Sub-NUMA Cluster divides CPUs sharing an L3 cache into separate NUMA nodes. Systems may support splitting into either two, three, four or six nodes. When SNC mode is enabled the effective amount of L3 cache available for allocation is divided by the number of nodes per L3.

It's possible to detect which SNC mode is active by comparing the number of CPUs that share a cache with CPU0, with the number of CPUs on node0.

Detect SNC mode once and let other tests inherit that information.

To check if SNC detection is reliable one can check the /sys/devices/system/cpu/offline file. If it's empty, it means all cores are operational and the ratio should be calculated correctly. If it has any contents, it means the detected SNC mode can't be trusted and should be disabled.

Check if detection was not reliable due to offline cpus. If it was skip running tests since the results couldn't be trusted.

Co-developed-by: Tony Luck tony.luck@intel.com Signed-off-by: Tony Luck tony.luck@intel.com Signed-off-by: Maciej Wieczor-Retman maciej.wieczor-retman@intel.com --- Changelog v6: - Redo snc_nodes_per_l3_cache() by removing old logic and taking into account existance of cpus_offline_empty(). - Remove MAX_SNC since now we just divide cache cpus by node cpus instead of trying to estimate the ratio. - Skip tests if snc_unreliable was set and get_vendor() returns ARCH_INTEL. - Move extern int snc_unreliable; from the next patch to this one since it's needed here already. - Set snc_unreliable if node or cache cpus cannot be counted for some reason. - Redo patch message after code changes.

Changelog v5: - Set snc_mode to 1 at the start of if(!snc_mode) block. - Move all resctrlfs.c code from patch 2/2 to this one. (Reinette) - Fix Co-developed-by tag. - Update SNC discovery comments for the case where it gets disabled by being unreliable. - Remove exclamation mark from ksft_perror() and add full file path instead of "offline CPUs file". - bit map -> bitmap. - Remove unnecessary empty line.

Changelog v4: - Make returned value a static local variable so the function only runs the logic once. (Reinette)

Changelog v3: - Add comparison between present and online cpus to test if the calculated SNC mode is credible. (Reinette) - Added comment to cache size modification to better explain why it is needed there. (Reinette) - Fix facts in patch message. (Reinette) - Change snc_ways() to snc_nodes_per_l3_cache(). (Reinette)

tools/testing/selftests/resctrl/resctrl.h | 4 + .../testing/selftests/resctrl/resctrl_tests.c | 9 +- tools/testing/selftests/resctrl/resctrlfs.c | 97 +++++++++++++++++++ 3 files changed, 109 insertions(+), 1 deletion(-)

diff --git a/tools/testing/selftests/resctrl/resctrl.h b/tools/testing/selftests/resctrl/resctrl.h index dab1953fc7a0..38dfe5a03fcd 100644 --- a/tools/testing/selftests/resctrl/resctrl.h +++ b/tools/testing/selftests/resctrl/resctrl.h @@ -11,6 +11,7 @@ #include <signal.h> #include <dirent.h> #include <stdbool.h> +#include <ctype.h> #include <sys/stat.h> #include <sys/ioctl.h> #include <sys/mount.h> @@ -156,8 +157,11 @@ struct perf_event_read { */ extern volatile int *value_sink;

+extern int snc_unreliable; + extern char llc_occup_path[1024];

+int snc_nodes_per_l3_cache(void); int get_vendor(void); bool check_resctrlfs_support(void); int filter_dmesg(void); diff --git a/tools/testing/selftests/resctrl/resctrl_tests.c b/tools/testing/selftests/resctrl/resctrl_tests.c index 3335af815b21..5154ffd821c4 100644 --- a/tools/testing/selftests/resctrl/resctrl_tests.c +++ b/tools/testing/selftests/resctrl/resctrl_tests.c @@ -118,7 +118,7 @@ static bool test_vendor_specific_check(const struct resctrl_test *test)

static void run_single_test(const struct resctrl_test *test, const struct user_params *uparams) { - int ret; + int ret, snc_mode;

if (test->disabled) return; @@ -128,8 +128,15 @@ static void run_single_test(const struct resctrl_test *test, const struct user_p return; }

