summaryrefslogtreecommitdiff
path: root/arch/arc
diff options
context:
space:
mode:
Diffstat (limited to 'arch/arc')
-rw-r--r--arch/arc/Kconfig18
-rw-r--r--arch/arc/Makefile13
-rw-r--r--arch/arc/boot/dts/axc003.dtsi15
-rw-r--r--arch/arc/boot/dts/axc003_idu.dtsi2
-rw-r--r--arch/arc/include/asm/Kbuild1
-rw-r--r--arch/arc/include/asm/arcregs.h8
-rw-r--r--arch/arc/include/asm/atomic.h86
-rw-r--r--arch/arc/include/asm/bitops.h35
-rw-r--r--arch/arc/include/asm/cache.h8
-rw-r--r--arch/arc/include/asm/cmpxchg.h22
-rw-r--r--arch/arc/include/asm/futex.h104
-rw-r--r--arch/arc/include/asm/mm-arch-hooks.h15
-rw-r--r--arch/arc/include/asm/perf_event.h23
-rw-r--r--arch/arc/include/asm/ptrace.h52
-rw-r--r--arch/arc/include/asm/spinlock.h538
-rw-r--r--arch/arc/include/asm/spinlock_types.h2
-rw-r--r--arch/arc/include/uapi/asm/ptrace.h20
-rw-r--r--arch/arc/kernel/entry-arcv2.S9
-rw-r--r--arch/arc/kernel/entry.S6
-rw-r--r--arch/arc/kernel/intc-arcv2.c1
-rw-r--r--arch/arc/kernel/intc-compact.c1
-rw-r--r--arch/arc/kernel/mcip.c26
-rw-r--r--arch/arc/kernel/perf_event.c275
-rw-r--r--arch/arc/kernel/process.c2
-rw-r--r--arch/arc/kernel/setup.c27
-rw-r--r--arch/arc/kernel/time.c40
-rw-r--r--arch/arc/kernel/troubleshoot.c1
-rw-r--r--arch/arc/kernel/unaligned.c6
-rw-r--r--arch/arc/lib/memcpy-archs.S2
-rw-r--r--arch/arc/lib/memset-archs.S43
-rw-r--r--arch/arc/mm/cache.c196
-rw-r--r--arch/arc/mm/dma.c22
-rw-r--r--arch/arc/plat-axs10x/axs10x.c19
33 files changed, 1323 insertions, 315 deletions
diff --git a/arch/arc/Kconfig b/arch/arc/Kconfig
index e7cee0a5c56d..78c0621d5819 100644
--- a/arch/arc/Kconfig
+++ b/arch/arc/Kconfig
@@ -8,6 +8,7 @@
config ARC
def_bool y
+ select ARCH_SUPPORTS_ATOMIC_RMW if ARC_HAS_LLSC
select BUILDTIME_EXTABLE_SORT
select COMMON_CLK
select CLONE_BACKWARDS
@@ -22,6 +23,7 @@ config ARC
select GENERIC_SMP_IDLE_THREAD
select HAVE_ARCH_KGDB
select HAVE_ARCH_TRACEHOOK
+ select HAVE_FUTEX_CMPXCHG
select HAVE_IOREMAP_PROT
select HAVE_KPROBES
select HAVE_KRETPROBES
@@ -115,6 +117,7 @@ if ISA_ARCOMPACT
config ARC_CPU_750D
bool "ARC750D"
+ select ARC_CANT_LLSC
help
Support for ARC750 core
@@ -312,11 +315,11 @@ config ARC_PAGE_SIZE_8K
config ARC_PAGE_SIZE_16K
bool "16KB"
- depends on ARC_MMU_V3
+ depends on ARC_MMU_V3 || ARC_MMU_V4
config ARC_PAGE_SIZE_4K
bool "4KB"
- depends on ARC_MMU_V3
+ depends on ARC_MMU_V3 || ARC_MMU_V4
endchoice
@@ -362,7 +365,12 @@ config ARC_CANT_LLSC
config ARC_HAS_LLSC
bool "Insn: LLOCK/SCOND (efficient atomic ops)"
default y
- depends on !ARC_CPU_750D && !ARC_CANT_LLSC
+ depends on !ARC_CANT_LLSC
+
+config ARC_STAR_9000923308
+ bool "Workaround for llock/scond livelock"
+ default y
+ depends on ISA_ARCV2 && SMP && ARC_HAS_LLSC
config ARC_HAS_SWAPE
bool "Insn: SWAPE (endian-swap)"
@@ -378,6 +386,10 @@ config ARC_HAS_LL64
dest operands with 2 possible source operands.
default y
+config ARC_HAS_DIV_REM
+ bool "Insn: div, divu, rem, remu"
+ default y
+
config ARC_HAS_RTC
bool "Local 64-bit r/o cycle counter"
default n
diff --git a/arch/arc/Makefile b/arch/arc/Makefile
index 6107062c0111..8a27a48304a4 100644
--- a/arch/arc/Makefile
+++ b/arch/arc/Makefile
@@ -36,8 +36,16 @@ cflags-$(atleast_gcc44) += -fsection-anchors
cflags-$(CONFIG_ARC_HAS_LLSC) += -mlock
cflags-$(CONFIG_ARC_HAS_SWAPE) += -mswape
+ifdef CONFIG_ISA_ARCV2
+
ifndef CONFIG_ARC_HAS_LL64
-cflags-$(CONFIG_ISA_ARCV2) += -mno-ll64
+cflags-y += -mno-ll64
+endif
+
+ifndef CONFIG_ARC_HAS_DIV_REM
+cflags-y += -mno-div-rem
+endif
+
endif
cflags-$(CONFIG_ARC_DW2_UNWIND) += -fasynchronous-unwind-tables
@@ -49,7 +57,8 @@ endif
ifndef CONFIG_CC_OPTIMIZE_FOR_SIZE
# Generic build system uses -O2, we want -O3
-cflags-y += -O3
+# Note: No need to add to cflags-y as that happens anyways
+ARCH_CFLAGS += -O3
endif
# small data is default for elf32 tool-chain. If not usable, disable it
diff --git a/arch/arc/boot/dts/axc003.dtsi b/arch/arc/boot/dts/axc003.dtsi
index 15c8d6226c9d..846481f37eef 100644
--- a/arch/arc/boot/dts/axc003.dtsi
+++ b/arch/arc/boot/dts/axc003.dtsi
@@ -12,7 +12,7 @@
/ {
compatible = "snps,arc";
- clock-frequency = <75000000>;
+ clock-frequency = <90000000>;
#address-cells = <1>;
#size-cells = <1>;
@@ -72,12 +72,13 @@
};
/*
- * This INTC is actually connected to DW APB GPIO
- * which acts as a wire between MB INTC and CPU INTC.
- * GPIO INTC is configured in platform init code
- * and here we mimic direct connection from MB INTC to
- * CPU INTC, thus we set "interrupts = <7>" instead of
- * "interrupts = <12>"
+ * The DW APB ICTL intc on MB is connected to CPU intc via a
+ * DT "invisible" DW APB GPIO block, configured to simply pass thru
+ * interrupts - setup accordinly in platform init (plat-axs10x/ax10x.c)
+ *
+ * So here we mimic a direct connection betwen them, ignoring the
+ * ABPG GPIO. Thus set "interrupts = <24>" (DW APB GPIO to core)
+ * instead of "interrupts = <12>" (DW APB ICTL to DW APB GPIO)
*
* This intc actually resides on MB, but we move it here to
* avoid duplicating the MB dtsi file given that IRQ from
diff --git a/arch/arc/boot/dts/axc003_idu.dtsi b/arch/arc/boot/dts/axc003_idu.dtsi
index 199d42820eca..2f0b33257db2 100644
--- a/arch/arc/boot/dts/axc003_idu.dtsi
+++ b/arch/arc/boot/dts/axc003_idu.dtsi
@@ -12,7 +12,7 @@
/ {
compatible = "snps,arc";
- clock-frequency = <75000000>;
+ clock-frequency = <90000000>;
#address-cells = <1>;
#size-cells = <1>;
diff --git a/arch/arc/include/asm/Kbuild b/arch/arc/include/asm/Kbuild
index c8f07ce1c757..0b10ef2a4372 100644
--- a/arch/arc/include/asm/Kbuild
+++ b/arch/arc/include/asm/Kbuild
@@ -22,6 +22,7 @@ generic-y += kvm_para.h
generic-y += local.h
generic-y += local64.h
generic-y += mcs_spinlock.h
+generic-y += mm-arch-hooks.h
generic-y += mman.h
generic-y += msgbuf.h
generic-y += param.h
diff --git a/arch/arc/include/asm/arcregs.h b/arch/arc/include/asm/arcregs.h
index 070f58827a5c..d8023bc8d1ad 100644
--- a/arch/arc/include/asm/arcregs.h
+++ b/arch/arc/include/asm/arcregs.h
@@ -35,6 +35,7 @@
#define ARC_REG_RTT_BCR 0xF2
#define ARC_REG_IRQ_BCR 0xF3
#define ARC_REG_SMART_BCR 0xFF
+#define ARC_REG_CLUSTER_BCR 0xcf
/* status32 Bits Positions */
#define STATUS_AE_BIT 5 /* Exception active */
@@ -89,11 +90,10 @@
#define ECR_C_BIT_DTLB_LD_MISS 8
#define ECR_C_BIT_DTLB_ST_MISS 9
-
/* Auxiliary registers */
#define AUX_IDENTITY 4
#define AUX_INTR_VEC_BASE 0x25
-
+#define AUX_NON_VOL 0x5e
/*
* Floating Pt Registers
@@ -240,9 +240,9 @@ struct bcr_extn_xymem {
struct bcr_perip {
#ifdef CONFIG_CPU_BIG_ENDIAN
- unsigned int start:8, pad2:8, sz:8, pad:8;
+ unsigned int start:8, pad2:8, sz:8, ver:8;
#else
- unsigned int pad:8, sz:8, pad2:8, start:8;
+ unsigned int ver:8, sz:8, pad2:8, start:8;
#endif
};
diff --git a/arch/arc/include/asm/atomic.h b/arch/arc/include/asm/atomic.h
index 03484cb4d16d..c3ecda023e3a 100644
--- a/arch/arc/include/asm/atomic.h
+++ b/arch/arc/include/asm/atomic.h
@@ -23,33 +23,60 @@
#define atomic_set(v, i) (((v)->counter) = (i))
-#ifdef CONFIG_ISA_ARCV2
-#define PREFETCHW " prefetchw [%1] \n"
-#else
-#define PREFETCHW
+#ifdef CONFIG_ARC_STAR_9000923308
+
+#define SCOND_FAIL_RETRY_VAR_DEF \
+ unsigned int delay = 1, tmp; \
+
+#define SCOND_FAIL_RETRY_ASM \
+ " bz 4f \n" \
+ " ; --- scond fail delay --- \n" \
+ " mov %[tmp], %[delay] \n" /* tmp = delay */ \
+ "2: brne.d %[tmp], 0, 2b \n" /* while (tmp != 0) */ \
+ " sub %[tmp], %[tmp], 1 \n" /* tmp-- */ \
+ " rol %[delay], %[delay] \n" /* delay *= 2 */ \
+ " b 1b \n" /* start over */ \
+ "4: ; --- success --- \n" \
+
+#define SCOND_FAIL_RETRY_VARS \
+ ,[delay] "+&r" (delay),[tmp] "=&r" (tmp) \
+
+#else /* !CONFIG_ARC_STAR_9000923308 */
+
+#define SCOND_FAIL_RETRY_VAR_DEF
+
+#define SCOND_FAIL_RETRY_ASM \
+ " bnz 1b \n" \
+
+#define SCOND_FAIL_RETRY_VARS
+
#endif
#define ATOMIC_OP(op, c_op, asm_op) \
static inline void atomic_##op(int i, atomic_t *v) \
{ \
- unsigned int temp; \
+ unsigned int val; \
+ SCOND_FAIL_RETRY_VAR_DEF \
\
__asm__ __volatile__( \
- "1: \n" \
- PREFETCHW \
- " llock %0, [%1] \n" \
- " " #asm_op " %0, %0, %2 \n" \
- " scond %0, [%1] \n" \
- " bnz 1b \n" \
- : "=&r"(temp) /* Early clobber, to prevent reg reuse */ \
- : "r"(&v->counter), "ir"(i) \
+ "1: llock %[val], [%[ctr]] \n" \
+ " " #asm_op " %[val], %[val], %[i] \n" \
+ " scond %[val], [%[ctr]] \n" \
+ " \n" \
+ SCOND_FAIL_RETRY_ASM \
+ \
+ : [val] "=&r" (val) /* Early clobber to prevent reg reuse */ \
+ SCOND_FAIL_RETRY_VARS \
+ : [ctr] "r" (&v->counter), /* Not "m": llock only supports reg direct addr mode */ \
+ [i] "ir" (i) \
: "cc"); \
} \
#define ATOMIC_OP_RETURN(op, c_op, asm_op) \
static inline int atomic_##op##_return(int i, atomic_t *v) \
{ \
- unsigned int temp; \
+ unsigned int val; \
+ SCOND_FAIL_RETRY_VAR_DEF \
\
/* \
* Explicit full memory barrier needed before/after as \
@@ -58,19 +85,21 @@ static inline int atomic_##op##_return(int i, atomic_t *v) \
smp_mb(); \
\
__asm__ __volatile__( \
- "1: \n" \
- PREFETCHW \
- " llock %0, [%1] \n" \
- " " #asm_op " %0, %0, %2 \n" \
- " scond %0, [%1] \n" \
- " bnz 1b \n" \
- : "=&r"(temp) \
- : "r"(&v->counter), "ir"(i) \
+ "1: llock %[val], [%[ctr]] \n" \
+ " " #asm_op " %[val], %[val], %[i] \n" \
+ " scond %[val], [%[ctr]] \n" \
+ " \n" \
+ SCOND_FAIL_RETRY_ASM \
+ \
+ : [val] "=&r" (val) \
+ SCOND_FAIL_RETRY_VARS \
+ : [ctr] "r" (&v->counter), \
+ [i] "ir" (i) \
: "cc"); \
\
smp_mb(); \
\
- return temp; \
+ return val; \
}
#else /* !CONFIG_ARC_HAS_LLSC */
@@ -143,13 +172,20 @@ static inline int atomic_##op##_return(int i, atomic_t *v) \
ATOMIC_OPS(add, +=, add)
ATOMIC_OPS(sub, -=, sub)
-ATOMIC_OP(and, &=, and)
-#define atomic_clear_mask(mask, v) atomic_and(~(mask), (v))
+#define atomic_andnot atomic_andnot
+
+ATOMIC_OP(and, &=, and)
+ATOMIC_OP(andnot, &= ~, bic)
+ATOMIC_OP(or, |=, or)
+ATOMIC_OP(xor, ^=, xor)
#undef ATOMIC_OPS
#undef ATOMIC_OP_RETURN
#undef ATOMIC_OP
+#undef SCOND_FAIL_RETRY_VAR_DEF
+#undef SCOND_FAIL_RETRY_ASM
+#undef SCOND_FAIL_RETRY_VARS
/**
