This was part of a debate about DPRINTF a while ago.  There is code in
src/mem/slicc/ast/FuncCallExprAST.py that hard codes DPRINTF in slicc
to be RubySlicc.  I argued against the hard coding of things and it
seems that we partially avoided that (since the DPRINTF statements use
RubySlicc instead of that being implicit).

Having just looked at the code, it inserts line numbers.  We could
easily improve DPRINTF to pass in __FILE__ and __LINE__ implicitly and
add a tracing option to turn on the location stuff.  If you do that,
the code should output #file and #line statements throughout.

  Nate

> I noticed that no matter what debug flag you provide in a SLICC definition,
> you are only able to view that output by using the "RubySlicc" flag.
>
> Example:
> You place this statement in "myruby.sm":
> DPRINTF(RubyMemory, "Testing").
>
> But when you test, it's only available through:
> /m5.opt --debug-flags="RubySlicc" ..
>
> instead of:
> /m5.opt --debug-flags="RubyMemory" ...
>
> This seems like a relic of the initial port of m5->ruby, but I think
> eventually you should be able to place arbitrary flags in the SLICC code.
> Has anyone else seen this? Or better yet, does anyone with knowledge of the
> SLICC parser have a suggested fix?
>
> Although, I can't dedicate the time toward SLICC features right now, I
> thought it'd be good to at least bring this up to the dev-list and see what
> people think.
>
> Thoughts?
>
> --
> - Korey
> _______________________________________________
> gem5-dev mailing list
> [email protected]
> http://m5sim.org/mailman/listinfo/gem5-dev
>
_______________________________________________
gem5-dev mailing list
[email protected]
http://m5sim.org/mailman/listinfo/gem5-dev

Reply via email to