On Sun, Jan 26, 2003 at 07:37:42PM -0800, David Brownell wrote: > Have you ruled out actually receiving a babble error from the > hardware? Like by switching cables or ports? Or best by seeing > on a CATC that all is fine. Strictly speaking, I don't think any > of the HCDs (can) handle that correctly ... the USB spec says > something about disabling hub ports when those are seen.
I'm pretty sure it isn't the hardware. The same configuration works under
2.4.x (and I mean the same -- don't touch a cable, just reboot into a
different kernel).
> Or as if for some reason something snuck a zero-byte read into the
> queue ... if you suspect that's the case, apply some debug printks.
> (Buffer is zero bytes, device provides a full packet ... tadaa,
> that's an overflow situation.)
That's an interesting possibility. Hrm... I guess the best place to add
that would be to the sg transfer code?
Matt
--
Matthew Dharm Home: [EMAIL PROTECTED]
Maintainer, Linux USB Mass Storage Driver
Why am I talking to a toilet brush?
-- CEO
User Friendly, 4/30/1998
msg11106/pgp00000.pgp
Description: PGP signature
