OS X 1.1 Users bits and source, Current Mac drivers for the U12

[Bytewraith]

Here is:
An installer which will install a kext and a Scripting Addition.
Sample Scripts which use the scripting addition.
Source for a Mac OS X version of the LabJack library, scripting addition, and test code.

The code has been updated since 1.0.1. A few memory/resource leaks were fixed. It was updated to the latest library code. The installer is a lot prettier and asks for a reboot (it was too much effort to force it to load).

Dave

Attached File(s)

 LabJack.dmg.zip ( 234.01K ) Number of downloads: 1128

 

[fuelcelldave]

Thanks, Dave, as always!

I was trying to use 2 LJs, one each from a separate application, using the 1.01 libraries, but only the first app launched could find its LJ. Hadn't gotten too detailed with trying to figure out why yet. Will try the 1.1 stuff and see how it goes.

Dave

[fuelcelldave]

the 1.1 source still won't let me use 2 labjacks from two different apps at the same time.

[Bytewraith]

To speed things up the first call to Open() opens all LabJacks and puts them in a table. I suppose it would be possible to make a temporary table of discovered units and a second table of units requested in the Open(). There will still be abritration issues, such as what if both applications ask for the "default" (first) unit? There wouldn't be an easy way for the two applications to say who has what open.

Dave

[LabJack Support]

This topic has been pinned by LabJack Support, as we believe these are the latest user-submitted Mac drivers for the U12.

[pfossey]

Thank you for posting the Mac OS X code.
I am in the process of building a Mac OS X / Windows application that uses the
LabJack. The work you have done has saved me a lot of time.

I have build a simple test application that just asks for the LabJack firmware
version stamp.

The application crashes once in a while.
If I run the program 20 times, it will crash with a seg fault 1 or 2 times.

Before I start digging into the root cause, I was wondering if you are aware of
any problems with the interface code on the Mac.

I am testing on a G5 Dual 2.5 Ghz system running OS 10.3.9.


Thanks,
Paul

[Bytewraith]

There is a problem where on a dual processor system the thread that sets up the notification can take longer than the main thread causing a failure. I have attached code that fixes the problem. Unfortunately, I don't have time to do a complete release package with installer.

Dave

Attached File(s)

 ljackmac_fix.zip ( 57.43K ) Number of downloads: 428

 

[pfossey]

Thanks Dave.

The callback function: MyNotifyProc() does get called eventually in all cases.

I added the following code to the open function to address the issue:
(A semaphore would be better, but this trick works)

// NEW CODE
// Wait to make sure the callback function has been called so we
// have a valid gNotifyPort
int passCount;
for( passCount = 0; passCount < 3; ++passCount )
{
if( gNotifyPort != 0 )
break;
sleep(1);
}

// Now set up a notification to be called when a device is first matched by I/O Kit.
if (!gNotifyPort)
{
*errorcode = OPEN_HANDLE_ERROR_LJ;
return -1;
}

I ran ran my test program several hundred times, and never observed more than
a single call to the sleep(1) in the for( passCount ) loop.

Thanks for your help,
Paul

[Bytewraith]

The code in the zip file I posted has the semiphore added.

Dave

[David]

I have had similar crashing in same routine quite often and have ignored it for now. I have not tried the code posted but this crashing occurs on a new powerbook 1.67Mhz, which is a single processor machine..

[klassen6g]

Is there any way I can get a fixed copy of the complete U12 drivers, with both the fix below and the one by Kris for Intel Macs:
Kris' Intel Mac Fix Thread

I've tried recompiling the source code but only have XCode version 2.5 and it doesn't seem to build anything I can use.

peace,
Grant

Thanks Dave.

The callback function: MyNotifyProc() does get called eventually in all cases.

I added the following code to the open function to address the issue:
(A semaphore would be better, but this trick works)

// NEW CODE
// Wait to make sure the callback function has been called so we
// have a valid gNotifyPort
int passCount;
for( passCount = 0; passCount < 3; ++passCount )
{
if( gNotifyPort != 0 )
break;
sleep(1);
}

// Now set up a notification to be called when a device is first matched by I/O Kit.
if (!gNotifyPort)
{
*errorcode = OPEN_HANDLE_ERROR_LJ;
return -1;
}

I ran ran my test program several hundred times, and never observed more than
a single call to the sleep(1) in the for( passCount ) loop.

Thanks for your help,
Paul