Skip to contentSkip to conversation

Patch that allows >=2.6.12 kernel to build on nls free systems

PATCH Yuri Vasilevski 4 participants 6 messages Oct 26, 2005 – Nov 1, 2005 Kconfig

lore
Share & tools
lore
#0 lore raw

Hi all,

I made a patch that detects if libintl.h (needed for nls) is present on the host system and if it's not, it nls support is disabled by providing dummies for the used nls functions.

This way if there is nls support on the host system the *config targets will build according to Arnaldo Carvalho de Melo's i18n modifications, else it just uses the original English messages.

I have also made a bug report at kernel's bugzilla:
http://bugzilla.kernel.org/show_bug.cgi?id=5501
And there is a discussion about this problem in Gentoo's bugzilla:
http://bugs.gentoo.org/show_bug.cgi?id=99810

scripts/kconfig/Makefile +9 -0
diff -Naur linux-2.6.14_rc2.orig/scripts/kconfig/Makefile linux-2.6.14_rc2/scripts/kconfig/Makefile
@@ -116,6 +116,15 @@
 clean-files	:= lkc_defs.h qconf.moc .tmp_qtcheck \
 		   .tmp_gtkcheck zconf.tab.c zconf.tab.h lex.zconf.c
 
+# Needed for systems without gettext
+KBUILD_HAVE_NLS := $(shell \
+     if echo "\#include <libint.h>" | $(HOSTCC) $(HOSTCFLAGS) -E - > /dev/null 2>&1 ; \
+     then echo yes ; \
+     else echo no ; fi)
+ifeq ($(KBUILD_HAVE_NLS),no)
+HOSTCFLAGS	+= -DKBUILD_NO_NLS
+endif
+
 # generated files seem to need this to find local include files
 HOSTCFLAGS_lex.zconf.o	:= -I$(src)
 HOSTCFLAGS_zconf.tab.o	:= -I$(src)
scripts/kconfig/lkc.h +7 -1
diff -Naur linux-2.6.14_rc2.orig/scripts/kconfig/lkc.h linux-2.6.14_rc2/scripts/kconfig/lkc.h
@@ -8,7 +8,13 @@
 
 #include "expr.h"
 
-#include <libintl.h>
+#ifndef KBUILD_NO_NLS
+# include <libintl.h>
+#else
+# define gettext(Msgid) ((const char *) (Msgid))
+# define textdomain(Domainname) ((const char *) (Domainname))
+# define bindtextdomain(Domainname, Dirname) ((const char *) (Dirname))
+#endif
 
 #ifdef __cplusplus
 extern "C" {

Yuri.

PS: Please CC me on replay as I'm not in the kernel mailing list.

#1 lore raw

Hi,

Quoted 14 lines · Yuri Vasilevski

I have also made a bug report at kernel's bugzilla:
http://bugzilla.kernel.org/show_bug.cgi?id=5501
And there is a discussion about this problem in Gentoo's bugzilla:
http://bugs.gentoo.org/show_bug.cgi?id=99810

diff -Naur linux-2.6.14_rc2.orig/scripts/kconfig/Makefile linux-2.6.14_rc2/scripts/kconfig/Makefile

scripts/kconfig/Makefile +2 -0
@@ -116,6 +116,15 @@
 clean-files	:= lkc_defs.h qconf.moc .tmp_qtcheck \
 		   .tmp_gtkcheck zconf.tab.c zconf.tab.h lex.zconf.c
 
+# Needed for systems without gettext
+KBUILD_HAVE_NLS := $(shell \

scripts/kconfig/Makefile line 130

Quoted 10 lines
scripts/kconfig/Makefile +7 -0
+     if echo "\#include <libint.h>" | $(HOSTCC) $(HOSTCFLAGS) -E - > /dev/null 2>&1 ; \
+     then echo yes ; \
+     else echo no ; fi)
+ifeq ($(KBUILD_HAVE_NLS),no)
+HOSTCFLAGS	+= -DKBUILD_NO_NLS
+endif
+
 # generated files seem to need this to find local include files
 HOSTCFLAGS_lex.zconf.o	:= -I$(src)
 HOSTCFLAGS_zconf.tab.o	:= -I$(src)

Sam, I was wondering about this kind runtime configuration stuff, if we could cache the information and don't run the tests every time.

bye, Roman

#2 lore raw
In reply to Yuri Vasilevski
Quoted 9 lines · Yuri Vasilevski

Hi all,

I made a patch that detects if libintl.h (needed for nls) is present on the host system and if it's not, it nls support is disabled by providing dummies for the used nls functions.

