[Adeos-main] adeos_enter/exit_syscall

[Adeos-main] adeos_enter/exit_syscall

[aaron durbin]

I have a question concerning the adeos_enter_syscall routine.  Is there
normally event monitors for syscall entry (SYSCALL_PROLOGUE)?  If so and
handle_event returns non-zero then one automatically restores the registers
and returns w/o calling the system call.  Doesn't this destroy the call?

-Aaron

Re: [Adeos-main] adeos_enter/exit_syscall

[Philippe Gerum]

On Thu, 2004-08-05 at 00:16, aaron durbin wrote:
> I have a question concerning the adeos_enter_syscall routine.  Is there
> normally event monitors for syscall entry (SYSCALL_PROLOGUE)? 

Not normally when no domain is loaded. The Linux domain by itself does
not intercept (its own) syscalls. But other domains might want this
(RTAI does).

>  If so and
> handle_event returns non-zero then one automatically restores the registers
> and returns w/o calling the system call.  Doesn't this destroy the call?
> 

It prevents its execution by the Linux kernel actually. Think of a more
prioritary (in the sense higher in the pipeline than Linux) that wants
to intercept some specific syscalls issued by regular Linux apps, and
provides a replacement for them. In such a case, you might not want the
original syscall to be handled by Linux too; so you make sure that it is
not propagated to Linux -- this translates in having handle_event return
non-zero (because adeos_propagate_event is not called for the said
syscall in the upper domain handler).

> -Aaron
> 
> _______________________________________________
> Adeos-main mailing list
> Adeos-main@domain.hid
> https://mail.gna.org/listinfo/adeos-main
-- 

Philippe.

Re: [Adeos-main] adeos_enter/exit_syscall

[Marc Kleine-Budde]

Hi!

Let's hope I've understand the concept... :)

On Wed, Aug 04, 2004 at 05:16:07PM -0500, aaron durbin wrote:
> I have a question concerning the adeos_enter_syscall routine.  Is there
> normally event monitors for syscall entry (SYSCALL_PROLOGUE)?  If so and

There can be a monitor, but there must not be one. 

> handle_event returns non-zero then one automatically restores the registers
> and returns w/o calling the system call.  Doesn't this destroy the call?

Yes - It should - This way you can handle syscall in the RTHAL domain
und substitute it with you own syscall, doing the same, but better then
linux (e.g. nanosleep).

Or the syscalls to the nucleus and skins, they are completely handled
outside of the linux kernel.

hth - Marc

-- 
#!/bin/sh
set - `type $0` 'tr "[a-zA-Z]" "[n-za-mN-ZA-M]"';while [ "$2" != "" ];do \
shift;done; echo 'frq -a -rc '`echo "$0"| $1 `'>$UBZR/.`rpub signature|'`\
echo $1|$1`'`;rpub "Jr ner fvtangher bs obet. Erfvfgnapr vf shgvyr!"'|$1|sh