[PATCH v1 13/20] ARC: [plat-eznps] Use dedicated bitops/atomic/cmpxchg
Noam Camus
noamc at ezchip.com
Sat Oct 31 06:15:20 PDT 2015
From: Noam Camus <noamc at ezchip.com>
We need our own implementaions since we lack of LLSC.
Our extended ISA provided with optimized solution for all 32bit
operations we see in those three headers.
Signed-off-by: Noam Camus <noamc at ezchip.com>
---
arch/arc/include/asm/atomic.h | 69 +++++++++++++++++++++++++++++++
arch/arc/include/asm/bitops.h | 49 ++++++++++++++++++++++
arch/arc/include/asm/cmpxchg.h | 49 ++++++++++++++++++++++
arch/arc/plat-eznps/include/plat/ctop.h | 1 +
4 files changed, 168 insertions(+), 0 deletions(-)
diff --git a/arch/arc/include/asm/atomic.h b/arch/arc/include/asm/atomic.h
index c3ecda0..ca318c2 100644
--- a/arch/arc/include/asm/atomic.h
+++ b/arch/arc/include/asm/atomic.h
@@ -17,6 +17,74 @@
#include <asm/barrier.h>
#include <asm/smp.h>
+#ifdef CONFIG_ARC_PLAT_EZNPS
+static inline int atomic_read(const atomic_t *v)
+{
+ int temp;
+
+ __asm__ __volatile__(
+ " ld.di %0, [%1]"
+ : "=r"(temp)
+ : "r"(&v->counter)
+ : "memory");
+ return temp;
+}
+
+static inline void atomic_set(atomic_t *v, int i)
+{
+ __asm__ __volatile__(
+ " st.di %0,[%1]"
+ :
+ : "r"(i), "r"(&v->counter)
+ : "memory");
+}
+
+#define ATOMIC_OP(op, c_op, asm_op) \
+static inline void atomic_##op(int i, atomic_t *v) \
+{ \
+ __asm__ __volatile__( \
+ " mov r2, %0\n" \
+ " mov r3, %1\n" \
+ " .word %2\n" \
+ : \
+ : "r"(i), "r"(&v->counter), "i"(asm_op) \
+ : "r2", "r3", "memory"); \
+} \
+
+#define ATOMIC_OP_RETURN(op, c_op, asm_op) \
+static inline int atomic_##op##_return(int i, atomic_t *v) \
+{ \
+ unsigned int temp = i; \
+ \
+ __asm__ __volatile__( \
+ " mov r2, %0\n" \
+ " mov r3, %1\n" \
+ " .word %2\n" \
+ " mov %0, r2" \
+ : "+r"(temp) \
+ : "r"(&v->counter), "i"(asm_op) \
+ : "r2", "r3", "memory"); \
+ \
+ return v->counter; \
+}
+
+#define ATOMIC_OPS(op, c_op, asm_op) \
+ ATOMIC_OP(op, c_op, asm_op) \
+ ATOMIC_OP_RETURN(op, c_op, asm_op)
+
+ATOMIC_OPS(add, +=, CTOP_INST_AADD_DI_R2_R2_R3)
+#define atomic_sub(i, v) atomic_add(-(i), (v))
+#define atomic_sub_return(i, v) atomic_add_return(-(i), (v))
+
+ATOMIC_OP(and, &=, CTOP_INST_AAND_DI_R2_R2_R3)
+#define atomic_andnot(mask, v) atomic_and(~(mask), (v))
+ATOMIC_OP(or, |=, CTOP_INST_AOR_DI_R2_R2_R3)
+ATOMIC_OP(xor, ^=, CTOP_INST_AXOR_DI_R2_R2_R3)
+
+#undef ATOMIC_OPS
+#undef ATOMIC_OP_RETURN
+#undef ATOMIC_OP
+#else /* CONFIG_ARC_PLAT_EZNPS */
#define atomic_read(v) ((v)->counter)
#ifdef CONFIG_ARC_HAS_LLSC
@@ -186,6 +254,7 @@ ATOMIC_OP(xor, ^=, xor)
#undef SCOND_FAIL_RETRY_VAR_DEF
#undef SCOND_FAIL_RETRY_ASM
#undef SCOND_FAIL_RETRY_VARS
+#endif /* CONFIG_ARC_PLAT_EZNPS */
/**
* __atomic_add_unless - add unless the number is a given value
diff --git a/arch/arc/include/asm/bitops.h b/arch/arc/include/asm/bitops.h
index 57c1f33..54ecbe4 100644
--- a/arch/arc/include/asm/bitops.h
+++ b/arch/arc/include/asm/bitops.h
@@ -22,6 +22,48 @@
#include <asm/smp.h>
#endif
+#ifdef CONFIG_ARC_PLAT_EZNPS
+#define BIT_OP(op, c_op, asm_op) \
+static inline void op##_bit(unsigned long nr, volatile unsigned long *m)\
+{ \
+ m += nr >> 5; \
+ \
+ nr = (1UL << (nr & 0x1f)); \
+ if (asm_op == CTOP_INST_AAND_DI_R2_R2_R3) \
+ nr = ~nr; \
+ \
+ __asm__ __volatile__( \
+ " mov r2, %0\n" \
+ " mov r3, %1\n" \
+ " .word %2\n" \
+ : \
+ : "r"(nr), "r"(m), "i"(asm_op) \
+ : "r2", "r3", "memory"); \
+}
+
+#define TEST_N_BIT_OP(op, c_op, asm_op) \
