Risk of static conditions of race FPGA

Similar Questions

• Performance of Modbus using DSC static Variables

  I'm fairly new to using Modbus with LabVIEW.  On some dozens tools and APIs that can be used for a project, I train, I decided to try using an alias of shared Variables in Modbus registers in the project, which is a tool of DSC.  It seemed like a smart to go way.  I used Variables shared in the past, however, and I am aware of some of the questions that surround them, especially when the number of them begins to increase.  I will have only about 120 variables, so I don't think that's not too bad, but I'm getting a little worried...

  The way I started to do this was to create a new shared variable for each data point.  What I have noticed since then is that there is a mechanism to process multiple records at once using a table of values.  (Unfortunately, even if I wanted to use the table method, I probably couldn't.)  Modbus points that I am interfacing are for a custom device and the programmer does not disturb the consecutive use records...)  But anyway, I was wondering what might be performance issues that surround this API.

  I guess:
  (1) all caveates of shared variables apply.  These are really shared variables, but only DSC taught the engine of the SV to go read.  Is this fair?

  And I wonder:

  (2) there is no improvement in performance for reading a table of variables consecutive rather than reading each variable individually?

  (3) there performance above problems what shared variables have normally when you use Modbus specifically?  (E.g. how many times can you read a few hundred points of the device even Modbus?)

  Thank you

  DaveT

  Hi Dave,.

  Thanks for your good questions!

  (1) you are right that the caveates of shared variables apply. Generally, the major issues (conditions of race etc.) are not met because these variables are generally used as I/O.

  (2) with a large number of shared variables that are located on separate machines, it is best to use a table. However, with 120 variables read consecutively will not affect your overall performance, especially if you don't need to read everything at the same time.

  (3) overall, there is no any concerns of performance outside normal common variable concerns. Modbus and DSC are designed to be efficient and do a good job to maximize performance for you.

  I hope this helps.

  Best regards

  Anna L

• FPGA dynamic Refnum

  I am trying to auto index a table of fixed-size FPGA of IO constants. The wire and the loop tunnel in the loop for the error even "possibility of dynamic Refnum not supported for the current target. What does that mean and what someone has suggestions to work around this problem.

  Hey Waldo, I would recommend having a look at the following KBs. Occurancs and refnums must be statically defined in LV-FPGA applications. This is by far the most common reason for this error.

  By passing an instance of a Subvi in LabVIEW FPGA

  http://digital.NI.com/public.nsf/allkb/30496B921D344B8686256FEF005224D6?OpenDocument

  Can I get a reference to a Subvi, which runs in the FPGA?

  http://AE.natinst.com/operations/AE/public.nsf/Web/searchinternal/9bb42c865405ef3e86256fc6005107ed?O...

  I hope this is useful,

  Best wishes

• Not updated by Labview VI Step property

  Hello

  I am currently using 4 TS and LV 8.2 with Win7 Pro.

  I created a custom step which employs a lower Edit level.

  The lower Edit level name successfully a Labview vi, which allows the user to change the values.

  The intention is to copy the user values in the properties of the custom calling step, using the value property Value.vi of the library of Teststand.

  After that step is changed, the vi is closed and I check the step variables expected variables to be updated, but what I see in the Variables Panel does not always show the new property values of step.

  Called by the lower Edit level Labview vi receives the context of the sequence and the correct step property string name.

  Have I correctly picked the Value.vi property set to the task of updating the step of step of the custom properties?

  In addition, when the properties of the stage show the new value, the sequence containing the custom step does not indicate

  the need to save that is in front of my expectations, else how is remined user to save the changes?

  Thank you

  Gary.

  Gary,

  I recommend most often to developers to interface with the EditSubstep modules using parameters.

  For an input parameter, I pass the current values of the step variables.

  For the output parameter, I connect the same variables that they get old/new values to the EditSubstep module.

