Path: utzoo!utgpu!news-server.csri.toronto.edu!clyde.concordia.ca!uunet!chinacat!chip
From: chip@chinacat.Unicom.COM (Chip Rosenthal)
Newsgroups: comp.unix.i386
Subject: Re: FAS on SCO ODT ???
Message-ID: <1326@chinacat.Unicom.COM>
Date: 16 Jun 90 19:12:51 GMT
References: <1990Jun15.020958.594@dynsim2.uucp> <646@mecky.UUCP>
Organization: Unicom Systems Development, Austin, TX
Lines: 25

>> I waded into the driver a little and found that fasinit() was calling a
>> UART test routine that in turn called delay() [...]
>> I stubbed the test routine out and the driver came up fine.

That test routine is what decides whether or not the UART has a FIFO.  I
don't think you want to chuck it entirely.

I played around with FAS a bit with SCO XENIX, which has the same limitations
of calling delay() in an xxinit() routine.  I just #define'ed delay() to
call a procedure which does a counting loop.  Since the delay is just to
"let the UART settle a bit" after configuring it and before pounding data
through it, the timing is not critical.  However, I didn't feel comfortable
chucking the delay entirely.  Although I didn't contact the author or
experiment around with it, it doesn't seem to be the sort of thing one
adds without a good reason.  The busy loop stuff is definately ugly, but
given that it runs once during initialization time, it isn't that bad a
compromise.

BTW...to forestall the obvious next question, I never fully debugged the
faswrite() routine, so I don't have a FAS for XENIX I can send out.  Sorry.

-- 
Chip Rosenthal                            |  You aren't some icon carved out
chip@chinacat.Unicom.COM                  |  of soap, sent down here to clean
Unicom Systems Development, 512-482-8260  |  up my reputation.  -John Hiatt