+ snc_mode = snc_nodes_per_l3_cache(); + ksft_print_msg("Starting %s test ...\n", test->name);

+ if (snc_mode == 1 && snc_unreliable && get_vendor() == ARCH_INTEL) { + ksft_test_result_skip("SNC detection unreliable due to offline CPUs. Test results may not be accurate if SNC enabled.\n"); + return; + } + if (test_prepare(test)) { ksft_exit_fail_msg("Abnormal failure when preparing for the test\n"); return; diff --git a/tools/testing/selftests/resctrl/resctrlfs.c b/tools/testing/selftests/resctrl/resctrlfs.c index d38d6dd90be4..50561993d37c 100644 --- a/tools/testing/selftests/resctrl/resctrlfs.c +++ b/tools/testing/selftests/resctrl/resctrlfs.c @@ -13,6 +13,8 @@

#include "resctrl.h"

+int snc_unreliable; + static int find_resctrl_mount(char *buffer) { FILE *mounts; @@ -156,6 +158,90 @@ int get_domain_id(const char *resource, int cpu_no, int *domain_id) return 0; }

+/* + * Count number of CPUs in a /sys bitmap + */ +static unsigned int count_sys_bitmap_bits(char *name) +{ + FILE *fp = fopen(name, "r"); + int count = 0, c; + + if (!fp) + return 0; + + while ((c = fgetc(fp)) != EOF) { + if (!isxdigit(c)) + continue; + switch (c) { + case 'f': + count++; + case '7': case 'b': case 'd': case 'e': + count++; + case '3': case '5': case '6': case '9': case 'a': case 'c': + count++; + case '1': case '2': case '4': case '8': + count++; + } + } + fclose(fp); + + return count; +} + +static bool cpus_offline_empty(void) +{ + char offline_cpus_str[64]; + FILE *fp; + + fp = fopen("/sys/devices/system/cpu/offline", "r"); + if (fscanf(fp, "%s", offline_cpus_str) < 0) { + if (!errno) { + fclose(fp); + return 1; + } + ksft_perror("Could not read /sys/devices/system/cpu/offline"); + } + + fclose(fp); + + return 0; +} + +/* + * Detect SNC by comparing #CPUs in node0 with #CPUs sharing LLC with CPU0. + * If any CPUs are offline declare the detection as unreliable and skip the + * tests. + */ +int snc_nodes_per_l3_cache(void) +{ + int node_cpus, cache_cpus; + static int snc_mode; + + if (!snc_mode) { + snc_mode = 1; + if (!cpus_offline_empty()) { + ksft_print_msg("Runtime SNC detection unreliable due to offline CPUs.\n"); + ksft_print_msg("Setting SNC mode to disabled.\n"); + snc_unreliable = 1; + return snc_mode; + } + node_cpus = count_sys_bitmap_bits("/sys/devices/system/node/node0/cpumap"); + cache_cpus = count_sys_bitmap_bits("/sys/devices/system/cpu/cpu0/cache/index3/shared_cpu_map"); + + if (!node_cpus || !cache_cpus) { + ksft_print_msg("Could not determine Sub-NUMA Cluster mode.\n"); + snc_unreliable = 1; + return snc_mode; + } + snc_mode = cache_cpus / node_cpus; + + if (snc_mode > 1) + ksft_print_msg("SNC-%d mode discovered.\n", snc_mode); + } + + return snc_mode; +} + /* * get_cache_size - Get cache size for a specified CPU * @cpu_no: CPU number @@ -211,6 +297,17 @@ int get_cache_size(int cpu_no, const char *cache_type, unsigned long *cache_size break; }

+ /* + * The amount of cache represented by each bit in the masks + * in the schemata file is reduced by a factor equal to SNC + * nodes per L3 cache. + * E.g. on a SNC-2 system with a 100MB L3 cache a test that + * allocates memory from its local SNC node (default behavior + * without using libnuma) will only see 50 MB llc_occupancy + * with a fully populated L3 mask in the schemata file. + */ + if (cache_num == 3) + *cache_size /= snc_nodes_per_l3_cache(); return 0; }