Re: ext3fs errors with kernel 2.6.18 but not with 2.4.27
On 8/3/07, Brad Sawatzky <email@example.com> wrote:
> Hi Francois,
> I agree with Doug: CRC errors shouldn't be ignored. At _best_ they are a
> sign that something in your system is marginal. At worst you end up
> reading and/or writing bogus data. The fact that the errors persisted
> after you changed hard drives suggest either a bad cable (most likely), bad
> secondary device on that cable, or bad motherboard (unlikely).
I tried to change the IDE cables (I had a few spares from my father's computer).
I happened to notice that I had a 40 pins cable. I replaced it with a 80 pins
cable. I have two, I tried both: nothing changed. I still have CRC errors. BTW,
there is only one drive on that cable.
However, I noticed that the motherboard's chipset's fan is dead. I knew it was
having a hard time spinning lately, but now, it's gone. Would it be the
motherboard's chipset that causes those errors because it's too hot?
> > I will try to see tonight or tomorrow if I can manage to get some logs as
> > Douglas suggested. I will also try to look at what driver is in use also.
> > Looking at my config-2.6.18 file:
> > http://www-etud.iro.umontreal.ca/~duranlef/linux-config/config-2.6.18
> > it's hard to guess. All I know is that all SATA support is disabled.
> Your config file says that you're using the 'old' (stable) IDE driver, not
> the newer PATA drivers. That's (probably) good.
> FWIW, you could try booting with the kernel option 'hda=autotune" or
> "idea=autotune" and see what happens. That should allow the driver/chipset
> to fall back to a slower PIO mode if it sees CRC errors. (Though I don't
> know what, if anything, it will do if DMA is enabled...)
Well, once in a while, the kernel resets ide0, and then DMA is dropped (and I
force it back on afterwards), which I guess is not a good idea.