selftests/resctrl: Remove nested calls in perf event handling
authorIlpo Järvinen <ilpo.jarvinen@linux.intel.com>
Fri, 15 Dec 2023 15:05:00 +0000 (17:05 +0200)
committerShuah Khan <skhan@linuxfoundation.org>
Tue, 13 Feb 2024 20:56:44 +0000 (13:56 -0700)
Perf event handling has functions that are the sole caller of another
perf event handling related function:
  - reset_enable_llc_perf() calls perf_event_open_llc_miss()
  - perf_event_measure() calls get_llc_perf()

Remove the extra layer of calls to make the code easier to follow by
moving the code into the calling function.

Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Reviewed-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: Shuah Khan <skhan@linuxfoundation.org>
tools/testing/selftests/resctrl/cache.c

index 11d1a6ce14c21108e0af48a21693fcd6dc695336..4f846a2e5e26633208dbef241c92e2fe810d463a 100644 (file)
@@ -29,24 +29,13 @@ static void initialize_perf_event_attr(void)
        pea_llc_miss.disabled = 1;
 }
 
+/* Start counters to log values */
 static void ioctl_perf_event_ioc_reset_enable(void)
 {
        ioctl(fd_lm, PERF_EVENT_IOC_RESET, 0);
        ioctl(fd_lm, PERF_EVENT_IOC_ENABLE, 0);
 }
 
-static int perf_event_open_llc_miss(pid_t pid, int cpu_no)
-{
-       fd_lm = perf_event_open(&pea_llc_miss, pid, cpu_no, -1,
-                               PERF_FLAG_FD_CLOEXEC);
-       if (fd_lm == -1) {
-               ksft_perror("Error opening leader");
-               return -1;
-       }
-
-       return 0;
-}
-
 static void initialize_llc_perf(void)
 {
        memset(&pea_llc_miss, 0, sizeof(struct perf_event_attr));
@@ -62,42 +51,13 @@ static void initialize_llc_perf(void)
 
 static int reset_enable_llc_perf(pid_t pid, int cpu_no)
 {
-       int ret = 0;
-
-       ret = perf_event_open_llc_miss(pid, cpu_no);
-       if (ret < 0)
-               return ret;
-
-       /* Start counters to log values */
-       ioctl_perf_event_ioc_reset_enable();
-
-       return 0;
-}
-
-/*
- * get_llc_perf:       llc cache miss through perf events
- * @llc_perf_miss:     LLC miss counter that is filled on success
- *
- * Perf events like HW_CACHE_MISSES could be used to validate number of
- * cache lines allocated.
- *
- * Return: =0 on success.  <0 on failure.
- */
-static int get_llc_perf(__u64 *llc_perf_miss)
-{
-       int ret;
-
-       /* Stop counters after one span to get miss rate */
-
-       ioctl(fd_lm, PERF_EVENT_IOC_DISABLE, 0);
-
-       ret = read(fd_lm, &rf_cqm, sizeof(struct read_format));
-       if (ret == -1) {
-               ksft_perror("Could not get llc misses through perf");
+       fd_lm = perf_event_open(&pea_llc_miss, pid, cpu_no, -1, PERF_FLAG_FD_CLOEXEC);
+       if (fd_lm == -1) {
+               ksft_perror("Error opening leader");
                return -1;
        }
 
-       *llc_perf_miss = rf_cqm.values[0].value;
+       ioctl_perf_event_ioc_reset_enable();
 
        return 0;
 }
@@ -178,14 +138,18 @@ static int print_results_cache(const char *filename, int bm_pid, __u64 llc_value
  */
 static int perf_event_measure(const char *filename, int bm_pid)
 {
-       __u64 llc_perf_miss = 0;
        int ret;
 
-       ret = get_llc_perf(&llc_perf_miss);
-       if (ret < 0)
-               return ret;
+       /* Stop counters after one span to get miss rate */
+       ioctl(fd_lm, PERF_EVENT_IOC_DISABLE, 0);
+
+       ret = read(fd_lm, &rf_cqm, sizeof(struct read_format));
+       if (ret == -1) {
+               ksft_perror("Could not get perf value");
+               return -1;
+       }
 
-       return print_results_cache(filename, bm_pid, llc_perf_miss);
+       return print_results_cache(filename, bm_pid, rf_cqm.values[0].value);
 }
 
 /*