2.5.31 hda: lost interrupt

2.5.31 hda: lost interrupt

[Kees Bakker]

With 2.5.31 I am getting
 hda: lost interrupt
2.5.30 was booting OK (but had some other problems).

My machine has a MSI K7T266 Pro motherboard with Athlon 1.3GHz. It has a
VIA chipset, 82C686b+VT8233. Harddisk: IBM Deskstar 60GXP, 40Gb.
-- 
**************************************
Kees Bakker
Senior Software Designer
Altium - Think it, Design it, Build it
Phone  : +31 33 455 8584
Fax    : +31 33 455 5503
E-Mail : Kees.Bakker@altium.nl
URL    : http://www.altium.com
**************************************
Computing is a terminal disease.....

Re: 2.5.31 hda: lost interrupt

[Irwan Hadi]

On Mon, Aug 12, 2002 at 09:07:07AM +0200, Kees Bakker wrote:

> With 2.5.31 I am getting
>  hda: lost interrupt
> 2.5.30 was booting OK (but had some other problems).
> 
> My machine has a MSI K7T266 Pro motherboard with Athlon 1.3GHz. It has a
> VIA chipset, 82C686b+VT8233. Harddisk: IBM Deskstar 60GXP, 40Gb.

Well on my machine, with Maxtor DiamondMax 40 and Asus A7A255 ->
AliMagic chipset, and with kernel 2.5.26 I was having the same problem
too.
It seems the problem might be because I was using ext3fs, which soon I
found out corrupt the filesystem because of this lost interrupt thing.
Or this problem might occur because my system is an AMD Athlon.

My solution was to move back to ext2fs and kernel 2.4.18, although for
this I needed to fsck the hard drive a couple times because of the
occured corruption to the filesystem.

Re: 2.5.31 hda: lost interrupt

[Alan Cox]

On Mon, 2002-08-12 at 16:31, Irwan Hadi wrote:
> Well on my machine, with Maxtor DiamondMax 40 and Asus A7A255 ->
> AliMagic chipset, and with kernel 2.5.26 I was having the same problem
> too.
> It seems the problem might be because I was using ext3fs, which soon I
> found out corrupt the filesystem because of this lost interrupt thing.
> Or this problem might occur because my system is an AMD Athlon.

It happened because you ran 2.5. IDE on 2.5 is not stable (especially on
2.5.26)

> My solution was to move back to ext2fs and kernel 2.4.18, although for
> this I needed to fsck the hard drive a couple times because of the
> occured corruption to the filesystem.

ext3 is stable on 2.4 systems. 

Re: 2.5.31 hda: lost interrupt

[Kees Bakker]

>>>>> "Alan" == Alan Cox <alan@lxorguk.ukuu.org.uk> writes:

Alan> On Mon, 2002-08-12 at 16:31, Irwan Hadi wrote:
>> Well on my machine, with Maxtor DiamondMax 40 and Asus A7A255 ->
>> AliMagic chipset, and with kernel 2.5.26 I was having the same problem
>> too.
>> It seems the problem might be because I was using ext3fs, which soon I
>> found out corrupt the filesystem because of this lost interrupt thing.
>> Or this problem might occur because my system is an AMD Athlon.

Alan> It happened because you ran 2.5. IDE on 2.5 is not stable (especially on
Alan> 2.5.26)

It happened because of 2.5.26, not just 2.5 in general.

Last night I changed a few things in the 2.5.31 configuration and now it
boots OK.
1) switched over to ACPI (was using APM before)
2) enabled CONFIG_IDEDMA_PCI_AUTO (which I didn't do before because I
   wasn't sure about the "VIA V2" warning in relation to my VIA chipset)
3) removed ATA_F_NOADMA from the chipset flags in ide-pci.c (chip id:
   PCI_DEVICE_ID_VIA_82C586_1)

I get the feeling that it is all DMA related, especially with my VIA
chip. I haven't seen any evidence of corruption, if that's what you mean by
'not stable'.

>> My solution was to move back to ext2fs and kernel 2.4.18, although for
>> this I needed to fsck the hard drive a couple times because of the
>> occured corruption to the filesystem.

Alan> ext3 is stable on 2.4 systems.

And on 2.5 systems, right?
-- 

Re: 2.5.31 hda: lost interrupt

[Alan Cox]

> >> My solution was to move back to ext2fs and kernel 2.4.18, although for
> >> this I needed to fsck the hard drive a couple times because of the
> >> occured corruption to the filesystem.
> 
> Alan> ext3 is stable on 2.4 systems.
> 
> And on 2.5 systems, right?

Thats tricky to answer until all the other bugs in 2.5 are fixed, since
any one of them could turn out to be ext3. It should work, but 'stable'
isnt a word I'd apply to the development tree at least in part because
there is so much new code that hasnt been beaten on hard