[PATCH 14/15] misc: Configurable number of supported IDE interfaces

[PATCH 14/15] misc: Configurable number of supported IDE interfaces

[Matt Mackall]

Configurable number of supported IDE interfaces

This overrides the default limit (which may be set per arch with
CONFIG_IDE_MAX_HWIFS). This is the result of setting interfaces to 1:

   text    data     bss     dec     hex filename
3330172  529036  190556 4049764  3dcb64 vmlinux-baseline
3329352  528928  172124 4030404  3d7fc4 vmlinux

Signed-off-by: Matt Mackall <mpm@selenic.com>

Index: 2.6.14-misc/drivers/ide/setup-pci.c
===================================================================
--- 2.6.14-misc.orig/drivers/ide/setup-pci.c	2005-10-27 17:02:08.000000000 -0700
+++ 2.6.14-misc/drivers/ide/setup-pci.c	2005-11-09 11:27:23.000000000 -0800
@@ -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: 2.6.14-misc/include/linux/ide.h
===================================================================
--- 2.6.14-misc.orig/include/linux/ide.h	2005-11-01 10:54:33.000000000 -0800
+++ 2.6.14-misc/include/linux/ide.h	2005-11-09 11:27:23.000000000 -0800
@@ -309,6 +309,11 @@ static inline void ide_init_hwif_ports(h
 }
 #endif /* IDE_ARCH_OBSOLETE_INIT */
 
+#if defined(CONFIG_IDE_HWIFS) && CONFIG_IDE_HWIFS > 0
+#undef MAX_HWIFS
+#define MAX_HWIFS CONFIG_IDE_HWIFS
+#endif
+
 /* Currently only m68k, apus and m8xx need it */
 #ifndef IDE_ARCH_ACK_INTR
 # define ide_ack_intr(hwif) (1)
Index: 2.6.14-misc/init/Kconfig
===================================================================
--- 2.6.14-misc.orig/init/Kconfig	2005-11-09 11:27:20.000000000 -0800
+++ 2.6.14-misc/init/Kconfig	2005-11-09 11:27:23.000000000 -0800
@@ -457,6 +457,15 @@ config CC_ALIGN_JUMPS
 	  no dummy operations need be executed.
 	  Zero means use compiler's default.
 
+config IDE_HWIFS
+	depends IDE
+	int "Number of IDE hardware interfaces (0 for default)" if EMBEDDED
+	range 0 20
+	default 0
+	help
+	  Select the maximum number of IDE interfaces (0 for default).
+          Saves up to 14k.
+
 endmenu		# General setup
 
 config TINY_SHMEM

Re: [PATCH 14/15] misc: Configurable number of supported IDE interfaces

[Bartlomiej Zolnierkiewicz]

You are duplicating functionality of CONFIG_IDE_MAX_HWIFS,
please find a way to use it for EMBEDDED.

Also please cc: linux-ide on IDE related patches.

