Please create an account to participate in the Slashdot moderation system

 


Forgot your password?
Close
typodupeerror
×
Linux Software

GCC Instability Problems With SuSE/AMD K6? 31

Posted by Cliff from the what's-the-failure-point dept.
cmickelson asks: "I've been running SuSE Linux 6.x on my laptop for over a year now and everytime I would compile something I would have to run make several times because of errors during compiling. Now my laptop is very unstable, Windows is unusable because of constant crashes, but Linux generally runs well except for Netscape core dumps, and the compiling problems. Anyway I just assumed the errors were due the system's instablility. However, I recently assembled a new desktop computer and installed SuSE 6.4 on that and the same thing happened when I tried to compile Samba. I had to run make five times before the whole source tree was completed. Is this typical? I find it hard to believe that gcc/gpp could be that unstable, but why else would this happen with two different systems? Is this due to the AMD K6-2 processors (which both laptops have), or is SuSE the problem?"
This discussion has been archived. No new comments can be posted.

GCC Instability Problems with SuSE/AMD K6? More

GCC Instability Problems with SuSE/AMD K6?

Comments Filter:

  • Re:OMG! (Score:1)

    by Anonymous Coward
    Fucking lay off Redhat you fuckwit. All the fucking whining here on Slashdot, and yet I've been using Redhat for nearly 2 years, used x.0 versions had absolutely ZERO problems. SuSE however is fucking terrible. I used 6.2 and a friend has used 6.3 and we both dumped it. If you're gonna pick on a distro, pick on SuSE. I've used 8 different distros and SuSE is BY FAR the worst of the lot. Random crashes requiring a hard reset and that shit. Redhat never did that, so before you follow the Slashdot party line QUIT YOUR FUCKING WHINING and get a fucking clue.

  • crash during compiling (Score:1)

    by Anonymous Coward
    I've had the same problem on my desktop some time ago. After some search I found out that the reason for crashes was the power saving feature of the bios. When I started compiling stuff and left the computer untouched, the power saving feature would be activated after some time and the computer crashed. Switching of this feature was the easiest solution for me. Solution nr. 2: Keep moving around your mouse pointer or bang on your keyboard every now and then so that the computer will think your still active.

  • But who uses a kernel off an install cd anyway?

    By stock kernel, I mean one available off of ftp://ftp.[country code].kernel.org/pub/linux/kernels/v2.x/linux-2.x. y.tar.bz2. RedHat adds custom shit ot the kernel, causing problems all over the place unless you stick to RedHat released RPMs.

  • this really sounds like flaky memory, or possibly an overheated processor. if at all possible, try swapping out the memory with some known good chips.
    ---
  • never had any problems compiling xith SuSE 6.4. It was very stable. I was using an AMD K6-3. Two other computers I have ran K6-2 with SuSE 6.4 and had no trouble either.
  • it will run at 350 (3.5x 100), but it does bizarre things... everything detects its speed to be 333, and things would crash when they shouldn't.....

    i clocked it back to 266 for a long time and it liked that... i've had it at 350 since septemeber and its had strange problems.... streaming mp3s off of it w/ samba works fine, but sending an iso craps out halfway through...

    i'm trying it at 333 (4 x 83) right now, it seems to work, but that may change...

  • i don't have the board's model number handy. (i think it may predate the 5300 though. might not even be a socket 7 board now that i think about it. maybe a 50x0 series?)

    it actually has a jumper on it to select whether you're dropping in 5v edo dimms or 3.3v sdram. the board and memory are from probably about 3 years ago. so, while i wouldn't deny the possibility of having older, more tolerant sdram, i would hope that the board runs at 3.3v if it has a jumper to do so. ;-)

    and you're correct, they don't have all the manuals online.

    food for thought.

    # cat .sig

  • i have a cheap a-trend mobo here that supports 5v simms and 3.3v dimms simultaneously and it worked fine when i used to do that. however, i was using early (read: slower, 12ns i think, pc66) sdram in the 3.3v slots.

    in another system i had a pc66 10 or 12ns chip combined with a pc100 6ns chip, both 3.3v sdram and the system was unstable. either chip works fine alone or with like chips, can't use them together though. perhaps it's more of a timing problem?

