From: Jeff Dike <jdike@ka...> - 2002-10-28 03:28:41
This patch cleans up some of the debris.
CONFIG_MODE_TT and CONFIG_MODE_SKAS are now configurable, rather than being
hard-wired in config.in. UML now builds and runs with only one built in,
although running a skas UML on a non-skas host is likely to be nasty.
SMP builds again, although I punted on SMP skas for the time being.
I cleaned up the register handling and separated out more code into the mode
I kludged UML into building as a normal dynamically linked binary, although
that required linker script surgery which I am not including in this patch.
The reason for this is to get valgrind working on UML. After some persuasion,
it sort of does, but bombs on the first call to clone, which is fairly early.
This looks like it might be easy to fix, so stay tuned.
From: David Coulson <david@da...> - 2002-10-28 14:33:27
Jeff Dike wrote:
> CONFIG_MODE_TT and CONFIG_MODE_SKAS are now configurable, rather than being
> hard-wired in config.in. UML now builds and runs with only one built in,
> although running a skas UML on a non-skas host is likely to be nasty.
I'm still having questionable success with UML on UML. I first compiled
it with my regular .config, and I got a "I'm tracing myself" error. I
then disabled SMP and highmem (not that I was actually using either of
them anyway), and now when I run the UML in the UML, it seg. faults.
I'm going to take a better look at it today and try to figure out if
it's something specific in my configuration or if the UML running on my
laptop is going screwy when I run the other one... :-/
David Coulson http://davidcoulson.net/