It all depends on what the device driver does. I've written device drivers
that have both Real-Time and normal Linux portions which communicate via
shared/global memory.
Norm
- Original Message -
From: Michael Nielsen [EMAIL PROTECTED]
To: [EMAIL PROTECTED]
Sent: Sunday, February
Umm... What's that mean? Would you please translate the mysterious acronym
SMM.
TIA
Norm
- Original Message -
From: [EMAIL PROTECTED]
To: [EMAIL PROTECTED]
Sent: Thursday, February 07, 2002 1:22 PM
Subject: Re: [rtl] Long latency on laptop
Is all power saving and the
Depends.
That is one way. Another is to wrap your rtlinux task with a normal linux
device driver which can then communicate with users using ordinary
read/write/ioctl functions. You'd need some way to notify the rtlinux task
that there's something for it to do. If it's already a periodic
I can't help you solve your problem because I've never seen anything like
that happen that wasn't my fault.
What I can tell you is that for my periodic threads, I use
pthread_create()
followed by
pthread_make_periodic_np()
and in the while-loop, I suspend with
rt_task_wait().
I've
Scan the archives. I've given the prescription a number of times for
extracting this function from the gcc library and linking it with your
program.
Norm
- Original Message -
From: Pablo Alvarez [EMAIL PROTECTED]
To: [EMAIL PROTECTED]
Sent: Wednesday, January 23, 2002 11:30 AM
Well, from the preprocessor's point of view, () are extra tokens.
What is being undefined is the string __cli, not the function
__cli(). Perhaps earlier preprocessors have been more lenient in
letting things like this through.
Norm
- Original Message -
From: Massimiliano Cialdi
Are you running X-window? If so, there's probably be a large number
of FP calculations you're unaware of.
Any sound cards? Things like that are often stealth uses of
resources.
The real question isn't having your thread's FP registers saved for
it, but the possibility that your thread my
- Original Message -
From: Calin A. Culianu [EMAIL PROTECTED]
To: [EMAIL PROTECTED]
Sent: Thursday, December 13, 2001 5:06 PM
Subject: Re: [rtl] make devices ?
You can write a small C program to essentially act as a wrapper to
the
mknod(2) system call. Then set this program to be
Ken.
Please don't think that I was advocating the abandonment of all
thoughts of creating a real-time memory allocator under all possible
sets of conditions -- No. What I was arguing was that it's
impossible in general, that there is no one-size-fits-all. Many
problems have finite
You can use some in rt-threads if you set the thread attribute to
use_fp before the first time it uses the math coprocessor.
The safest way to find out what math functions you can use is to
extract the modules you want from the math library with the 'ar'
command, then use objdump to see what
- Original Message -
From: Massimiliano Cialdi [EMAIL PROTECTED]
To: [EMAIL PROTECTED]
Sent: Sunday, November 25, 2001 5:29 AM
Subject: Re: [rtl] oneshot thread
Massimiliano Cialdi wrote:
Norm Dresner wrote:
If you have a 2 second deadline and a routine that
takes,
say
If you have a 2 second deadline and a routine that takes,
say, 150 milliseconds to execute, then set the sleep for
1.85 seconds and you'll meet your deadline. Unfortunately
there is nothing that I've found any more sophisticated
than this in RTLinux.
Norm
- Original Message -
From:
V3 of Real-Time Linux supports -- so the documentation
says -- POSIX (kernel) pthreads and sending signals to
(kernel) pthreads. The normal Linux kernel also supports a
version of pthreads for user-tasks. Can a real-time
Linux thread send a signal to a user-process/thread just by
knowing its
I remember answering this a month ago but here's the answer
again.
1. If by access runtime image of a module from another
module you mean to access globally defined data or
functions in one module from another, you have to use --
assuming you want an int variable called myData --
Changing the threads attributes to enable saving and
restoring the state of the FP coprocessor doesn't change
the function you call when you reference rtl_printf().
It's incapable of printing floating point without the
attribute change and it remains incapable with it.
There are several things
I need two features in RT-Linux that I had in AMX:
1. to wake up a sleeping task with a message. Message
means nothing more than a copy of a byte string of some
pre-defined length at an address specified by the caller.
