PCI configuration changes

PCI configuration changes

[Andrzej Krzysztofowicz]

Hi,
   Looking at the latest drivers/net/Config.in changes I noticed, that
two (modified earlier) lines are buggy:

      dep_tristate '    3c523 "EtherLink/MC" support' CONFIG_ELMC $CONFIG_MCA
      dep_tristate '    3c527 "EtherLink/MC 32" support (EXPERIMENTAL)' CONFIG_ELMC_II $CONFIG_MCA $CONFIG_EXPERIMENTAL

Note, that as CONFIG_MCA is defined only for i386 the dependencies on 
$CONFIG_MCA are no-op for other architectures (in Configure/Menuconfig).
Either CONFIG_MCA should be defined for all architectures or there should be
if ... fi around these lines.

BTW, is there any reason for not replacing 

   bool '  Other ISA cards' CONFIG_NET_ISA

by

  dep_bool '  Other ISA cards' CONFIG_NET_ISA $CONFIG_ISA

to eliminate more drivers from non-ISA arch configs ?

Andrzej

-- 
=======================================================================
  Andrzej M. Krzysztofowicz               ankry@mif.pg.gda.pl
  phone (48)(58) 347 14 61
Faculty of Applied Phys. & Math.,   Technical University of Gdansk
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: PCI configuration changes

[Peter Samuelson]

[Andrzej Krzysztofowicz]
> Note, that as CONFIG_MCA is defined only for i386 the dependencies on
> $CONFIG_MCA are no-op for other architectures (in
> Configure/Menuconfig).  Either CONFIG_MCA should be defined for all
> architectures or there should be if ... fi around these lines.

The former, I think.  Less confusing in the long run.

> BTW, is there any reason for not replacing 
>    bool '  Other ISA cards' CONFIG_NET_ISA
> by
>   dep_bool '  Other ISA cards' CONFIG_NET_ISA $CONFIG_ISA
> to eliminate more drivers from non-ISA arch configs ?

Looks good to me.  Anything to remove clutter from config menus....

Peter


diff -urk.orig 2.4.0test11pre4/arch/alpha/config.in.orig 2.4.0test11pre4/arch/alpha/config.in
--- 2.4.0test11pre4/arch/alpha/config.in.orig	Mon Nov 13 01:44:55 2000
+++ 2.4.0test11pre4/arch/alpha/config.in	Thu Nov 16 09:11:23 2000
@@ -69,6 +69,7 @@
 define_bool CONFIG_ISA y
 define_bool CONFIG_EISA y
 define_bool CONFIG_SBUS n
+define_bool CONFIG_MCA n
 
 if [ "$CONFIG_ALPHA_JENSEN" = "y" ]
 then
diff -urk.orig 2.4.0test11pre4/arch/arm/config.in.orig 2.4.0test11pre4/arch/arm/config.in
--- 2.4.0test11pre4/arch/arm/config.in.orig	Mon Nov 13 01:44:02 2000
+++ 2.4.0test11pre4/arch/arm/config.in	Thu Nov 16 09:11:48 2000
@@ -7,6 +7,7 @@
 define_bool CONFIG_ARM y
 define_bool CONFIG_EISA n
 define_bool CONFIG_SBUS n
+define_bool CONFIG_MCA n
 define_bool CONFIG_UID16 y
 
 
diff -urk.orig 2.4.0test11pre4/arch/ia64/config.in.orig 2.4.0test11pre4/arch/ia64/config.in
--- 2.4.0test11pre4/arch/ia64/config.in.orig	Mon Nov 13 01:44:02 2000
+++ 2.4.0test11pre4/arch/ia64/config.in	Thu Nov 16 09:17:26 2000
@@ -22,6 +22,7 @@
 
 define_bool CONFIG_ISA n
 define_bool CONFIG_EISA n
+define_bool CONFIG_MCA n
 define_bool CONFIG_SBUS n
 
 choice 'IA-64 system type'					\
diff -urk.orig 2.4.0test11pre4/arch/m68k/config.in.orig 2.4.0test11pre4/arch/m68k/config.in
--- 2.4.0test11pre4/arch/m68k/config.in.orig	Mon Nov 13 01:44:02 2000
+++ 2.4.0test11pre4/arch/m68k/config.in	Thu Nov 16 09:17:10 2000
@@ -26,6 +26,7 @@
 
 define_bool CONFIG_ISA n
 define_bool CONFIG_EISA n
+define_bool CONFIG_MCA n
 define_bool CONFIG_PCMCIA n
 
 bool 'Amiga support' CONFIG_AMIGA
diff -urk.orig 2.4.0test11pre4/arch/mips/config.in.orig 2.4.0test11pre4/arch/mips/config.in
--- 2.4.0test11pre4/arch/mips/config.in.orig	Mon Nov 13 01:44:02 2000
+++ 2.4.0test11pre4/arch/mips/config.in	Thu Nov 16 09:16:45 2000
@@ -39,6 +39,7 @@
 unset CONFIG_VIDEO_G364
 unset CONFIG_PC_KEYB
 
