Skip to contentSkip to conversation

GENERIC still failing?

DISCUSSION David Mosberger 5 participants 10 messages Oct 15, 2003

lore
Share & tools
lore
#0 lore raw

According to this page:

http://www.gelato.unsw.edu.au/kerncomp/

GENERIC builds are still failing:

gcc -Wp,-MD,fs/ext2/.balloc.o.d -nostdinc -iwithprefix include -D__KERNEL__ -Iinclude -D__KERNEL__ -Iinclude -Wall -Wstrict-prototypes -Wno-trigraphs -O2 -fno-strict-aliasing -fno-common -pipe -ffixed-r13 -mfixed-rangeñ2-f15,f32-f127 -falign-functions2 -frename-registers -fomit-frame-pointer -mconstant-gp -DKBUILD_BASENAMEºlloc -DKBUILD_MODNAME=ext2 -c -o fs/ext2/balloc.o fs/ext2/balloc.c
arch/ia64/sn/kernel/setup.c:94: error: MAX_PHYSNODE_ID' undeclared here (not in a function) arch/ia64/sn/kernel/setup.c:94: error: conflicting types for physical_node_map' include/asm/sn/sn_cpuid.h:126: error: previous declaration of `physical_node_map'

Anybody care to send me a fix for this?

	--david
#1 lore raw
Quoted 4 lines · David Mosberger

According to this page:

http://www.gelato.unsw.edu.au/kerncomp/

Quoted 8 lines

GENERIC builds are still failing:

gcc -Wp,-MD,fs/ext2/.balloc.o.d -nostdinc -iwithprefix include -D__KERNEL__ -Iinclude -D__KERNEL__ -Iinclude -Wall -Wstrict-prototypes -Wno-trigraphs -O2 -fno-strict-aliasing -fno-common -pipe -ffixed-r13 -mfixed-rangeñ2-f15,f32-f127 -falign-functions2 -frename-registers -fomit-frame-pointer -mconstant-gp -DKBUILD_BASENAMEºlloc -DKBUILD_MODNAME=ext2 -c -o fs/ext2/balloc.o fs/ext2/balloc.c
arch/ia64/sn/kernel/setup.c:94: error: MAX_PHYSNODE_ID' undeclared here (not in a function) arch/ia64/sn/kernel/setup.c:94: error: conflicting types for physical_node_map' include/asm/sn/sn_cpuid.h:126: error: previous declaration of `physical_node_map'

Anybody care to send me a fix for this?

Which tree is this? Your to-linus-2.5 tree? Let me know and I'll fix it.

Jesse

#2 lore raw
In reply to David Mosberger
Quoted 5 lines · jbarnes
Quoted 4 lines · David Mosberger

According to this page:

http://www.gelato.unsw.edu.au/kerncomp/

Quoted 8 lines
Quoted 8 lines

GENERIC builds are still failing:

gcc -Wp,-MD,fs/ext2/.balloc.o.d -nostdinc -iwithprefix include -D__KERNEL__ -Iinclude -D__KERNEL__ -Iinclude -Wall -Wstrict-prototypes -Wno-trigraphs -O2 -fno-strict-aliasing -fno-common -pipe -ffixed-r13 -mfixed-rangeñ2-f15,f32-f127 -falign-functions2 -frename-registers -fomit-frame-pointer -mconstant-gp -DKBUILD_BASENAMEºlloc -DKBUILD_MODNAME=ext2 -c -o fs/ext2/balloc.o fs/ext2/balloc.c
arch/ia64/sn/kernel/setup.c:94: error: MAX_PHYSNODE_ID' undeclared here (not in a function) arch/ia64/sn/kernel/setup.c:94: error: conflicting types for physical_node_map' include/asm/sn/sn_cpuid.h:126: error: previous declaration of `physical_node_map'

Anybody care to send me a fix for this?

Looks like CONFIG_NUMA and CONFIG_DISCONTIGMEM are missing from the config being built. A generic kernel won't work without them.

Jesse

#3 lore raw
In reply to David Mosberger
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line · On Tue, 14 Oct 2003 17:12:17 -0700, [email protected] (Jesse Barnes) said:

Jesse> On Tue, Oct 14, 2003 at 05:05:14PM -0700, David Mosberger

Quoted 4 lines · Jesse> wrote:
Quoted 4 lines

According to this page:
http://www.gelato.unsw.edu.au/kerncomp/
GENERIC builds are still failing:
Anybody care to send me a fix for this?

