[PATCH] [6/6] i386: Move CONFIG_DOUBLEFAULT into arch/i386 where it belongs.

[PATCH] [6/6] i386: Move CONFIG_DOUBLEFAULT into arch/i386 where it belongs.

[Andi Kleen]

Signed-off-by: Andi Kleen <ak@suse.de>

---
 arch/i386/Kconfig.debug |    9 +++++++++
 init/Kconfig            |    9 ---------
 2 files changed, 9 insertions(+), 9 deletions(-)

Index: linux/arch/i386/Kconfig.debug
===================================================================
--- linux.orig/arch/i386/Kconfig.debug
+++ linux/arch/i386/Kconfig.debug
@@ -81,4 +81,13 @@ config X86_MPPARSE
 	depends on X86_LOCAL_APIC && !X86_VISWS
 	default y
 
+config DOUBLEFAULT
+	default y
+	bool "Enable doublefault exception handler" if EMBEDDED
+	help
+          This option allows trapping of rare doublefault exceptions that
+          would otherwise cause a system to silently reboot. Disabling this
+          option saves about 4k and might cause you much additional grey
+          hair.
+
 endmenu
Index: linux/init/Kconfig
===================================================================
--- linux.orig/init/Kconfig
+++ linux/init/Kconfig
@@ -374,15 +374,6 @@ config SLAB
 	  SLOB is more space efficient but does not scale well and is
 	  more susceptible to fragmentation.
 
-config DOUBLEFAULT
-	default y
-	bool "Enable doublefault exception handler" if EMBEDDED && X86_32
-	help
-          This option allows trapping of rare doublefault exceptions that
-          would otherwise cause a system to silently reboot. Disabling this
-          option saves about 4k and might cause you much additional grey
-          hair.
-
 endmenu		# General setup
 
 config TINY_SHMEM

Re: [PATCH] [6/6] i386: Move CONFIG_DOUBLEFAULT into arch/i386 where it belongs.

[Adrian Bunk]

On Tue, Apr 18, 2006 at 12:35:22PM +0200, Andi Kleen wrote:
> 
> Signed-off-by: Andi Kleen <ak@suse.de>
>...

NAK.

When submitting a patch that is the revert of a patch that went 
into Linus' tree just 8 days ago [1], I'd expect at least:
- a Cc to the people involved with the patch you are reverting
- a note that you are reverting a recent patch in your patch
  description
- an explanation why you disagree with the patch you are reverting

If you disagree with a patch, please speak up when it's submitted or 
discuss it after you've seen it in the tree. But don't play such silly
revert-and-hope-they-don't-notice-I've-reverted-it games.

cu
Adrian

[1] commit e39632faa0efbddc3aed4f8658f2fa0a8afa2717

-- 

       "Is there not promise of rain?" Ling Tan asked suddenly out
        of the darkness. There had been need of rain for many days.
       "Only a promise," Lao Er said.
                                       Pearl S. Buck - Dragon Seed

Re: [discuss] Re: [PATCH] [6/6] i386: Move CONFIG_DOUBLEFAULT into arch/i386 where it belongs.

[Andi Kleen]

On Tuesday 18 April 2006 21:05, Adrian Bunk wrote:
> On Tue, Apr 18, 2006 at 12:35:22PM +0200, Andi Kleen wrote:
> > 
> > Signed-off-by: Andi Kleen <ak@suse.de>
> >...
> 
> NAK.
> 
> When submitting a patch that is the revert of a patch that went 
> into Linus' tree just 8 days ago [1], I'd expect at least:
> - a Cc to the people involved with the patch you are reverting
> - a note that you are reverting a recent patch in your patch
>   description
> - an explanation why you disagree with the patch you are reverting

The subject was very clear. i386 options belong into arch/i386.

> If you disagree with a patch, please speak up when it's submitted or 
> discuss it after you've seen it in the tree. But don't play such silly
> revert-and-hope-they-don't-notice-I've-reverted-it games.

I moved it because I noticed that my x86-64 configuration files 
had this strange new symbol. I also did a grep and no other architecture
other than i386 uses it.

i386 specific hacks belong into arch/i386

-Andi (who actually thinks the whole thing was always a bad idea - saving
a few K but giving up such debugging is a poor trade off)

Re: [discuss] Re: [PATCH] [6/6] i386: Move CONFIG_DOUBLEFAULT into arch/i386 where it belongs.

[Randy.Dunlap]

On Tue, 18 Apr 2006 22:12:13 +0200 Andi Kleen wrote:

> On Tuesday 18 April 2006 21:05, Adrian Bunk wrote:
> > On Tue, Apr 18, 2006 at 12:35:22PM +0200, Andi Kleen wrote:
> > > 
> > > Signed-off-by: Andi Kleen <ak@suse.de>
> > >...
> > 
> > NAK.
> >  submitting a patch that is the revert of a patch that went 
> > into Linus' tree just 8 days ago [1], I'd expect at least:
> > - a Cc to the people involved with the patch you are reverting
> > - a note that you are reverting a recent patch in your patch
> >   description
> > - an explanation why you disagree with the patch you are reverting
> 
> The subject was very clear. i386 options belong into arch/i386.

Yes, the timing could have been better.  Whatever.

I agree with Andi that it should be moved back to the ix86 Processor
menu, but not where he moved it to.  My patch is below.

> -Andi (who actually thinks the whole thing was always a bad idea - saving
> a few K but giving up such debugging is a poor trade off)

It does default to Y and can only be changed if EMBEDDED is enabled.
I don't think that it really should be in the Kernel hacking menu.

---

From: Randy Dunlap <rdunlap@xenotime.net>

Put DOUBLEFAULT option back in the i386 Processor menu, but
not at the very top of it like it was originally, where it
was out of place in the menu structure.

