Make IDE_HWIFS configurable if EMBEDDED
This lets us lop as much as 16k off an x86 build. It's a little ugly,
but it's dead simple. Note the fix for HWIFS < 2.
Sizing interfaces dynamically unfortunately turns out to be pretty
major surgery.
add/remove: 0/1 grow/shrink: 0/11 up/down: 0/-16182 (-16182)
function old new delta
ide_hwifs 16920 1692 -15228
init_irq 1113 750 -363
ideprobe_init 283 138 -145
ide_pci_setup_ports 1329 1193 -136
save_match 85 - -85
ide_register_hw_with_fixup 367 287 -80
ide_setup 1364 1308 -56
is_chipset_set 40 4 -36
create_proc_ide_interfaces 225 205 -20
init_ide_data 84 67 -17
ide_probe_for_cmd640x 1198 1183 -15
ide_unregister 1452 1451 -1
Signed-off-by: Matt Mackall <[email protected]>
Index: 2.6/drivers/ide/Kconfig
===================================================================
--- 2.6.orig/drivers/ide/Kconfig 2006-04-20 17:01:05.000000000 -0500
+++ 2.6/drivers/ide/Kconfig 2006-05-11 15:10:58.000000000 -0500
@@ -54,7 +54,7 @@ if IDE
config IDE_MAX_HWIFS
int "Max IDE interfaces"
- depends on ALPHA || SUPERH || IA64
+ depends on ALPHA || SUPERH || IA64 || EMBEDDED
default 4
help
This is the maximum number of IDE hardware interfaces that will
Index: 2.6/drivers/ide/setup-pci.c
===================================================================
--- 2.6.orig/drivers/ide/setup-pci.c 2006-05-11 15:07:32.000000000 -0500
+++ 2.6/drivers/ide/setup-pci.c 2006-05-11 15:13:51.000000000 -0500
@@ -102,7 +102,7 @@ static ide_hwif_t *ide_match_hwif(unsign
return hwif; /* pick an unused entry */
}
}
- for (h = 0; h < 2; ++h) {
+ for (h = 0; h < 2 && h < MAX_HWIFS; ++h) {
hwif = ide_hwifs + h;
if (hwif->chipset == ide_unknown)
return hwif; /* pick an unused entry */
Index: 2.6/include/linux/ide.h
===================================================================
--- 2.6.orig/include/linux/ide.h 2006-05-11 15:07:32.000000000 -0500
+++ 2.6/include/linux/ide.h 2006-05-12 14:01:53.000000000 -0500
@@ -252,7 +252,8 @@ static inline void ide_std_init_ports(hw
#include <asm/ide.h>
-#ifndef MAX_HWIFS
+#if !defined(MAX_HWIFS) || defined(CONFIG_EMBEDDED)
+#undef MAX_HWIFS
#define MAX_HWIFS CONFIG_IDE_MAX_HWIFS
#endif
--
Mathematics is the supreme nostalgia of our time.
On Fri, May 12, 2006 at 05:29:52PM -0500, Matt Mackall wrote:
>...
> --- 2.6.orig/include/linux/ide.h 2006-05-11 15:07:32.000000000 -0500
> +++ 2.6/include/linux/ide.h 2006-05-12 14:01:53.000000000 -0500
> @@ -252,7 +252,8 @@ static inline void ide_std_init_ports(hw
>
> #include <asm/ide.h>
>
> -#ifndef MAX_HWIFS
> +#if !defined(MAX_HWIFS) || defined(CONFIG_EMBEDDED)
> +#undef MAX_HWIFS
> #define MAX_HWIFS CONFIG_IDE_MAX_HWIFS
> #endif
Why do you need this?
cu
Adrian
--
"Is there not promise of rain?" Ling Tan asked suddenly out
of the darkness. There had been need of rain for many days.
"Only a promise," Lao Er said.
Pearl S. Buck - Dragon Seed
On Tue, May 16, 2006 at 06:02:50PM +0200, Adrian Bunk wrote:
> On Fri, May 12, 2006 at 05:29:52PM -0500, Matt Mackall wrote:
> >...
