Re: [PATCH v12 3/3]: perf record: extend trace writing to multi AIO

[Alexey Budankov <alexey.budankov@linux.intel.com>]

On 11.10.2018 16:46, Jiri Olsa wrote:
> On Tue, Oct 09, 2018 at 11:59:31AM +0300, Alexey Budankov wrote:
> 
> SNIP
> 
>> @@ -709,12 +729,13 @@ static int record__mmap_read_evlist(struct record *rec, struct perf_evlist *evli
>>  					goto out;
>>  				}
>>  			} else {
>> +				int idx;
>>  				/*
>>  				 * Call record__aio_sync() to wait till map->data buffer
>>  				 * becomes available after previous aio write request.
>>  				 */
>> -				record__aio_sync(map);
>> -				if (perf_mmap__aio_push(map, rec, record__aio_pushfn, &off) != 0) {
>> +				idx = record__aio_sync(map, false);
> 
> looks like record__aio_sync can return -1,
> how is it not breaking the perf_mmap__aio_push call below?

Currently -1 may be returned only when calling with 
sync_all==true i.e. record__aio_sync(, true).

Thanks,
Alexey

> 
>> +				if (perf_mmap__aio_push(map, rec, idx, record__aio_pushfn, &off) != 0) {
>>  					record__aio_set_pos(trace_fd, off);
>>  					rc = -1;
>>  					goto out;
> 
> SNIP
>