Subject: Re: Volunteers to test some kernel code...
To: Curt Sampson <>
From: Brett Lymn <>
List: tech-kern
Date: 06/22/1999 16:25:32
According to Curt Sampson:
>Isn't that going to happen automatically once we get UBC?

Dunno - I may have missed something that Chuck (I assume Chuck is
doing UBC?) has mailed out.  If it does happen then I would be happy.

> I don't
>see how it can't, because if I read() that file, I'm dealing with
>the same physical page in the buffer cache that's mapped in for
>the text segment of that program, and thus it's got to be COW,
>which it can't be if I'm write()ing it because then my writes go
>to swap instead of the file.

I suspect this is one of those hairy situations that make a UBC so,
ummmm, interesting to code.  I am not sure that you are correct
though, you would have to allow writes to files that are being read -
that is a valid thing to do.  What we need to do is make the
distinction between a data file and a currently executing binary.

> (Or we could just allow people to
>modify running programs. :-))

Like we do now :-)

Brett Lymn, Computer Systems Administrator, British Aerospace Australia