parse-events.l

  1
  2%option reentrant
  3%option bison-bridge
  4%option prefix="parse_events_"
  5%option stack
  6%option bison-locations
  7%option yylineno
  8%option noyywrap
  9
 10%{
 11#include <errno.h>
 12#include <stdlib.h>
 13#include <stdio.h>
 14#include "parse-events.h"
 15#include "parse-events-bison.h"
 16
 17char *parse_events_get_text(yyscan_t yyscanner);
 18YYSTYPE *parse_events_get_lval(yyscan_t yyscanner);
 19int parse_events_get_column(yyscan_t yyscanner);
 20int parse_events_get_leng(yyscan_t yyscanner);
 21
 22static int get_column(yyscan_t scanner)
 23{
 24	return parse_events_get_column(scanner) - parse_events_get_leng(scanner);
 25}
 26
 27static int value(struct parse_events_state *parse_state, yyscan_t scanner, int base)
 28{
 29	YYSTYPE *yylval = parse_events_get_lval(scanner);
 30	char *text = parse_events_get_text(scanner);
 31	u64 num;
 32
 33	errno = 0;
 34	num = strtoull(text, NULL, base);
 35	if (errno) {
 36		struct parse_events_error *error = parse_state->error;
 37		char *help = NULL;
 38
 39		if (asprintf(&help, "Bad base %d number \"%s\"", base, text) > 0)
 40			parse_events_error__handle(error, get_column(scanner), help , NULL);
 41
 42		return PE_ERROR;
 43	}
 44
 45	yylval->num = num;
 46	return PE_VALUE;
 47}
 48
 49static int str(yyscan_t scanner, int token)
 50{
 51	YYSTYPE *yylval = parse_events_get_lval(scanner);
 52	char *text = parse_events_get_text(scanner);
 53
 54	yylval->str = strdup(text);
 55	return token;
 56}
 57
 58static int quoted_str(yyscan_t scanner, int token)
 59{
 60	YYSTYPE *yylval = parse_events_get_lval(scanner);
 61	char *text = parse_events_get_text(scanner);
 62
 63	/*
 64	 * If a text tag specified on the command line
 65	 * contains opening single quite ' then it is
 66	 * expected that the tag ends with single quote
 67	 * as well, like this:
 68	 *     name=\'CPU_CLK_UNHALTED.THREAD:cmask=1\'
 69	 * quotes need to be escaped to bypass shell
 70	 * processing.
 71	 */
 72	yylval->str = strndup(&text[1], strlen(text) - 2);
 73	return token;
 74}
 75
 76/*
 77 * This function is called when the parser gets two kind of input:
 78 *
 79 * 	@cfg1 or @cfg2=config
 80 *
 81 * The leading '@' is stripped off before 'cfg1' and 'cfg2=config' are given to
 82 * bison.  In the latter case it is necessary to keep the string intact so that
 83 * the PMU kernel driver can determine what configurable is associated to
 84 * 'config'.
 85 */
 86static int drv_str(yyscan_t scanner, int token)
 87{
 88	YYSTYPE *yylval = parse_events_get_lval(scanner);
 89	char *text = parse_events_get_text(scanner);
 90
 91	/* Strip off the '@' */
 92	yylval->str = strdup(text + 1);
 93	return token;
 94}
 95
 96/*
 97 * Use yyless to return all the characaters to the input. Update the column for
 98 * location debugging. If __alloc is non-zero set yylval to the text for the
 99 * returned token's value.
