Skip to contentSkip to conversation

arm: Add armv6 and armv7 optimized swab functions

PATCH New Rob Herring 3 participants 7 messages Dec 15, 2010 – Dec 18, 2010 Arm (32-bit)

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 arm: Add armv6 and armv7 optimized swab functions

From: Rob Herring <[email protected]>

swab functions are heavily used by FDT code, so enable optimized assembly code for ARMv6 and later.

---
 arch/arm/include/asm/byteorder.h |   16 ++++++++++++++++
 1 files changed, 16 insertions(+), 0 deletions(-)
arch/arm/include/asm/byteorder.h +16 -0
index c3489f1..9df5844 100644
@@ -23,6 +23,22 @@
 #  define __SWAB_64_THRU_32__
 #endif
 
+#if defined(__ARM_ARCH_7A__) || defined(__ARM_ARCH_6__)
+static inline __u16 __attribute__((const)) ___arch_swab16(__u16 x)
+{
+	__asm__ ("rev16 %0, %1" : "=r" (x) : "r" (x));
+	return x;
+}
+#define __arch_swab16 ___arch_swab16
+
+static inline __u32 __attribute__((const)) ___arch_swab32(__u32 x)
+{
+	__asm__ ("rev %0, %1" : "=r" (x) : "r" (x));
+	return x;
+}
+#define __arch_swab32 ___arch_swab32
+#endif
+
 #ifdef __ARMEB__
 #include <linux/byteorder/big_endian.h>
 #else
Signature · 1 line
1.7.1
#1 lore raw

PATCH 1/1 arm: Add armv6 and armv7 optimized swab functions

Dear Rob Herring,

Quoted 9 lines · In message <[email protected]> you wrote:

From: Rob Herring <[email protected]>

swab functions are heavily used by FDT code, so enable optimized assembly code for ARMv6 and later.

---
 arch/arm/include/asm/byteorder.h |   16 ++++++++++++++++
 1 files changed, 16 insertions(+), 0 deletions(-)

Do you have any numbers if this changes gives any measurable improvement?

Best regards,

Wolfgang Denk

Signature · 4 lines
DENX Software Engineering GmbH,     MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de
A modem is a baudy house.
#2 lore raw

PATCH 1/1 arm: Add armv6 and armv7 optimized swab functions

Wolfgang,

Quoted 8 lines · Wolfgang Denk

Dear Rob Herring,

Quoted 5 lines · In message<[email protected]> you wrote:

From: Rob Herring<[email protected]>

swab functions are heavily used by FDT code, so enable optimized assembly code for ARMv6 and later.

Quoted 7 lines
Quoted 4 lines
---
  arch/arm/include/asm/byteorder.h |   16 ++++++++++++++++
  1 files changed, 16 insertions(+), 0 deletions(-)

Do you have any numbers if this changes gives any measurable improvement?

I have an instruction trace capture and see repeated calls to swab32 by the fdt code. It's an obvious low hanging fruit. The boot time for device tree vs. non-device tree is noticeably longer, but I don't have any formal measurements.

Rob

#3 lore raw

PATCH 1/1 arm: Add armv6 and armv7 optimized swab functions

In reply to Rob Herring
Quoted 15 lines · Rob Herring

From: Rob Herring <[email protected]>

swab functions are heavily used by FDT code, so enable optimized assembly code for ARMv6 and later.

---
 arch/arm/include/asm/byteorder.h |   16 ++++++++++++++++
 1 files changed, 16 insertions(+), 0 deletions(-)
arch/arm/include/asm/byteorder.h +0 -0
index c3489f1..9df5844 100644

arch/arm/include/asm/byteorder.h line 31

Quoted 9 lines
arch/arm/include/asm/byteorder.h +6 -0
@@ -23,6 +23,22 @@
 #  define __SWAB_64_THRU_32__
 #endif
 
+#if defined(__ARM_ARCH_7A__) || defined(__ARM_ARCH_6__)
+static inline __u16 __attribute__((const)) ___arch_swab16(__u16 x)
+{
+	__asm__ ("rev16 %0, %1" : "=r" (x) : "r" (x));
+	return x;
+}

Pay close attention to what gcc does with this as it is prone to add unnecessary masking of the low halfword. If the callers are well-behaved (argument having top halfword clear), making the parameter and return types here plain unsigned (or u32) gives better code.