  I declare this as my EditSubstep modules normally contain a button "Cancel". If the user presses that, I simply passes the output values that the module received as input values.

  Input values are also used to update correctly orders from the EditSubstep module. Otherwise, the user would see only the default values, but not the values of that step has already been changed.

  Output parameters are important have not changed (cancel) / update the values of the step. By selecting the appropriate container, the values are transmitted with the closure of the EditSubstep, so there is no risk of running into a race condition, which could explain the behavior you're seeing.

  hope this helps,

  Norbert

• Synchronization with a server of VI

  I currently have a server / multi-client application I implement with server of VI in LabVIEW 8.6.1.  Call to the server to execute different screws to store/retrieve/process data and the screw store data in files as well as globals function, i.e. an uninitialized moving records within a while loop.  I have a few questions about this implementation:

  1 are the calls made via VI server synchronized by VI Server; in other words, not realizing the reentrant Server VI VI never will launch multiple instances of the VI on the server system (I won't for obvious reasons)?

  2. I currently use some of the screws in my app server as static screw and also allow others to call in some of them from remote hosts via the server of VI, and I am concerned about the integrity of the data within my service globals (I took the precaution of open handles for all THE screws I share with VI Server so that they never go out of scope to be used as static or not).  Is this bad?  If I used only the VI server calls in my server application, it would be better (instead of using the 'shared' statically screw)?

  3. when a remote client is called on the server via a VI server, what happens when the VI is busy?  Does my remote call via an error call-by-reference node, or it fortunately waits his turn until the VI can be accessed (assuming that the #1 question is that one instance can never be executed).  I do have several systems to check the conditions of race.

  4. I use this implementation of VI server between Mac and Windows, do I need to be careful (in addition to the obvious path problems)?

  Thank you!

  -Danny

  TarantulaDiaz wrote:

  not realizing the reentrant Server VI VI never will launch multiple instances of the VI on server system?

  Only if it's a LIVING.

  In addition, even if it is reentrant, the VI server will not open multiple copies unless you set the flag in the options of the Ottawa Valley Railway (you can see the exact value using the primitive).

  If I used only the VI server calls in my server application, this would be better?

  I prefer the static approach for three reasons:

• Display of sizes of different icons on different devices

  I'm trying to dynamically set a different icon based on the device. I've set up another point of entry in the xml of the application descriptor file. It passes an argument of "icon" in my application main().

  The entry point is triggered as expected (installation of the application and to the power of the cycle), but he has problems when calling updateIcon() on some phones (for example BB 8900 T-Mobile version 4.6.1.231 (Platform 4.2.0.108)). Looks like it fails on phones with OS 4.7 and works very well on phones on or above this level.

  The error I get is IllegalArgumentException: "Module with handle [XXXX] and [X] index has no application entry point. Try to set the index to both 0 and 1 on updateIcon() has the same problem.

  It is only happening when I install the application OTA (works on the 8900 charging cable).

  I have my code below from from the section of this topic:

  public static void main(String[] args) { if ( args != null && args.length > 0 && args[0].equals("icon") ){      MyApp instance = new MyApp(true); } else {      MyApp instance = my MyApp(false); } instance.enterEventDispatcher();}
  
  public MyApp(boolean autostart){   if(autostart){        Bitmap loadedIcon;        int width = HomeScreen.getPreferredIconWidth();       int height = HomeScreen.getPreferredIconHeight();
  
