Code review comment for lp:~martin-lp/hipl/n900-build-fix

Hi,

On Mon, Oct 10, 2011 at 3:52 PM, Christof Mroz <email address hidden> wrote:
> On Mon, 10 Oct 2011 14:35:06 +0200, Diego Biurrun <email address hidden> wrote:
>
>>>>> + tcase_set_timeout(tc_core, 120);
>>>>
>>>> I wonder if we should not set this in a slightly more global place.
>>>
>>> Any more thoughts on this? I've fixed the cosmetics and it would be
>>> nice to get
>>> this fix into trunk.
>>
>> I tend to suspect that we may run into this problem in other places in
>> the future, so I'm inclined to set this (more) globally, but I don't
>> much mind either way.
>
> As a compromise, we could do that the next time we encounter a timeout on
> the n900.

Think so too.

> But with a lower timeout, since 120 seconds are excessive even
> on a phone for most operations. In this case (if I see it correctly),
> multiple tests were bundled into one which stretched the runtime even
> further.

It's just the key generation that takes so long, the rest is more or less
done instantly. The 1024 bit key takes a moment, 2048 bit takes a while
and 3072 bit already feels like ages on the N900 (long enough that you start
to wonder if it crapped out on you ;)). Not sure if it's the limited CPU (600MHz?)
or the OpenSSL implementation though, maybe a bit of both.