[PATCH] kbuild: fix oldnoconfig to do the right thing

[PATCH] kbuild: fix oldnoconfig to do the right thing

[Kyle McMartin]

From: Kyle McMartin <kyle@redhat.com>

Commit 861b4ea4 broke oldnoconfig when removed the oldnoconfig checks on
                       if (input_mode == nonint_oldconfig ||
                           input_mode == oldnoconfig) {
                               if (input_mode == nonint_oldconfig &&
                                   sym->name &&
                                    !sym_is_choice_value(sym)) {
to avoid oldnoconfig chugging through the else stanza.

Fix that to restore expected behaviour (which I've confirmed in the
Fedora kernel build that the configs end up looking the same.)

Signed-off-by: Kyle McMartin <kyle@redhat.com>
---
diff --git a/scripts/kconfig/conf.c b/scripts/kconfig/conf.c
index 5b7c86e..7ef429c 100644
--- a/scripts/kconfig/conf.c
+++ b/scripts/kconfig/conf.c
@@ -427,7 +427,7 @@ static void check_conf(struct menu *menu)
 				if (sym->name && !sym_is_choice_value(sym)) {
 					printf("CONFIG_%s\n", sym->name);
 				}
-			} else {
+			} else if (input_mode != oldnoconfig) {
 				if (!conf_cnt++)
 					printf(_("*\n* Restart config...\n*\n"));
 				rootEntry = menu_get_parent_menu(menu);

Re: [PATCH] kbuild: fix oldnoconfig to do the right thing

[Sam Ravnborg]

On Wed, Sep 01, 2010 at 11:21:30AM -0400, Kyle McMartin wrote:
> From: Kyle McMartin <kyle@redhat.com>
> 
> Commit 861b4ea4 broke oldnoconfig when removed the oldnoconfig checks on
>                        if (input_mode == nonint_oldconfig ||
>                            input_mode == oldnoconfig) {
>                                if (input_mode == nonint_oldconfig &&
>                                    sym->name &&
>                                     !sym_is_choice_value(sym)) {
> to avoid oldnoconfig chugging through the else stanza.
> 
> Fix that to restore expected behaviour (which I've confirmed in the
> Fedora kernel build that the configs end up looking the same.)

I played aroud with this.
I looks like the only difference I got was that before
your patch oldnoconfig produced a lot of output to stdout.
With your patch is was silent.

The resulting config was the same.

We obviously want the silent variant.

Acked-by: Sam Ravnborg <sam@ravnborg.org>

Re: [PATCH] kbuild: fix oldnoconfig to do the right thing

[Michal Marek]

On Sun, Sep 05, 2010 at 08:17:58AM +0200, Sam Ravnborg wrote:
> On Wed, Sep 01, 2010 at 11:21:30AM -0400, Kyle McMartin wrote:
> > From: Kyle McMartin <kyle@redhat.com>
> > 
> > Commit 861b4ea4 broke oldnoconfig when removed the oldnoconfig checks on
> >                        if (input_mode == nonint_oldconfig ||
> >                            input_mode == oldnoconfig) {
> >                                if (input_mode == nonint_oldconfig &&
> >                                    sym->name &&
> >                                     !sym_is_choice_value(sym)) {
> > to avoid oldnoconfig chugging through the else stanza.
> > 
> > Fix that to restore expected behaviour (which I've confirmed in the
> > Fedora kernel build that the configs end up looking the same.)
> 
> I played aroud with this.
> I looks like the only difference I got was that before
> your patch oldnoconfig produced a lot of output to stdout.
> With your patch is was silent.
> 
> The resulting config was the same.

I found that when running make oldnoconfig on the kconfig branch of
kbuild.git (which is currently a 2.6.35 + kconfig changes) against a
2.6.35-rc1 x86_64 defconfig, it mysteriously switches from
CONFIG_KERNEL_GZIP=y to CONFIG_KERNEL_LZO=y. With Kyle's patch it
doesn't do this. I applied the patch now.

