[PATCH] kmemcheck: Disable SLUB and SLAB debugging when kmemcheck is enabled

[PATCH] kmemcheck: Disable SLUB and SLAB debugging when kmemcheck is enabled

[Pekka Enberg]

From: Pekka Enberg <penberg@cs.helsinki.fi>

Enabling kmemcheck in Kconfig is a painful experience. Lets make it a
tad bit easier by inverting the dependencies to SLUB and SLAB debugging.
As kmemcheck is more powerful than slab debugging, disable the latter
functionality if an user enables kmemcheck.

Cc: Vegard Nossum <vegard.nossum@gmail.com>
Cc: Christoph Lameter <cl@linux-foundation.org>
Signed-off-by: Pekka Enberg <penberg@cs.helsinki.fi>
---
 arch/x86/Kconfig.debug |    2 +-
 init/Kconfig           |    2 +-
 lib/Kconfig.debug      |    2 +-
 3 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/arch/x86/Kconfig.debug b/arch/x86/Kconfig.debug
index de186c5..3d1d593 100644
--- a/arch/x86/Kconfig.debug
+++ b/arch/x86/Kconfig.debug
@@ -276,7 +276,7 @@ menuconfig KMEMCHECK
 	bool "kmemcheck: trap use of uninitialized memory"
 	depends on X86
 	depends on !X86_USE_3DNOW
-	depends on (SLUB && !SLUB_DEBUG_ON) || (SLAB && !DEBUG_SLAB)
+	depends on SLUB || SLAB
 	depends on !CC_OPTIMIZE_FOR_SIZE
 	depends on !DEBUG_PAGEALLOC
 	depends on !FUNCTION_TRACER
diff --git a/init/Kconfig b/init/Kconfig
index f068071..36617a4 100644
--- a/init/Kconfig
+++ b/init/Kconfig
@@ -890,7 +890,7 @@ config PCI_QUIRKS
 config SLUB_DEBUG
 	default y
 	bool "Enable SLUB debugging support" if EMBEDDED
-	depends on SLUB && SYSFS
+	depends on SLUB && SYSFS && !KMEMCHECK
 	help
 	  SLUB has extensive debug support features. Disabling these can
 	  result in significant savings in code size. This also disables
diff --git a/lib/Kconfig.debug b/lib/Kconfig.debug
index 29044f5..768e426 100644
--- a/lib/Kconfig.debug
+++ b/lib/Kconfig.debug
@@ -262,7 +262,7 @@ config DEBUG_OBJECTS_ENABLE_DEFAULT
 
 config DEBUG_SLAB
 	bool "Debug slab memory allocations"
-	depends on DEBUG_KERNEL && SLAB
+	depends on DEBUG_KERNEL && SLAB && !KMEMCHECK
 	help
 	  Say Y here to have the kernel do limited verification on memory
 	  allocation as well as poisoning memory on free to catch use of freed
-- 
1.5.4.3

Re: [PATCH] kmemcheck: Disable SLUB and SLAB debugging when kmemcheck is enabled

[Sitsofe Wheeler]

On Mon, Feb 23, 2009 at 11:35:12AM +0200, Pekka Enberg wrote:
> From: Pekka Enberg <penberg@cs.helsinki.fi>
> 
> Enabling kmemcheck in Kconfig is a painful experience. Lets make it a
> tad bit easier by inverting the dependencies to SLUB and SLAB debugging.
> As kmemcheck is more powerful than slab debugging, disable the latter
> functionality if an user enables kmemcheck.

You may also want to add something in the Documentation/kmemcheck.txt
saying that it conflicts with CONFIG_DEBUG_PAGEALLOC and
CONFIG_SLUB_DEBUG so that people realise why it is not presented as an
option...

-- 
Sitsofe | http://sucs.org/~sits/

Re: [PATCH] kmemcheck: Disable SLUB and SLAB debugging when kmemcheck is enabled

[Christoph Lameter]

On Mon, 23 Feb 2009, Pekka Enberg wrote:

> Enabling kmemcheck in Kconfig is a painful experience. Lets make it a
> tad bit easier by inverting the dependencies to SLUB and SLAB debugging.
> As kmemcheck is more powerful than slab debugging, disable the latter
> functionality if an user enables kmemcheck.

