| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <e2755213-0ea1-4b77-886c-294068940bd1@amd.com>
Date: Thu, 20 Jun 2024 11:18:40 -0500
From: "Moger, Babu" <babu.moger@....com>
To: Reinette Chatre <reinette.chatre@...el.com>, fenghua.yu@...el.com,
shuah@...nel.org
Cc: linux-kernel@...r.kernel.org, linux-kselftest@...r.kernel.org,
ilpo.jarvinen@...ux.intel.com, maciej.wieczor-retman@...el.com,
peternewman@...gle.com, eranian@...gle.com
Subject: Re: [PATCH v3 3/4] selftests/resctrl: Add support for MBM and MBA
tests on AMD
Hi Reinette,
On 6/14/24 13:39, Reinette Chatre wrote:
> Hi Babu,
>
> On 6/5/24 3:45 PM, Babu Moger wrote:
>> Add support to read UMC (Unified Memory Controller) perf events to compare
>> the numbers with QoS monitor for AMD.
>>
>> Signed-off-by: Babu Moger <babu.moger@....com>
>> ---
>> v3: Made read_from_mc_dir function generic to both AMD and Intel.
>> Rest are mostly related to rebase.
>>
>> v2: Replace perror with ksft_perror.
>> ---
>> tools/testing/selftests/resctrl/resctrl_val.c | 80 ++++++++++++-------
>> 1 file changed, 50 insertions(+), 30 deletions(-)
>>
>> diff --git a/tools/testing/selftests/resctrl/resctrl_val.c
>> b/tools/testing/selftests/resctrl/resctrl_val.c
>> index 23c0e0a1d845..ffacafb535cd 100644
>> --- a/tools/testing/selftests/resctrl/resctrl_val.c
>> +++ b/tools/testing/selftests/resctrl/resctrl_val.c
>> @@ -11,6 +11,7 @@
>> #include "resctrl.h"
>> #define UNCORE_IMC "uncore_imc_"
>> +#define AMD_UMC "amd_umc_"
>> #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"
>> @@ -128,7 +129,7 @@ static int open_perf_event(int i, int cpu_no, int j)
>> }
>> /* Get type and config (read and write) of an MC counter */
>> -static int read_from_mc_dir(char *mc_dir, int count)
>> +static int read_from_mc_dir(char *mc_dir, int count, int vendor)
>> {
>> char cas_count_cfg[1024], mc_counter_cfg[1024],
>> mc_counter_type[1024];
>> FILE *fp;
>> @@ -152,41 +153,56 @@ static int read_from_mc_dir(char *mc_dir, int count)
>> mc_counters_config[count][WRITE].type =
>> mc_counters_config[count][READ].type;
>> - /* Get read config */
>> - sprintf(mc_counter_cfg, "%s%s", mc_dir, READ_FILE_NAME);
>> - fp = fopen(mc_counter_cfg, "r");
>> - if (!fp) {
>> - ksft_perror("Failed to open MC config file");
>> + if (vendor == ARCH_AMD) {
>> + /*
>> + * Setup the event and umasks for UMC events
>> + * Number of CAS commands sent for reads:
>> + * EventCode = 0x0a, umask = 0x1
>> + * Number of CAS commands sent for writes:
>> + * EventCode = 0x0a, umask = 0x2
>> + */
>> + mc_counters_config[count][READ].event = 0xa;
>> + mc_counters_config[count][READ].umask = 0x1;
>> - return -1;
>> - }
>> - if (fscanf(fp, "%s", cas_count_cfg) <= 0) {
>> - ksft_perror("Could not get MC cas count read");
>> + mc_counters_config[count][WRITE].event = 0xa;
>> + mc_counters_config[count][WRITE].umask = 0x2;
>> + } else {
>> + /* Get read config */
>> + sprintf(mc_counter_cfg, "%s%s", mc_dir, READ_FILE_NAME);