Jesse> Which tree is this? Your to-linus-2.5 tree? Let me know and Jesse> I'll fix it.

That's for the official Linus tree. It shouldn't matter though: Linus recently pulled my tree, so he should have all of your fixes already.

	--david
#4 lore raw
In reply to David Mosberger
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line · On Tue, 14 Oct 2003 17:22:05 -0700, [email protected] (Jesse Barnes) said:

Jesse> Looks like CONFIG_NUMA and CONFIG_DISCONTIGMEM are missing Jesse> from the config being built. A generic kernel won't work Jesse> without them.

Shouldn't they be forced for GENERIC, then?

	--david
#5 lore raw
In reply to David Mosberger
Quoted 2 lines · Jesse Barnes

Looks like CONFIG_NUMA and CONFIG_DISCONTIGMEM are missing from the config being built. A generic kernel won't work without them.

mea culpa, as usual. Is there some way to make it when you select generic it always requires CONFIG_NUMA/DISCONTIGMEM?

Also, trying to look at the help text, it seems that the sub-items of a 'choice' in Kconfig can't have individual help texts?

For example

choice
prompt "IA-64 processor type"
---help---
this will show when you select help on 'IA-64 processor type'
config ITANIUM
bool "Itanium"
---help---
this text appears not to show when you select help on 'Itanium' in the choice.
endchoice

I'm not sure if this is the desired effect, but certainly some of the help texts appear mis-placed under an option of the choice rather than at the choice it's self (see attached for example).

-i
[email protected]
http://www.gelato.unsw.edu.au

===== arch/ia64/Kconfig 1.47 vs edited =====

arch/ia64/Kconfig +12 -12
@@ -36,27 +36,22 @@
 choice
 	prompt "IA-64 processor type"
 	default ITANIUM
-
-config ITANIUM
-	bool "Itanium"
 	help
 	  Select your IA-64 processor type.  The default is Intel Itanium.
 	  This choice is safe for all IA-64 systems, but may not perform
 	  optimally on systems with, say, Itanium 2 or newer processors.
 
+config ITANIUM
+	bool "Itanium"
+
 config MCKINLEY
 	bool "Itanium 2"
-	help
-	  Select this to configure for an Itanium 2 (McKinley) processor.
 
 endchoice
 
 choice
 	prompt "IA-64 system type"
 	default IA64_GENERIC
-
-config IA64_GENERIC
-	bool "generic"
 	---help---
 	  This selects the system type of your hardware.  A "generic" kernel
 	  will run on any supported IA-64 system.  However, if you configure
@@ -74,7 +69,12 @@
 	  DIG-compliant  For DIG ("Developer's Interface Guide") compliant
 	  systems.
 
-	  If you don't know what to do, choose "generic".
+	  If you don't know what to do, choose "generic".  Note that
+	  "generic" also requires  you enable NUMA support and Discontiguous 
+	  memory support.
+
+config IA64_GENERIC
+	bool "generic"
 
 config IA64_DIG
 	bool "DIG-compliant"
@@ -97,9 +97,6 @@
 choice
 	prompt "Kernel page size"
 	default IA64_PAGE_SIZE_16KB
-
-config IA64_PAGE_SIZE_4KB
-	bool "4KB"
 	---help---
 	  This lets you select the page size of the kernel.  For best IA-64
 	  performance, a page size of 8KB or 16KB is recommended.  For best
@@ -114,6 +111,9 @@
 	  64KB               Requires Itanium 2 or newer processor.
 
 	  If you don't know what to do, choose 16KB.
+
+config IA64_PAGE_SIZE_4KB
+	bool "4KB"
 
 config IA64_PAGE_SIZE_8KB
 	bool "8KB"
#6 lore raw
In reply to David Mosberger

On Wed, 15 Oct 2003 12:15:48 +1000 Ian Wienand <[email protected]> wrote:

| On Tue, Oct 14, 2003 at 05:22:05PM -0700, Jesse Barnes wrote:
| > Looks like CONFIG_NUMA and CONFIG_DISCONTIGMEM are missing from the
| > config being built.  A generic kernel won't work without them.
| 
| mea culpa, as usual. Is there some way to make it when you select
| generic it always requires CONFIG_NUMA/DISCONTIGMEM?

