| Summary: | [3.3/3.4 regression] Fortran ICE in dwarf2out:add_abstract_origin_attribute with -g -O -finline-functions | ||
|---|---|---|---|
| Product: | gcc | Reporter: | jiri.pittner |
| Component: | debug | Assignee: | Richard Henderson <rth> |
| Status: | RESOLVED FIXED | ||
| Severity: | normal | CC: | gcc-bugs, janis187, reichelt, s.bosscher, steven |
| Priority: | P3 | Keywords: | ice-on-valid-code |
| Version: | 3.2 | ||
| Target Milestone: | 3.3.1 | ||
| Host: | Target: | ||
| Build: | Known to work: | ||
| Known to fail: | Last reconfirmed: | 2003-05-28 23:11:28 | |
| Attachments: |
master.f
bug.f |
||
State-Changed-From-To: open->analyzed
State-Changed-Why: Confirmed. Actually an ICE in the debug writer:
g/x> /home/bangerth/bin/gcc-3.2.2-pre/bin/g77 -c -O3 -g -w -fno-globals master.f
master.F:566: Internal compiler error in add_abstract_origin_attribute, at dwarf2out.c:9458
Please submit a full bug report,
This didn't use to ICE in 2.95 (even with -gdwarf-2), but
ICEs in 3.2.2, 3.3 and mainline, so is a regression.
W.
From: Steven Bosscher <s.bosscher@student.tudelft.nl> To: gcc-gnats@gcc.gnu.org, jiri.pittner@jh-inst.cas.cz, gcc-bugs@gcc.gnu.org, toon@moene.indiv.nluug.nl, nobody@gcc.gnu.org, gcc-prs@gcc.gnu.org Cc: Subject: Re: debug/9864: [3.2/3.3/3.4 regression] Fortran ICE in dwarf2out:add_abstract_origin_attribute with -g -O -finline-functions Date: Wed, 26 Mar 2003 10:13:19 +0100 http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=9864 I've just analyzed this PR a bit more and I've reduced the test case to managable proportions (12 lines instead of almost 500 :-). This looks like something specific to Fortran (i.e. something we'll only be able to reproduce with the Fortran front end), so I've added Toon to the "Who's interested" list. Greetz Steven From: Janis Johnson <janis187@us.ibm.com> To: gcc-gnats@gcc.gnu.org, jiri.pittner@jh-inst.cas.cz, gcc-bugs@gcc.gnu.org, toon@moene.indiv.nluug.nl, nobody@gcc.gnu.org, gcc-prs@gcc.gnu.org, aoliva@redhat.com Cc: Subject: Re: debug/9864: [3.2/3.3/3.4 regression] Fortran ICE in dwarf2out:add_abstract_origin_attribute with -g -O -finline-functions Date: Wed, 26 Mar 2003 16:34:08 -0800 The regression appeared with this patch: 2002-02-26 Alexandre Oliva <aoliva@redhat.com> * dwarf2out.c (gen_inlined_subroutine_die): If block is abstract, generate a die for the lexical block. The hunt used the small test case on i686-pc-linux-gnu with the options "-O -finline-functions -g", with mainline. http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=9864 Responsible-Changed-From-To: unassigned->aoliva Responsible-Changed-Why: You broke this, and you know more about dwarf2 than most :-) From: Volker Reichelt <reichelt@igpm.rwth-aachen.de> To: gcc-gnats@gcc.gnu.org, jiri.pittner@jh-inst.cas.cz, gcc-bugs@gcc.gnu.org, toon@moene.indiv.nluug.nl, aoliva@gcc.gnu.org Cc: Subject: Re: debug/9864: [3.2/3.3/3.4 regression] Fortran ICE in dwarf2out:add_abstract_origin_attribute with -g -O -finline-functions Date: Wed, 16 Apr 2003 22:35:55 +0200 (CEST) Here's an even shorter testcase without all the warnings (which still ICEs): -----------------------snip here------------------------ SUBROUTINE FOO CALL BAR END SUBROUTINE BAZ ENTRY BAR CALL QUUS END -----------------------snip here------------------------ Regards, Volker http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=9864 Looking at it. Subject: Bug 9864 CVSROOT: /cvs/gcc Module name: gcc Branch: gcc-3_3-branch Changes by: rth@gcc.gnu.org 2003-06-14 00:36:06 Modified files: gcc/f : ChangeLog com.c Log message: PR debug/9864 * com.c (ffecom_sym_transform_): Install FFEINFO_whereGLOBAL symbols in the global binding level. Patches: http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/f/ChangeLog.diff?cvsroot=gcc&only_with_tag=gcc-3_3-branch&r1=1.503.2.26&r2=1.503.2.27 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/f/com.c.diff?cvsroot=gcc&only_with_tag=gcc-3_3-branch&r1=1.182.4.2&r2=1.182.4.3 The bug seems to be not closed as the following produces ICE.
-------
c g77.331 -c -g -O -finline-functions bug.f -> ICE
c
subroutine aaaa
implicit none
external cccc
return
entry bbbb
end
------
bug.f:3: internal compiler error: in add_abstract_origin_attribute, at dwarf2out.c:9886
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://gcc.gnu.org/bugs.html> for instructions.
Release:
gcc 3.3.1
The ICE you are referring to is filled under PR 11522 so there is no need to reopen this bug. |
g77-3.3 bug.f -c -O -finline-functions -g where bug.f looks like this: ------------------------ SUBROUTINE MASTER(Q,MAXQ) INTEGER MAXQ DOUBLE PRECISION Q DIMENSION Q(MAXQ) CALL ANHARM(Q,MAXQ) END SUBROUTINE DUMMY ENTRY ANHARM CALL REPORT('a bit optimistic - come back next year') RETURN END ------------------------ bug.f: In subroutine `dummy': bug.f:5: warning: CALL ANHARM(Q,MAXQ) 1 bug.f:9: (continued): ENTRY ANHARM 2 Too few arguments for `anharm' at (1) versus invocation at (2) [info -f g77 M GLOBALS] bug.f: Outside of any program unit: bug.f:8: internal compiler error: in add_abstract_origin_attribute, at dwarf2out.c:9879 Please submit a full bug report, with preprocessed source if appropriate. See <URL:http://www.gnu.org/software/gcc/bugs.html> for instructions. Release: gcc 3.2 Environment: SuSE Linux/386 How-To-Repeat: The critical flags to trigger this are -g and -finline-functions (which requires -O).