Firmware 2.9.0 RC4 Ready For Testing

[Andrew Stiegmann]

RC4 is ready for testing.  Only two minor patches since RC3.  If you are hitting the track this weekend load this up and give it a go.  Do remember that this is a major release... so BACKUP YOUR CONFIGS.

--

Autosport Labs

Open Source Motor Sports

[Max Kosenko]

https://www.autosportlabs.org/viewtopic.php?p=23932#23932

[Max Kosenko]

So 2.9.0 RC4 hangs much stronger than 2.8.7 with that script. It's not responding on a push buttons during start, firmware change and resetConfig, while 2.8.7 is still manageable.

Script doesn't have anything vehicle specific running, so you should have no problem repeating behavior. 

[Andrew Stiegmann]

> So 2.9.0 RC4 hangs much stronger than 2.8.7 with that script. It's not responding on a push buttons during start, firmware change and resetConfig, while 2.8.7 is still manageable.

Sounds like a stack/memory issue again.  I assume that without the script things are peachy?

--
You received this message because you are subscribed to the Google Groups "AutosportLabs Beta Testers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to autosportlabs-beta-...@googlegroups.com.
To post to this group, send email to autosportlabs...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/autosportlabs-beta-testers/1c1aaf11-4a61-45b3-8e6a-480ea9f5b1dc%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

[Max Kosenko]

W/o script they are fine. Script was working - I made tiny small insignificant change and it's gone. 2.8.7 reacts differently on this script. Try it...

[Andrew Stiegmann]

Hey Max,

   Do you have a version of this script that is not minified? 

[Max Kosenko]

Yes. https://gist.github.com/kosenko-max/059c125aafdc0a8a4f0c12c5cc4ab09c

[Andrew Stiegmann]

So took some time to look at this script.  It seems the script will get stuck in an infinite loop between lines 54 - 58.  With the canCon string being "0123456789ABCDEF", none of those ASCII characters yield a byte value of 44 (','), 59 (';') or 64 ('@').  Hence the counter will keep counting up, and i will go beyond the length of the string.  When this happens, string.byte will return no value, and thus we stay in this loop forever.  With that print statement there, you are blowing out the RCP log buffer, sending an endless stream or "\r\n" characters back to our app, which slows down trying to process them.

Right now there is a patch in progress to allow users to flash the script if it gets stuck like this.  See https://github.com/autosportlabs/RaceCapture-Pro_firmware/pull/504 for more info.  I intend to get it into 2.9.0 and resolve this issue. 

[Max Kosenko]

Wow. Thanks for finding out. I didn't thought I actually made infinite loop and overloaded log file. I still think it's a good test for finding some bottlenecks that make system unrestorable. So if your patch will fix it - that would be great. I believe scripting should not render system unusable anyway. It was especially scary to be unable to resetConfig through the terminal.

[Andrew Stiegmann]

> I believe scripting should not render system unusable anyway.

Completely agree.  Hence why I'm fixing this before 2.9.0 goes out the door.  I had never hit this problem until recently... and I can see how much of a PITA it is.  Here is the latest.... try and see if you can get lua into an un-programmable state.

[Max Kosenko]

Just tried - log viewer is hanging (I see you have ticket about that), but no disconnections and can update script. Thanks!