[PATCH 5/6] d3dx9_36/tests: Added trigonometric functions test to HLSL test suite

[Joris Huizer]

Hello Travis Athougies,

In this patch, you have this code:

+    void *ret = NULL;
//...
+    ret = HeapAlloc(GetProcessHeap(), 0, 4 * samples);
+    for(i = 0;i < 32;i++) {
//...
+        memcpy((void*) (((DWORD) ret) + (4 * i)), lr.pBits, 4);
//...
+    }

I'm not sure, but I think this will cause 64-bit issues (truncating to a 32-bit value?)
Even if it doesn't, it seems simpler (more readable) to use a char pointer instead of a void pointer, like

+    char *ret;
//...
+    ret = HeapAlloc(GetProcessHeap(), 0, 4 * samples);
//...
+        memcpy(ret + (4 * i), lr.pBits, 4);

HTH,
regards, Joris

[Octavian Voicu]

On Thu, Aug 5, 2010 at 3:58 PM, Joris Huizer  wrote:
> In this patch, you have this code:
> - void *ret = NULL;
>
> //...
> - ret = HeapAlloc(GetProcessHeap(), 0, 4 * samples);
> - for(i = 0;i < 32;i++) {
>
> //...
> - memcpy((void*) (((DWORD) ret) + (4 * i)), lr.pBits, 4);
>
> //...
> - }
>
> I'm not sure, but I think this will cause 64-bit issues (truncating to a 32-bit value?)
> Even if it doesn't, it seems simpler (more readable) to use a char pointer instead of a void pointer, like
> - char *ret;
>
> //...
> - ret = HeapAlloc(GetProcessHeap(), 0, 4 * samples);
>
> //...
> - memcpy(ret + (4 * i), lr.pBits, 4);

Or just use a DWORD pointer:

+    DWORD *ret;
//...
+    ret = HeapAlloc(GetProcessHeap(), 0, 4 * samples);
+    for(i = 0;i < 32;i++) {
//...
+        ret[i] = lr.pBits;
//...
+    }