How to boot with -ptcpip [was: netstat and route]

bridged with qdn.public.qnxrtp.os
Post Reply
Xiaodan Tang

Re: Thead Priority

Post by Xiaodan Tang » Thu May 30, 2002 1:10 am

camz@passageway.com wrote:
Chris McKillop <cdm@qnx.com> wrote:

the case of multiple machines with the same named resource. Perhaps using
an open standard like LDAP or some other open scheme.

I hadn't thought of LDAP. That's an interesting idea, but is also pretty
heavy in terms of solutions, especially when working with small embedded systems.
For the larger systems that might have LDAP for other reasons, it's a perfect
fit.
Since our request/reply is simple, maybe it is not that HEAVY.
I am actually investigating this one. So the "service look up" actually
sends out a LDAP message, and we will have a tiny LDAP server to answer
just those request.
Another thing I found light-years ahead of QNX4 was the push to use resmgrs
so that all applications can use the POSIX APIs to communicate to the
managers over the network.

Using a global name is pretty much useless if you want to use perl to talk
to your service.

Well, this depends. :-) If you use name_attach() and name_locate(), then you
could have a resmgr manage the /var/net space and perform a redirect/symlink
type service to point to the actual resource in /net/machine/... then you COULD.
Mind you, there isn't any reason why you can't have a resmgr sit "on top" of /var/net
and examine those requests to determine if it needed to let them pass-thru or
be redirected, somewhat like fs-pkg does already.
Exactly. A "service" could simply be a symlink to remote node. Thus, use
a "service" without knowing it's server node, is very important. Think
you take an iPaq to company, and just ask "where is tcpip service", and
"SOCK=<tcpip service> voyager", off you go! You don't actually care who
IS providing tcpip :)
There is a related topic (cdm: you knew this was coming), which is remote spawn,
which is, IMO the other requirement of being able to actually create distributed
systems.

Right now remote spawn *IS* possible, but it needs to be wrapped up in an API call
of some sort to move it out of the realm of "guru black magic".
This got to be a set of APIs, daemons. Like a system that have daemons
exist in each node of a distirbute network, keep track of each others
CPU usage, and "spawn" process to those spare nodes. And better, with
support of HAT, we could sort of doing remote "fork()", to push a
job from busy nodes to spare nodes :)

You running mozilla, everytime you click a link, the actually process
fly from one node to another :)

The PVM is actually something like that, except they can only decided
which node to start a job.

-xtang

David Gibbs

Re: Thead Priority

Post by David Gibbs » Thu May 30, 2002 3:14 pm

camz@passageway.com wrote:
Agreed. I actually liked the QNX2 method where the uniqueness of
global names was enforced (ie. you could not attach a duplicate name).
Unfortunately, true uniqueness enforcement is a non-trivial problem.
For instance, how do you deal with a network split, two processes
one on each side of the split register the "unique" name, which is
unique for their currently visible network, then the network rejoins?

-David
--
QNX Training Services
http://www.qnx.com/support/training/
Please followup in this newsgroup if you have further questions.

Guest

Re: Thead Priority

Post by Guest » Thu May 30, 2002 6:26 pm

David Gibbs <dagibbs@qnx.com> wrote:
Unfortunately, true uniqueness enforcement is a non-trivial problem.
For instance, how do you deal with a network split, two processes
one on each side of the split register the "unique" name, which is
unique for their currently visible network, then the network rejoins?
I know, I know... there are also race conditions and such. I don't
expect to actually have globally unique names, we've managed to live without
them in QNX4 for years now without any real negative effects.

I do like the option of being able to specify that the name be globally
unique, but I realize that it isn't overly realistic to actually achieve
it.

Cheers,
Camz.

Art Hays

Re: EIDE

Post by Art Hays » Thu Mar 13, 2003 11:22 pm

Does anyone know how involved this is? Someone has told me
it would be just a few lines of code to support the ICH4 if you
already had support for the ICH2.

