I'm using Codewarrior 10.4 with a PXS2010 board and have enabled floating point support in the project by adding:
#define _EWL_FLOATING_POINT 1
to the build prefix files. The project is set to SPFP floating point support, however logf(x) is giving very strange results. Following on from comments in this thread...
https://community.freescale.com/message/302198#302198
I tried adding:
#pragma dont_inline on
...
#pragma dont_inline reset
around the code using the maths libraries which has fixed the problem. The previous thread mentioning problem this marked this issue as solved in version 10.3 libraries but I seem to be having this problem with 10.4 libraries. Is there a fix for this issue?
Hi twp,
Can you perhaps post the example project? I cannot reproduce it at the moment on my side.
Did you select "SPFP" or "SPFP_Only" in Floating point support dialog when the project was created?
Did you add #include <math.h>?
What results do you get with and without #pragma dont_inline?
Thanks for additional details.
Stan