>> i didn't see anything related to scsi here: 
>> [...]
>
>Yes, I've seen that there was no SCSI implementation; my grumbling
>was more about the limited userland interface.
>Darren is still working on it as just seen in the commit list,
>so he'll turn it into something better...

ah, i get it.  okay.

>> i'd like to ask my dying drive which blocks it thinks are
>> toast, so that i can set about finding out which files are zorched
>
>In the cases I've experienced, IDE disks did die rather quickly.

i can use up a laptop drive in about a year.  sometimes less.

>The drive failed completely before I had a chance to recover anything.
>With alpha/OSF1 and SCSI disks I would have had a chance - one gets
>a message when a bad sector is found. Obviously it gets replaced automaticly.
>If the drive's pool of spare sectors is exhausted, it gets serious.
>If I had read the log messages, I'd been warned before:-)

i've been told that modern drives come with space sectors to which bad
ones are remapped on the fly.  the question of whether such activity
is reported back to the operating system is the important point.

at the moment, i'm getting read errors.  those can't really be
remapped very easily, so i just lose.  dumps take forever, and i
typically just give up, intending to attack the problem from a
different angle later.  if there was an easy to access list of blocks
that were "bad" (as opposed to skimming /var/log/messages), i'd
happier.  :)

-- 
|-----< "CODE WARRIOR" >-----|
codewarrior@daemon.org             * "ah!  i see you have the internet
twofsonet@graffiti.com (Andrew Brown)                that goes *ping*!"
werdna@squooshy.com       * "information is power -- share the wealth."