Thus
structmystructreal_data = { 1 , 2
... };
The initial Segmentation Fault is undoubtedly due to some
unexpected operation(s) in your own module. Once that
happens during removal, the module is still in the kernel
and in the state TERMINATING (or REMOVING or something
similar). If it is in this state, it can't be removed
because its
What I would do at this point if I were faced with the
problem is to write a hello-world program in user-space C++
and I would determine if atexit were being called in the
startup. I would do this by one of three ways:
1. gdb
2. write a substitute in which I set a global flag if
it was
fopen is a normal Linux user-space function which AFAIK is
not available to any kernel-space task under any
conditions. As I understand, you'll have to create a
user-task to open the file and use it to send the data down
to the kernel either via ioctl() or via an rtlinux FIFO.
Norm
-
There were (at least) three sample implementations of a
C-function to read the Pentium's time-stamp counter. I
originally implemented two and when I discovered a problem
with the results of one, I implemented the third which was
supposed to be an improved version of the second. The
results of
to a newer version of rtl?
Thanks
Norm
- Original Message -
From: Norm Dresner [EMAIL PROTECTED]
To: rtlinux [EMAIL PROTECTED]
Sent: Thursday, October 25, 2001 8:42 PM
Subject: checking a pthread-ID
According to the map-page, pthread_wakeup_np always
returns
a zero value
According to the map-page, pthread_wakeup_np always returns
a zero value and consequently can't be used to determine if
the target thread in fact no longer exists. I have two
questions:
1. What -- if anything -- will happen if the thread has
been destroyed, i.e. either it has executed
No reason to apologize, your English is almost perfect.
Anyway, there may be other means, but here are some:
1. Global variables. You can export symbols in the module
that's loaded first and have each of the modules put into
these variables what it wishes to share and take out what
it needs
- Original Message -
From: Calin A. Culianu [EMAIL PROTECTED]
To: [EMAIL PROTECTED]
Sent: Wednesday, October 17, 2001 6:30 PM
Subject: Re: [rtl] Locking Memory in Kernel
. Do the GFP_KERNEL and GFP_USER flags have any
effect on this?
YES. Best advice is to get Rubini's book or
Calin.
No offense intended. I was just trying to point out that
there are many resources out there which will answer all
kinds of questions (basic and otherwise) about Linux kernel
programming, and that I thought that the Rubini book was
the best investment.
Norm
- Original Message
I've been running 1300 Hz on P-133 and PMMX-233 for
billions of cycles with no observed deviation from the
programmed frequency. On the P-133, my RT-kernel code
takes about 40% of my CPU's bandwidth but it seems to be
less than 10% on the PMMX-233. What do you have your
thread doing?
Oct 2001, Norm Dresner
wrote:
Inside a function called by a real-time task, I coded a
simple routine to read the contents from a port.
Because
I wanted not to even try to read from the address if
the
corresponding board wasn't there, I put an if-statement
in
front of the i/o read
Inside a function called by a real-time task, I coded a
simple routine to read the contents from a port. Because
I wanted not to even try to read from the address if the
corresponding board wasn't there, I put an if-statement in
front of the i/o read.
The code I wanted to write was:
...
Alternately, if you know the approximate range of the
number -- you do know the approximate range of the number,
don't you? -- you can do some integer math and compute the
integer and fractional parts and create an ASCII
representation that way.
Norm
- Original Message -
From:
- Original Message -
From: Calin A. Culianu [EMAIL PROTECTED]
To: [EMAIL PROTECTED]
Sent: Saturday, September 29, 2001 9:24 PM
Subject: Re: [rtl] RT-Linux patch proposal for RTF size
queryage
On Fri, 28 Sep 2001, Victor Yodaiken wrote:
You're welcome to send a patch and your tone
Not quite.
There is a big difference between Library Calls
and System Calls. Some Library calls are
nothing much more than calls to your own
functions. System Calls are what you're referring
to. There are library calls that are safe to make
and library calls that call system calls that are
If the function is listed in the documentation as
having an RT-Linux version then you are not
calling the real user-space routine but an
RT-Linux one that has (approximately) the same
description and calling sequence.
