Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!apple!netcom!hinton
From: hinton@netcom.UUCP (Greg Hinton)
Newsgroups: comp.os.minix
Subject: Re: mkfs bombing out
Summary: mystery solved
Message-ID: <1709@netcom.UUCP>
Date: 9 Jul 89 03:14:20 GMT
References: <1695@netcom.UUCP>
Reply-To: hinton@netcom.UUCP (Greg Hinton)
Organization: NetCom -- San Jose, CA
Lines: 36

In article <1695@netcom.UUCP> I write:
> ....
>I've run across a problem with
>mkfs: on partitions 2 - 4 (/dev/hd[2-4]), mkfs crashes if the file system size
>is greater than about 7200 with the message:
>	Error: put_block couldn't write
>	Line 1 being processed when error detected.
>
>Mkfs works just fine, however, on the first partition (/dev/hd1) with any
>file system size.
> ....

It's the little things that can drive you bonkers!

I assumed that since at_wini.c used an intelligent method of determining
the number of heads & sectors on a disk, that utilities like mkfs & fsck
would do the same.  I forgot that these programs must run on non-AT machines
& therefore have those values hard-coded into them.  Once I changed that,
everything ran smoothly.

The first partition wasn't affected earlier because the sector offset into
it is always zero.  But the second partition was actually overlapped with
the first.  I'm lucky I didn't wipe out any DOS stuff!

Are these the only two files that hard-code disk parameters, or are there
other gotcha's lurking out there?

Also, does anyone have a list of which commands require the -i (separate
I/D) switch when compiled?  If so, please post.

Thanks.

Greg Hinton
INET: hinton@netcom.uucp
UUCP: ...!uunet!apple!netcom!hinton
"I've never let school get in the way of my education."