#include "r600_context.h"
#include "r600_emit.h"
-#if defined(USE_X86_ASM)
-#define COPY_DWORDS( dst, src, nr ) \
-do { \
- int __tmp; \
- __asm__ __volatile__( "rep ; movsl" \
- : "=%c" (__tmp), "=D" (dst), "=S" (__tmp) \
- : "0" (nr), \
- "D" ((long)dst), \
- "S" ((long)src) ); \
-} while (0)
-#else
-#define COPY_DWORDS( dst, src, nr ) \
-do { \
- int j; \
- for ( j = 0 ; j < nr ; j++ ) \
- dst[j] = ((int *)src)[j]; \
- dst += nr; \
-} while (0)
-#endif
-
-static void r600EmitVec4(uint32_t *out, GLvoid * data, int stride, int count)
-{
- int i;
-
- if (RADEON_DEBUG & DEBUG_VERTS)
- fprintf(stderr, "%s count %d stride %d out %p data %p\n",
- __FUNCTION__, count, stride, (void *)out, (void *)data);
-
- if (stride == 4)
- COPY_DWORDS(out, data, count);
- else
- for (i = 0; i < count; i++) {
- out[0] = *(int *)data;
- out++;
- data += stride;
- }
-}
-
-static void r600EmitVec8(uint32_t *out, GLvoid * data, int stride, int count)
-{
- int i;
-
- if (RADEON_DEBUG & DEBUG_VERTS)
- fprintf(stderr, "%s count %d stride %d out %p data %p\n",
- __FUNCTION__, count, stride, (void *)out, (void *)data);
-
- if (stride == 8)
- COPY_DWORDS(out, data, count * 2);
- else
- for (i = 0; i < count; i++) {
- out[0] = *(int *)data;
- out[1] = *(int *)(data + 4);
- out += 2;
- data += stride;
- }
-}
-
-static void r600EmitVec12(uint32_t *out, GLvoid * data, int stride, int count)
-{
- int i;
-
- if (RADEON_DEBUG & DEBUG_VERTS)
- fprintf(stderr, "%s count %d stride %d out %p data %p\n",
- __FUNCTION__, count, stride, (void *)out, (void *)data);
-
- if (stride == 12) {
- COPY_DWORDS(out, data, count * 3);
- }
- else
- for (i = 0; i < count; i++) {
- out[0] = *(int *)data;
- out[1] = *(int *)(data + 4);
- out[2] = *(int *)(data + 8);
- out += 3;
- data += stride;
- }
-}
-
-static void r600EmitVec16(uint32_t *out, GLvoid * data, int stride, int count)
-{
- int i;
-
- if (RADEON_DEBUG & DEBUG_VERTS)
- fprintf(stderr, "%s count %d stride %d out %p data %p\n",
- __FUNCTION__, count, stride, (void *)out, (void *)data);
-
- if (stride == 16)
- COPY_DWORDS(out, data, count * 4);
- else
- for (i = 0; i < count; i++) {
- out[0] = *(int *)data;
- out[1] = *(int *)(data + 4);
- out[2] = *(int *)(data + 8);
- out[3] = *(int *)(data + 12);
- out += 4;
- data += stride;
- }
-}
-
-/* Emit vertex data to GART memory
- * Route inputs to the vertex processor
- * This function should never return R600_FALLBACK_TCL when using software tcl.
- */
-int r600EmitArrays(GLcontext * ctx)
-{
-
- return R600_FALLBACK_NONE;
-}
-
void r600EmitCacheFlush(r600ContextPtr rmesa)
{
BATCH_LOCALS(&rmesa->radeon);
return GL_TRUE;
}
-
-GLboolean r600EmitVec(GLcontext * ctx,
- struct radeon_aos *aos,
- GLvoid * data,
- int size,
- int stride,
- int count)
-{
- radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
- uint32_t *out;
-
- if (stride == 0)
- {
- radeonAllocDmaRegion(rmesa, &aos->bo, &aos->offset, size * count * 4, 32);
- aos->stride = 0;
- }
- else
- {
- radeonAllocDmaRegion(rmesa, &aos->bo, &aos->offset, size * count * 4, 32);
- aos->stride = size;
- }
-
- aos->components = size;
- aos->count = count;
-
- out = (uint32_t*)((char*)aos->bo->ptr + aos->offset);
- switch (size) {
- case 1: r600EmitVec4(out, data, stride, count); break;
- case 2: r600EmitVec8(out, data, stride, count); break;
- case 3: r600EmitVec12(out, data, stride, count); break;
- case 4: r600EmitVec16(out, data, stride, count); break;
- default:
- assert(0);
- break;
- }
-
- return GL_TRUE;
-}
-
-void r600ReleaseVec(GLcontext * ctx)
-{
- radeonReleaseArrays(ctx, ~0);
-}
-
-void r600FreeDmaRegion(context_t *context,
- void * shaderbo)
-{
- struct radeon_bo *pbo = (struct radeon_bo *)shaderbo;
- if(pbo)
- {
- radeon_bo_unref(pbo);
- }
-}