100 */
101#define REWIND(__alloc)				\
102do {								\
103	YYSTYPE *__yylval = parse_events_get_lval(yyscanner);	\
104	char *text = parse_events_get_text(yyscanner);		\
105								\
106	if (__alloc)						\
107		__yylval->str = strdup(text);			\
108								\
109	yycolumn -= strlen(text);				\
110	yyless(0);						\
111} while (0)
112
113static int term(yyscan_t scanner, enum parse_events__term_type type)
114{
115	YYSTYPE *yylval = parse_events_get_lval(scanner);
116
117	yylval->term_type = type;
118	return PE_TERM;
119}
120
121static void modifiers_error(struct parse_events_state *parse_state, yyscan_t scanner,
122			    int pos, char mod_char, const char *mod_name)
123{
124	struct parse_events_error *error = parse_state->error;
125	char *help = NULL;
126
127	if (asprintf(&help, "Duplicate modifier '%c' (%s)", mod_char, mod_name) > 0)
128		parse_events_error__handle(error, get_column(scanner) + pos, help , NULL);
129}
130
131static int modifiers(struct parse_events_state *parse_state, yyscan_t scanner)
132{
133	YYSTYPE *yylval = parse_events_get_lval(scanner);
134	char *text = parse_events_get_text(scanner);
135	struct parse_events_modifier mod = { .precise = 0, };
136
137	for (size_t i = 0, n = strlen(text); i < n; i++) {
138#define CASE(c, field)							\
139		case c:							\
140			if (mod.field) {				\
141				modifiers_error(parse_state, scanner, i, c, #field); \
142				return PE_ERROR;			\
143			}						\
144			mod.field = true;				\
145			break
146
147		switch (text[i]) {
148		CASE('u', user);
149		CASE('k', kernel);
150		CASE('h', hypervisor);
151		CASE('I', non_idle);
152		CASE('G', guest);
153		CASE('H', host);
154		case 'p':
155			mod.precise++;
156			/*
157			 * precise ip:
158			 *
159			 *  0 - SAMPLE_IP can have arbitrary skid
160			 *  1 - SAMPLE_IP must have constant skid
161			 *  2 - SAMPLE_IP requested to have 0 skid
162			 *  3 - SAMPLE_IP must have 0 skid
163			 *
164			 *  See also PERF_RECORD_MISC_EXACT_IP
165			 */
166			if (mod.precise > 3) {
167				struct parse_events_error *error = parse_state->error;
168				char *help = strdup("Maximum precise value is 3");
169
170				if (help) {
171					parse_events_error__handle(error, get_column(scanner) + i,
172								   help , NULL);
173				}
174				return PE_ERROR;
175			}
176			break;
177		CASE('P', precise_max);
178		CASE('S', sample_read);
179		CASE('D', pinned);
180		CASE('W', weak);
181		CASE('e', exclusive);
182		CASE('b', bpf);
183		CASE('R', retire_lat);
184		CASE('X', dont_regroup);
185		default:
186			return PE_ERROR;
187		}
188#undef CASE
189	}
190	yylval->mod = mod;
191	return PE_MODIFIER_EVENT;
192}
193
194#define YY_USER_ACTION					\
195do {							\
196	yylloc->last_column  = yylloc->first_column;	\
197	yylloc->first_column = yycolumn;		\
198	yycolumn += yyleng;				\
199} while (0);
200
201%}
202
203%x mem
204%s config
205%x event
206
207group		[^,{}/]*[{][^}]*[}][^,{}/]*
208event_pmu	[^,{}/]+[/][^/]*[/][^,{}/]*
209event		[^,{}/]+
210
211num_dec		[0-9]+
212num_hex		0x[a-fA-F0-9]{1,16}
213num_raw_hex	[a-fA-F0-9]{1,16}
214/* Regular pattern to match the token PE_NAME. */
215name_start      [a-zA-Z0-9_*?\[\]]
216name		{name_start}[a-zA-Z0-9_*?.\[\]!\-]*
217/* PE_NAME token when inside a config term list, allows ':'. */
218term_name	{name_start}[a-zA-Z0-9_*?.\[\]!\-:]*
219/*
220 * PE_NAME token when quoted, allows ':,.='.
221 * Matches the RHS of terms like: name='COMPLEX_CYCLES_NAME:orig=cycles,desc=chip-clock-ticks'.
