NetNews Usenet Archive 1992 #20/NN_1992_20.iso/spool/comp/os/linux/10310

Text File | 1992-09-09 | 2.7 KB | 84 lines

Newsgroups: comp.os.linux
From: [e-mail] (David Metcalfe)
Path: sparky!uunet!pipex!demon!prism.demon.co.uk!david
ReplyTo: [e-mail]
Subject: Re: 0.97pl4 breaks SCSI extended partition code
References: <[e-mail]>
Distribution: world
X-Mailer: cppnews $Revision: 1.15 $
Organization: ** none **
Lines: 69
Date: Wed, 9 Sep 1992 20:41:24 +0000
Message-ID: <[e-mail]>
Sender: [e-mail]


In article <[e-mail]> [e-mail] (Eric Youngdale) writes:

> In article <[e-mail]> [e-mail] (Chris Newbold) writes:
> >Okay. This looks like a generic problem with the SCSI extended partition code.
> >I have seen several senarios which result in the problems described above.
> >
> >    1. Adding the CD-ROM filesystem extentions did it for me.
> >    2. Compiling the kernel under GCC v2.2.2d did it for some
> >    3. Now, looks like pl4 has blown it in this case.
> >
> >I was trying to work with Eric (the author of the CD-ROM stuff) to find
> >out what goes wrong, but my disk controller flaked out and I haven't
> >had a machine for the last week...
>
>     I found another guinea pig, and it appears as if we are closing in
> on the solution.  What I do know is that make_request in ll_rw_blk.c returns
> because it believes that the read is beyond the end of the partition.
> Specifically, the problem shows up in the following segment:
>
>     count = bh->b_size >> 9;
>     sector = bh->b_blocknr * count;
>     if (blk_size[major])
>         if (blk_size[major][MINOR(bh->b_dev)] < (sector + count)>>1) {
>             bh->b_dirt = bh->b_uptodate = 0;
>             return;
>         }
>
> As of yet, I do not know whether the blk_size array is incorrect or if sector
> or count is corrupt, or whether we are not converting something somewhere.
> I hope that within a day that we will have enough information to squash this
> bug.
>
I have now confirmed that my problem occurs in exactly the same
place.  The values of the key variables with pl4 (the one that fails)
are

blk_size[major][MINOR(bh->b_dev)] = 0
sector = 0
count = 2

When I repeat the same exercise with pl2 (which works), the values of
the variables on the first pass through the routine are

blk_size[major][MINOR(bh->b_dev)] = 507905
sector = 0
count = 2

So the problem would appear to be with the blk_size array.  I will
leave someone else to work out what the problem is because I'm not
really a kernel hacker :-)

> -Eric
>
>
> --
> Eric Youngdale
> [e-mail]
>
>

David

-----------------------------------------------------
David Metcalfe          [e-mail]
[phone]         [e-mail]