[Intel-gfx] [PATCH i-g-t 1/1] tools/intel_guc_logger: Send GuC log level in new i915 expected format

[Sagar Arun Kamble]

On 1/12/2018 6:51 PM, Michal Wajdeczko wrote:
> On Fri, 12 Jan 2018 07:52:04 +0100, Sagar Arun Kamble 
> <sagar.a.kamble at intel.com> wrote:
>
>> i915 expects GuC log level to be specified as:
>>     0: disabled
>>     1: enabled (verbosity level 0 = min)
>>     2: enabled (verbosity level 1)
>>     3: enabled (verbosity level 2)
>>     4: enabled (verbosity level 3 = max)
>>
>> Remove the earlier internal layout based logging control from
>> guc_log_control and send new expected values.
>>
>> Signed-off-by: Sagar Arun Kamble <sagar.a.kamble at intel.com>
>> Cc: Michal Wajdeczko <michal.wajdeczko at intel.com>
>> Cc: Daniele Ceraolo Spurio <daniele.ceraolospurio at intel.com>
>> Cc: Chris Wilson <chris at chris-wilson.co.uk>
>> ---
>
> Reviewed-by: Michal Wajdeczko <michal.wajdeczko at intel.com>
Thanks for the review. Will shared updated rev.
>
> with small bikeshedding below...
>
>>  tools/intel_guc_logger.c | 21 +++++++++++++++------
>>  1 file changed, 15 insertions(+), 6 deletions(-)
>>
>> diff --git a/tools/intel_guc_logger.c b/tools/intel_guc_logger.c
>> index 031fd84..e695497 100644
>> --- a/tools/intel_guc_logger.c
>> +++ b/tools/intel_guc_logger.c
>> @@ -51,18 +51,27 @@ uint32_t test_duration, max_filesize;
>>  pthread_cond_t underflow_cond, overflow_cond;
>>  bool stop_logging, discard_oldlogs, capturing_stopped;
>> -static void guc_log_control(bool enable_logging)
>> +static void guc_log_control(bool enable, uint32_t log_level)
>>  {
>>      int control_fd;
>>      char data[19];
>> -    uint64_t val;
>> +    uint64_t val = 0;
>>      int ret;
>
> Btw, this shouldn't hurt:
>
>     igt_assert_lte(log_level, 3);
>
Yes
>> +    /*
>> +     * i915 expects GuC log level to be specified as:
>> +     * 0: disabled
>> +     * 1: enabled (verbosity level 0 = min)
>> +     * 2: enabled (verbosity level 1)
>> +     * 3: enabled (verbosity level 2)
>> +     * 4: enabled (verbosity level 3 = max)
>> +     */
>> +    if (enable)
>> +        val = log_level + 1;
>
> control = enable ? log_level + 1 : 0;
>
Ok
>> +
>>      control_fd = igt_debugfs_open(-1, CONTROL_FILE_NAME, O_WRONLY);
>>      igt_assert_f(control_fd >= 0, "couldn't open the guc log control 
>> file\n");
>> -    val = enable_logging ? ((verbosity_level << 4) | 0x1) : 0;
>> -
>>      ret = snprintf(data, sizeof(data), "0x%" PRIx64, val);
>
> Btw, I'm wondering why we didn't use "fprintf(control_fd, ...)" here
>
I remember that using read instead of fread was optimization to speed up 
the buffering of logs.
For consistency I believe write is used instead fwrite.
>>      igt_assert(ret > 2 && ret < sizeof(data));
>> @@ -288,7 +297,7 @@ static void init_main_thread(void)
>>      /* Enable the logging, it may not have been enabled from boot 
>> and so
>>       * the relay file also wouldn't have been created.
>>       */
>> -    guc_log_control(true);
>> +    guc_log_control(true, verbosity_level);
>>     open_relay_file();
>>      open_output_file();
>> @@ -420,7 +429,7 @@ int main(int argc, char **argv)
>>      } while (!stop_logging);
>>     /* Pause logging on the GuC side */
>> -    guc_log_control(false);
>> +    guc_log_control(false, 0);
>>     /* Signal flusher thread to make an exit */
>>      capturing_stopped = 1;