        //conditional logic to initialize "loadedIcon" based on device icon width
  
        final Bitmap icon = loadedIcon;       invokeLater(new Runnable(){                 public void run(){                    ApplicationManager myAppManager = ApplicationManager.getApplicationManager();                     boolean keepGoing = true;                     while (keepGoing){                        //Check if the BlackBerry has completed its startup process.                          if (myAppManager.inStartup()){                            //The BlackBerry is still starting up, sleep for 1 second.                            try {                                 Thread.sleep(1000);                           } catch (Exception ex) {                                //Couldn't sleep, handle exception.                             }                         } else {                              //The BlackBerry has finished its startup process.                            //Set the rollover icons.                             try {                                 HomeScreen.updateIcon(icon, 0);                               HomeScreen.setRolloverIcon(icon, 0);                              } catch(Exception e){                                   System.out.println("setting icon caught: " + e.getMessage());                                 e.printStackTrace();                            }                             keepGoing = false;                        }                     }                     //Exit the application.                   System.exit(0);               }             }); } else {      //load the app normally, with UI. }}
  

  I use Eclipse plug-in 1.1.2 to build my app, but article is older, obsolete 1.0 plugin, maybe the instructions no longer apply and must be updated?

  Understood the problem, basically it is a condition of race as described here which is also in addition to restarting the device. It happens when the application is installed OTA and the entry point to attempts to define the icon when the application is not fully registered with the device. In a Word, it is the valid response (in invokeLater):

  int failedTries = 0; //failed attempt counter (maximum of 5 attempts for this example)public void run(){                     ApplicationManager myAppManager = ApplicationManager.getApplicationManager();                     boolean keepGoing = true;                     while (keepGoing){                        //Check if the BlackBerry has completed its startup process.                          if (myAppManager.inStartup()){                            //The BlackBerry is still starting up, sleep for 1 second.                            try {                                 Thread.sleep(1000);                           } catch (Exception ex) {                                //Couldn't sleep, handle exception.                             }                         } else {                              //The BlackBerry has finished its startup process.                            //Set the rollover icons.                             try {                                 HomeScreen.updateIcon(icon, 0);                               HomeScreen.setRolloverIcon(icon, 0);                                          keepGoing = false;                                          //Icon should of been set at this point                             } catch(Exception e){                                 System.out.println("setting icon caught: " + e.getMessage());                                 e.printStackTrace();                                          failedTries++;                                    try {                                     Thread.sleep(2500);                               } catch (Exception ex) {                                    //Couldn't sleep, handle exception.                                 }                                          //Since the attempt failed, we sleep for two and a half secs and try again.                              }                             if(failedTries > 4){                               //Worst case. Should revert to a default icon if it fails to be set after five tries.                                 keepGoing = false;                            }                         }                     }                     //Exit the application.                   System.exit(0);               }
  

  Catch the exception, and then sleep for about two seconds and leave the loop to set the icon again, because it will succeed on the second or third try (at that time that the app must be registered with the device, including its entry point). In addition, to ensure a maximum test because you do not want to create a possibility of an infinite loop if for some reason, it fails to set the icon each time.

• My version of FF running on Windows XP, SP3 was just updated (by FF, apparently) now it crashes regularly (4 times in the last 2 hours) what is happening. ?

  Yesterday (19 Feb) FF worked very well and I had not experienced a crash in a few weeks. This morning, the program had crashed, and when it is retrieved, he said that it has been updated. Since then, he has crashed to 4 or 5 times, usually just in a static condition (not to raise new pages). My computer has not changed. The only add-ons currently running are McAfee Site Advisor 3.6.0, 2.6.5.7 Noscript and Xmarks 4.1.3

  The reset Firefox feature can solve a lot of problems in restaurant Firefox to its factory default condition while saving your vital information.
  Note: This will make you lose all the Extensions, open Web sites and preferences.

  To reset Firefox, perform the following steps:

  1. Go to Firefox > help > troubleshooting information.
  2. Click on the button 'Reset Firefox'.
  3. Firefox will close and reset. After Firefox is finished, it will display a window with the imported information. Click Finish.
  4. Firefox opens with all the default settings applied.

  Information can be found in the article Firefox Refresh - reset the settings and Add-ons .

  This solve your problems? Please report to us!

• MacBook Pro 2012 sporadically slowdown since the upgrade: SSD ALL SSAY!

