// SPDX-License-Identifier: GPL-2.0 /* * Memory bandwidth monitoring and allocation library * * Copyright (C) 2018 Intel Corporation * * Authors: * Sai Praneeth Prakhya <sai.praneeth.prakhya@intel.com>, * Fenghua Yu <fenghua.yu@intel.com> */ #include "resctrl.h" #define UNCORE_IMC "uncore_imc" #define READ_FILE_NAME "events/cas_count_read" #define WRITE_FILE_NAME "events/cas_count_write" #define DYN_PMU_PATH "/sys/bus/event_source/devices" #define SCALE 0.00006103515625 #define MAX_IMCS 20 #define MAX_TOKENS 5 #define READ 0 #define WRITE 1 #define CON_MON_MBM_LOCAL_BYTES_PATH \ "%s/%s/mon_groups/%s/mon_data/mon_L3_%02d/mbm_local_bytes" #define CON_MBM_LOCAL_BYTES_PATH \ "%s/%s/mon_data/mon_L3_%02d/mbm_local_bytes" #define MON_MBM_LOCAL_BYTES_PATH \ "%s/mon_groups/%s/mon_data/mon_L3_%02d/mbm_local_bytes" #define MBM_LOCAL_BYTES_PATH \ "%s/mon_data/mon_L3_%02d/mbm_local_bytes" #define CON_MON_LCC_OCCUP_PATH \ "%s/%s/mon_groups/%s/mon_data/mon_L3_%02d/llc_occupancy" #define CON_LCC_OCCUP_PATH \ "%s/%s/mon_data/mon_L3_%02d/llc_occupancy" #define MON_LCC_OCCUP_PATH \ "%s/mon_groups/%s/mon_data/mon_L3_%02d/llc_occupancy" #define LCC_OCCUP_PATH \ "%s/mon_data/mon_L3_%02d/llc_occupancy" struct membw_read_format { __u64 value; /* The value of the event */ __u64 time_enabled; /* if PERF_FORMAT_TOTAL_TIME_ENABLED */ __u64 time_running; /* if PERF_FORMAT_TOTAL_TIME_RUNNING */ __u64 id; /* if PERF_FORMAT_ID */ }; struct imc_counter_config { __u32 type; __u64 event; __u64 umask; struct perf_event_attr pe; struct membw_read_format return_value; int fd; }; static char mbm_total_path[1024]; static int imcs; static struct imc_counter_config imc_counters_config[MAX_IMCS][2]; void membw_initialize_perf_event_attr(int i, int j) { memset(&imc_counters_config[i][j].pe, 0, sizeof(struct perf_event_attr)); imc_counters_config[i][j].pe.type = imc_counters_config[i][j].type; imc_counters_config[i][j].pe.size = sizeof(struct perf_event_attr); imc_counters_config[i][j].pe.disabled = 1; imc_counters_config[i][j].pe.inherit = 1; imc_counters_config[i][j].pe.exclude_guest = 0; imc_counters_config[i][j].pe.config = imc_counters_config[i][j].umask << 8 | imc_counters_config[i][j].event; imc_counters_config[i][j].pe.sample_type = PERF_SAMPLE_IDENTIFIER; imc_counters_config[i][j].pe.read_format = PERF_FORMAT_TOTAL_TIME_ENABLED | PERF_FORMAT_TOTAL_TIME_RUNNING; } void membw_ioctl_perf_event_ioc_reset_enable(int i, int j) { ioctl(imc_counters_config[i][j].fd, PERF_EVENT_IOC_RESET, 0); ioctl(imc_counters_config[i][j].fd, PERF_EVENT_IOC_ENABLE, 0); } void membw_ioctl_perf_event_ioc_disable(int i, int j) { ioctl(imc_counters_config[i][j].fd, PERF_EVENT_IOC_DISABLE, 0); } /* * get_event_and_umask: Parse config into event and umask * @cas_count_cfg: Config * @count: iMC number * @op: Operation (read/write) */ void get_event_and_umask(char *cas_count_cfg, int count, bool op) { char *token[MAX_TOKENS]; int i = 0; strcat(cas_count_cfg, ","); token[0] = strtok(cas_count_cfg, "=,"); for (i = 1; i < MAX_TOKENS; i++) token[i] = strtok(NULL, "=,"); for (i = 0; i < MAX_TOKENS; i++) { if (!token[i]) break; if (strcmp(token[i], "event") == 0) { if (op == READ) imc_counters_config[count][READ].event = strtol(token[i + 1], NULL, 16); else imc_counters_config[count][WRITE].event = strtol(token[i + 1], NULL, 16); } if (strcmp(token[i], "umask") == 0) { if (op == READ) imc_counters_config[count][READ].umask = strtol(token[i + 1], NULL, 16); else imc_counters_config[count][WRITE].umask = strtol(token[i + 1], NULL, 16); } } } static int open_perf_event(int i, int cpu_no, int j) { imc_counters_config[i][j].fd = perf_event_open(&imc_counters_config[i][j].pe, -1, cpu_no, -1, PERF_FLAG_FD_CLOEXEC); if (imc_counters_config[i][j].fd == -1) { fprintf(stderr, "Error opening leader %llx\n", imc_counters_config[i][j].pe.config); return -1; } return 0; } /* Get type and config (read and write) of an iMC counter */ static int read_from_imc_dir(char *imc_dir, int count) { char cas_count_cfg[1024], imc_counter_cfg[1024], imc_counter_type[1024]; FILE *fp; /* Get type of iMC counter */ sprintf(imc_counter_type, "%s%s", imc_dir, "type"); fp = fopen(imc_counter_type, "r"); if (!fp) { perror("Failed to open imc counter type file"); return -1; } if (fscanf(fp, "%u", &imc_counters_config[count][READ].type) <= 0) { perror("Could not get imc type"); fclose(fp); return -1; } fclose(fp); imc_counters_config[count][WRITE].type = imc_counters_config[count][READ].type; /* Get read config */ sprintf(imc_counter_cfg, "%s%s", imc_dir, READ_FILE_NAME); fp = fopen(imc_counter_cfg, "r"); if (!fp) { perror("Failed to open imc config file"); return -1; } if (fscanf(fp, "%s", cas_count_cfg) <= 0) { perror("Could not get imc cas count read"); fclose(fp); return -1; } fclose(fp); get_event_and_umask(cas_count_cfg, count, READ); /* Get write config */ sprintf(imc_counter_cfg, "%s%s", imc_dir, WRITE_FILE_NAME); fp = fopen(imc_counter_cfg, "r"); if (!fp) { perror("Failed to open imc config file"); return -1; } if (fscanf(fp, "%s", cas_count_cfg) <= 0) { perror("Could not get imc cas count write"); fclose(fp); return -1; } fclose(fp); get_event_and_umask(cas_count_cfg, count, WRITE); return 0; } /* * A system can have 'n' number of iMC (Integrated Memory Controller) * counters, get that 'n'. For each iMC counter get it's type and config. * Also, each counter has two configs, one for read and the other for write. * A config again has two parts, event and umask. * Enumerate all these details into an array of structures. * * Return: >= 0 on success. < 0 on failure. */ static int num_of_imcs(void) { char imc_dir[512], *temp; unsigned int count = 0; struct dirent *ep; int ret; DIR *dp; dp = opendir(DYN_PMU_PATH); if (dp) { while ((ep = readdir(dp))) { temp = strstr(ep->d_name, UNCORE_IMC); if (!temp) continue; /* * imc counters are named as "uncore_imc_<n>", hence * increment the pointer to point to <n>. Note that * sizeof(UNCORE_IMC) would count for null character as * well and hence the last underscore character in * uncore_imc'_' need not be counted. */ temp = temp + sizeof(UNCORE_IMC); /* * Some directories under "DYN_PMU_PATH" could have * names like "uncore_imc_free_running", hence, check if * first character is a numerical digit or not. */ if (temp[0] >= '0' && temp[0] <= '9') { sprintf(imc_dir, "%s/%s/", DYN_PMU_PATH, ep->d_name); ret = read_from_imc_dir(imc_dir, count); if (ret) { closedir(dp); return ret; } count++; } } closedir(dp); if (count == 0) { perror("Unable find iMC counters!