2010-02-03 20:24:17

by Borislav Petkov

[permalink] [raw]
Subject: mtrr/state.c unused?

It looks like <arch/x86/kernel/cpu/mtrr/state.c> is not used by
anything in the kernel.

Besides cleanups, the only relevant patch in the git history is:

commit 2ec1df4130c60d1eb49dc0fa0ed15858fede6b05
Author: Thomas Gleixner <[email protected]>
Date: Thu Oct 11 11:16:28 2007 +0200

i386: move kernel/cpu/mtrr


which adds it. Hmm...

A

git log -i --pickaxe-all -Sset_mtrr_prepare_save

for example, shows that

commit 9a6b344ea967efa0bb5ca4cb5405f840652b66c4
Author: Harvey Harrison <[email protected]>
Date: Mon Feb 4 16:48:01 2008 +0100

x86: remove long dead cyrix mtrr code

has removed the last callers of those helpers and they're now unused.
They should go if we don't need them anymore.

--
Regards/Gruss,
Boris.

--
Advanced Micro Devices, Inc.
Operating Systems Research Center


2010-02-04 08:24:58

by Ingo Molnar

[permalink] [raw]
Subject: Re: mtrr/state.c unused?


* Borislav Petkov <[email protected]> wrote:

> It looks like <arch/x86/kernel/cpu/mtrr/state.c> is not used by
> anything in the kernel.

Good catch - i dont see it used either. Mind sending a patch?

>
> Besides cleanups, the only relevant patch in the git history is:
>
> commit 2ec1df4130c60d1eb49dc0fa0ed15858fede6b05
> Author: Thomas Gleixner <[email protected]>
> Date: Thu Oct 11 11:16:28 2007 +0200
>
> i386: move kernel/cpu/mtrr
>
>
> which adds it. Hmm...

Might be a dead code artifact from the original unification.

Thanks,

Ingo

2010-02-04 08:51:38

by Borislav Petkov

[permalink] [raw]
Subject: Re: mtrr/state.c unused?

On Thu, Feb 04, 2010 at 09:24:28AM +0100, Ingo Molnar wrote:
>
> * Borislav Petkov <[email protected]> wrote:
>
> > It looks like <arch/x86/kernel/cpu/mtrr/state.c> is not used by
> > anything in the kernel.
>
> Good catch - i dont see it used either. Mind sending a patch?

Yep, here we go:

--
>From 059e59699f15c824c35a85774c377ffcae6bffdf Mon Sep 17 00:00:00 2001
From: Borislav Petkov <[email protected]>
Date: Thu, 4 Feb 2010 09:36:03 +0100
Subject: [PATCH] x86, mtrr: Remove unused mtrr/state.c

The last reference to the helpers in <arch/x86/kernel/cpu/mtrr/state.c>
went away with 9a6b344ea967efa0bb5ca4cb5405f840652b66c4 leaving unused
code. Remove it.

Signed-off-by: Borislav Petkov <[email protected]>
---
arch/x86/kernel/cpu/mtrr/Makefile | 2 +-
arch/x86/kernel/cpu/mtrr/state.c | 94 -------------------------------------
2 files changed, 1 insertions(+), 95 deletions(-)
delete mode 100644 arch/x86/kernel/cpu/mtrr/state.c

diff --git a/arch/x86/kernel/cpu/mtrr/Makefile b/arch/x86/kernel/cpu/mtrr/Makefile
index f4361b5..ad9e5ed 100644
--- a/arch/x86/kernel/cpu/mtrr/Makefile
+++ b/arch/x86/kernel/cpu/mtrr/Makefile
@@ -1,3 +1,3 @@
-obj-y := main.o if.o generic.o state.o cleanup.o
+obj-y := main.o if.o generic.o cleanup.o
obj-$(CONFIG_X86_32) += amd.o cyrix.o centaur.o

