Question about NMIs and Xenomai.

孙世龙 sunshilong sunshilong369 at gmail.com
Tue Jul 14 10:14:57 CEST 2020


Thank you for taking the time to answer my question.

>Neiher Xenomai nor I-pipe block NMIs. If Linux is configuring them for
>certain debugging or observation purposes, e.g. perf, they hit also the
>real-time domain.
So, the running Xenomai threads may be interrupted by NMIs. Am I right?
If it's right, should NMIs(e.g NMI watchdog) be disabled to achieve
lower latency?

Thank you for your attention to this matter.
Yours sincerely.
孙世龙

On Tue, Jul 14, 2020 at 3:45 PM Jan Kiszka <jan.kiszka at siemens.com> wrote:
>
> On 14.07.20 08:52, 孙世龙 sunshilong via Xenomai wrote:
> > Hi, list.
> >
> > I have googled for a long time and read the posts on the official
> > website(besides <<Life-with-Adeos>>) carefully, but I still can't find
> > the answers which directly answer these questions about NMIs:
> > *Is it possible that NMI watchdog influences the latency of Xenomai?
> > *Is it possible that Xenomai threads(which are busy running for
> > several seconds) block the Linux domain from handling the non-maskable
> > interrupts?
> >
> > I would be grateful if somebody shed some light on these questions.
> > Yours sincerely.
> >
>
> Neiher Xenomai nor I-pipe block NMIs. If Linux is configuring them for
> certain debugging or observation purposes, e.g. perf, they hit also the
> real-time domain.
>
> Jan
>
> --
> Siemens AG, Corporate Technology, CT RDA IOT SES-DE
> Corporate Competence Center Embedded Linux



More information about the Xenomai mailing list