[PATCH] NUMA-Q xquad_portio declaration

[PATCH] NUMA-Q xquad_portio declaration

[Martin J. Bligh]

This patch is from Matt Dobson. It corrects the definition of
xquad_portio, getting rid of a compile warning.
Tested on NUMA-Q and std 2-way SMP system through LTP.

Please apply,

Martin.

diff -Nur linux-2.5.25-vanilla/arch/i386/boot/compressed/misc.c linux-2.5.25-patched/arch/i386/boot/compressed/misc.c
--- linux-2.5.25-vanilla/arch/i386/boot/compressed/misc.c	Fri Jul  5 16:42:31 2002
+++ linux-2.5.25-patched/arch/i386/boot/compressed/misc.c	Thu Jul 11 15:30:03 2002
@@ -121,7 +121,7 @@
 static int lines, cols;
 
 #ifdef CONFIG_MULTIQUAD
-static void * const xquad_portio = NULL;
+static void * xquad_portio = NULL;
 #endif
 
 #include "../../../../lib/inflate.c"

Re: [PATCH] NUMA-Q xquad_portio declaration

[Alan Cox]

On Tue, 2002-08-06 at 00:53, Martin J. Bligh wrote:
> This patch is from Matt Dobson. It corrects the definition of
> xquad_portio, getting rid of a compile warning.

Marcelo - I have a much cleaner change for this.

Re: [PATCH] NUMA-Q xquad_portio declaration

[Martin J. Bligh]

>> This patch is from Matt Dobson. It corrects the definition of
>> xquad_portio, getting rid of a compile warning.
> 
> Marcelo - I have a much cleaner change for this.

Can you publish it? ;-)

Thanks,

M.

Re: [PATCH] NUMA-Q xquad_portio declaration

[Alan Cox]

On Tue, 2002-08-06 at 14:47, Martin J. Bligh wrote:
> >> This patch is from Matt Dobson. It corrects the definition of
> >> xquad_portio, getting rid of a compile warning.
> > 
> > Marcelo - I have a much cleaner change for this.
> 
> Can you publish it? ;-)

I did - its in -ac4 

Re: [PATCH] NUMA-Q xquad_portio declaration

[Martin J. Bligh]

>> >> This patch is from Matt Dobson. It corrects the definition of
>> >> xquad_portio, getting rid of a compile warning.
>> > 
>> > Marcelo - I have a much cleaner change for this.
>> 
>> Can you publish it? ;-)
> 
> I did - its in -ac4 

The STANDALONE thing? I'm not convinced that's really any cleaner,
it makes even more of a mess of io.h than there was already (though
we could consider that a lost cause ;-)). 

What's your objection to just throwing in a defn of xquad_portio?
A preference for burying the messy stuff in header files? Seems to
me that as you have to define STANDALONE now, the point is moot.

M.

Re: [PATCH] NUMA-Q xquad_portio declaration

[Alan Cox]

On Tue, 2002-08-06 at 16:06, Martin J. Bligh wrote:
> The STANDALONE thing? I'm not convinced that's really any cleaner,
> it makes even more of a mess of io.h than there was already (though
> we could consider that a lost cause ;-)). 
> 
> What's your objection to just throwing in a defn of xquad_portio?
> A preference for burying the messy stuff in header files? Seems to
> me that as you have to define STANDALONE now, the point is moot.

Because you are assuming there will be -one- kind of wackomatic PC
system - IBM's. The chances are there will be more than one as other
vendors like HP, Compaq and Dell begin shipping stuff. Having
__STANDALONE__ works for all the cases instead of exporting xquad this
hpmagic that and compaq the other in an ever growing cess pit

Re: [PATCH] NUMA-Q xquad_portio declaration

[Martin J. Bligh]

>> The STANDALONE thing? I'm not convinced that's really any cleaner,
>> it makes even more of a mess of io.h than there was already (though
>> we could consider that a lost cause ;-)). 
>> 
>> What's your objection to just throwing in a defn of xquad_portio?
>> A preference for burying the messy stuff in header files? Seems to
>> me that as you have to define STANDALONE now, the point is moot.
> 
> Because you are assuming there will be -one- kind of wackomatic PC
> system - IBM's. The chances are there will be more than one as other
> vendors like HP, Compaq and Dell begin shipping stuff. Having
> __STANDALONE__ works for all the cases instead of exporting xquad this
> hpmagic that and compaq the other in an ever growing cess pit