* __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 99fe118d3730..57c1f33844d4 100644
--- a/arch/arc/include/asm/bitops.h
+++ b/arch/arc/include/asm/bitops.h
@@ -50,8 +50,7 @@ static inline void op##_bit(unsigned long nr, volatile unsigned long *m)\
* done for const @nr, but no code is generated due to gcc \
* const prop. \
*/ \
- if (__builtin_constant_p(nr)) \
- nr &= 0x1f; \
+ nr &= 0x1f; \
\
__asm__ __volatile__( \
"1: llock %0, [%1] \n" \
@@ -82,8 +81,7 @@ static inline int test_and_##op##_bit(unsigned long nr, volatile unsigned long *
\
m += nr >> 5; \
\
- if (__builtin_constant_p(nr)) \
- nr &= 0x1f; \
+ nr &= 0x1f; \
\
/* \
* Explicit full memory barrier needed before/after as \
@@ -129,16 +127,13 @@ static inline void op##_bit(unsigned long nr, volatile unsigned long *m)\
unsigned long temp, flags; \
m += nr >> 5; \
\
- if (__builtin_constant_p(nr)) \
- nr &= 0x1f; \
- \
/* \
* spin lock/unlock provide the needed smp_mb() before/after \
*/ \
bitops_lock(flags); \
\
temp = *m; \
- *m = temp c_op (1UL << nr); \
+ *m = temp c_op (1UL << (nr & 0x1f)); \
\
bitops_unlock(flags); \
}
@@ -149,17 +144,14 @@ static inline int test_and_##op##_bit(unsigned long nr, volatile unsigned long *
unsigned long old, flags; \
m += nr >> 5; \
\
- if (__builtin_constant_p(nr)) \
- nr &= 0x1f; \
- \
bitops_lock(flags); \
\
old = *m; \
- *m = old c_op (1 << nr); \
+ *m = old c_op (1UL << (nr & 0x1f)); \
\
bitops_unlock(flags); \
\
- return (old & (1 << nr)) != 0; \
+ return (old & (1UL << (nr & 0x1f))) != 0; \
}
#endif /* CONFIG_ARC_HAS_LLSC */
@@ -174,11 +166,8 @@ static inline void __##op##_bit(unsigned long nr, volatile unsigned long *m) \
unsigned long temp; \
m += nr >> 5; \
\
- if (__builtin_constant_p(nr)) \
- nr &= 0x1f; \
- \
temp = *m; \
- *m = temp c_op (1UL << nr); \
+ *m = temp c_op (1UL << (nr & 0x1f)); \
}
#define __TEST_N_BIT_OP(op, c_op, asm_op) \
@@ -187,13 +176,10 @@ static inline int __test_and_##op##_bit(unsigned long nr, volatile unsigned long
unsigned long old; \
m += nr >> 5; \
\
- if (__builtin_constant_p(nr)) \
- nr &= 0x1f; \
- \
old = *m; \
- *m = old c_op (1 << nr); \
+ *m = old c_op (1UL << (nr & 0x1f)); \
\
- return (old & (1 << nr)) != 0; \
+ return (old & (1UL << (nr & 0x1f))) != 0; \
}
#define BIT_OPS(op, c_op, asm_op) \
@@ -224,10 +210,7 @@ test_bit(unsigned int nr, const volatile unsigned long *addr)
addr += nr >> 5;
- if (__builtin_constant_p(nr))
- nr &= 0x1f;
-
- mask = 1 << nr;
+ mask = 1UL << (nr & 0x1f);
return ((mask & *addr) != 0);
}
diff --git a/arch/arc/include/asm/cache.h b/arch/arc/include/asm/cache.h
index d67345d3e2d4..e23ea6e7633a 100644
--- a/arch/arc/include/asm/cache.h
+++ b/arch/arc/include/asm/cache.h
@@ -53,6 +53,8 @@ extern void arc_cache_init(void);
extern char *arc_cache_mumbojumbo(int cpu_id, char *buf, int len);
extern void read_decode_cache_bcr(void);
+extern int ioc_exists;
+
#endif /* !__ASSEMBLY__ */
/* Instruction cache related Auxiliary registers */
@@ -94,4 +96,10 @@ extern void read_decode_cache_bcr(void);
#define SLC_CTRL_BUSY 0x100
#define SLC_CTRL_RGN_OP_INV 0x200
+/* IO coherency related Auxiliary registers */
+#define ARC_REG_IO_COH_ENABLE 0x500
+#define ARC_REG_IO_COH_PARTIAL 0x501
+#define ARC_REG_IO_COH_AP0_BASE 0x508
+#define ARC_REG_IO_COH_AP0_SIZE 0x509
+
#endif /* _ASM_CACHE_H */
diff --git a/arch/arc/include/asm/cmpxchg.h b/arch/arc/include/asm/cmpxchg.h
index 44fd531f4d7b..af7a2db139c9 100644
--- a/arch/arc/include/asm/cmpxchg.h
+++ b/arch/arc/include/asm/cmpxchg.h
@@ -110,18 +110,18 @@ static inline unsigned long __xchg(unsigned long val, volatile void *ptr,
sizeof(*(ptr))))
/*
- * On ARC700, EX insn is inherently atomic, so by default "vanilla" xchg() need
- * not require any locking. However there's a quirk.
- * ARC lacks native CMPXCHG, thus emulated (see above), using external locking -
- * incidently it "reuses" the same atomic_ops_lock used by atomic APIs.
- * Now, llist code uses cmpxchg() and xchg() on same data, so xchg() needs to
- * abide by same serializing rules, thus ends up using atomic_ops_lock as well.
+ * xchg() maps directly to ARC EX instruction which guarantees atomicity.
+ * However in !LLSC config, it also needs to be use @atomic_ops_lock spinlock
+ * due to a subtle reason:
+ * - For !LLSC, cmpxchg() needs to use that lock (see above) and there is lot
+ * of kernel code which calls xchg()/cmpxchg() on same data (see llist.h)
+ * Hence xchg() needs to follow same locking rules.
*
- * This however is only relevant if SMP and/or ARC lacks LLSC
- * if (UP or LLSC)
- * xchg doesn't need serialization
- * else <==> !(UP or LLSC) <==> (!UP and !LLSC) <==> (SMP and !LLSC)
- * xchg needs serialization
+ * Technically the lock is also needed for UP (boils down to irq save/restore)
+ * but we can cheat a bit since cmpxchg() atomic_ops_lock() would cause irqs to
+ * be disabled thus can't possibly be interrpted/preempted/clobbered by xchg()
+ * Other way around, xchg is one instruction anyways, so can't be interrupted
+ * as such
*/
#if !defined(CONFIG_ARC_HAS_LLSC) && defined(CONFIG_SMP)
diff --git a/arch/arc/include/asm/futex.h b/arch/arc/include/asm/futex.h
index 05b5aaf5b0f9..11e1b1f3acda 100644
--- a/arch/arc/include/asm/futex.h
+++ b/arch/arc/include/asm/futex.h
@@ -16,18 +16,22 @@
#include <linux/uaccess.h>
#include <asm/errno.h>
+#ifdef CONFIG_ARC_HAS_LLSC
+
#define __futex_atomic_op(insn, ret, oldval, uaddr, oparg)\
\
+ smp_mb(); \
__asm__ __volatile__( \
- "1: ld %1, [%2] \n" \
+ "1: llock %1, [%2] \n" \
insn "\n" \
- "2: st %0, [%2] \n" \
+ "2: scond %0, [%2] \n" \
+ " bnz 1b \n" \
" mov %0, 0 \n" \
"3: \n" \
" .section .fixup,\"ax\" \n" \
" .align 4 \n" \
"4: mov %0, %4 \n" \
- " b 3b \n" \
+ " j 3b \n" \
" .previous \n" \
" .section __ex_table,\"a\" \n" \
" .align 4 \n" \
@@ -37,7 +41,37 @@
\
: "=&r" (ret), "=&r" (oldval) \
: "r" (uaddr), "r" (oparg), "ir" (-EFAULT) \
- : "cc", "memory")
+ : "cc", "memory"); \
+ smp_mb() \
+
+#else /* !CONFIG_ARC_HAS_LLSC */
+
+#define __futex_atomic_op(insn, ret, oldval, uaddr, oparg)\
+ \
+ smp_mb(); \
+ __asm__ __volatile__( \
+ "1: ld %1, [%2] \n" \
+ insn "\n" \
+ "2: st %0, [%2] \n" \
+ " mov %0, 0 \n" \
+ "3: \n" \
+ " .section .fixup,\"ax\" \n" \
+ " .align 4 \n" \
+ "4: mov %0, %4 \n" \
+ " j 3b \n" \
+ " .previous \n" \
+ " .section __ex_table,\"a\" \n" \
+ " .align 4 \n" \
+ " .word 1b, 4b \n" \
+ " .word 2b, 4b \n" \
+ " .previous \n" \
+ \
+ : "=&r" (ret), "=&r" (oldval) \
+ : "r" (uaddr), "r" (oparg), "ir" (-EFAULT) \
+ : "cc", "memory"); \
+ smp_mb() \
+
+#endif
static inline int futex_atomic_op_inuser(int encoded_op, u32 __user *uaddr)
{
@@ -53,6 +87,9 @@ static inline int futex_atomic_op_inuser(int encoded_op, u32 __user *uaddr)
if (!access_ok(VERIFY_WRITE, uaddr, sizeof(int)))
return -EFAULT;
+#ifndef CONFIG_ARC_HAS_LLSC
+ preempt_disable(); /* to guarantee atomic r-m-w of futex op */
+#endif
pagefault_disable();
switch (op) {
@@ -60,6 +97,7 @@ static inline int futex_atomic_op_inuser(int encoded_op, u32 __user *uaddr)
__futex_atomic_op("mov %0, %3", ret, oldval, uaddr, oparg);
break;
case FUTEX_OP_ADD:
+ /* oldval = *uaddr; *uaddr += oparg ; ret = *uaddr */
__futex_atomic_op("add %0, %1, %3", ret, oldval, uaddr, oparg);
break;
case FUTEX_OP_OR:
@@ -76,6 +114,9 @@ static inline int futex_atomic_op_inuser(int encoded_op, u32 __user *uaddr)
}
pagefault_enable();
+#ifndef CONFIG_ARC_HAS_LLSC
+ preempt_enable();
+#endif
if (!ret) {
switch (cmp) {
@@ -104,48 +145,57 @@ static inline int futex_atomic_op_inuser(int encoded_op, u32 __user *uaddr)
return ret;
}
-/* Compare-xchg with pagefaults disabled.
- * Notes:
- * -Best-Effort: Exchg happens only if compare succeeds.
- * If compare fails, returns; leaving retry/looping to upper layers
- * -successful cmp-xchg: return orig value in @addr (same as cmp val)
- * -Compare fails: return orig value in @addr
- * -user access r/w fails: return -EFAULT
+/*
+ * cmpxchg of futex (pagefaults disabled by caller)
+ * Return 0 for success, -EFAULT otherwise
*/
static inline int
-futex_atomic_cmpxchg_inatomic(u32 *uval, u32 __user *uaddr, u32 oldval,
- u32 newval)
+futex_atomic_cmpxchg_inatomic(u32 *uval, u32 __user *uaddr, u32 expval,
+ u32 newval)
{
- u32 val;
+ int ret = 0;
+ u32 existval;
- if (!access_ok(VERIFY_WRITE, uaddr, sizeof(int)))
+ if (!access_ok(VERIFY_WRITE, uaddr, sizeof(u32)))
return -EFAULT;
- pagefault_disable();
+#ifndef CONFIG_ARC_HAS_LLSC
+ preempt_disable(); /* to guarantee atomic r-m-w of futex op */
+#endif
+ smp_mb();
- /* TBD : can use llock/scond */
__asm__ __volatile__(
- "1: ld %0, [%3] \n"
- " brne %0, %1, 3f \n"
- "2: st %2, [%3] \n"
+#ifdef CONFIG_ARC_HAS_LLSC
+ "1: llock %1, [%4] \n"
+ " brne %1, %2, 3f \n"
+ "2: scond %3, [%4] \n"
+ " bnz 1b \n"
+#else
+ "1: ld %1, [%4] \n"
+ " brne %1, %2, 3f \n"
+ "2: st %3, [%4] \n"
+#endif
"3: \n"
" .section .fixup,\"ax\" \n"
- "4: mov %0, %4 \n"
- " b 3b \n"
+ "4: mov %0, %5 \n"
+ " j 3b \n"
" .previous \n"
" .section __ex_table,\"a\" \n"
" .align 4 \n"
" .word 1b, 4b \n"
" .word 2b, 4b \n"
" .previous\n"
- : "=&r"(val)
- : "r"(oldval), "r"(newval), "r"(uaddr), "ir"(-EFAULT)
+ : "+&r"(ret), "=&r"(existval)
+ : "r"(expval), "r"(newval), "r"(uaddr), "ir"(-EFAULT)
: "cc", "memory");
- pagefault_enable();
+ smp_mb();
- *uval = val;
- return val;
+#ifndef CONFIG_ARC_HAS_LLSC
+ preempt_enable();
+#endif
+ *uval = existval;
+ return ret;
}
#endif
diff --git a/arch/arc/include/asm/mm-arch-hooks.h b/arch/arc/include/asm/mm-arch-hooks.h
deleted file mode 100644
index c37541c5f8ba..000000000000
--- a/arch/arc/include/asm/mm-arch-hooks.h
+++ /dev/null
@@ -1,15 +0,0 @@
-/*
- * Architecture specific mm hooks
- *
- * Copyright (C) 2015, IBM Corporation
- * Author: Laurent Dufour <ldufour@linux.vnet.ibm.com>
- *
- * This program is free software; you can redistribute it and/or modify
- * it under the terms of the GNU General Public License version 2 as
- * published by the Free Software Foundation.
