~mmach/netext73/xorg-intel

« back to all changes in this revision

Viewing changes to src/sna/gen2_render.c

Committer: mmach
Date: 2021-03-17 18:24:35 UTC
Revision ID: netbit73@gmail.com-20210317182435-7okt04d35etrqgqp

lipiec

files added:

AUTHORS

COPYING

Makefile.am

NEWS

README

RELEASING

autogen.sh

benchmarks

benchmarks/.gitignore

benchmarks/Makefile.am

benchmarks/dri2-swap.c

benchmarks/dri3-swap.c

configure.ac

debian

debian/README.Debian

debian/README.source

debian/changelog

debian/control

debian/copyright

debian/patches

debian/patches/series

debian/rules

debian/source

debian/source/format

debian/upstream.changelog

debian/watch

debian/xorg.conf

debian/xserver-xorg-video-intel.docs

debian/xserver-xorg-video-intel.install

debian/xserver-xorg-video-intel.install.kfreebsd-amd64

debian/xserver-xorg-video-intel.install.kfreebsd-i386

debian/xserver-xorg-video-intel.links

debian/xserver-xorg-video-intel.lintian-overrides

debian/xserver-xorg-video-intel.maintscript

libobj

libobj/Makefile.am

libobj/alloca.c

libobj/getline.c

libobj/ignore.c

m4/ac_define_dir.m4

man/Makefile.am

man/intel.man

man/meson.build

meson.build

meson_options.txt

src/Makefile.am

src/backlight.c

src/backlight.h

src/compat-api.h

src/fd.c

src/fd.h

src/i915_pciids.h

src/intel_device.c

src/intel_driver.h

src/intel_list.h

src/intel_module.c

src/intel_options.c

src/intel_options.h

src/legacy

src/legacy/Makefile.am

src/legacy/README

src/legacy/i810

src/legacy/i810/Makefile.am

src/legacy/i810/i810.h

src/legacy/i810/i810_accel.c

src/legacy/i810/i810_common.h

src/legacy/i810/i810_cursor.c

src/legacy/i810/i810_dga.c

src/legacy/i810/i810_dri.c

src/legacy/i810/i810_dri.h

src/legacy/i810/i810_driver.c

src/legacy/i810/i810_hwmc.c

src/legacy/i810/i810_memory.c

src/legacy/i810/i810_reg.h

src/legacy/i810/i810_ring.h

src/legacy/i810/i810_video.c

src/legacy/i810/i810_wmark.c

src/legacy/i810/i810_xaa.c

src/legacy/i810/meson.build

src/legacy/i810/xvmc

src/legacy/i810/xvmc/I810XvMC.c

src/legacy/i810/xvmc/I810XvMC.h

src/legacy/i810/xvmc/Makefile.am

src/legacy/i810/xvmc/meson.build

src/legacy/legacy.h

src/meson.build

src/render_program

src/render_program/Makefile.am

src/render_program/exa_sf.g4a

src/render_program/exa_sf.g4b

src/render_program/exa_sf.g4b.gen5

src/render_program/exa_sf.g5a

src/render_program/exa_sf.g5b

src/render_program/exa_sf_mask.g4a

src/render_program/exa_sf_mask.g4b

src/render_program/exa_sf_mask.g4b.gen5

src/render_program/exa_sf_mask.g5a

src/render_program/exa_sf_mask.g5b

src/render_program/exa_wm.g4i

src/render_program/exa_wm_affine.g4i

src/render_program/exa_wm_affine.g6i

src/render_program/exa_wm_ca.g4a

src/render_program/exa_wm_ca.g4b

src/render_program/exa_wm_ca.g4b.gen5

src/render_program/exa_wm_ca.g5a

src/render_program/exa_wm_ca.g5b

src/render_program/exa_wm_ca.g6a

src/render_program/exa_wm_ca.g6b

src/render_program/exa_wm_ca_srcalpha.g4a

src/render_program/exa_wm_ca_srcalpha.g4b