You should never call user-space functions (unless
you really know what you
Peter.
I think you'll still need to use ioremap() to get a virtual address with
which to access the memory -- and don't forget to iounmap() afterwards. If
you don't have a copy of Rubini's book, Linux Device Drivers, the QD
account is
VirtualAddress = ioremap( PhysicalAddress , size );
Exactly what do you mean by port it to rtl? (notice that CAPS are
unnecessary).
If all you want to do is run this driver as a linux driver on an
rtl-enhanced computer, then there's nothing to do. If you want it to be a
real-time driver, then you'll probably have a lot of work.
Norm
-
I'll second the problem of using an A-to-D card with a mux. I've had
several projects where I've had to use a PC-Labs PCL-818 High Speed Data
Acquisition Card. I found that to get complete settling of the mux takes
over a millisecond -- an eternity compared to the 3-5 microseconds
conversion
I'll second the notion that there are no real problems with fwvm(2) which I
used with RTL v1 for over a year with no noticeable degradation. If -- and
this is a big IF that I'm going to have to investigate -- the only problem
with the Gnome desktop is with the automount features, they can be
PROTECTED]
To: Norm Dresner [EMAIL PROTECTED]
Cc: rtlinux, [EMAIL PROTECTED]
Sent: Sunday, August 19, 2001 7:18 AM
Subject: Re: [rtl] Examples and such
So examples/v1api/frank doesn't compile?
Did you enable CONFIG_RTL_USE_V1_API and recompile RTL?
Norm Dresner ([EMAIL PROTECTED]) wrote:
Okay
On Friday 17 August 2001 01:31, Calin Culianu wrote:
Hey, what are people doing running MS software anyway? :)
Dunno. Maybe they have better nerves than we have? (Not that that would
make it possible to do anything *really* interesting...) ;-)))
It's not totally my choice at work
For the x86 architecture, AIUI, the frequencies have to be expressed
relative to the motherboard oscillator (1.19...MHz) so I'd guess that sets
the resolution.
Norm
- Original Message -
From: fred august [EMAIL PROTECTED]
To: rtl [EMAIL PROTECTED]
Sent: Thursday, August 16, 2001
What version of Linux (and gcc) are you using?
Norm
- Original Message -
From: Dan Peters [EMAIL PROTECTED]
To: [EMAIL PROTECTED]
Sent: Friday, August 03, 2001 12:36 PM
Subject: [rtl] kernel compile problem
Can somebody tell me why I always get the following error when trying to
AIUI, the math coprocessor on an x86 CPU can be used in an RTL-module by
declaring that it uses FP math. I don't believe that there are these kinds
of restrictions in other architectures.
Norm
- Original Message -
From: Theo ten Brummelaar [EMAIL PROTECTED]
To: [EMAIL PROTECTED]
. The errors I
get there are similar to (some of) what I see in my own program so I feel
that if I can get past frank, I'll be a long way toward getting my program
compiled too.
All help is appreciated. BTW, the system is RH 6.2 w/2.2.18 and RTL 3.0
Norm Dresner
Fellow Systems Engineer
Dean.
While I understand that you're trying to port software with as few changes
as possible, I don't see why using RTLinux shared memory (not Linux shared
memory) wouldn't give you the functionality you need.
Norm
- Original Message -
From: Dean W. Anneser [EMAIL PROTECTED]
To:
Janet,
Here's the output from the lsmod command on my system here:
Module Size Used by
lvi 130360
rtl_sched 40936 0 (lvi)
rtl_time10164 0 [rtl_sched]
rtl25984 0 [rtl_sched rtl_time]
The only order that will work is
rmmod
I've tried to dual boot RTL 3.0 over Linux 2.2.18 and RTL 0.9 over 2.0.36.
While I can boot both systems, I can't access (that is load) the modules in
the "2nd" system. [In this context, the first system is the one whose LILO
I've last run and the second system is the other.]
I'm almost
1. The more things can be standardized, the less they have to be reinvented
with each new implementation. I approve of /usr/lib/xxx where xxx is
something mnemonic. I've already done that for each major project I have
on my development machines.
2. I have my own "standard" header files. For
46 matches
Mail list logo