- */
-
-#ifndef _ASM_ARC_MM_ARCH_HOOKS_H
-#define _ASM_ARC_MM_ARCH_HOOKS_H
-
-#endif /* _ASM_ARC_MM_ARCH_HOOKS_H */
diff --git a/arch/arc/include/asm/perf_event.h b/arch/arc/include/asm/perf_event.h
index 2b8880e953a2..5f071762fb1c 100644
--- a/arch/arc/include/asm/perf_event.h
+++ b/arch/arc/include/asm/perf_event.h
@@ -1,6 +1,7 @@
/*
* Linux performance counter support for ARC
*
+ * Copyright (C) 2014-2015 Synopsys, Inc. (www.synopsys.com)
* Copyright (C) 2011-2013 Synopsys, Inc. (www.synopsys.com)
*
* This program is free software; you can redistribute it and/or modify
@@ -12,8 +13,8 @@
#ifndef __ASM_PERF_EVENT_H
#define __ASM_PERF_EVENT_H
-/* real maximum varies per CPU, this is the maximum supported by the driver */
-#define ARC_PMU_MAX_HWEVENTS 64
+/* Max number of counters that PCT block may ever have */
+#define ARC_PERF_MAX_COUNTERS 32
#define ARC_REG_CC_BUILD 0xF6
#define ARC_REG_CC_INDEX 0x240
@@ -28,15 +29,22 @@
#define ARC_REG_PCT_CONFIG 0x254
#define ARC_REG_PCT_CONTROL 0x255
#define ARC_REG_PCT_INDEX 0x256
+#define ARC_REG_PCT_INT_CNTL 0x25C
+#define ARC_REG_PCT_INT_CNTH 0x25D
+#define ARC_REG_PCT_INT_CTRL 0x25E
+#define ARC_REG_PCT_INT_ACT 0x25F
+
+#define ARC_REG_PCT_CONFIG_USER (1 << 18) /* count in user mode */
+#define ARC_REG_PCT_CONFIG_KERN (1 << 19) /* count in kernel mode */
#define ARC_REG_PCT_CONTROL_CC (1 << 16) /* clear counts */
#define ARC_REG_PCT_CONTROL_SN (1 << 17) /* snapshot */
struct arc_reg_pct_build {
#ifdef CONFIG_CPU_BIG_ENDIAN
- unsigned int m:8, c:8, r:6, s:2, v:8;
+ unsigned int m:8, c:8, r:5, i:1, s:2, v:8;
#else
- unsigned int v:8, s:2, r:6, c:8, m:8;
+ unsigned int v:8, s:2, i:1, r:5, c:8, m:8;
#endif
};
@@ -95,10 +103,13 @@ static const char * const arc_pmu_ev_hw_map[] = {
/* counts condition */
[PERF_COUNT_HW_INSTRUCTIONS] = "iall",
- [PERF_COUNT_HW_BRANCH_INSTRUCTIONS] = "ijmp",
+ [PERF_COUNT_HW_BRANCH_INSTRUCTIONS] = "ijmp", /* Excludes ZOL jumps */
[PERF_COUNT_ARC_BPOK] = "bpok", /* NP-NT, PT-T, PNT-NT */
+#ifdef CONFIG_ISA_ARCV2
+ [PERF_COUNT_HW_BRANCH_MISSES] = "bpmp",
+#else
[PERF_COUNT_HW_BRANCH_MISSES] = "bpfail", /* NP-T, PT-NT, PNT-T */
-
+#endif
[PERF_COUNT_ARC_LDC] = "imemrdc", /* Instr: mem read cached */
[PERF_COUNT_ARC_STC] = "imemwrc", /* Instr: mem write cached */
diff --git a/arch/arc/include/asm/ptrace.h b/arch/arc/include/asm/ptrace.h
index 91755972b9a2..69095da1fcfd 100644
--- a/arch/arc/include/asm/ptrace.h
+++ b/arch/arc/include/asm/ptrace.h
@@ -20,20 +20,20 @@
struct pt_regs {
/* Real registers */
- long bta; /* bta_l1, bta_l2, erbta */
+ unsigned long bta; /* bta_l1, bta_l2, erbta */
- long lp_start, lp_end, lp_count;
+ unsigned long lp_start, lp_end, lp_count;
- long status32; /* status32_l1, status32_l2, erstatus */
- long ret; /* ilink1, ilink2 or eret */
- long blink;
- long fp;
- long r26; /* gp */
+ unsigned long status32; /* status32_l1, status32_l2, erstatus */
+ unsigned long ret; /* ilink1, ilink2 or eret */
+ unsigned long blink;
+ unsigned long fp;
+ unsigned long r26; /* gp */
- long r12, r11, r10, r9, r8, r7, r6, r5, r4, r3, r2, r1, r0;
+ unsigned long r12, r11, r10, r9, r8, r7, r6, r5, r4, r3, r2, r1, r0;
- long sp; /* user/kernel sp depending on where we came from */
- long orig_r0;
+ unsigned long sp; /* User/Kernel depending on where we came from */
+ unsigned long orig_r0;
/*
* To distinguish bet excp, syscall, irq
@@ -55,13 +55,13 @@ struct pt_regs {
unsigned long event;
};
- long user_r25;
+ unsigned long user_r25;
};
#else
struct pt_regs {
- long orig_r0;
+ unsigned long orig_r0;
union {
struct {
@@ -76,26 +76,26 @@ struct pt_regs {
unsigned long event;
};
- long bta; /* bta_l1, bta_l2, erbta */
+ unsigned long bta; /* bta_l1, bta_l2, erbta */
- long user_r25;
+ unsigned long user_r25;
- long r26; /* gp */
- long fp;
- long sp; /* user/kernel sp depending on where we came from */
+ unsigned long r26; /* gp */
+ unsigned long fp;
+ unsigned long sp; /* user/kernel sp depending on where we came from */
- long r12;
+ unsigned long r12;
/*------- Below list auto saved by h/w -----------*/
- long r0, r1, r2, r3, r4, r5, r6, r7, r8, r9, r10, r11;
+ unsigned long r0, r1, r2, r3, r4, r5, r6, r7, r8, r9, r10, r11;
- long blink;
- long lp_end, lp_start, lp_count;
+ unsigned long blink;
+ unsigned long lp_end, lp_start, lp_count;
- long ei, ldi, jli;
+ unsigned long ei, ldi, jli;
- long ret;
- long status32;
+ unsigned long ret;
+ unsigned long status32;
};
#endif
@@ -103,7 +103,7 @@ struct pt_regs {
/* Callee saved registers - need to be saved only when you are scheduled out */
struct callee_regs {
- long r25, r24, r23, r22, r21, r20, r19, r18, r17, r16, r15, r14, r13;
+ unsigned long r25, r24, r23, r22, r21, r20, r19, r18, r17, r16, r15, r14, r13;
};
#define instruction_pointer(regs) ((regs)->ret)
@@ -142,7 +142,7 @@ struct callee_regs {
static inline long regs_return_value(struct pt_regs *regs)
{
- return regs->r0;
+ return (long)regs->r0;
}
#endif /* !__ASSEMBLY__ */
diff --git a/arch/arc/include/asm/spinlock.h b/arch/arc/include/asm/spinlock.h
index e1651df6a93d..db8c59d1eaeb 100644
--- a/arch/arc/include/asm/spinlock.h
+++ b/arch/arc/include/asm/spinlock.h
@@ -18,9 +18,518 @@
#define arch_spin_unlock_wait(x) \
do { while (arch_spin_is_locked(x)) cpu_relax(); } while (0)
+#ifdef CONFIG_ARC_HAS_LLSC
+
+/*
+ * A normal LLOCK/SCOND based system, w/o need for livelock workaround
+ */
+#ifndef CONFIG_ARC_STAR_9000923308
+
static inline void arch_spin_lock(arch_spinlock_t *lock)
{
- unsigned int tmp = __ARCH_SPIN_LOCK_LOCKED__;
+ unsigned int val;
+
+ smp_mb();
+
+ __asm__ __volatile__(
+ "1: llock %[val], [%[slock]] \n"
+ " breq %[val], %[LOCKED], 1b \n" /* spin while LOCKED */
+ " scond %[LOCKED], [%[slock]] \n" /* acquire */
+ " bnz 1b \n"
+ " \n"
+ : [val] "=&r" (val)
+ : [slock] "r" (&(lock->slock)),
+ [LOCKED] "r" (__ARCH_SPIN_LOCK_LOCKED__)
+ : "memory", "cc");
+
+ smp_mb();
+}
+
+/* 1 - lock taken successfully */
+static inline int arch_spin_trylock(arch_spinlock_t *lock)
+{
+ unsigned int val, got_it = 0;
+
+ smp_mb();
+
+ __asm__ __volatile__(
+ "1: llock %[val], [%[slock]] \n"
+ " breq %[val], %[LOCKED], 4f \n" /* already LOCKED, just bail */
+ " scond %[LOCKED], [%[slock]] \n" /* acquire */
+ " bnz 1b \n"
+ " mov %[got_it], 1 \n"
+ "4: \n"
+ " \n"
+ : [val] "=&r" (val),
+ [got_it] "+&r" (got_it)
+ : [slock] "r" (&(lock->slock)),
+ [LOCKED] "r" (__ARCH_SPIN_LOCK_LOCKED__)
+ : "memory", "cc");
+
+ smp_mb();
+
+ return got_it;
+}
+
+static inline void arch_spin_unlock(arch_spinlock_t *lock)
+{
+ smp_mb();
+
+ lock->slock = __ARCH_SPIN_LOCK_UNLOCKED__;
+
+ smp_mb();
+}
+
+/*
+ * Read-write spinlocks, allowing multiple readers but only one writer.
+ * Unfair locking as Writers could be starved indefinitely by Reader(s)
+ */
+
+static inline void arch_read_lock(arch_rwlock_t *rw)
+{
+ unsigned int val;
+
+ smp_mb();
+
+ /*
+ * zero means writer holds the lock exclusively, deny Reader.
+ * Otherwise grant lock to first/subseq reader
+ *
+ * if (rw->counter > 0) {
+ * rw->counter--;
+ * ret = 1;
+ * }
+ */
+
+ __asm__ __volatile__(
+ "1: llock %[val], [%[rwlock]] \n"
+ " brls %[val], %[WR_LOCKED], 1b\n" /* <= 0: spin while write locked */
+ " sub %[val], %[val], 1 \n" /* reader lock */
+ " scond %[val], [%[rwlock]] \n"
+ " bnz 1b \n"
+ " \n"
+ : [val] "=&r" (val)
+ : [rwlock] "r" (&(rw->counter)),
+ [WR_LOCKED] "ir" (0)
+ : "memory", "cc");
+
+ smp_mb();
+}
+
+/* 1 - lock taken successfully */
+static inline int arch_read_trylock(arch_rwlock_t *rw)
+{
+ unsigned int val, got_it = 0;
+
+ smp_mb();
+
+ __asm__ __volatile__(
+ "1: llock %[val], [%[rwlock]] \n"
+ " brls %[val], %[WR_LOCKED], 4f\n" /* <= 0: already write locked, bail */
+ " sub %[val], %[val], 1 \n" /* counter-- */
+ " scond %[val], [%[rwlock]] \n"
+ " bnz 1b \n" /* retry if collided with someone */
+ " mov %[got_it], 1 \n"
+ " \n"
+ "4: ; --- done --- \n"
+
+ : [val] "=&r" (val),
+ [got_it] "+&r" (got_it)
+ : [rwlock] "r" (&(rw->counter)),
+ [WR_LOCKED] "ir" (0)
+ : "memory", "cc");
+
+ smp_mb();
+
+ return got_it;
+}
+
+static inline void arch_write_lock(arch_rwlock_t *rw)
+{
+ unsigned int val;
+
+ smp_mb();
+
+ /*
+ * If reader(s) hold lock (lock < __ARCH_RW_LOCK_UNLOCKED__),
+ * deny writer. Otherwise if unlocked grant to writer
+ * Hence the claim that Linux rwlocks are unfair to writers.
+ * (can be starved for an indefinite time by readers).
