Re: [Mj-tools-devel] Re: mj-xfer crashing my Visor Platinum!

[Frederic T. <te...@my...>]

Hello Arnim!

>
> Until now I was guessing that Bob's problems are caused by his Alpha
> platform. Unfortunately there is now another one reporting the same
> problem. Maybe it is really related to the UHCI driver.
>

You mean Bob run it on an Alpha processor systems? Do you know if they are 
other people are using the minijam tools successfully beside you? It might be 
a good thing to to see if the other people that have it working are on OHCI 
or UHCI.

>
> Frederic, you wrote that you use SuSE Linux 7.1 - which kernel do you
> use? If it is 2.4.2 or lower, did you apply the patch to visor.c as
> described in the mj-xfer source? The patch is not necessery for 2.4.3
> (and future ones).

At the moment i am on the 2.2.18 kernel! I would like to be running 2.4.0 
(which is also install with SuSE 7.1, but sadly by SoundBlaster crappy 
Vibra16 failed under that!  

Nope i saw that last night while reading your source files ;-), ill first 
need to install the 2.2.18 kernel source.  I try to apply the patch on the 
2.4.0 kernel (the source for it were install), but i am a bit a newbie with 
those kernel recompiling stuff 8-(.  How can i recompile just the visor 
kernel modules? I dont really wanna recompile the whole kernel.

> If this does not help, and one of you would like to experiment, here is
> a proposal for another patch. Looks as if you both are stuck with the
> problem that MJLoader does not accept packetized data from the host. So
> some experimenting could clarify things.

Ill try to apply this patch first and let you know when ill be running the 
new revised visor modules.  Well if i remember they are some issue with 
packet stuff with the UHCI compare to the OHCI (they were some explanation on 
usb.org forum about it).

> I looked through usb-uhci.c and there is some code that I cannot
> correlate to usb-ohci.c. In function uhci_submit_bulk_urb() there is a
> mechanism to split the transferred data into smaller packets. The code
> in question is in lines 892/893 for kernel 2.4.3:
>
>   if (pktsze > maxsze)
>     pktsze = maxsze;
>
> Maybe if you just comment out these two lines and recompile/reload the
> module. Perhaps this will temporarily work around the problem, but be
> careful. If I find me an UHCI card/mobo I could try and do some debug
> work myself :-(
>

Ill try this later if the first solution doesnt improve anything.

>   Good luck and backup well!
>

Thanks again for your collaboration, hopefully by helping each other we 
should be able to get ride of this annoying bugs (yuk yuk today i have to use 
Windows2000 to upload some new mp3 on my innogear!)

>     Arnim
>
>
>
> P.S: The documentation in the tar.gz distribution is quite short. You
> could have a look at the README in the CVS tree which is not yet
> distributed.

This is what i did ;-), since the readme in your dir was only saying "use the 
force Luke" hehehe ;-).

Fred