can't open after fixing smi issues

andypugh
Offline
Moderator

25 May 2012 23:16 #20402 by andypugh

Replied by andypugh on topic Re:can't open after fixing smi issues

pfred1 wrote:

Really Andy, the only time you _must_ reboot Linux is when you want to change monolithic kernel code. Something RTAI isn't.

For fun, save this as fault.comp then sudo comp --install fault.comp

component fault "designed specifically to break LinuxCNC";
license "GPL";
author "Andy Pugh";
option extra_setup yes;
pin out unsigned dummy = 666;
function _;

;;
FUNCTION(_){
//It will never even get this far
}

EXTRA_SETUP(){
*(int*)0 = 0;
}

then:
halrun
loadrt fault.comp
exit

Then try to recover the system so that halrun can be invoked again without a reboot.
I hope you can find a way, it will save me hours.
(currently I am struggling with a slightly more subtle issue of passing fixed-length strings between modules, which may or may not be initialised but the effect is the same)

MESSAGE FOR THE OP:
Don't worry about this, it is not relevant to the typical user experience.

Please Log in or Create an account to join the conversation.

pfred1
Offline
Junior Member

26 May 2012 01:38 #20404 by pfred1

Replied by pfred1 on topic Re:can't open after fixing smi issues

It is your experiment Andy so you try it,

sudo init 1

login as root

init 2

Because quite frankly I don't know the first thing you're on about with any of this. I still give fair odds that'd work. Technically it isn't rebooting either. Although I'll admit it looks an awful lot like rebooting. There may be less invasive fixes though. But I feel like I'm getting a bit of the mushroom treatment now. You know kept in the dark, and fed ....

So that is the best I can come up with for now. Well, maybe not, but it's the best I'm going to come up with considering what I have to work with.

Please Log in or Create an account to join the conversation.

andypugh
Offline
Moderator

26 May 2012 01:49 - 26 May 2012 01:49 #20406 by andypugh

Replied by andypugh on topic Re:can't open after fixing smi issues

pfred1 wrote:

It is your experiment Andy so you try it,

I did, I had to reboot. In fact dmesg said:
[ 4057.142301] RTAPI: Task 1[eed35800]: Fault with vec=14, signo=11 ip=fb598d71.
[ 4057.142305] RTAPI: This fault may not be recoverable without rebooting.

sudo init 1
login as root
init 2

I can try it, I am open to suggestions.

Because quite frankly I don't know the first thing you're on about with any of this. I still give fair odds that'd work. Technically it isn't rebooting either. Although I'll admit it looks an awful lot like rebooting. There may be less invasive fixes though. But I feel like I'm getting a bit of the mushroom treatment now. You know kept in the dark, and fed

You have me puzzled now. I thought you were saying that there were simpler ways to untangle a mess of kernel modules other than rebooting. I was rather hoping you were right.
I just provided a minimal realtime module that puts the system in the state I was describing.

I really do know next-to-nothing about Linux. It might seem odd that someone like me would mess about with Kernel modules, but it's true. I like my hobbies to be beyond my capabilities.

Last edit: 26 May 2012 01:49 by andypugh.

Please Log in or Create an account to join the conversation.

Banana Jack
Offline
New Member

26 May 2012 02:32 #20407 by Banana Jack

Replied by Banana Jack on topic Re:can't open after fixing smi issues

MESSAGE FOR THE OP:
Don't worry about this, it is not relevant to the typical user experience.

yeah i realized that... real fast.
on a side note... :woohoo:

YOU GUYS ARE AWESOME.

what a beautiful, beautiful sight...it's been much too long...

Please Log in or Create an account to join the conversation.

andypugh
Offline
Moderator

26 May 2012 02:34 #20408 by andypugh

Replied by andypugh on topic Re:can't open after fixing smi issues

Banana Jack wrote:

what a beautiful, beautiful sight...it's been much too long...

It's not that beautiful, you still have the "Realtime Delay" message :-(

Please Log in or Create an account to join the conversation.

pfred1
Offline
Junior Member

26 May 2012 03:35 #20409 by pfred1

Replied by pfred1 on topic Re:can't open after fixing smi issues

Things are rarely fun if they're not challenging.

[ 4057.142305] RTAPI: This fault may not be recoverable without rebooting.

In any event reading your dmesg error I think you might have better luck in June, and that isn't so far off now. May is a silly month!

Next thing you know you'll be singing jigs!

Don't ever learn anything about Linux, because as soon as you do some clown will just end up changing it all on you anyways. I just learned termcap no longer exists! I can't make heads or tails out of terminfo either. How am I supposed to pick colors for my terminals now?

I did learn about this bizarre little program called grc though. It doesn't do much for me with dmesg, but it sure can make syslog pretty.

Please Log in or Create an account to join the conversation.

