PDA - Memory Leak LV 8.0.1 PDA using Slide Indicators

[Greycat]

I originally encountered this error when, for no obivous reason, my application would crash after a certain amount of time.  I began to write a message on this board and while I was writing it, it occured to me that this crash certainly reaked of a memory leak.  I went back to my application and added in some memory monitoring displays and sure enough there was a leak.  After about 2 hours of removing everything else that I thought could have possibly caused the error, I removed the slide indicators and low and behold ... Memory Leak gone.  The Vi speaks for itself but I would really like some NI feedback on this.  I'm using LV PDA 8.0.1 under WM5.0 on an iPAQ hx2490.  The leak also occurs with a tank indicator, any slide indicator, a control slide changed to an indicator, and a dial indicator.  It does not, however, occur with the Meter indicator or the thermometer indicator.  Thanks in advance for the feedback NI.

If anyone else would like to verify that this is truly the case and I'm not just seeing things that would be great.

Greycat

[M_Kunze]

Hi Greycat.

I havent seen this problem before. I am using evc 4 SP 4 and not the
SDK for windows mobil 5. My target runs Win CE5.

It sounds a little bit strange to me but I had this problem with picture controls.
After Investigation the NI LV routines I saw that adding new data to a exiting picture
costs the amount of memory that is added to the picture. It's not the way programmed
a that the memory is released after the manipulation of the picture.

I know this is a litte bit of topic 😉

[Greycat]

Should not the memory used by a variable be reused rather then using new memory everytime something is updated, like in a loop for instance? Or, if it does require the cost of having the new data and the most recent data in memory at the time of update, the portion that is no longer used should be released after processing is complete. This is usually standard procedure, so you don't run out memory.  The fact that this eventually leads to a crash indicates something is definately not right.  Did NI address your problem with the picture controls?  If so, what was thier response?  Did you find a solution to your problem?

Greycat

BTW - Thanks for the response

[Greycat]

Any comment from NI?

Greycat

[tst]

---

@Greycat wrote:
Any comment from NI?

---

Not likely during the 4th of July, and I wouldn't be surprised if some of the americans would take the rest of the week off as well.

[Greycat]

Ah Yes ... thats right ... well maybe I'll bump it in a couple of days ...  Thanks tst

Greycat

[Chris_Cilino]

Greycat,

You couldn't be more right. I was able to duplicate exactly what you're seeing. There is certainly a memory leak when you have at least a slide indicator wired up on the block diagram. I'm going to file a bug report to R&D. THANK YOU so much for the code you posted that CLEARLY demonstrates the issue. I'm really sorry for the trouble and I very much appreciate the time you took to trouble shoot your code, find the problem, and even explore the problem. We really appreciate your patience.

Chris C

National Instruments

Applications Engineering

[sarahk]

Hi, Greycat.

I've never heard of anything like this, and I can't find any reference to it internally, either. I'll get a PDA and try this out and let you know what I find. In the meantime, has anybody else seen this behavior?

Whether or not it's reproducible, thanks for bringing it to our attention!

[Greycat]

Great ... I look forward to some resolution to this ... it makes running report screens in a loop a little difficult when you are restricted by what you can put on the front panel.  Thanks for looking into this and taking it to R&D.

Greycat

[Chris_Cilino]

Greycat,

Make sure to keep a look out for updates at the Drivers and Updates page.

Chris C