment slop won't get recovered.

551889a6e2a24a9c06fd453ea03b57b7746ffdc0 (the reason you CC'd me) only
changed this from unsigned long[1024] to pte_t[PAGE_SIZE/sizeof(pte_t)]

-static __initdata unsigned long bm_pte[1024]
+static __initdata pte_t bm_pte[PAGE_SIZE/sizeof(pte_t)]
                                __attribute__((aligned(PAGE_SIZE)));

I don't think that would have changed anything in this respect. It seems
to have been that way since 0947b2f31ca1ea1211d3cde2dbd8fcec579ef395
when it was added (Huang Ying CC'd).

> Don't we have a special section for page-aligned crap so it doesn't
> waste most of two pages?

We have .bss.page_aligned and it seems appropriate to use it.

Ian.

--- 
From ac954372c1ed7830148bdea1e4283856c05c354f Mon Sep 17 00:00:00 2001
From: Ian Campbell <ijc@hellion.org.uk>
Date: Sun, 17 Feb 2008 19:08:34 +0000
Subject: [PATCH] x86: bm_pte can be placed in .bss.page_aligned.

    text	   data	    bss	    dec	    hex	filename
    -   3388	   8236	      4	  11628	   2d6c	../build-32/arch/x86/mm/ioremap.o
    +   3388	     48	   4100	   7536	   1d70	../build-32/arch/x86/mm/ioremap.o

Signed-off-by: Ian Campbell <ijc@hellion.org.uk>
Cc: Matt Mackall <mpm@selenic.com>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Huang Ying <ying.huang@intel.com>
---
 arch/x86/mm/ioremap.c |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/arch/x86/mm/ioremap.c b/arch/x86/mm/ioremap.c
index 69f4981..c62ecbd 100644
--- a/arch/x86/mm/ioremap.c
+++ b/arch/x86/mm/ioremap.c
@@ -262,8 +262,8 @@ static int __init early_ioremap_debug_setup(char *str)
 early_param("early_ioremap_debug", early_ioremap_debug_setup);
 
 static __initdata int after_paging_init;
-static __initdata pte_t bm_pte[PAGE_SIZE/sizeof(pte_t)]
-				__attribute__((aligned(PAGE_SIZE)));
+static pte_t bm_pte[PAGE_SIZE/sizeof(pte_t)]
+		__attribute__((__section__(".bss.page_aligned")));
 
 static inline pmd_t * __init early_ioremap_pmd(unsigned long addr)
 {
-- 
1.5.4


-- 
Ian Campbell

In which level of metalanguage are you now speaking?

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: arch/x86/mm/ioremap unification grew by 10x

[Sam Ravnborg]

