1 files changed, 298 insertions, 156 deletions

diff --git a/tools/testing/selftests/resctrl/cat_test.c b/tools/testing/selftests/resctrl/cat_test.c
index 5da43767b973..4cb991be8e31 100644
--- a/tools/testing/selftests/resctrl/cat_test.c
+++ b/tools/testing/selftests/resctrl/cat_test.c
@@ -11,240 +11,382 @@
 #include "resctrl.h"
 #include <unistd.h>
 
-#define RESULT_FILE_NAME1	"result_cat1"
-#define RESULT_FILE_NAME2	"result_cat2"
+#define RESULT_FILE_NAME	"result_cat"
 #define NUM_OF_RUNS		5
-#define MAX_DIFF_PERCENT	4
-#define MAX_DIFF		1000000
-
-int count_of_bits;
-char cbm_mask[256];
-unsigned long long_mask;
-unsigned long cache_size;
 
 /*
- * Change schemata. Write schemata to specified
- * con_mon grp, mon_grp in resctrl FS.
- * Run 5 times in order to get average values.
+ * Minimum difference in LLC misses between a test with n+1 bits CBM to the
+ * test with n bits is MIN_DIFF_PERCENT_PER_BIT * (n - 1). With e.g. 5 vs 4
+ * bits in the CBM mask, the minimum difference must be at least
+ * MIN_DIFF_PERCENT_PER_BIT * (4 - 1) = 3 percent.
+ *
+ * The relationship between number of used CBM bits and difference in LLC
+ * misses is not expected to be linear. With a small number of bits, the
+ * margin is smaller than with larger number of bits. For selftest purposes,
+ * however, linear approach is enough because ultimately only pass/fail
+ * decision has to be made and distinction between strong and stronger
+ * signal is irrelevant.
  */
-static int cat_setup(int num, ...)
+#define MIN_DIFF_PERCENT_PER_BIT	1UL
+
+static int show_results_info(__u64 sum_llc_val, int no_of_bits,
+			     unsigned long cache_span,
+			     unsigned long min_diff_percent,
+			     unsigned long num_of_runs, bool platform,
+			     __s64 *prev_avg_llc_val)
 {
-	struct resctrl_val_param *p;
-	char schemata[64];
-	va_list param;
+	__u64 avg_llc_val = 0;
+	float avg_diff;
 	int ret = 0;
 
-	va_start(param, num);
-	p = va_arg(param, struct resctrl_val_param *);
-	va_end(param);
+	avg_llc_val = sum_llc_val / num_of_runs;
+	if (*prev_avg_llc_val) {
+		float delta = (__s64)(avg_llc_val - *prev_avg_llc_val);
 
-	/* Run NUM_OF_RUNS times */
-	if (p->num_of_runs >= NUM_OF_RUNS)
-		return -1;
+		avg_diff = delta / *prev_avg_llc_val;
+		ret = platform && (avg_diff * 100) < (float)min_diff_percent;
 
-	if (p->num_of_runs == 0) {
-		sprintf(schemata, "%lx", p->mask);
-		ret = write_schemata(p->ctrlgrp, schemata, p->cpu_no,
-				     p->resctrl_val);
+		ksft_print_msg("%s Check cache miss rate changed more than %.1f%%\n",
+			       ret ? "Fail:" : "Pass:", (float)min_diff_percent);
+
+		ksft_print_msg("Percent diff=%.1f\n", avg_diff * 100);
 	}
-	p->num_of_runs++;
+	*prev_avg_llc_val = avg_llc_val;
+
+	show_cache_info(no_of_bits, avg_llc_val, cache_span, true);
 
