demo stuff

[Dave Taht]

On Sat, Mar 17, 2012 at 2:05 PM, Dave Taht <dave.taht at gmail.com> wrote:
> On Sat, Mar 17, 2012 at 1:58 PM, Francis Dupont <fdupont at isc.org> wrote:
>>> assuming it all compiles and actually works on mips.
>>
>> => I have to dig in Francisco's file to get the picture of the demo
>> but for a CPE the software is proved to work on MIPS/ARM/etc,
>
> "a version of" the software is proved to work
>
>> for the AFTR I resisted to put some assembly to speedup IP checksum.
>> So you should get no issue from the processor itself.

An example of a problem that we ran into was that packets can come in
misaligned, and the default code generated spent much time in
unaligned traps.

While this is fixed in the ipv4 path, now, it isn't in the ipv6 path. It also
crops up in the ipsec path, so far as I know, and isn't fixed in my build...

http://www.bufferbloat.net/issues/195

While the unaligned code does indeed work, it's definately a lot slower.

Slow enough to affect the demo? no. on my todo list to profile and fix? yes.
Sometime after all the crash and security bugs are resolved.

For further elucidation on the problems working on embedded hardware
can induce vs working on easy stuff like the mainline architectures,
please view

http://www.bufferbloat.net/projects/cerowrt/issues

To see the closed bugs, turn off the filter and search again. This
bug list reflects a year's worth of effort by over a dozen core developers
in addition to the work that goes on in the mainline kernel and packages.

I would hope that viewing that would start to induce some of the same
kind of pessimism I have in the general case and a better feel for what
it takes to build a solid embedded OS.



-- 
Dave Täht
SKYPE: davetaht
US Tel: 1-239-829-5608
http://www.bufferbloat.net