This way if there is nls support on the host system the *config targets will build according to Arnaldo Carvalho de Melo's i18n modifications, else it just uses the original English messages.

Haven't tested this myself, but looks like the way to go.

Thanks,

- Arnaldo

#3 lore raw
In reply to Yuri Vasilevski
Quoted 43 lines · Yuri Vasilevski

Hi all,

I made a patch that detects if libintl.h (needed for nls) is present on the host system and if it's not, it nls support is disabled by providing dummies for the used nls functions.

This way if there is nls support on the host system the *config targets will build according to Arnaldo Carvalho de Melo's i18n modifications, else it just uses the original English messages.

I have also made a bug report at kernel's bugzilla:
http://bugzilla.kernel.org/show_bug.cgi?id=5501
And there is a discussion about this problem in Gentoo's bugzilla:
http://bugs.gentoo.org/show_bug.cgi?id=99810

diff -Naur linux-2.6.14_rc2.orig/scripts/kconfig/Makefile linux-2.6.14_rc2/scripts/kconfig/Makefile

scripts/kconfig/Makefile 2005-11-18 03:52:03 +0000 renamed from scripts/kconfig/Makefile 2005-11-06 04:13:01 +0000 +9 -0
@@ -116,6 +116,15 @@
 clean-files    := lkc_defs.h qconf.moc .tmp_qtcheck \
                   .tmp_gtkcheck zconf.tab.c zconf.tab.h lex.zconf.c
 
+# Needed for systems without gettext
+KBUILD_HAVE_NLS := $(shell \
+     if echo "\#include <libint.h>" | $(HOSTCC) $(HOSTCFLAGS) -E - > /dev/null 2>&1 ; \
+     then echo yes ; \
+     else echo no ; fi)
+ifeq ($(KBUILD_HAVE_NLS),no)
+HOSTCFLAGS     += -DKBUILD_NO_NLS
+endif
+
 # generated files seem to need this to find local include files
 HOSTCFLAGS_lex.zconf.o := -I$(src)
 HOSTCFLAGS_zconf.tab.o := -I$(src)

diff -Naur linux-2.6.14_rc2.orig/scripts/kconfig/lkc.h linux-2.6.14_rc2/scripts/kconfig/lkc.h

scripts/kconfig/lkc.h 2005-11-18 02:23:07 +0000 renamed from scripts/kconfig/lkc.h 2005-11-06 04:13:01 +0000 +0 -1
@@ -8,7 +8,13 @@
 
 #include "expr.h"
 
-#include <libintl.h>

scripts/kconfig/lkc.h 2005-11-18 02:23:07 +0000 line 20

Quoted 10 lines
scripts/kconfig/lkc.h 2005-11-18 02:23:07 +0000 renamed from scripts/kconfig/lkc.h 2005-11-06 04:13:01 +0000 +7 -0
+#ifndef KBUILD_NO_NLS
+# include <libintl.h>
+#else
+# define gettext(Msgid) ((const char *) (Msgid))
+# define textdomain(Domainname) ((const char *) (Domainname))
+# define bindtextdomain(Domainname, Dirname) ((const char *) (Dirname))
+#endif
 
 #ifdef __cplusplus
 extern "C" {

Looks like this patch was merged:

http://www.kernel.org/git/gitweb.cgi?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=70a6a0cb92f24fd6bbe2e75299168909f735676a

I noticed with builds of -git3/-git4, I get the following complaints from oldconfig:

scripts/kconfig/mconf.c: In function `main':
scripts/kconfig/mconf.c:1048: warning: statement with no effect scripts/kconfig/mconf.c:1049: warning: statement with no effect

Not a big deal, just more complaints to have to see during the build process (with CONFIG_NLS=y) :)

Thanks,
Nish

#4 lore raw

Hi,

It was caused by my copy/pasting error.

