Skip to contentSkip to conversation

ARM: move #include <asm/cp15.h> into cacheflush.h

PATCH In review Brian Norris 3 participants 8 messages Sep 30, 2014

Subsystems (7)
lore
Share & tools
lore
#0 lore raw

cover ARM: move #include <asm/cp15.h> into cacheflush.h

There are several places where an explicit include of <asm/cp15.h> is needed just because cacheflush.h uses one of its macros in v7_exit_coherency_flush(). Let's put the include in the proper header.

These obviously have some dependencies, so I'd focus on:
(1) Is patch 1 acceptable? If so, then:
(2) Make sure patch 1 is serialized before the rest

I did a cursory scan of every file that includes cp15.h, and these are the only ones I found that only needed it for the sake of v7_exit_coherency_flush().

These are barely compile tested.

Thanks,
Brian

Brian Norris (5):
ARM: include missing cp15.h header
ARM: mvebu: drop unnecessary include
ARM: rockchip: drop unnecessary include
ARM: vexpress: drop unnecessary include
ARM: brcmstb: drop unnecessary include

 arch/arm/include/asm/cacheflush.h   | 1 +
 arch/arm/mach-bcm/platsmp-brcmstb.c | 1 -
 arch/arm/mach-mvebu/pmsu.c          | 1 -
 arch/arm/mach-rockchip/platsmp.c    | 1 -
 arch/arm/mach-vexpress/dcscb.c      | 1 -
 arch/arm/mach-vexpress/tc2_pm.c     | 1 -
 6 files changed, 1 insertion(+), 5 deletions(-)
Signature · 1 line
1.9.1
#1 lore raw

PATCH 1/5 ARM: include missing cp15.h header

As of this commit:

commit 39792c7cf3111d69dc4aa0923859d8b929e9039f
ARM: 7861/1: cacheflush: consolidate single-CPU ARMv7 cache disabling code

cacheflush.h now utilizes the CR_C macro from cp15.h. Let's include it.

This also means several users of v7_exit_coherency_flush() no longer have to include cp15.h directly when they don't directly use any #defines from it.

---
 arch/arm/include/asm/cacheflush.h | 1 +
 1 file changed, 1 insertion(+)
arch/arm/include/asm/cacheflush.h +1 -0
index 79ecb4f34ffb..866e229d3129 100644
@@ -16,6 +16,7 @@
 #include <asm/shmparam.h>
 #include <asm/cachetype.h>
 #include <asm/outercache.h>
+#include <asm/cp15.h>
 
 #define CACHE_COLOUR(vaddr)	((vaddr & (SHMLBA - 1)) >> PAGE_SHIFT)
 
Signature · 1 line
1.9.1
#2 lore raw

PATCH 2/5 ARM: mvebu: drop unnecessary include

In reply to Brian Norris

<asm/cp15.h> is only being included because of the implicit requirements of v7_exit_coherency_flush(). Now that the implicit include is provided for us, we can drop it from our explicit list.

Cc Jason Cooper <[email protected]>
Cc Andrew Lunn <[email protected]>
Cc Gregory Clement <[email protected]>
Cc Sebastian Hesselbarth <[email protected]>
---
 arch/arm/mach-mvebu/pmsu.c | 1 -
 1 file changed, 1 deletion(-)
arch/arm/mach-mvebu/pmsu.c +0 -1
index 8a70a51533fd..2d283cbc5cc7 100644
@@ -33,7 +33,6 @@
 #include <linux/slab.h>
 #include <linux/smp.h>
 #include <asm/cacheflush.h>
-#include <asm/cp15.h>
 #include <asm/smp_scu.h>
 #include <asm/smp_plat.h>
 #include <asm/suspend.h>
Signature · 1 line
1.9.1
#3 lore raw

PATCH 3/5 ARM: rockchip: drop unnecessary include

In reply to Brian Norris

<asm/cp15.h> is only being included because of the implicit requirements of v7_exit_coherency_flush(). Now that the implicit include is provided for us, we can drop it from our explicit list.

