static Bool
RADEONUploadToScreenCS(PixmapPtr pDst, int x, int y, int w, int h,
char *src, int src_pitch)
{
ScreenPtr pScreen = pDst->drawable.pScreen;
RINFO_FROM_SCREEN(pScreen);
struct radeon_exa_pixmap_priv *driver_priv;
struct radeon_bo *scratch = NULL;
struct radeon_bo *copy_dst;
unsigned char *dst;
unsigned size;
uint32_t datatype = 0;
uint32_t dst_domain;
uint32_t dst_pitch_offset;
unsigned bpp = pDst->drawable.bitsPerPixel;
uint32_t scratch_pitch = RADEON_ALIGN(w * bpp / 8, 64);
uint32_t copy_pitch;
uint32_t swap = RADEON_HOST_DATA_SWAP_NONE;
int ret;
Bool flush = TRUE;
Bool r;
int i;
if (bpp < 8)
return FALSE;
driver_priv = exaGetPixmapDriverPrivate(pDst);
if (!driver_priv || !driver_priv->bo)
return FALSE;
#if X_BYTE_ORDER == X_BIG_ENDIAN
switch (bpp) {
case 32:
swap = RADEON_HOST_DATA_SWAP_32BIT;
break;
case 16:
swap = RADEON_HOST_DATA_SWAP_16BIT;
break;
}
#endif
/* If we know the BO won't be busy / in VRAM, don't bother with a scratch */
copy_dst = driver_priv->bo;
copy_pitch = pDst->devKind;
if (!(driver_priv->tiling_flags & (RADEON_TILING_MACRO | RADEON_TILING_MICRO))) {
if (!radeon_bo_is_referenced_by_cs(driver_priv->bo, info->cs)) {
flush = FALSE;
if (!radeon_bo_is_busy(driver_priv->bo, &dst_domain) &&
!(dst_domain & RADEON_GEM_DOMAIN_VRAM))
goto copy;
}
/* use cpu copy for fast fb access */
if (info->is_fast_fb)
goto copy;
}
size = scratch_pitch * h;
scratch = radeon_bo_open(info->bufmgr, 0, size, 0, RADEON_GEM_DOMAIN_GTT, 0);
if (scratch == NULL) {
goto copy;
}
radeon_cs_space_reset_bos(info->cs);
radeon_add_pixmap(info->cs, pDst, 0, RADEON_GEM_DOMAIN_VRAM);
radeon_cs_space_add_persistent_bo(info->cs, scratch, RADEON_GEM_DOMAIN_GTT, 0);
ret = radeon_cs_space_check(info->cs);
if (ret) {
goto copy;
}
copy_dst = scratch;
copy_pitch = scratch_pitch;
flush = FALSE;
copy:
if (flush)
radeon_cs_flush_indirect(pScrn);
ret = radeon_bo_map(copy_dst, 0);
if (ret) {
r = FALSE;
goto out;
}
r = TRUE;
size = w * bpp / 8;
dst = copy_dst->ptr;
if (copy_dst == driver_priv->bo)
dst += y * copy_pitch + x * bpp / 8;
for (i = 0; i < h; i++) {
RADEONCopySwap(dst + i * copy_pitch, (uint8_t*)src, size, swap);
src += src_pitch;
}
radeon_bo_unmap(copy_dst);
if (copy_dst == scratch) {
RADEONGetDatatypeBpp(pDst->drawable.bitsPerPixel, &datatype);
RADEONGetPixmapOffsetPitch(pDst, &dst_pitch_offset);
RADEON_SWITCH_TO_2D();
RADEONBlitChunk(pScrn, scratch, driver_priv->bo, datatype, scratch_pitch << 16,
dst_pitch_offset, 0, 0, x, y, w, h,
RADEON_GEM_DOMAIN_GTT, RADEON_GEM_DOMAIN_VRAM);
}
out:
if (scratch)
radeon_bo_unref(scratch);
return r;
}
static Bool
RADEONDownloadFromScreenCS(PixmapPtr pSrc, int x, int y, int w,
int h, char *dst, int dst_pitch)
{
