OpenOffice 1.0.2-r1 compile fails

Jamon · Apprentice Joined: 03 May 2002 Posts: 173

Hello! I recently upgraded my PC's hardware to an Abit NF7-S (NForce2) with an AMD 2500+ Barton. My previous system was an AMD Thunderbird 1400MHz, with an MSI K7T266 Pro Motherboard (Via KT266 chipset). I made the many changes to my configuration, and I am working okay now without problems. My problem is, that the OpenOffice 1.0.2-r1 ebuild always fails, but not always at the same spot. I tried all forms of CFLAGS, including my old tbird ones, so that is not the problem. My old PC compiled OpenOffice 1.0.2 (no -r) without problems before the upgrade. The messages are something like this:

Making: ../../../unxlngi4.pro/slo/BIndexColumns.obj
g++ -fmessage-length=0 -c -I. -I. -I../inc -I../../inc -I../../../inc -I../../../unx/inc -I../../../unxlngi4.pro/inc -I. -I/var/tmp/portage/openoffice-1.0.2-r1/work/oo_1.0.2_src/solver/641/unxlngi4.pro/inc/stl -I/var/tmp/portage/openoffice-1.0.2-r1/work/oo_1.0.2_src/solver/641/unxlngi4.pro/inc/external -I/var/tmp/portage/openoffice-1.0.2-r1/work/oo_1.0.2_src/solver/641/unxlngi4.pro/inc -I/var/tmp/portage/openoffice-1.0.2-r1/work/oo_1.0.2_src/solenv/unxlngi4/inc -I/var/tmp/portage/openoffice-1.0.2-r1/work/oo_1.0.2_src/solenv/inc -I/var/tmp/portage/openoffice-1.0.2-r1/work/oo_1.0.2_src/res -I/var/tmp/portage/openoffice-1.0.2-r1/work/oo_1.0.2_src/solver/641/unxlngi4.pro/inc/stl -I/var/tmp/portage/openoffice-1.0.2-r1/work/oo_1.0.2_src/solenv/inc/Xp31 -I/opt/blackdown-jdk-1.3.1/include -I/opt/blackdown-jdk-1.3.1/include/linux -I/opt/blackdown-jdk-1.3.1/include/native_threads/include -I/usr/X11R6/include -I. -I../../../res -I. -march=athlon-xp -O2 -pipe -mmmx -m3dnow -msse -DTT_CONFIG_OPTION_BYTECODE_INTERPRETER -DTT_CONFIG_OPTION_BYTECODE_INTERPRETER -pipe -fno-for-scope -fpermissive -fexceptions -fno-enforce-eh-specs -fpic -DLINUX -DUNX -DVCL -DGCC -DC300 -DINTEL -DCVER=C300 -D_USE_NAMESPACE -DGLIBC=2 -DX86 -D_PTHREADS -D_REENTRANT -DNEW_SOLAR -D_USE_NAMESPACE=1 -DSTLPORT_VERSION=400 -D__DMAKE -DUNIX -DCPPU_ENV=gcc3 -DSUPD=641 -DBUILD=7663 -DPRODUCT -DNDEBUG -DPRODUCT_FULL -DOPTIMIZE -DEXCEPTIONS_ON -DCUI -DSOLAR_JAVA -DSRC641 -DSHAREDLIB -D_DLL_ -DMULTITHREAD -w -o ../../../unxlngi4.pro/slo/BIndexColumns.o /var/tmp/portage/openoffice-1.0.2-r1/work/oo_1.0.2_src/connectivity/source/drivers/adabas/BIndexColumns.cxx
In file included from ../../../unxlngi4.pro/inc/com/sun/star/sdbc/XRow.hdl:63,
from ../../../unxlngi4.pro/inc/com/sun/star/sdbc/XRow.hpp:5,
from /var/tmp/portage/openoffice-1.0.2-r1/work/oo_1.0.2_src/connectivity/source/drivers/adabas/BIndexColumns.cxx:69:
../../../unxlngi4.pro/inc/com/sun/star/util/Date.hdl:50: internal error: Segmentation fault
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:https://bugs.gentoo.org/> for instructions.
dmake: Error code 1, while making '../../../unxlngi4.pro/slo/BIndexColumns.obj'
---* TG_SLO.MK *---