+define_bool CONFIG_MCA n
 define_bool CONFIG_SBUS n
 
 if [ "$CONFIG_ALGOR_P4032" = "y" ]; then
diff -urk.orig 2.4.0test11pre4/arch/mips64/config.in.orig 2.4.0test11pre4/arch/mips64/config.in
--- 2.4.0test11pre4/arch/mips64/config.in.orig	Mon Nov 13 01:44:02 2000
+++ 2.4.0test11pre4/arch/mips64/config.in	Thu Nov 16 09:16:33 2000
@@ -66,6 +66,7 @@
    define_bool CONFIG_PCI n
 fi
 
+define_bool CONFIG_MCA n
 define_bool CONFIG_SBUS n
 
 mainmenu_option next_comment
diff -urk.orig 2.4.0test11pre4/arch/ppc/config.in.orig 2.4.0test11pre4/arch/ppc/config.in
--- 2.4.0test11pre4/arch/ppc/config.in.orig	Mon Nov 13 01:44:02 2000
+++ 2.4.0test11pre4/arch/ppc/config.in	Thu Nov 16 09:15:21 2000
@@ -101,6 +101,9 @@
 define_bool CONFIG_EISA n
 define_bool CONFIG_SBUS n
 
+# Yes MCA RS/6000s exist but Linux-PPC does not currently support any
+define_bool CONFIG_MCA n
+
 if [ "$CONFIG_APUS" = "y" -o "$CONFIG_4xx" = "y" -o \
      "$CONFIG_8260" = "y" ]; then
    define_bool CONFIG_PCI n
diff -urk.orig 2.4.0test11pre4/arch/s390/config.in.orig 2.4.0test11pre4/arch/s390/config.in
--- 2.4.0test11pre4/arch/s390/config.in.orig	Mon Nov 13 01:44:03 2000
+++ 2.4.0test11pre4/arch/s390/config.in	Thu Nov 16 09:13:39 2000
@@ -5,6 +5,7 @@
 
 define_bool CONFIG_ISA n
 define_bool CONFIG_EISA n
+define_bool CONFIG_MCA n
 define_bool CONFIG_UID16 y
 
 mainmenu_name "Linux Kernel Configuration"
diff -urk.orig 2.4.0test11pre4/arch/sh/config.in.orig 2.4.0test11pre4/arch/sh/config.in
--- 2.4.0test11pre4/arch/sh/config.in.orig	Mon Nov 13 01:44:03 2000
+++ 2.4.0test11pre4/arch/sh/config.in	Thu Nov 16 09:13:26 2000
@@ -67,6 +67,7 @@
 
 define_bool CONFIG_ISA n
 define_bool CONFIG_EISA n
+define_bool CONFIG_MCA n
 define_bool CONFIG_SBUS n
 
 bool 'Networking support' CONFIG_NET
diff -urk.orig 2.4.0test11pre4/arch/sparc/config.in.orig 2.4.0test11pre4/arch/sparc/config.in
--- 2.4.0test11pre4/arch/sparc/config.in.orig	Mon Nov 13 01:44:03 2000
+++ 2.4.0test11pre4/arch/sparc/config.in	Thu Nov 16 09:12:31 2000
@@ -32,6 +32,7 @@
 # Global things across all Sun machines.
 define_bool CONFIG_ISA n
 define_bool CONFIG_EISA n
+define_bool CONFIG_MCA n
 define_bool CONFIG_PCMCIA n
 define_bool CONFIG_SBUS y
 define_bool CONFIG_SBUSCHAR y
diff -urk.orig 2.4.0test11pre4/arch/sparc64/config.in.orig 2.4.0test11pre4/arch/sparc64/config.in
--- 2.4.0test11pre4/arch/sparc64/config.in.orig	Mon Nov 13 01:44:03 2000
+++ 2.4.0test11pre4/arch/sparc64/config.in	Thu Nov 16 09:12:45 2000
@@ -30,6 +30,7 @@
 define_bool CONFIG_HAVE_DEC_LOCK y
 define_bool CONFIG_ISA n
 define_bool CONFIG_EISA n
+define_bool CONFIG_MCA n
 define_bool CONFIG_PCMCIA n
 define_bool CONFIG_SBUS y
 define_bool CONFIG_SBUSCHAR y
diff -urk.orig 2.4.0test11pre4/drivers/net/Config.in.orig 2.4.0test11pre4/drivers/net/Config.in
--- 2.4.0test11pre4/drivers/net/Config.in.orig	Mon Nov 13 01:44:09 2000
+++ 2.4.0test11pre4/drivers/net/Config.in	Thu Nov 16 09:20:13 2000
@@ -100,7 +100,7 @@
    if [ "$CONFIG_ISA" = "y" -o "$CONFIG_EISA" = "y" -o "$CONFIG_PCI" = "y" ]; then
       tristate '  HP 10/100VG PCLAN (ISA, EISA, PCI) support' CONFIG_HP100
    fi
-   bool '  Other ISA cards' CONFIG_NET_ISA
+   dep_bool '  Other ISA cards' CONFIG_NET_ISA $CONFIG_ISA
    if [ "$CONFIG_NET_ISA" = "y" ]; then
       tristate '    Cabletron E21xx support' CONFIG_E2100
       if [ "$CONFIG_OBSOLETE" = "y" ]; then
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: PCI configuration changes