+ *
+ * if (rw->counter == __ARCH_RW_LOCK_UNLOCKED__) {
+ * rw->counter = 0;
+ * ret = 1;
+ * }
+ */
+
+ __asm__ __volatile__(
+ "1: llock %[val], [%[rwlock]] \n"
+ " brne %[val], %[UNLOCKED], 1b \n" /* while !UNLOCKED spin */
+ " mov %[val], %[WR_LOCKED] \n"
+ " scond %[val], [%[rwlock]] \n"
+ " bnz 1b \n"
+ " \n"
+ : [val] "=&r" (val)
+ : [rwlock] "r" (&(rw->counter)),
+ [UNLOCKED] "ir" (__ARCH_RW_LOCK_UNLOCKED__),
+ [WR_LOCKED] "ir" (0)
+ : "memory", "cc");
+
+ smp_mb();
+}
+
+/* 1 - lock taken successfully */
+static inline int arch_write_trylock(arch_rwlock_t *rw)
+{
+ unsigned int val, got_it = 0;
+
+ smp_mb();
+
+ __asm__ __volatile__(
+ "1: llock %[val], [%[rwlock]] \n"
+ " brne %[val], %[UNLOCKED], 4f \n" /* !UNLOCKED, bail */
+ " mov %[val], %[WR_LOCKED] \n"
+ " scond %[val], [%[rwlock]] \n"
+ " bnz 1b \n" /* retry if collided with someone */
+ " mov %[got_it], 1 \n"
+ " \n"
+ "4: ; --- done --- \n"
+
+ : [val] "=&r" (val),
+ [got_it] "+&r" (got_it)
+ : [rwlock] "r" (&(rw->counter)),
+ [UNLOCKED] "ir" (__ARCH_RW_LOCK_UNLOCKED__),
+ [WR_LOCKED] "ir" (0)
+ : "memory", "cc");
+
+ smp_mb();
+
+ return got_it;
+}
+
+static inline void arch_read_unlock(arch_rwlock_t *rw)
+{
+ unsigned int val;
+
+ smp_mb();
+
+ /*
+ * rw->counter++;
+ */
+ __asm__ __volatile__(
+ "1: llock %[val], [%[rwlock]] \n"
+ " add %[val], %[val], 1 \n"
+ " scond %[val], [%[rwlock]] \n"
+ " bnz 1b \n"
+ " \n"
+ : [val] "=&r" (val)
+ : [rwlock] "r" (&(rw->counter))
+ : "memory", "cc");
+
+ smp_mb();
+}
+
+static inline void arch_write_unlock(arch_rwlock_t *rw)
+{
+ smp_mb();
+
+ rw->counter = __ARCH_RW_LOCK_UNLOCKED__;
+
+ smp_mb();
+}
+
+#else /* CONFIG_ARC_STAR_9000923308 */
+
+/*
+ * HS38x4 could get into a LLOCK/SCOND livelock in case of multiple overlapping
+ * coherency transactions in the SCU. The exclusive line state keeps rotating
+ * among contenting cores leading to a never ending cycle. So break the cycle
+ * by deferring the retry of failed exclusive access (SCOND). The actual delay
+ * needed is function of number of contending cores as well as the unrelated
+ * coherency traffic from other cores. To keep the code simple, start off with
+ * small delay of 1 which would suffice most cases and in case of contention
+ * double the delay. Eventually the delay is sufficient such that the coherency
+ * pipeline is drained, thus a subsequent exclusive access would succeed.
+ */
+
+#define SCOND_FAIL_RETRY_VAR_DEF \
+ unsigned int delay, tmp; \
+
+#define SCOND_FAIL_RETRY_ASM \
+ " ; --- scond fail delay --- \n" \
+ " mov %[tmp], %[delay] \n" /* tmp = delay */ \
+ "2: brne.d %[tmp], 0, 2b \n" /* while (tmp != 0) */ \
+ " sub %[tmp], %[tmp], 1 \n" /* tmp-- */ \
+ " rol %[delay], %[delay] \n" /* delay *= 2 */ \
+ " b 1b \n" /* start over */ \
+ " \n" \
+ "4: ; --- done --- \n" \
+
+#define SCOND_FAIL_RETRY_VARS \
+ ,[delay] "=&r" (delay), [tmp] "=&r" (tmp) \
+
+static inline void arch_spin_lock(arch_spinlock_t *lock)
+{
+ unsigned int val;
+ SCOND_FAIL_RETRY_VAR_DEF;
+
+ smp_mb();
+
+ __asm__ __volatile__(
+ "0: mov %[delay], 1 \n"
+ "1: llock %[val], [%[slock]] \n"
+ " breq %[val], %[LOCKED], 0b \n" /* spin while LOCKED */
+ " scond %[LOCKED], [%[slock]] \n" /* acquire */
+ " bz 4f \n" /* done */
+ " \n"
+ SCOND_FAIL_RETRY_ASM
+
+ : [val] "=&r" (val)
+ SCOND_FAIL_RETRY_VARS
+ : [slock] "r" (&(lock->slock)),
+ [LOCKED] "r" (__ARCH_SPIN_LOCK_LOCKED__)
+ : "memory", "cc");
+
+ smp_mb();
+}
+
+/* 1 - lock taken successfully */
+static inline int arch_spin_trylock(arch_spinlock_t *lock)
+{
+ unsigned int val, got_it = 0;
+ SCOND_FAIL_RETRY_VAR_DEF;
+
+ smp_mb();
+
+ __asm__ __volatile__(
+ "0: mov %[delay], 1 \n"
+ "1: llock %[val], [%[slock]] \n"
+ " breq %[val], %[LOCKED], 4f \n" /* already LOCKED, just bail */
+ " scond %[LOCKED], [%[slock]] \n" /* acquire */
+ " bz.d 4f \n"
+ " mov.z %[got_it], 1 \n" /* got it */
+ " \n"
+ SCOND_FAIL_RETRY_ASM
+
+ : [val] "=&r" (val),
+ [got_it] "+&r" (got_it)
+ SCOND_FAIL_RETRY_VARS
+ : [slock] "r" (&(lock->slock)),
+ [LOCKED] "r" (__ARCH_SPIN_LOCK_LOCKED__)
+ : "memory", "cc");
+
+ smp_mb();
+
+ return got_it;
+}
+
+static inline void arch_spin_unlock(arch_spinlock_t *lock)
+{
+ smp_mb();
+
+ lock->slock = __ARCH_SPIN_LOCK_UNLOCKED__;
+
+ smp_mb();
+}
+
+/*
+ * Read-write spinlocks, allowing multiple readers but only one writer.
+ * Unfair locking as Writers could be starved indefinitely by Reader(s)
+ */
+
+static inline void arch_read_lock(arch_rwlock_t *rw)
+{
+ unsigned int val;
+ SCOND_FAIL_RETRY_VAR_DEF;
+
+ smp_mb();
+
+ /*
+ * zero means writer holds the lock exclusively, deny Reader.
+ * Otherwise grant lock to first/subseq reader
+ *
+ * if (rw->counter > 0) {
+ * rw->counter--;
+ * ret = 1;
+ * }
+ */
+
+ __asm__ __volatile__(
+ "0: mov %[delay], 1 \n"
+ "1: llock %[val], [%[rwlock]] \n"
+ " brls %[val], %[WR_LOCKED], 0b\n" /* <= 0: spin while write locked */
+ " sub %[val], %[val], 1 \n" /* reader lock */
+ " scond %[val], [%[rwlock]] \n"
+ " bz 4f \n" /* done */
+ " \n"
+ SCOND_FAIL_RETRY_ASM
+
+ : [val] "=&r" (val)
+ SCOND_FAIL_RETRY_VARS
+ : [rwlock] "r" (&(rw->counter)),
+ [WR_LOCKED] "ir" (0)
+ : "memory", "cc");
+
+ smp_mb();
+}
+
+/* 1 - lock taken successfully */
+static inline int arch_read_trylock(arch_rwlock_t *rw)
+{
+ unsigned int val, got_it = 0;
+ SCOND_FAIL_RETRY_VAR_DEF;
+
+ smp_mb();
+
+ __asm__ __volatile__(
+ "0: mov %[delay], 1 \n"
+ "1: llock %[val], [%[rwlock]] \n"
+ " brls %[val], %[WR_LOCKED], 4f\n" /* <= 0: already write locked, bail */
+ " sub %[val], %[val], 1 \n" /* counter-- */
+ " scond %[val], [%[rwlock]] \n"
+ " bz.d 4f \n"
+ " mov.z %[got_it], 1 \n" /* got it */
+ " \n"
+ SCOND_FAIL_RETRY_ASM
+
+ : [val] "=&r" (val),
+ [got_it] "+&r" (got_it)
+ SCOND_FAIL_RETRY_VARS
+ : [rwlock] "r" (&(rw->counter)),
+ [WR_LOCKED] "ir" (0)
+ : "memory", "cc");
+
+ smp_mb();
+
+ return got_it;
+}
+
+static inline void arch_write_lock(arch_rwlock_t *rw)
+{
+ unsigned int val;
+ SCOND_FAIL_RETRY_VAR_DEF;
+
+ smp_mb();
+
+ /*
+ * If reader(s) hold lock (lock < __ARCH_RW_LOCK_UNLOCKED__),
+ * deny writer. Otherwise if unlocked grant to writer
+ * Hence the claim that Linux rwlocks are unfair to writers.
+ * (can be starved for an indefinite time by readers).
+ *
+ * if (rw->counter == __ARCH_RW_LOCK_UNLOCKED__) {
+ * rw->counter = 0;
+ * ret = 1;
+ * }
+ */
+
+ __asm__ __volatile__(
+ "0: mov %[delay], 1 \n"
+ "1: llock %[val], [%[rwlock]] \n"
+ " brne %[val], %[UNLOCKED], 0b \n" /* while !UNLOCKED spin */
+ " mov %[val], %[WR_LOCKED] \n"
+ " scond %[val], [%[rwlock]] \n"
+ " bz 4f \n"
+ " \n"
+ SCOND_FAIL_RETRY_ASM
+
+ : [val] "=&r" (val)
+ SCOND_FAIL_RETRY_VARS
+ : [rwlock] "r" (&(rw->counter)),
+ [UNLOCKED] "ir" (__ARCH_RW_LOCK_UNLOCKED__),
+ [WR_LOCKED] "ir" (0)
+ : "memory", "cc");
+
+ smp_mb();
+}
+
+/* 1 - lock taken successfully */
+static inline int arch_write_trylock(arch_rwlock_t *rw)
+{
+ unsigned int val, got_it = 0;
+ SCOND_FAIL_RETRY_VAR_DEF;
+
+ smp_mb();
+
+ __asm__ __volatile__(
+ "0: mov %[delay], 1 \n"
+ "1: llock %[val], [%[rwlock]] \n"
+ " brne %[val], %[UNLOCKED], 4f \n" /* !UNLOCKED, bail */
+ " mov %[val], %[WR_LOCKED] \n"
+ " scond %[val], [%[rwlock]] \n"
+ " bz.d 4f \n"
+ " mov.z %[got_it], 1 \n" /* got it */
+ " \n"
+ SCOND_FAIL_RETRY_ASM
+
+ : [val] "=&r" (val),
+ [got_it] "+&r" (got_it)
+ SCOND_FAIL_RETRY_VARS
+ : [rwlock] "r" (&(rw->counter)),
+ [UNLOCKED] "ir" (__ARCH_RW_LOCK_UNLOCKED__),
+ [WR_LOCKED] "ir" (0)
+ : "memory", "cc");
+
+ smp_mb();
+
+ return got_it;
+}
+
+static inline void arch_read_unlock(arch_rwlock_t *rw)
+{
+ unsigned int val;
+
+ smp_mb();
+
+ /*
+ * rw->counter++;
+ */
+ __asm__ __volatile__(
+ "1: llock %[val], [%[rwlock]] \n"
+ " add %[val], %[val], 1 \n"
+ " scond %[val], [%[rwlock]] \n"
+ " bnz 1b \n"
+ " \n"
+ : [val] "=&r" (val)
+ : [rwlock] "r" (&(rw->counter))
+ : "memory", "cc");
+
+ smp_mb();
+}
+
+static inline void arch_write_unlock(arch_rwlock_t *rw)
+{
+ unsigned int val;
+
+ smp_mb();
+
+ /*
+ * rw->counter = __ARCH_RW_LOCK_UNLOCKED__;
+ */
+ __asm__ __volatile__(
+ "1: llock %[val], [%[rwlock]] \n"
+ " scond %[UNLOCKED], [%[rwlock]]\n"
+ " bnz 1b \n"
+ " \n"
+ : [val] "=&r" (val)
+ : [rwlock] "r" (&(rw->counter)),
+ [UNLOCKED] "r" (__ARCH_RW_LOCK_UNLOCKED__)
+ : "memory", "cc");
+
+ smp_mb();
+}
+
+#undef SCOND_FAIL_RETRY_VAR_DEF
+#undef SCOND_FAIL_RETRY_ASM
+#undef SCOND_FAIL_RETRY_VARS
+
+#endif /* CONFIG_ARC_STAR_9000923308 */
+
+#else /* !CONFIG_ARC_HAS_LLSC */
+
+static inline void arch_spin_lock(arch_spinlock_t *lock)
+{
+ unsigned int val = __ARCH_SPIN_LOCK_LOCKED__;
/*
* This smp_mb() is technically superfluous, we only need the one
@@ -33,7 +542,7 @@ static inline void arch_spin_lock(arch_spinlock_t *lock)
__asm__ __volatile__(
"1: ex %0, [%1] \n"
" breq %0, %2, 1b \n"
- : "+&r" (tmp)
+ : "+&r" (val)
: "r"(&(lock->slock)), "ir"(__ARCH_SPIN_LOCK_LOCKED__)
: "memory");
@@ -48,26 +557,27 @@ static inline void arch_spin_lock(arch_spinlock_t *lock)
smp_mb();
}
+/* 1 - lock taken successfully */
static inline int arch_spin_trylock(arch_spinlock_t *lock)
{
- unsigned int tmp = __ARCH_SPIN_LOCK_LOCKED__;
+ unsigned int val = __ARCH_SPIN_LOCK_LOCKED__;
smp_mb();
__asm__ __volatile__(
"1: ex %0, [%1] \n"
- : "+r" (tmp)
+ : "+r" (val)
: "r"(&(lock->slock))
: "memory");
smp_mb();
- return (tmp == __ARCH_SPIN_LOCK_UNLOCKED__);
+ return (val == __ARCH_SPIN_LOCK_UNLOCKED__);
}
static inline void arch_spin_unlock(arch_spinlock_t *lock)
{
- unsigned int tmp = __ARCH_SPIN_LOCK_UNLOCKED__;
+ unsigned int val = __ARCH_SPIN_LOCK_UNLOCKED__;
/*
* RELEASE barrier: given the instructions avail on ARCv2, full barrier
@@ -77,7 +587,7 @@ static inline void arch_spin_unlock(arch_spinlock_t *lock)
__asm__ __volatile__(
" ex %0, [%1] \n"
- : "+r" (tmp)
+ : "+r" (val)
: "r"(&(lock->slock))
: "memory");
@@ -90,19 +600,12 @@ static inline void arch_spin_unlock(arch_spinlock_t *lock)
/*
* Read-write spinlocks, allowing multiple readers but only one writer.
