Gentoo Archives: gentoo-embedded

From: wireless <wireless@×××××××××××.com>
To: gentoo-embedded@l.g.o
Subject: Re: [gentoo-embedded] Hard Real Time Systems with Gentoo?
Date: Sun, 14 Jan 2007 00:38:44
Message-Id: 45A93480.7080708@tampabay.rr.com
In Reply to: Re: [gentoo-embedded] Hard Real Time Systems with Gentoo? by Bob Paddock
1 not.
2 Bob Paddock wrote:
3
4 > The way I posed the question was meant to result in:
5
6 > "Yes, there is a RTOS that works well with Gentoo-Embedded,
7 > and it is XYZ at http...."
8 >
9 > How would you have asked the question?
10
11
12 I think you just did. You seem to be very versed
13 in the RTOS offerings, commercial or opensource.
14 Notice the name RTOS, it does not differential
15 between soft and hard. When it all boils down to the essence,
16 that distinction is meaningless. What the should be called
17 is reliable, low-latency executive. But everybody
18 accepts the moniker RTOS as OS for embedded systems.
19 RTOS do not have State Machine semantics, unless the
20 firmware engineer that using the RTOS imposes that
21 rigidly in the software design. The RTOS might have
22 available mechanisms and resources for the firmware
23 engineer to create a state machine, using their RTOS,
24 but it's not ensured, just by selecting a given RTOS.
25
26
27 >> Soft real time, means that timing constraints are soft
28 >> or guildlines for process performance metrics.
29
30 > Erlang is a 'Soft' Real Time System, not what I need at the moment.
31
32 >> Hard real time, means the timing constraints are hard
33 >> or an absolute requirement.
34
35 > I am aware of the issues, that is why I made the distinction,
36 > in what I was after.
37
38 >> to not be 'real time'. But think about if a missile,
39 >> is suppose to change it's rudder in response to an
40 >> AD input, in 150ns, does it mean that the missile as
41 >> as system is a failure if the rudder response is
42 >> 151 ns or 150.002 ns?
43
44 > If it is in the targeting system, it makes the difference between
45 > hitting the target or not. A 1 ns difference, is a large
46 > area to transverse when your missile is traveling at Mach-6.
47 > A missed target is a failed Missile System.
48
49 possibly, possible not.
50
51
52 > TechUpdate from the Missile Defense Agency
53 > comes out quarterly. MDA funds a lot of research in other
54 > areas. The Summer 2006 laying here beside me is
55 > about "Phasing out Brest Cancer" using Microwaves.
56 > "Fuzzy-logic software" on page ten might be more relevant to this
57 > list. Check it out at: http://www.mdatechnology.net/
58 > Maybe someone can make a ebuild for their screen saver?
59
60 Yes many of those publications and organizations participate
61 in promulgation of pop-science. However, that does not
62 affect the reality of mathematics but it does taint the use
63 of precise language constructs used in math, science and engineering.
64
65 I guess if Webster's includes a word from Eubonics into the
66 English dictionary it becomes acceptable English?
67
68
69 >> Determinism is another funny character. Real embedded
70 >> systems use a 'state machine' design where every
71 >> state or transition is fully characterized and defined.
72
73 >> ON a large complex system, it is virtually impossible to
74 >> discover, characterize, define and test all possible states
75 >> of a complex system.
76
77 > Use the appropriate tools and languages such as Esterel.
78 > http://www.esterel-technologies.com/
79
80 Or state machine design and native assembler or ansi C.
81 If either system meets the (arbitrarily set timing constraints
82 both are classified as RT..... by the masses of pop-science.
83
84 What happens when after months of deployment a failure
85 in the OO dev tools appears? Not very common with
86 a tested state machine design, but too often common when
87 OO tools are used.
88
89 >> Embedded linux anything, currently, can never be fully deterministic.
90 > Embedded Linux can be run as a task of a fully deterministic RTOS,
91 > built with a Finite State Machine if you like.
92
93
94 Are you kidding? Deterministic means all possible state/transitions
95 have been defined, tested and verified. Do you really think that
96 has ever been accomplished on any Unix/Linux system?
97
98 Nasa did it decades ago on a processor and spent so much time
99 and money that it has never been attempted again, except as
100 pop-science projects using magic and pop-statistical models.
101
102 > Any done with Gentoo-Embedded?
103
104 Nope and never with any Unix/linux/bsd operation system.
105 However, Bill Gates will probable tell you that windowXP
106 has this feature (again pop-science).
107
108 > I know what I'm building, as well as being well aware of the high
109 > importance of System Design Requirements.
110 > But will I find one that actually answered my question? :-)
111
112 If you ask a question with common mathematical misconceptions,
113 some may point out the invalid requests. And no you did not
114 ask for RTOS options, you asked for
115 "Does anyone know of any Hard Real Time Systems,
116 where deterministic response is required, that work
117 with Gentoo-Embedded?"
118
119 Although RTOS is a common term, it does not ensure HRT performance,
120 but, the user accepts 'best effort' reliable, timing constraints
121 when using an RTOS embedded operating systems. If others want
122 to incorrectly use terms such as hard and soft real time,
123 then it's either pop-science or marketing; you pick.
124
125 (Fully) Deterministic systems are an enormous level above HRT and
126 rarely exist except in smaller systems. The very nature of a uinx
127 OS (embedded or not) is gargantuan when you look at the possible
128 number of states and transitions.
129
130
131 James
132
133
134
135
136 --
137 gentoo-embedded@g.o mailing list