Michal

Re: [PATCH] kbuild: fix oldnoconfig to do the right thing

[Kyle McMartin]

On Wed, Sep 08, 2010 at 01:55:53PM +0200, Michal Marek wrote:
> I found that when running make oldnoconfig on the kconfig branch of
> kbuild.git (which is currently a 2.6.35 + kconfig changes) against a
> 2.6.35-rc1 x86_64 defconfig, it mysteriously switches from
> CONFIG_KERNEL_GZIP=y to CONFIG_KERNEL_LZO=y. With Kyle's patch it
> doesn't do this. I applied the patch now.
> 

Yes, sorry, I had meant to followup with a diff of the resulting
configs. I noticed without the patch, that menu config options, like the
i386 processor selection, or the compression type, were getting set
incorrectly despite being set in the input oldconfig correctly.

Thanks Michal and Sam for the review!
	Kyle

Re: [PATCH] kbuild: fix oldnoconfig to do the right thing

[Kyle McMartin]

On Wed, Sep 08, 2010 at 01:55:53PM +0200, Michal Marek wrote:
> I found that when running make oldnoconfig on the kconfig branch of
> kbuild.git (which is currently a 2.6.35 + kconfig changes) against a
> 2.6.35-rc1 x86_64 defconfig, it mysteriously switches from
> CONFIG_KERNEL_GZIP=y to CONFIG_KERNEL_LZO=y. With Kyle's patch it
> doesn't do this. I applied the patch now.
> 

Hi Michal,

Can we get this sent to Linus before 2.6.36? It would be awkward to
release 2.6.36 with oldnoconfig working the broken way, and then patch
it after release.

--Kyle

Re: [PATCH] kbuild: fix oldnoconfig to do the right thing

[Michal Marek]

On 8.10.2010 16:36, Kyle McMartin wrote:
> On Wed, Sep 08, 2010 at 01:55:53PM +0200, Michal Marek wrote:
>> I found that when running make oldnoconfig on the kconfig branch of
>> kbuild.git (which is currently a 2.6.35 + kconfig changes) against a
>> 2.6.35-rc1 x86_64 defconfig, it mysteriously switches from
>> CONFIG_KERNEL_GZIP=y to CONFIG_KERNEL_LZO=y. With Kyle's patch it
>> doesn't do this. I applied the patch now.
>>
> 
> Hi Michal,
> 
> Can we get this sent to Linus before 2.6.36? It would be awkward to
> release 2.6.36 with oldnoconfig working the broken way, and then patch
> it after release.

Oh, sorry, I didn't realize that this is a regression from 2.6.35 and
applied it to the kconfig branch, which is part of for-next. I'll try to
send a pull request with this patch and two other later today, but I
don't know if Linus is going to accept it.

Michal

Re: [PATCH] kbuild: fix oldnoconfig to do the right thing

[Kyle McMartin]

On Fri, Oct 08, 2010 at 04:54:38PM +0200, Michal Marek wrote:
> Oh, sorry, I didn't realize that this is a regression from 2.6.35 and
> applied it to the kconfig branch, which is part of for-next. I'll try to
> send a pull request with this patch and two other later today, but I
> don't know if Linus is going to accept it.
> 

It's not a regression from .35, just trying to avoid unexpected changes
to how parts of the build system work after release (which means it
could be perceived as a regression from .36 to .37)

IE: Someone begins to expect this behaviour in 2.6.36, and then
 the patch goes into 2.6.37-rc1, so 2.6.37 behaves differently.

regards, Kyle

Re: [PATCH] kbuild: fix oldnoconfig to do the right thing

[Michal Marek]

On 8.10.2010 17:12, Kyle McMartin wrote:
> On Fri, Oct 08, 2010 at 04:54:38PM +0200, Michal Marek wrote:
>> Oh, sorry, I didn't realize that this is a regression from 2.6.35 and
>> applied it to the kconfig branch, which is part of for-next. I'll try to
>> send a pull request with this patch and two other later today, but I
>> don't know if Linus is going to accept it.
>>
> 
> It's not a regression from .35, just trying to avoid unexpected changes

Yeah, I meant "bug introduced in 2.6.36-rc1".


> to how parts of the build system work after release (which means it
> could be perceived as a regression from .36 to .37)
> 
> IE: Someone begins to expect this behaviour in 2.6.36, and then
>  the patch goes into 2.6.37-rc1, so 2.6.37 behaves differently.

I see.

Michal