+ * Unfair locking as Writers could be starved indefinitely by Reader(s)
*
* The spinlock itself is contained in @counter and access to it is
* serialized with @lock_mutex.
- *
- * Unfair locking as Writers could be starved indefinitely by Reader(s)
*/
-/* Would read_trylock() succeed? */
-#define arch_read_can_lock(x) ((x)->counter > 0)
-
-/* Would write_trylock() succeed? */
-#define arch_write_can_lock(x) ((x)->counter == __ARCH_RW_LOCK_UNLOCKED__)
-
/* 1 - lock taken successfully */
static inline int arch_read_trylock(arch_rwlock_t *rw)
{
@@ -173,6 +676,11 @@ static inline void arch_write_unlock(arch_rwlock_t *rw)
arch_spin_unlock(&(rw->lock_mutex));
}
+#endif
+
+#define arch_read_can_lock(x) ((x)->counter > 0)
+#define arch_write_can_lock(x) ((x)->counter == __ARCH_RW_LOCK_UNLOCKED__)
+
#define arch_read_lock_flags(lock, flags) arch_read_lock(lock)
#define arch_write_lock_flags(lock, flags) arch_write_lock(lock)
diff --git a/arch/arc/include/asm/spinlock_types.h b/arch/arc/include/asm/spinlock_types.h
index 662627ced4f2..4e1ef5f650c6 100644
--- a/arch/arc/include/asm/spinlock_types.h
+++ b/arch/arc/include/asm/spinlock_types.h
@@ -26,7 +26,9 @@ typedef struct {
*/
typedef struct {
volatile unsigned int counter;
+#ifndef CONFIG_ARC_HAS_LLSC
arch_spinlock_t lock_mutex;
+#endif
} arch_rwlock_t;
#define __ARCH_RW_LOCK_UNLOCKED__ 0x01000000
diff --git a/arch/arc/include/uapi/asm/ptrace.h b/arch/arc/include/uapi/asm/ptrace.h
index 76a7739aab1c..0b3ef63d4a03 100644
--- a/arch/arc/include/uapi/asm/ptrace.h
+++ b/arch/arc/include/uapi/asm/ptrace.h
@@ -32,20 +32,20 @@
*/
struct user_regs_struct {
- long pad;
+ unsigned long pad;
struct {
- long bta, lp_start, lp_end, lp_count;
- long status32, ret, blink, fp, gp;
- long r12, r11, r10, r9, r8, r7, r6, r5, r4, r3, r2, r1, r0;
- long sp;
+ unsigned long bta, lp_start, lp_end, lp_count;
+ unsigned long status32, ret, blink, fp, gp;
+ unsigned long r12, r11, r10, r9, r8, r7, r6, r5, r4, r3, r2, r1, r0;
+ unsigned long sp;
} scratch;
- long pad2;
+ unsigned long pad2;
struct {
- long r25, r24, r23, r22, r21, r20;
- long r19, r18, r17, r16, r15, r14, r13;
+ unsigned long r25, r24, r23, r22, r21, r20;
+ unsigned long r19, r18, r17, r16, r15, r14, r13;
} callee;
- long efa; /* break pt addr, for break points in delay slots */
- long stop_pc; /* give dbg stop_pc after ensuring brkpt trap */
+ unsigned long efa; /* break pt addr, for break points in delay slots */
+ unsigned long stop_pc; /* give dbg stop_pc after ensuring brkpt trap */
};
#endif /* !__ASSEMBLY__ */
diff --git a/arch/arc/kernel/entry-arcv2.S b/arch/arc/kernel/entry-arcv2.S
index bd7105d3172f..8fa76567e402 100644
--- a/arch/arc/kernel/entry-arcv2.S
+++ b/arch/arc/kernel/entry-arcv2.S
@@ -57,13 +57,8 @@ VECTOR handle_interrupt ; (23) End of fixed IRQs
.section .text, "ax",@progbits
-res_service: ; processor restart
- flag 0x1 ; not implemented
- nop
- nop
-
-reserved: ; processor restart
- rtie ; jump to processor initializations
+reserved:
+ flag 1 ; Unexpected event, halt
;##################### Interrupt Handling ##############################
diff --git a/arch/arc/kernel/entry.S b/arch/arc/kernel/entry.S
index f7a82fd4d601..589abf5172d6 100644
--- a/arch/arc/kernel/entry.S
+++ b/arch/arc/kernel/entry.S
@@ -42,7 +42,7 @@ ENTRY(ret_from_fork)
; when the forked child comes here from the __switch_to function
; r0 has the last task pointer.
; put last task in scheduler queue
- bl @schedule_tail
+ jl @schedule_tail
ld r9, [sp, PT_status32]
brne r9, 0, 1f
@@ -320,7 +320,7 @@ resume_user_mode_begin:
; --- (Slow Path #1) task preemption ---
bbit0 r9, TIF_NEED_RESCHED, .Lchk_pend_signals
mov blink, resume_user_mode_begin ; tail-call to U mode ret chks
- b @schedule ; BTST+Bnz causes relo error in link
+ j @schedule ; BTST+Bnz causes relo error in link
.Lchk_pend_signals:
IRQ_ENABLE r10
@@ -381,7 +381,7 @@ resume_kernel_mode:
bbit0 r9, TIF_NEED_RESCHED, .Lrestore_regs
; Invoke PREEMPTION
- bl preempt_schedule_irq
+ jl preempt_schedule_irq
; preempt_schedule_irq() always returns with IRQ disabled
#endif
diff --git a/arch/arc/kernel/intc-arcv2.c b/arch/arc/kernel/intc-arcv2.c
index 6208c630abed..26c156827479 100644
--- a/arch/arc/kernel/intc-arcv2.c
+++ b/arch/arc/kernel/intc-arcv2.c
@@ -12,7 +12,6 @@
#include <linux/of.h>
#include <linux/irqdomain.h>
#include <linux/irqchip.h>
-#include "../../drivers/irqchip/irqchip.h"
#include <asm/irq.h>
/*
diff --git a/arch/arc/kernel/intc-compact.c b/arch/arc/kernel/intc-compact.c
index fcdddb631766..039fac30b5c1 100644
--- a/arch/arc/kernel/intc-compact.c
+++ b/arch/arc/kernel/intc-compact.c
@@ -12,7 +12,6 @@
#include <linux/of.h>
#include <linux/irqdomain.h>
#include <linux/irqchip.h>
-#include "../../drivers/irqchip/irqchip.h"
#include <asm/irq.h>
/*
diff --git a/arch/arc/kernel/mcip.c b/arch/arc/kernel/mcip.c
index 30284e8de6ff..4ffd1855f1bd 100644
--- a/arch/arc/kernel/mcip.c
+++ b/arch/arc/kernel/mcip.c
@@ -175,7 +175,6 @@ void mcip_init_early_smp(void)
#include <linux/irqchip.h>
#include <linux/of.h>
#include <linux/of_irq.h>
-#include "../../drivers/irqchip/irqchip.h"
/*
* Set the DEST for @cmn_irq to @cpu_mask (1 bit per core)
@@ -218,11 +217,28 @@ static void idu_irq_unmask(struct irq_data *data)
raw_spin_unlock_irqrestore(&mcip_lock, flags);
}
+#ifdef CONFIG_SMP
static int
-idu_irq_set_affinity(struct irq_data *d, const struct cpumask *cpumask, bool f)
+idu_irq_set_affinity(struct irq_data *data, const struct cpumask *cpumask,
+ bool force)
{
+ unsigned long flags;
+ cpumask_t online;
+
+ /* errout if no online cpu per @cpumask */
+ if (!cpumask_and(&online, cpumask, cpu_online_mask))
+ return -EINVAL;
+
+ raw_spin_lock_irqsave(&mcip_lock, flags);
+
+ idu_set_dest(data->hwirq, cpumask_bits(&online)[0]);
+ idu_set_mode(data->hwirq, IDU_M_TRIG_LEVEL, IDU_M_DISTRI_RR);
+
+ raw_spin_unlock_irqrestore(&mcip_lock, flags);
+
return IRQ_SET_MASK_OK;
}
+#endif
static struct irq_chip idu_irq_chip = {
.name = "MCIP IDU Intc",
@@ -236,9 +252,10 @@ static struct irq_chip idu_irq_chip = {
static int idu_first_irq;
-static void idu_cascade_isr(unsigned int core_irq, struct irq_desc *desc)
+static void idu_cascade_isr(struct irq_desc *desc)
{
struct irq_domain *domain = irq_desc_get_handler_data(desc);
+ unsigned int core_irq = irq_desc_get_irq(desc);
unsigned int idu_irq;
idu_irq = core_irq - idu_first_irq;
@@ -330,8 +347,7 @@ idu_of_init(struct device_node *intc, struct device_node *parent)
if (!i)
idu_first_irq = irq;
- irq_set_handler_data(irq, domain);
- irq_set_chained_handler(irq, idu_cascade_isr);
+ irq_set_chained_handler_and_data(irq, idu_cascade_isr, domain);
}
__mcip_cmd(CMD_IDU_ENABLE, 0);
diff --git a/arch/arc/kernel/perf_event.c b/arch/arc/kernel/perf_event.c
index 1287388c258a..0c08bb1ce15a 100644
--- a/arch/arc/kernel/perf_event.c
+++ b/arch/arc/kernel/perf_event.c
@@ -1,7 +1,7 @@
/*
* Linux performance counter support for ARC700 series
*
- * Copyright (C) 2013 Synopsys, Inc. (www.synopsys.com)
+ * Copyright (C) 2013-2015 Synopsys, Inc. (www.synopsys.com)
*
* This code is inspired by the perf support of various other architectures.
*
@@ -11,6 +11,7 @@
*
*/
#include <linux/errno.h>
+#include <linux/interrupt.h>
#include <linux/module.h>
#include <linux/of.h>
#include <linux/perf_event.h>
@@ -20,12 +21,25 @@
struct arc_pmu {
struct pmu pmu;
- int counter_size; /* in bits */
+ unsigned int irq;
int n_counters;
- unsigned long used_mask[BITS_TO_LONGS(ARC_PMU_MAX_HWEVENTS)];
+ u64 max_period;
int ev_hw_idx[PERF_COUNT_ARC_HW_MAX];
};
+struct arc_pmu_cpu {
+ /*
+ * A 1 bit for an index indicates that the counter is being used for
+ * an event. A 0 means that the counter can be used.
+ */
+ unsigned long used_mask[BITS_TO_LONGS(ARC_PERF_MAX_COUNTERS)];
+
+ /*
+ * The events that are active on the PMU for the given index.
+ */
+ struct perf_event *act_counter[ARC_PERF_MAX_COUNTERS];
+};
+
struct arc_callchain_trace {
int depth;
void *perf_stuff;
@@ -65,6 +79,7 @@ perf_callchain_user(struct perf_callchain_entry *entry, struct pt_regs *regs)
}
static struct arc_pmu *arc_pmu;
+static DEFINE_PER_CPU(struct arc_pmu_cpu, arc_pmu_cpu);
/* read counter #idx; note that counter# != event# on ARC! */
static uint64_t arc_pmu_read_counter(int idx)
@@ -88,18 +103,15 @@ static uint64_t arc_pmu_read_counter(int idx)
static void arc_perf_event_update(struct perf_event *event,
struct hw_perf_event *hwc, int idx)
{
- uint64_t prev_raw_count, new_raw_count;
- int64_t delta;
-
- do {
- prev_raw_count = local64_read(&hwc->prev_count);
- new_raw_count = arc_pmu_read_counter(idx);
- } while (local64_cmpxchg(&hwc->prev_count, prev_raw_count,
- new_raw_count) != prev_raw_count);
-
- delta = (new_raw_count - prev_raw_count) &
- ((1ULL << arc_pmu->counter_size) - 1ULL);
+ uint64_t prev_raw_count = local64_read(&hwc->prev_count);
+ uint64_t new_raw_count = arc_pmu_read_counter(idx);
+ int64_t delta = new_raw_count - prev_raw_count;
+ /*
+ * We don't afaraid of hwc->prev_count changing beneath our feet
+ * because there's no way for us to re-enter this function anytime.