  Hello

  I recently updated from a Samsung 850 Evo 256 GB SSD to the 1 TB version. Everything worked perfectly on El Capitan before the upgrade.

  Since then, my Macbook (15-inch, mid-2012, 9.1) is very reliable. It has constantly been extremely hot, that fans speed up at bottom, unless they are overridden by third-party applications (fan SSD command), and even simple things like streaming video may cause the thing to grind to a halt. The battery life is also significantly more bad for exchanging the two SSDS.

  1. almost directly after installation of the SSD using my Time Machine (from the 256 GB SSD), underside of the Macbook backup was extremely hot to the touch and the fans were at full blast all the time. I have reset the SMC and downloaded SSD fan control to override the fans. This seemed to help for a few days, but then...

  2. on the accusation, it slows to a halt, which makes the Macbook unusable. I can't understand why, because it is not as if a particular application is the cause on several occasions. I noticed that kernel_task consuming CPU during one of these "slowdowns", so I tried the fix involving passing the model identifier of IOPluginPlatformFamily.kext after you turn off the csrutil of the recovery partition. But my model identifier was not yet which are listed there, there is nothing to move.

  3. the SSD seems to work very well. Repeated with BlackMagic Disk Speed Test assessment tests show reading and writing almost 500 MB/s.

  3. I ran the Apple hardware test and he told me that he had "no problems found".

  4. I have checked inside the case for excessive dust and has no problem.

  I scoured the internet to find answers and tried every possible solution I found, everything to nothing does not. Any help would be greatly appreciated - I am considering looking for a new laptop if I can't understand what is the problem, as I can't work

  I partitioned the SSD drive immediately after the Time Machine backup - this could be the problem?

  Thank you very much for reading this and for all the advice!

  Tom

  A number of users have had problems with the new Samsung SSD.  Note that Samsung omits Mac compatibility on their web site.  Whether or not this SSD is the problem is unknown at this time.  You might consider to replace it with the prior SSD and see if there is a change.

  First of all, I suggest that reset you the SMC and NVRAM:

  https://support.Apple.com/en-us/HT201295

  https://support.Apple.com/en-us/HT204063

  Try a safe boot:

  https://support.Apple.com/en-us/HT201262

  If no change, remember to reinstall os x.

  txl879 wrote:

  the fans speed up at bottom, unless they are overridden by third-party applications (fan SSD command),

  Application of fan control do you use?  The reason why I ask, is that all of these applications I know, allow to increase the speed of the fan, does not decrease the speed.  To do this involves an element of risk because the conditions may require the speed high ventilation and by the user, reducing can result in thermal shutdown.

  Ciao.

• Addition of new global variables

  I change code that uses one or more global variables in a global folder of vi. I would like to add a few variables more and potentially change what exists and ask them to be in the same globals vi. I can easily add a variable of the same type by copying and pasting a variable of the same type on the Panel before the globals vi, but I can't add variables of different types.

  This post comes closest: http://forums.ni.com/t5/LabVIEW/how-to-associate-existing-global-variables/m-p/1079535/highlight/tru... but I did not understand this line:

  • Add FP (there may be several elements in a global variable VI)

  I read some of the warnings against the use of global variables as it can cause race conditions / lead to General illegibility. I'm curious what is recommended if global variables are used as the constants that are defined at the beginning of the program. In general, my LabVIEW code gets pretty unreadable if I have a lot of threads for all of these constants. Is there something better?

  Thank you!

  Kristen

  Hello kllurie,

  The scenario you mentioned (using globals to replace duplicate constants) is probably the best use cases for them - conditions of race etc. problems arise when they are used to the stream or in several scenarios of the writer.  Readability is perhaps a matter of concern, because it is not immediately clear what could be reading from or writing to a global when a thread would suffice.

  With regard to the modification of the globals vi, all you need to do is drop objects on the front of the range of controls (right click).