 	return ret;
 }
 
-static void show_cache_info(unsigned long sum_llc_perf_miss, int no_of_bits,
-			    unsigned long span)
+/* Remove the highest bit from CBM */
+static unsigned long next_mask(unsigned long current_mask)
 {
-	unsigned long allocated_cache_lines = span / 64;
-	unsigned long avg_llc_perf_miss = 0;
-	float diff_percent;
-
-	avg_llc_perf_miss = sum_llc_perf_miss / (NUM_OF_RUNS - 1);
-	diff_percent = ((float)allocated_cache_lines - avg_llc_perf_miss) /
-				allocated_cache_lines * 100;
-
-	printf("%sok CAT: cache miss rate within %d%%\n",
-	       !is_amd && abs((int)diff_percent) > MAX_DIFF_PERCENT ?
-	       "not " : "", MAX_DIFF_PERCENT);
-	tests_run++;
-	printf("# Percent diff=%d\n", abs((int)diff_percent));
-	printf("# Number of bits: %d\n", no_of_bits);
-	printf("# Avg_llc_perf_miss: %lu\n", avg_llc_perf_miss);
-	printf("# Allocated cache lines: %lu\n", allocated_cache_lines);
+	return current_mask & (current_mask >> 1);
 }
 
-static int check_results(struct resctrl_val_param *param)
+static int check_results(struct resctrl_val_param *param, const char *cache_type,
+			 unsigned long cache_total_size, unsigned long full_cache_mask,
+			 unsigned long current_mask)
 {
 	char *token_array[8], temp[512];
-	unsigned long sum_llc_perf_miss = 0;
-	int runs = 0, no_of_bits = 0;
+	__u64 sum_llc_perf_miss = 0;
+	__s64 prev_avg_llc_val = 0;
+	unsigned long alloc_size;
+	int runs = 0;
+	int fail = 0;
+	int ret;
 	FILE *fp;
 
-	printf("# Checking for pass/fail\n");
+	ksft_print_msg("Checking for pass/fail\n");
 	fp = fopen(param->filename, "r");
 	if (!fp) {
-		perror("# Cannot open file");
+		ksft_perror("Cannot open file");
 
-		return errno;
+		return -1;
 	}
 
 	while (fgets(temp, sizeof(temp), fp)) {
 		char *token = strtok(temp, ":\t");
 		int fields = 0;
+		int bits;
 
 		while (token) {
 			token_array[fields++] = token;
 			token = strtok(NULL, ":\t");
 		}
-		/*
-		 * Discard the first value which is inaccurate due to monitoring
-		 * setup transition phase.
-		 */
-		if (runs > 0)
-			sum_llc_perf_miss += strtoul(token_array[3], NULL, 0);
+
+		sum_llc_perf_miss += strtoull(token_array[3], NULL, 0);
 		runs++;
+
+		if (runs < NUM_OF_RUNS)
+			continue;
+
+		if (!current_mask) {
+			ksft_print_msg("Unexpected empty cache mask\n");
+			break;
+		}
+
+		alloc_size = cache_portion_size(cache_total_size, current_mask, full_cache_mask);
+
+		bits = count_bits(current_mask);
+
+		ret = show_results_info(sum_llc_perf_miss, bits,
+					alloc_size / 64,
+					MIN_DIFF_PERCENT_PER_BIT * (bits - 1),
+					runs, get_vendor() == ARCH_INTEL,
+					&prev_avg_llc_val);
+		if (ret)
+			fail = 1;
+
+		runs = 0;
+		sum_llc_perf_miss = 0;
+		current_mask = next_mask(current_mask);
 	}
 
 	fclose(fp);
-	no_of_bits = count_bits(param->mask);
 
-	show_cache_info(sum_llc_perf_miss, no_of_bits, param->span);
-
-	return 0;
+	return fail;
 }
 
 void cat_test_cleanup(void)
 {
-	remove(RESULT_FILE_NAME1);
-	remove(RESULT_FILE_NAME2);
+	remove(RESULT_FILE_NAME);
 }
 
