Date: Thu, 6 Jul 1995 20:53:56 +1596657 (MET DST) From: Wilko Bulte <wilko@yedi.iaf.nl> To: terry@cs.weber.edu (Terry Lambert) Cc: davidg@root.com, freebsd-hackers@freebsd.org Subject: Re: Stabikity/Usability of 2.0.5R Message-ID: <199507061853.UAA01585@yedi.iaf.nl> In-Reply-To: <9507052305.AA06876@cs.weber.edu> from "Terry Lambert" at Jul 5, 95 05:05:16 pm
next in thread | previous in thread | raw e-mail | index | archive | help
> Probably also an AT&T/NCR machine?
No. It is a 'genuine' (ugh) Intel 302 board. So, 386/25 and 64k cache.
> > BIOS? Mine doesn't. Disabling the BIOS of the WD and selecting the
> > untranslated geometry in the system BIOS also gives me the
> > n * 35 badblocks.
>
> I can't select anything in BIOS. The ESDI drives are, by default, not
> translated anyway. It's only BSD that wants to assume a non-default
They are if they are > 1024 cyls and things like DOS etc are to be used/
booted. The BIOS on my WD1007 allows different translations.
I just found my 'Preliminary engineering specification for WD1007A',
dated Oct 2, 1987. Of any use ??
> geometry for the things. These were built as SVR4 boxes in the first
> place; it was never expected that there would be DOS & therefore no
> DOS correction is necessary.
Mine was not for SVR4, although the company I worked for (Philips
Information SYstems) sold them as SVR3 servers. That worked
perfectly BTW.
> > I have a faint recollection that I had to tell 1.1.5. to use 34
> > sectors (when the drive actually has 35) before it wanted to install.
> > Telling 2.05R to use 34 does not work.
>
> This is critically strange anyway, in either case: there should be no
> knowledge of C/H/S values in any case in BSD once it's up; they are
> simply a convenience for any DOS partitions and DOS partition table
> crap that happens to be there.
>
> The odd thing is that it dies every 35th, so it seems that the head is
> used as a multiplier and pushes it past the end of the disk. One would
> expect that that behaviour would cause normal accesses to skip all over
> the disk (hence the "drive chatter" reports). The point being that if
> head 1 worked and head 17 worked, then head 35 sould work too. Clearly,
> what's happening is that the head multiplier is more significant in
> determining the sector offset than it should be.
OK. From the manual mentioned above:
" The WD also supports high cap ESDI drives, with more than 1024 cyls,
15 heads and at least 35 physical sectors per track (34 user, 1 spare).
For this type of drives the following algorithm is used:
x = 63 * (LH mod 8) + LS - 1
PS= x mod 34 + 1
PH= x div 34
PC= LC * 2 + (LH div 8)
with PX is physical (sector, head, cyl)
LX is logical ( " )
This algorithm is based on the fact that a physical cylinder contains
510 (34 * 15) physical user sectors. Using half the number of phys cyls
as the log cyl gives a sector count of 1020 per log cyl. Dividing
by 16 yields 63.75 sectors per track.
In the translate mode, the spare sectors will not be accesible by the
host (!!! WKB) and will require the firmware top check the requested cyl, head
and sector values before doing the translation.
Translated mode is invoked by issuing a set params (91hex) at 17 or 34,
which is performed by the DOS Bios (??? WKB) at boot time. This option
may be disabled if jumper W14 is installed. W8 if jumped and W14 open
enables WD1005-WAH compat mode, which employs only the double translation (
1 phys track is 2 log tracks). Other operating systems may like to utilise
the ESDI drives in a pure physical manner"
The (!!! WKB) and (??? WKB) were added by me. If I can improve insights
by copying the manual and snailmailing it to someone please tell me so.
Might take some time (transatlantic).
It appears to me that the card reports 35 sectors/tr of which 34
are data and 1 is inacessible (see (!!! WKB)). This might explain why
n * 35 is a show stopper. Also, this seems to correlate with me
using 34 sect/tr on 115R.
I just tried installing W8 and removing W14. Now the 205R probe sees
36 sect/tr.
> Well, I have three more days, and I'll be away from the equipment yet
> again, so if anyone wants a diagnostic run on their code, now is the
> time to get it if you want it done by me. Otherwise Wilko will have
> to be your remote pilot.
Is fine with me. Expect some delay because my ftp capability is
only at work.
> Terry Lambert
> terry@cs.weber.edu
Cheers, Wilko
_ __________________________________________________________________________
| / o / / _ Wilko Bulte email: wilko@yedi.iaf.nl
|/|/ / / /( (_) Private FreeBSD site - Arnhem - The Netherlands
--------------------------------------------------------------------------------
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199507061853.UAA01585>