ERROR: Error 65280 occurred while making /var/tmp/portage/openoffice-1.0.2-r1/work/oo_1.0.2_src/connectivity/source/drivers/adabas

!!! ERROR: app-office/openoffice-1.0.2-r1 failed.
!!! Function src_compile, Line 400, Exitcode 1
!!! Build failed!

A few things to note - It is not always this file, or in this location. I am not overclocking, and this error (or one like it) still occurs. (It occurs at default clockspeed and when overclocked.) Most other ebuilds complete fine. Even GLIBC compiled without errors, and that one is pretty long, too. Any thoughts/ideas? I searched without luck. Thanks!

Jamon
_________________
Looking for a Linux PC? Check us out at
http://www.progresssystemsinc.com!

AlterEgo · Veteran Joined: 25 Apr 2002 Posts: 1619

The segfault suggests memory or heat problems.
Run a memtest86, and monitor your temps and/or 12/5/Vcore value using lm-sensors under changing loads.

Jamon · Apprentice Joined: 03 May 2002 Posts: 173

I was afraid you might say that.. I have ran memtest86 without problems when I first got the PC. As for heat, I cannot check in Linux because lm_sensors does not recognize the sensors on my motherboard. Any idea on how to get it to work with an NF7-S? My 5 volts are fine, while 12 are around 11.8. Do you think that would be a problem?

Jamon
_________________
Looking for a Linux PC? Check us out at
http://www.progresssystemsinc.com!

-=Beelzebub=- · n00b Joined: 02 Apr 2003 Posts: 30 Location: ~beelzebub

I also got the same error as Jamon! Can't recall if it was on the same line though. Memtest reports no errors on this system and cpu is cool, ~40C (watercooled). Very strange...

Jamon · Apprentice Joined: 03 May 2002 Posts: 173

Do you have the same motherboard? What do you have? I have tried as many voltage combos I can think of.. I tried raising and lowering nearly everything possible. I got lm_sensors to work, and found some interesting things, like how the vcore voltage will go a bit lower as the system works harder, and the default voltages for many things are high. Even when I adjust them with softmenu, it still occurs. I am really close to starting to rant on the ABIT forums

. Any more ideas? And your (or anyone else's) situation with regards to the systems having compilation errors would be most appreciated. Also, could anyone else with an NForce2 chime in here? And point out your kernel, too. I am using 2.4.20-gaming-r1. Thanks!

Frustrated but still trying.....

Jamon
_________________
Looking for a Linux PC? Check us out at
http://www.progresssystemsinc.com!

-=Beelzebub=- · n00b Joined: 02 Apr 2003 Posts: 30 Location: ~beelzebub

Jamon · Apprentice Joined: 03 May 2002 Posts: 173

What kind of CPU do you have? How about video, etc.? I am really interested in figuring this out.. I know that the AMD board does use the same IDE chipset driver. I wonder. What filesystem do you use? CFLAGS?

Jamon
_________________
Looking for a Linux PC? Check us out at
http://www.progresssystemsinc.com!

jbuberel · n00b Joined: 02 Feb 2003 Posts: 31 Location: Sunnyvale, CA

One way to avoid the problems with installing OpenOffice.org from source is to install the binary ebuild:

Jamon · Apprentice Joined: 03 May 2002 Posts: 173

I know about openoffice-bin, but that is not really a solution, as my system can sometimes segfault on other ebuilds too, it's just that I can most likely restart it and have it complete (ccache helps). Openoffice is much to large for this. Basically, I do not believe in band-aids - I want to fix the underlying problem. I am far too much of a perfectionist - and indeed have too much pride in my system's stability - to be content with any ebuild not working because of hardware problems. It really disappoints me..