-int cat_perf_miss_val(int cpu_no, int n, char *cache_type)
+/*
+ * cat_test - Execute CAT benchmark and measure cache misses
+ * @test:		Test information structure
+ * @uparams:		User supplied parameters
+ * @param:		Parameters passed to cat_test()
+ * @span:		Buffer size for the benchmark
+ * @current_mask	Start mask for the first iteration
+ *
+ * Run CAT selftest by varying the allocated cache portion and comparing the
+ * impact on cache misses (the result analysis is done in check_results()
+ * and show_results_info(), not in this function).
+ *
+ * One bit is removed from the CAT allocation bit mask (in current_mask) for
+ * each subsequent test which keeps reducing the size of the allocated cache
+ * portion. A single test flushes the buffer, reads it to warm up the cache,
+ * and reads the buffer again. The cache misses are measured during the last
+ * read pass.
+ *
+ * Return:		0 when the test was run, < 0 on error.
+ */
+static int cat_test(const struct resctrl_test *test,
+		    const struct user_params *uparams,
+		    struct resctrl_val_param *param,
+		    size_t span, unsigned long current_mask)
 {
-	unsigned long l_mask, l_mask_1;
-	int ret, pipefd[2], sibling_cpu_no;
-	char pipe_message;
+	char *resctrl_val = param->resctrl_val;
+	struct perf_event_read pe_read;
+	struct perf_event_attr pea;
+	cpu_set_t old_affinity;
+	unsigned char *buf;
+	char schemata[64];
+	int ret, i, pe_fd;
 	pid_t bm_pid;
 
-	cache_size = 0;
+	if (strcmp(param->filename, "") == 0)
+		sprintf(param->filename, "stdio");
 
-	ret = remount_resctrlfs(true);
+	bm_pid = getpid();
+
+	/* Taskset benchmark to specified cpu */
+	ret = taskset_benchmark(bm_pid, uparams->cpu, &old_affinity);
 	if (ret)
 		return ret;
 
-	if (!validate_resctrl_feature_request("cat"))
-		return -1;
+	/* Write benchmark to specified con_mon grp, mon_grp in resctrl FS*/
+	ret = write_bm_pid_to_resctrl(bm_pid, param->ctrlgrp, param->mongrp,
+				      resctrl_val);
+	if (ret)
+		goto reset_affinity;
+
+	perf_event_attr_initialize(&pea, PERF_COUNT_HW_CACHE_MISSES);
+	perf_event_initialize_read_format(&pe_read);
+	pe_fd = perf_open(&pea, bm_pid, uparams->cpu);
+	if (pe_fd < 0) {
+		ret = -1;
+		goto reset_affinity;
+	}
 
-	/* Get default cbm mask for L3/L2 cache */
-	ret = get_cbm_mask(cache_type);
+	buf = alloc_buffer(span, 1);
+	if (!buf) {
+		ret = -1;
+		goto pe_close;
+	}
+
+	while (current_mask) {
+		snprintf(schemata, sizeof(schemata), "%lx", param->mask & ~current_mask);
+		ret = write_schemata("", schemata, uparams->cpu, test->resource);
+		if (ret)
+			goto free_buf;
+		snprintf(schemata, sizeof(schemata), "%lx", current_mask);
+		ret = write_schemata(param->ctrlgrp, schemata, uparams->cpu, test->resource);
+		if (ret)
+			goto free_buf;
+
+		for (i = 0; i < NUM_OF_RUNS; i++) {
+			mem_flush(buf, span);
+			fill_cache_read(buf, span, true);
+
+			ret = perf_event_reset_enable(pe_fd);
+			if (ret)
+				goto free_buf;
+
+			fill_cache_read(buf, span, true);
+
+			ret = perf_event_measure(pe_fd, &pe_read, param->filename, bm_pid);
+			if (ret)
+				goto free_buf;
+		}
+		current_mask = next_mask(current_mask);
+	}
+
+free_buf:
+	free(buf);
+pe_close:
+	close(pe_fd);
+reset_affinity:
+	taskset_restore(bm_pid, &old_affinity);
+
+	return ret;
+}
+
+static int cat_run_test(const struct resctrl_test *test, const struct user_params *uparams)
+{
+	unsigned long long_mask, start_mask, full_cache_mask;
+	unsigned long cache_total_size = 0;
+	int n = uparams->bits;
+	unsigned int start;
+	int count_of_bits;
+	size_t span;
+	int ret;
+
+	ret = get_full_cbm(test->resource, &full_cache_mask);
+	if (ret)
+		return ret;
+	/* Get the largest contiguous exclusive portion of the cache */
+	ret = get_mask_no_shareable(test->resource, &long_mask);
 	if (ret)
 		return ret;
-
-	long_mask = strtoul(cbm_mask, NULL, 16);
 