On 11/11/05, Matt Mackall <mpm@selenic.com> wrote:
> Configurable number of supported IDE interfaces
>
> This overrides the default limit (which may be set per arch with
> CONFIG_IDE_MAX_HWIFS). This is the result of setting interfaces to 1:
>
>    text    data     bss     dec     hex filename
> 3330172  529036  190556 4049764  3dcb64 vmlinux-baseline
> 3329352  528928  172124 4030404  3d7fc4 vmlinux
>
> Signed-off-by: Matt Mackall <mpm@selenic.com>
>
> Index: 2.6.14-misc/drivers/ide/setup-pci.c
> ===================================================================
> --- 2.6.14-misc.orig/drivers/ide/setup-pci.c    2005-10-27 17:02:08.000000000 -0700
> +++ 2.6.14-misc/drivers/ide/setup-pci.c 2005-11-09 11:27:23.000000000 -0800
> @@ -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: 2.6.14-misc/include/linux/ide.h
> ===================================================================
> --- 2.6.14-misc.orig/include/linux/ide.h        2005-11-01 10:54:33.000000000 -0800
> +++ 2.6.14-misc/include/linux/ide.h     2005-11-09 11:27:23.000000000 -0800
> @@ -309,6 +309,11 @@ static inline void ide_init_hwif_ports(h
>  }
>  #endif /* IDE_ARCH_OBSOLETE_INIT */
>
> +#if defined(CONFIG_IDE_HWIFS) && CONFIG_IDE_HWIFS > 0
> +#undef MAX_HWIFS
> +#define MAX_HWIFS CONFIG_IDE_HWIFS
> +#endif
> +
>  /* Currently only m68k, apus and m8xx need it */
>  #ifndef IDE_ARCH_ACK_INTR
>  # define ide_ack_intr(hwif) (1)
> Index: 2.6.14-misc/init/Kconfig
> ===================================================================
> --- 2.6.14-misc.orig/init/Kconfig       2005-11-09 11:27:20.000000000 -0800
> +++ 2.6.14-misc/init/Kconfig    2005-11-09 11:27:23.000000000 -0800
> @@ -457,6 +457,15 @@ config CC_ALIGN_JUMPS
>           no dummy operations need be executed.
>           Zero means use compiler's default.
>
> +config IDE_HWIFS
> +       depends IDE
> +       int "Number of IDE hardware interfaces (0 for default)" if EMBEDDED
> +       range 0 20
> +       default 0
> +       help
> +         Select the maximum number of IDE interfaces (0 for default).
> +          Saves up to 14k.
> +
>  endmenu                # General setup
>
>  config TINY_SHMEM

Re: [PATCH 14/15] misc: Configurable number of supported IDE interfaces

[Nick Warne]

Bartlomiej Zolnierkiewicz wrote:

> You are duplicating functionality of CONFIG_IDE_MAX_HWIFS,
> please find a way to use it for EMBEDDED.
> 
> Also please cc: linux-ide on IDE related patches.
> 
> On 11/11/05, Matt Mackall <mpm@selenic.com> wrote:
>> Configurable number of supported IDE interfaces
>>
>> This overrides the default limit (which may be set per arch with
>> CONFIG_IDE_MAX_HWIFS). This is the result of setting interfaces to 1:

This is very similar to my unaccepted patch a few months ago:

http://lkml.org/lkml/2005/6/25/69

Nick
-- 
http://sourceforge.net/projects/quake2plus

"Person who say it cannot be done should not interrupt person doing it."
-Chinese Proverb

Re: [PATCH 14/15] misc: Configurable number of supported IDE interfaces

[Matt Mackall]

On Fri, Nov 11, 2005 at 03:33:13PM +0000, Nick Warne wrote:
> Bartlomiej Zolnierkiewicz wrote:
> 
> > You are duplicating functionality of CONFIG_IDE_MAX_HWIFS,
> > please find a way to use it for EMBEDDED.
> > 
> > Also please cc: linux-ide on IDE related patches.
> > 
> > On 11/11/05, Matt Mackall <mpm@selenic.com> wrote:
> >> Configurable number of supported IDE interfaces
> >>
> >> This overrides the default limit (which may be set per arch with
> >> CONFIG_IDE_MAX_HWIFS). This is the result of setting interfaces to 1:
> 
> This is very similar to my unaccepted patch a few months ago:
> 
> http://lkml.org/lkml/2005/6/25/69

And it's practically identical to the one that's been in my -tiny tree
for two years.

I agree with Alan's complaint that this should in fact be done with
dynamic allocation. However, when I last set out to do that, it
appeared to be an _extremely_ invasive change so I quickly abandoned
it.

Meanwhile, this is currently the largest static allocation in a
typical kernel. Something has to be done.

-- 
Mathematics is the supreme nostalgia of our time.

Re: [PATCH 14/15] misc: Configurable number of supported IDE interfaces

[Matt Mackall]

On Fri, Nov 11, 2005 at 11:14:08AM +0100, Bartlomiej Zolnierkiewicz wrote:
[top-posting adjusted]
> > This overrides the default limit (which may be set per arch with
> > CONFIG_IDE_MAX_HWIFS). This is the result of setting interfaces to 1:
>
> You are duplicating functionality of CONFIG_IDE_MAX_HWIFS,
> please find a way to use it for EMBEDDED.

