Signed-off-by: Matteo Bruni mbruni@codeweavers.com --- dlls/wined3d/device.c | 5 ++- dlls/wined3d/stateblock.c | 63 ++++++++++++++++++++++++---------- dlls/wined3d/wined3d_private.h | 4 +-- 3 files changed, 48 insertions(+), 24 deletions(-)
diff --git a/dlls/wined3d/device.c b/dlls/wined3d/device.c index 84056437414..2297d7de916 100644 --- a/dlls/wined3d/device.c +++ b/dlls/wined3d/device.c @@ -27,7 +27,6 @@ #include "config.h" #include "wine/port.h"
-#include <stdio.h> #ifdef HAVE_FLOAT_H # include <float.h> #endif @@ -3848,7 +3847,7 @@ void CDECL wined3d_device_apply_stateblock(struct wined3d_device *device, count = 0; for (i = 0; i < d3d_info->limits.vs_uniform_count; ++i) { - if (stateblock->changed.vs_consts_f[i]) + if (stateblock->changed.vs_consts_f[i >> 5] & (1u << (i & 0x1f))) ++count; else if (count) { @@ -3890,7 +3889,7 @@ void CDECL wined3d_device_apply_stateblock(struct wined3d_device *device, count = 0; for (i = 0; i < d3d_info->limits.ps_uniform_count; ++i) { - if (stateblock->changed.ps_consts_f[i]) + if (stateblock->changed.ps_consts_f[i >> 5] & (1u << (i & 0x1f))) ++count; else if (count) { diff --git a/dlls/wined3d/stateblock.c b/dlls/wined3d/stateblock.c index 499d23fbbd3..6d1562cc60e 100644 --- a/dlls/wined3d/stateblock.c +++ b/dlls/wined3d/stateblock.c @@ -189,7 +189,7 @@ static const DWORD vertex_states_sampler[] = WINED3D_SAMP_DMAP_OFFSET, };
-static inline void stateblock_set_bits(DWORD *map, UINT map_size) +static inline void stateblock_set_all_bits(DWORD *map, UINT map_size) { DWORD mask = (1u << (map_size & 0x1f)) - 1; memset(map, 0xff, (map_size >> 5) * sizeof(*map)); @@ -201,7 +201,6 @@ static void stateblock_savedstates_set_all(struct wined3d_saved_states *states, { unsigned int i;
- /* Single values */ states->indices = 1; states->material = 1; states->viewport = 1; @@ -210,12 +209,11 @@ static void stateblock_savedstates_set_all(struct wined3d_saved_states *states, states->vertexShader = 1; states->scissorRect = 1;
- /* Fixed size arrays */ states->streamSource = 0xffff; states->streamFreq = 0xffff; states->textures = 0xfffff; - stateblock_set_bits(states->transform, WINED3D_HIGHEST_TRANSFORM_STATE + 1); - stateblock_set_bits(states->renderState, WINEHIGHEST_RENDER_STATE + 1); + stateblock_set_all_bits(states->transform, WINED3D_HIGHEST_TRANSFORM_STATE + 1); + stateblock_set_all_bits(states->renderState, WINEHIGHEST_RENDER_STATE + 1); for (i = 0; i < WINED3D_MAX_TEXTURES; ++i) states->textureState[i] = 0x3ffff; for (i = 0; i < WINED3D_MAX_COMBINED_SAMPLERS; ++i) states->samplerState[i] = 0x3ffe; states->clipplane = (1u << WINED3D_MAX_CLIP_DISTANCES) - 1; @@ -224,9 +222,8 @@ static void stateblock_savedstates_set_all(struct wined3d_saved_states *states, states->vertexShaderConstantsB = 0xffff; states->vertexShaderConstantsI = 0xffff;
- /* Dynamically sized arrays */ - memset(states->ps_consts_f, TRUE, sizeof(BOOL) * ps_consts); - memset(states->vs_consts_f, TRUE, sizeof(BOOL) * vs_consts); + memset(states->ps_consts_f, 0xffu, sizeof(states->ps_consts_f)); + memset(states->vs_consts_f, 0xffu, sizeof(states->vs_consts_f)); }
static void stateblock_savedstates_set_pixel(struct wined3d_saved_states *states, const DWORD num_constants) @@ -252,7 +249,7 @@ static void stateblock_savedstates_set_pixel(struct wined3d_saved_states *states states->pixelShaderConstantsB = 0xffff; states->pixelShaderConstantsI = 0xffff;
- memset(states->ps_consts_f, TRUE, sizeof(BOOL) * num_constants); + memset(states->ps_consts_f, 0xffu, sizeof(states->ps_consts_f)); }
static void stateblock_savedstates_set_vertex(struct wined3d_saved_states *states, const DWORD num_constants) @@ -279,12 +276,12 @@ static void stateblock_savedstates_set_vertex(struct wined3d_saved_states *state states->vertexShaderConstantsB = 0xffff; states->vertexShaderConstantsI = 0xffff;
- memset(states->vs_consts_f, TRUE, sizeof(BOOL) * num_constants); + memset(states->vs_consts_f, 0xffu, sizeof(states->vs_consts_f)); }