On Sun, Feb 17, 2008 at 07:09:42PM +0000, Ian Campbell wrote:
> 
> On Fri, 2008-02-15 at 17:17 -0600, Matt Mackall wrote:
> > On Fri, 2008-02-15 at 15:21 -0600, Matt Mackall wrote:
> > > On Fri, 2008-02-15 at 21:32 +0100, Sam Ravnborg wrote:
> > > > On Fri, Feb 15, 2008 at 02:25:54PM -0600, Matt Mackall wrote:
> > > > > In 2.6.24 defconfig, my build stats show ioremap_32.o was 1.8k. In
> > > > > 2.6.25-rc1, the unified ioremap.o is 20.8k.
> > > > 
> > > > Just an observation - 17 commits touches said file after
> > > > the unification (at least in latest -linus).
> > > 
> > > Correction: those numbers should be halved. So we're going from .9k to
> > > 10.4k.
> > 
> > And here's most of the cause:
> > 
> > 000002b8 00000124 T early_ioremap
> > 00001000 00001000 t bm_pte
> > 00002000 00000004 T early_ioremap_debug
> > 
> > static __initdata pte_t bm_pte[PAGE_SIZE/sizeof(pte_t)]
> > 				__attribute__((aligned(PAGE_SIZE)));
> > 
> > Double ouch. First, this isn't in BSS. Second, even though it's
> > initdata, the alignment slop won't get recovered.
> 
> 551889a6e2a24a9c06fd453ea03b57b7746ffdc0 (the reason you CC'd me) only
> changed this from unsigned long[1024] to pte_t[PAGE_SIZE/sizeof(pte_t)]
> 
> -static __initdata unsigned long bm_pte[1024]
> +static __initdata pte_t bm_pte[PAGE_SIZE/sizeof(pte_t)]
>                                 __attribute__((aligned(PAGE_SIZE)));
> 
> I don't think that would have changed anything in this respect. It seems
> to have been that way since 0947b2f31ca1ea1211d3cde2dbd8fcec579ef395
> when it was added (Huang Ying CC'd).
> 
> > Don't we have a special section for page-aligned crap so it doesn't
> > waste most of two pages?
> 
> We have .bss.page_aligned and it seems appropriate to use it.
> 
> Ian.
> 
> --- 
> From ac954372c1ed7830148bdea1e4283856c05c354f Mon Sep 17 00:00:00 2001
> From: Ian Campbell <ijc@hellion.org.uk>
> Date: Sun, 17 Feb 2008 19:08:34 +0000
> Subject: [PATCH] x86: bm_pte can be placed in .bss.page_aligned.
> 
>     text	   data	    bss	    dec	    hex	filename
>     -   3388	   8236	      4	  11628	   2d6c	../build-32/arch/x86/mm/ioremap.o
>     +   3388	     48	   4100	   7536	   1d70	../build-32/arch/x86/mm/ioremap.o
> 
> Signed-off-by: Ian Campbell <ijc@hellion.org.uk>
> Cc: Matt Mackall <mpm@selenic.com>
> Cc: Sam Ravnborg <sam@ravnborg.org>
> Cc: Ingo Molnar <mingo@elte.hu>
> Cc: Huang Ying <ying.huang@intel.com>
> ---
>  arch/x86/mm/ioremap.c |    4 ++--
>  1 files changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/arch/x86/mm/ioremap.c b/arch/x86/mm/ioremap.c
> index 69f4981..c62ecbd 100644
> --- a/arch/x86/mm/ioremap.c
> +++ b/arch/x86/mm/ioremap.c
> @@ -262,8 +262,8 @@ static int __init early_ioremap_debug_setup(char *str)
>  early_param("early_ioremap_debug", early_ioremap_debug_setup);
>  
>  static __initdata int after_paging_init;
> -static __initdata pte_t bm_pte[PAGE_SIZE/sizeof(pte_t)]
> -				__attribute__((aligned(PAGE_SIZE)));
> +static pte_t bm_pte[PAGE_SIZE/sizeof(pte_t)]

Replace
> +		__attribute__((__section__(".bss.page_aligned")));

with the shorter form:
> +		__section(.bss.page_aligned)));


	Sam

Re: arch/x86/mm/ioremap unification grew by 10x

[Ingo Molnar]

* Sam Ravnborg <sam@ravnborg.org> wrote:

> >  static __initdata int after_paging_init;
> > -static __initdata pte_t bm_pte[PAGE_SIZE/sizeof(pte_t)]
> > -				__attribute__((aligned(PAGE_SIZE)));
> > +static pte_t bm_pte[PAGE_SIZE/sizeof(pte_t)]
> 
> Replace
> > +		__attribute__((__section__(".bss.page_aligned")));
> 
> with the shorter form:
> > +		__section(.bss.page_aligned)));

i've picked up Ian's patch into x86.git and have done the above change. 
Thanks,

	Ingo

Re: arch/x86/mm/ioremap unification grew by 10x

[Huang, Ying]