src/render_program/exa_wm_ca_srcalpha.g4b.gen5

src/render_program/exa_wm_ca_srcalpha.g5a

src/render_program/exa_wm_ca_srcalpha.g5b

src/render_program/exa_wm_ca_srcalpha.g6a

src/render_program/exa_wm_ca_srcalpha.g6b

src/render_program/exa_wm_mask_affine.g4a

src/render_program/exa_wm_mask_affine.g4b

src/render_program/exa_wm_mask_affine.g4b.gen5

src/render_program/exa_wm_mask_affine.g5a

src/render_program/exa_wm_mask_affine.g5b

src/render_program/exa_wm_mask_affine.g6a

src/render_program/exa_wm_mask_affine.g6b

src/render_program/exa_wm_mask_affine.g7a

src/render_program/exa_wm_mask_affine.g7b

src/render_program/exa_wm_mask_projective.g4a

src/render_program/exa_wm_mask_projective.g4b

src/render_program/exa_wm_mask_projective.g4b.gen5

src/render_program/exa_wm_mask_projective.g5a

src/render_program/exa_wm_mask_projective.g5b

src/render_program/exa_wm_mask_projective.g6a

src/render_program/exa_wm_mask_projective.g6b

src/render_program/exa_wm_mask_projective.g7a

src/render_program/exa_wm_mask_projective.g7b

src/render_program/exa_wm_mask_sample_a.g4a

src/render_program/exa_wm_mask_sample_a.g4b

src/render_program/exa_wm_mask_sample_a.g4b.gen5

src/render_program/exa_wm_mask_sample_a.g5a

src/render_program/exa_wm_mask_sample_a.g5b

src/render_program/exa_wm_mask_sample_a.g6a

src/render_program/exa_wm_mask_sample_a.g6b

src/render_program/exa_wm_mask_sample_a.g7a

src/render_program/exa_wm_mask_sample_a.g7b

src/render_program/exa_wm_mask_sample_argb.g4a

src/render_program/exa_wm_mask_sample_argb.g4b

src/render_program/exa_wm_mask_sample_argb.g4b.gen5

src/render_program/exa_wm_mask_sample_argb.g5a

src/render_program/exa_wm_mask_sample_argb.g5b

src/render_program/exa_wm_mask_sample_argb.g6a

src/render_program/exa_wm_mask_sample_argb.g6b

src/render_program/exa_wm_mask_sample_argb.g7a

src/render_program/exa_wm_mask_sample_argb.g7b

src/render_program/exa_wm_noca.g4a

src/render_program/exa_wm_noca.g4b

src/render_program/exa_wm_noca.g4b.gen5

src/render_program/exa_wm_noca.g5a

src/render_program/exa_wm_noca.g5b

src/render_program/exa_wm_noca.g6a

src/render_program/exa_wm_noca.g6b

src/render_program/exa_wm_projective.g4i

src/render_program/exa_wm_sample_nv12.g4i

src/render_program/exa_wm_sample_planar.g4i

src/render_program/exa_wm_src_affine.g4a

src/render_program/exa_wm_src_affine.g4b

src/render_program/exa_wm_src_affine.g4b.gen5

src/render_program/exa_wm_src_affine.g5a

src/render_program/exa_wm_src_affine.g5b

src/render_program/exa_wm_src_affine.g6a

src/render_program/exa_wm_src_affine.g6b

src/render_program/exa_wm_src_affine.g7a

src/render_program/exa_wm_src_affine.g7b

src/render_program/exa_wm_src_affine.g8a

src/render_program/exa_wm_src_affine.g8b

src/render_program/exa_wm_src_projective.g4a

src/render_program/exa_wm_src_projective.g4b

src/render_program/exa_wm_src_projective.g4b.gen5

src/render_program/exa_wm_src_projective.g5a

src/render_program/exa_wm_src_projective.g5b