Quoted 11 lines · On Tue, 1 Nov 2005 09:13:50 -0800 Nish Aravamudan <[email protected]>…

scripts/kconfig/Makefile 2005-11-18 03:52:03 +0000 line 121

Quoted 10 lines · Yuri Vasilevski

diff -Naur linux-2.6.14_rc2.orig/scripts/kconfig/Makefile linux-2.6.14_rc2/scripts/kconfig/Makefile

scripts/kconfig/Makefile 2005-11-18 03:52:03 +0000 renamed from scripts/kconfig/Makefile 2005-11-06 04:13:01 +0000 +3 -0
@@ -116,6 +116,15 @@
 clean-files    := lkc_defs.h qconf.moc .tmp_qtcheck \
                   .tmp_gtkcheck zconf.tab.c zconf.tab.h lex.zconf.c
 
+# Needed for systems without gettext
+KBUILD_HAVE_NLS := $(shell \
+     if echo "\#include <libint.h>" | $(HOSTCC) $(HOSTCFLAGS) -E - > /dev/null 2>&1 ; \

The file name is libintl.h and not libint.h (patch in the next e-mail)

Quoted 11 lines

Looks like this patch was merged:

http://www.kernel.org/git/gitweb.cgi?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=70a6a0cb92f24fd6bbe2e75299168909f735676a

I noticed with builds of -git3/-git4, I get the following complaints from oldconfig:

scripts/kconfig/mconf.c: In function `main':
scripts/kconfig/mconf.c:1048: warning: statement with no effect scripts/kconfig/mconf.c:1049: warning: statement with no effect

This should be the output on nls free systems, but all systems were detected as nls free because of that error.

Quoted 5 lines

Not a big deal, just more complaints to have to see during the build process (with CONFIG_NLS=y) :)

Thanks,
Nish

Sorry for this mistake.

Yuri.

#5 lore raw
Quoted 15 lines · Yuri Vasilevski

Hi,

It was caused by my copy/pasting error.

Quoted 8 lines · On Tue, 1 Nov 2005 09:13:50 -0800 Nish Aravamudan <[email protected]>…
Quoted 7 lines · Yuri Vasilevski

diff -Naur linux-2.6.14_rc2.orig/scripts/kconfig/Makefile linux-2.6.14_rc2/scripts/kconfig/Makefile

scripts/kconfig/Makefile 2005-11-18 03:52:03 +0000 renamed from scripts/kconfig/Makefile 2005-11-06 04:13:01 +0000 +0 -0
@@ -116,6 +116,15 @@
 clean-files    := lkc_defs.h qconf.moc .tmp_qtcheck \
                   .tmp_gtkcheck zconf.tab.c zconf.tab.h lex.zconf.c
 
Quoted 5 lines
Quoted 3 lines

scripts/kconfig/Makefile 2005-11-18 03:52:03 +0000 line 121

Quoted 3 lines
scripts/kconfig/Makefile 2005-11-18 03:52:03 +0000 renamed from scripts/kconfig/Makefile 2005-11-06 04:13:01 +0000 +3 -0
+# Needed for systems without gettext
+KBUILD_HAVE_NLS := $(shell \
+     if echo "\#include <libint.h>" | $(HOSTCC) $(HOSTCFLAGS) -E - > /dev/null 2>&1 ; \

The file name is libintl.h and not libint.h (patch in the next e-mail)

Makes sense.

Quoted 7 lines
Quoted 7 lines

Looks like this patch was merged:

http://www.kernel.org/git/gitweb.cgi?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=70a6a0cb92f24fd6bbe2e75299168909f735676a

I noticed with builds of -git3/-git4, I get the following complaints from oldconfig:

Quoted 6 lines
Quoted 3 lines

scripts/kconfig/mconf.c: In function `main':
scripts/kconfig/mconf.c:1048: warning: statement with no effect scripts/kconfig/mconf.c:1049: warning: statement with no effect

This should be the output on nls free systems, but all systems were detected as nls free because of that error.

Ah, I see.

Quoted 7 lines
Quoted 5 lines

Not a big deal, just more complaints to have to see during the build process (with CONFIG_NLS=y) :)

Thanks,
Nish

Sorry for this mistake.

No worries, it didn't make the compile fail :) Thanks for the quick response!

Thanks,
Nish