Re: recent MIPS mdebug-related changes cause gcc testsuite failures.

[Daniel Jacobowitz <drow at mvista dot com>]

On Fri, Jun 21, 2002 at 11:18:26AM -0700, cgd@broadcom.com wrote:
> FYI:
> 
> It looks like the recent MIPS mdebug-related changes cause gcc testuite
> (specifically, g++.dg/debug) failures.  E.g., see the g++ test .log
> file output below.

Oy.

> It looks like -gcoff1 doesn't produce a message which matches:
> 
>         if { ! [string match "*: unknown or unsupported -g option*" \
>                 $comp_output] } {
> 
> so the tests try to run that combination of options, and fail.
> 
> 
> Not really sure what the right thing to do is.  maybe turn on mdebug
> if people ask of -gcoff?  hmm, looks like in the current compiler,
> subtarget_asm_debugging_spec is:

What happened before?  Did the test show as unsupported, or did .def
etc. actually get handled?  Similarly, if you run the test with
-Wa,-mdebug does it succeed?

Never mind, I have a build tree here... yes, adding -mdebug is the
correct thing.




> 
> *subtarget_asm_debugging_spec: {g} %{g0} %{g1} %{g2} %{g3} %{ggdb:-g}
> %%{ggdb0:-g0} %{ggdb1:-g1} %{ggdb2:-g2} %{ggdb3:-g3} %{gstabs:-g}
> %%{gstabs0:-g0} %{gstabs1:-g1} %{gstabs2:-g2} %{gstabs3:-g3}
> %%{gstabs+:-g} %{gstabs+0:-g0} %{gstabs+1:-g1} %{gstabs+2:-g2}
> %%{gstabs+3:-g3} %{gcoff:-g} %{gcoff0:-g0} %{gcoff1:-g1}
> %%{gcoff2:-g2} %{gcoff3:-g3}
> 
> Doesn't look like that flag will be passed unmolested to the
> assembler.  "-mdebug" would have to be passed in via the specs, it
> seems...
> 
> 
> Thoughts?

That spec is just wrong, period.  According to standard practice, if we
have compiler debugging we do not desire assembler debugging; -g* (*
not equal to 0, at least) should cause -g0 to be passed to the
assembler.

-gcoff{,1,2,3} should cause -mdebug to be passed to the assembler.

Would you mind submitting the appropriate GCC patch?


-- 
Daniel Jacobowitz                           Carnegie Mellon University
MontaVista Software                         Debian GNU/Linux Developer