+static inline int test_and_##op##_bit(unsigned long nr, volatile unsigned long *m)\
+{ \
+ unsigned long old; \
+ \
+ m += nr >> 5; \
+ \
+ old = (1UL << (nr & 0x1f)); \
+ if (asm_op == CTOP_INST_AAND_DI_R2_R2_R3) \
+ old = ~old; \
+ \
+ __asm__ __volatile__( \
+ " mov r2, %0\n" \
+ " mov r3, %1\n" \
+ " .word %2\n" \
+ " mov %0, r2" \
+ : "+r"(old) \
+ : "r"(m), "i"(asm_op) \
+ : "r2", "r3", "memory"); \
+ \
+ return (old & (1 << nr)) != 0; \
+}
+#else /* CONFIG_ARC_PLAT_EZNPS */
#if defined(CONFIG_ARC_HAS_LLSC)
/*
@@ -155,6 +197,7 @@ static inline int test_and_##op##_bit(unsigned long nr, volatile unsigned long *
}
#endif /* CONFIG_ARC_HAS_LLSC */
+#endif /* CONFIG_ARC_PLAT_EZNPS */
/***************************************
* Non atomic variants
@@ -196,9 +239,15 @@ static inline int __test_and_##op##_bit(unsigned long nr, volatile unsigned long
/* __test_and_set_bit(), __test_and_clear_bit(), __test_and_change_bit() */\
__TEST_N_BIT_OP(op, c_op, asm_op)
+#ifdef CONFIG_ARC_PLAT_EZNPS
+BIT_OPS(set, |, CTOP_INST_AOR_DI_R2_R2_R3)
+BIT_OPS(clear, & ~, CTOP_INST_AAND_DI_R2_R2_R3)
+BIT_OPS(change, ^, CTOP_INST_AXOR_DI_R2_R2_R3)
+#else
BIT_OPS(set, |, bset)
BIT_OPS(clear, & ~, bclr)
BIT_OPS(change, ^, bxor)
+#endif
/*
* This routine doesn't need to be atomic.
diff --git a/arch/arc/include/asm/cmpxchg.h b/arch/arc/include/asm/cmpxchg.h
index af7a2db..8578586 100644
--- a/arch/arc/include/asm/cmpxchg.h
+++ b/arch/arc/include/asm/cmpxchg.h
@@ -14,6 +14,53 @@
#include <asm/barrier.h>
#include <asm/smp.h>
+#ifdef CONFIG_ARC_PLAT_EZNPS
+static inline unsigned long
+__cmpxchg(volatile void *ptr, unsigned long expected, unsigned long new)
+{
+ write_aux_reg(CTOP_AUX_GPA1, expected);
+
+ __asm__ __volatile__(
+ " mov r2, %0\n"
+ " mov r3, %1\n"
+ " .word %2\n"
+ " mov %0, r2"
+ : "+r"(new)
+ : "r"(ptr), "i"(CTOP_INST_EXC_DI_R2_R2_R3)
+ : "r2", "r3", "memory");
+
+ return new;
+}
+
+#define cmpxchg(ptr, o, n) ((typeof(*(ptr)))__cmpxchg((ptr), \
+ (unsigned long)(o), (unsigned long)(n)))
+#define atomic_cmpxchg(v, o, n) ((int)cmpxchg(&((v)->counter), (o), (n)))
+
+static inline unsigned long __xchg(unsigned long val, volatile void *ptr,
+ int size)
+{
+ extern unsigned long __xchg_bad_pointer(void);
+
+ switch (size) {
+ case 4:
+ __asm__ __volatile__(
+ " mov r2, %0\n"
+ " mov r3, %1\n"
+ " .word %2\n"
+ " mov %0, r2\n"
+ : "+r"(val)
+ : "r"(ptr), "i"(CTOP_INST_XEX_DI_R2_R2_R3)
+ : "r2", "r3", "memory");
+
+ return val;
+ }
+ return __xchg_bad_pointer();
+}
+
+#define xchg(ptr, with) ((typeof(*(ptr)))__xchg((unsigned long)(with), (ptr), \
+ sizeof(*(ptr))))
+#define atomic_xchg(v, new) (xchg(&((v)->counter), new))
+#else /* CONFIG_ARC_PLAT_EZNPS */
#ifdef CONFIG_ARC_HAS_LLSC
static inline unsigned long
@@ -158,4 +205,6 @@ static inline unsigned long __xchg(unsigned long val, volatile void *ptr,
*/
#define atomic_xchg(v, new) (xchg(&((v)->counter), new))
+#endif /* CONFIG_ARC_PLAT_EZNPS */
+
#endif
diff --git a/arch/arc/plat-eznps/include/plat/ctop.h b/arch/arc/plat-eznps/include/plat/ctop.h
index b708f9f..655a860 100644
--- a/arch/arc/plat-eznps/include/plat/ctop.h
+++ b/arch/arc/plat-eznps/include/plat/ctop.h
@@ -34,6 +34,7 @@
#define AUX_REG_TSI1 (CTOP_AUX_BASE + 0x050)
#define CTOP_AUX_EFLAGS (CTOP_AUX_BASE + 0x080)
#define CTOP_AUX_IACK (CTOP_AUX_BASE + 0x088)
+#define CTOP_AUX_GPA1 (CTOP_AUX_BASE + 0x08C)
#define CTOP_AUX_UDMC (CTOP_AUX_BASE + 0x300)
/* EZchip core instructions */
--
1.7.1
More information about the linux-snps-arc
mailing list