    # cat .sig

  • K6 processors may have accidentally obtained their reputation of instability. There probably is a correlation between a K6 processor and the overall system 'tweakedness' factor (overclocking, fiddling with memory timings, lots of cruel and unusual peripherals etc.), because the K6 processor is popular among computer hobbyists. Many processors in tweaked systems naturally leads to many reports of problems, and the correlation will make it appear that the processor is the component to blame, since it is practically always mentioned in the reports. The same phenomenon will probably affect the Athlon line of processors as well.
  • I have a k6-2 333 with 256 mb ram, and its fine. Ive had probs with video with a crappy generic video card, but I havent ran into sig 11 or compiling probs, and I am using SuSE 6.4 as well.
  • I could really see it being an overheated cpu. I remember a few years back reading an article about the mobile K6 series was sometimes a problem because they ran very hot and not only were there cpu problems but the heat affected other, nearby, components. Don't know if this applies to the K6-2 but it's a thought. I've run SuSE 6.X on a K6-III 450 for well over a year now with no instability problems I can think of. Don't even get me started about Red Hat. But anyway to each his own. If you're happy with Red Hat stick with it. I'm pretty sure however that those red hat users that are blasting SuSE had problems with it because it's not as dumbed down (although it seems headed that way).
  • I've been running an OpenBSD-based firewall on a K6-200 for this school year, and I couldn't recompile the kernel or any userland binaries until I dropped the memory form 64MB to 32MB. So, if it's the >32MB bug that's affecting you, installing OpenBSD might not help.
  • I seem to recall there was a bad batch of k6's that had a bug that woudln't allow them to compile in gcc ... I had a friend who would always ask me to compile shit for him because his machine couldn't do it ...

    I'm sorry I don't recall more then that but I'm willing to bet this is your problem

  • I have K6-2 266Mhz 64MB SDRAM and MVP3 motherboard running Mandrake 7.2. I got this used computer from a friend for free and the OS for free (FTP install) :D. It was 100% stable.

    I tried to compile 2.4 kernel and there's no error. However after I boot into 2.4.0-test9 kernel, make generates some error. Also after 1 crash in 2.4 kernel somewhere in the file system (/usr/src/linux/.....) was corrupted. I use ReiserFS and I dunno how to do a fsck with it. Everytime I rm -rf linux/ the system will HALT :(.
  • Actually, i have some really weird Intel hardware from 93, you can't install anything but Win31 or a UNIX variant on it. I used to get alot of Sig 11's when i ran linux or freebsd, but i switched to OpenBSD about a year ago and havn't really had any problems since.
  • I checked out what I think is a-trend's website. www.atrend.com

    The only motherboard they have a manual for that is Socket 7 and has both DIMM and SIMM slots is the ATC5300.

    If this is the board you have, the manual says you can't mix 3.3v SDRAM and regular DRAM (in Chinaglish. I'd give $$$, just once, for a motherboard manual actually written in reasonable English.)

    Some very old (4-5 years) SDRAM could handle 5 volts. Maybe you've got that?

    But maybe they don't have all the manuals online. It wouldn't surprise me. :-)

    >perhaps it's more of a timing problem?

    That's also very possible.
  • Just a random thought, but I recently played with a Dual-PPro motherboard (only one CPU inserted) that exhibited a strange problem. When running Linux, only 64MB RAM would work. The memory sticks were fine, but 128 MB of RAM would just cause the kernel to OOPS its brains out during Mandrake installation startup. The paramater mem= didn't do a thing.

    I finally fixed it by bring up the BIOS and doing a [Load BIOS defaults] which must have done something to the chipset settings because suddenly it started working. Now 128MB of ram works beautifully and the system has been handling a load very well.

    The BIOS settings may not be optimal for speed, but it can make a difference. Try writing down your BIOS settings and do a defaults reload.
  • If it was a hardware issue, it would happen on only one system or the other.

    You might run RAM diagnostics and if it does come up with some thing, you can patch the kernel with this [zonnet.nl].

    I'm also working under the assumption that you've checked out SUSE's web site for bug reports, and done other basic research before posting here.

  • Give me a break

    This s a question from a SUSE user, only comments so far have been in relation to SUSE

    So where exactly did your flame about RH come from

    (yes - I do run RH7 on numerous boxen and I am getting seriously pissed off by anti RH comments on Slashdat by people who don't use it)

  • The Way I see it is that if the computer is able to run it is in close enough to good condition to be able to compile stuff.
    I help my Drafting teacher out in the Lab at my school, t set up computers, and whenever we couldn't get something to work, he would want to blame it on the condition of the computer, but I see it is as if it can run at all, its in condition enough to be able to run programs and such.

    Unless, you have cheap-ram, if thats the case it most likely could be the source of the problem
  • ideat

  • Aside from the stuff already mentioned (overheating cpu, bad mem, windows problems with >400mhz amd), I have seen a few mb/processor combinations that just won't work (usually cheap mb's). You could try stepping down the speed of the cpu (in which case, the cpu could be defective), or trying to clock down the fsb to 66mhz (if you're doing 100mhz... especially if you don't have 100mhz mem ;), but it might be worth a try... if that works, and you do have 100mhz ram, see if you can find an upgrade from the mb manufacturer.

    FWIW, I've been running a k6-2 300 on a tmc mb for a few years without a hitch (moving to duron soon :)
  • Using SuSe 6.2 and an AMD-K6-2 550 MHz, I also received Signal 11s at random during compilations.

    If I started the compiler again, it would compile the code where it aborted, just to sig 11 at another location.

    I then lowered the CPU clock from 550 MHz to 533 Mhz, and I have never seen it happen again.

