From mboxrd@z Thu Jan  1 00:00:00 1970
From: Scott Branden <sbranden@broadcom.com>
Subject: Re: [PATCH v2 2/2] hwrng: iproc-rng200 - Add Broadcom IPROC RNG driver
Date: Thu, 26 Feb 2015 14:26:02 -0800
Message-ID: <54EF9D7A.3090707@broadcom.com>
References: <1424888184-22180-3-git-send-email-sbranden@broadcom.com> <3062825.7b1oYP4yuL@wuerfel> <54EF75F0.7000901@broadcom.com> <4488852.TuVxKbAYde@wuerfel>
Mime-Version: 1.0
Content-Type: text/plain; charset="windows-1252"; format=flowed
Content-Transfer-Encoding: 7bit
Return-path: <linux-kernel-owner@vger.kernel.org>
In-Reply-To: <4488852.TuVxKbAYde@wuerfel>
Sender: linux-kernel-owner@vger.kernel.org
To: Arnd Bergmann <arnd@arndb.de>
Cc: bcm-kernel-feedback-list@broadcom.com, Rob Herring <robh+dt@kernel.org>, Pawel Moll <pawel.moll@arm.com>, Mark Rutland <mark.rutland@arm.com>, Ian Campbell <ijc+devicetree@hellion.org.uk>, Kumar Gala <galak@codeaurora.org>, Matt Mackall <mpm@selenic.com>, Herbert Xu <herbert@gondor.apana.org.au>, Grant Likely <grant.likely@linaro.org>, Ray Jui <rjui@broadcom.com>, Jonathan Richardson <jonathar@broadcom.com>, Dmitry Torokhov <dtor@google.com>, Anatol Pomazao <anatol@google.com>, devicetree@vger.kernel.org, linux-arm-kernel@lists.infradead.org, linux-kernel@vger.kernel.org
List-Id: devicetree@vger.kernel.org

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@broadcom.com>
>>>> Signed-off-by: Scott Branden <sbranden@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
>