[LAD] Test app for LADSPA plugins

[Damon Chaplin]

On Thu, 2009-07-30 at 11:55 +0200, xmag wrote:
> Hi,
> 
> I wrote all the Glame* filters, did you send me an email, and if, where
> did you send it :-)

I sent Steve Harris an email about some issues in the swh plugins, but
no problems were spotted in the Glame* filters by my test app.


> What is your testapp doing?
> I guess, your probably just calling instantiate and then cleanup.

Just this:

  if ((handle = instantiate(descriptor, SRATE))) {
    descriptor->cleanup(handle);
  }

> The structs are initialized within the activate function. I don't know
> whether it is mandatory to call this function or not.

I don't think it is mandatory. I can't see that in the spec.


> A quick look in my code shows, that in util/iir.h the function
> free_iirf_t does not check if the pointers are different from NULL
> before freeing them. This probably causes the segfault.
> Maybe this should be fixed :)

Yep, that fixes all of them, in iir.h and iir.c.

Thanks.

Damon


--- iir.h.orig  2006-08-08 16:53:50.000000000 +0100
+++ iir.h       2009-07-30 11:09:59.000000000 +0100
@@ -63,9 +63,11 @@ static inline iirf_t* init_iirf_t(iir_st
 
 static inline void free_iirf_t(iirf_t* iirf, iir_stage_t* gt) {
        int i;
-       for(i=0;i<gt->availst;i++){
-               free(iirf[i].iring);
-               free(iirf[i].oring);
+       if (gt){
+         for(i=0;i<gt->availst;i++){
+           free(iirf[i].iring);
+           free(iirf[i].oring);
+         }
        }
        free(iirf);
 };
--- iir.c.orig  2003-09-13 12:55:23.000000000 +0100
+++ iir.c       2009-07-30 11:12:45.000000000 +0100
@@ -85,10 +85,12 @@ void combine_iir_stages(int mode, iir_st
 
 void free_iir_stage(iir_stage_t *gt){
        int i;
-       for(i=0;i<gt->availst;i++)
-               free(gt->coeff[i]);
-       free(gt->coeff);
-       free(gt);
+       if (gt){
+         for(i=0;i<gt->availst;i++)
+           free(gt->coeff[i]);
+         free(gt->coeff);
+         free(gt);
+       }
 }
 
 /* center: frequency already normalized between 0 and 0.5 of sampling