On Friday 05 January 2001 21:47, you wrote:
> Thanks for the info civileme.
>
> If the problem is that the kernel has been patch to include udma100 and
> this causes the problem I have been having, is their a fix????? I would
> love to get Mandrake working again.
> just for info I have the following configuration:-
>
> ide0  udma33  IBM6.4GB        Master
>               hda1    /
>               hda2    swap
>               hda3    /winbackup
> ide1          DVD     /cdrom  Master
>               CDRW    /cdrom1 Slave
> ide2  udma66  Quantum KA9     cable select end of cable1of ultra66 controller
>               hde1    C:      win98
>               hde5    swap
>               hde6    /backup
> ide3  udma66  Quantum CR13    cable select middle of cable1of ultra66 controller
>               hdf2    /oldhome
>               hdf3    E:winbackup
>               hdf5    F:windata
>               hdf6    G:diskcopy
> cdrom Master end connector of cable2 of ultra66 controller
> scsi  IBM 2.2GB
>               sda1    swap
>               sda2    /home
> Thanks for any input or advice
> Paul
>

That does not cause the problem per se.  It seems manufacturers of hdds, 
particularly ides, discovered that the broad tolerances allowed by most 
software made it possible to relax QA and sell out of spec drives.  Now with 
their newer ATA/66 and ATA/100 tech, they are supplying drivers that are very 
tolerant, but the linux-ide folks BELIEVED the specs and wrote with that 
target.  

The fix is to hit F1 at the splash screen and type 

expert linux idebus=33

--just a workaround....  Of course avoid selecting hd optimisation and put an 

idebus=33 in the append statement blank of every single bootable linux on the 
system.

You can alter this, once you are up, by patching /etc/rc.sysinit with hdparm 
statements right after the paths are defined (somewhere in the first 10 
lines) that slow down the suspect disks.  

Mostly this has been WD disks but some others have also failed.  And at least 
one has been reported to fail on kernel 2.4.0....  This appears to be a 
hardware problem as well because plain vanilla unpatched 2.4 is showing write 
errors on some DMA 33 drives where plain 2.2.18 kernel-linus did not.

The fix this time probably has to be to start returning out of spec 
merchandise.  It might be possible for two kernels to be on your boot menu, 
one for good hardware and one for garbage, but they might be for different 
versions of the distro, or have other services changed.

It should(tm) be possible to simply replace a 7.2 kernel with 7.1's version 
or to use the kernel-linus shipped with 7.2   For the method to do that, 
check here:

http://www.mandrakeforum.com/article.php3?sid=20001206025807

Only the portion relating to the copying of kernel info to /boot need concern 
you.

Civileme


Civileme

Reply via email to