RINFO_FROM_SCREEN(pSrc->drawable.pScreen);
struct radeon_exa_pixmap_priv *driver_priv;
struct radeon_bo *scratch = NULL;
struct radeon_bo *copy_src;
unsigned size;
uint32_t datatype = 0;
uint32_t src_domain = 0;
uint32_t src_pitch_offset;
unsigned bpp = pSrc->drawable.bitsPerPixel;
uint32_t scratch_pitch = RADEON_ALIGN(w * bpp / 8, 64);
uint32_t copy_pitch;
uint32_t swap = RADEON_HOST_DATA_SWAP_NONE;
int ret;
Bool flush = FALSE;
Bool r;
if (bpp < 8)
return FALSE;
driver_priv = exaGetPixmapDriverPrivate(pSrc);
if (!driver_priv || !driver_priv->bo)
return FALSE;
#if X_BYTE_ORDER == X_BIG_ENDIAN
switch (bpp) {
case 32:
swap = RADEON_HOST_DATA_SWAP_32BIT;
break;
case 16:
swap = RADEON_HOST_DATA_SWAP_16BIT;
break;
}
#endif
/* If we know the BO won't end up in VRAM anyway, don't bother with a scratch */
copy_src = driver_priv->bo;
copy_pitch = pSrc->devKind;
if (!(driver_priv->tiling_flags & (RADEON_TILING_MACRO | RADEON_TILING_MICRO))) {
if (radeon_bo_is_referenced_by_cs(driver_priv->bo, info->cs)) {
src_domain = radeon_bo_get_src_domain(driver_priv->bo);
if ((src_domain & (RADEON_GEM_DOMAIN_GTT | RADEON_GEM_DOMAIN_VRAM)) ==
(RADEON_GEM_DOMAIN_GTT | RADEON_GEM_DOMAIN_VRAM))
src_domain = 0;
else /* A write may be scheduled */
flush = TRUE;
}
if (!src_domain)
radeon_bo_is_busy(driver_priv->bo, &src_domain);
if (src_domain & ~(uint32_t)RADEON_GEM_DOMAIN_VRAM)
goto copy;
}
size = scratch_pitch * h;
scratch = radeon_bo_open(info->bufmgr, 0, size, 0, RADEON_GEM_DOMAIN_GTT, 0);
if (scratch == NULL) {
goto copy;
}
radeon_cs_space_reset_bos(info->cs);
radeon_add_pixmap(info->cs, pSrc, RADEON_GEM_DOMAIN_GTT | RADEON_GEM_DOMAIN_VRAM, 0);
radeon_cs_space_add_persistent_bo(info->cs, scratch, 0, RADEON_GEM_DOMAIN_GTT);
ret = radeon_cs_space_check(info->cs);
if (ret) {
goto copy;
}
RADEONGetDatatypeBpp(pSrc->drawable.bitsPerPixel, &datatype);
RADEONGetPixmapOffsetPitch(pSrc, &src_pitch_offset);
RADEON_SWITCH_TO_2D();
RADEONBlitChunk(pScrn, driver_priv->bo, scratch, datatype, src_pitch_offset,
scratch_pitch << 16, x, y, 0, 0, w, h,
RADEON_GEM_DOMAIN_VRAM | RADEON_GEM_DOMAIN_GTT,
RADEON_GEM_DOMAIN_GTT);
copy_src = scratch;
copy_pitch = scratch_pitch;
flush = TRUE;
copy:
if (flush)
FLUSH_RING();
ret = radeon_bo_map(copy_src, 0);
if (ret) {
ErrorF("failed to map pixmap: %d\n", ret);
r = FALSE;
goto out;
}
r = TRUE;
w *= bpp / 8;
if (copy_src == driver_priv->bo)
size = y * copy_pitch + x * bpp / 8;
else
size = 0;
while (h--) {
RADEONCopySwap((uint8_t*)dst, copy_src->ptr + size, w, swap);
size += copy_pitch;
dst += dst_pitch;
}
radeon_bo_unmap(copy_src);
out:
if (scratch)
radeon_bo_unref(scratch);
return r;
}
/* Calculate appropriate tiling and pitch for a pixmap and allocate a BO that
* can hold it.
