From: Philippe Gerum <rpm@xenomai.org>
To: Giuseppe Lipari <lipari@domain.hid>
Cc: xenomai@xenomai.org
Subject: Re: [Xenomai-help] events and domains
Date: Tue, 13 Dec 2005 19:02:55 +0100 [thread overview]
Message-ID: <439F0CCF.4060609@domain.hid> (raw)
In-Reply-To: <439F0806.4070206@domain.hid>
Giuseppe Lipari wrote:
> Dear all,
>
> We are new to Xenomai and Adeos. We have some questions and I could not
> find answers to these in the documentation (but maybe we are not very
> good at finding things!). We hope someone can help!
>
> 1) Adeos is able to handle events that are passed to domains.
> I understand that events are hw interrupts and "other things". In the
> docs, I see that an event can be a "soft interrupt". As soft interrupt
> can have different meanings in different contexts, what do you actually
> mean with "soft interrupt"??
>
Virtual interrupts (virq) actually. Pseudo-IRQ channels that are not
mapped to real hardware but rather allocated by software to propagate
software-generated events. For instance, Xeno uses a virq for carrying
out asynchronous procedure calls from the primary to the secondary
domain (see /proc/xenomai/apc).
> 2) Also, what exactly are hooks? In particular, when there is a
> context-switch between two non-Xenomai (Linux) threads running in user
> space, is any hook-event generated?
>
Hooks == callbacks. Yes, Xeno uses this to track and monitor what the
hell Linux is doing at any time wrt user-space tasks mapped to shadow
Xeno threads. In nucleus/shadow.c, see do_schedule_event().
> 3) when a Xenomai thread starts executing, does it start in primary or
> in secondary mode?
>
Primary, always. It's written in stone.
> 4) is the memory of a Xenomai thread allocated in user space? (I guess
> yes, but just to be sure!)
>
No difference from regular Linux threads. Actually, a Xenomai thread
running in user-space is a Linux thread on steroïd. To this end, what we
call a shadow thread context is mapped to the initial Linux task
context, so that the Xenomai scheduler can handle it too, in parallel to
the regular Linux scheduler (in a mutally exclusive fashion though).
> 5) if two UVMs are active at the same time, how they are scheduled with
> respect to each other? (I guess SCHED\_FIFO, but it looks somewhat
> strange to me: why not SCHED\_RR, at least for the application tasks?).
>
It's explained in the "Introduction to UVMs" document.
>
> Giuseppe Lipari and Paolo Gai
>
>
>
--
Philippe.
prev parent reply other threads:[~2005-12-13 18:02 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2005-12-13 17:42 [Xenomai-help] events and domains Giuseppe Lipari
2005-12-13 18:02 ` Philippe Gerum [this message]
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=439F0CCF.4060609@domain.hid \
--to=rpm@xenomai.org \
--cc=lipari@domain.hid \
--cc=xenomai@xenomai.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.