Jamon
_________________
Looking for a Linux PC? Check us out at
http://www.progresssystemsinc.com!

jbuberel · n00b Joined: 02 Feb 2003 Posts: 31 Location: Sunnyvale, CA

Install and run memtest:

Jamon · Apprentice Joined: 03 May 2002 Posts: 173

I also have ran memtest86; I let it run for over 10 hours one day. I have no problems as well really; I rarely might see a segfault but not too often. Otherwise, like you, I have a perfectly fine running system. But this really does bug me. It's like if I had a porsche, and I needed to rebuild the engine (not that I could) and someone said "I can give you this chevy 4 cylinder" so I would not have to do the work (assuming it would fit) - who in their right mind would do that? I have a porsche-like PC hardware, and I want my Linux (including OpenOffice) to be the same way! Just because both engines can't legally drive any faster doesn't mean I should be stuch with a slower one..

Jamon
_________________
Looking for a Linux PC? Check us out at
http://www.progresssystemsinc.com!

kyron · Posted: Sun Apr 20, 2003 2:35 am Post subject: Causes for Segfaults ...

While browsing through this thread I noticed that everyone seems to be concentrating on RAM as the cause for Segfaults. I think it's worth noting that Segfaults can be cause by any type of flakey hardware. Here are a few examples:

Overheating CPU
Bad RAM
Bad HDD controller
Bad HDD connector (IDE _and_/_or_ power)
Bad HDD

And I am only listing the ones I have personnaly experienced withoug starting to theorize on other probable sources like bad cards in PCI/ISA/AGP slots and defective chips.
_________________
M$ Windows: When in doubt, REBOOT
Linux: When in doubt, RTFM ;-)

ben_h · Posted: Wed Apr 23, 2003 7:10 am Post subject:

You're right, although I think borked memory is the most common cause.

In my experience memtest86 doesn't pick up errors very well. It's given the all-clear to a known-bad system at least once for me.

download prime95 (just google for it -- get the statically compiled version) and run the torture test. If you can run that for an hour with no errors or warnings, you've probably got a stable box. Run it for 24 hours with no problems and you can guarantee it's a rock-solid box.

In my experience, though, it picks up errors within 10 - 20 minutes.

kyron · Posted: Wed Apr 23, 2003 3:27 pm Post subject: Neat!!!

I am defenately trying that one out!!!! I have some systems for which I don't trust the hardware, this will come in rather handy!

_________________
M$ Windows: When in doubt, REBOOT
Linux: When in doubt, RTFM ;-)

montster · Posted: Thu Apr 24, 2003 6:22 pm Post subject:

Ok, I too am having this problem. It seems that the 1.0.2-r[whatever] and 1.0.3 ebuilds are not as stable as we would like. Is it only those of us running on AMD processors that are having the problems? Has anyone experienced any of these symptoms on processors other than Athlon/Duron?

I built openoffice-1.0.2 without any trouble whatsoever on my 700MHz Duron (512MB RAM, ultra-thouroughly tested with very expensive memory tester), but when the 1.0.2-r? version was marked stable, I couldn't build it. I didn't try very hard since openoffice-1.0.3 was out the next day (when I next ran 'emerge sync').

I have good RAM, a solid CPU and no excessive heat (that system stays under 95F all the time). I do not believe that these problems (that so many of us seem to be having) are due to (nearly?) imperceptable hardware glitches. As complicated as OpenOffice is (160+ million bytes of source code!), I believe there is some other, completely software, explaination. These are the kind of bugs that we all hate with a vengence.

I am wondering if one of the compile tools is the cause. More precisely, is there a bug in the openoffice-1.0.2r? and openoffice-1.0.3 build code (not the ebuild, but the Makefiles and such)? There may be a misconfiguration that works elsewhere but dies on Athlon due to miscompiled code in one of the tools?

