[PATCH] ARM: armv6 & armv7 no need update SCTLR for alignment

From: Wuqixuan
Date: Wed Apr 23 2014 - 21:34:36 EST


Now in the code, when interrupt and exception disturbs the user-space, always
update SCTLR with cr_alignment. In our ARMv7 Cortex A15 SOC, the statements
bracketed by CONFIG_ALIGNMENT_TRAP need about 100 cycles. it impacts our
high realtime interrupt requirement scenario. Considering in ARMv6 and ARMv7
versions, cr_alignment and cr_no_alignment are always same without A bit,
they are initialized by clearing A bit in alignment_init. So there is no
need to update SCTLR if interrupt and exception disturbs the user-space
in vector_swi function or alignment_trap macro, in the ARMv6 and ARMv7 chip.
The patch has already verified in our ARMv7 Cortex A15 SOC.
Signed-off-by: Wuqixuan <wuqixuan@xxxxxxxxxx>
---
arch/arm/kernel/entry-common.S | 4 ++--
arch/arm/kernel/entry-header.S | 2 +-
2 files changed, 3 insertions(+), 3 deletions(-)
diff --git a/arch/arm/kernel/entry-common.S b/arch/arm/kernel/entry-common.S
index a2dcafd..4b7686a 100644
--- a/arch/arm/kernel/entry-common.S
+++ b/arch/arm/kernel/entry-common.S
@@ -366,7 +366,7 @@ ENTRY(vector_swi)
#endif
zero_fp

-#ifdef CONFIG_ALIGNMENT_TRAP
+#if defined(CONFIG_ALIGNMENT_TRAP) && !(__LINUX_ARM_ARCH__ >= 6)
ldr ip, __cr_alignment
ldr ip, [ip]
mcr p15, 0, ip, c1, c0 @ update control register
@@ -490,7 +490,7 @@ __sys_trace_return:
b ret_slow_syscall

.align 5
-#ifdef CONFIG_ALIGNMENT_TRAP
+#if defined(CONFIG_ALIGNMENT_TRAP) && !(__LINUX_ARM_ARCH__ >= 6)
.type __cr_alignment, #object
__cr_alignment:
.word cr_alignment
diff --git a/arch/arm/kernel/entry-header.S b/arch/arm/kernel/entry-header.S
index 1420725..5eb9e59 100644
--- a/arch/arm/kernel/entry-header.S
+++ b/arch/arm/kernel/entry-header.S
@@ -38,7 +38,7 @@
.endm

.macro alignment_trap, rtemp
-#ifdef CONFIG_ALIGNMENT_TRAP
+#if defined(CONFIG_ALIGNMENT_TRAP) && !(__LINUX_ARM_ARCH__ >= 6)
ldr \rtemp, .LCcralign
ldr \rtemp, [\rtemp]
mcr p15, 0, \rtemp, c1, c0
--
1.7.6--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/