Yes, use "select" (see Documentation/kbuild/kconfig-language.txt). E.g.,

config GENERIC

	bool "blah"
	select NUMA
	select DISCONTIGMEM
	---help---
| Also, trying to look at the help text, it seems that the sub-items of
| a 'choice' in Kconfig can't have individual help texts?
| 
| For example 
|  
| choice
| 	prompt "IA-64 processor type"
| 	---help---
| 	 this will show when you select help on 'IA-64 processor type'
| config ITANIUM
| 	bool "Itanium"
| 	---help---
| 	 this text appears not to show when you select help on 'Itanium' 
| 	 in the choice.
| endchoice
| 
| I'm not sure if this is the desired effect, but certainly some of the
| help texts appear mis-placed under an option of the choice rather than
| at the choice it's self (see attached for example).

--
~Randy

#7 lore raw
In reply to David Mosberger
Quoted 9 lines · Randy.Dunlap

On Wed, 15 Oct 2003 12:15:48 +1000 Ian Wienand <[email protected]> wrote:

| On Tue, Oct 14, 2003 at 05:22:05PM -0700, Jesse Barnes wrote:
| > Looks like CONFIG_NUMA and CONFIG_DISCONTIGMEM are missing from the
| > config being built.  A generic kernel won't work without them.
| 
| mea culpa, as usual. Is there some way to make it when you select
| generic it always requires CONFIG_NUMA/DISCONTIGMEM?
Quoted 8 lines

Yes, use "select" (see Documentation/kbuild/kconfig-language.txt). E.g.,

config GENERIC

	bool "blah"
	select NUMA
	select DISCONTIGMEM
	---help---

Ok, thanks. I'll send out a patch tomorrow unless someone beats me to it.

Jesse

#8 lore raw
In reply to David Mosberger
Quoted 16 lines · Ian Wienand

Also, trying to look at the help text, it seems that the sub-items of a 'choice' in Kconfig can't have individual help texts?

For example

choice
prompt "IA-64 processor type"
---help---
this will show when you select help on 'IA-64 processor type'
config ITANIUM
bool "Itanium"
---help---
this text appears not to show when you select help on 'Itanium' in the choice.
endchoice

Quoted 3 lines

I'm not sure if this is the desired effect, but certainly some of the help texts appear mis-placed under an option of the choice rather than at the choice it's self (see attached for example).

Which *config tool are you using? make config/oldconfig/defconfig have a bug which Roman hasn't got round to pushing upstream yet.

This whitespace-damaged diff shows what needs to be fixed:

Index: scripts//kconfig/conf.c
=================================RCS file: /var/cvs/linux-2.6/scripts/kconfig/conf.c,v retrieving revision 1.3.2.1
retrieving revision 1.3

/kconfig/conf.c 28 Sep 2003 04:06:34 -0000 1.3 renamed from /kconfig/conf.c 8 Oct 2003 20:38:32 -0000 1.3.2.1 +1 -1
diff -u -p -r1.3.2.1 -r1.3
@@ -175,7 +175,7 @@ int conf_string(struct menu *menu)
                        break;
                case '?':
                        /* print help */
-                       if (line[1] = 0) {
+                       if (line[1] = '\n') {
                                help = nohelp_text;
                                if (menu->sym->help)
                                        help = menu->sym->help;
Signature · 3 lines
"It's not Hollywood.  War is real, war is primarily not about defeat or
victory, it is about death.  I've seen thousands and thousands of dead bodies.
Do you think I want to have an academic debate on this subject?" -- Robert Fisk
#9 lore raw
In reply to David Mosberger
Quoted 4 lines · Matthew Wilcox

Which *config tool are you using? make config/oldconfig/defconfig have a bug which Roman hasn't got round to pushing upstream yet.

This whitespace-damaged diff shows what needs to be fixed:

Thanks but that patch didn't seem to make any difference; a make config seems to work as expected.

I'm using menuconfig; the problems I see appear two fold.

Firstly, I would have expected that when I selected 'Help' on, say,

IA-64 system type (generic) --->

I would get the help text that says something like what is actually assocated with the IA64_GENERIC option. This seems to be an ordering decision in the ia64 Kconfig file.

But, as that help text is associated with IA64_GENERIC, I would have expected as I when I was in the choice and selected help for

(X) generic

I would have got that big spiel

"This selects the system type of your hardware. A "generic" kernel ..."

If I use make config, when I select '1?' in the IA-64 system type choice, I receive the help text associated with it. But with make config, I can't see a way to get help for the choice overall.

-i
[email protected]
http://www.gelato.unsw.edu.au