\n"); return -1; } } else { perror("Unable to open PMU directory!\n"); return -1; } return count; } static int initialize_mem_bw_imc(void) { int imc, j; imcs = num_of_imcs(); if (imcs <= 0) return imcs; /* Initialize perf_event_attr structures for all iMC's */ for (imc = 0; imc < imcs; imc++) { for (j = 0; j < 2; j++) membw_initialize_perf_event_attr(imc, j); } return 0; } /* * get_mem_bw_imc: Memory band width as reported by iMC counters * @cpu_no: CPU number that the benchmark PID is binded to * @bw_report: Bandwidth report type (reads, writes) * * Memory B/W utilized by a process on a socket can be calculated using * iMC counters. Perf events are used to read these counters. * * Return: = 0 on success. < 0 on failure. */ static int get_mem_bw_imc(int cpu_no, char *bw_report, float *bw_imc) { float reads, writes, of_mul_read, of_mul_write; int imc, j, ret; /* Start all iMC counters to log values (both read and write) */ reads = 0, writes = 0, of_mul_read = 1, of_mul_write = 1; for (imc = 0; imc < imcs; imc++) { for (j = 0; j < 2; j++) { ret = open_perf_event(imc, cpu_no, j); if (ret) return -1; } for (j = 0; j < 2; j++) membw_ioctl_perf_event_ioc_reset_enable(imc, j); } sleep(1); /* Stop counters after a second to get results (both read and write) */ for (imc = 0; imc < imcs; imc++) { for (j = 0; j < 2; j++) membw_ioctl_perf_event_ioc_disable(imc, j); } /* * Get results which are stored in struct type imc_counter_config * Take over flow into consideration before calculating total b/w */ for (imc = 0; imc < imcs; imc++) { struct imc_counter_config *r = &imc_counters_config[imc][READ]; struct imc_counter_config *w = &imc_counters_config[imc][WRITE]; if (read(r->fd, &r->return_value, sizeof(struct membw_read_format)) == -1) { perror("Couldn't get read b/w through iMC"); return -1; } if (read(w->fd, &w->return_value, sizeof(struct membw_read_format)) == -1) { perror("Couldn't get write bw through iMC"); return -1; } __u64 r_time_enabled = r->return_value.time_enabled; __u64 r_time_running = r->return_value.time_running; if (r_time_enabled != r_time_running) of_mul_read = (float)r_time_enabled / (float)r_time_running; __u64 w_time_enabled = w->return_value.time_enabled; __u64 w_time_running = w->return_value.time_running; if (w_time_enabled != w_time_running) of_mul_write = (float)w_time_enabled / (float)w_time_running; reads += r->return_value.value * of_mul_read * SCALE; writes += w->return_value.value * of_mul_write * SCALE; } for (imc = 0; imc < imcs; imc++) { close(imc_counters_config[imc][READ].fd); close(imc_counters_config[imc][WRITE].fd); } if (strcmp(bw_report, "reads") == 0) { *bw_imc = reads; return 0; } if (strcmp(bw_report, "writes") == 0) { *bw_imc = writes; return 0; } *bw_imc = reads + writes; return 0; } void set_mbm_path(const char *ctrlgrp, const char *mongrp, int resource_id) { if (ctrlgrp && mongrp) sprintf(mbm_total_path, CON_MON_MBM_LOCAL_BYTES_PATH, RESCTRL_PATH, ctrlgrp, mongrp, resource_id); else