--3eH4Qcq5fItR5cpy
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

On Wed, Dec 08, 2004 at 04:58:54PM +0200, Brad du Plessis wrote:
> > a) make sure the TAs create and obey cts/rts
> >
> > b) make sure your machine's serial ports are used with crt/rts (your
> >    variant of ppp will have to be configured for this; for "normal" pppd
> >    it's the "crtscts" keyword, I think)
>=20
> The TA's were configured to use cts/rts flow control and I'd also configu=
red
> pppd to use cts/rts. Is the fact the the TA's are USB likely to be a prob=
lem
> in this regard? Surely the USB flow control works in exactly the same way=
 as
> serial?
>=20
> > 1. serial handshaking problem, resulting in bytes lost resulting in ppp
> >    packet  loss
>=20
> Is there any test that I can perform to explicitly confirm that serial
> handshaking is the problem?

Ah, simulated serial. Well, you can't easily look at the simulated
handshaking lines, so I don't know how to check...

But this is likely your problem when you're doing highspeed (for a
serial) transfers with lots of handshaking events. At least, somebody
explained it a few days ago on tech-hm... was it kern?, like this:

all handshaking events on the simulated serial need an USB packet to
be exchanged; so the latency for handshaking events is much higher than
on a real RS232 port.

Regards,
	-is

--3eH4Qcq5fItR5cpy
Content-Type: application/pgp-signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: 2.6.i

iQEVAgUBQbcbCzCn4om+4LhpAQHslwf+Ka031Pwig1B0+Seeqs2hyr13wFif5Z1w
3Z/bOH5StmXN3NkuQ3Mbt/wELlmz5kxDypfDVEQvEHunJut9mS+I3OmxDcXqPSU+
UquhyJzRgv/VVq7QOtW6WH6wbmE0UeVSVPEByxuDW8sx0TBT5RmfMMGFZ/zBsdBr
lfOr3euv6mqTdJg8ndB2Q/iRJBXc1J9bcYdh/XcsQkn/ppunDVCDDhaBFqi7tZNr
wn9g3q41f0wcI5VMOgWtWIR4vpqhIwp4IC+YMIvTc2f//eGG3EINxSiURtFTPxAf
dD2uIRfqVh2smYUA1xUZvWZbGi480pa/v97QPftBB+j8bLb2GndWCw==
=bzvM
-----END PGP SIGNATURE-----

--3eH4Qcq5fItR5cpy--