| View previous topic :: View next topic |
| Author |
Message |
nekranos
n00b
Joined: 07 Apr 2004
Posts: 52
|
 Posted: Mon Jul 11, 2005 4:08 am Post subject: Bootstrapping = GCC segfaults |
 |
|
Tried to bootstrap from a stage1 tarball, GCC segfaulted.
Very similar to this:
https://forums.gentoo.org/viewtopic-t-350156.html
I don't think it's a hardware issue (except for the semi-not-updated BIOS since DFI doesn't offer any newer stable releases). Also, the machine runs Windows perfecty fine, I ran Prime95 for ~6 hours a week or so ago?
Relevant hardware:
DFI LanParty UT nF4 SLI-DR
Venice 3200+
OCZ Platinum Rev.2 2x512 2-2-2-5
Everything is running at stock.  |
|
| Back to top |
|
 |
Ariakus
n00b
Joined: 15 Dec 2004
Posts: 4
|
| Posted: Mon Jul 11, 2005 11:23 am Post subject: Problem |
|
|
Hello,
I had alot of problems bootstrapping and compiling some stuff with a NF4-SLID.
In the end, i found the problem to be at the memory setup ( bios ). I had all stock, and was not stable at all. I had to tweak TREF value, and the voltage value. I would increased the voltage by aounr 0.2V, and try to see if get's ok.
Try to increment your voltage, or to raise the timings.
Best Regard's |
|
| Back to top |
|
|
nekranos
n00b
Joined: 07 Apr 2004
Posts: 52
|
| Posted: Mon Jul 11, 2005 3:24 pm Post subject: |
|
|
Raised voltage but didn't do anything
Trying in single-channel mode w/ each stick at a time.... |
|
| Back to top |
|
|
nekranos
n00b
Joined: 07 Apr 2004
Posts: 52
|
| Posted: Mon Jul 11, 2005 6:21 pm Post subject: |
|
|
Nope, doesn't work. Voltages at 2.87, single stick fails to compile gcc as well. Tried using 2.5-3-3-8, system refuses to run at all.
Help |
|
| Back to top |
|
|
dellaxim
Tux's lil' helper
Joined: 04 Mar 2005
Posts: 138
|
| Posted: Mon Jul 11, 2005 6:29 pm Post subject: |
|
|
| please post your make.conf |
|
| Back to top |
|
|
nekranos
n00b
Joined: 07 Apr 2004
Posts: 52
|
| Posted: Mon Jul 11, 2005 6:31 pm Post subject: |
|
|
# These settings were set by the catalyst build script that automatically built$
# Please consult /etc/make.conf.example for a more detailed example
CFLAGS="-O2 -march=athlon64 -pipe"
CHOST="x86_64-pc-linux-gnu"
CXXFLAGS="${CFLAGS}"
MAKEOPTS="-j2"
USE="-X -kde -qt" |
|
| Back to top |
|
|
Birtz
Apprentice
Joined: 09 Feb 2005
Posts: 272
Location: Osijek / Croatia
|
| Posted: Tue Jul 12, 2005 6:05 am Post subject: |
|
|
Just a shot in the dark, but you could try to use gentoo-sources-2.6.12 kernel because there are whole lotta "bunch-o-bugs" AMD64-wise corrected. But, I assume you are doing a new install, that way you have limited options for choosing the kernel. As mentioned countless times in this threads, you can try the conservative approach to LiveCD kernel like booting with:
On the side note, if Windows and especially Prime95 worked for hours without a glitch, good chances are that you are experiencing kernel problems. In your shoes I would leave BIOS settings at ease (the workable setting for Prime95 you mentioned at the beginning of this thread).
Regards
_________________
It is not enough to have a good mind. The main thing is to use it well.
-- Rene Descartes
Don't have a childhood hero? How about Rob Hubbard http://www.freenetpages.co.uk/hp/tcworh/profile.htm |
|
| Back to top |
|
|
nekranos
n00b
Joined: 07 Apr 2004
Posts: 52
|
| Posted: Tue Jul 12, 2005 6:10 am Post subject: |
|
|
| Birtz wrote: | | In your shoes I would leave BIOS settings at ease (the workable setting for Prime95 you mentioned at the beginning of this thread). |
Yeah, I took everything back down to the stock settings much earlier, ~2.7v for DIMMs at 2-2-2-7, ~2.4-2.5v for the 3200+. Looser timings at 2.5-3-3-8 (as suggested elsewhere) won't let the system run
A few of the guys in #gentoo-amd64 helped me out a bit by suggesting to use gentoo-sources-2.6.12-r4 instead of my insane love of ck-sources 
Which I did. It worked this afternoon when I did an "emerge gcc". However, just on a whim, I decided to do a "emerge -e world" not even ten minutes ago.
Same error, same segfault. There's a bug on bugs.gentoo.org about this. 98350.
