Problems with nested interrupts at high interrupt load ? (x

QNX Newsgroups (Archives) qdn.public.qnxrtp.os
21

Adam Mallory wrote:

Armin Steinhoff <> a-steinhoff@web.de> > wrote in message
news:> 3F45D9E5.A135827@web.de> …

[… ]
I asked simply:
“are there any problems known with nested interrupts
at a high interrupt load? (QNX 6.2.0A, x86)”

The answer is No.

Thanks, that’s a clear statement.

Have you actually verified that the interrupt line associated with the
resource manager which does not respond is being asserted (edge/level)
properly, even if the manager doesn’t see it?

Every resource manager works flawless at maximal interrupt load!
But the operating system isn’t able to run two of these resmgr at a
lower interrupt load!

Note: there isn’t only an interrupt problem, also the resmgr_block()
statement of the main thread doesn’t receive a message even if it is
receive blocked.

As you like to interpret, just my interpretation to that answer:

  1. he was not able to read and understand the question

Even I can’t understand some of the broken english that gets posted
to the newsgroups.

sure, and others have problems to understand posted slang.
But I think that my short question was understandable …

  1. examination of a snipped code can’t tell you if others saw the
    same problem

No, but if you made a mistake we could help you (related to this issue or
not). You’re human, and regardless of how long you’ve coded ISRs, you’re
not excluded from making mistakes.

Ditto … that’s also the reason why I test my code.

  1. the answer implies that I didn’t check my code before asking, and
    that I’ve no experience with simple handling of interrupts…

That’s utter nonsense. Asking for code is just a fast way for others to
understand what you’re doing (we all speak the same langauge in that
regard), it has no bearing on experience, knowledge or ability.

… reading the code of a (standard) interrupt handler again and again
let you understand nothing.

Important is to understand the missbehavior of the system!


And now you expect that I say “thanks anyway”… you can’t expect that
everyone in these NGs has American behaviour, just because of the
language is English and QSSL is a Canadian company …

Being polite has nothing to do with nationality, it has to do with working
with others.

Understanding what ‘polite’ means has much to do with nationality if you
believe or not. What’s polite for you can be annoying and extremely
unpolite for an other one. Have a look to qdn.cafe, Jutta followed up
this discussion …

If you wish not to be polite, then so be it; you can hardly
get shocked if others don’t respond the way you want them to.

May be you feel that the answer on my question was polite, for me it
was impertinent… just my interpretation.

Rennie wasn’t rude at all - it is your incorrect interpretation.

The quoted answer was from Kevin, not from Rennie!
… and notice, an interpretation is always subjective.

Anyway, thanks for your clear answer to the initial question,

Armin

22

Igor Kovalenko wrote:

“Armin Steinhoff” <> a-steinhoff@web.de> > wrote in message
news:> 3F45D9E5.A135827@web.de> …

[…]

Sorry Rennie, this is the OS NG and not the cafe NG …
I wanted a related answer, nothing else.

You wanted a related answer, huh?

Is it absurd? …

Have you forgotten Armin, this is not an
official support channel where you can demand an answer.

I didn’t demand an answer.

This is community place.

It’s called “QNX public newsgroups forums” and I thought it’s allowed to
ask whether a problem is known. If nobody saw that problem yet, so no
answer is ok… and much better than some statements which say ‘no’ in
a wrapped and ‘polite’ way…

BTW, no answer would be even a very clear answer as we are used that no
answer in NA means ‘no’ :wink:


When you post here, you are asking other people here for
courtesy. To give you an answer, related or not, they have to spend
some of their time for you. Do they deserve a little appreciation?

Oh yes … every answer for the topic is highly appreciated.

Now your question “is there a known problem” was simple, indeed.
Simple for someone working at QNX, but not for general public here.

Every reader knows if he knows about such a problem or not, no answers
would also be a clear answer. And that’s really simple, isn’t it?


