Re: fc5 seek errors - nash mount (built-in)

On Thu, 2006-03-30 at 17:31 -0500, James Olson wrote:
> > ----- Original Message -----
> > From: "Peter Jones" <[email protected]>
> > To: "ATARAID (eg, Promise Fasttrak,  Highpoint 370) related discussions" <[email protected]>
> > Subject: Re: fc5 seek errors - nash mount (built-in)
> > Date: Thu, 30 Mar 2006 11:22:39 -0500
> > 
> > 
> > On Thu, 2006-03-30 at 06:32 -0500, James Olson wrote:
> > 
> > > Thanks Peter! That does make testing easier (don't have to
> > > rename /bin/mount) Smiles!
> > >
> > > I came up with a patch to nash/block.c that seems to fix the problem
> > > for me, you're welcome to use it if you want.  I just added another
> > > modified function to populate the cache that doesn't probe devices or
> > > scan for block id tags (BLOCK_ID_DEV_FIND instead of
> > > BLOCK_ID_DEV_NORMAL).
> > 
> > This would break all the label and uuid mount cases.
> > 
> I don't think so.  The LABEL= and UUID cases still use the old
> function.  Those cases do generate seek errors but that is the
> expected result since if you use them, you have to probe all
> non-removable block devices to find them.

Yeah, you're right -- I eyeballed the patch and read it as changing
block_find_fs_by_label() .  This is why "diff -upN" is handy ;)

Still, it's not the best way to fix it.  Thanks for the patch though!
-- 
  Peter