perf_event.h 6.72 KB
Newer Older
1
/*
2 3
 * Linux performance counter support for ARC
 *
4
 * Copyright (C) 2014-2015 Synopsys, Inc. (www.synopsys.com)
5
 * Copyright (C) 2011-2013 Synopsys, Inc. (www.synopsys.com)
6 7 8 9 10 11 12 13 14 15
 *
 * 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_PERF_EVENT_H
#define __ASM_PERF_EVENT_H

16 17
/* Max number of counters that PCT block may ever have */
#define ARC_PERF_MAX_COUNTERS	32
18 19 20 21 22 23 24 25 26 27 28 29 30 31

#define ARC_REG_CC_BUILD	0xF6
#define ARC_REG_CC_INDEX	0x240
#define ARC_REG_CC_NAME0	0x241
#define ARC_REG_CC_NAME1	0x242

#define ARC_REG_PCT_BUILD	0xF5
#define ARC_REG_PCT_COUNTL	0x250
#define ARC_REG_PCT_COUNTH	0x251
#define ARC_REG_PCT_SNAPL	0x252
#define ARC_REG_PCT_SNAPH	0x253
#define ARC_REG_PCT_CONFIG	0x254
#define ARC_REG_PCT_CONTROL	0x255
#define ARC_REG_PCT_INDEX	0x256
32 33 34 35
#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
36

37 38 39
#define ARC_REG_PCT_CONFIG_USER	(1 << 18)	/* count in user mode */
#define ARC_REG_PCT_CONFIG_KERN	(1 << 19)	/* count in kernel mode */

40 41 42 43 44
#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
45
	unsigned int m:8, c:8, r:5, i:1, s:2, v:8;
46
#else
47
	unsigned int v:8, s:2, i:1, r:5, c:8, m:8;
48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64
#endif
};

struct arc_reg_cc_build {
#ifdef CONFIG_CPU_BIG_ENDIAN
	unsigned int c:16, r:8, v:8;
#else
	unsigned int v:8, r:8, c:16;
#endif
};

#define PERF_COUNT_ARC_DCLM	(PERF_COUNT_HW_MAX + 0)
#define PERF_COUNT_ARC_DCSM	(PERF_COUNT_HW_MAX + 1)
#define PERF_COUNT_ARC_ICM	(PERF_COUNT_HW_MAX + 2)
#define PERF_COUNT_ARC_BPOK	(PERF_COUNT_HW_MAX + 3)
#define PERF_COUNT_ARC_EDTLB	(PERF_COUNT_HW_MAX + 4)
#define PERF_COUNT_ARC_EITLB	(PERF_COUNT_HW_MAX + 5)
65 66 67 68
#define PERF_COUNT_ARC_LDC	(PERF_COUNT_HW_MAX + 6)
#define PERF_COUNT_ARC_STC	(PERF_COUNT_HW_MAX + 7)

#define PERF_COUNT_ARC_HW_MAX	(PERF_COUNT_HW_MAX + 8)
69 70

/*
71
 * Some ARC pct quirks:
72 73 74 75 76 77 78 79 80 81 82 83 84 85
 *
 * PERF_COUNT_HW_STALLED_CYCLES_BACKEND
 * PERF_COUNT_HW_STALLED_CYCLES_FRONTEND
 *	The ARC 700 can either measure stalls per pipeline stage, or all stalls
 *	combined; for now we assign all stalls to STALLED_CYCLES_BACKEND
 *	and all pipeline flushes (e.g. caused by mispredicts, etc.) to
 *	STALLED_CYCLES_FRONTEND.
 *
 *	We could start multiple performance counters and combine everything
 *	afterwards, but that makes it complicated.
 *
 *	Note that I$ cache misses aren't counted by either of the two!
 */