>> + fp = fopen(mc_counter_cfg, "r");
>> + if (!fp) {
>> + ksft_perror("Failed to open MC config file");
>> +
>> + return -1;
>> + }
>> + if (fscanf(fp, "%s", cas_count_cfg) <= 0) {
>> + ksft_perror("Could not get MC cas count read");
>> + fclose(fp);
>> +
>> + return -1;
>> + }
>> fclose(fp);
>> - return -1;
>> - }
>> - fclose(fp);
>> + get_event_and_umask(cas_count_cfg, count, READ);
>> - get_event_and_umask(cas_count_cfg, count, READ);
>> + /* Get write config */
>> + sprintf(mc_counter_cfg, "%s%s", mc_dir, WRITE_FILE_NAME);
>> + fp = fopen(mc_counter_cfg, "r");
>> + if (!fp) {
>> + ksft_perror("Failed to open MC config file");
>> - /* Get write config */
>> - sprintf(mc_counter_cfg, "%s%s", mc_dir, WRITE_FILE_NAME);
>> - fp = fopen(mc_counter_cfg, "r");
>> - if (!fp) {
>> - ksft_perror("Failed to open MC config file");
>> + return -1;
>> + }
>> + if (fscanf(fp, "%s", cas_count_cfg) <= 0) {
>> + ksft_perror("Could not get MC cas count write");
>> + fclose(fp);
>> - return -1;
>> - }
>> - if (fscanf(fp, "%s", cas_count_cfg) <= 0) {
>> - ksft_perror("Could not get MC cas count write");
>> + return -1;
>> + }
>> fclose(fp);
>> - return -1;
>> + get_event_and_umask(cas_count_cfg, count, WRITE);
>> }
>> - fclose(fp);
>> -
>> - get_event_and_umask(cas_count_cfg, count, WRITE);
>> return 0;
>> }
>> @@ -213,6 +229,8 @@ static int num_of_mem_controllers(void)
>> vendor = get_vendor();
>> if (vendor == ARCH_INTEL) {
>> sysfs_name = UNCORE_IMC;
>> + } else if (vendor == ARCH_AMD) {
>> + sysfs_name = AMD_UMC;
>> } else {
>> ksft_perror("Unsupported vendor!\n");
>> return -1;
>> @@ -228,6 +246,7 @@ static int num_of_mem_controllers(void)
>> /*
>> * imc counters are named as "uncore_imc_<n>", hence
>> * increment the pointer to point to <n>.
>> + * For AMD, it will be amd_umc_<n>.
>> */
>> temp = temp + strlen(sysfs_name);
>> @@ -239,7 +258,7 @@ static int num_of_mem_controllers(void)
>> if (temp[0] >= '0' && temp[0] <= '9') {
>> sprintf(mc_dir, "%s/%s/", DYN_PMU_PATH,
>> ep->d_name);
>> - ret = read_from_mc_dir(mc_dir, count);
>> + ret = read_from_mc_dir(mc_dir, count, vendor);
>> if (ret) {
>> closedir(dp);
>> @@ -250,8 +269,9 @@ static int num_of_mem_controllers(void)
>> }
>> closedir(dp);
>> if (count == 0) {
>> - ksft_print_msg("Unable to find MC counters\n");
>> -
>> + ksft_print_msg("Unable to find iMC/UMC counters\n");
>> + if (vendor == ARCH_AMD)
>> + ksft_print_msg("Try loading amd_uncore module\n");
>> return -1;
>> }
>> } else {
>
> Can all the vendor checking be contained in num_of_mem_controllers()
> instead of
> scattered through multiple layers? There can be two vendor specific
> functions to
> initialize mc_counters_config[][]. Only the type setting code ends up
> being shared so that can be split into a function that is called by both
> vendor functions?
Yes, We can do that. Will add it in v4.
--
Thanks
Babu Moger
Powered by blists - more mailing lists