I'll try that cmdline when I start from scratch tomorrow. I don't like letting this bug beat me. Ugh. I don't see an option to disable APIC in gentoo-sources, though? |
|
| Back to top |
|
|
nekranos
n00b
Joined: 07 Apr 2004
Posts: 52
|
| Posted: Tue Jul 12, 2005 4:34 pm Post subject: |
|
|
Nope, passing the noacpi and noapic commands to the kernel doesn't seem to help GCC still segfaults at the same place.
Help |
|
| Back to top |
|
|
nekranos
n00b
Joined: 07 Apr 2004
Posts: 52
|
| Posted: Wed Jul 13, 2005 5:57 am Post subject: bump |
|
|
Bump. Loving the plethora of support replies.
| Code: | insn-attrtab.c: In function `internal_dfa_insn_code':
insn-attrtab.c:22564: internal compiler error: Segmentation fault
Please submit a full bug report,
with preprocessed source if appropriate. |
https://bugs.gentoo.org/show_bug.cgi?id=98350 <-- same settings as that user (emerge --info), only with the LiveCD kernel instead. |
|
| Back to top |
|
|
genneth
Apprentice
Joined: 24 Mar 2003
Posts: 152
Location: UK
|
| Posted: Wed Jul 13, 2005 7:50 am Post subject: |
|
|
Can you try re-bootstrapping without ~arch please? Then report back.
_________________
[img]http://www.kordix.com/rpi/?u=genneth[/img] |
|
| Back to top |
|
|
nekranos
n00b
Joined: 07 Apr 2004
Posts: 52
|
| Posted: Wed Jul 13, 2005 7:54 am Post subject: |
|
|
| genneth wrote: | | Can you try re-bootstrapping without ~arch please? Then report back. |
Can you elaborate on that? Without ~arch? How would I do that?
If you mean without ~amd64: no, I do not have that set as ACCEPT_KEYWORDS. I'm using whatever it is that the stage1 tarball comes as stock, which I'm assuming is just "amd64" - unlike the original bug reporter. |
|
| Back to top |
|
|
genneth
Apprentice
Joined: 24 Mar 2003
Posts: 152
Location: UK
|
| Posted: Wed Jul 13, 2005 2:01 pm Post subject: |
|
|
| nekranos wrote: | | genneth wrote: | | Can you try re-bootstrapping without ~arch please? Then report back. |
Can you elaborate on that? Without ~arch? How would I do that?
If you mean without ~amd64: no, I do not have that set as ACCEPT_KEYWORDS. I'm using whatever it is that the stage1 tarball comes as stock, which I'm assuming is just "amd64" - unlike the original bug reporter. |
I did mean that. Could you try with a stage 3?
_________________
[img]http://www.kordix.com/rpi/?u=genneth[/img] |
|
| Back to top |
|
|
dellaxim
Tux's lil' helper
Joined: 04 Mar 2005
Posts: 138
|
| Posted: Wed Jul 13, 2005 4:07 pm Post subject: |
|
|
I used to get the same bootstrapping segmentfault
then I tried to use
CFLAGS="-O2 -march=k8"
then it compiled ok...I used to compile it on an AMD64 mobile |
|
| Back to top |
|
|
nekranos
n00b
Joined: 07 Apr 2004
Posts: 52
|
| Posted: Wed Jul 13, 2005 4:45 pm Post subject: |
|
|
| genneth wrote: | | I did mean that. Could you try with a stage 3? |
Yep. Same thing. stage3 + 2.6.12-gentoo-r4 (or any kernel for that matter) + emerge -e world = segfault.
stage3 + kernel + emerge gcc = segfault.
| dellaxim wrote: | | CFLAGS="-O2 -march=k8" |
Huh, minus the -pipe flag (k8 = athlon64)...I'll give it a shot. |
|
| Back to top |
|
|
nekranos
n00b
Joined: 07 Apr 2004
Posts: 52
|
| Posted: Wed Jul 13, 2005 10:30 pm Post subject: |
|
|
Nope, still no good. Same segfault error.
C'mon, someone has to have an idea... |
|
| Back to top |
|
|
genneth
Apprentice
Joined: 24 Mar 2003
Posts: 152
Location: UK
|
| Posted: Thu Jul 14, 2005 7:42 am Post subject: |
|
|
| nekranos wrote: | Nope, still no good. Same segfault error.
C'mon, someone has to have an idea... |
Hmmm.... this is very odd. Have you run memtest? AFAIK, Prime only stresses the cpu; memtest will check for memory integrity. Leave it to run for a few hours.
_________________
[img]http://www.kordix.com/rpi/?u=genneth[/img] |
|
| Back to top |
|
|
nekranos
n00b
Joined: 07 Apr 2004
Posts: 52
|
| Posted: Fri Jul 15, 2005 3:22 am Post subject: |
|
|
| genneth wrote: | | Hmmm.... this is very odd. Have you run memtest? AFAIK, Prime only stresses the cpu; memtest will check for memory integrity. Leave it to run for a few hours. |
Yes.