 	/* Get L3/L2 cache size */
-	ret = get_cache_size(cpu_no, cache_type, &cache_size);
+	ret = get_cache_size(uparams->cpu, test->resource, &cache_total_size);
 	if (ret)
 		return ret;
-	printf("cache size :%lu\n", cache_size);
+	ksft_print_msg("Cache size :%lu\n", cache_total_size);
 
-	/* Get max number of bits from default-cabm mask */
-	count_of_bits = count_bits(long_mask);
+	count_of_bits = count_contiguous_bits(long_mask, &start);
 
-	if (n < 1 || n > count_of_bits - 1) {
-		printf("Invalid input value for no_of_bits n!\n");
-		printf("Please Enter value in range 1 to %d\n",
-		       count_of_bits - 1);
-		return -1;
-	}
+	if (!n)
+		n = count_of_bits / 2;
 
-	/* Get core id from same socket for running another thread */
-	sibling_cpu_no = get_core_sibling(cpu_no);
-	if (sibling_cpu_no < 0)
+	if (n > count_of_bits - 1) {
+		ksft_print_msg("Invalid input value for no_of_bits n!\n");
+		ksft_print_msg("Please enter value in range 1 to %d\n",
+			       count_of_bits - 1);
 		return -1;
+	}
+	start_mask = create_bit_mask(start, n);
 
 	struct resctrl_val_param param = {
-		.resctrl_val	= "cat",
-		.cpu_no		= cpu_no,
-		.mum_resctrlfs	= 0,
-		.setup		= cat_setup,
+		.resctrl_val	= CAT_STR,
+		.ctrlgrp	= "c1",
+		.filename	= RESULT_FILE_NAME,
+		.num_of_runs	= 0,
 	};
+	param.mask = long_mask;
+	span = cache_portion_size(cache_total_size, start_mask, full_cache_mask);
 
-	l_mask = long_mask >> n;
-	l_mask_1 = ~l_mask & long_mask;
+	remove(param.filename);
 
-	/* Set param values for parent thread which will be allocated bitmask
-	 * with (max_bits - n) bits
-	 */
-	param.span = cache_size * (count_of_bits - n) / count_of_bits;
-	strcpy(param.ctrlgrp, "c2");
-	strcpy(param.mongrp, "m2");
-	strcpy(param.filename, RESULT_FILE_NAME2);
-	param.mask = l_mask;
-	param.num_of_runs = 0;
-
-	if (pipe(pipefd)) {
-		perror("# Unable to create pipe");
-		return errno;
-	}
+	ret = cat_test(test, uparams, &param, span, start_mask);
+	if (ret)
+		goto out;
 
-	bm_pid = fork();
+	ret = check_results(&param, test->resource,
+			    cache_total_size, full_cache_mask, start_mask);
+out:
+	cat_test_cleanup();
 
-	/* Set param values for child thread which will be allocated bitmask
-	 * with n bits
-	 */
-	if (bm_pid == 0) {
-		param.mask = l_mask_1;
-		strcpy(param.ctrlgrp, "c1");
-		strcpy(param.mongrp, "m1");
-		param.span = cache_size * n / count_of_bits;
-		strcpy(param.filename, RESULT_FILE_NAME1);
-		param.num_of_runs = 0;
-		param.cpu_no = sibling_cpu_no;
-	}
+	return ret;
+}
 
-	remove(param.filename);
+static int noncont_cat_run_test(const struct resctrl_test *test,
+				const struct user_params *uparams)
+{
+	unsigned long full_cache_mask, cont_mask, noncont_mask;
+	unsigned int eax, ebx, ecx, edx, sparse_masks;
+	int bit_center, ret;
+	char schemata[64];
 
-	ret = cat_val(&param);
+	/* Check to compare sparse_masks content to CPUID output. */
+	ret = resource_info_unsigned_get(test->resource, "sparse_masks", &sparse_masks);
 	if (ret)
 		return ret;
 
-	ret = check_results(&param);
-	if (ret)
-		return ret;
+	if (!strcmp(test->resource, "L3"))
+		__cpuid_count(0x10, 1, eax, ebx, ecx, edx);
+	else if (!strcmp(test->resource, "L2"))
+		__cpuid_count(0x10, 2, eax, ebx, ecx, edx);
+	else
+		return -EINVAL;
 
