Randy Dunlap | 2 Mar 04:42 2009
Picon

Re: [PATCH 0/6] [KBUILD] configuration system improvements

Cheng Renquan wrote:
> This group of patches improve the usability of kernel configuration system,
> mainly,
> 
> 1. add symbol's value shown accompanied in config item's help message;
> 2. move the real plumbing functions code from mconf.c to menu.c, make them
>    be able to be shared with other configuration methods;
> 3. add "symbol's value shown" support for gconfig/xconfig/config;
> 
>   0001--KBUILD-add-symbol-value-to-help-find-the-real-depe.patch
>   0002--KBUILD-add-menu_get_ext_help-function-to-display-m.patch
>   0003--KBUILD-menuconfig-improvements.patch
>   0004--KBUILD-make-use-of-menu_get_ext_help-in-gconfig.patch
>   0005--KBUILD-make-use-of-menu_get_ext_help-in-qconfig.patch
>   0006--KBUILD-make-use-of-menu_get_ext_help-in-make-conf.patch
> 
> I found gconfig still lack of "search by symbol" support, and that would be
> in another patch.
> 
> Request for comments, thanks.

Hi,

I still like this idea a lot.  However, with "make gconfig" I am
seeing a segfault when I click on "Firmware Drivers" in the main menu.
I only see this with these patches applied.

Other than that, they look good to me.

Here's the gconfig info:
(Continue reading)

Cheng Renquan | 3 Mar 05:09 2009
Picon

Re: [PATCH 0/6] [KBUILD] configuration system improvements

On Mon, Mar 2, 2009 at 11:42 AM, Randy Dunlap <randy.dunlap <at> oracle.com> wrote:
> Cheng Renquan wrote:
>> This group of patches improve the usability of kernel configuration system,
>> mainly,
>>
>> 1. add symbol's value shown accompanied in config item's help message;
>> 2. move the real plumbing functions code from mconf.c to menu.c, make them
>>    be able to be shared with other configuration methods;
>> 3. add "symbol's value shown" support for gconfig/xconfig/config;
>>
>>   0001--KBUILD-add-symbol-value-to-help-find-the-real-depe.patch
>>   0002--KBUILD-add-menu_get_ext_help-function-to-display-m.patch
>>   0003--KBUILD-menuconfig-improvements.patch
>>   0004--KBUILD-make-use-of-menu_get_ext_help-in-gconfig.patch
>>   0005--KBUILD-make-use-of-menu_get_ext_help-in-qconfig.patch
>>   0006--KBUILD-make-use-of-menu_get_ext_help-in-make-conf.patch
>>
>> I found gconfig still lack of "search by symbol" support, and that would be
>> in another patch.
>>
>> Request for comments, thanks.
>
> Hi,
>
> I still like this idea a lot.  However, with "make gconfig" I am
> seeing a segfault when I click on "Firmware Drivers" in the main menu.
> I only see this with these patches applied.
>
> Other than that, they look good to me.
>
(Continue reading)

Randy Dunlap | 3 Mar 19:24 2009
Picon

Re: [PATCH 0/6] [KBUILD] configuration system improvements

Cheng Renquan wrote:
> On Mon, Mar 2, 2009 at 11:42 AM, Randy Dunlap <randy.dunlap <at> oracle.com> wrote:
>> Cheng Renquan wrote:
>>> This group of patches improve the usability of kernel configuration system,
>>> mainly,
>>>
>>> 1. add symbol's value shown accompanied in config item's help message;
>>> 2. move the real plumbing functions code from mconf.c to menu.c, make them
>>>    be able to be shared with other configuration methods;
>>> 3. add "symbol's value shown" support for gconfig/xconfig/config;
>>>
>>>   0001--KBUILD-add-symbol-value-to-help-find-the-real-depe.patch
>>>   0002--KBUILD-add-menu_get_ext_help-function-to-display-m.patch
>>>   0003--KBUILD-menuconfig-improvements.patch
>>>   0004--KBUILD-make-use-of-menu_get_ext_help-in-gconfig.patch
>>>   0005--KBUILD-make-use-of-menu_get_ext_help-in-qconfig.patch
>>>   0006--KBUILD-make-use-of-menu_get_ext_help-in-make-conf.patch
>>>
>>> I found gconfig still lack of "search by symbol" support, and that would be
>>> in another patch.
>>>
>>> Request for comments, thanks.
>> Hi,
>>
>> I still like this idea a lot.  However, with "make gconfig" I am
>> seeing a segfault when I click on "Firmware Drivers" in the main menu.
>> I only see this with these patches applied.
>>
>> Other than that, they look good to me.
>>
(Continue reading)

