c/6300: sparcv9-sun-solaris2.7 gcc-3.1 C testsuite failure in gcc.dg/cpp/charconst.c

Tue Apr 16 17:26:00 GMT 2002

The following reply was made to PR c/6300; it has been noted by GNATS.

From: "Kaveh R. Ghazi" <ghazi@caip.rutgers.edu>
To: neil@daikokuya.demon.co.uk
Cc: gcc-gnats@gcc.gnu.org, ghazi@caip.rutgers.edu, zack@codesourcery.com
Subject: Re: c/6300: sparcv9-sun-solaris2.7 gcc-3.1 C testsuite failure in gcc.dg/cpp/charconst.c
Date: Tue, 16 Apr 2002 20:19:15 -0400 (EDT)

  > From: "Kaveh R. Ghazi" <ghazi@caip.rutgers.edu>
  > 
  >   > From: Neil Booth <neil@daikokuya.demon.co.uk>
  >   > 
  >   > Kaveh R. Ghazi wrote:-
  >   > 
  >   > >  > FAIL: gcc.dg/cpp/charconst.c (test for excess errors)
  >   > > 
  >   > > >From gcc.log, the excess error is:
  >   > > 
  >   > >  > Excess errors:
  >   > >  > gcc.dg/cpp/charconst.c:28: warning: overflow in implicit constant conversion
  >   > > 
  >   > > It seems that sparcv9- (i.e. not sparc-) is necessary to trigger
  >   > the bug.  See: http://gcc.gnu.org/ml/gcc-testresults/2002-04/msg00484.html
  >   > > vs: http://gcc.gnu.org/ml/gcc-testresults/2002-04/msg00482.html
  >   > > 
  >   > > *** This is a regression from 3.0.3.
  >   > 
  >   > Any idea what patch caused it?
  >   > Neil.
  >  
  >  No.  I traced back and found that it didn't happen here:
  >  6/7/2001 http://gcc.gnu.org/ml/gcc-testresults/2001-06/msg00210.html
  >  
  >  but it did occur here:
  >  7/10/2001 http://gcc.gnu.org/ml/gcc-testresults/2001-07/msg00160.html
  >  7/17/2001 http://gcc.gnu.org/ml/gcc-testresults/2001-07/msg00219.html
  >  
  >  This interval seems to have been a particularing unstable period
  >  for solaris2 on what was then the 3.1 trunk.  So unfortunately
  >  there aren't any reports in between those listed above which
  >  narrow down the exact date when it started for solaris2.

 I narrowed the search by building cc1 using various "cvs up -D <date>"
 and found that 6/9/2001 was ok and 6/10/2001 had the problem.  I've
 included the ChangeLog diff between those two dates.  If I had to
 guess, I'd say that the reenabling of the solaris2 bi-arch was the
 trigger.  However that probably only enabled the bug and wasn't
 necessarily the underlying cause. :-(

 > 2001-06-09  Alexandre Oliva  <aoliva@redhat.com>
 > 
 > 	* doc/invoke.texi (C Dialect Options): Document -aux-info.
 > 	(Debugging Options): Document -print-multi-directory and
 > 	-print-multi-lib.
 > 	* doc/gcc.1: Rebuilt.
 > 
 > 2001-06-09  Alexandre Oliva  <aoliva@redhat.com>
 > 
 > 	* toplev.c (independent_decode_option): Require `=' between
 > 	`-aux-info' and filename in the same argument.
 > 	* gcc.c: Don't pass -aux-info=filename twice.
 > 
 > 2001-06-09  Mark Mitchell  <mark@codesourcery.com>
 > 
 > 	* expr.c (expand_expr, case ARRAY_REF): Do not replace
 > 	an array element with the known initializing value if it has
 > 	side-effects.
 > 
 > 	* toplev.c (rest_of_compilation): Remove dead code before
 > 	purge_addressof.
 > 	
 > 2001-06-09  Neil Booth  <neil@daikokuya.demon.co.uk>
 > 
 > 	* cppinit.c (print_help): Document -fpreprocessed.
 > 	* invoke.texi: Similarly.
 > 
 > Sun Jun 10 00:35:37 CEST 2001  Jan Hubicka  <jh@suse.cz>
 > 
 > 	* predict.def: New file.
 > 	* predict.h: New file.
 > 	* predict.c: Include predict.h.
 > 	(predictor_info): New structure and array.
 > 	(predict_edge, predict_insn): Rewrite; make global.
 > 	(predict_edge_def, predict_insn_def): New global functions.
 > 	(dump_prediction, combine_predictions_for_insns): New
 > 	static functions.
 > 	(estimate_probability): Change calls to predict_edge
 > 	to predict_edge_def and calls to predict_insn to predict_insn_def;
 > 	combine probabilities.
 > 	(expected_value_to_br_prob): Behave as predictor.
 > 	* rtl.h (REG_BR_PRED): Update coment.
 > 	(invert_br_probabilities): New.
 > 	* jump.c: Include predict.h
 > 	(duplicate_loop_exit_test): Drop branch prediction notes.
 > 	(invert_jump): Use invert_br_probabilities.
 > 	* Makefile.in: Add dependancy on predict.h for jump and predict pass.
 > 	* ifcvt.c (dead_or_predicable): Use invert_br_probabilities.
 > 	* combine.c (distribute_notes): Handle BR_PRED.
 > 
 > Sat Jun  9 23:29:41 CEST 2001  Jan Hubicka  <jh@suse.cz>
 > 
 > 	* predict.c (predict_insn, predict_edge): New static functions.
 > 	(estimate_probability): Revamp to use new functions;
 > 	fix loop header heruistics; add loop exist heruistics
 > 
 > 2001-06-09  Alexandre Oliva  <aoliva@redhat.com>
 > 
 > 	* config.gcc: Re-enable bi-arch sparc on Solaris 7 and above.
 > 
 > 2001-06-09  Peter Jakubek <pjak@snafu.de>
 > 
 > 	* config/m68k/m68k.md (ashrdi_const): Fix right shift by 16.
 > 
 > 2001-06-09  John David Anglin  <dave@hiauly1.hia.nrc.ca>
 > 
 > 	* vax.h (NOTICE_UPDATE_CC): Set CC_NO_OVERFLOW in cc_status.flags
 > 	for integer operations that don't update C bit.
 > 
 > 2001-06-09  Nicola Pero <nicola@brainstorm.co.uk>
 > 
 > 	* gthr-posix.h (__gthread_objc_init_thread_system): If
 > 	pthread_key_create fails, it's not clear what the return value is.
 > 	This obvious fix makes sure it always returns -1 if a problem
 > 	occurs.
 > 
 > 2001-06-09  Joseph S. Myers  <jsm28@cam.ac.uk>
 > 
 > 	* doc/gcc.texi: Use @defcodeindex for the index of options.  Add
 > 	more index entries.  Use more logical markup where appropriate.
 > 	Use TeX quotes and dashes.
 > 
 > 2001-06-09  Benjamin Kosnik  <bkoz@redhat.com>
 > 
 > 	* unwind-pe.h: Conditionally qualify references to abort with
 > 	std::.
 > 
 > 2001-06-09  Julian Foad <julian_foad@hotmail.com>
 > 
 > 	* doc/invoke.texi: Remove superfluous text.
 > 