LFS5.0 Chapter 6 GCC pg 123/289 -- FIXED PART 4

Mark Carey markcare at paradise.net.nz
Sat May 8 13:51:06 PDT 2004


Firstly thank you all for your replies.

Secondly gcc now compiles, I removed my 3COM 3C905B TX PCI ethernet adaptor and
the WOL cable and gcc compiled first time.  This was eth1 previously and it had
no cable plugged into its back.  So how this was causing errors I do not know,
but I am now writing from mozilla in fluxbox and thigs seem reasonable stable!!

I am running `make -k check' as I write.

Mr Nobby Wrote:
> Well it won't be the SCSI system reporting the error message as your Zip
> drive is IDE (/dev/hdc). Have you tried unplugging the Zip drive? Does
> that stop the error messages?

Your comments on IDE are interesting my CDROM is IDE-ATAPI but Knoppix (Kernel
2.4.x) insists on using ide-scsi for it, it isnt even a CD writer.  I mave also
seen mention on lkml that it is possible to use ide-scsi with zip drives perhaps
Knoppix does this?

> If you are confident that you have proved the RAM to ok then check all
> your data cables, especially to the HDD.

Cables are good.

> Re-seat the memory just in case, better still if you have more than one
> stick, swap them around, just unplugging/plugging will clean the
> contacts. Make sure the tabs are locked in place.
>
> Look at power problems, either dirty power (i.e. the DC voltage is not
> being regulated properly - may be the PSU is going to die soon), or you
> are overloading your PSU.
>
> What wattage is it and how much power is supplied on all the voltage
> rails? (you can get the info from the label on the PSU). Try paring your
> machine down to the minimum, i.e. Video, HDD only and see if the problem
> still occurs.

It was this suggestion that prompted me to remove the second network card, I
shall dig further but I believe the Enermax supply I have to be sufficient, but
it will not hurt to do the sums.

Ken Moffat Wrote:
> Hmm, it certainly smelt like memory errors in your earlier posting, but
> 12 hours ought to be enough to show /most/ problems in memtest (I've
> seen occasional reports on lkml where much longer testing was needed,
> but my recent experiences with one stick that failed and with trying to
> find "compatible" sticks for different old boxes suggest that if it
> completes a whole pass of the tests then it's probably ok).
>
>  The other problem I get here is on my AmigaOne (not the world's best
> box for running linux at the moment).  With gcc-3.3.3 I count myself
> lucky to complete a full LFS build without internal compiler errors.  I
> suspect a cooling problem (gcc-3.3.3 works it hard, sometimes the top of
> the case is quite warm).  Maybe you too have got a cooling problem.

I will be shortly replacing my CPU cooler, the ball bearings in it rattle like
nothing on earth, I call my computer the meat grinder for the fan noise it makes
while operating.

Tony Sauri Wrote:
> Some observations:
> a) I always understood that running a box with the covers open disturbed
> normal forced cooling airflows and increased the chances of overheating.

I have no was of checking the cpu temperature as I dont have lm_sensors compiled
yet, you may well be right.  I have however noted that the CPU heatsink is
clogged with dust so I will get the vaccum cleaner out later today.

> b) Have you been able to discount a faulty Knoppix CD?

No, but I have alread compiled gcc in Chapter 5 and glibc so would a faulty cd
not show up before now?

THanks for all your help, moving forward.

Mark Carey



More information about the lfs-support mailing list