Randy Dunlap | 3 Mar 20:55 2009
Picon

Re: [PATCH 0/6] [KBUILD] configuration system improvements

Randy Dunlap wrote:
> Cheng Renquan wrote:
>> On Mon, Mar 2, 2009 at 11:42 AM, Randy Dunlap <randy.dunlap <at> oracle.com> wrote:
>>> Cheng Renquan wrote:
>>>> This group of patches improve the usability of kernel configuration system,
>>>> mainly,
>>>>
>>>> 1. add symbol's value shown accompanied in config item's help message;
>>>> 2. move the real plumbing functions code from mconf.c to menu.c, make them
>>>>    be able to be shared with other configuration methods;
>>>> 3. add "symbol's value shown" support for gconfig/xconfig/config;
>>>>
>>>>   0001--KBUILD-add-symbol-value-to-help-find-the-real-depe.patch
>>>>   0002--KBUILD-add-menu_get_ext_help-function-to-display-m.patch
>>>>   0003--KBUILD-menuconfig-improvements.patch
>>>>   0004--KBUILD-make-use-of-menu_get_ext_help-in-gconfig.patch
>>>>   0005--KBUILD-make-use-of-menu_get_ext_help-in-qconfig.patch
>>>>   0006--KBUILD-make-use-of-menu_get_ext_help-in-make-conf.patch
>>>>
>>>> I found gconfig still lack of "search by symbol" support, and that would be
>>>> in another patch.
>>>>
>>>> Request for comments, thanks.
>>> Hi,
>>>
>>> I still like this idea a lot.  However, with "make gconfig" I am
>>> seeing a segfault when I click on "Firmware Drivers" in the main menu.
>>> I only see this with these patches applied.
>>>
>>> Other than that, they look good to me.
(Continue reading)

Cheng Renquan | 4 Mar 07:50 2009
Picon

Re: [PATCH 0/6] [KBUILD] configuration system improvements

On Wed, Mar 4, 2009 at 3:55 AM, Randy Dunlap <randy.dunlap <at> oracle.com> wrote:
>> This is on OpenSUSE 10.3 on x86_64.
>>
>> glibc-2.6.1-18
>> gtk2-2.12.0-5
>>
>> I don't know that I have any other gtk+ programs.
>>
>>> I have developed these patches on default Fedora 10 i686, it runs
>>> well, and today I have tested it under a latest Gentoo x86_64 desktop,
>>> all seems run well here.
>>
>> Here's a backtrace from gdb.  gconfig was built without -d, however.

Your glibc and gtk version not old, the building environment seems
good, but here I don't have a OpenSUSE 10.3 x86_64, really no idea on
how to resolve that problem, maybe I have chance to install a OpenSUSE
10.3 x86_64 these days, the I will try that,

Today I tested these patches on another pristine debian-5.0 (lenny)
x86_64 system, it also performed well.
On systems I tested (glibc 2.6,2.7,2.8, gtk-2.12 or gtk-2.14), it may
report warnings from gtkitemfactory.h and ununsed variables from
images.c, but doesn't matter, after "scripts/kconfig/gconf
arch/x86/Kconfig" run, the gtk gconfig window will display, with
symbol's value quoted well,

/usr/include/gtk-2.0/gtk/gtkitemfactory.h:47: warning: function
declaration isn’t a prototype

(Continue reading)

Randy Dunlap | 4 Mar 18:18 2009
Picon

Re: [PATCH 0/6] [KBUILD] configuration system improvements

Cheng Renquan wrote:

[snip]

> Maybe you can try to "make gconfig" from a vanilla
> linux-2.6.29-rc6-git5 kernel, without these kbuild patches, to check
> if your system can configure kernel with gconfig?

I had already reported that gconfig without this patch series works fine.

> Alternatively, you can use this to compile gconfig,
> 
> gcc -Wall -Wl,--as-needed -Wstrict-prototypes -O2 -fomit-frame-pointer
> -D LKC_DIRECT_LINK `pkg-config --cflags gtk+-2.0 gmodule-2.0
> libglade-2.0` -o scripts/kconfig/gconf scripts/kconfig/gconf.c
> scripts/kconfig/kconfig_load.c scripts/kconfig/zconf.tab.c -lncurses
> `pkg-config --libs gtk+-2.0 gmodule-2.0 libglade-2.0`

OK, did that.