+ */
+ local64_set(&hwc->prev_count, new_raw_count);
local64_add(delta, &event->count);
local64_sub(delta, &hwc->period_left);
}
@@ -142,22 +154,41 @@ static int arc_pmu_event_init(struct perf_event *event)
struct hw_perf_event *hwc = &event->hw;
int ret;
+ if (!is_sampling_event(event)) {
+ hwc->sample_period = arc_pmu->max_period;
+ hwc->last_period = hwc->sample_period;
+ local64_set(&hwc->period_left, hwc->sample_period);
+ }
+
+ hwc->config = 0;
+
+ if (is_isa_arcv2()) {
+ /* "exclude user" means "count only kernel" */
+ if (event->attr.exclude_user)
+ hwc->config |= ARC_REG_PCT_CONFIG_KERN;
+
+ /* "exclude kernel" means "count only user" */
+ if (event->attr.exclude_kernel)
+ hwc->config |= ARC_REG_PCT_CONFIG_USER;
+ }
+
switch (event->attr.type) {
case PERF_TYPE_HARDWARE:
if (event->attr.config >= PERF_COUNT_HW_MAX)
return -ENOENT;
if (arc_pmu->ev_hw_idx[event->attr.config] < 0)
return -ENOENT;
- hwc->config = arc_pmu->ev_hw_idx[event->attr.config];
+ hwc->config |= arc_pmu->ev_hw_idx[event->attr.config];
pr_debug("init event %d with h/w %d \'%s\'\n",
(int) event->attr.config, (int) hwc->config,
arc_pmu_ev_hw_map[event->attr.config]);
return 0;
+
case PERF_TYPE_HW_CACHE:
ret = arc_pmu_cache_event(event->attr.config);
if (ret < 0)
return ret;
- hwc->config = arc_pmu->ev_hw_idx[ret];
+ hwc->config |= arc_pmu->ev_hw_idx[ret];
return 0;
default:
return -ENOENT;
@@ -180,6 +211,47 @@ static void arc_pmu_disable(struct pmu *pmu)
write_aux_reg(ARC_REG_PCT_CONTROL, (tmp & 0xffff0000) | 0x0);
}
+static int arc_pmu_event_set_period(struct perf_event *event)
+{
+ struct hw_perf_event *hwc = &event->hw;
+ s64 left = local64_read(&hwc->period_left);
+ s64 period = hwc->sample_period;
+ int idx = hwc->idx;
+ int overflow = 0;
+ u64 value;
+
+ if (unlikely(left <= -period)) {
+ /* left underflowed by more than period. */
+ left = period;
+ local64_set(&hwc->period_left, left);
+ hwc->last_period = period;
+ overflow = 1;
+ } else if (unlikely(left <= 0)) {
+ /* left underflowed by less than period. */
+ left += period;
+ local64_set(&hwc->period_left, left);
+ hwc->last_period = period;
+ overflow = 1;
+ }
+
+ if (left > arc_pmu->max_period)
+ left = arc_pmu->max_period;
+
+ value = arc_pmu->max_period - left;
+ local64_set(&hwc->prev_count, value);
+
+ /* Select counter */
+ write_aux_reg(ARC_REG_PCT_INDEX, idx);
+
+ /* Write value */
+ write_aux_reg(ARC_REG_PCT_COUNTL, (u32)value);
+ write_aux_reg(ARC_REG_PCT_COUNTH, (value >> 32));
+
+ perf_event_update_userpage(event);
+
+ return overflow;
+}
+
/*
* Assigns hardware counter to hardware condition.
* Note that there is no separate start/stop mechanism;
@@ -194,13 +266,20 @@ static void arc_pmu_start(struct perf_event *event, int flags)
return;
if (flags & PERF_EF_RELOAD)
- WARN_ON_ONCE(!(event->hw.state & PERF_HES_UPTODATE));
+ WARN_ON_ONCE(!(hwc->state & PERF_HES_UPTODATE));
+
+ hwc->state = 0;
- event->hw.state = 0;
+ arc_pmu_event_set_period(event);
+
+ /* Enable interrupt for this counter */
+ if (is_sampling_event(event))
+ write_aux_reg(ARC_REG_PCT_INT_CTRL,
+ read_aux_reg(ARC_REG_PCT_INT_CTRL) | (1 << idx));
/* enable ARC pmu here */
- write_aux_reg(ARC_REG_PCT_INDEX, idx);
- write_aux_reg(ARC_REG_PCT_CONFIG, hwc->config);
+ write_aux_reg(ARC_REG_PCT_INDEX, idx); /* counter # */
+ write_aux_reg(ARC_REG_PCT_CONFIG, hwc->config); /* condition */
}
static void arc_pmu_stop(struct perf_event *event, int flags)
@@ -208,6 +287,17 @@ static void arc_pmu_stop(struct perf_event *event, int flags)
struct hw_perf_event *hwc = &event->hw;
int idx = hwc->idx;
+ /* Disable interrupt for this counter */
+ if (is_sampling_event(event)) {
+ /*
+ * Reset interrupt flag by writing of 1. This is required
+ * to make sure pending interrupt was not left.
+ */
+ write_aux_reg(ARC_REG_PCT_INT_ACT, 1 << idx);
+ write_aux_reg(ARC_REG_PCT_INT_CTRL,
+ read_aux_reg(ARC_REG_PCT_INT_CTRL) & ~(1 << idx));
+ }
+
if (!(event->hw.state & PERF_HES_STOPPED)) {
/* stop ARC pmu here */
write_aux_reg(ARC_REG_PCT_INDEX, idx);
@@ -227,8 +317,12 @@ static void arc_pmu_stop(struct perf_event *event, int flags)
static void arc_pmu_del(struct perf_event *event, int flags)
{
+ struct arc_pmu_cpu *pmu_cpu = this_cpu_ptr(&arc_pmu_cpu);
+
arc_pmu_stop(event, PERF_EF_UPDATE);
- __clear_bit(event->hw.idx, arc_pmu->used_mask);
+ __clear_bit(event->hw.idx, pmu_cpu->used_mask);
+
+ pmu_cpu->act_counter[event->hw.idx] = 0;
perf_event_update_userpage(event);
}
@@ -236,20 +330,31 @@ static void arc_pmu_del(struct perf_event *event, int flags)
/* allocate hardware counter and optionally start counting */
static int arc_pmu_add(struct perf_event *event, int flags)
{
+ struct arc_pmu_cpu *pmu_cpu = this_cpu_ptr(&arc_pmu_cpu);
struct hw_perf_event *hwc = &event->hw;
int idx = hwc->idx;
- if (__test_and_set_bit(idx, arc_pmu->used_mask)) {
- idx = find_first_zero_bit(arc_pmu->used_mask,
+ if (__test_and_set_bit(idx, pmu_cpu->used_mask)) {
+ idx = find_first_zero_bit(pmu_cpu->used_mask,
arc_pmu->n_counters);
if (idx == arc_pmu->n_counters)
return -EAGAIN;
- __set_bit(idx, arc_pmu->used_mask);
+ __set_bit(idx, pmu_cpu->used_mask);
hwc->idx = idx;
}
write_aux_reg(ARC_REG_PCT_INDEX, idx);
+
+ pmu_cpu->act_counter[idx] = event;
+
+ if (is_sampling_event(event)) {
+ /* Mimic full counter overflow as other arches do */
+ write_aux_reg(ARC_REG_PCT_INT_CNTL, (u32)arc_pmu->max_period);
+ write_aux_reg(ARC_REG_PCT_INT_CNTH,
+ (arc_pmu->max_period >> 32));
+ }
+
write_aux_reg(ARC_REG_PCT_CONFIG, 0);
write_aux_reg(ARC_REG_PCT_COUNTL, 0);
write_aux_reg(ARC_REG_PCT_COUNTH, 0);
@@ -264,11 +369,82 @@ static int arc_pmu_add(struct perf_event *event, int flags)
return 0;
}
+#ifdef CONFIG_ISA_ARCV2
+static irqreturn_t arc_pmu_intr(int irq, void *dev)
+{
+ struct perf_sample_data data;
+ struct arc_pmu_cpu *pmu_cpu = this_cpu_ptr(&arc_pmu_cpu);
+ struct pt_regs *regs;
+ int active_ints;
+ int idx;
+
+ arc_pmu_disable(&arc_pmu->pmu);
+
+ active_ints = read_aux_reg(ARC_REG_PCT_INT_ACT);
+
+ regs = get_irq_regs();
+
+ for (idx = 0; idx < arc_pmu->n_counters; idx++) {
+ struct perf_event *event = pmu_cpu->act_counter[idx];
+ struct hw_perf_event *hwc;
+
+ if (!(active_ints & (1 << idx)))
+ continue;
+
+ /* Reset interrupt flag by writing of 1 */
+ write_aux_reg(ARC_REG_PCT_INT_ACT, 1 << idx);
+
+ /*
+ * On reset of "interrupt active" bit corresponding
+ * "interrupt enable" bit gets automatically reset as well.
+ * Now we need to re-enable interrupt for the counter.
+ */
+ write_aux_reg(ARC_REG_PCT_INT_CTRL,
+ read_aux_reg(ARC_REG_PCT_INT_CTRL) | (1 << idx));
+
+ hwc = &event->hw;
+
+ WARN_ON_ONCE(hwc->idx != idx);
+
+ arc_perf_event_update(event, &event->hw, event->hw.idx);
+ perf_sample_data_init(&data, 0, hwc->last_period);
+ if (!arc_pmu_event_set_period(event))
+ continue;
+
+ if (perf_event_overflow(event, &data, regs))
+ arc_pmu_stop(event, 0);
+ }
+
+ arc_pmu_enable(&arc_pmu->pmu);
+
+ return IRQ_HANDLED;
+}
+#else
+
+static irqreturn_t arc_pmu_intr(int irq, void *dev)
+{
+ return IRQ_NONE;
+}
+
+#endif /* CONFIG_ISA_ARCV2 */
+
+void arc_cpu_pmu_irq_init(void)
+{
+ struct arc_pmu_cpu *pmu_cpu = this_cpu_ptr(&arc_pmu_cpu);
+
+ arc_request_percpu_irq(arc_pmu->irq, smp_processor_id(), arc_pmu_intr,
+ "ARC perf counters", pmu_cpu);
+
+ /* Clear all pending interrupt flags */
+ write_aux_reg(ARC_REG_PCT_INT_ACT, 0xffffffff);
+}
+
static int arc_pmu_device_probe(struct platform_device *pdev)
{
struct arc_reg_pct_build pct_bcr;
struct arc_reg_cc_build cc_bcr;
- int i, j;
+ int i, j, has_interrupts;
+ int counter_size; /* in bits */
union cc_name {
struct {
@@ -284,7 +460,7 @@ static int arc_pmu_device_probe(struct platform_device *pdev)
pr_err("This core does not have performance counters!\n");
return -ENODEV;
}
- BUG_ON(pct_bcr.c > ARC_PMU_MAX_HWEVENTS);
+ BUG_ON(pct_bcr.c > ARC_PERF_MAX_COUNTERS);
READ_BCR(ARC_REG_CC_BUILD, cc_bcr);
BUG_ON(!cc_bcr.v); /* Counters exist but No countable conditions ? */
@@ -293,11 +469,16 @@ static int arc_pmu_device_probe(struct platform_device *pdev)
if (!arc_pmu)
return -ENOMEM;
+ has_interrupts = is_isa_arcv2() ? pct_bcr.i : 0;
+
arc_pmu->n_counters = pct_bcr.c;
- arc_pmu->counter_size = 32 + (pct_bcr.s << 4);
+ counter_size = 32 + (pct_bcr.s << 4);
- pr_info("ARC perf\t: %d counters (%d bits), %d countable conditions\n",
- arc_pmu->n_counters, arc_pmu->counter_size, cc_bcr.c);
+ arc_pmu->max_period = (1ULL << counter_size) / 2 - 1ULL;
+
+ pr_info("ARC perf\t: %d counters (%d bits), %d conditions%s\n",
+ arc_pmu->n_counters, counter_size, cc_bcr.c,
+ has_interrupts ? ", [overflow IRQ support]":"");
cc_name.str[8] = 0;
for (i = 0; i < PERF_COUNT_ARC_HW_MAX; i++)
@@ -332,8 +513,37 @@ static int arc_pmu_device_probe(struct platform_device *pdev)
.read = arc_pmu_read,
};
- /* ARC 700 PMU does not support sampling events */
- arc_pmu->pmu.capabilities |= PERF_PMU_CAP_NO_INTERRUPT;
+ if (has_interrupts) {
+ int irq = platform_get_irq(pdev, 0);
+ unsigned long flags;
+
+ if (irq < 0) {
+ pr_err("Cannot get IRQ number for the platform\n");
+ return -ENODEV;
+ }
+
+ arc_pmu->irq = irq;
+
+ /*
+ * arc_cpu_pmu_irq_init() needs to be called on all cores for
+ * their respective local PMU.
+ * However we use opencoded on_each_cpu() to ensure it is called
+ * on core0 first, so that arc_request_percpu_irq() sets up
+ * AUTOEN etc. Otherwise enable_percpu_irq() fails to enable
+ * perf IRQ on non master cores.
