[PATCH v2 2/2] hwrng: iproc-rng200 - Add Broadcom IPROC RNG driver

[Scott Branden]

Hi Arnd,

Latency is 32 us for 32bits of data - commented inline.  What delay call 
do you recommend in this case?

On 15-02-26 12:15 PM, Arnd Bergmann wrote:
> On Thursday 26 February 2015 11:37:20 Scott Branden wrote:
>> Response inline.
>>
>> On 15-02-25 11:17 AM, Arnd Bergmann wrote:
>>> On Wednesday 25 February 2015 10:16:24 Scott Branden wrote:
>>>> This adds a driver for random number generator present on Broadcom
>>>> IPROC devices.
>>>>
>>>> Reviewed-by: Ray Jui <rjui at broadcom.com>
>>>> Signed-off-by: Scott Branden <sbranden at broadcom.com>
>>>
>>> The driver looks reasonable overall, I have just one question about
>>> something that sticks out:
>>>
>>>> +    while ((num_remaining > 0) && time_before(jiffies, idle_endtime)) {
>>> ...
>>>> +
>>>> +            /* Are there any random numbers available? */
>>>> +            if ((ioread32(rng_base + RNG_FIFO_COUNT_OFFSET) &
>>>> +                            RNG_FIFO_COUNT_RNG_FIFO_COUNT_MASK) > 0) {
>>> ...
>>>> +            } else {
>>>> +                    if (!wait)
>>>> +                            /* Cannot wait, return immediately */
>>>> +                            return max - num_remaining;
>>>> +
>>>> +                    /* Can wait, give others chance to run */
>>>> +                    cpu_relax();
>>>> +            }
>>>> +    }
>>>> +
>>>
>>> It looks like you do a busy-loop around cpu_relax here if asked to wait.
>>> Is this intentional? I would normally expect either cond_resched() or
>>> some msleep() instead.
>>
>> This code was following examples of other open source drivers - bcm2835
>> and exynos both use cpu_relax.  I'll have to look into this more to
>> understand.
>>
>
> The majority of the driver apparently use udelay(10) to wait, which is
> not much better but at least consistent. The cpu_relax() call probably
> gives better throughput.
>
> I don't understand why none of the drivers actually attempts to
> msleep(), but that may be because the delay is much too long.
>
> Can you find out what the expected latency is for new data to
> become available on your hardware?
RNG generates at a nominal 1 Mbps.  So to generate 32 bits of data takes
approximately 32 us.

>
> 	Arnd
>