void CDECL wined3d_stateblock_init_contained_states(struct wined3d_stateblock *stateblock) { - const struct wined3d_d3d_info *d3d_info = &stateblock->device->adapter->d3d_info; + const unsigned int word_bit_count = sizeof(*stateblock->changed.vs_consts_f) * CHAR_BIT; unsigned int i, j;
for (i = 0; i <= WINEHIGHEST_RENDER_STATE >> 5; ++i) @@ -311,11 +308,14 @@ void CDECL wined3d_stateblock_init_contained_states(struct wined3d_stateblock *s } }
- for (i = 0; i < d3d_info->limits.vs_uniform_count; ++i) + for (i = 0; i < ARRAY_SIZE(stateblock->changed.vs_consts_f); ++i) { - if (stateblock->changed.vs_consts_f[i]) + DWORD bitmask = stateblock->changed.vs_consts_f[i]; + + while (bitmask) { - stateblock->contained_vs_consts_f[stateblock->num_contained_vs_consts_f] = i; + j = wined3d_bit_scan(&bitmask); + stateblock->contained_vs_consts_f[stateblock->num_contained_vs_consts_f] = i * word_bit_count + j; ++stateblock->num_contained_vs_consts_f; } } @@ -338,11 +338,14 @@ void CDECL wined3d_stateblock_init_contained_states(struct wined3d_stateblock *s } }
- for (i = 0; i < d3d_info->limits.ps_uniform_count; ++i) + for (i = 0; i < ARRAY_SIZE(stateblock->changed.ps_consts_f); ++i) { - if (stateblock->changed.ps_consts_f[i]) + DWORD bitmask = stateblock->changed.ps_consts_f[i]; + + while (bitmask) { - stateblock->contained_ps_consts_f[stateblock->num_contained_ps_consts_f] = i; + j = wined3d_bit_scan(&bitmask); + stateblock->contained_ps_consts_f[stateblock->num_contained_ps_consts_f] = i * word_bit_count + j; ++stateblock->num_contained_ps_consts_f; } } @@ -1280,6 +1283,28 @@ void CDECL wined3d_stateblock_set_vertex_shader(struct wined3d_stateblock *state stateblock->changed.vertexShader = TRUE; }
+static void wined3d_bitmask_set_bits(DWORD *bitmask, unsigned int offset, unsigned int count) +{ + const unsigned int word_bit_count = sizeof(*bitmask) * CHAR_BIT; + const unsigned int shift = offset % word_bit_count; + + bitmask += offset / word_bit_count; + *bitmask |= ~0u >> (word_bit_count - min(count, word_bit_count)) << shift; + ++bitmask; + count -= min(count, word_bit_count - shift); + if (!count) + return; + if (count >= word_bit_count) + { + memset(bitmask, 0xffu, count / CHAR_BIT); + bitmask += count / word_bit_count; + count = count % word_bit_count; + if (!count) + return; + } + *bitmask |= (1u << count) - 1; +} + HRESULT CDECL wined3d_stateblock_set_vs_consts_f(struct wined3d_stateblock *stateblock, unsigned int start_idx, unsigned int count, const struct wined3d_vec4 *constants) { @@ -1290,7 +1315,7 @@ HRESULT CDECL wined3d_stateblock_set_vs_consts_f(struct wined3d_stateblock *stat return WINED3DERR_INVALIDCALL;
memcpy(&stateblock->stateblock_state.vs_consts_f[start_idx], constants, count * sizeof(*constants)); - memset(&stateblock->changed.vs_consts_f[start_idx], 1, count * sizeof(*stateblock->changed.vs_consts_f)); + wined3d_bitmask_set_bits(stateblock->changed.vs_consts_f, start_idx, count); return WINED3D_OK; }
@@ -1356,7 +1381,7 @@ HRESULT CDECL wined3d_stateblock_set_ps_consts_f(struct wined3d_stateblock *stat return WINED3DERR_INVALIDCALL;
memcpy(&stateblock->stateblock_state.ps_consts_f[start_idx], constants, count * sizeof(*constants)); - memset(&stateblock->changed.ps_consts_f[start_idx], 1, count * sizeof(*stateblock->changed.ps_consts_f)); + wined3d_bitmask_set_bits(stateblock->changed.ps_consts_f, start_idx, count); return WINED3D_OK; }
diff --git a/dlls/wined3d/wined3d_private.h b/dlls/wined3d/wined3d_private.h index 0e1e55d52bf..c8b22832e65 100644 --- a/dlls/wined3d/wined3d_private.h +++ b/dlls/wined3d/wined3d_private.h @@ -3919,10 +3919,10 @@ struct wined3d_saved_states DWORD clipplane; /* WINED3D_MAX_USER_CLIP_PLANES, 32 */ WORD pixelShaderConstantsB; /* WINED3D_MAX_CONSTS_B, 16 */ WORD pixelShaderConstantsI; /* WINED3D_MAX_CONSTS_I, 16 */ - BOOL ps_consts_f[WINED3D_MAX_PS_CONSTS_F]; + DWORD ps_consts_f[WINED3D_MAX_PS_CONSTS_F >> 5]; WORD vertexShaderConstantsB; /* WINED3D_MAX_CONSTS_B, 16 */ WORD vertexShaderConstantsI; /* WINED3D_MAX_CONSTS_I, 16 */ - BOOL vs_consts_f[WINED3D_MAX_VS_CONSTS_F]; + DWORD vs_consts_f[WINED3D_MAX_VS_CONSTS_F >> 5]; DWORD textures : 20; /* WINED3D_MAX_COMBINED_SAMPLERS, 20 */ DWORD indices : 1; DWORD material : 1;