Whats the problem with my IDE...

Whats the problem with my IDE...

[Manuel Reimer]

Hello,

after my problems with my DVD drive, I've unplugged this device and 
jumpered my burning device as master, to start my linux setup from there.

Now I don't longer get those silly DVD errors, but now I get something, 
which is much more worrying for me:

http://pastebin.com/f6fc7552f

After mounting my four LVs, formatted with XFS, here, I'm starting to 
delete all files on those LVs to be able to do a fresh linux install. 
While deleting, I get *many* DMA errors and "BAD CRC" messages.

Maybe someone has a minute of time to tell me who is responsible for 
this errors. Does the kernel not work very well with my IDE controller? 
Is there a defect in the IDE controller? Is there a problem with my HDD?

Thank you *very* much in advance for any information about this...

Yours

Manuel

Re: Whats the problem with my IDE...

[Alan Cox]

> After mounting my four LVs, formatted with XFS, here, I'm starting to 
> delete all files on those LVs to be able to do a fresh linux install. 
> While deleting, I get *many* DMA errors and "BAD CRC" messages.

Bad CRC indicates a data transfer problem between the drive and the
controller. The CRC is computed one end and verified the other. The OS
isn't directly involved.

This usually means either
#1 A 40 wire cable in use for > UDMA33
#2 A drive plugged into the middle of the cable and none on the end
#3 The short part of the cable connected between a drive and the
controller not the long part

And occassionally it may indicate power, iffy connectors or electrical
noise problems.

Theoretically it can also be caused by the kernel misprogramming the
timings but the old VIA IDE driver is very well tested in this area and
the one bug I know it has with timing it always sets to be excessively
safe not problematic.

Re: Whats the problem with my IDE...

[Manuel Reimer]

At first: Thank you very much for the detailled information!

Alan Cox wrote:

> Bad CRC indicates a data transfer problem between the drive and the
> controller. The CRC is computed one end and verified the other. The OS
> isn't directly involved.
> 
> This usually means either
> #1 A 40 wire cable in use for > UDMA33

Not the case. Yesterday, I even replaced the 40 wire cable, in use for 
the DVD device, with a 80 wire one.

> #2 A drive plugged into the middle of the cable and none on the end

Yes, this was the case. I plugged the drive on the end now, but the 
drive, which was plugged into the middle of the cable, wasn't the one, 
which errors, it was the DVD device (hdc).

> #3 The short part of the cable connected between a drive and the
> controller not the long part

Not the case.

> And occassionally it may indicate power, iffy connectors or electrical
> noise problems.

... don't know about how to find and remove such problems...

I still have those messages, but now there seem to be much less of them.

It would really help me, if someone could tell me about how critical 
such error messages are.

Is this a common problem of IDE connections? May I just ignore those 
messages, if I tried anything to get rid of them, or do they tell me 
"get a new hardware, the old one will fail, soon"? Will my data be 
read/written without mistakes, even with those errors?

Its interesting, that I only get those messages using the Slackware 
install CD, which has many kernel modules built into the kernel.

With the "slimmed down" kernel of slackware, which even needs an initrd 
for ext2 root devices, I don't get this errors, so currently I only get 
them while installing and not while my daily work on my system...

Maybe I'll install the "huge" kernel, with the many "built in" stuff, 
onto my regular Slackware install, tomorrow, to see if this will cause 
those errors.

Thank you *very* much for any help and information!

Yours

Manuel

Re: Whats the problem with my IDE...

[Alan Cox]

> I still have those messages, but now there seem to be much less of them.
> 
> It would really help me, if someone could tell me about how critical 
> such error messages are.

The CRC errors are indicative of cable or similar problems but each time
you see a CRC error the transmission is repeated so it should never cause
corruption. If there are repeated CRC errors it will give up and use PIO
instead.

> With the "slimmed down" kernel of slackware, which even needs an initrd 
> for ext2 root devices, I don't get this errors, so currently I only get 
> them while installing and not while my daily work on my system...

Could be one driver is upsetting another, or perhaps triggering more
hardware which in turn triggers this. You'd really need to know what the
extra driver which triggers the problem is ?