  Kind regards

• HSDIO - deterministic written submission during the generation

  Hello

  I use a card HSDIO (PXI-6542) to control a device (DUT) with a predefined fixed length, waveform continuously running.

  That maybe had TO actually benefit from the dynamic changes in digital signals according to the measurements on the INSTRUMENT itself.

  At the moment I just pass off the coast of the ASE and map HSDIO, write the new waveform on the map, turn to ESA and the generation of digital signals.

  Of course, this is not optimal and I would like to change the way signal 'live', while the DUT is running.

  Just crush a waveform, whereas it is generated should not be working smoothly due to a condition of race as explained here: http://digital.ni.com/public.nsf/allkb/14CE41C9CB9F10A88625766A005CEE47

  I think that I've found a workaround, but need a confirmation about this.

  Instead of just a waveform, I could use 2 and select which is generated using a trigger.

  The corresponding script might look like this:

  script myScript
  
    Repeat forever
  
            If scriptTrigger0
  
              generate a waveform1
  
            on the other
  
              generate a waveform2
  
            end if
  
    end repeat
  
  

  For example, the idea is to generate a waveform1.

  Then according to that CSA should I calculates waveform2 and transfer it to the HSDIO map.

  I think here is overwrite waveform2 with the data of the same size will not create any small problem because it is not currently used to generate signals.

  After this using the trigger, I could start using waveform2. If I need additional adjustment of wave shape I could do this by editing waveform1, so it is not used for the generation and thus now alternating serves really what waveform.

  Am I correct in that a waveform in memory but are not currently used can be replaced without causing glitch on an another waveform that is currently?

  Kind regards

  Baptist

  Hello, Baptist,.

  Yes, you are right. It is a method to dynamically change the waveforms on the fly that was already used in the past. If you waveform1 to generate and download waveform2 then it is idle, it will not affect the generation of waveform1. Then, triggering via scripttrigger to waveform2, then you can download the next waveform, you want to use for waveform1 without affecting the production of waveform2.

• Reference open vi

  Hello

  can someone tell me why LabVIEW does not execute the function "open reference vi" when I interact with for example the context menu? I have attached a picture for more understanding.

  

  Kind regards

  Thomas

  "Root loop."

  See here.

  in order to keep things straight avoid the conditions of race, dogs and cats living together, there are some operations that take place in what he calls for 'root loop' that everything else is blocked. See the link I provided and then on the link sinned this thread.

  Ben

• command or the indicator?

  I have a Subvi used by the user to change some parameters used by the main vi. That is to say, click on settings button and the Subvi appears.

  Anyway, the Subvi reads a text file to load the previously saved settings.  It is, the text file must be put in order, or the indicator?  The user must be able to change the values in the Subvi Panel and then save again.  Looks like I can do this with the orders or the lights.

  -Mike

  KSU Flyer says:

  Looks like I can do this with the orders or the lights.

  Not quite. Indicators can only be changed by a user when the VI is running. So, in your case, you would like to use controls. When you read the file for the saved values you can simply update the values of the controls by using local variables or property nodes. Be careful about the conditions of race with local variables.

• Can I drag the event target source drag?

  I have adds drag and drop features of a tree control on a VI at several targets of slipping on different VI.  For example, they can drag the tree for a XY Chart on another VI, a table to an another VI, a digital indicator on an another VI, etc...  In addition, when they abandon the data, I need to know where he left it down and initiate an action from the drag source (tree control) VI.

  My problem is, I see no way to determine what VI and control the move action took place the.  I'm basically looking to the control reference of the operation drag and drop target once the drop event occurs.

  I know that I should add a few cases of event for all the different targets, but I try to keep the minimal code and keep most of the complexity of the VI with the source.  If I have, I will use some sort of a global or the driving force to write an identifier of the target VI to a place where the source VI can be read, but I'd rather not do that in many places so I don't have to.