  • I don't know about your laptop, but I had similar problems with my K6-2 300 (running @333) desktop machine. It's on a FIC board, and running with 128Meg of SIMMS. For the longest time I could never get a make to finish.

    Your problems, though, seem indicative of an overheating CPU--don't you think? You say you run a make and it fails--you try again, it might get a little farther (it's skipping the parts it finished previously), and then dies, right?

    That was my problem, too. My CPU had the original heatsink and FAN from AMD mounted on it. Unfortunately, the heatsink is mounted onto the CPU with some really crappy epoxy that will often just lose its contact with the CPU--so I'll have to go over to the machine and push down on the heatsink to make contact with the chip again. I know, I know...I should just replace the damned epoxy before I push too hard and crack the motherboard :)

    Just an idear, though--dunno if you've got the same stuff.
  • "My hardware is flaky, why do I have trouble compiling?!"

    Because your hardware is flaky. Some cheapie motherboards don't do well. Or maybe your processor is faulty. I've had no trouble with a K6-450 and a K6-500 on linux here. Including compiling all sorts of things.

  • Agreed on the overheated processor. My K6 started flaking out during kernel compiles and Quakeworld games. I found that the CPU fan had died, replaced it and had more more problems.
  • My home server runs a K6-2 @ 266MHz, and I used to have *lots* of problems running Linux on it. make would sometimes spontaneously die (compiling a kernel would always do it), and it would even crash (several nasty kernel-oops'es) once in a while. "su" and other common programs would sometimes segfault, which usually happened right before it would simply just stop taking network connections. I could ping it, but telnet and ssh were dead, and if I were to try to login locally, it would take my username and password, but then just show me the login screen again! And it apparently didn't log anything during that time either.

    My theory was that it was either overheating or that it was memory related. Although the CPU fan was wearing out (it was gradually becoming very noisy), it still kept it really cool, so I dismissed overheating as a possible cause. At the time, the memory was 32MB of 168-pin and 16MB of 72-pin, and that 48MB was horribly inadequate, as it was using it's swap partition heavily.

    It now has 96MB (all 168-pin), a new CPU fan, new hard drive, and a clean install of Mandrake 7.1, and it's working better than ever before. It just made 100 hours of uptime today, which would have been nearly impossible before. I would suspect that 16MB of 72-pin memory, but it seems to be working fine in a Windows machine, so maybe it was my motherboard or something to do with the swapping???

  • I have been running FreeBSD (supposedly a VERY stable OS) for a few months on my AMD K6-2, and I have noticed the same problems. In fact, I can't even recompile the kernel to fix the problem that is specific to my machine, even though there is a fix for it in the FreeBSD kernel source. It is just plain depressing. make has become the most frustrating aspect of *nix for me, simply because my hardware prevents it from working properly.

    BTW, it has to do with some kind of memory hole. I always get signal 11s when trying to make anything.

    I'm gonna try my luck with OpenBSD now, since there's not really anything on the machine I care about...
  • [more info...]

    Also, I can't even install Windows 98, because it will crash during the installation process. My proc is an AMD K6-2 400Mhz.

    Try looking through /usr/src/sys/i386/conf/LINT to find the processor-specific option for fixing this particular problem, if you can get a look at a FreeBSD machine.

  • Maybe hardware (Score:3)

    by levendis ( 67993 ) on Friday November 10, 2000 @05:04PM (#631034) Homepage
    It is possible that its related to hardware. Check out http://www.bitwizard.nl/sig11/ [bitwizard.nl] - grep the page for "K6". If you get a signal 11 (no, not the Signal 11), its almost always CPU or memory related. I have an AMD-K6 350 that exhibits the same symptoms, unless I disable the CPU cache in the BIOS.
    ----

  • Re:It happens to me too! (Score:3)

    by shepd ( 155729 ) <slashdot@org.gmail@com> on Friday November 10, 2000 @08:02PM (#631035) Homepage Journal
    Your problem was probably the mixture of SIMMS (72-pin) and DIMMS (168-pin).

    SIMMS run at 5 volts. DIMMS usually run at 3.3 volts (unless you spent a lot of dough on them, they will be running at 3.3 volts). Running a DIMM at 5 volts will either cause it to burn out, or will cause it to be unstable.

    Running a SIMM at 3.3 volts will certainly cause it to be unstable, but will not cause it to burn out.

    Since I have never, ever, heard of even the most expensive motherboard supporting both SIMMS and DIMMS at the same time (unless they are all 5 volts... which is why some manuals say they can use both at the same time), I would say your board didn't support it.

    So, you were either ruining your DIMMS, or not supplying enough voltage to the SIMMS. Since you say the DIMMS and SIMMS by themselves appear to still work, I'll assume the latter. You're pretty lucky actually... :-)

    [if I'm wrong, please correct me...]

Slashdot Top Deals

An authority is a person who can tell you more about something than you really care to know.

Close