Quote:

Humm, I can't see any need to write the driver is 68k asm if it's what you mean.

A clean C implementation could have been recompiled easily for MorphOS. (And maybe Amithlon x86 ???).
Native PowerPC C is faster than emulated 68k asm.

In my mind, wait for a Poseidon/MorphOS update in order to get better USB 2.0 support.

Bye


[ Edited by NicoPPC on 2005/5/10 16:19 ]

[ Edited by NicoPPC on 2005/5/10 16:21 ]

@Schlonz

>>Use it to search for 202c 000c 0300 and replace it by 202c ffff ffff.
>>Are you sure? I could not find that string, but there was 203c 000c 0300. Did you mean this?
My fault. Of course it's 203c. My Fault. I also provide a patched version on my homepage.

Please consult the readme for any further questions. I don't garantee it's working.

http://www.geit.de/stupid/OpenUSB.lha

@NicoPPC

For USB you need massiv endian swaps and little endian linked listest, which must be set up by the CPU. It's consuming much CPU load, even in the optimized version. Regarding having a 030 CPU this is relevant.

Geit

Geit

Thx alot for the advice...


Also, I love the

blah/stupid/blah



magnetic

I coludn't get it working at all. Which chipsets does it support? I have a VIA PCI USB2.0 card (don't know it's number, but it is working with MOS in 1.1 mode).

I installed the device, added to poseidon, but it didn't go online. (It says:
openusb.png
)

What is wrong?

@Schlonz

Use Unit 0! Since your card is UHCI there is no other unit than EHCI.

Geit

Does it work now? And if yes what is the speed?

Geit,

Why does the byteswapping requirement somehow make coding in C so much difficult? If it is 68k asm for the byteswapping routines I can understand, but remember PowerPC has a set of instructions for loading and storing with an implicit byteswap - there should be some fairly simple replacements for your heavy 68k asm.

The entire driver can't be one great big byteswapping routine. What is your benefit for coding in 68k asm otherwise? Did you also think that an x86-native driver under Amithlon would have worked better too (x86 has the same kind of byteswap routines.. I don't know how Amithlon presents the data though..)

Neko

I bought a USB 2.0 Card with the Via VT6212L Chip but i can't boot into MOS when the Card is pluged in. I got MOS1.4.5 installed and Poseidon V2.0 installed.

>Why does the byteswapping requirement somehow make coding in C so much difficult?
Where did I say that?

My driver was ment to run on 68K Systems. Point.

Making it in C would cause a speed drop by around 20%, depending on the used CPU. The available USB controller are crap. They need more CPU load, than you are expecting. I would never ever use a mouse or keyboard via USB. The cpu load required to setup and parse the stuff is huge. I guess you should take a look into the ?HCI manuals for details on the stuff.

Geit

Maybe is needed a C version with PowerPC assembly optimizations for those non-68k systems?

Or if you can do it, a fully PowerPC assembly version

@Geit:

Thanks, using Unit 0 did work. Well, kind of. Poseidon went online telling me that OpenUSB.device unit 0 is an EHCI controller, but no USB devices (I tried a scanner, an USB stick and a Hub) were recognized - inserting USB devices into the USB controller card just did no result, just like no USB device were inserted.

Could you imagine if an OHCI/EHCI controller card works better or would there be no difference?

@ Schlonz and all

A bit offtopic but...

I have the problem, that an USB stick does not mount twice! I explain: I insert the stick, it mounts, I copy something do it, take it out.
Without a reset of the Peg in between, if I use the USB stick again it does not mount anymore.

Is this due to the Peg's controller?, is an external controller better?

@ms

i have the same probleme with my camera olympus c460.

We worked all night on our LAN party to find the problem.

It seems the EHCI controller is positioned missaligned into the address space, which causes the card not to be working. So currently no chance!

Geit

Hi Geit,

thanks for trying to find the problem.

Do you know if this missaligne-problem is caused by MorphOS or by OpenPCI?

>Do you know if this missaligne-problem is caused by MorphOS or by OpenPCI?
No! Regarding to the base registers the card is at the right position, but when checking the cards position by locating specific register fields it's not.

The same card within Amithlon shows the card at the right position and the registers are where they should be.

It's very strange. Thanks to Axel Knabe I now have an USB 2.0 card in the Peg, so I can do more testing.

Guido Mersmann

>Thanks to Axel Knabe I now have an USB 2.0 card
>in the Peg, so I can do more testing.

Glad to hear that.

Please continue your great work.

Geit

Alright! You are a valued dev in our community. Thank you, many appreciate your work.

magnetic

@ GK_LKA

>Good morning, it is 2005, and an average photo size on my digital camera is 2MB.
>I want to watch films with my PDA (= I have to copy lots of MB's over USB).
>Copying films and photos take much time...
>I have a 2.0 USB PCI card for over 3 months...
>I'm very tired, that these simple operations are so slow...

@ Schlonz

>Sorry, but this is not true... Guido Mersmann,
>Author of ArakAttack, told me that it runs at
>about 6 MB / sec on his Amithlon system.
>
>And Spider does provide about 3 MB/sec.
>But will the Spider work with Pegasos? I dont know...
[...]
>Yes, but right now I will only get USB 1.1 speed with that cheap card,
>and me and a lot of other people want USB 2.0 speed in the Peggy.

@ NicoPPC

>By the way, believe it or not but my brother's XP box does not support USB 2.0 high speed
>;-P

@ SoundSquare

>XP supports USB2 with SP1
>maybe ur bro doesn't even have some USB2 hardware...?


I want to remember you all (as SoundSquare said about Brother Printer) that if you connect USB1.x (old) hardware to USB 2.0 port, then that port degrades its transfer-rate capabilities to fit 1.x specifications.

This fact could explain some low-speed problems you encounterd.


Although now here follows a question risen in my mind:

"Is USB 2.0 so intelligent to degrade to USB1.x specs ONLY THAT PORT connected to "old" hardware???"


Or maybe

"ALL 2.0 system degrades itself to 1.x specs in that situation?"

If the answer is:
"it degrades only the port connected to 1.x hardware" then you all just need to connect your new 2.0 Hardware to those USB ports you leave free from 1.x hardware.

If the answer is:
"It degraded it all", then USB 2.0 is not so robust as I previously believe, and you could do nothing.

>USB 2.0 so intelligent to degrade to USB1.x specs ONLY THAT PORT connected to "old" hardware???"

On USB2.0 cards/board each USB connector is connected to an UHCI/OHCI and an EHCI by a switch. If you plug in a USB1.1 device EHCI (USB2.0) simply turns the switch and the UHCI/OHCI controller is taking over (owning) this port until the user unplugs the device again.

Guido Mersmann