Andrew Morton wrote:
>
> Wilken Boie wrote:
> >
> > David Olofson schrieb:
> > >
> > > On Tuesday 20 March 2001 10:37, Erwin Rol wrote:
> > > > The low latency ppl also warm against using the linux framebuffer
> > > > drivers, although they
> > > > don't have hardrealtime so this warning might not be of importance for
> > > > RTAI.
> > >
> > > Right; Linux/lowlatency relies on the Linux scheduler, and the framebuffer
> > > problem (possibly about to be fixed) is that the fb drivers stay in kernel
> > > space for extended amounts of time. A fully preemptive RT kernel that
> > > bypasses Linux (such as RTL or RTAI) wouldn't be affected by that.
> > >
> > Is there anybody who could be a bit more specific about this? E.g. I
> > remember I heard this occurs when _scrolling_ the fb. Does this in turn
> > mean we're fine without scrolling?
>
> Pretty much.
>
> Try http://www.uow.edu.au/~andrewm/linux/console.html and
> follow the first link...
>
> > I would appreciate, if somebody could point me to the right resources for this
> > issue (mailing list, fixing status etc.). The reason is, that there is in deed
> > a world beside X and to a good part that's based on fb (e.g. QT embedded),
> > and I would like to use that under low latency conditions.
>
> The patches which are in 2.4.2-ac20 work OK.
>
> Problem is, the entire patch is so big, hairy, scary and late that
> I'm uncertain about the chances of getting Linus to accept it for 2.4.
>
> So I'm currently wondering about kludgy alternatives, breaking the
> writes into smaller chunks.
>
> But I'm told that a single "\n" on vesafb can block interrupts for
> 100s of milliseconds. I don't understand why this is at present.
This also seemed to happen in Xfree 4.0.x one [enter] in a gnome console
or xterm and latency of about 10msec follows. A time back we (some xfree
ppl and i) looked for the cause but never could find the exact cause. I
disabled the hard interupt enabling and disabeling but that didn't help
, we forced all the PCI writes to check the FIFO before writing and that
also didn't help. only disabling the accelerator helped.
>
> So. Still a work in progress. But it has to be fixed somehow.
> The current behaviour is unacceptable. Alan agrees with this;
> not sure about Linus.
The soft IRQ disabeling should have no effect on RTAI or RTL, so it is
just an secodary problem from most ppl on this list. But if it is caused
by something else that also effect RTAI and RTL than it really is a
problem :-)
- Erwin
> -- [rtl] ---
> To unsubscribe:
> echo "unsubscribe rtl" | mail [EMAIL PROTECTED] OR
> echo "unsubscribe rtl <Your_email>" | mail [EMAIL PROTECTED]
> --
> For more information on Real-Time Linux see:
> http://www.rtlinux.org/rtlinux/
-- [rtl] ---
To unsubscribe:
echo "unsubscribe rtl" | mail [EMAIL PROTECTED] OR
echo "unsubscribe rtl <Your_email>" | mail [EMAIL PROTECTED]
--
For more information on Real-Time Linux see:
http://www.rtlinux.org/rtlinux/
