[K42-discussion] Problem with thinwire3

[Kevin Cormier]

I checked out several revisions of thinwire3.c and tried them out to see 
which version my Xserve was having trouble booting with.  It turns out 
that all of them before revision 1.8 worked fine.  Here is the cvs log 
for that revision:

revision 1.8
date: 2005/07/14 00:53:37;  author: mostrows;  state: Exp;  lines: +12 -7
Fix setting of flow control bits.

The exact problem seems to be a change to the serial_read function.  The 
body of the function was changed to be enclosed by the statements:

bitSet(ic->fd, TIOCM_RTS);
bitClear(ic->fd, TIOCM_RTS);

Commenting these lines out in the latest version worked, but of course 
they must be there for a reason.  I'm not sure exactly what they do.

Is there any configuration having to do with flow control that I might 
not have set up properly?  Here's how I got the latest revision working:

Index: tools/build/shared/thinwire/thinwire-src/thinwire3.c
===================================================================
RCS file: /u/kitchawa/cvsroot/thinwire-src/thinwire3.c,v
retrieving revision 1.9
diff -r1.9 thinwire3.c
314c314
<     bitSet(ic->fd, TIOCM_RTS);
---
 >     //bitSet(ic->fd, TIOCM_RTS);
320c320
<     bitClear(ic->fd, TIOCM_RTS);
---
 >     //bitClear(ic->fd, TIOCM_RTS);


Kevin Cormier wrote:

> Hi everyone,
>
> I'm a student in the CS department at U of T, and for quite awhile I'd 
> been trying to boot K42 for the first time on an Xserve, but I was 
> running into problems with thinwire3.
>
> The problem was solved by using an older copy of thinwire3 
> (timestamped April 12th) that the ECE department here uses, so I think 
> there may be a problem with newer versions, at least for Apple G5 
> Xserves.
>
> I'm going to try to pinpoint which revision of 
> tools/build/shared/thinwire/thinwire-src/thinwire3.c causes the 
> problems, and I'll follow up when I'm able to.  I think a likely 
> candidate is the latest revision (1.9), but again I'll have to try it 
> out to be sure.  The comment reads "Reordering of some of the speed 
> re-negotiation code.  Make sure things are done in the right 
> sequence.  Add a signal handler on SIGINT to clean-up -- eventually we 
> may want to hook some other stuff off this."  Most of the other 
> changes seem minor.
>
> -Kevin
>
>
> _______________________________________________
> K42-discussion mailing list
> K42-discussion at ozlabs.org
> https://ozlabs.org/mailman/listinfo/k42-discussion