Subject: Re: cp(1) vs. UVM/UBC
To: Matthias Scheler <firstname.lastname@example.org>
From: Ignatios Souvatzis <email@example.com>
Date: 07/19/2001 10:40:53
Content-Type: text/plain; charset=us-ascii
On Wed, Jul 18, 2001 at 06:37:54PM +0200, Matthias Scheler wrote:
> On Wed, Jul 18, 2001 at 09:05:01AM -0700, Jason R. Thorpe wrote:
> > > So turning that option on would break "cp" on LFS. How fatal would t=
> > > be? Is LFS usable at the moment despite the mmap() problem?
> > Somewhat usable, yes. Let's wait to flip the switch until LFS
> > is fixed.
> I've submitted a change request in PR bin/13502 which says that it
> should be turned on after LFS and NTFS are fixed.
There are other file systems with potential problems, if I recall right...
e.g., the DOS\0 and DOS\1 variants of adosfs, which always have some=20
metadata in their datablocks, so for this case, data copying always
Maybe we need to disable mmap()ing for such filesystems inside the kernel,
and make cp do the right thing if mmap fails? Is this possible?
-----BEGIN PGP SIGNATURE-----
-----END PGP SIGNATURE-----