if (!ctrlgrp && mongrp) sprintf(mbm_total_path, MON_MBM_LOCAL_BYTES_PATH, RESCTRL_PATH, mongrp, resource_id); else if (ctrlgrp && !mongrp) sprintf(mbm_total_path, CON_MBM_LOCAL_BYTES_PATH, RESCTRL_PATH, ctrlgrp, resource_id); else if (!ctrlgrp && !mongrp) sprintf(mbm_total_path, MBM_LOCAL_BYTES_PATH, RESCTRL_PATH, resource_id); } /* * initialize_mem_bw_resctrl: Appropriately populate "mbm_total_path" * @ctrlgrp: Name of the control monitor group (con_mon grp) * @mongrp: Name of the monitor group (mon grp) * @cpu_no: CPU number that the benchmark PID is binded to * @resctrl_val: Resctrl feature (Eg: mbm, mba.. etc) */ static void initialize_mem_bw_resctrl(const char *ctrlgrp, const char *mongrp, int cpu_no, char *resctrl_val) { int resource_id; if (get_resource_id(cpu_no, &resource_id) < 0) { perror("Could not get resource_id"); return; } if (!strncmp(resctrl_val, MBM_STR, sizeof(MBM_STR))) set_mbm_path(ctrlgrp, mongrp, resource_id); if (!strncmp(resctrl_val, MBA_STR, sizeof(MBA_STR))) { if (ctrlgrp) sprintf(mbm_total_path, CON_MBM_LOCAL_BYTES_PATH, RESCTRL_PATH, ctrlgrp, resource_id); else sprintf(mbm_total_path, MBM_LOCAL_BYTES_PATH, RESCTRL_PATH, resource_id); } } /* * Get MBM Local bytes as reported by resctrl FS * For MBM, * 1. If con_mon grp and mon grp are given, then read from con_mon grp's mon grp * 2. If only con_mon grp is given, then read from con_mon grp * 3. If both are not given, then read from root con_mon grp * For MBA, * 1. If con_mon grp is given, then read from it * 2. If con_mon grp is not given, then read from root con_mon grp */ static int get_mem_bw_resctrl(unsigned long *mbm_total) { FILE *fp; fp = fopen(mbm_total_path, "r"); if (!fp) { perror("Failed to open total bw file"); return -1; } if (fscanf(fp, "%lu", mbm_total) <= 0) { perror("Could not get mbm local bytes"); fclose(fp); return -1; } fclose(fp); return 0; } pid_t bm_pid, ppid; void ctrlc_handler(int signum, siginfo_t *info, void *ptr) { kill(bm_pid, SIGKILL); umount_resctrlfs(); tests_cleanup(); ksft_print_msg("Ending\n\n"); exit(EXIT_SUCCESS); } /* * Register CTRL-C handler for parent, as it has to kill * child process before exiting. */ int signal_handler_register(void) { struct sigaction sigact; int ret = 0; sigact.sa_sigaction = ctrlc_handler; sigemptyset(&sigact.sa_mask); sigact.sa_flags = SA_SIGINFO; if (sigaction(SIGINT, &sigact, NULL) || sigaction(SIGTERM, &sigact, NULL) || sigaction(SIGHUP, &sigact, NULL)) { perror("# sigaction"); ret = -1; } return ret; } /* * Reset signal handler to SIG_DFL. * Non-Value return because the caller should keep * the error code of other path even if sigaction fails. */ void signal_handler_unregister(void) { struct sigaction sigact; sigact.sa_handler = SIG_DFL; sigemptyset(&sigact.sa_mask); if (sigaction(SIGINT, &sigact, NULL) || sigaction(SIGTERM, &sigact, NULL) || sigaction(SIGHUP, &sigact, NULL)) { perror("# sigaction"); } } /* * print_results_bw: the memory bandwidth results are stored in a file * @filename: file that stores the results * @bm_pid: child pid that runs benchmark * @bw_imc: perf imc counter value * @bw_resc: memory bandwidth value * * Return: 0 on