JIM HOLSONBACK wrote in a message to ROY J. TELLASON:

 JH> Hello, Roy.  I overheard you telling Tom Walker about  - -

 RJT> Tom Walker wrote in a message to Wayne Chirnside:
 JH> 

 -> The card I'm looking at is made by Siig and specifically mentions
 -> Win 98 and Linux is supported, 35 bucks.

 JH> OUCH!  I wonder about the advisability of investing in a $35 card
 JH> for a system with mainboard worth prolly less than $5 in order to
 JH> jump up from UDMA33 to UDMA133 capability.  I think I'd be for
 JH> saving up for investing in a newer/better/faster mainboard.

That's Wayne,  not me.  I don't think I'd be doing that...

 TW>  I have a Siig ATA-100 Card that breaks the BIOS limit also. Since
 TW> it is an older card it only handles 128 Gig and smaller.
 TW> Interestingly it only has Win95/98/98Se/WinNT/Win2000 support.  NO
 TW> Linux

 RJT> That's because linux doesn't _need_ such cards to work with large
 RJT> drives in older systems...

 JH> Ah, so Wayne doesn't need to spend $35 on such a card at all, just
 JH> like Windows users don't need such cards, as long as they use the
 JH> HDD Mfgr's supplied work-around software, (like MaxBlast) - right??

Is that what maxblast does?  I thought so,  but wasn't sure if it was for
that purpose and the other package was for the diagnostics...  I've gotten
them both but haven't fired up the w98 box to get them unpacked yet,  since
I've worked around the problem with the drive,  it's on _this_ box,  and I
don't want to start on that mess when I have all this other stuff coming
up...

But yeah,  it was my impression that the card wasn't needed either,  unless
you got it for just the speed increase.

 RJT> You just need to arrange things so that the system sees the drive at
 RJT> all,  and can read some of it,  put the kernel in a small /boot
 RJT> partition at the start of the drive,  and once it boots,  the BIOS is
 RJT> irrelevant.

 JH> Just like HDD mfgr's hardware and BIOS -tricking overlay software,
 JH> right? I have a few questions - -

 JH> 1) In what way are Linux software BIOS limitation workarounds 
 JH> superior to the BIOS limitation workarounds which software like
 JH> MaxBlast provides? I know I don't want any more of those BIOS
 JH> limitation workaround "HDD overlays" for DOS/Win, and don't think
 JH> I'd want any of them If I were to give Linux a try, either.

I'm not real clear on how those work,  except that I remember reading about
one where it kept the actual partition info somewhere a bit further in than
the usual place,  and software that was loaded at boot time
"fixed" things to work with this.  If you were using one of those
for linux you'd need to deal with things during the install process so as
to use it,  rather than letting linux do its own thing.

One of the things I really like about linux is that unless you're talking
really weird hardware that has some special setup considerations,  you can
take a drive that was set up on one system and plug it into something else
and it'll work.  I did exactly that one time,  when I suspected some
hardware difficulties early on,  where I took the pair of HDs that I'd been
using with a P200 and plugged it in to a 486/66.  Not having shut the
system down properly could have had something to do with me needing to run
a filesystem check,  which took a *lot* longer than it would have on the
other MB,  but it ran,  and the system booted, and that was that.  None of
this "finding new hardaware" nonsense and the subsequent loading
and install of a bunch of drivers,  etc.  The second drive was (and still
is!) a 6.4G,  and I have no idea of what the 486 thought of that, nor did
it skip a beat.

 JH> 2) If there is a HDD problem,  HDD can't be properly recognized by
 JH> BIOS, and system won't boot, what software do you use to boot into
 JH> and do diagnostics and repair to the HDD when you're running LInux?

This a little gets complicated.  If the system won't see the drive at all
then it won't see it,  and won't boot it.  It may be possible in such a
circumstance to still boot from a floppy and then go on from there.  Seems
pretty likely to me,  in fact.  If there's a problem,  and portions of the
filesystem get trashed,  fsck (under various names like e2fsck, etc.) will
be invoked during the boot process and will deal with it from there, 
unless it's too messed up to do so in automatic mode,  in which case you
get prompted for the admin password and told you need to run it manually. 
If you have the sort of problem where the system doesn't want to see the
drive at all,  or can't read it properly after booting from a floppy,  then
you're getting into some territory that I haven't explored that much just
yet.  :-)

 JH> 3) When mainboard BIOS will only support UDMA33,  how fast do
 JH> newer/faster UDMA133 drives run under Linux?

I'm not real up on this stuff.  If the BIOS will only support that is it
because of some aspect of the hardware?  If not,  then maybe it'll go
faster, I'm not real sure.  Looking at the bootup messages on the screen to
my left with dmesg | less,  I can guess at what might be applicable to
this.  There's a line that says "ide: Assuming 33MHz system bus speed
for PIO modes; override with idebus=xx" which I'm guessing would be a
boot-time parameter.  Then some stuff where it finds the chipset,  says
"VP_IDE: not 100% native mode: will probe irqs later".  It
correctly IDs the drives in the system (at the moment :-) as:

hda: FUJITSU MPD3043AT, ATA DISK drive
hdb: ST36531A, ATA DISK drive
hdc: CENDYNE 481648AX, ATAPI CD/DVD-ROM drive
hdd: CREATIVE CD2422E, ATAPI CD/DVD-ROM drive,

then finds the IRQs on both channels,  and the i/o addresses.  Specifies
the amount of "queue" (buffer?) on each drive,  looks for
partitions,  loads the SCSI subsystem (which I have to use for the burner),
 loads the "md" (multiple device) subsystem,  which is in the
kernel by default in this setup,  does kernel,  network,  filesystem, 
soundcard,  PnP,  and SCSI (1542) stuff.

So anyway looking at that "idebus=xx" bit,  it looks to me like
this setup defaults to 33 MHz but can easily be overridden with the right
parameter passed to it if the hardware was capable of more performance. 
Hmm. Interesting stuff! I stick one of those cards in there I could see a
heck of a speed boost in disk performance,  is what it looks like. 
Particularly when the system starts to swap,  which is where I'm running
into problems right now.  That does tend to bog things down a bit.

It seems that there are a number of the folks from the linux echo in here,
lurking if nothing else.  I'm sure that if I've messed this up too badly
they'll jump all over it.  

---