On Wed, Jun 13, 2001 at 01:23:14PM -0400, Pete Zaitcev wrote:
> > From: Pavel Machek <[EMAIL PROTECTED]>
> > To: [EMAIL PROTECTED]
> > Date: Wed, 13 Jun 2001 00:42:16 +0200
>
> > > https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=39238
> > >
> > > I think I'll just take that flag off from the read URB for us.
> >
> > ...and you get 5x slowdown on PCI bandwidth when using modem...
>
> Users prefer 5x slowdown to not working modem, evidently.
Uh? The modem should work just fine with the NO_FSBR flag, just with its
speed limited to about 64 kbytes/second. (One 64-byte packet per USB
frame). And 64 kbytes/second is quite a lot for any modem out there,
even ISDN can't do more than 16 kbytes/second uncompressed.
However without the flag, the way the ACM specification was designed,
there needs to be a pending bulk read URB. And those happen to hog the
PCI bus completely with UHCI and FSBR, causing huge slowdown of the
whole computer all the time the serial device is open.
NO_FSBR is correct in acm.c
There may be different problem, though. I do have the 3Com ISDN TA, so I
can test it. I don't have an ISDN line, though. I'll try to enable
internal loopback in the TA.
> People all over the world keep buying crap, such as winmodems.
Their problem.
> Arguing with them postfactum is to piss against the wind.
> And Pavel, you were a staunch supporter of winmodems, and
> you are against a 5x slowdown now - how comes? Stop weathervaning.
Pavel was trying to get them working. That's different from saying they
are a good thing.
--
Vojtech Pavlik
SuSE Labs
_______________________________________________
[EMAIL PROTECTED]
To unsubscribe, use the last form field at:
http://lists.sourceforge.net/lists/listinfo/linux-usb-devel
