Have any ISA devices on the bus?
How many threads and what priorities?
Send code and we can test it on some other hardware.
On Tue, Sep 19, 2000 at 12:02:50PM -0400, Stephen D. Cohen wrote:
> Gang,
>
> I am troubled by some rather non-deterministic behavior that I am
> observing in the scheduler of RTL 3.0. Basically, I have an ISR that
> responds to an interrupt by doing a pthread_wakeup_np() on a thread that was
> sleeping on a pthread_suspend_np(). Just before doing the wakeup I capture
> the time from gethrtime(). Just after the wakeup, I capture the time with
> another gethrtime(). I then examined the time difference between the two.
>
> Unfortunately, rather than getting a large cluster of times centered
> around some reasonably small mean time, I got a distribution that spread
> over an order of magnitude. Here is the histogram (roughly):
>
> 0-2 uS 28,590,000
> 2-4 uS 28,150,000
> 4-6 uS 492,000
> 6-8 uS 293,000
> 8-10 uS 848,800
> 10-12 uS 1,051,000
> 12-14 uS 40,564
> 14-16 uS 1,272
> 16-18 uS 1,200
> 18-20 uS 1,094
> 20-22 uS 1,606
> 22-24 uS 3,084
> 24-26 uS 881
> 26-28 uS 10
> 28-30 uS 2
>
> This is running on a Celeron 400 with plenty of RAM. Linux is
> running a Tcl application that displays a front panel on an X window with no
> window manager running. I would describe the Linux load as "very light".
>
> I am greatly troubled by the very long delays here. I was expecting
> *all* of the delays to be in the sub-10 uS range and for the *vast majority*
> to fall in the sub-2 uS range. I am quite unpleasantly surprised.
>
> I need to depend on things to be deterministic. Obviously,
> something is preventing me from having any real semblance of determinism.
> Does anybody have any suggestions for places to look for solutions to this
> problem?
>
> Alternatively, if someone can provide a description of *why* I get
> some delays in the tens of micro-seconds and some assurance that I will not
> see *any* delay longer than X microseconds, that would help put my mind at
> ease.
>
> Regards,
>
> Steve
>
>
> ------------------------------------------
> Stephen D. Cohen, Engineering Manager
> Xybion Positioning Systems
> 11528 53rd Street North
> Clearwater, FLA 33760
> phone: (727) 299-0150, fax: (727) 299-0804
> ------------------------------------------
>
> -- [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/
--
---------------------------------------------------------
Victor Yodaiken
Finite State Machine Labs: The RTLinux Company.
www.fsmlabs.com www.rtlinux.com
-- [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/
