From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <gcc-prs-return-32407-listarch-gcc-prs=gcc.gnu.org@gcc.gnu.org>
Received: (qmail 27003 invoked by alias); 31 Mar 2003 19:16:01 -0000
Mailing-List: contact gcc-prs-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Archive: <http://gcc.gnu.org/ml/gcc-prs/>
List-Post: <mailto:gcc-prs@gcc.gnu.org>
List-Help: <mailto:gcc-prs-help@gcc.gnu.org>
Sender: gcc-prs-owner@gcc.gnu.org
Received: (qmail 26983 invoked by uid 71); 31 Mar 2003 19:16:01 -0000
Resent-Date: 31 Mar 2003 19:16:01 -0000
Resent-Message-ID: <20030331191601.26982.qmail@sources.redhat.com>
Resent-From: gcc-gnats@gcc.gnu.org (GNATS Filer)
Resent-Cc: gcc-prs@gcc.gnu.org, gcc-bugs@gcc.gnu.org
Resent-Reply-To: gcc-gnats@gcc.gnu.org, janis187@us.ibm.com
Received: (qmail 26754 invoked by uid 48); 31 Mar 2003 19:14:45 -0000
Message-Id: <20030331191445.26753.qmail@sources.redhat.com>
Date: Mon, 31 Mar 2003 19:18:00 -0000
From: janis187@us.ibm.com
Reply-To: janis187@us.ibm.com
To: gcc-gnats@gcc.gnu.org
X-Send-Pr-Version: gnatsweb-2.9.3 (1.1.1.1.2.31)
Subject: debug/10281: [3.4 regression] ICE in dwarf2out.c compiling glibc
X-SW-Source: 2003-03/txt/msg02137.txt.bz2
List-Id: <gcc-prs.sourceware.org>


>Number:         10281
>Category:       debug
>Synopsis:       [3.4 regression] ICE in dwarf2out.c compiling glibc
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    unassigned
>State:          open
>Class:          ice-on-legal-code
>Submitter-Id:   net
>Arrival-Date:   Mon Mar 31 19:16:01 UTC 2003
>Closed-Date:
>Last-Modified:
>Originator:     Janis Johnson
>Release:        gcc (GCC) 3.4 20030330 (experimental)
>Organization:
>Environment:
i686-pc-linux-gnu, powerpc64-unknown-linux-gnu
>Description:
The mainline C compiler ICEs when building glibc from CVS
sources using -g.  I first noticed this with
powerpc-unknown-linux-gnu, but the shorter test case also
ICEs for i686-pc-linux-gnu.  This does not affect the 3.3
or 3.2 branches.  The test case can probably be minimized
more, but I gave up at 59 lines.

Output of the compiler is:

vfprintf.c: In function `_IO_vfprintf':
vfprintf.c:1919: internal compiler error: in loc_descriptor_from_tree, at dwarf2
out.c:8963
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://gcc.gnu.org/bugs.html> for instructions.

A hunt for the patch that introduced the regression is in
progress.
>How-To-Repeat:
Compile the test case with -g.
>Fix:

>Release-Note:
>Audit-Trail:
>Unformatted:
----gnatsweb-attachment----
Content-Type: text/plain; name="ice.c"
Content-Disposition: inline; filename="ice.c"

struct dummy_locale_data
{
  const char *name;
  union dummy_locale_data_value
  {
    const char *string;
    unsigned int word;
  }
  values [];
};

typedef struct dummy_locale_struct
{
  struct dummy_locale_data *__locales[13];
} *dummy_locale_t;

extern int dummy1;
extern void *dummy2;

int
_IO_vfprintf (void *s, const char *format)
{
  int done;
  const unsigned char *f;
  static const int jump_table[4];

  {
    static const int step4_jumps[4];
    char spec;
    int offset;
    void * ptr;
    spec = (*++f);
    offset = ((spec) < ' ' || (spec) > 'z')
	      ? &&do_form_unknown - &&do_form_unknown
	      : step4_jumps[(jump_table[(int) (spec) - ' '])];
    ptr = &&do_form_unknown + offset;
    goto *ptr;

do_form_unknown:
    done++;
  }

  {
    static const int step4_jumps[2] = {
	&&do2_form_unsigned - &&do_form_unknown,
	&&do2_form_unknown - &&do_form_unknown };

do2_form_unsigned:
    {
      char buf[(((unsigned int) (((dummy_locale_t)
		({ dummy2; }))->__locales[0])->values[dummy1].word))];
    }

do2_form_unknown:
    done++;
  }

  return done;
}