*/
struct radeon_bo*
radeon_alloc_pixmap_bo(ScrnInfoPtr pScrn, int width, int height, int depth,
int usage_hint, int bitsPerPixel, int *new_pitch,
struct radeon_surface *new_surface, uint32_t *new_tiling)
{
RADEONInfoPtr info = RADEONPTR(pScrn);
int pitch, base_align;
uint32_t size, heighta;
int cpp = bitsPerPixel / 8;
uint32_t tiling = 0;
struct radeon_surface surface;
struct radeon_bo *bo;
int domain = RADEON_GEM_DOMAIN_VRAM;
if (usage_hint) {
if (info->allowColorTiling) {
if (usage_hint & RADEON_CREATE_PIXMAP_TILING_MACRO)
tiling |= RADEON_TILING_MACRO;
if (usage_hint & RADEON_CREATE_PIXMAP_TILING_MICRO)
tiling |= RADEON_TILING_MICRO;
}
if (usage_hint & RADEON_CREATE_PIXMAP_DEPTH)
tiling |= RADEON_TILING_MACRO | RADEON_TILING_MICRO;
#ifdef CREATE_PIXMAP_USAGE_SHARED
if ((usage_hint & 0xffff) == CREATE_PIXMAP_USAGE_SHARED) {
tiling = 0;
domain = RADEON_GEM_DOMAIN_GTT;
}
#endif
}
/* Small pixmaps must not be macrotiled on R300, hw cannot sample them
* correctly because samplers automatically switch to macrolinear. */
if (info->ChipFamily >= CHIP_FAMILY_R300 &&
info->ChipFamily <= CHIP_FAMILY_RS740 &&
(tiling & RADEON_TILING_MACRO) &&
!RADEONMacroSwitch(width, height, bitsPerPixel, tiling,
info->ChipFamily >= CHIP_FAMILY_RV350)) {
tiling &= ~RADEON_TILING_MACRO;
}
heighta = RADEON_ALIGN(height, drmmode_get_height_align(pScrn, tiling));
pitch = RADEON_ALIGN(width, drmmode_get_pitch_align(pScrn, cpp, tiling)) * cpp;
base_align = drmmode_get_base_align(pScrn, cpp, tiling);
size = RADEON_ALIGN(heighta * pitch, RADEON_GPU_PAGE_SIZE);
memset(&surface, 0, sizeof(struct radeon_surface));
if (info->ChipFamily >= CHIP_FAMILY_R600 && info->surf_man) {
if (width) {
surface.npix_x = width;
/* need to align height to 8 for old kernel */
surface.npix_y = RADEON_ALIGN(height, 8);
surface.npix_z = 1;
surface.blk_w = 1;
surface.blk_h = 1;
surface.blk_d = 1;
surface.array_size = 1;
surface.last_level = 0;
surface.bpe = cpp;
surface.nsamples = 1;
if (height < 128) {
/* disable 2d tiling for small surface to work around
* the fact that ddx align height to 8 pixel for old
* obscure reason i can't remember
*/
tiling &= ~RADEON_TILING_MACRO;
}
surface.flags = RADEON_SURF_SCANOUT;
/* we are requiring a recent enough libdrm version */
surface.flags |= RADEON_SURF_HAS_TILE_MODE_INDEX;
surface.flags |= RADEON_SURF_SET(RADEON_SURF_TYPE_2D, TYPE);
surface.flags |= RADEON_SURF_SET(RADEON_SURF_MODE_LINEAR, MODE);
if ((tiling & RADEON_TILING_MICRO)) {
surface.flags = RADEON_SURF_CLR(surface.flags, MODE);
surface.flags |= RADEON_SURF_SET(RADEON_SURF_MODE_1D, MODE);
}
if ((tiling & RADEON_TILING_MACRO)) {
surface.flags = RADEON_SURF_CLR(surface.flags, MODE);
surface.flags |= RADEON_SURF_SET(RADEON_SURF_MODE_2D, MODE);
}
if (usage_hint & RADEON_CREATE_PIXMAP_SZBUFFER) {
surface.flags |= RADEON_SURF_ZBUFFER;
surface.flags |= RADEON_SURF_SBUFFER;
}
if (radeon_surface_best(info->surf_man, &surface)) {
return NULL;
}
if (radeon_surface_init(info->surf_man, &surface)) {
return NULL;
}
size = surface.bo_size;
base_align = surface.bo_alignment;
pitch = surface.level[0].pitch_bytes;
tiling = 0;
switch (surface.level[0].mode) {
case RADEON_SURF_MODE_2D:
tiling |= RADEON_TILING_MACRO;
tiling |= surface.bankw << RADEON_TILING_EG_BANKW_SHIFT;
tiling |= surface.bankh << RADEON_TILING_EG_BANKH_SHIFT;
tiling |= surface.mtilea << RADEON_TILING_EG_MACRO_TILE_ASPECT_SHIFT;
tiling |= eg_tile_split(surface.tile_split) << RADEON_TILING_EG_TILE_SPLIT_SHIFT;
tiling |= eg_tile_split(surface.stencil_tile_split) << RADEON_TILING_EG_STENCIL_TILE_SPLIT_SHIFT;
break;
case RADEON_SURF_MODE_1D:
tiling |= RADEON_TILING_MICRO;
break;
default:
break;
}
}
}
bo = radeon_bo_open(info->bufmgr, 0, size, base_align,
domain, 0);
if (bo && tiling && radeon_bo_set_tiling(bo, tiling, pitch) == 0)
*new_tiling = tiling;
*new_surface = surface;
*new_pitch = pitch;
return bo;
}
