RFC: stack trace generation

[David Daney]

Andrew Haley wrote:
> David Daney writes:
>  > Arn't there two issues?
>  > 
>  > 1) Generate the raw stack trace.
>  > 
>  > 2) Make a pretty printout.
>  > 
>  > These patches deal with #2.
>  > 
>  > #1 is easily done where ever DWARF 2 unwinding is available (shameless
>  > plug: libjava/sysdep/dwarf2-backtrace.cc).
> 
> #1 is no longer a problem because we have functions in libgcc to do
> it, and these functions work with any unwinder format, not just DWARF
> 2.
> 

Yes.  The dwarf2-backtrace.cc is probably mis-named as it uses the
functions that I believe you are referring to.

Being a very lazy person, I am only expending GCJ effort on the
platforms that I am currently paid to support (mipsel-linux*).  But I
have to wonder if it would not be a win to use dwarf2-backtrace.cc with
the i386-linux-gnu target and enable -fomit-frame-pointer.  From reading
some of the posts to this list, it seems that having the extra register
would be a good thing.  It is a small amount of configure hackery to see
if it would work, but something I don't have on my to-do list.

David Daney