lkml.org 
[lkml]   [2005]   [Mar]   [9]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH] make st seekable again
On Wed, 9 Mar 2005, Alan Cox wrote:

> On Maw, 2005-03-08 at 17:25, Linux Kernel Mailing List wrote:
> > ChangeSet 1.2030, 2005/03/08 09:25:05-08:00, kai.makisara@kolumbus.fi
> >
> > [PATCH] make st seekable again
> >
> > Apparently `tar' errors out if it cannot perform lseek() against a tape. Work
> > around that in-kernel.
>
> Unfortunately this isn't a good idea. Allowing tar to read the tape
> position makes sense, allowing it to zero the position might but you
> have to do major surgery on the driver first because
>
> 1. It doesn't use ppos
> 2. It doesn't do locking on the ppos at all
>
> Also allowing apps to randomly seek and report "ok" when they are
> backing up to tape and might really need to see the error is not what
> I'd call stable, professional or quality code.
>
The proper fix is to fix tar. I have sent an analysis of the problem and a
suggestion how to fix this to the bug-tar list on March 5 but it is still
waiting for moderator approval.

While waiting for the application to be fixed, it was decided to restore
the old behaviour of the tape drivers.

lseek on a tape is not a good fit (addressed by block, blocks on tape can
have any size, etc.). I don't know any Unix that would really implement
lseek on tapes but they usually don't return error. This is probably why
the tar bug has not been found earlier.

There has been one useful way of using lseek() with tapes in some systems.
Those refuse reads and writes if the file pointer reaches 2 GB. Resetting
it with lseek(fd,0,0) now and then has allowed writing/reading more than 2
GB.

I don't think implementing proper read-only lseek for tapes is worth the
trouble (reliable tracking of the current location is tricky). Purist
kernels can refuse lseeks. Pragmatic kernels can allow lseeks until
refusing those won't break common applications.

--
Kai
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 14:11    [from the cache]
©2003-2011 Jasper Spaans