222 */
223quoted_name	[\']{name_start}[a-zA-Z0-9_*?.\[\]!\-:,\.=]*[\']
224drv_cfg_term	[a-zA-Z0-9_\.]+(=[a-zA-Z0-9_*?\.:]+)?
225/*
226 * If you add a modifier you need to update modifiers().
227 * Also, the letters in modifier_event must not be in modifier_bp.
228 */
229modifier_event	[ukhpPGHSDIWebRX]{1,17}
230modifier_bp	[rwx]{1,3}
231digit		[0-9]
232non_digit	[^0-9]
233
234%%
235
236%{
237	struct parse_events_state *_parse_state = parse_events_get_extra(yyscanner);
238	{
239		int start_token = _parse_state->stoken;
240
241		if (start_token == PE_START_TERMS)
242			BEGIN(config);
243		else if (start_token == PE_START_EVENTS)
244			BEGIN(event);
245
246		if (start_token) {
247			_parse_state->stoken = 0;
248			/*
249			 * The flex parser does not init locations variable
250			 * via the scan_string interface, so we need do the
251			 * init in here.
252			 */
253			yycolumn = 0;
254			return start_token;
255		}
256         }
257%}
258
259<event>{
260
261{group}		{
262			BEGIN(INITIAL);
263			REWIND(0);
264		}
265
266{event_pmu}	|
267{event}		{
268			BEGIN(INITIAL);
269			REWIND(1);
270			return PE_EVENT_NAME;
271		}
272
273<<EOF>>		{
274			BEGIN(INITIAL);
275			REWIND(0);
276		}
277,		{
278			return ',';
279		}
280}
281
282<config>{
283	/*
284	 * Please update config_term_names when new static term is added.
285	 */
286config			{ return term(yyscanner, PARSE_EVENTS__TERM_TYPE_CONFIG); }
287config1			{ return term(yyscanner, PARSE_EVENTS__TERM_TYPE_CONFIG1); }
288config2			{ return term(yyscanner, PARSE_EVENTS__TERM_TYPE_CONFIG2); }
289config3			{ return term(yyscanner, PARSE_EVENTS__TERM_TYPE_CONFIG3); }
290config4			{ return term(yyscanner, PARSE_EVENTS__TERM_TYPE_CONFIG4); }
291name			{ return term(yyscanner, PARSE_EVENTS__TERM_TYPE_NAME); }
292period			{ return term(yyscanner, PARSE_EVENTS__TERM_TYPE_SAMPLE_PERIOD); }
293freq			{ return term(yyscanner, PARSE_EVENTS__TERM_TYPE_SAMPLE_FREQ); }
294branch_type		{ return term(yyscanner, PARSE_EVENTS__TERM_TYPE_BRANCH_SAMPLE_TYPE); }
295time			{ return term(yyscanner, PARSE_EVENTS__TERM_TYPE_TIME); }
296call-graph		{ return term(yyscanner, PARSE_EVENTS__TERM_TYPE_CALLGRAPH); }
297stack-size		{ return term(yyscanner, PARSE_EVENTS__TERM_TYPE_STACKSIZE); }
298max-stack		{ return term(yyscanner, PARSE_EVENTS__TERM_TYPE_MAX_STACK); }
299nr			{ return term(yyscanner, PARSE_EVENTS__TERM_TYPE_MAX_EVENTS); }
300inherit			{ return term(yyscanner, PARSE_EVENTS__TERM_TYPE_INHERIT); }
301no-inherit		{ return term(yyscanner, PARSE_EVENTS__TERM_TYPE_NOINHERIT); }
302overwrite		{ return term(yyscanner, PARSE_EVENTS__TERM_TYPE_OVERWRITE); }
303no-overwrite		{ return term(yyscanner, PARSE_EVENTS__TERM_TYPE_NOOVERWRITE); }
304percore			{ return term(yyscanner, PARSE_EVENTS__TERM_TYPE_PERCORE); }
305aux-output		{ return term(yyscanner, PARSE_EVENTS__TERM_TYPE_AUX_OUTPUT); }
306aux-action		{ return term(yyscanner, PARSE_EVENTS__TERM_TYPE_AUX_ACTION); }
