Ok. I admit defeat. I once said that it would take 1/2 hour
to make a memory allocator on RTLinux and was off by a small
multiplier: although when I said it, it took me less than 1/2
hour to remember how to compile a module.
C++ users seem to have a real need, so here we go.
My opinion is that there should be separate buffer
pools for each component so that a memory allocation in one 
component cannot cause failures in another. 

Anyways: the curious can see two allocators in
www.fsmlabs.com/~yodaiken/hacks (does the name give a clue as
to whether we will support these yet?).

the better allocator is simply an adaption of the nice allocator
from Fourmilabs. This is in the bget files. I added a mutex and
made it into a module. That took longer than expected, since I 
got too interested in the type structure. Testing included 0 minutes
so use at your own risk.

The more interesting, from my point of view, code is rtl_vmalloc
which shows how to call an arbitrary Linux function from RTL and
includes an improved version of the code using a kernel thread
as a server.  The example is vmalloc. I've got a much more
ambitious version of this in the works, but it's been in the works
so long, I decided to release something.

Comments and suggestions are welcomed as always.



-- 
---------------------------------------------------------
Victor Yodaiken 
Finite State Machine Labs: The RTLinux Company.
 www.fsmlabs.com  www.rtlinux.com

-- [rtl] ---
To unsubscribe:
echo "unsubscribe rtl" | mail [EMAIL PROTECTED] OR
echo "unsubscribe rtl <Your_email>" | mail [EMAIL PROTECTED]
--
For more information on Real-Time Linux see:
http://www.rtlinux.org/rtlinux/

Reply via email to