Re: [MiNT] Network and USB problems with NetUSBee

[Mariusz Buras <mariusz.buras@gmail.com>]

I have similar issues. When I download stuff with AFTP it's usually fine, but with wget it almost always breaks the connection and then resumes it. Wget seems to be working much faster (no GEM window to update?) so that would be in line with the theory that faster transfer rates are causing troubles.

I know that Netusbee driver is hooking itself to 200Hz interrupt to pull for packets. For 10Mbit ethernet we would be getting about ~830 packets per seconds.  Maybe doubling (or x4) the timer interval would help ? 

Greetings/Pozdrawiam,
M.Buras

On 4 August 2011 11:48, Lars Schmidbauer <latzmaster@gmx.de> wrote:

Hi,

exept m0n0, nobody replied to my posting in german Atari forum yet (just a poll
if there's anybody who can confirm this problem). :-(
But it's impossible, i cannot be alone with this, happens here reproducable on two
Falcons (with slightly different hardware) but nowhere else in this world??????
Anyway, we sorted out some things:
- no problem with MiNT 1-16-3 (EasyMiNT)
- problem occures with FreeMiNT >= 1-17-0, same network settings
- NOT caused by inet4.xdd and enec*.xif (EasyMiNT's inet4.xdd, version-check removed from Alan)
- when writing huge trace_logs with curl, DL works but slower than with 1-16-3
Sure this is a workaround for downloads (thanks to m0n0), but other apllications (mail, netradio,...)
are still not working correct.
How can we find the main problem??? I don't want to be stucked to 1-16-3 for proper network usage...
Exept this, 1-17-0 (1-18-CUR) are sooo much better :-)

Regards,
Lars

On Wed, 03 Aug 2011 21:50:16 , "Helmut Karlowski" <helmut.karlowski@ish.de> wrote:
> Am 03.08.2011, 16:01 Uhr, schrieb m0n0 <ole@monochrom.net>:
>
> >  Does anyone have an idea how this tracing can affect the network/socket
> >  driver? Maybe someones alarm-bell rings... :)
>
> My first guess was that disk-IO blocks the transfer and the connection
> gets confused, but I do not know much about MiNT-net.
>