[LLVMdev] Enable debug for MSP430

[Eric Christopher]

On Wed, Sep 3, 2014 at 12:57 PM, Rafael Espíndola <
rafael.espindola at gmail.com> wrote:

> On 3 September 2014 10:32, Iain Sandoe <iain at codesourcery.com> wrote:
> > Hi Gents,
> >
> > For those of us with out-of-tree backends which are not 32bit, the
> msp430 backend is a useful vehicle for examining changes and testing out
> ideas.
> >
> > So I was wondering about enabling debug output on the MSP430 backend so
> that I can illustrate a few issues to Adrian and you on the variable pieces
> side.  (there doesn't appear to be any specific person claiming the msp430
> code right now).
> >
> > I have a question tho.
> >
> > the following patch :
> >
> > diff --git a/llvm/lib/Target/MSP430/MCTargetDesc/MSP430MCAsmInfo.cpp
> b/llvm/lib/Target/MSP430/MCTargetDesc/MSP430MCAsmInfo.cpp
> > index df1aa1a..844db93 100644
> > --- a/llvm/lib/Target/MSP430/MCTargetDesc/MSP430MCAsmInfo.cpp
> > +++ b/llvm/lib/Target/MSP430/MCTargetDesc/MSP430MCAsmInfo.cpp
> > @@ -24,4 +24,6 @@ MSP430MCAsmInfo::MSP430MCAsmInfo(StringRef TT) {
> >
> >    AlignmentIsInBytes = false;
> >    UsesELFSectionDirectiveForBSS = true;
> > +
> > +  SupportsDebugInformation = true;
> >  }
> >
> >
> > does, indeed enable debug - but NOT .debug_frame (and IIRC some of the
> variable range info goes missing).
> >
> > it needs:
> >
> > diff --git a/llvm/lib/Target/MSP430/MCTargetDesc/MSP430MCAsmInfo.cpp
> b/llvm/lib/Target/MSP430/MCTargetDesc/MSP430MCAsmInfo.cpp
> > index df1aa1a..e7e0550 100644
> > --- a/llvm/lib/Target/MSP430/MCTargetDesc/MSP430MCAsmInfo.cpp
> > +++ b/llvm/lib/Target/MSP430/MCTargetDesc/MSP430MCAsmInfo.cpp
> > @@ -24,4 +24,7 @@ MSP430MCAsmInfo::MSP430MCAsmInfo(StringRef TT) {
> >
> >    AlignmentIsInBytes = false;
> >    UsesELFSectionDirectiveForBSS = true;
> > +
> > +  SupportsDebugInformation = true;
> > +  ExceptionsType = ExceptionHandling::DwarfCFI;
> >  }
> >
> > to make the debug_frame stuff happen.
> >
> > Is that intentional?
> > (it's not a big deal, just a bit non-intuitive for backends which have
> no exception support).
>
> I think that is the case. ExceptionsType is probably just a bad name
> for the frame unwind format.
>

Yes. This is the case.

-eric
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20140903/3ec84f35/attachment.html>