With 1.0.3, it dies in EXACTLY the same place every time for me. That isn't hardware. I've also noticed that there is a string of errors occuring for a few minutes before reaching the one that finally nails the coffin shut. I've checked all the CFLAGS stuff. That isn't causing the problem.

I am intentionally not including the output from my failed build attempt at this time. I think that we need to know what categor(y/ies) of Gentooers are experiencing this difficulty first. Then, let's get some official bug reporting going on, if we can not find the solution, based on the common elements in our failures.

Ah, I just remembered that the OpenOffice-1.0.3 ebuild tells you at the very beginning that this thing is a really sensitive ebuild. Obviously, someone realised that this program was going to cause problems for some people. Why have none of them jumped into this thread, yet (I don't know who any of "them" are, so if you are here, sorry to have not recognized you).

I still have 1.0.2 running perfectly. There are no troubles whatsoever with any other ebuilds that I have (except when I filled up the whole hard drive during an emerge -e the other day...OOPS!).

Jamon · Apprentice Joined: 03 May 2002 Posts: 173

I tell you what, I did get my PC to build OpenOffice 1.0.3, and I did try various voltages, etc. but I cannot be certain that they had any real effect. I did succeed in an emerge -e world - and since I have not had that problem. I rebuilt the whole system with common CFLAGS, since I had changed CPUs before I changed my CFLAGS (from athon-tbird to athlon-xp). Though I am not certain of the whole problem, things are looking up now!!

Jamon
_________________
Looking for a Linux PC? Check us out at
http://www.progresssystemsinc.com!

kyron · Posted: Thu Apr 24, 2003 8:36 pm Post subject: ooffice does build

Note the I DID manage to build OpenOffice 1.0.3. The only problem I encountered is that, being extremely sensitive, it won't build with ccache nor distcc. I had to go and manually disable distcc and ccache in the ebuild since it kept ignoring my FEATURES in my make.conf file. Apart from that, 1.0.3 seems to be running smoothly though it annoyingly screws up during the installation when it tries to copy some stuff into the ~/.kde2 directory (which is a sybolic link to ~/.kde3.1 in my case).

All in all, openoffice does build without any problems, as long as you don't try any exotic tricks to make it build faster.....

_________________
M$ Windows: When in doubt, REBOOT
Linux: When in doubt, RTFM ;-)

TheCoop · Posted: Sat May 24, 2003 12:55 pm Post subject:

same problems, kt400 chipset this time with an overclocked 2500+ barton to 2.2ghz (you gotta love it, saved myself ~£150). Since other peopel are having the same problem I suspect it has nothing to do with the overclock, memtest works fine, system/cpu stays below 50C, no other problems apart from OOo
_________________
95% of all computer errors occur between chair and keyboard (TM)

"One World, One web, One program" - Microsoft Promo ad.
"Ein Volk, Ein Reich, Ein Führer" - Adolf Hitler

Change the world - move a rock

jbuberel · n00b Joined: 02 Feb 2003 Posts: 31 Location: Sunnyvale, CA

In an effort to save others from having to figure this out for themselves, I just tried to build the new openoffice-1.0.3-r1 ebuild, and ran into what amounts to the same problem:

frbie · n00b Joined: 11 May 2003 Posts: 63

Hello,

same error here with the 1.0.3-r1

I use a Athlon XP 1900+.

I really hate to emerge binaries

ph_flippy

GentooOpus · n00b Joined: 31 Mar 2003 Posts: 25 Location: Asheboro, NC

I've been hacking at this issure for about a week now.

I had an existing system running 1200 mhz Tbird, on an Abit KT7A.

This was a super stable system that had been in operation for 18 months.

I just upgraded it with a AthlonXP 2500+ and a Gigabyte KT400 mb (GA-K7VAX) My existing gentoo had been built from a stage 3 install of 1.4 with optimizations for cpu athlon-tbird.

