Re: [PATCH] dmaengine: dmatest: timeout value of -1 should specify infinite wait

[Gary R Hook]

On 6/4/19 7:23 AM, Vinod Koul wrote:
> [CAUTION: External Email]
> 
> On 31-05-19, 19:43, Hook, Gary wrote:
>> The dmatest module parameter 'timeout' is documented as accepting a
>> -1 to mean "infinite timeout". Change the parameter to to signed
>> integer, and check the value to call the appropriate wait_event()
>> function.
>>
>> Signed-off-by: Gary R Hook <gary.hook@xxxxxxx>
>> ---
>>   drivers/dma/dmatest.c |   11 ++++++++---
>>   1 file changed, 8 insertions(+), 3 deletions(-)
>>
>> diff --git a/drivers/dma/dmatest.c b/drivers/dma/dmatest.c
>> index b96814a7dceb..28a237686578 100644
>> --- a/drivers/dma/dmatest.c
>> +++ b/drivers/dma/dmatest.c
>> @@ -63,7 +63,7 @@ MODULE_PARM_DESC(pq_sources,
>>                "Number of p+q source buffers (default: 3)");
>>
>>   static int timeout = 3000;
>> -module_param(timeout, uint, S_IRUGO | S_IWUSR);
>> +module_param(timeout, int, S_IRUGO | S_IWUSR);
>>   MODULE_PARM_DESC(timeout, "Transfer Timeout in msec (default: 3000), "
>>                 "Pass -1 for infinite timeout");
>>
>> @@ -795,8 +795,13 @@ static int dmatest_func(void *data)
>>                }
>>                dma_async_issue_pending(chan);
>>
>> -             wait_event_freezable_timeout(thread->done_wait, done->done,
>> -                                          msecs_to_jiffies(params->timeout));
>> +             /* A timeout value of -1 means infinite wait */
>> +             if (timeout == -1)
>> +                     wait_event_freezable(thread->done_wait, done->done);
> 
> well i am not too happy that we have a infinite wait and no way to
> cancel, maybe remove this case?

Well, I was uncomfortable with documentation that didn't match behavior.

I see two choices (and I just chose one to start a conversation):

1) Accept this patch, with an infinite timeout, or
2) Leave the data type alone, but change the description to state that 
timeout values up to hex 0xFFFFFFFF / decimal 4294967295 can be used, 
emulating an "infinite" wait. A very long wait that eventually pops a 
timer is probably preferable. I don't think there are any conversion 
issues since the jiffy parameter to wait_event_freezable_timeout() is 
converted to a long. I could be wrong about that.

I'm happy to go with option (2). Please suggest a course of action.

grh

> 
>> +             else
>> +                     wait_event_freezable_timeout(thread->done_wait,
>> +                                     done->done,
>> +                                     msecs_to_jiffies(params->timeout));
>>
>>                status = dma_async_is_tx_complete(chan, cookie, NULL, NULL);