This is the mail archive of the
mailing list for the GCC project.
Re: Another 3 debug hooks
- To: Neil Booth <neil at daikokuya dot demon dot co dot uk>
- Subject: Re: Another 3 debug hooks
- From: David Edelsohn <dje at watson dot ibm dot com>
- Date: Thu, 26 Jul 2001 01:35:55 -0400
- cc: gcc-patches at gcc dot gnu dot org, Mark Mitchell <mark at codesourcery dot com>
Even after I work around the declarations of
extern char *xcoff_bss_section_name;
extern char *xcoff_private_data_section_name;
extern char *xcoff_read_only_section_name;
which possibly should be in rs6000.h and defined in rs6000.c, I still end
up with dbxout_global_decl undefined at link time.
dbxout_global_decl is referenced in the gcc_debug_hooks
structureof dbxout.c for XCOFF_DEBUGGING_INFO, but the definition of the
function is inside
#if defined (DBX_DEBUGGING_INFO)
#endif /* DBX_DEBUGGING_INFO */
so it never gets defined for XCOFF. I suspect that dbxout_global_decl()
should be outside the #ifdef because xcoffout.c also references
dbxout_symbol(), but I am not sure.
I am very disturbed that you are making significant changes
affecting xcoffout, but you do not appear to be bootstrapping or testing
on AIX. It is possible to build a powerpc-ibm-aix4.3 target as a
cross-compiler to verify correctness at this level. You should be much
more careful when making these types of changes.
All of these problems are before we even get to the point of
testing whether the debug information still works on AIX. Your glib
statements when the patches were submitted about removing other XCOFF
references do not give me a lot of confidence. We need to verify that
file-scope global variable debugging information is emitted at the proper
point so that the assembler generates a correct stab entry.
If this does not get tested and fixed quickly, all of these
patches will have to be reverted.