[Jeff Garzik]

Peter Samuelson wrote:
> 
> [Andrzej Krzysztofowicz]
> > Note, that as CONFIG_MCA is defined only for i386 the dependencies on
> > $CONFIG_MCA are no-op for other architectures (in
> > Configure/Menuconfig).  Either CONFIG_MCA should be defined for all
> > architectures or there should be if ... fi around these lines.
> 
> The former, I think.  Less confusing in the long run.
> 
> > BTW, is there any reason for not replacing
> >    bool '  Other ISA cards' CONFIG_NET_ISA
> > by
> >   dep_bool '  Other ISA cards' CONFIG_NET_ISA $CONFIG_ISA
> > to eliminate more drivers from non-ISA arch configs ?
> 
> Looks good to me.  Anything to remove clutter from config menus....

Patch looks ok to me, applied.

-- 
Jeff Garzik             |
Building 1024           | The chief enemy of creativity is "good" sense
MandrakeSoft            |          -- Picasso
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: PCI configuration changes

[Andrzej Krzysztofowicz]

> Peter Samuelson wrote:
> > [Andrzej Krzysztofowicz]
> > > Note, that as CONFIG_MCA is defined only for i386 the dependencies on
> > > $CONFIG_MCA are no-op for other architectures (in
> > > Configure/Menuconfig).  Either CONFIG_MCA should be defined for all
> > > architectures or there should be if ... fi around these lines.
> > 
> > Looks good to me.  Anything to remove clutter from config menus....
> 
> Patch looks ok to me, applied.

I think the following i386 chunk should be added to the patch:


--- arch/i386/config.in.old	Tue Nov 14 23:18:12 2000
+++ arch/i386/config.in	Tue Nov 14 23:19:10 2000
@@ -198,6 +198,8 @@
 
 if [ "$CONFIG_VISWS" != "y" ]; then
    bool 'MCA support' CONFIG_MCA
+else
+   define_bool CONFIG_MCA n
 fi
 
 bool 'Support for hot-pluggable devices' CONFIG_HOTPLUG


-- 
=======================================================================
  Andrzej M. Krzysztofowicz               ankry@mif.pg.gda.pl
  phone (48)(58) 347 14 61
Faculty of Applied Phys. & Math.,   Technical University of Gdansk
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: PCI configuration changes

[Jeff Garzik]

Oh yeah, another MCA cleanup to consider -- like EISA, there exists a
'MCA_bus' variable which is 0 or 1, depending on the absence or presence
of MCA bus on the current system.

When CONFIG_MCA is enabled, this should be variable like it currently is
[on x86].  When CONFIG_MCA==n, MCA_bus should be unconditionally defined
to zero.

Look at how 'EISA_bus' is handled in test11-pre5...

	Jeff



-- 
Jeff Garzik             |
Building 1024           | The chief enemy of creativity is "good" sense
MandrakeSoft            |          -- Picasso
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: PCI configuration changes

[Peter Samuelson]

[Jeff Garzik <jgarzik@mandrakesoft.com>]
> Oh yeah, another MCA cleanup to consider -- like EISA, there exists a
> 'MCA_bus' variable which is 0 or 1, depending on the absence or presence
> of MCA bus on the current system.

OK, workin' on it..

Peter
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: PCI configuration changes

[Peter Samuelson]

[Jeff Garzik <jgarzik@mandrakesoft.com>]
> For drivers that are 100% MCA, they do not need to test MCA_bus,
> because that test can be done in Config.in.

I think Andrzej was concerned with a driver assuming that just because
CONFIG_MCA is defined, there *is* an MCA bus on the machine.  This is
of course an invalid assumption.  I will try to make sure that the
driver init in each case has something like

	if (!MCA_bus)
		return -ENODEV;

Peter
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/