No errors after several hours. |
|
| Back to top |
|
|
nekranos
n00b
Joined: 07 Apr 2004
Posts: 52
|
| Posted: Sun Jul 17, 2005 10:54 pm Post subject: |
|
|
Bump. Again.  |
|
| Back to top |
|
|
crazycat
l33t
Joined: 26 Aug 2003
Posts: 838
Location: Hamburg, Germany
|
| Posted: Mon Jul 18, 2005 5:29 am Post subject: |
|
|
| I think this problem is specific to your board. Have you take a look at dfi forum. What about changing the mainboard. This board is for overclockers and costs much so I'd think it might have something non standard extras/settings so if I had to fix that , I'd start trying to config bios to be as standard as possible. Have you tried for example freebsd on it ? or knoppix? How about of googling for dfi lanparty gcc segfault. |
|
| Back to top |
|
|
nekranos
n00b
Joined: 07 Apr 2004
Posts: 52
|
| Posted: Mon Jul 18, 2005 6:39 am Post subject: |
|
|
| crazycat wrote: | | I think this problem is specific to your board. Have you take a look at dfi forum. What about changing the mainboard. This board is for overclockers and costs much so I'd think it might have something non standard extras/settings so if I had to fix that , I'd start trying to config bios to be as standard as possible. Have you tried for example freebsd on it ? or knoppix? How about of googling for dfi lanparty gcc segfault. |
I think otherwise. I've seen other users here with similar configurations with no problems at all. Asking for support on the DFI-Street board about Linux is like...well, it has something to do with cyanide, I can tell you that. Changing the mainboard == $$$
Knoppix is fine.
Everything is "as standard as possible." BIOS settings are at the most basic and standard that will still allow Windows to run (except for OCZ memory, which only works at 2-2-2-7 for some odd reason, can't loosen timings).
I'm pretty sure it's not a hardware/board/BIOS setting issue.
Google brings up nothing. Did anyone even bother to read the bug link that I posted earlier - it's not just me who is having this problem with gcc, apparently. I can't pull much info off of the box since 2-3 minutes after gcc fails the kernel panics and dies, but it's pretty close to that bugreport, sans ~amd64. Weird, huh? |
|
| Back to top |
|
|
genneth
Apprentice
Joined: 24 Mar 2003
Posts: 152
Location: UK
|
| Posted: Mon Jul 18, 2005 7:22 am Post subject: |
|
|
You say that Knoppix is fine... could you try boostrapping a 32bit system?
_________________
[img]http://www.kordix.com/rpi/?u=genneth[/img] |
|
| Back to top |
|
|
RaZoR1394
Guru
Joined: 09 Jan 2005
Posts: 356
|
| Posted: Mon Jul 18, 2005 11:39 am Post subject: |
|
|
nekranos. I have a DFI nf4 board myself and I had to increase voltage and change some other settings before getting compiling to run stable. I suggest you visit DFI-Street and follow Rgones "get started" guide. It contains a lot of values to be tweaked before starting the board up. Some settings also depend on what CPU, graphics card and memory you have.
And as you run Ocz memory you may also wan't to check the official OCZ DFI startup guide. There is a link to it on DFI street.
Last edited by RaZoR1394 on Tue Jul 19, 2005 1:20 pm; edited 1 time in total |
|
| Back to top |
|
|
nekranos
n00b
Joined: 07 Apr 2004
Posts: 52
|
| Posted: Mon Jul 18, 2005 9:00 pm Post subject: |
|
|
| RaZoR1394 wrote: | I have a DFI nf4 board myself and I had to increase voltage and change some other settings before getting compiling to run run stable. I suggest you visit DFI-Street and follow Rgones "get started" guide. It contains a lot of values to be tweaked before starting the board up. Some settings also depend on what CPU, graphics card and memory you have.
And as you run Ocz memory you may also wan't to check the official Ocz DFI startup guide. There is a ling to it on DFI street. |
Thank you, I already have. I'm already running the CPU's Vcore and the memory's voltage to their highest while staying within the operating specs to not invalidate the warranty. And as I have already said, my memory is fine, I am running it with the settings given by one of the OCZ employess (who, as well, has given corrections that I am already using). Wouldn't want to invalidate the warranty now, would we?
I don't know how much I have to stress that the chances of it being a problem with hardware/settings is slim to none at this point; there was someone in #gentoo-amd64 that agreed.
| genneth wrote: | | You say that Knoppix is fine... could you try boostrapping a 32bit system? |
Will do. i686 and -march=athlon-xp? |
|
| Back to top |
|
|
genneth
Apprentice
Joined: 24 Mar 2003
Posts: 152
Location: UK
|
| Posted: Tue Jul 19, 2005 7:30 am Post subject: |
|
|
| nekranos wrote: | | genneth wrote: | | You say that Knoppix is fine... could you try boostrapping a 32bit system? |
Will do. i686 and -march=athlon-xp? |
Yep. And I agree that would be helper should read the entire thread first. We're not dealing with a complete newbie here
_________________
[img]http://www.kordix.com/rpi/?u=genneth[/img] |
|
| Back to top |
|
|
|
Gentoo on AMD64