Cc Heiko Stuebner <[email protected]>
Cc linux-rockchip at lists.infradead.org
---
 arch/arm/mach-rockchip/platsmp.c | 1 -
 1 file changed, 1 deletion(-)
arch/arm/mach-rockchip/platsmp.c +0 -1
index 189684f55927..9b72e84e6494 100644
@@ -21,7 +21,6 @@
 #include <linux/of_address.h>
 
 #include <asm/cacheflush.h>
-#include <asm/cp15.h>
 #include <asm/smp_scu.h>
 #include <asm/smp_plat.h>
 #include <asm/mach/map.h>
Signature · 1 line
1.9.1
#4 lore raw

PATCH 4/5 ARM: vexpress: drop unnecessary include

In reply to Brian Norris

<asm/cp15.h> is only being included because of the implicit requirements of v7_exit_coherency_flush(). Now that the implicit include is provided for us, we can drop it from our explicit list.

---
 arch/arm/mach-vexpress/dcscb.c  | 1 -
 arch/arm/mach-vexpress/tc2_pm.c | 1 -
 2 files changed, 2 deletions(-)
arch/arm/mach-vexpress/dcscb.c +0 -1
index 30b993399ed7..28b0a33c91e1 100644
@@ -22,7 +22,6 @@
 #include <asm/proc-fns.h>
 #include <asm/cacheflush.h>
 #include <asm/cputype.h>
-#include <asm/cp15.h>
 
 
 #define RST_HOLD0	0x0
arch/arm/mach-vexpress/tc2_pm.c +0 -1
index 2fb78b4648cb..bc596f7d0d85 100644
@@ -26,7 +26,6 @@
 #include <asm/proc-fns.h>
 #include <asm/cacheflush.h>
 #include <asm/cputype.h>
-#include <asm/cp15.h>
 
 #include <linux/arm-cci.h>
 
Signature · 1 line
1.9.1
#5 lore raw

PATCH 5/5 ARM: brcmstb: drop unnecessary include

In reply to Brian Norris

<asm/cp15.h> is only being included because of the implicit requirements of v7_exit_coherency_flush(). Now that the implicit include is provided for us, we can drop it from our explicit list.

---
This is based on code queued for 3.18

 arch/arm/mach-bcm/platsmp-brcmstb.c | 1 -
 1 file changed, 1 deletion(-)
arch/arm/mach-bcm/platsmp-brcmstb.c +0 -1
index 31c87a284a34..3911ca7a0c3c 100644
@@ -25,7 +25,6 @@
 #include <linux/mfd/syscon.h>
 
 #include <asm/cacheflush.h>
-#include <asm/cp15.h>
 #include <asm/mach-types.h>
 #include <asm/smp_plat.h>
 
Signature · 1 line
1.9.1
#6 lore raw
In reply to Brian Norris
Quoted 6 lines · Brian Norris

There are several places where an explicit include of <asm/cp15.h> is needed just because cacheflush.h uses one of its macros in v7_exit_coherency_flush(). Let's put the include in the proper header.

These obviously have some dependencies, so I'd focus on:
(1) Is patch 1 acceptable? If so, then:

Well, I'm wondering if it is really a gain to force a dependency for cp15.h on every user of cacheflush.h just because of the seldomly used v7_exit_coherency_flush(). But I don't mind either ways.

Nicolas

#7 lore raw
Quoted 6 lines · Nicolas Pitre
Quoted 4 lines · Brian Norris

There are several places where an explicit include of <asm/cp15.h> is needed just because cacheflush.h uses one of its macros in v7_exit_coherency_flush(). Let's put the include in the proper header.

Quoted 6 lines
Quoted 2 lines

These obviously have some dependencies, so I'd focus on:
(1) Is patch 1 acceptable? If so, then:

Well, I'm wondering if it is really a gain to force a dependency for cp15.h on every user of cacheflush.h just because of the seldomly used v7_exit_coherency_flush(). But I don't mind either ways.

I don't like unnecessary dependencies. We have too many of them already, and it basically means that touching any header file triggers a full kernel rebuild.

If we go down the path of adding unnecessary dependencies, we might as well do away with make and just script the damn thing, and that'll probably be faster since there'll be no dependencies to track. Just throw everything through ccache and let it deal with anything which has changed. Just don't allow your disk to fill up during the build, ccache really doesn't like that.

Signature · 2 lines
FTTC broadband for 0.8mile line: currently at 9.5Mbps down 400kbps up
according to speedtest.net.