From: Jeff Dike <jdike@ka...> - 2001-08-12 23:59:29
> From UM-linux instead, the write succeeds and the process terminates
> normally. The mmap-ed file is not affected by the write though, so I
> guess the UM-linux forces a copy-on-write in this case...
Interesting. I'll have a look and let you know what I find.
> any reason
> for this difference?
Yeah, it sure looks like a UML bug to me.
From: Jeff Dike <jdike@ka...> - 2001-09-11 02:26:36
> If I try to write the returned pointer from the host linux, the
> process causes a segmentation fault, as the pages are read-only. From
> UM-linux instead, the write succeeds and the process terminates
I just checked this, and this bug was fixed a while ago when I started
figuring out why crashme behaved differently on the host and UML.