86 87 88 89 90 91 92 93 94
/*
 * ARC PCT has hardware conditions with fixed "names" but variable "indexes"
 * (based on a specific RTL build)
 * Below is the static map between perf generic/arc specific event_id and
 * h/w condition names.
 * At the time of probe, we loop thru each index and find it's name to
 * complete the mapping of perf event_id to h/w index as latter is needed
 * to program the counter really
 */
95
static const char * const arc_pmu_ev_hw_map[] = {
96
	/* count cycles */
97 98 99
	[PERF_COUNT_HW_CPU_CYCLES] = "crun",
	[PERF_COUNT_HW_REF_CPU_CYCLES] = "crun",
	[PERF_COUNT_HW_BUS_CYCLES] = "crun",
100

101 102
	[PERF_COUNT_HW_STALLED_CYCLES_FRONTEND] = "bflush",
	[PERF_COUNT_HW_STALLED_CYCLES_BACKEND] = "bstall",
103 104 105

	/* counts condition */
	[PERF_COUNT_HW_INSTRUCTIONS] = "iall",
106 107
	/* All jump instructions that are taken */
	[PERF_COUNT_HW_BRANCH_INSTRUCTIONS] = "ijmptak",
108
	[PERF_COUNT_ARC_BPOK]         = "bpok",	  /* NP-NT, PT-T, PNT-NT */
109 110 111
#ifdef CONFIG_ISA_ARCV2
	[PERF_COUNT_HW_BRANCH_MISSES] = "bpmp",
#else
112
	[PERF_COUNT_HW_BRANCH_MISSES] = "bpfail", /* NP-T, PT-NT, PNT-T */
113
#endif
114 115 116
	[PERF_COUNT_ARC_LDC] = "imemrdc",	/* Instr: mem read cached */
	[PERF_COUNT_ARC_STC] = "imemwrc",	/* Instr: mem write cached */

117 118 119 120 121
	[PERF_COUNT_ARC_DCLM] = "dclm",		/* D-cache Load Miss */
	[PERF_COUNT_ARC_DCSM] = "dcsm",		/* D-cache Store Miss */
	[PERF_COUNT_ARC_ICM] = "icm",		/* I-cache Miss */
	[PERF_COUNT_ARC_EDTLB] = "edtlb",	/* D-TLB Miss */
	[PERF_COUNT_ARC_EITLB] = "eitlb",	/* I-TLB Miss */
122 123 124 125 126 127 128 129
};

