linux-kernel - Re: [PATCH 2/2] hwrng: mtk: Add driver for hardware random generator on MT7623 SoC

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Fri, 14 Apr 2017 10:27:51 +0530
From:   PrasannaKumar Muralidharan <prasannatsmkumar@...il.com>
To:     Sean Wang <sean.wang@...iatek.com>
Cc:     Herbert Xu <herbert@...dor.apana.org.au>,
        Matt Mackall <mpm@...enic.com>,
        Rob Herring <robh+dt@...nel.org>,
        Mark Rutland <mark.rutland@....com>,
        Corentin LABBE <clabbe.montjoie@...il.com>,
        Romain Perier <romain.perier@...e-electrons.com>,
        shannon.nelson@...cle.com, Wei Yongjun <weiyongjun1@...wei.com>,
        devicetree@...r.kernel.org, linux-crypto@...r.kernel.org,
        linux-mediatek@...ts.infradead.org,
        linux-arm-kernel@...ts.infradead.org, linux-kernel@...r.kernel.org,
        sean wang <keyhaede@...il.com>
Subject: Re: [PATCH 2/2] hwrng: mtk: Add driver for hardware random generator
 on MT7623 SoC

On 14 April 2017 at 09:28, Sean Wang <sean.wang@...iatek.com> wrote:
>
> Hi PrasannaKumar,
>
> Add my comments inline
>
>>
>> Use readl_poll_timeout_atomic's return value or -EIO instead of
>> !!ready. This will simplify mtk_rng_read.
>>
>
> !!ready provided is in order to let blocking/non-blocking case could
> share same code path. And readl_poll_timeout_atomic only handles
> blocking case.

Missed this point. Makes sense. My previous comment about return value
in mtk_rng_read is invalid as I based it on a wrong assumption.

>
>> > +static int mtk_rng_read(struct hwrng *rng, void *buf, size_t max, bool wait)
>> > +{
>> > +       struct mtk_rng *priv = to_mtk_rng(rng);
>> > +       int retval = 0;
>> > +
>> > +       while (max >= sizeof(u32)) {
>> > +               if (!mtk_rng_wait_ready(rng, wait))
>> > +                       break;
>> > +
>> > +               *(u32 *)buf = readl(priv->base + RNG_DATA);
>> > +               retval += sizeof(u32);
>> > +               buf += sizeof(u32);
>> > +               max -= sizeof(u32);
>> > +       }
>> > +
>> > +       if (unlikely(wait && max))
>> > +               dev_warn(priv->dev, "timeout might be not properly set\n");
>>
>> Is this really necessary? Better to choose proper timeout than
>> providing this warning message. In rare cases if the timeout could
>> occur due to some reason (may be a hardware fault) print appropriate
>> warning message.
>
> It is good, I will choose the proper timeout and remove the log in the
> next one.
>
>>
>> > +       return retval || !wait ? retval : -EIO;
>> > +}
>>
>> Set retavl to mtk_rng_wait_ready and return retval.
>>
>
> Maybe i didn't get your points exactly. Adding some explanation about
> thoughts here.
>
> "return retval || !wait ? retval : -EIO;" I use can also help handling
> the both cases in one line which i think is elegant enough.
>
> And retval is accumulated with each round if some data's existing in
> hardware, so we don't return the value from mtk_rng_wait_ready().

retval can be 0 only when mkt_rng_wait_ready fails, returning 0 when
wait is true is confusing. Expected return value when 0 bytes is read
from device and wait is true is not clearly documented.

"return retval || !wait ? retval : -EIO;" is also fine.

Overall the code looks good to me. You can add:
Reviewed-by: PrasannaKumar Muralidharan <prasannatsmkumar@...il.com>.

Regards,
PrasannaKumar