[Bug optimization/11319] [3.3/3.4 regression] loop miscompiled on ppc32
PLEASE REPLY TO [EMAIL PROTECTED] ONLY, *NOT* [EMAIL PROTECTED] http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11319 --- Additional Comments From wilson at gcc dot gnu dot org 2003-09-07 05:35 --- The bug has been fixed by the alias.c patch. However, there were questions raised about the effect of the patch on the performance of gcc compiled code. I have left the bug report open for now as a reminder to look into this issue. I have checked in a patch to fix the loop_regs_update !INSN_P problem I mentioned earlier. http://gcc.gnu.org/ml/gcc-patches/2003-09/msg00397.html I have done SPEC testing on a Pentium4 system running GNU/Linux. I used -O2 -march=pentium4 as the gcc optimization options. I see no significant effect on the performance for those benchmarks that I could get running. I do see a noticable drop in 301.apsi performance after the first loop_regs_update patch went in, but it went back up after the alias.c patch went in, so the problem seems to have corrected itself. PowerPC performance tests are pending. --- You are receiving this mail because: --- You are on the CC list for the bug, or are watching someone who is.
Processed: Moved.
Processing commands for [EMAIL PROTECTED]: > submitter 180853 ! Bug#180853: Code within #if/#endif incorrectly highlighted Changed Bug submitter from Bart Samwel <[EMAIL PROTECTED]> to Bart Samwel <[EMAIL PROTECTED]>. > submitter 180856 ! Bug#180856: Build command not run in file's directory Changed Bug submitter from Bart Samwel <[EMAIL PROTECTED]> to Bart Samwel <[EMAIL PROTECTED]>. > submitter 187599 ! Bug#187599: Cannot subtract from streampos Changed Bug submitter from Bart Samwel <[EMAIL PROTECTED]> to Bart Samwel <[EMAIL PROTECTED]>. > submitter 163053 ! Bug#163053: galeon-nautilus: Zoom spinner not correct after session restore Changed Bug submitter from "Bart Samwel" <[EMAIL PROTECTED]> to Bart Samwel <[EMAIL PROTECTED]>. > submitter 163348 ! Bug#163348: kcalc: No exit in Sawfish Changed Bug submitter from "Bart Samwel" <[EMAIL PROTECTED]> to Bart Samwel <[EMAIL PROTECTED]>. (By the way, that Bug is currently marked as done.) > submitter 166775 ! Bug#166775: Font p052004l.pfb crashes gnome-font-install in libfreetype Changed Bug submitter from Bart Samwel <[EMAIL PROTECTED]> to Bart Samwel <[EMAIL PROTECTED]>. > submitter 187500 ! Bug#187500: Insert Ascii File as Lines into Lyx-Code drops empty lines Changed Bug submitter from Bart Samwel <[EMAIL PROTECTED]> to Bart Samwel <[EMAIL PROTECTED]>. > submitter 191655 ! Bug#191655: lyx: Changing size unit in graphic dialog does not enable "Apply" and "OK" Changed Bug submitter from Bart Samwel <[EMAIL PROTECTED]> to Bart Samwel <[EMAIL PROTECTED]>. > submitter 208737 ! Bug#208737: arson: Arson segfaults at startup most of the time Changed Bug submitter from Bart Samwel <[EMAIL PROTECTED]> to Bart Samwel <[EMAIL PROTECTED]>. > submitter 208807 ! Bug#208807: mozilla-browser: When minimum font size set, pages rendered with overlapping fonts Changed Bug submitter from Bart Samwel <[EMAIL PROTECTED]> to Bart Samwel <[EMAIL PROTECTED]>. > submitter 208803 ! Bug#208803: galeon: Zoom spinner unusable for large system font sizes Changed Bug submitter from Bart Samwel <[EMAIL PROTECTED]> to Bart Samwel <[EMAIL PROTECTED]>. > submitter 208805 ! Bug#208805: galeon: Tabs menu does not contain "Close Tab" option Changed Bug submitter from Bart Samwel <[EMAIL PROTECTED]> to Bart Samwel <[EMAIL PROTECTED]>. > submitter 201432 ! Bug#201432: gimp1.3: Keyboard shortcuts for "Close" and "Cancel" both C! Changed Bug submitter from Bart Samwel <[EMAIL PROTECTED]> to Bart Samwel <[EMAIL PROTECTED]>. > submitter 192779 ! Bug#192779: spamassassin: Would like to be able to combine -d and -t Changed Bug