[Eda-dev] 'bad switch yylook' error when running Graywolf in QFlow tutorial

Tim Edwards tim at opencircuitdesign.com
Tue Apr 17 17:09:52 EDT 2018


Hello Jamie,

 > I have just installed QFlow and its components from source and have run
 > into a problem when following the tutorial
 > (http://opencircuitdesign.com/qflow/).
 >
 > The 'synthesise' stage ran fine, but the ‘place’ stage failed with a
 > parsing error:
 >
 > $ qflow synthesize place route map9v3
 >
 > ...
 > Running blif2cel to generate input files for graywolf
 > blif2cel.tcl --blif /home/jamie/synth/tutorial/map9v3.blif --lef
 > /home/jamie/synth/qflow-1.1/../share/qflow/tech/osu035\
 > /osu035_stdcells.lef --cel /home/jamie/synth/tutorial/map9v3.cel
 > Preparing pin placement hints from map9v3.cel2
 > Running GrayWolf placement
 > graywolf  map9v3
 > graywolf failed with exit status 255
 > Premature exit.
 >
 > From place.log:
 >
 > ...
 > TimberWolfSC switches:
 >  Graphics mode on
 >  TimberWolfSC will inherit window
 >
 > TimberWolfSC version:v6.0 date:Mon May 25 21:19:07 EDT 1992
 > Row-Based Placement and Global Routing Program
 > Authors: Carl Sechen, Kai-Win Lee, and Bill Swartz,
 >         Yale University
 > bad switch yylook 6754668bad switch yylook 6754748bad switch yylook
 > ...
 > TimberWolfSC terminated abnormally with 1 error[s] and 0 warning[s]
 >
 > This looks like it might be a bug with Graywolf, but I would be
 > interested if anyone has any suggestions if there may be another way
 > to resolve it.

Ah, you picked just the wrong time to download graywolf!  I just found
out about this error this week, and only figured out that it was due to
graywolf today.  I debugged it earlier today and whipped off a message
to Ruben Undheim, who rolled back the git version of graywolf on github
to remove the unfortunate error that crept in with the last set of
commits, which was from about three weeks ago.

Specifically, there are two routines that got recast from having no
type (one of those ways plain C allows programmers to be sloppy) to
type "void", but for a routine that was depending on the default type
of "int" and returning a value.  If you or anyone is interested in the
details, then I can give you a specific patch for it.  However, since
I'm not certain that those are the only such errors, it's probably better
just to pull the graywolf corrections from github and recompile.

(https://github.com/rubund/graywolf)

Sorry about the mixup, and thanks for the bug report.

					Regards,
					Tim

+--------------------------------+-------------------------------------+
| R. Timothy Edwards (Tim)       | email: tim at opencircuitdesign.com    |
| Open Circuit Design            | web:   http://opencircuitdesign.com |
| 19601 Jerusalem Road           | phone: (240) 489-3255               |
| Poolesville, MD 20837          | cell:  (408) 828-8212               |
+--------------------------------+-------------------------------------+



More information about the Eda-dev mailing list