[PATCH] cpufreq: Add test cases to validate the frequency and pstate table

[Shilpasri G Bhat]

Hi,

On 04/22/2016 01:47 PM, Colin Ian King wrote:
> On 22/04/16 05:51, Shilpasri G Bhat wrote:
>> Add test case to parse CPU frequency and pstate table to check for
>> invalid entries that are beyond the specified minimum and maximum
>> limits.
>>
>> Signed-off-by: Shilpasri G Bhat <shilpa.bhat at linux.vnet.ibm.com>
>> ---
>>  src/cpu/cpufreq/cpufreq.c | 139 ++++++++++++++++++++++++++++++++++++++++++++++
>>  1 file changed, 139 insertions(+)
>>
>> diff --git a/src/cpu/cpufreq/cpufreq.c b/src/cpu/cpufreq/cpufreq.c
>> index 8ab5292..1a9c9fd 100644
>> --- a/src/cpu/cpufreq/cpufreq.c
>> +++ b/src/cpu/cpufreq/cpufreq.c
>> @@ -30,6 +30,7 @@
>>  #include <unistd.h>
>>  #include <sys/types.h>
>>  #include <sys/stat.h>
>> +#include <fcntl.h>
>>  #include <limits.h>
>>  #include <dirent.h>
>>  #include <stdint.h>
>> @@ -393,6 +394,142 @@ static int cpufreq_test_cpu_performance(fwts_framework *fw)
>>  	return FWTS_OK;
>>  }
>>  
>> +static int cpufreq_test_frequency_limits(fwts_framework *fw)
>> +{
>> +	char path[PATH_MAX];
>> +	int min, max, i;
>> +	bool ok = true;
>> +
>> +	cpu_mkpath(path, sizeof(path), &cpus[0], "cpuinfo_min_freq");
>> +	if (fwts_get_int(path, &min) != FWTS_OK) {
>> +		fwts_log_warning(fw, "Failed to read the minimum CPU frequency %s",
>> +				 path);
>> +		return FWTS_ERROR;
>> +	}
>> +
>> +	cpu_mkpath(path, sizeof(path), &cpus[0], "cpuinfo_max_freq");
>> +	if (fwts_get_int(path, &max) != FWTS_OK) {
>> +		fwts_log_warning(fw, "Failed to read the maximum CPU frequency %s",
>> +				 path);
>> +		return FWTS_ERROR;
>> +	}
>> +
>> +	for (i = 0; i < cpus[0].n_freqs; i++) {
>> +		if (cpus[0].freqs[i].Hz > (uint64_t)max) {
>> +			fwts_log_warning(fw, "%s is greater than max frequency %s",
>> +					 hz_to_human(cpus[0].freqs[i].Hz),
>> +					 hz_to_human(max));
>> +			ok = false;
>> +		}
>> +		if (cpus[0].freqs[i].Hz < (uint64_t)min) {
>> +			fwts_log_warning(fw, "%s is lesser than min frequency %s",
>> +					 hz_to_human(cpus[0].freqs[i].Hz),
>> +					 hz_to_human(min));
>> +			ok = false;
>> +		}
>> +	}
>> +	if (ok)
>> +		fwts_passed(fw, "CPU frequencies are within the min and max limits");
>> +	else
>> +		fwts_failed(fw, LOG_LEVEL_MEDIUM, "CPUFreqLimitsTest",
>> +			     "Contains frequencies beyond max and min limits");
>> +	return FWTS_OK;
>> +}
>> +
>> +static int read_device_tree_file(const char *path, int *val)
>> +{
>> +	int fd;
>> +
>> +	fd = open(path, O_RDONLY);
>> +	if (fd < 0)
>> +		return fd;
>> +
>> +	if (read(fd, val, sizeof(*val)) < 0)
>> +		return FWTS_ERROR;
>> +
>> +	*val = be32toh(*val);
>> +	close(fd);
>> +	return FWTS_OK;
>> +}
>> +
>> +static int read_device_tree_array(const char *path, int *arr, int *len)
>> +{
>> +	int fd;
>> +	int i = 0;
>> +
>> +	fd = open(path, O_RDONLY);
>> +	if (fd < 0)
>> +		return fd;
>> +
>> +	while(1) {