src/render_program/exa_wm_src_projective.g6a

src/render_program/exa_wm_src_projective.g6b

src/render_program/exa_wm_src_projective.g7a

src/render_program/exa_wm_src_projective.g7b

src/render_program/exa_wm_src_sample_a.g4a

src/render_program/exa_wm_src_sample_a.g4b

src/render_program/exa_wm_src_sample_a.g4b.gen5

src/render_program/exa_wm_src_sample_a.g5a

src/render_program/exa_wm_src_sample_a.g5b

src/render_program/exa_wm_src_sample_a.g6a

src/render_program/exa_wm_src_sample_a.g6b

src/render_program/exa_wm_src_sample_a.g7a

src/render_program/exa_wm_src_sample_a.g7b

src/render_program/exa_wm_src_sample_argb.g4a

src/render_program/exa_wm_src_sample_argb.g4b

src/render_program/exa_wm_src_sample_argb.g4b.gen5

src/render_program/exa_wm_src_sample_argb.g4i

src/render_program/exa_wm_src_sample_argb.g5a

src/render_program/exa_wm_src_sample_argb.g5b

src/render_program/exa_wm_src_sample_argb.g6a

src/render_program/exa_wm_src_sample_argb.g6b

src/render_program/exa_wm_src_sample_argb.g7a

src/render_program/exa_wm_src_sample_argb.g7b

src/render_program/exa_wm_src_sample_argb.g8a

src/render_program/exa_wm_src_sample_argb.g8b

src/render_program/exa_wm_src_sample_argb_ayuv.g8a

src/render_program/exa_wm_src_sample_argb_ayuv.g8b

src/render_program/exa_wm_src_sample_nv12.g4a

src/render_program/exa_wm_src_sample_nv12.g4b

src/render_program/exa_wm_src_sample_nv12.g4b.gen5

src/render_program/exa_wm_src_sample_nv12.g5a

src/render_program/exa_wm_src_sample_nv12.g5b

src/render_program/exa_wm_src_sample_nv12.g6a

src/render_program/exa_wm_src_sample_nv12.g6b

src/render_program/exa_wm_src_sample_nv12.g7a

src/render_program/exa_wm_src_sample_nv12.g7b

src/render_program/exa_wm_src_sample_nv12.g8a

src/render_program/exa_wm_src_sample_nv12.g8b

src/render_program/exa_wm_src_sample_planar.g4a

src/render_program/exa_wm_src_sample_planar.g4b

src/render_program/exa_wm_src_sample_planar.g4b.gen5

src/render_program/exa_wm_src_sample_planar.g5a

src/render_program/exa_wm_src_sample_planar.g5b

src/render_program/exa_wm_src_sample_planar.g6a

src/render_program/exa_wm_src_sample_planar.g6b

src/render_program/exa_wm_src_sample_planar.g7a

src/render_program/exa_wm_src_sample_planar.g7b

src/render_program/exa_wm_src_sample_planar.g8a

src/render_program/exa_wm_src_sample_planar.g8b

src/render_program/exa_wm_write.g4a

src/render_program/exa_wm_write.g4b

src/render_program/exa_wm_write.g4b.gen5

src/render_program/exa_wm_write.g5a

src/render_program/exa_wm_write.g5b

src/render_program/exa_wm_write.g6a

src/render_program/exa_wm_write.g6b

src/render_program/exa_wm_write.g6i

src/render_program/exa_wm_write.g7a

src/render_program/exa_wm_write.g7b

src/render_program/exa_wm_write.g8a

src/render_program/exa_wm_write.g8b

src/render_program/exa_wm_xy.g4a

src/render_program/exa_wm_xy.g4b

src/render_program/exa_wm_xy.g4b.gen5

src/render_program/exa_wm_xy.g5a

src/render_program/exa_wm_xy.g5b

src/render_program/exa_wm_yuv_rgb_bt601.g4a

