Using a left shift of an unsigned int by 15 and converting the result to double
does not work on sparc and creates broken code on solaris.
To reproduce, enter the following code and compile with -O2 and run:
/* snip */
#include stdio.h
#include stdlib.h
unsigned char SampleDepthOf() { return (
--- Comment #1 from thor at math dot tu-berlin dot de 2006-12-25 12:41
---
Addition: The following compiler switches also seem to be necessary to
reproduce the bug:
-fpic (or -fPIC), -On (n0) and -mcpu=v9
-mcpu=v8 *does not* have the problem.
--
--
pinskia at gcc dot gnu dot org changed:
What|Removed |Added
Severity|critical|normal
Component|c |middle-end
--- Comment #2 from pinskia at gcc dot gnu dot org 2006-12-25 16:56 ---
Can you try out 4.0.x or 4.1.x as 3.4.x is no longer being maintained?
--
pinskia at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #3 from burnus at gcc dot gnu dot org 2006-12-25 17:55 ---
I think the proper fix is to add . to the search list of directories where
include files may live.
I don't see how this will help; adding . for
'/home/allan/slot2usl/physcons.inc', searches at
--- Comment #3 from thor at math dot tu-berlin dot de 2006-12-25 18:36
---
Subject: Re: left-shift by 16 + int to double conversion
broken in optimized code
pinskia at gcc dot gnu dot org wrote:
--- Comment #2 from pinskia at gcc dot gnu dot org 2006-12-25 16:56
---
Can
--- Comment #17 from jvdelisle at gcc dot gnu dot org 2006-12-25 22:53
---
Subject: Bug 30200
Author: jvdelisle
Date: Mon Dec 25 22:53:29 2006
New Revision: 120199
URL: http://gcc.gnu.org/viewcvs?root=gccview=revrev=120199
Log:
2006-12-25 Jerry DeLisle [EMAIL PROTECTED]
PR
--- Comment #11 from jvdelisle at gcc dot gnu dot org 2006-12-25 22:57
---
Subject: Bug 30145
Author: jvdelisle
Date: Mon Dec 25 22:56:54 2006
New Revision: 120200
URL: http://gcc.gnu.org/viewcvs?root=gccview=revrev=120200
Log:
2006-12-25 Jerry DeLisle [EMAIL PROTECTED]
PR
--- Comment #12 from jvdelisle at gcc dot gnu dot org 2006-12-25 23:08
---
Subject: Bug 30145
Author: jvdelisle
Date: Mon Dec 25 23:08:17 2006
New Revision: 120201
URL: http://gcc.gnu.org/viewcvs?root=gccview=revrev=120201
Log:
2006-12-25 Jerry DeLisle [EMAIL PROTECTED]
PR
--- Comment #18 from jvdelisle at gcc dot gnu dot org 2006-12-25 23:08
---
Subject: Bug 30200
Author: jvdelisle
Date: Mon Dec 25 23:08:17 2006
New Revision: 120201
URL: http://gcc.gnu.org/viewcvs?root=gccview=revrev=120201
Log:
2006-12-25 Jerry DeLisle [EMAIL PROTECTED]
PR
--- Comment #19 from jvdelisle at gcc dot gnu dot org 2006-12-25 23:10
---
Fixed on 4.1
--
jvdelisle at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #1 from tromey at gcc dot gnu dot org 2006-12-26 00:24 ---
I think the rationale here is just that, if -MP added the main source
file as a phony target, then 'make' would not do anything at all if the
main file was missing. But, that result seems wrong -- for this to
--- Comment #2 from tromey at gcc dot gnu dot org 2006-12-26 00:33 ---
This is a bit weird but it is the documented behavior of -MMD.
To get a different target you must use -MT or -MQ.
My first reaction is that changing this may be unfriendly
(we don't know who, if anyone, is relying on
--- Comment #2 from tromey at gcc dot gnu dot org 2006-12-26 00:37 ---
FWIW I do not see this bug with the fedora core 6 gcc:
mopsy. gcc --version
gcc (GCC) 4.1.1 20061011 (Red Hat 4.1.1-30)
Copyright (C) 2006 Free Software Foundation, Inc.
This is free software; see the source for
--- Comment #2 from carlton at bactrian dot org 2006-12-26 03:24 ---
I'm trying to remember the details; I think I must have moved the
file from one directory to the other, and been pulling in directories
with vpath.
And one could certainly argue that any use of vpath is a
Makefile
--- Comment #4 from pluto at agmk dot net 2006-12-26 05:19 ---
$ sparc-sun-solaris2.9-g++ -Wall pr30290.c \
-o pr30290 --save-temps -fPIC -O2 -mcpu=v9 -g0 -s
pr30290-v8: ELF 32-bit MSB executable, SPARC, version 1 (SYSV), dynamically
linked (uses shared libs), stripped
pr30290-v9:
--- Comment #5 from pluto at agmk dot net 2006-12-26 05:22 ---
v8:
SampleDepthOf():
0: 81 c3 e0 08 retl
4: 90 10 20 10 mov 0x10, %o0
8: 81 c3 e0 08 retl
c: ae 03 c0 17 add %o7, %l7, %l7
0010 pow2test():
10: 9d e3 bf 88 save
Compilation by naive gcj usage results in ICE instead of a complaint.
gcc-4.1.1, sources are the same in HEAD.
$ gcc -o hello hello.java
hello.java:1: internal compiler error: in uses_jv_markobj_p, at
java/boehm.c:245
Please submit a full bug report,
with preprocessed source if appropriate.
See
--- Comment #1 from jan dot kratochvil at redhat dot com 2006-12-26 05:38
---
Created an attachment (id=12843)
-- (http://gcc.gnu.org/bugzilla/attachment.cgi?id=12843action=view)
Fix using a new gcc.c variable.
4.1.1 testsuite results not affected.
--
19 matches
Mail list logo