diff --git a/arch/x86/kernel/cpu/mtrr/state.c b/arch/x86/kernel/cpu/mtrr/state.c
deleted file mode 100644
index dfc80b4..0000000
--- a/arch/x86/kernel/cpu/mtrr/state.c
+++ /dev/null
@@ -1,94 +0,0 @@
-#include <linux/init.h>
-#include <linux/io.h>
-#include <linux/mm.h>
-
-#include <asm/processor-cyrix.h>
-#include <asm/processor-flags.h>
-#include <asm/mtrr.h>
-#include <asm/msr.h>
-
-#include "mtrr.h"
-
-/* Put the processor into a state where MTRRs can be safely set */
-void set_mtrr_prepare_save(struct set_mtrr_context *ctxt)
-{
- unsigned int cr0;
-
- /* Disable interrupts locally */
- local_irq_save(ctxt->flags);
-
- if (use_intel() || is_cpu(CYRIX)) {
-
- /* Save value of CR4 and clear Page Global Enable (bit 7) */
- if (cpu_has_pge) {
- ctxt->cr4val = read_cr4();
- write_cr4(ctxt->cr4val & ~X86_CR4_PGE);
- }
-
- /*
- * Disable and flush caches. Note that wbinvd flushes the TLBs
- * as a side-effect
- */
- cr0 = read_cr0() | X86_CR0_CD;
- wbinvd();
- write_cr0(cr0);
- wbinvd();
-
- if (use_intel()) {
- /* Save MTRR state */
- rdmsr(MSR_MTRRdefType, ctxt->deftype_lo, ctxt->deftype_hi);
- } else {
- /*
- * Cyrix ARRs -
- * everything else were excluded at the top
- */
- ctxt->ccr3 = getCx86(CX86_CCR3);
- }
- }
-}
-
-void set_mtrr_cache_disable(struct set_mtrr_context *ctxt)
-{
- if (use_intel()) {
- /* Disable MTRRs, and set the default type to uncached */
- mtrr_wrmsr(MSR_MTRRdefType, ctxt->deftype_lo & 0xf300UL,
- ctxt->deftype_hi);
- } else {
- if (is_cpu(CYRIX)) {
- /* Cyrix ARRs - everything else were excluded at the top */
- setCx86(CX86_CCR3, (ctxt->ccr3 & 0x0f) | 0x10);
- }
- }
-}
-
-/* Restore the processor after a set_mtrr_prepare */
-void set_mtrr_done(struct set_mtrr_context *ctxt)
-{
- if (use_intel() || is_cpu(CYRIX)) {
-
- /* Flush caches and TLBs */
- wbinvd();
-
- /* Restore MTRRdefType */
- if (use_intel()) {
- /* Intel (P6) standard MTRRs */
- mtrr_wrmsr(MSR_MTRRdefType, ctxt->deftype_lo,
- ctxt->deftype_hi);
- } else {
- /*
- * Cyrix ARRs -
- * everything else was excluded at the top
- */
- setCx86(CX86_CCR3, ctxt->ccr3);
- }
-
- /* Enable caches */
- write_cr0(read_cr0() & 0xbfffffff);
-
- /* Restore value of CR4 */
- if (cpu_has_pge)
- write_cr4(ctxt->cr4val);
- }
- /* Re-enable interrupts locally (if enabled previously) */
- local_irq_restore(ctxt->flags);
-}
--
1.6.6

--
Regards/Gruss,
Boris.

2010-02-04 09:58:24

by Borislav Petkov

[permalink] [raw]
Subject: [tip:x86/cleanups] x86, mtrr: Remove unused mtrr/state.c

Commit-ID: 34d2819f20782feb60f9434470ecfb200875fd41
Gitweb: http://git.kernel.org/tip/34d2819f20782feb60f9434470ecfb200875fd41
Author: Borislav Petkov <[email protected]>
AuthorDate: Thu, 4 Feb 2010 09:51:28 +0100
Committer: Ingo Molnar <[email protected]>
CommitDate: Thu, 4 Feb 2010 10:01:38 +0100

x86, mtrr: Remove unused mtrr/state.c

The last reference to the helpers in
<arch/x86/kernel/cpu/mtrr/state.c> went away with
9a6b344ea967efa0bb5ca4cb5405f840652b66c4 leaving unused code.
Remove it.

Signed-off-by: Borislav Petkov <[email protected]>
LKML-Reference: <[email protected]>
Signed-off-by: Ingo Molnar <[email protected]>
---
arch/x86/kernel/cpu/mtrr/Makefile | 2 +-
arch/x86/kernel/cpu/mtrr/state.c | 94 -------------------------------------
2 files changed, 1 insertions(+), 95 deletions(-)

diff --git a/arch/x86/kernel/cpu/mtrr/Makefile b/arch/x86/kernel/cpu/mtrr/Makefile
index f4361b5..ad9e5ed 100644
--- a/arch/x86/kernel/cpu/mtrr/Makefile
+++ b/arch/x86/kernel/cpu/mtrr/Makefile
@@ -1,3 +1,3 @@
-obj-y := main.o if.o generic.o state.o cleanup.o
+obj-y := main.o if.o generic.o cleanup.o
obj-$(CONFIG_X86_32) += amd.o cyrix.o centaur.o