307aux-sample-size		{ return term(yyscanner, PARSE_EVENTS__TERM_TYPE_AUX_SAMPLE_SIZE); }
308metric-id		{ return term(yyscanner, PARSE_EVENTS__TERM_TYPE_METRIC_ID); }
309cpu			{ return term(yyscanner, PARSE_EVENTS__TERM_TYPE_CPU); }
310ratio-to-prev		{ return term(yyscanner, PARSE_EVENTS__TERM_TYPE_RATIO_TO_PREV); }
311legacy-hardware-config 	{ return term(yyscanner, PARSE_EVENTS__TERM_TYPE_LEGACY_HARDWARE_CONFIG); }
312legacy-cache-config	{ return term(yyscanner, PARSE_EVENTS__TERM_TYPE_LEGACY_CACHE_CONFIG); }
313r{num_raw_hex}		{ return str(yyscanner, PE_RAW); }
314r0x{num_raw_hex}	{ return str(yyscanner, PE_RAW); }
315,			{ return ','; }
316"/"			{ BEGIN(INITIAL); return '/'; }
317{num_dec}		{ return value(_parse_state, yyscanner, 10); }
318{num_hex}		{ return value(_parse_state, yyscanner, 16); }
319{term_name}		{ return str(yyscanner, PE_NAME); }
320@{drv_cfg_term}		{ return drv_str(yyscanner, PE_DRV_CFG_TERM); }
321}
322
323<mem>{
324{modifier_bp}		{ return str(yyscanner, PE_MODIFIER_BP); }
325	/*
326	 * The colon before memory access modifiers can get mixed up with the
327	 * colon before event modifiers. Fortunately none of the option letters
328	 * are the same, so trailing context can be used disambiguate the two
329	 * cases.
330	 */
331":"/{modifier_bp}	{ return PE_BP_COLON; }
332	/*
333	 * The slash before memory length can get mixed up with the slash before
334	 * config terms. Fortunately config terms do not start with a numeric
335	 * digit, so trailing context can be used disambiguate the two cases.
336	 */
337"/"/{digit}		{ return PE_BP_SLASH; }
338"/"/{non_digit}		{ BEGIN(config); return '/'; }
339{num_dec}		{ return value(_parse_state, yyscanner, 10); }
340{num_hex}		{ return value(_parse_state, yyscanner, 16); }
341	/*
342	 * We need to separate 'mem:' scanner part, in order to get specific
343	 * modifier bits parsed out. Otherwise we would need to handle PE_NAME
344	 * and we'd need to parse it manually. During the escape from <mem>
345	 * state we need to put the escaping char back, so we dont miss it.
346	 */
347.			{ unput(*yytext); BEGIN(INITIAL); }
348	/*
349	 * We destroy the scanner after reaching EOF,
350	 * but anyway just to be sure get back to INIT state.
351	 */
352<<EOF>>			{ BEGIN(INITIAL); }
353}
354
355mem:			{ BEGIN(mem); return PE_PREFIX_MEM; }
356r{num_raw_hex}		{ return str(yyscanner, PE_RAW); }
357{num_dec}		{ return value(_parse_state, yyscanner, 10); }
358{num_hex}		{ return value(_parse_state, yyscanner, 16); }
359
360{modifier_event}	{ return modifiers(_parse_state, yyscanner); }
361{name}			{ return str(yyscanner, PE_NAME); }
362{quoted_name}		{ return quoted_str(yyscanner, PE_NAME); }
363"/"			{ BEGIN(config); return '/'; }
364,			{ BEGIN(event); return ','; }
365:			{ return ':'; }
366"{"			{ BEGIN(event); return '{'; }
367"}"			{ return '}'; }
368=			{ return '='; }
369\n			{ }
370.			{ }
371
372%%