+ * see arc_request_percpu_irq()
+ */
+ preempt_disable();
+ local_irq_save(flags);
+ arc_cpu_pmu_irq_init();
+ local_irq_restore(flags);
+ smp_call_function((smp_call_func_t)arc_cpu_pmu_irq_init, 0, 1);
+ preempt_enable();
+
+ /* Clean all pending interrupt flags */
+ write_aux_reg(ARC_REG_PCT_INT_ACT, 0xffffffff);
+ } else
+ arc_pmu->pmu.capabilities |= PERF_PMU_CAP_NO_INTERRUPT;
return perf_pmu_register(&arc_pmu->pmu, pdev->name, PERF_TYPE_RAW);
}
@@ -341,6 +551,7 @@ static int arc_pmu_device_probe(struct platform_device *pdev)
#ifdef CONFIG_OF
static const struct of_device_id arc_pmu_match[] = {
{ .compatible = "snps,arc700-pct" },
+ { .compatible = "snps,archs-pct" },
{},
};
MODULE_DEVICE_TABLE(of, arc_pmu_match);
@@ -348,7 +559,7 @@ MODULE_DEVICE_TABLE(of, arc_pmu_match);
static struct platform_driver arc_pmu_driver = {
.driver = {
- .name = "arc700-pct",
+ .name = "arc-pct",
.of_match_table = of_match_ptr(arc_pmu_match),
},
.probe = arc_pmu_device_probe,
diff --git a/arch/arc/kernel/process.c b/arch/arc/kernel/process.c
index 44092456776f..91d5a0f1f3f7 100644
--- a/arch/arc/kernel/process.c
+++ b/arch/arc/kernel/process.c
@@ -65,7 +65,7 @@ asmlinkage void ret_from_fork(void);
* ------------------
* | r25 | <==== top of Stack (thread.ksp)
* ~ ~
- * | --to-- | (CALLEE Regs of user mode)
+ * | --to-- | (CALLEE Regs of kernel mode)
* | r13 |
* ------------------
* | fp |
diff --git a/arch/arc/kernel/setup.c b/arch/arc/kernel/setup.c
index a3d186211ed3..cabde9dc0696 100644
--- a/arch/arc/kernel/setup.c
+++ b/arch/arc/kernel/setup.c
@@ -47,6 +47,7 @@ static void read_arc_build_cfg_regs(void)
struct bcr_perip uncached_space;
struct bcr_generic bcr;
struct cpuinfo_arc *cpu = &cpuinfo_arc700[smp_processor_id()];
+ unsigned long perip_space;
FIX_PTR(cpu);
READ_BCR(AUX_IDENTITY, cpu->core);
@@ -56,7 +57,12 @@ static void read_arc_build_cfg_regs(void)
cpu->vec_base = read_aux_reg(AUX_INTR_VEC_BASE);
READ_BCR(ARC_REG_D_UNCACH_BCR, uncached_space);
- BUG_ON((uncached_space.start << 24) != ARC_UNCACHED_ADDR_SPACE);
+ if (uncached_space.ver < 3)
+ perip_space = uncached_space.start << 24;
+ else
+ perip_space = read_aux_reg(AUX_NON_VOL) & 0xF0000000;
+
+ BUG_ON(perip_space != ARC_UNCACHED_ADDR_SPACE);
READ_BCR(ARC_REG_MUL_BCR, cpu->extn_mpy);
@@ -142,17 +148,22 @@ static void read_arc_build_cfg_regs(void)
}
static const struct cpuinfo_data arc_cpu_tbl[] = {
+#ifdef CONFIG_ISA_ARCOMPACT
{ {0x20, "ARC 600" }, 0x2F},
{ {0x30, "ARC 700" }, 0x33},
{ {0x34, "ARC 700 R4.10"}, 0x34},
{ {0x35, "ARC 700 R4.11"}, 0x35},
- { {0x50, "ARC HS38" }, 0x51},
+#else
+ { {0x50, "ARC HS38 R2.0"}, 0x51},
+ { {0x52, "ARC HS38 R2.1"}, 0x52},
+#endif
{ {0x00, NULL } }
};
-#define IS_AVAIL1(v, str) ((v) ? str : "")
-#define IS_USED(cfg) (IS_ENABLED(cfg) ? "" : "(not used) ")
-#define IS_AVAIL2(v, str, cfg) IS_AVAIL1(v, str), IS_AVAIL1(v, IS_USED(cfg))
+#define IS_AVAIL1(v, s) ((v) ? s : "")
+#define IS_USED_RUN(v) ((v) ? "" : "(not used) ")
+#define IS_USED_CFG(cfg) IS_USED_RUN(IS_ENABLED(cfg))
+#define IS_AVAIL2(v, s, cfg) IS_AVAIL1(v, s), IS_AVAIL1(v, IS_USED_CFG(cfg))
static char *arc_cpu_mumbojumbo(int cpu_id, char *buf, int len)
{
@@ -226,7 +237,7 @@ static char *arc_cpu_mumbojumbo(int cpu_id, char *buf, int len)
n += scnprintf(buf + n, len - n, "mpy[opt %d] ", opt);
}
n += scnprintf(buf + n, len - n, "%s",
- IS_USED(CONFIG_ARC_HAS_HW_MPY));
+ IS_USED_CFG(CONFIG_ARC_HAS_HW_MPY));
}
n += scnprintf(buf + n, len - n, "%s%s%s%s%s%s%s%s\n",
@@ -325,6 +336,10 @@ static void arc_chk_core_config(void)
pr_warn("CONFIG_ARC_FPU_SAVE_RESTORE needed for working apps\n");
else if (!cpu->extn.fpu_dp && fpu_enabled)
panic("FPU non-existent, disable CONFIG_ARC_FPU_SAVE_RESTORE\n");
+
+ if (is_isa_arcv2() && IS_ENABLED(CONFIG_SMP) && cpu->isa.atomic &&
+ !IS_ENABLED(CONFIG_ARC_STAR_9000923308))
+ panic("llock/scond livelock workaround missing\n");
}
/*
diff --git a/arch/arc/kernel/time.c b/arch/arc/kernel/time.c
index 3364d2bbc515..4294761a2b3e 100644
--- a/arch/arc/kernel/time.c
+++ b/arch/arc/kernel/time.c
@@ -203,34 +203,24 @@ static int arc_clkevent_set_next_event(unsigned long delta,
return 0;
}
-static void arc_clkevent_set_mode(enum clock_event_mode mode,
- struct clock_event_device *dev)
+static int arc_clkevent_set_periodic(struct clock_event_device *dev)
{
- switch (mode) {
- case CLOCK_EVT_MODE_PERIODIC:
- /*
- * At X Hz, 1 sec = 1000ms -> X cycles;
- * 10ms -> X / 100 cycles
- */
- arc_timer_event_setup(arc_get_core_freq() / HZ);
- break;
- case CLOCK_EVT_MODE_ONESHOT:
- break;
- default:
- break;
- }
-
- return;
+ /*
+ * At X Hz, 1 sec = 1000ms -> X cycles;
+ * 10ms -> X / 100 cycles
+ */
+ arc_timer_event_setup(arc_get_core_freq() / HZ);
+ return 0;
}
static DEFINE_PER_CPU(struct clock_event_device, arc_clockevent_device) = {
- .name = "ARC Timer0",
- .features = CLOCK_EVT_FEAT_ONESHOT | CLOCK_EVT_FEAT_PERIODIC,
- .mode = CLOCK_EVT_MODE_UNUSED,
- .rating = 300,
- .irq = TIMER0_IRQ, /* hardwired, no need for resources */
- .set_next_event = arc_clkevent_set_next_event,
- .set_mode = arc_clkevent_set_mode,
+ .name = "ARC Timer0",
+ .features = CLOCK_EVT_FEAT_ONESHOT |
+ CLOCK_EVT_FEAT_PERIODIC,
+ .rating = 300,
+ .irq = TIMER0_IRQ, /* hardwired, no need for resources */
+ .set_next_event = arc_clkevent_set_next_event,
+ .set_state_periodic = arc_clkevent_set_periodic,
};
static irqreturn_t timer_irq_handler(int irq, void *dev_id)
@@ -240,7 +230,7 @@ static irqreturn_t timer_irq_handler(int irq, void *dev_id)
* irq_set_chip_and_handler() asked for handle_percpu_devid_irq()
*/
struct clock_event_device *evt = this_cpu_ptr(&arc_clockevent_device);
- int irq_reenable = evt->mode == CLOCK_EVT_MODE_PERIODIC;
+ int irq_reenable = clockevent_state_periodic(evt);
/*
* Any write to CTRL reg ACks the interrupt, we rewrite the
diff --git a/arch/arc/kernel/troubleshoot.c b/arch/arc/kernel/troubleshoot.c
index 807f7d61d7a7..a6f91e88ce36 100644
--- a/arch/arc/kernel/troubleshoot.c
+++ b/arch/arc/kernel/troubleshoot.c
@@ -58,7 +58,6 @@ static void show_callee_regs(struct callee_regs *cregs)
static void print_task_path_n_nm(struct task_struct *tsk, char *buf)
{
- struct path path;
char *path_nm = NULL;
struct mm_struct *mm;
struct file *exe_file;
diff --git a/arch/arc/kernel/unaligned.c b/arch/arc/kernel/unaligned.c
index 74db59b6f392..abd961f3e763 100644
--- a/arch/arc/kernel/unaligned.c
+++ b/arch/arc/kernel/unaligned.c
@@ -34,7 +34,7 @@
" .section .fixup,\"ax\"\n" \
" .align 4\n" \
"3: mov %0, 1\n" \
- " b 2b\n" \
+ " j 2b\n" \
" .previous\n" \
" .section __ex_table,\"a\"\n" \
" .align 4\n" \
@@ -82,7 +82,7 @@
" .section .fixup,\"ax\"\n" \
" .align 4\n" \
"4: mov %0, 1\n" \
- " b 3b\n" \
+ " j 3b\n" \
" .previous\n" \
" .section __ex_table,\"a\"\n" \
" .align 4\n" \
@@ -113,7 +113,7 @@
" .section .fixup,\"ax\"\n" \
" .align 4\n" \
"6: mov %0, 1\n" \
- " b 5b\n" \
+ " j 5b\n" \
" .previous\n" \
" .section __ex_table,\"a\"\n" \
" .align 4\n" \
diff --git a/arch/arc/lib/memcpy-archs.S b/arch/arc/lib/memcpy-archs.S
index 1b2b3acfed52..0cab0b8a57c5 100644
--- a/arch/arc/lib/memcpy-archs.S
+++ b/arch/arc/lib/memcpy-archs.S
@@ -206,7 +206,7 @@ unalignedOffby3:
ld.ab r6, [r1, 4]
prefetch [r1, 28] ;Prefetch the next read location
ld.ab r8, [r1,4]
- prefetch [r3, 32] ;Prefetch the next write location
+ prefetchw [r3, 32] ;Prefetch the next write location
SHIFT_1 (r7, r6, 8)
or r7, r7, r5
diff --git a/arch/arc/lib/memset-archs.S b/arch/arc/lib/memset-archs.S
index 92d573c734b5..365b18364815 100644
--- a/arch/arc/lib/memset-archs.S
+++ b/arch/arc/lib/memset-archs.S
@@ -10,12 +10,6 @@
#undef PREALLOC_NOT_AVAIL
-#ifdef PREALLOC_NOT_AVAIL
-#define PREWRITE(A,B) prefetchw [(A),(B)]
-#else
-#define PREWRITE(A,B) prealloc [(A),(B)]
-#endif
-
ENTRY(memset)
prefetchw [r0] ; Prefetch the write location
mov.f 0, r2
@@ -51,9 +45,15 @@ ENTRY(memset)
;;; Convert len to Dwords, unfold x8
lsr.f lp_count, lp_count, 6
+
lpnz @.Lset64bytes
;; LOOP START
- PREWRITE(r3, 64) ;Prefetch the next write location
+#ifdef PREALLOC_NOT_AVAIL
+ prefetchw [r3, 64] ;Prefetch the next write location
+#else
+ prealloc [r3, 64]
+#endif
+#ifdef CONFIG_ARC_HAS_LL64
std.ab r4, [r3, 8]
std.ab r4, [r3, 8]
std.ab r4, [r3, 8]
@@ -62,16 +62,45 @@ ENTRY(memset)
std.ab r4, [r3, 8]
std.ab r4, [r3, 8]
std.ab r4, [r3, 8]
+#else
+ st.ab r4, [r3, 4]
+ st.ab r4, [r3, 4]
+ st.ab r4, [r3, 4]
+ st.ab r4, [r3, 4]
+ st.ab r4, [r3, 4]
+ st.ab r4, [r3, 4]
+ st.ab r4, [r3, 4]
+ st.ab r4, [r3, 4]
+ st.ab r4, [r3, 4]
+ st.ab r4, [r3, 4]
+ st.ab r4, [r3, 4]
+ st.ab r4, [r3, 4]
+ st.ab r4, [r3, 4]
+ st.ab r4, [r3, 4]
+ st.ab r4, [r3, 4]
+ st.ab r4, [r3, 4]
+#endif
.Lset64bytes:
lsr.f lp_count, r2, 5 ;Last remaining max 124 bytes
lpnz .Lset32bytes
;; LOOP START
prefetchw [r3, 32] ;Prefetch the next write location
+#ifdef CONFIG_ARC_HAS_LL64
std.ab r4, [r3, 8]
std.ab r4, [r3, 8]
std.ab r4, [r3, 8]
std.ab r4, [r3, 8]
+#else
+ st.ab r4, [r3, 4]
+ st.ab r4, [r3, 4]
+ st.ab r4, [r3, 4]
+ st.ab r4, [r3, 4]
+ st.ab r4, [r3, 4]
+ st.ab r4, [r3, 4]
+ st.ab r4, [r3, 4]
+ st.ab r4, [r3, 4]
+#endif
.Lset32bytes:
and.f lp_count, r2, 0x1F ;Last remaining 31 bytes
diff --git a/arch/arc/mm/cache.c b/arch/arc/mm/cache.c
index b29d62ed4f7e..0d1a6e96839f 100644
--- a/arch/arc/mm/cache.c
+++ b/arch/arc/mm/cache.c
@@ -22,15 +22,22 @@
#include <asm/setup.h>
static int l2_line_sz;
+int ioc_exists;
+volatile int slc_enable = 1, ioc_enable = 1;
void (*_cache_line_loop_ic_fn)(unsigned long paddr, unsigned long vaddr,
unsigned long sz, const int cacheop);
+void (*__dma_cache_wback_inv)(unsigned long start, unsigned long sz);
+void (*__dma_cache_inv)(unsigned long start, unsigned long sz);
+void (*__dma_cache_wback)(unsigned long start, unsigned long sz);
+
char *arc_cache_mumbojumbo(int c, char *buf, int len)
{
int n = 0;
struct cpuinfo_arc_cache *p;
+#define IS_USED_RUN(v) ((v) ? "" : "(disabled) ")
#define PR_CACHE(p, cfg, str) \
if (!(p)->ver) \
n += scnprintf(buf + n, len - n, str"\t\t: N/A\n"); \
@@ -45,10 +52,18 @@ char *arc_cache_mumbojumbo(int c, char *buf, int len)
PR_CACHE(&cpuinfo_arc700[c].icache, CONFIG_ARC_HAS_ICACHE, "I-Cache");
PR_CACHE(&cpuinfo_arc700[c].dcache, CONFIG_ARC_HAS_DCACHE, "D-Cache");
+ if (!is_isa_arcv2())
+ return buf;
+
p = &cpuinfo_arc700[c].slc;
if (p->ver)
n += scnprintf(buf + n, len - n,
- "SLC\t\t: %uK, %uB Line\n", p->sz_k, p->line_len);
+ "SLC\t\t: %uK, %uB Line%s\n",
+ p->sz_k, p->line_len, IS_USED_RUN(slc_enable));
+
+ if (ioc_exists)
+ n += scnprintf(buf + n, len - n, "IOC\t\t:%s\n",
+ IS_USED_RUN(ioc_enable));
return buf;
}
@@ -58,18 +73,9 @@ char *arc_cache_mumbojumbo(int c, char *buf, int len)
* the cpuinfo structure for later use.