#define C(_x)			PERF_COUNT_HW_CACHE_##_x
#define CACHE_OP_UNSUPPORTED	0xffff

static const unsigned arc_pmu_cache_map[C(MAX)][C(OP_MAX)][C(RESULT_MAX)] = {
	[C(L1D)] = {
		[C(OP_READ)] = {
130
			[C(RESULT_ACCESS)]	= PERF_COUNT_ARC_LDC,
131 132 133
			[C(RESULT_MISS)]	= PERF_COUNT_ARC_DCLM,
		},
		[C(OP_WRITE)] = {
134
			[C(RESULT_ACCESS)]	= PERF_COUNT_ARC_STC,
135 136 137 138 139 140 141 142 143
			[C(RESULT_MISS)]	= PERF_COUNT_ARC_DCSM,
		},
		[C(OP_PREFETCH)] = {
			[C(RESULT_ACCESS)]	= CACHE_OP_UNSUPPORTED,
			[C(RESULT_MISS)]	= CACHE_OP_UNSUPPORTED,
		},
	},
	[C(L1I)] = {
		[C(OP_READ)] = {
144
			[C(RESULT_ACCESS)]	= PERF_COUNT_HW_INSTRUCTIONS,
145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171
			[C(RESULT_MISS)]	= PERF_COUNT_ARC_ICM,
		},
		[C(OP_WRITE)] = {
			[C(RESULT_ACCESS)]	= CACHE_OP_UNSUPPORTED,
			[C(RESULT_MISS)]	= CACHE_OP_UNSUPPORTED,
		},
		[C(OP_PREFETCH)] = {
			[C(RESULT_ACCESS)]	= CACHE_OP_UNSUPPORTED,
			[C(RESULT_MISS)]	= CACHE_OP_UNSUPPORTED,
		},
	},
	[C(LL)] = {
		[C(OP_READ)] = {
			[C(RESULT_ACCESS)]	= CACHE_OP_UNSUPPORTED,
			[C(RESULT_MISS)]	= CACHE_OP_UNSUPPORTED,
		},
		[C(OP_WRITE)] = {
			[C(RESULT_ACCESS)]	= CACHE_OP_UNSUPPORTED,
			[C(RESULT_MISS)]	= CACHE_OP_UNSUPPORTED,
		},
		[C(OP_PREFETCH)] = {
			[C(RESULT_ACCESS)]	= CACHE_OP_UNSUPPORTED,
			[C(RESULT_MISS)]	= CACHE_OP_UNSUPPORTED,
		},
	},
	[C(DTLB)] = {
		[C(OP_READ)] = {
172
			[C(RESULT_ACCESS)]	= PERF_COUNT_ARC_LDC,
173 174
			[C(RESULT_MISS)]	= PERF_COUNT_ARC_EDTLB,
		},
175
			/* DTLB LD/ST Miss not segregated by h/w*/
176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228
		[C(OP_WRITE)] = {
			[C(RESULT_ACCESS)]	= CACHE_OP_UNSUPPORTED,
			[C(RESULT_MISS)]	= CACHE_OP_UNSUPPORTED,
		},
		[C(OP_PREFETCH)] = {
			[C(RESULT_ACCESS)]	= CACHE_OP_UNSUPPORTED,
			[C(RESULT_MISS)]	= CACHE_OP_UNSUPPORTED,
		},
	},
	[C(ITLB)] = {
		[C(OP_READ)] = {
			[C(RESULT_ACCESS)]	= CACHE_OP_UNSUPPORTED,
			[C(RESULT_MISS)]	= PERF_COUNT_ARC_EITLB,
		},
		[C(OP_WRITE)] = {
			[C(RESULT_ACCESS)]	= CACHE_OP_UNSUPPORTED,
			[C(RESULT_MISS)]	= CACHE_OP_UNSUPPORTED,
		},
		[C(OP_PREFETCH)] = {
			[C(RESULT_ACCESS)]	= CACHE_OP_UNSUPPORTED,
			[C(RESULT_MISS)]	= CACHE_OP_UNSUPPORTED,
		},
	},
	[C(BPU)] = {
		[C(OP_READ)] = {
			[C(RESULT_ACCESS)] = PERF_COUNT_HW_BRANCH_INSTRUCTIONS,
			[C(RESULT_MISS)]	= PERF_COUNT_HW_BRANCH_MISSES,
		},
		[C(OP_WRITE)] = {
			[C(RESULT_ACCESS)]	= CACHE_OP_UNSUPPORTED,
			[C(RESULT_MISS)]	= CACHE_OP_UNSUPPORTED,
		},
		[C(OP_PREFETCH)] = {
			[C(RESULT_ACCESS)]	= CACHE_OP_UNSUPPORTED,
			[C(RESULT_MISS)]	= CACHE_OP_UNSUPPORTED,
		},
	},
	[C(NODE)] = {
		[C(OP_READ)] = {
			[C(RESULT_ACCESS)]	= CACHE_OP_UNSUPPORTED,
			[C(RESULT_MISS)]	= CACHE_OP_UNSUPPORTED,
		},
		[C(OP_WRITE)] = {
			[C(RESULT_ACCESS)]	= CACHE_OP_UNSUPPORTED,
			[C(RESULT_MISS)]	= CACHE_OP_UNSUPPORTED,
		},
		[C(OP_PREFETCH)] = {
			[C(RESULT_ACCESS)]	= CACHE_OP_UNSUPPORTED,
			[C(RESULT_MISS)]	= CACHE_OP_UNSUPPORTED,
		},
	},
};

229
#endif /* __ASM_PERF_EVENT_H */