submitter from Bart Samwel <[EMAIL PROTECTED]> to Bart Samwel <[EMAIL PROTECTED]>. > submitter 191439 ! Bug#191439: Bookmarks created after crash on date (null). Changed Bug submitter from Bart Samwel <[EMAIL PROTECTED]> to Bart Samwel <[EMAIL PROTECTED]>. (By the way, that Bug is currently marked as done.) > submitter 192420 ! Bug#192420: galeon: Going back to page containing POSTDATA asks to press "OK" but only "No" and "Yes" available. Changed Bug submitter from Bart Samwel <[EMAIL PROTECTED]> to Bart Samwel <[EMAIL PROTECTED]>. (By the way, that Bug is currently marked as done.) > thanks Stopping processing here. Please contact me if you need assistance. Debian bug tracking system administrator (administrator, Debian Bugs database)
nearbyint returns always 895
>Submitter-Id: net
>Originator:Stefan Karrmann
>Organization:
>Confidential: no
>Synopsis: nearbyint returns always 895
>Severity: serious
>Priority: medium
>Category: c
>Class: wrong-code
>Release: 3.0.3 (Debian testing/unstable)
>Environment:
System: Linux johann 2.4.21 #2 Sun Aug 17 20:52:39 CEST 2003 i686 unknown
Architecture: i686
host: i686-pc-linux-gnu
build: i686-pc-linux-gnu
target: i686-pc-linux-gnu
configured with: ./configure --enable-languages=c++
>Description:
Calling nearbyint returns always 895.0
>How-To-Repeat:
Compile and run:
#include
#include
void show (double x)
{
printf("nearbyint(%+.3e) = %d\n", x, (int)(nearbyint(x)));
if(0.0 == x) return;
x = -x;
printf("nearbyint(%+.3e) = %d\n", x, (int)(nearbyint(x)));
x = 1/x;
printf("nearbyint(%+.3e) = %d\n", x, (int)(nearbyint(x)));
x = -x;
printf("nearbyint(%+.3e) = %d\n", x, (int)(nearbyint(x)));
}
int main ()
{
show(1.1);
show(11);
show(111.1);
}
>Fix:
Re: Bug#204654: glibc 2.3.2-2 breaks apache on ia64
On Sat, Aug 09, 2003 at 03:32:12PM +0900, GOTO Masanori wrote: > At Sat, 9 Aug 2003 02:52:01 +0100, > Matthew Wilcox wrote: > > [EMAIL PROTECTED]:~$ sudo /etc/init.d/apache start > > Starting web server: apache/etc/init.d/apache: line 70: 29184 Segmentation > > fault start-stop-daemon --start --pidfile $PIDFILE --exec $DAEMON > > failed > > > > [EMAIL PROTECTED]:~# /usr/sbin/apache > > Segmentation fault > > Could you tell me which apache version do you use? > > > Program received signal SIGSEGV, Segmentation fault. > > server_type (cmd=0x600282e3, dummy=0x6001a598, > > arg=0x6 > > 282d8 "standalone") at http_core.c:2032 > > 2032http_core.c: No such file or directory. > > in http_core.c > > http_core.c:2032 is not server_type() in apache 1.3.27.0-2, so your > apache is not the latest. I've started investigating this myself, and your statement is incorrect. server_type() occupies lines 2018 to 2036 of http_core.c in apache 1.3.27.0-2. The bug I reported here is the same as that reported by Dann Frazier in bug #200698. Why a recompile fixed it, I don't know. A recompile of 1.3.27.0-2 still fixes it (and btw means that the patch from Dann isn't necessary ...). Could this be some exceptionally strange gcc bug?! Let's summarise: Apache 1.3.27.0-2 in testing does not work against glibc 2.3.1-16 or 2.3.2-5 Apache 1.3.27-0.2 rebuilt in unstable against glibc 2.3.2-5 headers with gcc 1:3.3.2ds0-0pre1 works against glibc 2.3.2-5. Apache 1.3.27-0.2 rebuilt in a testing chroot against 2.3.1-16 headers with gcc 1:3.3.1-0pre0 does not work against glibc 2.3.1-16 or 2.3.2-5 This does seem to point to gcc, not glibc, being the problem. And also, btw, points to me being mistaken about glibc triggering this problem. My apologies. I guess I got apache and glibc on the same update. -- "It's not Hollywood. War is real, war is primarily not about defeat or victory, it is about death. I've seen thousands and thousands of dead bodies. Do you think I want to have an academic debate on this subject?" -- Robert Fisk
Processed: per the last message to the bug...
