I'm new to the RT world, but I'm curious... what significant challenges are
there when working with dynamic memory and real time? Just the term
"dynamic" gives us an indication that the mechanism would be
undeterministic... but what's worse case here? Why is it impossible to have
a "real time" dynamic memory manager? At worst, nothing is "real time",
it's a matter of perception. If I lived in a world where my timing
resolution is 1 sec at a time, then having a "dynamic" system ranging
between 0 and 0.999999s would be perceived as "real time", keeping in mind
that we ignore any other "timing" details involved... maybe I don't quite
understand the discussion.... but bottom line, what are the challenges
involved here???
- Tony
-----Original Message-----
From: Paolo Mantegazza <[EMAIL PROTECTED]>
To: [EMAIL PROTECTED] <[EMAIL PROTECTED]>; [EMAIL PROTECTED]
<[EMAIL PROTECTED]>
Date: Thursday, September 07, 2000 2:52 PM
Subject: Re: [rtl] RTAI and RTLinux
>[EMAIL PROTECTED] wrote:
>>
>> On Thu, Sep 07, 2000 at 06:58:48AM +0200, Paolo Mantegazza wrote:
>> > [EMAIL PROTECTED] wrote:
>> > - ..... do we want to bet how long will it take to see a real time
>> > memory manager in RTL?
>>
>> There is no such thing as a dynamic real time memory manager.
>
>In fact I was just betting there will be one in the future :-).
>
>> It's been some years since RTL has been available and we have
>> many times had this silly discussion.
>>
>> The code for a simple version is trivial.
>>
>>
>> Step 1: write some module code:
>> In Linux init:
>> ask for a soft RTL interrupt
>> install a handler
>>
>> handler:
>> read from request queue (could be a fifo)
>> v = kmalloc( requested.space)
>> *requested.result = v;
>> pthread_wakeup_np(requested.thread);
>>
>> Step 2: write some RTL thread code.
>>
>> In rtl periodic thread;
>>
>> queue request.
>> pthread_kill(pthread_linux(),kmalloc_irq);
>> while( !request.result)
>> pthread_wait_np();
>>
>> Step3 : receive Nobel Prize.
>
>Since you like joking, I feel invited to go on ;-).
>
>Once more RTAI has no problem in doing that. RTAI fifos were already
>based on the use of its srqs when RTL was still patching the kernel to
>protect Linux tasks queues to be used with fifos in real time.
>
>In fact we (I, Steve and Pierre) discussed a similar solution before
>they implemented theirs ideas, but they denied its usefulness because
>was too much not deterministic, and they needed a truly fast minimal
>latency dynamic memory manager for their applications.
>So I was just pleased to have them add, and copyright, what I think is
>an usefull service for RTAI. Recall that I support the idea that a reach
>toolbox of mechanisms enhances user politics implementation freedom.
>
>Note also that RTAI calls what you name soft interrupts as system
>requests and they are structured with two handlers, one for user and one
>for kernel space. So it could be simple for a supervisory user space
>process, acting as a user interface, to pre allocate memory in kernel
>space for a new stack of a to be launched rt_task, or a larger one onto
>which save an already in use stack, so that a real time task could be
>dynamically resumed on a larger stack if needed, and many other fancy
>things. However all those things were believed not satisfactory, because
>tightened to Linux, and thus miles away from a minimal latency memory
>management.
>
>Nonetheless RTAI exploits the idea of using an srq in relation to memory
>managment, but just for kfreeing kmalloced memory when tasks are
>deleted.
>
>Clearly I applied to have a Nobel Prize. They already promised me one
>when the euro will be worth .001 dollars, and it seems it will happen
>quite soon.
>
>Ciao, Paolo.
>-- [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/
>
-- [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/
