Bug 65522 - [5 Regression] Svn revision 221590 fails bootstrap - ../libiberty/libiberty.a(cplus-dem.o): In function `ada_demangle': cplus-dem.c:(.text+0xdb8): multiple definition of `ada_demangle' ada/adadecode.o:adadecode.c:(.text+0x863): first defined here
Summary: [5 Regression] Svn revision 221590 fails bootstrap - ../libiberty/libiberty.a...
Status: RESOLVED FIXED
Alias: None
Product: gcc
Classification: Unclassified
Component: bootstrap (show other bugs)
Version: 5.0
: P1 normal
Target Milestone: 5.0
Assignee: Jakub Jelinek
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2015-03-23 10:31 UTC by David Kredba
Modified: 2015-03-23 15:50 UTC (History)
5 users (show)

See Also:
Host:
Target:
Build:
Known to work:
Known to fail:
Last reconfirmed: 2015-03-23 00:00:00

Attachments
gcc5-pr65522.patch (995 bytes, patch)
2015-03-23 12:57 UTC, Jakub Jelinek 		Details | Diff
gcc5-pr65522.patch (647 bytes, text/plain)
2015-03-23 14:30 UTC, Jakub Jelinek 		Details
Show Obsolete (1) View All Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description David Kredba 2015-03-23 10:31:22 UTC 
../libiberty/libiberty.a(cplus-dem.o): In function `ada_demangle':
cplus-dem.c:(.text+0xdb8): multiple definition of `ada_demangle'
ada/adadecode.o:adadecode.c:(.text+0x863): first defined here
collect2: error: ld returned 1 exit status
/var/tmp/portage/sys-devel/gcc-4.10.0_pre20150322/work/gcc-4.10.0-20150322/gcc/ada/gcc-interface/Make-lang.in:600: návod pro cíl „gnat1“ selhal

Revision 221558 was OK.
Comment 1 Richard Biener 2015-03-23 11:56:41 UTC 
Confirmed by Andreas.
Comment 2 Jakub Jelinek 2015-03-23 12:57:16 UTC 
Created attachment 35109 [details]
gcc5-pr65522.patch

Seems the ada/adadecode.[ch] (ada_demangle) function is totally unused.
Comment 3 Jakub Jelinek 2015-03-23 13:13:07 UTC 
Ah, except ada_demangle is also exported from libgnat*.so - wonder why the RTS is built with IN_GCC.
In that case, guess the Ada folks need to decide which one of the two to keep, and/or if one of them (either adadecode.[ch], or {include,libiberty}/* one shouldn't be renamed to something else.
I haven't found any uses of either of those, except the libiberty demangler that calls ada_demangle but it could very well call a static routine instead.
Comment 4 Eric Botcazou 2015-03-23 13:36:50 UTC 
> Ah, except ada_demangle is also exported from libgnat*.so - wonder why the
> RTS is built with IN_GCC.

This is explained in gcc-interface/Makefile.in.

> In that case, guess the Ada folks need to decide which one of the two to
> keep, and/or if one of them (either adadecode.[ch], or {include,libiberty}/*
> one shouldn't be renamed to something else.

I think your patch is fine, but I'll let Arno approve it.
Comment 5 Jakub Jelinek 2015-03-23 14:30:48 UTC 
Created attachment 35112 [details]
gcc5-pr65522.patch

I'm actually testing a slightly adjusted variant that keeps the compatibility ada_demangle in libgnat* for ABI reasons, but not in gnat1 which doesn't use it and thus can link against libiberty.
Comment 6 Jakub Jelinek 2015-03-23 15:49:34 UTC 
Author: jakub
Date: Mon Mar 23 15:49:02 2015
New Revision: 221599

URL: https://gcc.gnu.org/viewcvs?rev=221599&root=gcc&view=rev
Log:
	PR bootstrap/65522
	* ipa-devirt.c: Remove duplicate demangle.h include.

	* adadecode.c (ada_demangle): Guard with IN_RTS instead of IN_GCC.

Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/ada/ChangeLog
    trunk/gcc/ada/adadecode.c
    trunk/gcc/ipa-devirt.c
Comment 7 Jakub Jelinek 2015-03-23 15:50:07 UTC 
Fixed.