On Sun, 2008-02-17 at 19:09 +0000, Ian Campbell wrote:
> On Fri, 2008-02-15 at 17:17 -0600, Matt Mackall wrote:
> > On Fri, 2008-02-15 at 15:21 -0600, Matt Mackall wrote:
> > > On Fri, 2008-02-15 at 21:32 +0100, Sam Ravnborg wrote:
> > > > On Fri, Feb 15, 2008 at 02:25:54PM -0600, Matt Mackall wrote:
> > > > > In 2.6.24 defconfig, my build stats show ioremap_32.o was 1.8k. In
> > > > > 2.6.25-rc1, the unified ioremap.o is 20.8k.
> > > > 
> > > > Just an observation - 17 commits touches said file after
> > > > the unification (at least in latest -linus).
> > > 
> > > Correction: those numbers should be halved. So we're going from .9k to
> > > 10.4k.
> > 
> > And here's most of the cause:
> > 
> > 000002b8 00000124 T early_ioremap
> > 00001000 00001000 t bm_pte
> > 00002000 00000004 T early_ioremap_debug
> > 
> > static __initdata pte_t bm_pte[PAGE_SIZE/sizeof(pte_t)]
> > 				__attribute__((aligned(PAGE_SIZE)));
> > 
> > Double ouch. First, this isn't in BSS. Second, even though it's
> > initdata, the alignment slop won't get recovered.
> 
> 551889a6e2a24a9c06fd453ea03b57b7746ffdc0 (the reason you CC'd me) only
> changed this from unsigned long[1024] to pte_t[PAGE_SIZE/sizeof(pte_t)]
> 
> -static __initdata unsigned long bm_pte[1024]
> +static __initdata pte_t bm_pte[PAGE_SIZE/sizeof(pte_t)]
>                                 __attribute__((aligned(PAGE_SIZE)));
> 
> I don't think that would have changed anything in this respect. It seems
> to have been that way since 0947b2f31ca1ea1211d3cde2dbd8fcec579ef395
> when it was added (Huang Ying CC'd).
> 
> > Don't we have a special section for page-aligned crap so it doesn't
> > waste most of two pages?
> 
> We have .bss.page_aligned and it seems appropriate to use it.

But .bss.page_aligned is not an init section, this will waste some
memory after boot. Do you think that it is more appropriate to add
another section like .init.bss and .init.bss.page_aligned.

Best Regards,
Huang Ying

Re: arch/x86/mm/ioremap unification grew by 10x

[Ingo Molnar]

* Huang, Ying <ying.huang@intel.com> wrote:

> > > And here's most of the cause:
> > > 
> > > 000002b8 00000124 T early_ioremap
> > > 00001000 00001000 t bm_pte
> > > 00002000 00000004 T early_ioremap_debug
> > > 
> > > static __initdata pte_t bm_pte[PAGE_SIZE/sizeof(pte_t)]
> > > 				__attribute__((aligned(PAGE_SIZE)));
> > > 
> > > Double ouch. First, this isn't in BSS. Second, even though it's
> > > initdata, the alignment slop won't get recovered.
> > 
> > 551889a6e2a24a9c06fd453ea03b57b7746ffdc0 (the reason you CC'd me) only
> > changed this from unsigned long[1024] to pte_t[PAGE_SIZE/sizeof(pte_t)]
> > 
> > -static __initdata unsigned long bm_pte[1024]
> > +static __initdata pte_t bm_pte[PAGE_SIZE/sizeof(pte_t)]
> >                                 __attribute__((aligned(PAGE_SIZE)));
> > 
> > I don't think that would have changed anything in this respect. It seems
> > to have been that way since 0947b2f31ca1ea1211d3cde2dbd8fcec579ef395
> > when it was added (Huang Ying CC'd).
> > 
> > > Don't we have a special section for page-aligned crap so it doesn't
> > > waste most of two pages?
> > 
> > We have .bss.page_aligned and it seems appropriate to use it.
> 
> But .bss.page_aligned is not an init section, this will waste some 
> memory after boot. Do you think that it is more appropriate to add 
> another section like .init.bss and .init.bss.page_aligned.

agreed, and i'd suggest we do so before this information gets forgotten 
and someone has to re-discover the fix the hard way.

	Ingo