It's intentional. The current CONFIG_IDE_MAX_HWIFS is a hidden
variable that sets a per architecture maximum. To the best of my
knowledge, there's no way to do, say:

   default 4 if ARCH_FOO
   default 1 if ARCH_BAR

..so I'm stuck with using two config symbols anyway.

I've thought about it, this is the best I could come up with. If you
can come up with something cleaner, I'm all ears.

-- 
Mathematics is the supreme nostalgia of our time.

Re: [PATCH 14/15] misc: Configurable number of supported IDE interfaces

[Roman Zippel]

Hi,

On Fri, 11 Nov 2005, Matt Mackall wrote:

> It's intentional. The current CONFIG_IDE_MAX_HWIFS is a hidden
> variable that sets a per architecture maximum. To the best of my
> knowledge, there's no way to do, say:
> 
>    default 4 if ARCH_FOO
>    default 1 if ARCH_BAR
> 
> ..so I'm stuck with using two config symbols anyway.

Where is the problem? This should work fine.
With the latest kernel you can even use a dynamic range:

config IDE_HWIFS
	int "..."
	range 1 IDE_MAX_HWIFS

bye, Roman

Re: [PATCH 14/15] misc: Configurable number of supported IDE interfaces

[Matt Mackall]

On Fri, Nov 11, 2005 at 06:34:27PM +0100, Roman Zippel wrote:
> Hi,
> 
> On Fri, 11 Nov 2005, Matt Mackall wrote:
> 
> > It's intentional. The current CONFIG_IDE_MAX_HWIFS is a hidden
> > variable that sets a per architecture maximum. To the best of my
> > knowledge, there's no way to do, say:
> > 
> >    default 4 if ARCH_FOO
> >    default 1 if ARCH_BAR
> > 
> > ..so I'm stuck with using two config symbols anyway.
> 
> Where is the problem? This should work fine.

Does it? Didn't work when last I checked (which was a while ago).

> With the latest kernel you can even use a dynamic range:
> 
> config IDE_HWIFS
> 	int "..."
> 	range 1 IDE_MAX_HWIFS

But this suggests a good reason to hold on to both variables.

-- 
Mathematics is the supreme nostalgia of our time.

Re: [PATCH 14/15] misc: Configurable number of supported IDE interfaces

[Matt Mackall]

On Fri, Nov 11, 2005 at 09:37:37AM -0800, Matt Mackall wrote:
> > With the latest kernel you can even use a dynamic range:
> > 
> > config IDE_HWIFS
> > 	int "..."
> > 	range 1 IDE_MAX_HWIFS
> 
> But this suggests a good reason to hold on to both variables.

..except that not all arches define it. In fact, only sh and Alpha do.

-- 
Mathematics is the supreme nostalgia of our time.

Re: [PATCH 14/15] misc: Configurable number of supported IDE interfaces

[Roman Zippel]

Hi,

On Fri, 11 Nov 2005, Matt Mackall wrote:

> > > It's intentional. The current CONFIG_IDE_MAX_HWIFS is a hidden
> > > variable that sets a per architecture maximum. To the best of my
> > > knowledge, there's no way to do, say:
> > > 
> > >    default 4 if ARCH_FOO
> > >    default 1 if ARCH_BAR
> > > 
> > > ..so I'm stuck with using two config symbols anyway.
> > 
> > Where is the problem? This should work fine.
> 
> Does it? Didn't work when last I checked (which was a while ago).

I don't know what you tried, for me it does.

> > With the latest kernel you can even use a dynamic range:
> > 
> > config IDE_HWIFS
> > 	int "..."
> > 	range 1 IDE_MAX_HWIFS
> 
> But this suggests a good reason to hold on to both variables.

You _can_ use it, I didn't say you have to use it.

bye, Roman