src/render_program/exa_wm_yuv_rgb_bt601.g4b

src/render_program/exa_wm_yuv_rgb_bt601.g4b.gen5

src/render_program/exa_wm_yuv_rgb_bt601.g5a

src/render_program/exa_wm_yuv_rgb_bt601.g5b

src/render_program/exa_wm_yuv_rgb_bt601.g6a

src/render_program/exa_wm_yuv_rgb_bt601.g6b

src/render_program/exa_wm_yuv_rgb_bt601.g7a

src/render_program/exa_wm_yuv_rgb_bt601.g7b

src/render_program/exa_wm_yuv_rgb_bt601.g8a

src/render_program/exa_wm_yuv_rgb_bt601.g8b

src/render_program/exa_wm_yuv_rgb_bt709.g4a

src/render_program/exa_wm_yuv_rgb_bt709.g4b

src/render_program/exa_wm_yuv_rgb_bt709.g4b.gen5

src/render_program/exa_wm_yuv_rgb_bt709.g5a

src/render_program/exa_wm_yuv_rgb_bt709.g5b

src/render_program/exa_wm_yuv_rgb_bt709.g6a

src/render_program/exa_wm_yuv_rgb_bt709.g6b

src/render_program/exa_wm_yuv_rgb_bt709.g7a

src/render_program/exa_wm_yuv_rgb_bt709.g7b

src/render_program/exa_wm_yuv_rgb_bt709.g8a

src/render_program/exa_wm_yuv_rgb_bt709.g8b

src/scripts

src/scripts/clock-graph.5c

src/scripts/clock.5c

src/scripts/fix.5c

src/scripts/tv.5c

src/sna

src/sna/Makefile.am

src/sna/README

src/sna/atomic.h

src/sna/blt.c

src/sna/brw

src/sna/brw/Makefile.am

src/sna/brw/brw.h

src/sna/brw/brw_disasm.c

src/sna/brw/brw_eu.c

src/sna/brw/brw_eu.h

src/sna/brw/brw_eu_emit.c

src/sna/brw/brw_sf.c

src/sna/brw/brw_test.c

src/sna/brw/brw_test.h

src/sna/brw/brw_test_gen4.c

src/sna/brw/brw_test_gen5.c

src/sna/brw/brw_test_gen6.c

src/sna/brw/brw_test_gen7.c

src/sna/brw/brw_wm.c

src/sna/brw/meson.build

src/sna/compiler.h

src/sna/debug.h

src/sna/fb

src/sna/fb/Makefile.am

src/sna/fb/README

src/sna/fb/fb.h

src/sna/fb/fbarc.c

src/sna/fb/fbarcbits.h

src/sna/fb/fbbitmap.c

src/sna/fb/fbblt.c

src/sna/fb/fbbltone.c

src/sna/fb/fbclip.c

src/sna/fb/fbclip.h

src/sna/fb/fbcopy.c

src/sna/fb/fbfill.c

src/sna/fb/fbgc.c

src/sna/fb/fbglyph.c

src/sna/fb/fbglyphbits.h

src/sna/fb/fbimage.c

src/sna/fb/fbline.c

src/sna/fb/fblinebits.h

src/sna/fb/fbpict.c

src/sna/fb/fbpict.h

src/sna/fb/fbpoint.c

src/sna/fb/fbpointbits.h

src/sna/fb/fbpush.c

src/sna/fb/fbrop.h

src/sna/fb/fbseg.c

src/sna/fb/fbsegbits.h

src/sna/fb/fbspan.c

src/sna/fb/fbstipple.c

src/sna/fb/fbtile.c

src/sna/fb/fbutil.c

src/sna/fb/meson.build

src/sna/fb/sfb.h

src/sna/gen2_render.c

src/sna/gen2_render.h

src/sna/gen3_render.c

src/sna/gen3_render.h

src/sna/gen4_common.c

src/sna/gen4_common.h

src/sna/gen4_render.c

src/sna/gen4_render.h

src/sna/gen4_source.c

src/sna/gen4_source.h

src/sna/gen4_vertex.c

src/sna/gen4_vertex.h

src/sna/gen5_render.c

src/sna/gen5_render.h

src/sna/gen6_common.c

src/sna/gen6_common.h

src/sna/gen6_render.c

src/sna/gen6_render.h

src/sna/gen7_render.c

src/sna/gen7_render.h

src/sna/gen8_eu.c

src/sna/gen8_eu.h

src/sna/gen8_render.c

src/sna/gen8_render.h

src/sna/gen8_vertex.c

src/sna/gen8_vertex.h

src/sna/gen9_render.c

src/sna/gen9_render.h

src/sna/git_version.h.in

src/sna/kgem.c

src/sna/kgem.h

src/sna/kgem_debug.c

src/sna/kgem_debug.h

