--- Additional Comments From cvs-commit at gcc dot gnu dot org 2005-06-11
19:39 ---
Subject: Bug 21333
CVSROOT:/cvs/gcc
Module name:gcc
Changes by: [EMAIL PROTECTED] 2005-06-11 19:39:13
Modified files:
libgfortran: Makefile.in Makefile.am libgfortran.h
--- Additional Comments From tkoenig at gcc dot gnu dot org 2005-06-02
15:49 ---
Created an attachment (id=9014)
--> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=9014&action=view)
Proposed patch
--
What|Removed |Added
-
--- Additional Comments From tkoenig at gcc dot gnu dot org 2005-05-24
22:40 ---
For 4-byte complex, an option is to check the alignment at runtime.
If the complex is aligned on an 8-byte boundary, it should be
perfectly OK to call the 8-byte-integer routines.
The check could be done wi
--- Additional Comments From tkoenig at gcc dot gnu dot org 2005-05-17
13:25 ---
For ia-64, this can be exposed with prctl:
$ gfortran packtest.f90
$ prctl --unaligned=signal gdb ./a.out
GNU gdb Red Hat Linux (6.1post-1.20040607.52rh)
Copyright 2004 Free Software Foundation, Inc.
GDB is
--- Additional Comments From steven at gcc dot gnu dot org 2005-05-11
22:40 ---
Output on ia64:
a.out(31029): unaligned access to 0x6fffafbc, ip=0x4000aaa0
a.out(31029): unaligned access to 0x6fffafb4, ip=0x4000aaa0
a.out(31029): unaligned access to 0x6
--- Additional Comments From tobi at gcc dot gnu dot org 2005-05-11 21:58
---
This is a real bug, but I can't confirm it since I can't test it. In other
words, EQUIVALENCE means that we always have to assume the smaller alignment for
complex types (unless we can prove they're correctly
--- Additional Comments From tobi at gcc dot gnu dot org 2005-05-11 21:32
---
You're a nasty person. Very intriguing. Unfortunately, I can't test this.
--
What|Removed |Added