> > --- 2.6.orig/include/linux/ide.h 2006-05-11 15:07:32.000000000 -0500
> > +++ 2.6/include/linux/ide.h 2006-05-12 14:01:53.000000000 -0500
> > @@ -252,7 +252,8 @@ static inline void ide_std_init_ports(hw
> >
> > #include <asm/ide.h>
> >
> > -#ifndef MAX_HWIFS
> > +#if !defined(MAX_HWIFS) || defined(CONFIG_EMBEDDED)
> > +#undef MAX_HWIFS
> > #define MAX_HWIFS CONFIG_IDE_MAX_HWIFS
> > #endif
>
> Why do you need this?
Doesn't work without it?
Most platforms define MAX_HWIFS.
--
Mathematics is the supreme nostalgia of our time.
On Tue, May 16, 2006 at 11:29:34AM -0500, Matt Mackall wrote:
> On Tue, May 16, 2006 at 06:02:50PM +0200, Adrian Bunk wrote:
> > On Fri, May 12, 2006 at 05:29:52PM -0500, Matt Mackall wrote:
> > >...
> > > --- 2.6.orig/include/linux/ide.h 2006-05-11 15:07:32.000000000 -0500
> > > +++ 2.6/include/linux/ide.h 2006-05-12 14:01:53.000000000 -0500
> > > @@ -252,7 +252,8 @@ static inline void ide_std_init_ports(hw
> > >
> > > #include <asm/ide.h>
> > >
> > > -#ifndef MAX_HWIFS
> > > +#if !defined(MAX_HWIFS) || defined(CONFIG_EMBEDDED)
> > > +#undef MAX_HWIFS
> > > #define MAX_HWIFS CONFIG_IDE_MAX_HWIFS
> > > #endif
> >
> > Why do you need this?
>
> Doesn't work without it?
>
> Most platforms define MAX_HWIFS.
OK, now I got it.
Setting this value is sometimes done in heder files and sometimes
done in the Kconfig file.
That is extremely ugly.
Bart, would you accept a patch to set in in the Kconfig file on all
architectures?
cu
Adrian
--
"Is there not promise of rain?" Ling Tan asked suddenly out
of the darkness. There had been need of rain for many days.
"Only a promise," Lao Er said.
Pearl S. Buck - Dragon Seed
I submitted a patch ages ago similar (but differnet ;-) ) to this:
http://lkml.org/lkml/2005/6/25/69
I couldn't/don't see why I get six ide probes when I know I have only
2 IDE interfaces - I thought it would be better to stick to default
_unless_ the kernel builder knew otherwise and specified the amount.
Alan Cox didn't like it as it introduces more config unnecessary config options.
But it would be nice though if this could be included with that enhancement too.
Nick
On 16/05/06, Adrian Bunk <[email protected]> wrote:
> On Tue, May 16, 2006 at 11:29:34AM -0500, Matt Mackall wrote:
> > On Tue, May 16, 2006 at 06:02:50PM +0200, Adrian Bunk wrote:
> > > On Fri, May 12, 2006 at 05:29:52PM -0500, Matt Mackall wrote:
> > > >...
> > > > --- 2.6.orig/include/linux/ide.h 2006-05-11 15:07:32.000000000 -0500
> > > > +++ 2.6/include/linux/ide.h 2006-05-12 14:01:53.000000000 -0500
> > > > @@ -252,7 +252,8 @@ static inline void ide_std_init_ports(hw
> > > >
> > > > #include <asm/ide.h>
> > > >
> > > > -#ifndef MAX_HWIFS
> > > > +#if !defined(MAX_HWIFS) || defined(CONFIG_EMBEDDED)
> > > > +#undef MAX_HWIFS
> > > > #define MAX_HWIFS CONFIG_IDE_MAX_HWIFS
> > > > #endif
> > >
> > > Why do you need this?
> >
> > Doesn't work without it?
> >
> > Most platforms define MAX_HWIFS.
>
> OK, now I got it.
>
> Setting this value is sometimes done in heder files and sometimes
> done in the Kconfig file.
>
> That is extremely ugly.
>
> Bart, would you accept a patch to set in in the Kconfig file on all
> architectures?
>
> cu
> Adrian
>
> --
>
> "Is there not promise of rain?" Ling Tan asked suddenly out
> of the darkness. There had been need of rain for many days.
> "Only a promise," Lao Er said.
> Pearl S. Buck - Dragon Seed
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
>