-	if (bm_pid == 0) {
-		/* Tell parent that child is ready */
-		close(pipefd[0]);
-		pipe_message = 1;
-		if (write(pipefd[1], &pipe_message, sizeof(pipe_message)) <
-		    sizeof(pipe_message)) {
-			close(pipefd[1]);
-			perror("# failed signaling parent process");
-			return errno;
-		}
+	if (sparse_masks != ((ecx >> 3) & 1)) {
+		ksft_print_msg("CPUID output doesn't match 'sparse_masks' file content!\n");
+		return 1;
+	}
 
-		close(pipefd[1]);
-		while (1)
-			;
-	} else {
-		/* Parent waits for child to be ready. */
-		close(pipefd[1]);
-		pipe_message = 0;
-		while (pipe_message != 1) {
-			if (read(pipefd[0], &pipe_message,
-				 sizeof(pipe_message)) < sizeof(pipe_message)) {
-				perror("# failed reading from child process");
-				break;
-			}
-		}
-		close(pipefd[0]);
-		kill(bm_pid, SIGKILL);
+	/* Write checks initialization. */
+	ret = get_full_cbm(test->resource, &full_cache_mask);
+	if (ret < 0)
+		return ret;
+	bit_center = count_bits(full_cache_mask) / 2;
+
+	/*
+	 * The bit_center needs to be at least 3 to properly calculate the CBM
+	 * hole in the noncont_mask. If it's smaller return an error since the
+	 * cache mask is too short and that shouldn't happen.
+	 */
+	if (bit_center < 3)
+		return -EINVAL;
+	cont_mask = full_cache_mask >> bit_center;
+
+	/* Contiguous mask write check. */
+	snprintf(schemata, sizeof(schemata), "%lx", cont_mask);
+	ret = write_schemata("", schemata, uparams->cpu, test->resource);
+	if (ret) {
+		ksft_print_msg("Write of contiguous CBM failed\n");
+		return 1;
 	}
 
-	cat_test_cleanup();
-	if (bm_pid)
-		umount_resctrlfs();
+	/*
+	 * Non-contiguous mask write check. CBM has a 0xf hole approximately in the middle.
+	 * Output is compared with support information to catch any edge case errors.
+	 */
+	noncont_mask = ~(0xfUL << (bit_center - 2)) & full_cache_mask;
+	snprintf(schemata, sizeof(schemata), "%lx", noncont_mask);
+	ret = write_schemata("", schemata, uparams->cpu, test->resource);
+	if (ret && sparse_masks)
+		ksft_print_msg("Non-contiguous CBMs supported but write of non-contiguous CBM failed\n");
+	else if (ret && !sparse_masks)
+		ksft_print_msg("Non-contiguous CBMs not supported and write of non-contiguous CBM failed as expected\n");
+	else if (!ret && !sparse_masks)
+		ksft_print_msg("Non-contiguous CBMs not supported but write of non-contiguous CBM succeeded\n");
+
+	return !ret == !sparse_masks;
+}
+
+static bool noncont_cat_feature_check(const struct resctrl_test *test)
+{
+	if (!resctrl_resource_exists(test->resource))
+		return false;
 
-	return 0;
+	return resource_info_file_exists(test->resource, "sparse_masks");
 }
+
+struct resctrl_test l3_cat_test = {
+	.name = "L3_CAT",
+	.group = "CAT",
+	.resource = "L3",
+	.feature_check = test_resource_feature_check,
+	.run_test = cat_run_test,
+};
+
+struct resctrl_test l3_noncont_cat_test = {
+	.name = "L3_NONCONT_CAT",
+	.group = "CAT",
+	.resource = "L3",
+	.feature_check = noncont_cat_feature_check,
+	.run_test = noncont_cat_run_test,
+};
+
+struct resctrl_test l2_noncont_cat_test = {
+	.name = "L2_NONCONT_CAT",
+	.group = "CAT",
+	.resource = "L2",
+	.feature_check = noncont_cat_feature_check,
+	.run_test = noncont_cat_run_test,
+};