[LTP] [PATCH v2] pty04: Limit the number of packets sent to avoid timeout

[Richard Palethorpe <rpalethorpe@suse.de>]

Hello,

Cyril Hrubis <chrubis@suse.cz> writes:

> Hi!
>> +static ssize_t try_async_write(int fd, const char *data, ssize_t size,
>> +			       ssize_t *done)
>>  {
>> -	ssize_t ret = write(fd, data, size);
>> +	ssize_t off = done ? *done : 0;
>> +	ssize_t ret = write(fd, data + off, size - off);
>>  
>>  	if (ret < 0)
>>  		return -(errno != EAGAIN);
>>  
>> -	return !written || (*written += ret) >= size;
>> +	if (!done)
>> +		return 1;
>> +
>> +	*done += ret;
>> +	return *done >= size;
>> +}
>> +
>> +static ssize_t try_async_read(int fd, char *data, ssize_t size,
>> +			      ssize_t *done)
>> +{
>> +	ssize_t off = done ? *done : 0;
>> +	ssize_t ret = read(fd, data + off, size - off);
>> +
>> +	if (ret < 0)
>> +		return -(errno != EAGAIN);
>> +
>> +	if (!done)
>> +		return 1;
>> +
>> +	*done += ret;
>> +	return *done >= size;
>>  }
>>  
>> -static void write_pty(const struct ldisc_info *ldisc)
>> +#define RETRY_ASYNC(fn) ({				      \
>> +	ssize_t done = 0;				      \
>> +	TST_RETRY_FUNC(try_async_##fn(ptmx, data, len, &done),\
>> +		       TST_RETVAL_NOTNULL);		      \
>> +})
>
> I do not like this macro that much. Maybe we can have two inline
> functions here one for read and one for write.

OK.

>
> So we do have one process that just reads and one that reads and writes
> right? I wonder if that is okay, maybe we should write twice as much as
> we read in the do_pty()?
>
> Other than that it looks fine.

They both read and write in the final loop. I will make this clearer in
the final while loop.

-- 
Thank you,
Richard.