> this will build a smaller scripts/kconfig/gconf, with less library
> dependencies, or try "-O2" to "-O0", "-ggdb" to compute more debuging
> information, thanks,

With -O0 -ggdb, this is the backtrace:

(gdb) bt
#0  0x000000000041a6ab in get_symbol_str (r=0x7fff41724130, sym=0x0)
    at scripts/kconfig/menu.c:497
#1  0x000000000041a87c in menu_get_ext_help (menu=0xd82370, 
(Continue reading)

Sam Ravnborg | 7 Mar 13:29 2009

Re: [PATCHv2] Fix 'make rpm' when CONFIG_LOCALVERSION_AUTO=y and using SCM tree

On Thu, Feb 19, 2009 at 11:45:31PM -0800, Josh Hunt wrote:
> Updated patch with Sam's fixes as per this discussion:
> http://lkml.org/lkml/2009/2/15/18
> 
> Running 'make rpm' fails when CONFIG_LOCALVERSION_AUTO=y and using a kernel
> source tree under SCM.  This is due to KERNELRELEASE being different when the
> initial make is run and when make is run from rpmbuild.
...
Hand applied - your patches continue to be wordwrapped.
Please see if you can fix this next time.

Thanks,
	Sam
--
To unsubscribe from this list: send the line "unsubscribe linux-kbuild" in
the body of a message to majordomo <at> vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Sam Ravnborg | 7 Mar 13:59 2009

Re: [PATCH] Fix mkspec to cleanup RPM_BUILD_ROOT

On Sun, Feb 22, 2009 at 10:54:55AM -0800, Josh Hunt wrote:
> The contents of the %clean section in mkspec is currently commented
> out leaving RPM_BUILD_ROOT and its contents on the build machine.
> This patch removes it once the rpm build process is complete.
> 
> Signed-off-by: Josh Hunt <josh <at> scalex86.org>

Applied,
	Sam
--
To unsubscribe from this list: send the line "unsubscribe linux-kbuild" in
the body of a message to majordomo <at> vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Kyle McMartin | 11 Mar 17:17 2009

[PATCH] scripts/unifdef.c: rename getline symbol to something else

From: Kyle McMartin <kyle <at> redhat.com>

Rawhide builds are currently failing to build unifdef.c, as the next
version of glibc changes the default _POSIX_C_SOURCE level, which
exposes getline() from <stdio.h>

scripts/unifdef.c:209: error: conflicting types for 'getline'
/usr/include/stdio.h:653: note: previous declaration of 'getline' was
here
make[2]: *** [scripts/unifdef] Error 1
make[1]: *** [__headers] Error 2
make: *** [vmlinux] Error 2

Rename the symbol in unifdef.c to parseline to avoid this conflicting
declaration.

Otherwise Jakub says we could add a -D_POSIX_C_SOURCE=200112L as a
workaround to unifdef CFLAGS, but this seems like it would be less
desirable in the long term.

Signed-off-by: Kyle McMartin <kyle <at> redhat.com>

---
diff --git a/scripts/unifdef.c b/scripts/unifdef.c
index 552025e..977e682 100644
--- a/scripts/unifdef.c
+++ b/scripts/unifdef.c
 <at>  <at>  -206,7 +206,7  <at>  <at>  static void             done(void);
 static void             error(const char *);
 static int              findsym(const char *);
(Continue reading)

Sam Ravnborg | 11 Mar 19:37 2009

Re: [PATCH] scripts/unifdef.c: rename getline symbol to something else

On Wed, Mar 11, 2009 at 12:17:33PM -0400, Kyle McMartin wrote:
> From: Kyle McMartin <kyle <at> redhat.com>
> 
> Rawhide builds are currently failing to build unifdef.c, as the next
> version of glibc changes the default _POSIX_C_SOURCE level, which
> exposes getline() from <stdio.h>
> 
> scripts/unifdef.c:209: error: conflicting types for 'getline'
> /usr/include/stdio.h:653: note: previous declaration of 'getline' was
> here
> make[2]: *** [scripts/unifdef] Error 1
> make[1]: *** [__headers] Error 2
> make: *** [vmlinux] Error 2
> 
> Rename the symbol in unifdef.c to parseline to avoid this conflicting
> declaration.

I have a similar patch queued in kbuild-fixes.git.

	Sam
--
To unsubscribe from this list: send the line "unsubscribe linux-kbuild" in
the body of a message to majordomo <at> vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html


Gmane