* No Validation done here, simply read/convert the BCRs
*/
-void read_decode_cache_bcr(void)
+static void read_decode_cache_bcr_arcv2(int cpu)
{
- struct cpuinfo_arc_cache *p_ic, *p_dc, *p_slc;
- unsigned int cpu = smp_processor_id();
- struct bcr_cache {
-#ifdef CONFIG_CPU_BIG_ENDIAN
- unsigned int pad:12, line_len:4, sz:4, config:4, ver:8;
-#else
- unsigned int ver:8, config:4, sz:4, line_len:4, pad:12;
-#endif
- } ibcr, dbcr;
-
+ struct cpuinfo_arc_cache *p_slc = &cpuinfo_arc700[cpu].slc;
struct bcr_generic sbcr;
struct bcr_slc_cfg {
@@ -80,6 +86,39 @@ void read_decode_cache_bcr(void)
#endif
} slc_cfg;
+ struct bcr_clust_cfg {
+#ifdef CONFIG_CPU_BIG_ENDIAN
+ unsigned int pad:7, c:1, num_entries:8, num_cores:8, ver:8;
+#else
+ unsigned int ver:8, num_cores:8, num_entries:8, c:1, pad:7;
+#endif
+ } cbcr;
+
+ READ_BCR(ARC_REG_SLC_BCR, sbcr);
+ if (sbcr.ver) {
+ READ_BCR(ARC_REG_SLC_CFG, slc_cfg);
+ p_slc->ver = sbcr.ver;
+ p_slc->sz_k = 128 << slc_cfg.sz;
+ l2_line_sz = p_slc->line_len = (slc_cfg.lsz == 0) ? 128 : 64;
+ }
+
+ READ_BCR(ARC_REG_CLUSTER_BCR, cbcr);
+ if (cbcr.c && ioc_enable)
+ ioc_exists = 1;
+}
+
+void read_decode_cache_bcr(void)
+{
+ struct cpuinfo_arc_cache *p_ic, *p_dc;
+ unsigned int cpu = smp_processor_id();
+ struct bcr_cache {
+#ifdef CONFIG_CPU_BIG_ENDIAN
+ unsigned int pad:12, line_len:4, sz:4, config:4, ver:8;
+#else
+ unsigned int ver:8, config:4, sz:4, line_len:4, pad:12;
+#endif
+ } ibcr, dbcr;
+
p_ic = &cpuinfo_arc700[cpu].icache;
READ_BCR(ARC_REG_IC_BCR, ibcr);
@@ -122,17 +161,8 @@ dc_chk:
p_dc->ver = dbcr.ver;
slc_chk:
- if (!is_isa_arcv2())
- return;
-
- p_slc = &cpuinfo_arc700[cpu].slc;
- READ_BCR(ARC_REG_SLC_BCR, sbcr);
- if (sbcr.ver) {
- READ_BCR(ARC_REG_SLC_CFG, slc_cfg);
- p_slc->ver = sbcr.ver;
- p_slc->sz_k = 128 << slc_cfg.sz;
- l2_line_sz = p_slc->line_len = (slc_cfg.lsz == 0) ? 128 : 64;
- }
+ if (is_isa_arcv2())
+ read_decode_cache_bcr_arcv2(cpu);
}
/*
@@ -468,10 +498,18 @@ static void __ic_line_inv_vaddr(unsigned long paddr, unsigned long vaddr,
noinline void slc_op(unsigned long paddr, unsigned long sz, const int op)
{
#ifdef CONFIG_ISA_ARCV2
+ /*
+ * SLC is shared between all cores and concurrent aux operations from
+ * multiple cores need to be serialized using a spinlock
+ * A concurrent operation can be silently ignored and/or the old/new
+ * operation can remain incomplete forever (lockup in SLC_CTRL_BUSY loop
+ * below)
+ */
+ static DEFINE_SPINLOCK(lock);
unsigned long flags;
unsigned int ctrl;
- local_irq_save(flags);
+ spin_lock_irqsave(&lock, flags);
/*
* The Region Flush operation is specified by CTRL.RGN_OP[11..9]
@@ -504,15 +542,10 @@ noinline void slc_op(unsigned long paddr, unsigned long sz, const int op)
while (read_aux_reg(ARC_REG_SLC_CTRL) & SLC_CTRL_BUSY);
- local_irq_restore(flags);
+ spin_unlock_irqrestore(&lock, flags);
#endif
}
-static inline int need_slc_flush(void)
-{
- return is_isa_arcv2() && l2_line_sz;
-}
-
/***********************************************************
* Exported APIs
*/
@@ -561,30 +594,74 @@ void flush_dcache_page(struct page *page)
}
EXPORT_SYMBOL(flush_dcache_page);
-void dma_cache_wback_inv(unsigned long start, unsigned long sz)
+/*
+ * DMA ops for systems with L1 cache only
+ * Make memory coherent with L1 cache by flushing/invalidating L1 lines
+ */
+static void __dma_cache_wback_inv_l1(unsigned long start, unsigned long sz)
{
__dc_line_op_k(start, sz, OP_FLUSH_N_INV);
+}
- if (need_slc_flush())
- slc_op(start, sz, OP_FLUSH_N_INV);
+static void __dma_cache_inv_l1(unsigned long start, unsigned long sz)
+{
+ __dc_line_op_k(start, sz, OP_INV);
}
-EXPORT_SYMBOL(dma_cache_wback_inv);
-void dma_cache_inv(unsigned long start, unsigned long sz)
+static void __dma_cache_wback_l1(unsigned long start, unsigned long sz)
+{
+ __dc_line_op_k(start, sz, OP_FLUSH);
+}
+
+/*
+ * DMA ops for systems with both L1 and L2 caches, but without IOC
+ * Both L1 and L2 lines need to be explicity flushed/invalidated
+ */
+static void __dma_cache_wback_inv_slc(unsigned long start, unsigned long sz)
+{
+ __dc_line_op_k(start, sz, OP_FLUSH_N_INV);
+ slc_op(start, sz, OP_FLUSH_N_INV);
+}
+
+static void __dma_cache_inv_slc(unsigned long start, unsigned long sz)
{
__dc_line_op_k(start, sz, OP_INV);
+ slc_op(start, sz, OP_INV);
+}
+
+static void __dma_cache_wback_slc(unsigned long start, unsigned long sz)
+{
+ __dc_line_op_k(start, sz, OP_FLUSH);
+ slc_op(start, sz, OP_FLUSH);
+}
- if (need_slc_flush())
- slc_op(start, sz, OP_INV);
+/*
+ * DMA ops for systems with IOC
+ * IOC hardware snoops all DMA traffic keeping the caches consistent with
+ * memory - eliding need for any explicit cache maintenance of DMA buffers
+ */
+static void __dma_cache_wback_inv_ioc(unsigned long start, unsigned long sz) {}
+static void __dma_cache_inv_ioc(unsigned long start, unsigned long sz) {}
+static void __dma_cache_wback_ioc(unsigned long start, unsigned long sz) {}
+
+/*
+ * Exported DMA API
+ */
+void dma_cache_wback_inv(unsigned long start, unsigned long sz)
+{
+ __dma_cache_wback_inv(start, sz);
+}
+EXPORT_SYMBOL(dma_cache_wback_inv);
+
+void dma_cache_inv(unsigned long start, unsigned long sz)
+{
+ __dma_cache_inv(start, sz);
}
EXPORT_SYMBOL(dma_cache_inv);
void dma_cache_wback(unsigned long start, unsigned long sz)
{
- __dc_line_op_k(start, sz, OP_FLUSH);
-
- if (need_slc_flush())
- slc_op(start, sz, OP_FLUSH);
+ __dma_cache_wback(start, sz);
}
EXPORT_SYMBOL(dma_cache_wback);
@@ -840,4 +917,41 @@ void arc_cache_init(void)
panic("Disable CONFIG_ARC_CACHE_VIPT_ALIASING\n");
}
}
+
+ if (is_isa_arcv2() && l2_line_sz && !slc_enable) {
+
+ /* IM set : flush before invalidate */
+ write_aux_reg(ARC_REG_SLC_CTRL,
+ read_aux_reg(ARC_REG_SLC_CTRL) | SLC_CTRL_IM);
+
+ write_aux_reg(ARC_REG_SLC_INVALIDATE, 1);
+
+ /* Important to wait for flush to complete */
+ while (read_aux_reg(ARC_REG_SLC_CTRL) & SLC_CTRL_BUSY);
+ write_aux_reg(ARC_REG_SLC_CTRL,
+ read_aux_reg(ARC_REG_SLC_CTRL) | SLC_CTRL_DISABLE);
+ }
+
+ if (is_isa_arcv2() && ioc_exists) {
+ /* IO coherency base - 0x8z */
+ write_aux_reg(ARC_REG_IO_COH_AP0_BASE, 0x80000);
+ /* IO coherency aperture size - 512Mb: 0x8z-0xAz */
+ write_aux_reg(ARC_REG_IO_COH_AP0_SIZE, 0x11);
+ /* Enable partial writes */
+ write_aux_reg(ARC_REG_IO_COH_PARTIAL, 1);
+ /* Enable IO coherency */
+ write_aux_reg(ARC_REG_IO_COH_ENABLE, 1);
+
+ __dma_cache_wback_inv = __dma_cache_wback_inv_ioc;
+ __dma_cache_inv = __dma_cache_inv_ioc;
+ __dma_cache_wback = __dma_cache_wback_ioc;
+ } else if (is_isa_arcv2() && l2_line_sz && slc_enable) {
+ __dma_cache_wback_inv = __dma_cache_wback_inv_slc;
+ __dma_cache_inv = __dma_cache_inv_slc;
+ __dma_cache_wback = __dma_cache_wback_slc;
+ } else {
+ __dma_cache_wback_inv = __dma_cache_wback_inv_l1;
+ __dma_cache_inv = __dma_cache_inv_l1;
+ __dma_cache_wback = __dma_cache_wback_l1;
+ }
}
diff --git a/arch/arc/mm/dma.c b/arch/arc/mm/dma.c
index 74a637a1cfc4..29a46bb198cc 100644
--- a/arch/arc/mm/dma.c
+++ b/arch/arc/mm/dma.c
@@ -19,6 +19,7 @@
#include <linux/dma-mapping.h>
#include <linux/dma-debug.h>
#include <linux/export.h>
+#include <asm/cache.h>
#include <asm/cacheflush.h>
/*
@@ -53,6 +54,20 @@ void *dma_alloc_coherent(struct device *dev, size_t size,
{
void *paddr, *kvaddr;
+ /*
+ * IOC relies on all data (even coherent DMA data) being in cache
+ * Thus allocate normal cached memory
+ *
+ * The gains with IOC are two pronged:
+ * -For streaming data, elides needs for cache maintenance, saving
+ * cycles in flush code, and bus bandwidth as all the lines of a
+ * buffer need to be flushed out to memory
+ * -For coherent data, Read/Write to buffers terminate early in cache
+ * (vs. always going to memory - thus are faster)
+ */
+ if (is_isa_arcv2() && ioc_exists)
+ return dma_alloc_noncoherent(dev, size, dma_handle, gfp);
+
/* This is linear addr (0x8000_0000 based) */
paddr = alloc_pages_exact(size, gfp);
if (!paddr)
@@ -60,8 +75,8 @@ void *dma_alloc_coherent(struct device *dev, size_t size,
/* This is kernel Virtual address (0x7000_0000 based) */
kvaddr = ioremap_nocache((unsigned long)paddr, size);
- if (kvaddr != NULL)
- memset(kvaddr, 0, size);
+ if (kvaddr == NULL)
+ return NULL;
/* This is bus address, platform dependent */
*dma_handle = (dma_addr_t)paddr;
@@ -85,6 +100,9 @@ EXPORT_SYMBOL(dma_alloc_coherent);
void dma_free_coherent(struct device *dev, size_t size, void *kvaddr,
dma_addr_t dma_handle)
{
+ if (is_isa_arcv2() && ioc_exists)
+ return dma_free_noncoherent(dev, size, kvaddr, dma_handle);
+
iounmap((void __force __iomem *)kvaddr);
free_pages_exact((void *)dma_handle, size);
diff --git a/arch/arc/plat-axs10x/axs10x.c b/arch/arc/plat-axs10x/axs10x.c
index 99f7da513a48..0a77b19e1df8 100644
--- a/arch/arc/plat-axs10x/axs10x.c
+++ b/arch/arc/plat-axs10x/axs10x.c
@@ -46,7 +46,7 @@ static void __init axs10x_enable_gpio_intc_wire(void)
* ------------------- -------------------
* | snps,dw-apb-gpio | | snps,dw-apb-gpio |
* ------------------- -------------------
- * | |
+ * | #12 |
* | [ Debug UART on cpu card ]
* |
* ------------------------
@@ -389,6 +389,23 @@ axs103_set_freq(unsigned int id, unsigned int fd, unsigned int od)
static void __init axs103_early_init(void)
{
+ /*
+ * AXS103 configurations for SMP/QUAD configurations share device tree
+ * which defaults to 90 MHz. However recent failures of Quad config
+ * revealed P&R timing violations so clamp it down to safe 50 MHz
+ * Instead of duplicating defconfig/DT for SMP/QUAD, add a small hack
+ *
+ * This hack is really hacky as of now. Fix it properly by getting the
+ * number of cores as return value of platform's early SMP callback
+ */
+#ifdef CONFIG_ARC_MCIP
+ unsigned int num_cores = (read_aux_reg(ARC_REG_MCIP_BCR) >> 16) & 0x3F;
+ if (num_cores > 2)
+ arc_set_core_freq(50 * 1000000);
+ else if (num_cores == 2)
+ arc_set_core_freq(75 * 1000000);
+#endif
+
switch (arc_get_core_freq()/1000000) {
case 33:
axs103_set_freq(1, 1, 1);