From: duperval@ERE.UMontreal.CA (Duperval Laurent) Subject: Re: FAQ addition: (was Re: Badblocks revisted (**sigh**)) Date: 18 Jul 1992 20:00:18 GMT
In article <1992Jul17.213244.4490@athena.mit.edu> Graham@DOCKMASTER.NCSC.MIL writes:
>|From: u31b3hs@cip-s02.informatik.rwth-aachen.de (Michael Haardt) writes:
>|Subject: FAQ addition: (was Re: Badblocks revisited (**sigh**))
>|Date: 16 Jul 92 08:21:05 GMT
>|
>|duperval@ERE.UMontreal.CA (Duperval Laurent) writes:
>|> [I have badblocks, how can I mark them?]
>|
[ Stuff about fixing with Minix deleted ]
>|Linux does not mark bad blocks, so you have to do that. It's not really
>|difficult, it is just that you need MINIX to do the job. Don't yell at
>|me ...
>
>This was definitely true in 0.12 but is this still true? The reason I ask
>is because when I run "mkfs -c /dev/hda2" It spits out errors on each
>bad block it finds, and you can even see it go back and retry. But if
>you let it run, you eventually have a .badblocks file. I don't think
>0.12 created this, but bootimage-0.96c and rootimage-0.96 do.
>
>If Linux doesn't mark badblocks then what is this file?
OK! I now have the (almost) definitive answer to my original post. The
problem with bad blocks isn't inherent to Linux. It's a bug (or an oversight)
in the mcc-interim installation package.
When install_root asks 'Do you want to replace all files' you have to say 'no'
if you want to keep your badblocks from being written to. The installation
package does a 'rm -rf' on all the directories it creates. therefore, any
information written by the mkfs command will be lost. I'm sure of it because
I just finished reinstalling Linux, from a clean disk and my .badblocks files
are still there.
So, if the person in charge of the mcc-interim package is listening, here are
my recommendations ;-) (and this goes to anyone who plans on releasing an install
package):
1- DON'T DO A 'rm -rf' ON A DISK. It's very dangerous and leads to unexpected
problems like the one above.
>Please post a reply here, because I have seen questions about this
>before, but I have never seen a public reply, and I'm sure there
>are other people out there that don't have IDE or SCSI drives.
>Thanks,
>
>Pat Graham
>Graham@Dockmaster.ncsc.mil
If my assumptions are correct, this problem would have nothing to do with the
type of drive used.
I'd like to thank the person who pointed this out first (sorry, I forgot your
name). Thank to you, I'll probably be able to compile without getting all
those I/O errors and those kernel panics!
-- Laurent Duperval duperval@ere.umontreal.ca duperval@jsp.umontreal.ca