Signature · 2 lines
M?ns Rullg?rd
mans at mansr.com
#4 lore raw

PATCH 1/1 arm: Add armv6 and armv7 optimized swab functions

Quoted 20 lines · M?ns Rullg?rd
Quoted 18 lines · Rob Herring

From: Rob Herring<[email protected]>

swab functions are heavily used by FDT code, so enable optimized assembly code for ARMv6 and later.

---
  arch/arm/include/asm/byteorder.h |   16 ++++++++++++++++
  1 files changed, 16 insertions(+), 0 deletions(-)
arch/arm/include/asm/byteorder.h +0 -0
index c3489f1..9df5844 100644
@@ -23,6 +23,22 @@
  #  define __SWAB_64_THRU_32__
  #endif
 
Quoted 12 lines

arch/arm/include/asm/byteorder.h line 31

Quoted 6 lines
arch/arm/include/asm/byteorder.h +6 -0
+#if defined(__ARM_ARCH_7A__) || defined(__ARM_ARCH_6__)
+static inline __u16 __attribute__((const)) ___arch_swab16(__u16 x)
+{
+	__asm__ ("rev16 %0, %1" : "=r" (x) : "r" (x));
+	return x;
+}

Pay close attention to what gcc does with this as it is prone to add unnecessary masking of the low halfword. If the callers are well-behaved (argument having top halfword clear), making the parameter and return types here plain unsigned (or u32) gives better code.

This straight from the Linux code and there are only a few users of swab16 (none in my build).

Rob

#5 lore raw

PATCH 1/1 arm: Add armv6 and armv7 optimized swab functions

Quoted 34 lines · Rob Herring
Quoted 20 lines · M?ns Rullg?rd
Quoted 18 lines · Rob Herring

From: Rob Herring<[email protected]>

swab functions are heavily used by FDT code, so enable optimized assembly code for ARMv6 and later.

---
  arch/arm/include/asm/byteorder.h |   16 ++++++++++++++++
  1 files changed, 16 insertions(+), 0 deletions(-)
arch/arm/include/asm/byteorder.h +0 -0
index c3489f1..9df5844 100644
@@ -23,6 +23,22 @@
  #  define __SWAB_64_THRU_32__
  #endif
 
Quoted 12 lines

arch/arm/include/asm/byteorder.h line 31

Quoted 6 lines
arch/arm/include/asm/byteorder.h +6 -0
+#if defined(__ARM_ARCH_7A__) || defined(__ARM_ARCH_6__)
+static inline __u16 __attribute__((const)) ___arch_swab16(__u16 x)
+{
+	__asm__ ("rev16 %0, %1" : "=r" (x) : "r" (x));
+	return x;
+}

Pay close attention to what gcc does with this as it is prone to add unnecessary masking of the low halfword. If the callers are well-behaved (argument having top halfword clear), making the parameter and return types here plain unsigned (or u32) gives better code.

Quoted 2 lines

This straight from the Linux code and there are only a few users of swab16 (none in my build).

Look at the generated code if you don't believe me.

Signature · 2 lines
M?ns Rullg?rd
mans at mansr.com
#6 lore raw

PATCH 1/1 arm: Add armv6 and armv7 optimized swab functions

In reply to Rob Herring

Dear Rob Herring,

Quoted 3 lines · In message <[email protected]> you wrote:

This straight from the Linux code and there are only a few users of swab16 (none in my build).

Given that we have no idea if this code really gives any measurable performance improvement, and that it appears to be dangerous as well, I tend to not include that as is.

Thanks.

Wolfgang Denk

Signature · 5 lines
DENX Software Engineering GmbH,     MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de
It is surely a great calamity for  a  human  being  to  have  no  ob-
sessions.                                                - Robert Bly