[gentoo-user] mysterious segfaults

[gentoo-user] mysterious segfaults

[matthew.garman]

Recently, programs on my computer have been victims of abrupt
segfaults.

For example, yesterday, I tried to start firefox, and before it even
displayed its window, it died with a segfault.

Even portage was a victim of this: trying to emerge anything, or
even do an "emerge search" resulted in an almost instant segfault.

The weird thing is, after waiting a while (say two or three hours),
the problem went away---everything just started working.

Last night, my email client, mutt, suddenly started having this
problem: I could use it, up until switching to the "gentoo-user"
folder, when it would crash with a segmentation fault.  X was hung,
so I rebooted.  Now I can get into my gentoo-user folder... but for
how long?

All of my system partitions (/, /usr, /var, /tmp) reside on an ext3
filesystem; I ran e2fsck on all partitions.  I've also run "smartctl
-t long" on the disk---no problems reported.  So I don't think my
hard drive is dying.

Anyone else ever experience anything like this?  Anyone have any
thoughts as to what the problem might be?

Thanks!
Matt

-- 
Matt Garman
email at: http://raw-sewage.net/index.php?file=email
-- 
gentoo-user@gentoo.org mailing list

Re: [gentoo-user] mysterious segfaults

[Holly Bostick]

matthew.garman@gmail.com schreef:
> Recently, programs on my computer have been victims of abrupt 
> segfaults.
<snip>
> The weird thing is, after waiting a while (say two or three hours), 
> the problem went away---everything just started working.
> 

This sounds awfully like a heat problem, especially if the problem goes
away after waiting a significant amount of time (the system cools down),
or only appears after a long time of use (the system heats up).

I did have a similar problem (but I knew it was heat, because I have the
gkrellm temp monitors displayed), and it was so bad that I could not
even run the computer without an external fan pointing into the box (the
sides have been off all the time). If I turned the fan off, I got
immediate segfaults, and I was unable to even boot the machine. Even
with the fan, it was running *extremely* hot (I'm talking about 70 to
over 100 degrees Celsius).

Finally, I realized that I was insane to even attempt to run like that,
and took the machine apart, to discover that the CPU fan was absolutely
clogged with dust and "stuff" (I have a cat, and her hair gets into
*everything*. A LOT of hair).

Cleaned out the CPU fan, and suddenly the box was running at like 48 degrees
C. Without the external fan.

It could be some kind of intermittent hardware issue in your case, but
heat is easy enough to check, and very well might be the problem.

HTH,
Holly
-- 
gentoo-user@gentoo.org mailing list

RE: [gentoo-user] mysterious segfaults

[Michael Kintzios]

> -----Original Message-----
> From: matthew.garman@gmail.com [mailto:matthew.garman@gmail.com] 
> Sent: 02 March 2006 13:29
> To: gentoo-user
> Subject: [gentoo-user] mysterious segfaults
> 
> 
> 
> Recently, programs on my computer have been victims of abrupt
> segfaults.
[snip...]
> Anyone else ever experience anything like this?  Anyone have any
> thoughts as to what the problem might be?

The random nature of your segfaults probably points to an overheating
problem.  I am saying that because if you waited for a while before the
segfaulting disappeared, the particular device (CPU, hard drive, memory)
cooled down and was able to function again.  Easy to test this
hypothesis with a domestic comfort cooling fan blowing towards an open
case.  More difficult to find out what particular device overheats.  A
problematic cooling fan (CPU, PSU, case) would usually become noisier as
its bearings are drying out (a drop of oil will provide an instant fix).
So would a hard drive.

If the application of a domestic cooling fan does not relieve the
problem, then it could well be faulty memory module(s), or a faulty
power supply.
-- 
Regards,
Mick

-- 
gentoo-user@gentoo.org mailing list

Re: [gentoo-user] mysterious segfaults

[matthew.garman]

On Thu, Mar 02, 2006 at 02:23:17PM -0000, Michael Kintzios wrote:
> If the application of a domestic cooling fan does not relieve the
> problem, then it could well be faulty memory module(s), or a
> faulty power supply.