>> +		if (read(fd, (arr + i), sizeof(arr[i])) <= 0)
>> +			break;
>> +		arr[i] = be32toh(arr[i]);
>> +		i++;
>> +	}
>> +
>> +	*len = (!i) ? i : i - 1;
>> +	close(fd);
>> +	return FWTS_OK;
>> +}
>> +
>> +static int cpufreq_test_pstate_limits(fwts_framework *fw)
>> +{
>> +	char *dev_tree_path = "/proc/device-tree/ibm,opal/power-mgt/";
>> +	int pstate_min, pstate_max, pstates[MAX_FREQS], nr_pstates, len;
>> +	char path[PATH_MAX];
>> +	bool ok = true;
>> +	int i;
>> +
>> +#if !defined(__PPC64__)
>> +	fwts_skipped(fw, "Test case applicable to PPC64 only");
>> +	return FWTS_SKIP;
>> +#endif
>> +
>> +	snprintf(path, sizeof(path), "%sibm,pstate-max", dev_tree_path);
>> +	if (read_device_tree_file(path, &pstate_max) < 0) {
>> +		fwts_log_warning(fw, "Failed to read the max pstate %s\n", path);
>> +		return FWTS_ERROR;
>> +	}
>> +
>> +	snprintf(path, sizeof(path), "%sibm,pstate-min", dev_tree_path);
>> +	if (read_device_tree_file(path, &pstate_min) < 0) {
>> +		fwts_log_warning(fw, "Failed to read the min pstate %s\n", path);
>> +		return FWTS_ERROR;
>> +	}
>> +
>> +	snprintf(path, sizeof(path), "%sibm,pstate-ids", dev_tree_path);
>> +	if (read_device_tree_array(path, pstates, &len) < 0 || !len) {
>> +		fwts_log_warning(fw, "Failed to read the pstate-id array %s\n", path);
>> +		return FWTS_ERROR;
>> +	}
>> +
>> +	nr_pstates = pstate_max - pstate_min;
>> +	if (len != nr_pstates)
>> +		fwts_log_warning(fw, "Wrong number of pstates. Expected %d pstates, but found %d pstates\n",
>> +				 nr_pstates, len);
>> +
>> +	for (i = 0; i < nr_pstates; i++) {
>> +		if (pstates[i] > pstate_max) {
>> +			fwts_log_warning(fw, "Invalid Pstate id %d greater than max pstate %d\n",
>> +					  pstates[i], pstate_max);
>> +			ok = false;
>> +		}
>> +		if (pstates[i] < pstate_min) {
>> +			fwts_log_warning(fw, "Invalid Pstate id %d lesser than min pstate %d\n",
>> +					  pstates[i], pstate_min);
>> +			ok = false;
>> +		}
>> +	}
>> +
>> +	if (ok)
>> +		fwts_passed(fw, "CPU pstates are within the min and max limits");
>> +	else
>> +		fwts_failed(fw, LOG_LEVEL_MEDIUM, "CPUPstateLimitsTest",
>> +			     "Contains pstates beyond max and min limits");
>> +	return FWTS_OK;
>> +}
>> +
>>  static int sw_tests_possible(fwts_framework *fw)
>>  {
>>  	int i, online_cpus = 0;
>> @@ -861,6 +998,8 @@ static fwts_framework_minor_test cpufreq_tests[] = {
>>  	{ cpufreq_test_sw_any,		"CPU frequency SW_ANY control" },
>>  	{ cpufreq_test_sw_all,		"CPU frequency SW_ALL control" },
>>  	{ cpufreq_test_cpu_performance,	"CPU frequency performance tests." },
>> +	{ cpufreq_test_frequency_limits, "CPU frequency limits test" },
> 
> Would it make more sense to make the following line a powerpc64 build
> only option:
> 
> + #if defined(__PPC64__)
>> +	{ cpufreq_test_pstate_limits,	"CPU Pstate limits test (device_tree)" },
> + #endif
> 

Yes I will make this change and post.

Thanks and Regards,
Shilpa


>>  	{ NULL, NULL }
>>  };
>>  
>>
>