[igt-dev] [PATCH i-g-t v2] tools/i915-perf: workaround overzelous compiler warnings

Lionel Landwerlin lionel.g.landwerlin at intel.com
Fri Feb 21 10:29:21 UTC 2020


On 21/02/2020 12:23, Petri Latvala wrote:
> On Fri, Feb 21, 2020 at 12:17:24PM +0200, Lionel Landwerlin wrote:
>> On 21/02/2020 12:09, Petri Latvala wrote:
>>> On Thu, Feb 20, 2020 at 03:34:16PM +0200, Lionel Landwerlin wrote:
>>>> Take 2.
>>>>
>>>> Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin at intel.com>
>>>> ---
>>>>    tools/i915-perf/i915_perf_recorder_commands.h | 3 ++-
>>>>    1 file changed, 2 insertions(+), 1 deletion(-)
>>>>
>>>> diff --git a/tools/i915-perf/i915_perf_recorder_commands.h b/tools/i915-perf/i915_perf_recorder_commands.h
>>>> index 4855d80f..44dd4438 100644
>>>> --- a/tools/i915-perf/i915_perf_recorder_commands.h
>>>> +++ b/tools/i915-perf/i915_perf_recorder_commands.h
>>>> @@ -35,5 +35,6 @@ struct recorder_command_base {
>>>>    };
>>>>    struct recorder_command_dump {
>>>> -	uint8_t path[0];
>>>> +	uint8_t unused;
>>>> +	uint8_t path[];
>>>>    };
>>> And now all your mallocs need + 1.
>>
>> You're right, I forgot the sizeof(struct recorder_command_dump).
>>
>>
>>> Why do you insist on having two structs?
>>
>> I want the person reading this code to understand what the packets exchanged
>> look like.
> Fair enough.
>
>>
>>>    The recorder_command_dump
>>> struct only acts as "here be an array" object and you don't do
>>> anything to it as an object anywhere.
>>>
>>> A drive-by note, i915_perf_recorder_commands.h lacks include guards.
>>>
>>>
>>>
>>> diff --git a/tools/i915-perf/i915_perf_control.c b/tools/i915-perf/i915_perf_control.c
>>> index a8d0d30f..be5996c0 100644
>>> --- a/tools/i915-perf/i915_perf_control.c
>>> +++ b/tools/i915-perf/i915_perf_control.c
>>> @@ -93,12 +93,12 @@ main(int argc, char *argv[])
>>>    				sizeof(struct recorder_command_base) + strlen(dump_file) + 1;
>>>    			struct {
>>>    				struct recorder_command_base base;
>>> -				struct recorder_command_dump dump;
>>> +				uint8_t dump[];
>>>    			} *data = malloc(total_len);
>>>    			data->base.command = RECORDER_COMMAND_DUMP;
>>>    			data->base.size = total_len;
>>> -			snprintf((char *) data->dump.path, strlen(dump_file) + 1, "%s", dump_file);
>>> +			snprintf((char *) data->dump, strlen(dump_file) + 1, "%s", dump_file);
>>>    			fwrite(data, total_len, 1, command_fifo_file);
>>>    		} else {
>>> @@ -107,12 +107,12 @@ main(int argc, char *argv[])
>>>    			uint32_t total_len = sizeof(struct recorder_command_base) + path_len;
>>>    			struct {
>>>    				struct recorder_command_base base;
>>> -				struct recorder_command_dump dump;
>>> +				uint8_t dump[];
>>>    			} *data = malloc(total_len);
>>>    			data->base.command = RECORDER_COMMAND_DUMP;
>>>    			data->base.size = total_len;
>>> -			snprintf((char *) data->dump.path, path_len, "%s/%s", cwd, dump_file);
>>> +			snprintf((char *) data->dump, path_len, "%s/%s", cwd, dump_file);
>>>    			fwrite(data, total_len, 1, command_fifo_file);
>>>    		}
>>> diff --git a/tools/i915-perf/i915_perf_recorder.c b/tools/i915-perf/i915_perf_recorder.c
>>> index 760cabf1..6bbc451e 100644
>>> --- a/tools/i915-perf/i915_perf_recorder.c
>>> +++ b/tools/i915-perf/i915_perf_recorder.c
>>> @@ -605,7 +605,7 @@ read_command_file(struct recording_context *ctx)
>>>    	switch (header.command) {
>>>    	case RECORDER_COMMAND_DUMP: {
>>>    		uint32_t len = header.size - sizeof(header), offset = 0;
>>> -		struct recorder_command_dump *dump = malloc(len);
>>> +		uint8_t *dump = malloc(len);
>>>    		FILE *file;
>>>    		while (offset < len &&
>>> @@ -616,9 +616,9 @@ read_command_file(struct recording_context *ctx)
>>>    				offset += ret;
>>>    		}
>>> -		fprintf(stdout, "Writing circular buffer to %s\n", dump->path);
>>> +		fprintf(stdout, "Writing circular buffer to %s\n", dump);
>>> -		file = fopen((const char *) dump->path, "w+");
>>> +		file = fopen((const char *) dump, "w+");
>>>    		if (file) {
>>>    			struct chunk chunks[2];
>>> @@ -634,11 +634,11 @@ read_command_file(struct recording_context *ctx)
>>>    			     fwrite(chunks[1].data, chunks[1].len, 1, file) != 1) ||
>>>    			    !write_correlation_timestamps(file, ctx->drm_fd)) {
>>>    				fprintf(stderr, "Unable to write circular buffer data in file '%s'\n",
>>> -					dump->path);
>>> +					dump);
>>>    			}
>>>    			fclose(file);
>>>    		} else
>>> -			fprintf(stderr, "Unable to write dump file '%s'\n", dump->path);
>>> +			fprintf(stderr, "Unable to write dump file '%s'\n", dump);
>>>    		free(dump);
>>>    		break;
>>> diff --git a/tools/i915-perf/i915_perf_recorder_commands.h b/tools/i915-perf/i915_perf_recorder_commands.h
>>> index 4855d80f..3db11327 100644
>>> --- a/tools/i915-perf/i915_perf_recorder_commands.h
>>> +++ b/tools/i915-perf/i915_perf_recorder_commands.h
>>> @@ -33,7 +33,3 @@ struct recorder_command_base {
>>>    	uint32_t command;
>>>    	uint32_t size;
>>>    };
>>> -
>>> -struct recorder_command_dump {
>>> -	uint8_t path[0];
>>> -};
>>>
>> Can you leave this structure in there but commented?
>>
>> Just so that there is a trace of what you is exchanged between the 2
>> processes.
> Sure, and the size field can have a comment explaining that it's the
> size of the header + dump.
>
> I'm reading that as an implied consent to wrapping the patch in a
> commit with your A-b, is that correct?
>
>
Yep :)


Thanks again,


-Lionel



More information about the igt-dev mailing list