HDL Coder failing for single-precision

I have a single-precision sub-module that is failing HDL generation when using the native floating point library with the following error in R2019b (both Update 4 and 6):

Compilation errors when generating code for: model/subsystem\reset_in Error during MATLAB code compilation: Error Path: /usr/local/MATLAB/R2019b/toolbox/hdlcoder/hdlcommon/emlauthoring/wireops/hdleml_switch_multiport.p Error Location: (@Row: --> 1 @Column: --> 1) This assignment writes a 'double' value into a 'single' type. Code generation does not support changing types through assignment. Check preceding assignments or input type specifications for type mismatches. Internal Error: This error occurred inside a MathWorks function.

I have a work-around so this isn't blocking me but can someone from Mathworks take a look "under the hood"?

Thanks,

Andy

Answer 1

[Andrew Nicol]

Adding more context: with a different model, I encountered this same issue. Interestingly, I can make it go away by unselecting the "Minimize intermediate signals" option. Hopefully that will help anyone else that gets hit by this...

Answer 2

[Kiran Kintali]

Thank you for reporting this error. This is unexpected.

Can you share a reproduction model? It will help verify this issue is resolved and/or publish a workaround or patch for the OriginatedIn release.

[Andrew Nicol]

Unfortunately, I can't share the actual model that I'm getting this error in and, so far, I've been unable to get any other models to throw the same error. Is there any other debug information or anything I could pull from the generation process that might help?

[Kiran Kintali]

Thank you for the response.

You can share the following information to help the development team isolate the issue and make sure it is resolved in the 20a or one of the upcoming releases.

1. Release you are into this issue (output of ver command)
2. HDL Coder options used during code generaiton hdlsaveparams('model/path_to_dut')
3. makehdl('model/path_to_dut', 'debug', 1/2/3/4) one of these debug levels sometimes isolates the issue and points to the block running into the issue.
4. If you already have a work-around (such a block replaced) you can provide a sample model with just the block that got replaced etc., if it can reproduce the same issue.

You can reach out to me [e-mail] or [e-mail] for further assistance if any needed.

[Andrew Nicol]

Hi Kiran - I sent you the log files via email. I'll close out this question and we can continue in email form :-)