Subject: Re: cp(1) vs. UVM/UBC
To: None <>
From: Christos Zoulas <>
List: tech-userlevel
Date: 07/19/2001 12:14:51
In article <>,
Matthias Scheler <> wrote:
>On Wed, Jul 18, 2001 at 12:16:32PM -0400, Andrew Brown wrote:
>> would there be any point to looking at the output of fstatfs() after
>> the open() to decide a code path?
>That would probably eat up the performance gain.

I think that mmap should be fixed so that it returns EOPNOTSUPP on filesystems
where it does not work, or will never work. Having it return an address and
then segfaulting when you dereference it, it suboptimal.