From: gert@greenie.muc.de (Gert Doering) Subject: Re: linux/mgetty again... Date: Wed, 15 Sep 1993 12:43:00 GMT
lan_kova@rcvie.co.at (Tom Kovar) writes:
> I am using the ACEEX 1496 faxmodem under Linux .99.8 and mgetty/sendfax
>0.13. After enforcing the hardware flow control (mapping the ioctl calls to
>tc{get,set}attr), sending faxes is working perfectly. The only problem is
Fine.
>faxrec. The received faxes are legible (somehow), but many lines are corrupted;
>even g3topbm reports a problem of premature EOF, so I had to hack it in order
>to get something at least. I fear the problem might still be the flow control,
>but dunno...
Uuuuh. Sounds like you're losing a *lot* data.
> Another surprising fact is that sending a fax at 14400 bps sendfax consumes
>only about 10% CPU (80386, 33 MHz), whereas receiving at 9600 mgetty consumes
>something in the region 70-95 % CPU !!
Well, are you sure that mgetty 0.13 is running? In older versions, this
was common (well... that's what you get if you use unbuffered one-byte
read()s for ease of implementing), but in 0.13 the fax reception stuff
has been changed to use buffered read()s.
With the old unbuffered reads, it was very likely that data is lost, but
with 0.13, I have tried receiving a five-page fax while the system load
was at 5 (!) - and no lost any data at all (on a 386/40).
Does your port use a 16550? Which port speed do you use for fax
reception?
>I seem to have noticed that even during
>the fax receive, OPOST and OCRNL are active; is it quite OK?
That will not do any harm, since only the direction *to* the modem is
postprocessed by the serial driver.
One problem *may* be that you use a fairly old linux version, I think in
the serial driver had been a lot changes since then, improving reliability
at high speeds.
gert
-- ... ich kann diese signature nicht mehr sehen!Gert Doering - Munich, Germany gert@greenie.muc.de fax: +49-89-3243328 gert.doering@physik.tu-muenchen.de