In my OpenCL kernel I use the following code :
#define BxDFType uint
#define BSDF_UNDEFINED 0
#define BSDF_REFLECTION 1<<0
#define BSDF_TRANSMISSION 1<<1
#define BSDF_DIFFUSE 1<<2
#define BSDF_GLOSSY 1<<3
#define BSDF_SPECULAR 1<<4
#define BSDF_ALL_TYPES BSDF_DIFFUSE | BSDF_GLOSSY | BSDF_SPECULAR
#define BSDF_ALL_REFLECTION BSDF_REFLECTION | BSDF_ALL_TYPES
#define BSDF_ALL_TRANSMISSION BSDF_TRANSMISSION | BSDF_ALL_TYPES
#define BSDF_ALL BSDF_ALL_REFLECTION | BSDF_ALL_TRANSMISSION
#define BSDF_CONDUCTOR 1<<5
#define BSDF_DIELECTRIC 1<<6
#define BSDF_MICROFACET_BLINN 1<<8
#define BSDF_MICROFACET_ANISOTROPIC 1<<9
And I do some operations like this :
1) Set a specific flag : type = (type | BSDF_CONDUCTOR)
2) Remove a specific flag : type = (type & ~ BSDF_CONDUCTOR)
3) Check if a flag is set : if (type & BSDF_CONDUCTOR) ...
4) Check if a flag is not set : if (!(type & BSDF_CONDUCTOR)) ...
5) Check that a flag is set : if (type == BSDF_UNDEFINED)... nothing defined !
But it sounds that there are some of theses operations that doesn't work well !
Because I'm unable to debug I can find where is my error. So, I just would like to check with you if
we can do theses operations in OpenCL ?
BSDF_CONDUCTOR has a effective value of 32(1<<5).
so we are doing all operation on flags reletive to this 5th bit only.But actually flag is a 32 bit uint and its value depends on the values at other bit positions.
in set flag you do type|BSDF_CONDUCTOR which definitely sets 5th bit of flag but other bits remain unchanged,resulting in a random number.
I hope it is clear.
In general, it's a good idea to wrap such defines with a paranthesis, like
#define BSDF_TRANSMISSION (1<<1)
#define BSDF_ALL_TYPES (BSDF_DIFFUSE | BSDF_GLOSSY | BSDF_SPECULAR)
Although I don't think that's the problem in your code, assuming you listed all the operations you use.
However, there is a potential for operator precedence effects to create really obscure bugs if you don't make sure the define is evaluated in its entirety before any outside operations can operate on the content.