Processing commands for [EMAIL PROTECTED]: > retitle 204654 gcc 3.3.2 breaks apache on ia64 Bug#204654: glibc 2.3.2-2 breaks apache on ia64 Changed Bug title. > reassign 204654 gcc-3.3 Bug#204654: gcc 3.3.2 breaks apache on ia64 Bug reassigned from package `libc6.1' to `gcc-3.3'. > -- Stopping processing here. Please contact me if you need assistance. Debian bug tracking system administrator (administrator, Debian Bugs database)
Processed: ...
Processing commands for [EMAIL PROTECTED]: > retitle 208717 gcc-3.3_1:3.3.2ds1-0pre2(unstable/ia64,alpha): FTBFS: ICE at > src/gcc/p/rts/rtsc.pas:60 Bug#208717: gcc-3.3_1:3.3.2ds1-0pre2(unstable/ia64): FTBFS: ICE Changed Bug title. > -- Stopping processing here. Please contact me if you need assistance. Debian bug tracking system administrator (administrator, Debian Bugs database)
Processed: oops
Processing commands for [EMAIL PROTECTED]: > retitle 204654 gcc 3.3.1 breaks apache on ia64 Bug#204654: gcc 3.3.2 breaks apache on ia64 Changed Bug title. > -- Stopping processing here. Please contact me if you need assistance. Debian bug tracking system administrator (administrator, Debian Bugs database)
Bug#209152: gcc-3.3: Large array problem on 64 bit platforms
Package: gcc-3.3 Version: 1:3.3.2-0pre1 Severity: normal Try this simple test program on IA64 --- test.c --- #define SZ (1L<<32) char x[SZ]; char y[SZ-1]; --- end --- you will see that gcc wraps the array size around --- test.s --- .file "test.c" .pred.safe_across_calls p1-p5,p16-p63 > .common x#,0,8 .common y#,4294967295,8 .ident "GCC: (GNU) 3.3.2 20030812 (Debian prerelease)" --- end --- This appears to be fixed in gcc CVS thanks to the checkin you can see from this : http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/varasm.c?rev=1.318.2.7&content-type=text/x-cvsweb-markup I am in the dark as to how the branching system works with gcc and what Debian is using; this bug report is requesting the changes to properly output HOST_WIDE_INT from that CVS log be integrated, but if someone could explain how it works that would be great too. -i [EMAIL PROTECTED] http://www.gelato.unsw.edu.au -- System Information: Debian Release: testing/unstable Architecture: ia64 Kernel: Linux tartufi 2.6.0-test4 #2 SMP Wed Aug 27 13:09:22 EST 2003 ia64 Locale: LANG=C, LC_CTYPE=C Versions of packages gcc-3.3 depends on: ii binutils 2.14.90.0.5-0.2 The GNU assembler, linker and bina ii cpp-3.3 1:3.3.2-0pre1 The GNU C preprocessor ii gcc-3.3-base 1:3.3.2-0pre1 The GNU Compiler Collection (base ii libc6.1 2.3.2-3 GNU C Library: Shared libraries an ii libgcc1 1:3.3.2-0pre1 GCC support library -- no debconf information