Signed-off-by: Randy Dunlap <rdunlap@xenotime.net>
---
 arch/i386/Kconfig |    9 +++++++++
 init/Kconfig      |    9 ---------
 2 files changed, 9 insertions(+), 9 deletions(-)

--- linux-2617-rc1g12.orig/arch/i386/Kconfig
+++ linux-2617-rc1g12/arch/i386/Kconfig
@@ -282,6 +282,15 @@ config X86_VISWS_APIC
 	depends on X86_VISWS
 	default y
 
+config DOUBLEFAULT
+	default y
+	bool "Enable doublefault exception handler" if EMBEDDED
+	help
+          This option allows trapping of rare doublefault exceptions that
+          would otherwise cause a system to silently reboot. Disabling this
+          option saves about 4k and might cause you much additional grey
+          hair.
+
 config X86_MCE
 	bool "Machine Check Exception"
 	depends on !X86_VOYAGER
--- linux-2617-rc1g12.orig/init/Kconfig
+++ linux-2617-rc1g12/init/Kconfig
@@ -374,15 +374,6 @@ config SLAB
 	  SLOB is more space efficient but does not scale well and is
 	  more susceptible to fragmentation.
 
-config DOUBLEFAULT
-	default y
-	bool "Enable doublefault exception handler" if EMBEDDED && X86_32
-	help
-          This option allows trapping of rare doublefault exceptions that
-          would otherwise cause a system to silently reboot. Disabling this
-          option saves about 4k and might cause you much additional grey
-          hair.
-
 endmenu		# General setup
 
 config TINY_SHMEM

---

Re: [discuss] Re: [PATCH] [6/6] i386: Move CONFIG_DOUBLEFAULT into arch/i386 where it belongs.

[Adrian Bunk]

On Tue, Apr 18, 2006 at 07:08:39PM -0700, Randy.Dunlap wrote:
> On Tue, 18 Apr 2006 22:12:13 +0200 Andi Kleen wrote:
> 
> > On Tuesday 18 April 2006 21:05, Adrian Bunk wrote:
> > > On Tue, Apr 18, 2006 at 12:35:22PM +0200, Andi Kleen wrote:
> > > > 
> > > > Signed-off-by: Andi Kleen <ak@suse.de>
> > > >...
> > > 
> > > NAK.
> > >  submitting a patch that is the revert of a patch that went 
> > > into Linus' tree just 8 days ago [1], I'd expect at least:
> > > - a Cc to the people involved with the patch you are reverting
> > > - a note that you are reverting a recent patch in your patch
> > >   description
> > > - an explanation why you disagree with the patch you are reverting
> > 
> > The subject was very clear. i386 options belong into arch/i386.
> 
> Yes, the timing could have been better.  Whatever.
> 
> I agree with Andi that it should be moved back to the ix86 Processor
> menu, but not where he moved it to.  My patch is below.
>...

I'd still disagree with Andi regarding this point (but it's not a very 
important issue).

My main problem with his patch is still the way he did it - sending a 
patch reverting a recently included patch with neither discussion before 
the patch nor mentioning in the patch that it's a revert nor a Cc to the 
people involved with the patch.

cu
Adrian

-- 

       "Is there not promise of rain?" Ling Tan asked suddenly out
        of the darkness. There had been need of rain for many days.
       "Only a promise," Lao Er said.
                                       Pearl S. Buck - Dragon Seed

Re: [discuss] Re: [PATCH] [6/6] i386: Move CONFIG_DOUBLEFAULT into arch/i386 where it belongs.

[Andi Kleen]

On Thursday 20 April 2006 13:49, Adrian Bunk wrote:

> My main problem with his patch is still the way he did it - sending a 
> patch reverting a recently included patch with neither discussion before 
> the patch nor mentioning in the patch that it's a revert nor a Cc to the 
> people involved with the patch.

I just noticed a problem (bogus symbols in my x86-64 config) and fixed it. I normally 
don't look at which patch it introduced for such trivial changes.

-Andi

Re: [discuss] Re: [PATCH] [6/6] i386: Move CONFIG_DOUBLEFAULT into arch/i386 where it belongs.

[Adrian Bunk]

On Thu, Apr 20, 2006 at 03:26:21PM +0200, Andi Kleen wrote:
> On Thursday 20 April 2006 13:49, Adrian Bunk wrote:
> 
> > My main problem with his patch is still the way he did it - sending a 
> > patch reverting a recently included patch with neither discussion before 
> > the patch nor mentioning in the patch that it's a revert nor a Cc to the 
> > people involved with the patch.
> 
> I just noticed a problem (bogus symbols in my x86-64 config) and fixed it. I normally 
> don't look at which patch it introduced for such trivial changes.

First of all, the "problem" of an unset config variable in the .config 
is at most a cosmetical issue.

And even if it had been a real problem, discussing such an issue and 
convincing people why you consider it to be bad is always better, since 
whoever gets conviced in the situation will have learned for similar 
future cases.

There are ways how things should work and we should follow them.

As an example, there is one trivial x86_64 specific patch removing the 
small bloat caused by an unused export I made in -mm Andrew has 
forwarded at least twice to you.

Until now, I'd have considered it an unfriendly act to forward such a 
patch to Linus through my trivial tree bypassing you as the maintainer, 
but looking at your statement it seems to be OK for you if I send such 
trivial patches directly to Linus without bothering to go through you as 
the maintainer.

> -Andi

cu
Adrian

-- 

       "Is there not promise of rain?" Ling Tan asked suddenly out
        of the darkness. There had been need of rain for many days.
       "Only a promise," Lao Er said.
                                       Pearl S. Buck - Dragon Seed