src/sna/kgem_debug_gen2.c

src/sna/kgem_debug_gen3.c

src/sna/kgem_debug_gen4.c

src/sna/kgem_debug_gen5.c

src/sna/kgem_debug_gen6.c

src/sna/kgem_debug_gen7.c

src/sna/meson.build

src/sna/rop.h

src/sna/sna.h

src/sna/sna_accel.c

src/sna/sna_acpi.c

src/sna/sna_blt.c

src/sna/sna_composite.c

src/sna/sna_cpu.c

src/sna/sna_cpuid.h

src/sna/sna_damage.c

src/sna/sna_damage.h

src/sna/sna_display.c

src/sna/sna_display_fake.c

src/sna/sna_dri2.c

src/sna/sna_dri3.c

src/sna/sna_driver.c

src/sna/sna_glyphs.c

src/sna/sna_gradient.c

src/sna/sna_io.c

src/sna/sna_module.h

src/sna/sna_present.c

src/sna/sna_reg.h

src/sna/sna_render.c

src/sna/sna_render.h

src/sna/sna_render_inline.h

src/sna/sna_stream.c

src/sna/sna_threads.c

src/sna/sna_tiling.c

src/sna/sna_transform.c

src/sna/sna_trapezoids.c

src/sna/sna_trapezoids.h

src/sna/sna_trapezoids_boxes.c

src/sna/sna_trapezoids_imprecise.c

src/sna/sna_trapezoids_mono.c

src/sna/sna_trapezoids_precise.c

src/sna/sna_vertex.c

src/sna/sna_video.c

src/sna/sna_video.h

src/sna/sna_video_hwmc.c

src/sna/sna_video_hwmc.h

src/sna/sna_video_overlay.c

src/sna/sna_video_sprite.c

src/sna/sna_video_textured.c

src/sna/xassert.h

src/uxa

src/uxa/Makefile.am

src/uxa/brw_defines.h

src/uxa/brw_structs.h

src/uxa/common.h

src/uxa/i830_3d.c

src/uxa/i830_reg.h

src/uxa/i830_render.c

src/uxa/i915_3d.c

src/uxa/i915_3d.h

src/uxa/i915_reg.h

src/uxa/i915_render.c

src/uxa/i915_video.c

src/uxa/i965_3d.c

src/uxa/i965_reg.h

src/uxa/i965_render.c

src/uxa/i965_video.c

src/uxa/intel.h

src/uxa/intel_batchbuffer.c

src/uxa/intel_batchbuffer.h

src/uxa/intel_display.c

src/uxa/intel_dri.c

src/uxa/intel_dri3.c

src/uxa/intel_driver.c

src/uxa/intel_hwmc.c

src/uxa/intel_memory.c

src/uxa/intel_present.c

src/uxa/intel_sync.c

src/uxa/intel_uxa.c

src/uxa/intel_uxa.h

src/uxa/intel_uxa_video.c

src/uxa/intel_video.c

src/uxa/intel_video.h

src/uxa/intel_video_overlay.c

src/uxa/intel_video_overlay.h

src/uxa/meson.build

src/uxa/uxa-accel.c

src/uxa/uxa-glyphs.c

src/uxa/uxa-priv.h

src/uxa/uxa-render.c

src/uxa/uxa-unaccel.c

src/uxa/uxa.c

src/uxa/uxa.h

src/uxa/uxa_module.h

test

test/.gitignore

test/DrawSegments.c

test/Makefile.am

test/README

test/basic-copyarea-size.c

test/basic-copyarea.c

test/basic-copyplane.c

test/basic-fillrect.c

test/basic-lines.c

test/basic-putimage.c

test/basic-rectangle.c

test/basic-stippledrect.c

test/basic-stress.c

test/basic-string.c

test/basic-tiledrect.c

test/cursor-test.c

test/dri2-race.c

test/dri2-speed.c

test/dri2-swap.c

test/dri2-test.c

test/dri2.c

test/dri2.h

test/dri3-test.c

test/dri3.c

test/dri3.h

test/lowlevel-blt-bench.c

test/mixed-stress.c

test/mkvsync.sh

test/present-race.c

test/present-speed.c

test/present-test.c

test/render-composite-solid-mask.c

test/render-composite-solid.c

test/render-copy-alphaless.c

test/render-copyarea-mask.c

test/render-copyarea-size.c

test/render-copyarea.c

test/render-fill-copy.c