  Create your own event type user 'Control' and save it in the same diagram as source Tree. On the event of filter "Drag start?" of the source tree, inject this event Ref use in drag data (it will take be flattened). Then, in the 'Drop' unflatten handler this Ref user event and send event user with the value of the event data "Ref" of this event handler 'drop '. It is a form of a callback function, which is much more reliable than a global engine or action that exposes the risks of racing conditions.

• Writing only to certain parts of the cluster in an array by reference

  Hello

  I have an array of clusters that I use as well to view and enter data, i.e. elements of the cluster are unmodifiable (disabled) controls used as indicators of "false" (numeric values, strings, LEDs) and some are normal witnesses (numerical values, buttons).

  The "indicator" part of this cluster table must be constantly updated (for example with the positions of the engines), normally by replacing the elements according to cluster in a loop and writing in the table. This can lead to conditions of race with the part 'control': If the user enters data in a 'bad' timing control, it gets immediately replaced by the old value - in this case the update process began just before the user input and completed just after the entrance of (which may be a simple click is enough) , so the old values of controls are rewritten on user controls, as the table should be rewritten as a whole.

  Now, I'm looking for a solution to this critical race condition without changing this 'mixed' approach of control/indicator (e.g. by control and indicator tables separated next to each other and paired scrolling or so - which would make it much more horrible GUI design, among other disadvantages). I know that it is possible to change the value of an element in the cluster without having to rewrite the entire cluster using the element reference. However, it's more complicated if you have an array of clusters, because you want the cluster to a certain index table and to my knowledge, there is no such property that gives you for example the reference to the item table located in an index of certain (who is a lack of long date in the table of Labview manipulation). If you change the 'value' of an element of the cluster property in a table, it seems to affect the 'last active' element of the array, at least it is correlated with the array element that was clicked on last. Maybe there's a way somehow programmatically set that ' last active ' array element and browse the table in this way (setting the flag "correspondent" elements of the cluster by reference) or maybe someone knows a solution 'Nice' and elegant?

  I hope you understand what I mean

  Thanks in advance!

  To avoid such conditions of race, make sure what you write on the Board in the same place change you and write only the data for the indicators. Practically, this means that if you have an event for the control change value, you must have another case of event in the same structure of update of the indicators (timeout or a user event) and make sure that the indicator event takes the rest of the data of the current value (for example through a terminal or a local variable or the DVR suggested Steve). Similarly, you can use the terminal control to the current value of the control for the indicators rather than depend on the event containing the correct value.

• Tab control sequence

  Hello

  I learn LabView and until I started to make applications I couldn't see how complicated it is.

  I tried to make VI to choose in the order of the parameters. After several attempts to solve the problem, the VI it works badly all the time. Could you give me some ideas how to do this? What architecture you choose?

  The idea is:

  -After starting VI Page1 is visible and active

  -When the conditions for Page1 are correct, we could push the next button and go to the next page

  -Page 2 is visible and active, and so on

  -When we on Page4 (should be called "Summary"), this summary page of ist, which I prepare yet

  -at any time, we should have the ability to go to any page using the tabs

  The problem is:

  -the following buttons are dimmed when I select the previous page

  -There is a problem with reliability, sometimes I push the next button and nothings happen or page is visible but not active

  I welcome all suggestions.

  Concerning

  GerdW has a very good point. You can move the structure of the event outside the structure of the case. Also, having a while loop within another time loop is not indicated.

  If you have, it's just an event structure and a structure of cases, all within a single while loop. And with the structure of the event to the outside of the structure of the case, you might have a thread that changes States and not a local variable (it is recommended to use those when you have no other choice, because not using them can save you a lot of headaches when it comes to flow conditions and race data).

  What I would use for such an application (and I would try to make it more modular as possible), is the architecture of producer-consumer, given the structure of the event in the producer and the structure of matter in the consumer. Something like this. You can also find some examples of programming HMI LabVIEW examples Finder, under Create User Interfaces

  Kind regards

  Rare