I'm afraid it's a random hardware failure.  I've been running
cpuburn for the last couple hours.  According to sensors, my cpu has
reached a max temp of 57 degress C.  No segfaults thus far.

It's been several months ago, but I did run about eight hours of
memtest86 on the memory.  Is it unusual for memory to work fine for
a while and *then* go bad?

I might try a new power supply anyway.

For what it's worth, mysterious problems on this box have come and
gone for probably a year now.  Every time something comes up, it's
so random that I don't even know where to start looking.  I'm this
-> <- close to building a whole new PC :)

Thanks again,
Matt

-- 
Matt Garman
email at: http://raw-sewage.net/index.php?file=email
-- 
gentoo-user@gentoo.org mailing list

RE: [gentoo-user] mysterious segfaults

[Michael Kintzios]

> -----Original Message-----
> From: matthew.garman@gmail.com [mailto:matthew.garman@gmail.com] 
> Sent: 02 March 2006 15:54
> To: gentoo-user@lists.gentoo.org
> Subject: Re: [gentoo-user] mysterious segfaults
> 
> 
> On Thu, Mar 02, 2006 at 02:23:17PM -0000, Michael Kintzios wrote:
> > If the application of a domestic cooling fan does not relieve the
> > problem, then it could well be faulty memory module(s), or a
> > faulty power supply.
> 
> I'm afraid it's a random hardware failure.  I've been running
> cpuburn for the last couple hours.  According to sensors, my cpu has
> reached a max temp of 57 degress C.  No segfaults thus far.
> 
> It's been several months ago, but I did run about eight hours of
> memtest86 on the memory.  Is it unusual for memory to work fine for
> a while and *then* go bad?
> 
> I might try a new power supply anyway.
> 
> For what it's worth, mysterious problems on this box have come and
> gone for probably a year now.  Every time something comes up, it's
> so random that I don't even know where to start looking.  I'm this
> -> <- close to building a whole new PC :)

No two PC's/MoBos are the same, but FWIW here's a bed time story:  I had
three incompatible memory sticks on mine which kept failing at random.
MEMTEST86+ did not show any errors.  Occasionally, a simple emerge
--sync was enough to crash the machine and needless to say all these
repeated crashes had started to corrupt my fs.  Running out of ideas I
decided to start removing memory sticks until I discovered that the best
result (in terms of stability) could only be arrived at if I left only
one 256M stick of branded memory in the box.  That was despite the fact
that the MoBo manual said you could mix 'n match memory modules without
any adverse effect on performance...  :p

In case you're suffering from the same problem, check whether a crash is
more likely if them machine is about to switch to the next bank of
memory/swap.  Mine invariably crashed most times it was getting ready to
swap data to the hard drive, or in any case as soon as it had used all
the memory on the first stick.  What drove me insane with this fault was
that it would only crash once and thereafter it was often good until the
next reboot.  Also, if the transition from the first memory stick to the
second or swap space, was caused by an application engaging in
aggressive resource usage (e.g. Opera loading font files when it hits a
website with Chinese characters) a crash was guaranteed.  Slowly
building up to it during a large emerge session would not cause any
crashes.

I hope this helps.
-- 
Regards,
Mick

-- 
gentoo-user@gentoo.org mailing list

Re: [gentoo-user] mysterious segfaults

[b.n.]

> It's been several months ago, but I did run about eight hours of
> memtest86 on the memory.  Is it unusual for memory to work fine for
> a while and *then* go bad?

Mmm. No.
Mine did exactly so.

> I might try a new power supply anyway.

A faulty PS would give you odd hw-related (mostly cd-related) errors in 
logs, random freezes and random shutdowns... I think it's unlikely to 
give segfaults. however...

> For what it's worth, mysterious problems on this box have come and
> gone for probably a year now.  Every time something comes up, it's
> so random that I don't even know where to start looking.  

I bet it's a faulty mobo...

m.
-- 
gentoo-user@gentoo.org mailing list