SLUB_DEBUG is fundamentally different from SLAB_DEBUG. It only includes
the code to be able to enable debugging at runtime. It does not enable
debugging.

Re: [PATCH] kmemcheck: Disable SLUB and SLAB debugging when kmemcheck is enabled

[Pekka Enberg]

On Mon, 2009-02-23 at 10:41 -0500, Christoph Lameter wrote:
> On Mon, 23 Feb 2009, Pekka Enberg wrote:
> 
> > Enabling kmemcheck in Kconfig is a painful experience. Lets make it a
> > tad bit easier by inverting the dependencies to SLUB and SLAB debugging.
> > As kmemcheck is more powerful than slab debugging, disable the latter
> > functionality if an user enables kmemcheck.
> 
> SLUB_DEBUG is fundamentally different from SLAB_DEBUG. It only includes
> the code to be able to enable debugging at runtime. It does not enable
> debugging.

Yes, I know that but we can't really let the user enable both, slub
debugging and kmemcheck debugging, at run-time. That's why I used
SLUB_DEBUG here.

Re: [PATCH] kmemcheck: Disable SLUB and SLAB debugging when kmemcheck is enabled

[Vegard Nossum]

2009/2/23 Pekka Enberg <penberg@cs.helsinki.fi>:
> On Mon, 2009-02-23 at 10:41 -0500, Christoph Lameter wrote:
>> On Mon, 23 Feb 2009, Pekka Enberg wrote:
>>
>> > Enabling kmemcheck in Kconfig is a painful experience. Lets make it a
>> > tad bit easier by inverting the dependencies to SLUB and SLAB debugging.
>> > As kmemcheck is more powerful than slab debugging, disable the latter
>> > functionality if an user enables kmemcheck.
>>
>> SLUB_DEBUG is fundamentally different from SLAB_DEBUG. It only includes
>> the code to be able to enable debugging at runtime. It does not enable
>> debugging.
>
> Yes, I know that but we can't really let the user enable both, slub
> debugging and kmemcheck debugging, at run-time. That's why I used
> SLUB_DEBUG here.

It is possible. Which is why it said SLUB_DEBUG_ON to begin with. This
is how it works:

If slub_debug=,xyz is specified on the command line, the SLUB
debugging will be used for the "xyz" cache, and kmemcheck will track
the rest (all non-slub-debugged caches). So if we only disable
SLUB_DEBUG_ON, then slub debugging and kmemcheck debugging can still
be mixed at boot-time, but kmemcheck will take precedence unless
slub_debug= is specified.

Can you rethink whether SLUB_DEBUG or SLUB_DEBUG_ON is the one we
really want to depend on !KMEMCHECK, and either submit a new patch or
convince me that this is sufficient?

Thanks!


Vegard

-- 
"The animistic metaphor of the bug that maliciously sneaked in while
the programmer was not looking is intellectually dishonest as it
disguises that the error is the programmer's own creation."
	-- E. W. Dijkstra, EWD1036

Re: [PATCH] kmemcheck: Disable SLUB and SLAB debugging when kmemcheck is enabled

[Christoph Lameter]

On Wed, 25 Feb 2009, Vegard Nossum wrote:

> Can you rethink whether SLUB_DEBUG or SLUB_DEBUG_ON is the one we
> really want to depend on !KMEMCHECK, and either submit a new patch or
> convince me that this is sufficient?

!SLUB_DEBUG_ON is the right thing to depend on.

Re: [PATCH] kmemcheck: Disable SLUB and SLAB debugging when kmemcheck is enabled

[Pekka Enberg]

On Wed, 25 Feb 2009, Vegard Nossum wrote:
> > Can you rethink whether SLUB_DEBUG or SLUB_DEBUG_ON is the one we
> > really want to depend on !KMEMCHECK, and either submit a new patch or
> > convince me that this is sufficient?

On Wed, 2009-02-25 at 16:23 -0500, Christoph Lameter wrote:
> !SLUB_DEBUG_ON is the right thing to depend on.