No matter how much
we “identify ourselves with QNX” (I am sure lot of people at QNX
are laughing) we just don’t know that answer. Yet being good willing
humans (sometimes anyway) some of us were trying to help anyway.
Let’s go to my question now - I asked some simple details about your
hardware setup (not even about your code!) - that may actually be
relevant to the problem, even if the OS is the problem here. Was it
that hard to understand? Or answer? Just out of courtesy?

I don’t expect questions which are out of topic…

I suppose when you walk into a store and a sales associate asks
“can I help you with something” you just tell him that you did not
ask for his help.

What would you say if you ask a sales associate for some shoes and you
will get offered help for some pants? :slight_smile:


My impression is that here are some participants in the NGs

plural !

who identify themselves with QNX and feel attacked personally
if the reason for a problem could be QNX…

it was a general statement to the style of some postings in qdn NGs.
Igor, you know exactly what I mean …

LOL. Armin, have you forgotten already about “the worst attack on QNX ever”
attributed to me?

So you will also recall the reasons… at that time were only a few
Europeans in comp.os.qnx. For some NA guys it was unbearable that guys
from an other continent have been critical to QNX, and when they hadn’t
still technical arguments, they started personal attacks…

Armin

23

Armin Steinhoff <a-steinhoff@web.de> wrote in message
news:3F4A0668.CDAE5ABA@web.de…

Have you actually verified that the interrupt line associated with the
resource manager which does not respond is being asserted (edge/level)
properly, even if the manager doesn’t see it?

Every resource manager works flawless at maximal interrupt load!
But the operating system isn’t able to run two of these resmgr at a
lower interrupt load!

Note: there isn’t only an interrupt problem, also the resmgr_block()
statement of the main thread doesn’t receive a message even if it is
receive blocked.

That isn’t an answer to the question I asked.

Ditto … that’s also the reason why I test my code.

Testing is not a method which yields 100% certainty.

… reading the code of a (standard) interrupt handler again and again
let you understand nothing.

Important is to understand the misbehaviour of the system!

You’re not really interesting in talking further about your issue - you had
a question, and it got answered, period.

Understanding what ‘polite’ means has much to do with nationality if you
believe or not. What’s polite for you can be annoying and extremely
unpolite for an other one. Have a look to qdn.cafe, Jutta followed up
this discussion …

That isn’t what I said. I said that being polite has nothing to do with
nationality. Understanding how to be polite obviously has nationality
involved.

Anyway, thanks for your clear answer to the initial question,

No problem.

-Adam

24

“Armin Steinhoff” <a-steinhoff@web.de> wrote in message
news:3F4A07B8.36EAD31B@web.de…

No matter how much
we “identify ourselves with QNX” (I am sure lot of people at QNX
are laughing) we just don’t know that answer. Yet being good willing
humans (sometimes anyway) some of us were trying to help anyway.
Let’s go to my question now - I asked some simple details about your
hardware setup (not even about your code!) - that may actually be
relevant to the problem, even if the OS is the problem here. Was it
that hard to understand? Or answer? Just out of courtesy?

I don’t expect questions which are out of topic…

That’s not out of topic at all (and I am surprized someone as experienced as
you would even say that). Even if QNX has a problem handling high interrupt
load with multiple handlers (which I do not find unbelievable, we actually
had an interesting discussion with Brian back in february on this subject in
the ddk.network group) then this problem could be likely tied to a certain
pattern of hardware configurations. Hardware has a lot to do with how well
interrupts are handled and you can’t just shrug it off as ‘out of topic’.

But since you’re not willing to say as much as ‘my hardware is …’ then I
don’t think anyome here will be interested in trying to help you any further
either. I think by now most people here have concluded that you’re avoiding
posting the code for some reason. I don’t know what your reasons are, but in
general when a programmer avoids showing the code, that usually does not
speak highly of his confidence in that code.

So you will also recall the reasons… at that time were only a few
Europeans in comp.os.qnx. For some NA guys it was unbearable that guys
from an other continent have been critical to QNX, and when they hadn’t
still technical arguments, they started personal attacks…

You’re confusing national specifics with your personal specifics. Aren’t you
‘identifying yourself’ with ‘Europeans’? And I am hardly a ‘NA guy’ either
… Too many generalisations.

– igor