Re: USB v1 and v2 compatibility -- OHCI v. EHCI ...

["Bryan J. Smith" <b.j.smith@ieee.org>]

On Sun, 2004-10-31 at 15:18, Robert Citek wrote:
> Unfortunately, my experience is that Linux has better support for USB2 
> than FireWire, but then we're still using the 2.4 kernel on our 
> servers. Anyone care to share experiences with the 2.6 kernel?

Red Hat kernels after 2.4.20 seem to work well with IEEE1394.
Lots of backported support.
I was patching my own IEEE1394 support in since 2.4.1.
IEEE1394 works flawlessly when you have the support.

IEEE1394 is very, very standardized, unlike USB.
Most standards have been retrofitted for USB, and many end-user devices
still lack them.
Don't get me started on how USB was designed.
There were far superior "universal" serial busses out there.

I think Intel really screwed it all up when it pulled out of support for
FireWire.  Anyone remember Device Bay?  Damn that was going to simplify
everything -- USB (character) and FireWire (block) devices in one,
external chassis.

On Sun, 2004-10-31 at 15:27, Casey Boone wrote: 
> i havent had any trouble accessing drives with 1394 under linux except
> on my laptop, but i think that either the 1394 port itself is a little
> touchy or the controller is.

IEEE1394 has a very intelligent host controller.
USB is ultra-dumb, CPU drives it completely.
USB was designed this way so it was easy for Intel and Microsoft.
That's why end-devices lagged 3+ years.
It's also why more than a few USB devices on a bus conflict.
Because the "brains" are in the end-device software driver.

> i have the same issues under windows with it.  i recently got a 1394
> hub on clearance for 5 bucks, i am going to attempt hooking up 2
> computers to a hard drive and play with one of the shared media
> file systems (the one redhat bought, i dont remember its name offhand).

Sistina Global File System (GFS).

> i dont know if this will work or not, as i know under windows only 1 pc
> can have the drive mounted, the other pc wont even attempt to use it.

It's more than just the connection and OS support, you have _lots_ of
coherency issues.  Now IEEE1394 _is_ capable as such, but typically
most of the _end_devices_ are not capable.

That's the problem.

> we shall see.

Silent corruption is never a good thing.
Especially since you don't know about it.

> i also want to attempt to network the machines over 1394 and stress
> test it to see if it breaks.

Now that should work.
IEEE1394 shouldn't have any host-to-host transfer issues.
It's when you have two hosts, one end-device, that's when you get trouble.


-- 
Bryan J. Smith                                  b.j.smith@ieee.org 
------------------------------------------------------------------ 
"Communities don't have rights. Only individuals in the community
 have rights. ... That idea of community rights is firmly rooted
 in the 'Communist Manifesto.'" -- Michael Badnarik



-
To unsubscribe, send email to majordomo@silug.org with
"unsubscribe silug-discuss" in the body.