Banana Jack
Offline
New Member

26 May 2012 18:32 - 26 May 2012 23:08 #20428 by Banana Jack

Replied by Banana Jack on topic Re:can't open after fixing smi issues

andypugh wrote:

Banana Jack wrote:
what a beautiful, beautiful sight...it's been much too long...

It's not that beautiful, you still have the "Realtime Delay" message

yeah after i posted that i just knew somebody would notice...i hadn't activated the smi bypass yet at that point. we should be good now. i'm about to go down there to play with it.
and i ALMOST got internet on it too, it blew my mind...followed some tutorial and had no idea what i had just done, and then it worked. :woohoo:

then i unplugged everything and put it back where it should be and then turned it on and it had no idea what the internet was anymore.

Last edit: 26 May 2012 23:08 by Banana Jack.

Please Log in or Create an account to join the conversation.

pfred1
Offline
Junior Member

26 May 2012 22:09 #20431 by pfred1

Replied by pfred1 on topic Re:can't open after fixing smi issues

Banana Jack wrote:

andypugh wrote:
Banana Jack wrote:
what a beautiful, beautiful sight...it's been much too long...

It's not that beautiful, you still have the "Realtime Delay" message

yeah after i posted that i just knew somebody would notice...i hadn't activated the smi bypass yet at that point. we should be good now. i'm about to go down there to play with it.
and i ALMOST got internet on it too, it blew my mind...followed some tutorial and had no idea what i had just done, and then it worked. then i unplugged everything and put it back where it should be and then turned it on and i had no idea what the internet was anymore.

No internet? That can make some things more difficult to do. Even your boot process is affected by not being able to establish a net connection. So there really is no way to get around the impact. Well, you could just leave the machine on I guess, but other than that you'll get hit any time you do boot up. This is a chart of my CNC machine booting up where you may be able to see it establishing a net conneciton during the process:

img87.imageshack.us/img87/2250/wedoct27bootchart.png

Stuff like ifup, and dhclient3 are related to network connectivity. So you might want to configure your machine so it can connect, or it'll be something holding you up at times.

Please Log in or Create an account to join the conversation.

Banana Jack
Offline
New Member

27 May 2012 01:30 - 27 May 2012 01:35 #20438 by Banana Jack

Replied by Banana Jack on topic Re:can't open after fixing smi issues

pfred1 wrote:Stuff like ifup, and dhclient3 are related to network connectivity. So you might want to configure your machine so it can connect, or it'll be something holding you up at times.[/quote]
yeah... i'm working on that...

ArcEye wrote:

i don't actually know what those numbers mean...i may have just put them in myself while setting up.
Try to find your drives here or search for a spec sheet for them.
wiki.linuxcnc.org/cgi-bin/wiki.pl?Stepper_Drive_Timing

so you think i should drop my acceleration?
Yes, your velocity is 1" / sec which is 5ft / min, this is pretty much what my small stepper mill is set to (25mm in my case)

However your STEPGEN_MAX_ACCEL is 37.5"/sec/sec which equates to 952mm . My mill runs smoothly with acceleration of 20mm/sec/sec.

I would just set your acceleration to 2" / sec to start with and then tune to suit

The thing that is going to affect your figures is the slow timing for the driver and the huge SCALE factor.
The first will limit the number of steps per second you can generate and the latter will require a huge number of steps to get anywhere.

Even if you have very fine ballscrews and high gearing on the stepper, I would guess you must have set micro-stepping very high, I would suggest that anything over 8:1 (1/8th step) is unnecessary and counter-productive and I rarely use higher than 4:1 (1/4).
One good reason for that, is that anything higher is interpolated positioning and no matter how good the driver it is far from precise.

Hopefully it will run now

regards

well i dropped my acceleration (yeah it was pretty up there, i don't know why...) but i can't figure out where those 42000s came from. i checked my setup, and they are actually dirhold 2000 dirsetup 2000, which is correct. i don't know why they show up in the hal file as 42000.
and what is this scale you speak of? i have not heard of that. i'm currently half stepping though, if that makes any difference. and my steppers are directly driving my leadscrews.

Last edit: 27 May 2012 01:35 by Banana Jack.

Please Log in or Create an account to join the conversation.

ArcEye
Offline
Junior Member

27 May 2012 08:27 #20441 by ArcEye

Replied by ArcEye on topic Re:can't open after fixing smi issues

what is this scale you speak of?

SCALE is to do with the number of pulses it takes to move one machine unit

linuxcnc.org/docs/2.5/html/config/ini_co...d_sub_axis_section_a

However I was forgetting that you are using inches, we haven't used imperial units in engineering in the UK since the early - mid 70's
(although there was a long messy dual system period)
When your scale is divided by 25.4, then it comes out as something much more like I expected, so forget that, my brain failure.

Looks like you need to re-introduce the smi module now and see if your real-time errors go away

regards

Please Log in or Create an account to join the conversation.

Time to create page: 0.422 seconds