0.10.19 crashed

[RomChip200]

There's something strange. I erased everything, incl. registry entries, to give it a try.

I freshly installed Nistune and then, Logworks3.
Then, I launched Nistune, config. menu and it crashed ..... yes, config. menu doesn't even display.
So, I had a careful look into the registry entries, I noticed WIDEBAND_COMPORT1= 2 and WIDEBAND_COMPORT2= 2.
My job PC is docked and has only one COM port on the docking station, nothing USB related plugged in.
I changed the values WIDEBAND_COMPORT1/WIDEBAND_COMPORT2 to 1 (just because COM2 doesn't exist).

I re-launched Nistune and this time, config. menu is displayed.
I tried to reproduce the issue by putting back 2 in WIDEBAND_COMPORT1/WIDEBAND_COMPORT2, but it doesn't crash anymore and NIstune detects COM1 correctly and affects it by default to the widebands.
Innovate is also detected correctly now.

There's something latent related to COM ports .... but I don't know what.

[Matt]

Strange. I'll see if I can repeat on my Win7 machine here

With regards to Dr Watson missing on Win Vista/7 it was replaced by 'Problem Reports and Solutions' which doesnt actually record any crash logs/dumps directly on the PC but sends them to Microsoft if the application is signed
http://windows.microsoft.com/en-US/wind ... -Dr-Watson

Basically the code needs to be signed using Verisign ($500USD/year just for the signature) and submitted to Microsoft via the net. Using the signature via Winqual I can then retrieve everyones reports. I'm working down the path of digital signing for the application and FTDI drivers (modified INF files for PID/UIDs require signing) but need to complete Microsoft logo testing etc for each software release which is intended to be completed for the next stable version

In the meantime I've found a crash debugging report solution (sent directly to me via email) to get the current problems and their dump files for the test releases so will put that in the next version

[Matt]

Config menu was not displaying because I think the MTS OCX registration was screwed up. Do you see 'Innovate MTS' now in the drop lists for wideband unit? This shows MTS is now detected and working

Wonder what were WIDEBAND_PORTIDX1 and WIDEBAND_PORTIDX2 set to in the registry? This is what MTS uses to store the indexes (rather than COM ports when other wideband units are used)

[bachig24u]

Matt I moved on to the next release. 0.10.20.
still having nightmares with LC-1 thru nis.

seems fine in logworks 3.07. but add the lc-1 to nis and it all falls apart rapidly.

just upgraded logworks to 3.08 and ran nis with LC-1.. so far, 2 minutes of engine idle data and no crash or consult data receive failures.

will start adding to the 0.10.20 thread.

[RomChip200]

Config menu was not displaying because I think the MTS OCX registration was screwed up. Do you see 'Innovate MTS' now in the drop lists for wideband unit? This shows MTS is now detected and working

I checked in the log, MTS is now detected correctly and Innovate MTS is now in the drop list.

Wonder what were WIDEBAND_PORTIDX1 and WIDEBAND_PORTIDX2 set to in the registry? This is what MTS uses to store the indexes (rather than COM ports when other wideband units are used)

They were 0 0 and they are still 0 0. I haven't even connect the PC to the widebands yet. This is a PC for my job, so I mainly use Nistune to check some logs.

[Matt]

just upgraded logworks to 3.08 and ran nis with LC-1.. so far, 2 minutes of engine idle data and no crash or consult data receive failures.

so the version of logworks from 3.07 to 3.08 improved your crash problems (so far) when using Nistune. It makes me wonder if there was a crash fix in 3.08

It is possible that crashes when using MTS with Nistune come from the DLL rather than Nistune. I have noticed this during previous debug runs with Innovate and seeing the stack trace die in the MTS.DLL component (which is called from Nistune code)

I'll wind back my logworks installation and try and test that theory since trying to cause crashes on the bench isnt replicating the issue at the moment.

[bachig24u]

doesn't explain why I've never experienced a logworks crash in all the last 4 years.

Has anyone else experienced logworks crash?

[Matt]

I've seen them occasionally on the R31 but having Type 1 and MTS connected its hard to tell where it comes from. However the MTS crashes I've seen on the bench looked like a logworks issue

[Matt]

Okay I've identified a problem with the Disconnect and usage of consult threading occurring causing a crash. I'm going through rewriting these sections of code to handle this properly. Hopefully this will resolve this issue

[bachig24u]

Thank you.

btw, just em'd a 2MB video of a input tracer issue from the other week.

[Matt]

okay thanks will look at it. Some of the threading stuff (fairly old code) needed cleaning up. I've pulled all the re-connection code out and rewrote it tonight to do the job better

[Matt]

An update. More work done on the threading code for Type 1 and now completed. There are quite a few code paths where there could have been potential issues so I've gone through each one

[Matt]

Still got another crash tonight on Type 1 whilst in the debugger during disconnect errors

Appears the FTDI DLL/driver is barfing on us when we get an unexpected disconnect. We get a read error from the board, but then attempt to close the device (as normal)

However the driver has hung up on us and the 'handle' is no longer available. Trying to close with the handle not available causes the crash. I've put a check in for NULL handle but this might not be enough (not sure how else to know the handle is bad)

So in addition to checking the handle I've wrapped the Close() code in an exception handler around it to limit the scope of the problem and it taking down the rest of the software

I've also seen the software 'hang' and found its getting stuck inside MTS.DLL . There is an infinite loop in the DLL (even though I'm not using the LC-1 at the time) which hangs the caller application (Nistune)

Not sure why its doing that but perhaps the 3.08 version of Logworks has fixed that issue

[johnus]

0.10.19 crashes before it even loads for me , giving me a ms error report ,i'm running win xp 32bit ,installed by upgrading 0.10.13R then tried fresh install same result. Gone back to 0.10.13R for now as i've fuelling issues to sort out, but hope you can help me sort it before 10.13R runs out.
cheers Paul

[Matt]

Will probably extend 0.10.13.R in the meantime if I cant sort that one out

Could you please let me know if 0.10.14 does the same thing? I'm seeing a crash running under Wine through Ubuntu Linux from that version onwards on startup so might be related.

Also added crash report emailing to the current version yesterday to send me a copy of where the crash is occurring so I can locate the cause quickly for resolving it

http://www.nistune.com/secret_squirrel/ ... _setup.exe