Here's a patch to do that.

			Pekka

>From bae2afd233c24d766e20e09bb92c67c0e5ea16e3 Mon Sep 17 00:00:00 2001
From: Pekka Enberg <penberg@cs.helsinki.fi>
Date: Mon, 23 Feb 2009 11:35:13 +0200
Subject: [PATCH] kmemcheck: Disable SLUB and SLAB debugging when kmemcheck is enabled

Enabling kmemcheck in Kconfig is a painful experience. Lets make it a
tad bit easier by inverting the dependencies to SLUB and SLAB debugging.
As kmemcheck is more powerful than slab debugging, disable the latter
functionality if an user enables kmemcheck.

Cc: Vegard Nossum <vegard.nossum@gmail.com>
Cc: Christoph Lameter <cl@linux-foundation.org>
Signed-off-by: Pekka Enberg <penberg@cs.helsinki.fi>
---
 arch/x86/Kconfig.debug |    2 +-
 lib/Kconfig.debug      |    4 ++--
 2 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/arch/x86/Kconfig.debug b/arch/x86/Kconfig.debug
index de186c5..3d1d593 100644
--- a/arch/x86/Kconfig.debug
+++ b/arch/x86/Kconfig.debug
@@ -276,7 +276,7 @@ menuconfig KMEMCHECK
 	bool "kmemcheck: trap use of uninitialized memory"
 	depends on X86
 	depends on !X86_USE_3DNOW
-	depends on (SLUB && !SLUB_DEBUG_ON) || (SLAB && !DEBUG_SLAB)
+	depends on SLUB || SLAB
 	depends on !CC_OPTIMIZE_FOR_SIZE
 	depends on !DEBUG_PAGEALLOC
 	depends on !FUNCTION_TRACER
diff --git a/lib/Kconfig.debug b/lib/Kconfig.debug
index 29044f5..e9a930c 100644
--- a/lib/Kconfig.debug
+++ b/lib/Kconfig.debug
@@ -262,7 +262,7 @@ config DEBUG_OBJECTS_ENABLE_DEFAULT
 
 config DEBUG_SLAB
 	bool "Debug slab memory allocations"
-	depends on DEBUG_KERNEL && SLAB
+	depends on DEBUG_KERNEL && SLAB && !KMEMCHECK
 	help
 	  Say Y here to have the kernel do limited verification on memory
 	  allocation as well as poisoning memory on free to catch use of freed
@@ -274,7 +274,7 @@ config DEBUG_SLAB_LEAK
 
 config SLUB_DEBUG_ON
 	bool "SLUB debugging on by default"
-	depends on SLUB && SLUB_DEBUG
+	depends on SLUB && SLUB_DEBUG && !KMEMCHECK
 	default n
 	help
 	  Boot with debugging on by default. SLUB boots by default with
-- 
1.5.4.3

Re: [PATCH] kmemcheck: Disable SLUB and SLAB debugging when kmemcheck is enabled

[Vegard Nossum]

2009/2/26 Pekka Enberg <penberg@cs.helsinki.fi>:
> From: Pekka Enberg <penberg@cs.helsinki.fi>
> Date: Mon, 23 Feb 2009 11:35:13 +0200
> Subject: [PATCH] kmemcheck: Disable SLUB and SLAB debugging when kmemcheck is enabled
>
> Enabling kmemcheck in Kconfig is a painful experience. Lets make it a
> tad bit easier by inverting the dependencies to SLUB and SLAB debugging.
> As kmemcheck is more powerful than slab debugging, disable the latter
> functionality if an user enables kmemcheck.
>
> Cc: Vegard Nossum <vegard.nossum@gmail.com>
> Cc: Christoph Lameter <cl@linux-foundation.org>
> Signed-off-by: Pekka Enberg <penberg@cs.helsinki.fi>

Thanks, applied!


Vegard

-- 
"The animistic metaphor of the bug that maliciously sneaked in while
the programmer was not looking is intellectually dishonest as it
disguises that the error is the programmer's own creation."
	-- E. W. Dijkstra, EWD1036