Art Hays

"S. Miller" <smiller@retia.cz> wrote in message news:3E6C365A.90BA2F5F@retia.cz...
Hi,

I am seeking a powerful mainboard for Pentium 4 or Xeon. But problem is
with eide busmastering with chipsets: with southbridge - Intel ICH4, SiS
963 (These we have tested). When will QNX support them? And is eide
busmastering supported on Intel E7501 (ICH3-S)?

Thanks
Svatopluk Miller

Guest

Re: Package mount points

Post by Guest » Mon Apr 21, 2003 7:16 am

william@bangel.demon.co.uk sed in <3E9D3CAE.BCCC7D1E@bangel.demon.co.uk>:
What determines the mount point reported for a package?

/pkgs/repository/GNU/GNUEmacs/core-21.2-bld8 on / type pkg

/pkgs/repository/waukes/htdoc/core-1.0 on / type pkg
Ignore them. They aren't useful (now).

Current fs-pkg shows spurious locations for source mountpoint
(or wasit the first line in /etc/system/packages/package; dunno)
--
kabe

William Morris

Re: Package mount points

Post by William Morris » Tue Apr 22, 2003 12:16 pm

kabe@sra-tohoku.co.jp wrote:
What determines the mount point reported for a package?
Ignore them. They aren't useful (now).
But if I try to remount / as read-only I get "resource busy".
It was indicated in another thread that the likely reason for
this was that something else is mounted on / - am I right to
assume that these otherwise-unuseful mount reports are
responsible? If so, I cannot ignore the failure of the re-mount.

Thanks
William

Adam Mallory

Re: mmap'ed memory access very slow

Post by Adam Mallory » Thu Jun 26, 2003 7:25 pm

Can't you just change the attributes of the memory block after the DMA
transfer and before the calculations, so that the memory can be cached,
using mprotect()?
You'd also have to make it PROT_NOCACHE, and flush the cache line(s) before
the DMA transfer starts otherwise a delayed writeback could clobber your new
data. I think it's not too easy or possible to get DMA transfer 'begin'
notification *and* hold off the DMA write until all the cache has been
written back.

-Adam

Paolo Messina

Re: mmap'ed memory access very slow

Post by Paolo Messina » Thu Jun 26, 2003 9:20 pm

Can't you just change the attributes of the memory block after the DMA
transfer and before the calculations, so that the memory can be cached,
using mprotect()?

You'd also have to make it PROT_NOCACHE, and flush the cache line(s)
before
the DMA transfer starts otherwise a delayed writeback could clobber your
new
data. I think it's not too easy or possible to get DMA transfer 'begin'
notification *and* hold off the DMA write until all the cache has been
written back.
I was supposing that changing the attributes to PROT_NOCACHE would flush the
cache. Is this the case?
As for the buffer, I was assuming that the calculations involved only read
operations on the data coming from the video grabber, while the result of
the processing was written to another buffer.
If that's the case then you should only need to flush the cache before
processing the data, right?

-Paolo

Wojtek Lerch

pterm & +'S'

Post by Wojtek Lerch » Mon Nov 17, 2003 3:55 pm

(Crossposted from qdn.public.qnxrtp.applications)

Wojtek Lerch <wojtek_l@yahoo.ca> wrote:
Pterm doesn't understand the difference between Ctrl-C and Ctrl-S. If one
works and the other doesn't, it must be because devc-pty treats them
differently.

"Bill Caroselli" <qtps@earthlink.net> wrote in message
news:bp3cf1$845$1@inn.qnx.com...
I have a text application that spits stuff to the screen very fast
indefinitely. I can't seem to stop pterm with <Ctrl>+'S'. If I put
any delay in the loop, I.E.
delay( 10 );
I can pause the output of pterm just fine and restart it when I'm
ready.

I can always <Ctrl>+'C' the progrm. That happens immediately.

Why can't I pause the output of a pterm?

Post Reply

Return to “qdn.public.qnxrtp.os”