success. non-zero on failure. */ static int print_results_bw(char *filename, int bm_pid, float bw_imc, unsigned long bw_resc) { unsigned long diff = fabs(bw_imc - bw_resc); FILE *fp; if (strcmp(filename, "stdio") == 0 || strcmp(filename, "stderr") == 0) { printf("Pid: %d \t Mem_BW_iMC: %f \t ", bm_pid, bw_imc); printf("Mem_BW_resc: %lu \t Difference: %lu\n", bw_resc, diff); } else { fp = fopen(filename, "a"); if (!fp) { perror("Cannot open results file"); return errno; } if (fprintf(fp, "Pid: %d \t Mem_BW_iMC: %f \t Mem_BW_resc: %lu \t Difference: %lu\n", bm_pid, bw_imc, bw_resc, diff) <= 0) { fclose(fp); perror("Could not log results."); return errno; } fclose(fp); } return 0; } static void set_cmt_path(const char *ctrlgrp, const char *mongrp, char sock_num) { if (strlen(ctrlgrp) && strlen(mongrp)) sprintf(llc_occup_path, CON_MON_LCC_OCCUP_PATH, RESCTRL_PATH, ctrlgrp, mongrp, sock_num); else if (!strlen(ctrlgrp) && strlen(mongrp)) sprintf(llc_occup_path, MON_LCC_OCCUP_PATH, RESCTRL_PATH, mongrp, sock_num); else if (strlen(ctrlgrp) && !strlen(mongrp)) sprintf(llc_occup_path, CON_LCC_OCCUP_PATH, RESCTRL_PATH, ctrlgrp, sock_num); else if (!strlen(ctrlgrp) && !strlen(mongrp)) sprintf(llc_occup_path, LCC_OCCUP_PATH, RESCTRL_PATH, sock_num); } /* * initialize_llc_occu_resctrl: Appropriately populate "llc_occup_path" * @ctrlgrp: Name of the control monitor group (con_mon grp) * @mongrp: Name of the monitor group (mon grp) * @cpu_no: CPU number that the benchmark PID is binded to * @resctrl_val: Resctrl feature (Eg: cat, cmt.. etc) */ static void initialize_llc_occu_resctrl(const char *ctrlgrp, const char *mongrp, int cpu_no, char *resctrl_val) { int resource_id; if (get_resource_id(cpu_no, &resource_id) < 0) { perror("# Unable to resource_id"); return; } if (!strncmp(resctrl_val, CMT_STR, sizeof(CMT_STR))) set_cmt_path(ctrlgrp, mongrp, resource_id); } static int measure_vals(struct resctrl_val_param *param, unsigned long *bw_resc_start) { unsigned long bw_resc, bw_resc_end; float bw_imc; int ret; /* * Measure memory bandwidth from resctrl and from * another source which is perf imc value or could * be something else if perf imc event is not available. * Compare the two values to validate resctrl value. * It takes 1sec to measure the data. */ ret = get_mem_bw_imc(param->cpu_no, param->bw_report, &bw_imc); if (ret < 0) return ret; ret = get_mem_bw_resctrl(&bw_resc_end); if (ret < 0) return ret; bw_resc = (bw_resc_end - *bw_resc_start) / MB; ret = print_results_bw(param->filename, bm_pid, bw_imc, bw_resc); if (ret) return ret; *bw_resc_start = bw_resc_end; return 0; } /* * resctrl_val: execute benchmark and measure memory bandwidth on * the benchmark * @benchmark_cmd: benchmark command and its arguments * @param: parameters passed to resctrl_val() * * Return: 0 on success. non-zero on failure. */ int resctrl_val(char **benchmark_cmd, struct resctrl_val_param *param) { char *resctrl_val = param->resctrl_val; unsigned long bw_resc_start = 0; struct sigaction sigact; int ret = 0, pipefd[2]; char pipe_message = 0; union sigval value; if (strcmp(param->filename, "") == 0) sprintf(param->filename, "stdio"); if (!strncmp(resctrl_val, MBA_STR, sizeof(MBA_STR)) || !strncmp(resctrl_val, MBM_STR, sizeof(MBM_STR))) { ret = validate_bw_report_request(param->bw_report); if (ret) return ret; } /* * If benchmark wasn't successfully started by child, then child should * kill parent, so save parent's pid */ ppid = getpid(); if (pipe(pipefd)) { perror("# Unable to create pipe"); return -1; } /* * Fork to start benchmark, save child's pid so that it can be killed * when needed */ fflush(stdout); bm_pid = fork(); if (bm_pid == -1) { perror("# Unable to fork"); return -1; } if (bm_pid == 0) { /* * Mask all signals except SIGUSR1, parent uses SIGUSR1 to * start benchmark */ sigfillset(&sigact.sa_mask); sigdelset(&sigact.sa_mask, SIGUSR1); sigact.sa_sigaction = run_benchmark; sigact.sa_flags = SA_SIGINFO; /* Register for "SIGUSR1" signal from parent */ if (sigaction(SIGUSR1, &sigact, NULL)) PARENT_EXIT("Can't register child for signal"); /* Tell parent that child is ready */ close(pipefd[0]); pipe_message = 1; if (write(pipefd[1], &pipe_message, sizeof(pipe_message)) < sizeof(pipe_message)) { perror("# failed signaling parent process"); close(pipefd[1]); return -1; } close(pipefd[1]); /* Suspend child until delivery of "SIGUSR1" from parent */ sigsuspend(&sigact.sa_mask); PARENT_EXIT("Child is done"); } ksft_print_msg("Benchmark PID: %d\n", bm_pid); ret = signal_handler_register(); if (ret) goto out; value.sival_ptr = benchmark_cmd; /* Taskset benchmark to specified cpu */ ret = taskset_benchmark(bm_pid, param->cpu_no); if (ret) goto unregister; /* Write benchmark to specified control&monitoring grp in resctrl FS */ ret = write_bm_pid_to_resctrl(bm_pid, param->ctrlgrp, param->mongrp, resctrl_val); if (ret) goto unregister; if (!strncmp(resctrl_val, MBM_STR, sizeof(MBM_STR)) || !strncmp(resctrl_val, MBA_STR, sizeof(MBA_STR))) { ret = initialize_mem_bw_imc(); if (ret) goto unregister; initialize_mem_bw_resctrl(param->ctrlgrp, param->mongrp, param->cpu_no, resctrl_val); } else if (!strncmp(resctrl_val, CMT_STR, sizeof(CMT_STR))) initialize_llc_occu_resctrl(param->ctrlgrp, param->mongrp, param->cpu_no, resctrl_val); /* Parent waits for child to be ready. */ close(pipefd[1]); while (pipe_message != 1) { if (read(pipefd[0], &pipe_message, sizeof(pipe_message)) < sizeof(pipe_message)) { perror("# failed reading message from child process"); close(pipefd[0]); goto unregister; } } close(pipefd[0]); /* Signal child to start benchmark */ if (sigqueue(bm_pid, SIGUSR1, value) == -1) { perror("# sigqueue SIGUSR1 to child"); ret = errno; goto unregister; } /* Give benchmark enough time to fully run */ sleep(1); /* Test runs until the callback setup() tells the test to stop. */ while (1) { ret = param->setup(param); if (ret == END_OF_TESTS) { ret = 0; break; } if (ret < 0) break; if (!strncmp(resctrl_val, MBM_STR, sizeof(MBM_STR)) || !strncmp(resctrl_val, MBA_STR, sizeof(MBA_STR))) { ret = measure_vals(param, &bw_resc_start); if (ret) break; } else if (!strncmp(resctrl_val, CMT_STR, sizeof(CMT_STR))) { sleep(1); ret = measure_cache_vals(param, bm_pid); if (ret) break; } } unregister: signal_handler_unregister(); out: kill(bm_pid, SIGKILL); return ret; }