[RFA] Fix Darwin failures on macscp.exp

[RFA] Fix Darwin failures on macscp.exp

[Tristan Gingold]

Hi,

On Darwin the debugging info are never in the final executable.  They  
either stay in object files or
a separate debug file is created.
When an executable is directly created from source files (ie gcc -g -o  
x x.c), the separate debug file
is created (as the object files are temporary).  The Darwin tool that  
creates the separate debug file,
dsymutil, doesn't put .dwarf_macinfo into it.
As a consequence, there are a lot of failures in macscp.exp (more than  
100).

Instead of disabling this test on Darwin, I'd prefer to do a two steps  
compilation: first create an object
file and then link.

Tristan.

gdb/testsuite:
2008-12-03  Tristan Gingold  <gingold@adacore.com>

	* gdb.base/macscp.exp: Generate an object file during compilation
	to work around Darwin dsymutil limitations.

RCS file: /cvs/src/src/gdb/testsuite/gdb.base/macscp.exp,v
retrieving revision 1.15
diff -c -r1.15 macscp.exp
*** testsuite/gdb.base/macscp.exp	9 Oct 2008 16:49:47 -0000	1.15
--- testsuite/gdb.base/macscp.exp	3 Dec 2008 13:34:21 -0000
***************
*** 23,28 ****
--- 23,29 ----

   set srcfile macscp1.c
   set testfile "macscp"
+ set objfile "macscp.o"
   set binfile ${objdir}/${subdir}/${testfile}

   set options { debug }
***************
*** 32,38 ****
       lappend options additional_flags=-g3
   }

! if  {[gdb_compile "${srcdir}/${subdir}/macscp1.c" "${binfile}"  
executable $options] != "" } {
       untested macscp.exp
       return -1
   }
--- 33,41 ----
       lappend options additional_flags=-g3
   }

! if  {[gdb_compile "${srcdir}/${subdir}/macscp1.c" "${objfile}" \
! 	  object $options] != ""
!      || [gdb_compile "${objfile}" "${binfile}" executable $options] ! 
= "" } {
       untested macscp.exp
       return -1
   }

Re: [RFA] Fix Darwin failures on macscp.exp

[Daniel Jacobowitz]

On Wed, Dec 03, 2008 at 02:39:16PM +0100, Tristan Gingold wrote:
> Instead of disabling this test on Darwin, I'd prefer to do a two steps  
> compilation: first create an object
> file and then link.

This is OK, but only with a comment; otherwise someone will clean it
up again.

-- 
Daniel Jacobowitz
CodeSourcery

Re: [RFA] Fix Darwin failures on macscp.exp

[Tristan Gingold]

On Dec 4, 2008, at 2:16 PM, Daniel Jacobowitz wrote:

> On Wed, Dec 03, 2008 at 02:39:16PM +0100, Tristan Gingold wrote:
>> Instead of disabling this test on Darwin, I'd prefer to do a two  
>> steps
>> compilation: first create an object
>> file and then link.
>
> This is OK, but only with a comment; otherwise someone will clean it
> up again.

I added a comment and committed it.
Thanks,

Tristan.