diff --git a/arch/x86/kernel/cpu/mtrr/state.c b/arch/x86/kernel/cpu/mtrr/state.c
deleted file mode 100644
index dfc80b4..0000000
--- a/arch/x86/kernel/cpu/mtrr/state.c
+++ /dev/null
@@ -1,94 +0,0 @@
-#include <linux/init.h>
-#include <linux/io.h>
-#include <linux/mm.h>
-
-#include <asm/processor-cyrix.h>
-#include <asm/processor-flags.h>
-#include <asm/mtrr.h>
-#include <asm/msr.h>
-
-#include "mtrr.h"
-
-/* Put the processor into a state where MTRRs can be safely set */
-void set_mtrr_prepare_save(struct set_mtrr_context *ctxt)
-{
- unsigned int cr0;
-
- /* Disable interrupts locally */
- local_irq_save(ctxt->flags);
-
- if (use_intel() || is_cpu(CYRIX)) {
-
- /* Save value of CR4 and clear Page Global Enable (bit 7) */
- if (cpu_has_pge) {
- ctxt->cr4val = read_cr4();
- write_cr4(ctxt->cr4val & ~X86_CR4_PGE);
- }
-
- /*
- * Disable and flush caches. Note that wbinvd flushes the TLBs
- * as a side-effect
- */
- cr0 = read_cr0() | X86_CR0_CD;
- wbinvd();
- write_cr0(cr0);
- wbinvd();
-
- if (use_intel()) {
- /* Save MTRR state */
- rdmsr(MSR_MTRRdefType, ctxt->deftype_lo, ctxt->deftype_hi);
- } else {
- /*
- * Cyrix ARRs -
- * everything else were excluded at the top
- */
- ctxt->ccr3 = getCx86(CX86_CCR3);
- }
- }
-}
-
-void set_mtrr_cache_disable(struct set_mtrr_context *ctxt)
-{
- if (use_intel()) {
- /* Disable MTRRs, and set the default type to uncached */
- mtrr_wrmsr(MSR_MTRRdefType, ctxt->deftype_lo & 0xf300UL,
- ctxt->deftype_hi);
- } else {
- if (is_cpu(CYRIX)) {
- /* Cyrix ARRs - everything else were excluded at the top */
- setCx86(CX86_CCR3, (ctxt->ccr3 & 0x0f) | 0x10);
- }
- }
-}
-
-/* Restore the processor after a set_mtrr_prepare */
-void set_mtrr_done(struct set_mtrr_context *ctxt)
-{
- if (use_intel() || is_cpu(CYRIX)) {
-
- /* Flush caches and TLBs */
- wbinvd();
-
- /* Restore MTRRdefType */
- if (use_intel()) {
- /* Intel (P6) standard MTRRs */
- mtrr_wrmsr(MSR_MTRRdefType, ctxt->deftype_lo,
- ctxt->deftype_hi);
- } else {
- /*
- * Cyrix ARRs -
- * everything else was excluded at the top
- */
- setCx86(CX86_CCR3, ctxt->ccr3);
- }
-
- /* Enable caches */
- write_cr0(read_cr0() & 0xbfffffff);
-
- /* Restore value of CR4 */
- if (cpu_has_pge)
- write_cr4(ctxt->cr4val);
- }
- /* Re-enable interrupts locally (if enabled previously) */
- local_irq_restore(ctxt->flags);
-}

2010-02-04 17:37:10

by Harvey Harrison

[permalink] [raw]
Subject: Re: mtrr/state.c unused?

On Thu, Feb 4, 2010 at 12:24 AM, Ingo Molnar <[email protected]> wrote:
>
> Might be a dead code artifact from the original unification.
>
> Thanks,
>
>        Ingo

I dimly recall there was another user at the time, not that I've
actually gone back and looked
however.

Cheers,

Harvey

2010-02-04 17:43:30

by H. Peter Anvin

[permalink] [raw]
Subject: Re: mtrr/state.c unused?

On 02/04/2010 09:29 AM, Harvey Harrison wrote:
>
> I dimly recall there was another user at the time, not that I've
> actually gone back and looked
> however.
>

Doesn't really matter, there isn't one now.

-hpa

--
H. Peter Anvin, Intel Open Source Technology Center
I work for Intel. I don't speak on their behalf.