Frederick Bruckman  (fb@enteract.com) wrote:

>On Tue, 12 Oct 1999, Bob Nestor wrote:
>
>> Frederick Bruckman  (fb@enteract.com) wrote:
>> >The way I read Tommy Smith's message, the _default_ Sector Size was
>> >256. I seem to recall Mkfs_1.45 coming up with odd defaults for the
>> >other paramaters, but never anything but 512 for the Sector Size. So
>> >is that the case, that Mkfs produced a _default_ Sector Size of 256,
>> >or is this simply a case of pilot error?
>> 
>> I didn't change anything in that section of Mkfs for the 1.46 update, so 
>> unless the new CodeWarrior Pro 5 compiler generated bad code this time 
>> around this shouldn't be any different between 1.45 and 1.46.
>
>Maybe this is an old problem. The poster who started this thread
>solved his problem by reverting to Mkfs_1.45, but that was something
>different.
>
Yes, I think this is an old problem.  So far I've only seen one or two 
people post problems they encountered with Mkfs 1.46 but haven't gotten 
anything solid I can work on.  I have gotten a number of private e-mail 
reports that Mkfs 1.46 worked fine and that it solved the problem that 
1.45 has with Linux formatted disk partitions.  At this point I'm just 
going to assume that there's either no interest in solving this problem 
in 1.46 or it isn't really a problem.

Besides, I see that the next version of Apple's Disk Setup that comes 
with MacOS 9 is supposted to allow one to setup UNIX partitions.  Maybe 
it will replace Mkfs for us.

>> It's been some time since I looked at that section, but as I recall 
>> it gets this value from the drive itself by doing a Mode Sense. 
>> That's why some have reported in the past that they were unable to
>> use certain disks with NetBSD because they were originally
>> formatted for block sizes other than 512.
>
>I seem to recall there's some kind of trickery involved in getting the
>geometry with a Mode Sense. You get different values depending on how
>you've set the "Notch" first: notch 0 for the whole drive synthetic
>geometry, notches 1 - ? for actual values. FWB Hard Disk Toolkit
>returns the notch 0 value by default; maybe Mkfs does it different,
>i.e. _wrong_?
>
As I recall there are four different sets of values, Default, Active, 
Current, and one other that seems to be media related.  When I played 
with them on all the disks I have I wasn't able to find any differences 
between the four sets.  At one time I thought the removable media disks 
like the Zip would return different values and worked on that trying to 
solve the Zip problems we had at the time.  That didn't pan out and I 
instead implemented a solution suggested by Bill Studenmund which was to 
pick a set of default values that was more or less universal for all SCSI 
devices.  That's what both 1.45 and 1.46 do now, although neither of them 
screw with the block size.

-bob
>> MacOS seems to be able to deal with this though.