OK, fair enough. Would a simpler approach to what you've done be
to do in io.h something like:

#ifdef CONFIG_MULTIQUAD
 #ifdef STANDALONE
  #define xquad_portio 0
 #else
  extern void *xquad_portio;    /* Where the IO area was mapped */
 #endif
#endif /* CONFIG_MULTIQUAD */

Or something along these lines ... ? Would make the changeset
somewhat smaller. Seems to work from 30 seconds thought, but 
haven't tried it (yet).

M.

Re: [PATCH] NUMA-Q xquad_portio declaration

[Matthew Dobson]

[-- Attachment #1: Type: text/plain, Size: 1366 bytes --]

Alan & Martin,
	How does this look?  I've combined what Martin suggested with what Alan has in 
his tree...  Comments?

-Matt

Martin J. Bligh wrote:
>>>The STANDALONE thing? I'm not convinced that's really any cleaner,
>>>it makes even more of a mess of io.h than there was already (though
>>>we could consider that a lost cause ;-)). 
>>>
>>>What's your objection to just throwing in a defn of xquad_portio?
>>>A preference for burying the messy stuff in header files? Seems to
>>>me that as you have to define STANDALONE now, the point is moot.
>>
>>Because you are assuming there will be -one- kind of wackomatic PC
>>system - IBM's. The chances are there will be more than one as other
>>vendors like HP, Compaq and Dell begin shipping stuff. Having
>>__STANDALONE__ works for all the cases instead of exporting xquad this
>>hpmagic that and compaq the other in an ever growing cess pit
> 
> 
> OK, fair enough. Would a simpler approach to what you've done be
> to do in io.h something like:
> 
> #ifdef CONFIG_MULTIQUAD
>  #ifdef STANDALONE
>   #define xquad_portio 0
>  #else
>   extern void *xquad_portio;    /* Where the IO area was mapped */
>  #endif
> #endif /* CONFIG_MULTIQUAD */
> 
> Or something along these lines ... ? Would make the changeset
> somewhat smaller. Seems to work from 30 seconds thought, but 
> haven't tried it (yet).
> 
> M.
> 
> 


[-- Attachment #2: xquad_fixup-2531.patch --]
[-- Type: text/plain, Size: 1252 bytes --]

diff -Nur linux-2.5.31-vanilla/arch/i386/boot/compressed/misc.c linux-2.5.31-xquad/arch/i386/boot/compressed/misc.c
--- linux-2.5.31-vanilla/arch/i386/boot/compressed/misc.c	Sat Aug 10 18:41:40 2002
+++ linux-2.5.31-xquad/arch/i386/boot/compressed/misc.c	Thu Aug 15 14:28:33 2002
@@ -9,6 +9,8 @@
  * High loaded stuff by Hans Lermen & Werner Almesberger, Feb. 1996
  */
 
+#define STANDALONE
+
 #include <linux/linkage.h>
 #include <linux/vmalloc.h>
 #include <linux/tty.h>
@@ -120,10 +122,6 @@
 static int vidport;
 static int lines, cols;
 
-#ifdef CONFIG_MULTIQUAD
-static void * const xquad_portio = NULL;
-#endif
-
 #include "../../../../lib/inflate.c"
 
 static void *malloc(int size)
diff -Nur linux-2.5.31-vanilla/include/asm-i386/io.h linux-2.5.31-xquad/include/asm-i386/io.h
--- linux-2.5.31-vanilla/include/asm-i386/io.h	Sat Aug 10 18:41:28 2002
+++ linux-2.5.31-xquad/include/asm-i386/io.h	Thu Aug 15 15:17:31 2002
@@ -303,7 +303,11 @@
 #endif
 
 #ifdef CONFIG_MULTIQUAD
-extern void *xquad_portio;    /* Where the IO area was mapped */
+ #ifdef STANDALONE
+  #define xquad_portio 0
+ #else /* !STANDALONE */
+  extern void *xquad_portio;    /* Where the IO area was mapped */
+ #endif /* STANDALONE */
 #endif /* CONFIG_MULTIQUAD */
 
 /*