Since the upgrade, I have had sporading faults while trying to compile many different packages. I first encounted problems with trying to recompile the kernel to take advantage of the new onboard hardware. Later problems with gcc and various other packages. Also, binary programs will fail sporadically.

Most of the kernel dumps refer to problems not being able to reach vm pages or address. ie VM: killing Galeon messages in logs.

I've even gone pack to a livecd and tried to reinstall with no success as I get segmentations faults a random points, with various packages regardless of make.conf optimizations.

I've also run the memtest kernel for 4 hours and found nothing wrong. However, it was not capable of correctly reading the cache configuration of the cpu it reported a L1 of 128kb and unknown L2

in realligy its L1 = 64kb, L2 512kb

Temps:

cpu ~41C
hdd ~ 99.3 F (surface at spindle)
ambient ~ 75.8 F

I'm going to find the prime95 test and if that doesn't finding anything wrong I may try installing another version of linux as Windows 2000 works fine from the dual boot partition.

Ok Back again, after a about 6 hours. Downloaded the prime95 benchmark and put in torture mode 3 Hours for a single client on windows now problem.

Switched to gentoo, got thinking maybe 1 isn't enough so I decided to really streas things and ran 4 separate instances of prime95 plus top to monitor the load. About 1 hour 15 minutes into the run, 3 of the clients caved stating that they could not access their temp file. What this means, I'm not sure. Could be end of partition space, end of swap, hdd failing, hdd controller failure. Overall temperature, characteristics are the same, although hdd spindle tmp is now 104.5 F.

System is still up and has not had a seg fault yet. Maybe this new kernel will work. I'm going to try a compile and see what happens.

jd

opus

hummus · n00b Joined: 17 Sep 2002 Posts: 40

its early monday, i just wanted to bump this thread since i have a presenetation due wednesday and my OOo build failed in the exact same way over the weekend (p3 500mhz)

is anyone else getting the feeling that all this memory testing might not be worth it given its happening on such various setups, especially setups that build everything else fine.

i tried quick google search and this came up, i'm not completely sure what to make of it
http://www.oooforum.org/forum/viewtopic.php?t=318

Jamon · Apprentice Joined: 03 May 2002 Posts: 173

I just wanted to let you all know that I (the originator of this thread) solved my problem. Yes, it was RAM. What?!? Well, it turns out that my Infineon RAM, even though it passed 10 hours of memtest86, was bad after all. I tried EVERYTHING for those less confident

. I was willing to live with it until a game I was playing - Dungeon Siege (in Windows) was constantly giving me exception errors. I investigated my problems again, and even though no other programs failed, as soon as I replaced the ram with some old PC2100 I had laying around, overclocked it to PC2700 speed (333MHz) the problems went away. Poof. Gone. Seriously. I RMA's my RAM with Kingston, and got shiny new Micron chips

and everything works fine. I compiled OOo-1.0.3-r1 with no problems at all - and my Windows game works fine too

(now if it could only WINE!

). So.. if you can borrow RAM to see if that's it, do it. I hope this helps.. (I was frustrated too)

Jamon

P.S. For those who don't remember or can't scroll up

I have a Barton 2500+ with an ABIT NF7-S (Nforce2). Have fun!

_________________
Looking for a Linux PC? Check us out at
http://www.progresssystemsinc.com!

GentooOpus · n00b Joined: 31 Mar 2003 Posts: 25 Location: Asheboro, NC

Well, i finally found the solution, at least for my woes and it wasn't ram. In sum, it was my own stupidity for not throughly resetting the CMOS on the MB prior to putting into service. The Gigabyte mb is a Dual
Bios variety and one of the last things that I tried was resetting all values and trying the backup bios. Once this was done, the system works like a charm. 24 hours of almost constant compiling with no seg faults!!!

My conclusion then with my situation was corrupt cmos values which may have affected ram timings etc. Once the bios area and cmos values were swapped and reset things began to work fine.

So sys upgrades follow the golden rule, zap the cmos first then put the board in the case.

opus