test/render-fill.c

test/render-glyphs.c

test/render-trapezoid-image.c

test/render-trapezoid.c

test/render-triangle.c

test/shm-test.c

test/tearing.mp4

test/test.h

test/test_display.c

test/test_image.c

test/test_log.c

test/test_render.c

test/virtual.conf

test/xvidmode.c

tools

tools/.gitignore

tools/Makefile.am

tools/backlight_helper.c

tools/cursor.c

tools/dri3info.c

tools/intel-virtual-output.man

tools/meson.build

tools/org.x.xf86-video-intel.backlight-helper.policy.in

tools/virtual.c

xvmc

xvmc/Makefile.am

xvmc/brw_defines.h

xvmc/brw_structs.h

xvmc/i830_reg.h

xvmc/i915_program.h

xvmc/i915_reg.h

xvmc/i915_structs.h

xvmc/i915_xvmc.c

xvmc/i915_xvmc.h

xvmc/i965_reg.h

xvmc/i965_xvmc.c

xvmc/intel_batchbuffer.c

xvmc/intel_batchbuffer.h

xvmc/intel_xvmc.c

xvmc/intel_xvmc.h

xvmc/intel_xvmc_dump.c

xvmc/intel_xvmc_private.h

xvmc/meson.build

xvmc/shader

xvmc/shader/Makefile.am

xvmc/shader/mc

xvmc/shader/mc/Makefile.am

xvmc/shader/mc/addidct.g4i

xvmc/shader/mc/addidct_igd.g4i

xvmc/shader/mc/block_clear.g4i

xvmc/shader/mc/dual_prime.g4a

xvmc/shader/mc/dual_prime.g4b

xvmc/shader/mc/dual_prime.g4b.gen5

xvmc/shader/mc/dual_prime_igd.g4a

xvmc/shader/mc/dual_prime_igd.g4b

xvmc/shader/mc/dual_prime_igd.g4b.gen5

xvmc/shader/mc/field_backward.g4a

xvmc/shader/mc/field_backward.g4b

xvmc/shader/mc/field_backward.g4b.gen5

xvmc/shader/mc/field_backward_igd.g4a

xvmc/shader/mc/field_backward_igd.g4b

xvmc/shader/mc/field_backward_igd.g4b.gen5

xvmc/shader/mc/field_f_b.g4a

xvmc/shader/mc/field_f_b.g4b

xvmc/shader/mc/field_f_b.g4b.gen5

xvmc/shader/mc/field_f_b_igd.g4a

xvmc/shader/mc/field_f_b_igd.g4b

xvmc/shader/mc/field_f_b_igd.g4b.gen5

xvmc/shader/mc/field_forward.g4a

xvmc/shader/mc/field_forward.g4b

xvmc/shader/mc/field_forward.g4b.gen5

xvmc/shader/mc/field_forward_igd.g4a

xvmc/shader/mc/field_forward_igd.g4b

xvmc/shader/mc/field_forward_igd.g4b.gen5

xvmc/shader/mc/frame_backward.g4a

xvmc/shader/mc/frame_backward.g4b

xvmc/shader/mc/frame_backward.g4b.gen5

xvmc/shader/mc/frame_backward_igd.g4a

xvmc/shader/mc/frame_backward_igd.g4b

xvmc/shader/mc/frame_backward_igd.g4b.gen5

xvmc/shader/mc/frame_f_b.g4a

xvmc/shader/mc/frame_f_b.g4b

xvmc/shader/mc/frame_f_b.g4b.gen5

xvmc/shader/mc/frame_f_b_igd.g4a

xvmc/shader/mc/frame_f_b_igd.g4b

xvmc/shader/mc/frame_f_b_igd.g4b.gen5

xvmc/shader/mc/frame_forward.g4a

xvmc/shader/mc/frame_forward.g4b

xvmc/shader/mc/frame_forward.g4b.gen5

xvmc/shader/mc/frame_forward_igd.g4a

xvmc/shader/mc/frame_forward_igd.g4b

xvmc/shader/mc/frame_forward_igd.g4b.gen5

xvmc/shader/mc/ipicture.g4a

xvmc/shader/mc/ipicture.g4b

xvmc/shader/mc/ipicture.g4b.gen5

xvmc/shader/mc/ipicture_igd.g4a

xvmc/shader/mc/ipicture_igd.g4b

xvmc/shader/mc/ipicture_igd.g4b.gen5

xvmc/shader/mc/lib_igd.g4a

xvmc/shader/mc/lib_igd.g4b

xvmc/shader/mc/lib_igd.g4b.gen5

xvmc/shader/mc/motion_field_uv.g4i

xvmc/shader/mc/motion_field_uv_igd.g4i

xvmc/shader/mc/motion_field_y.g4i

xvmc/shader/mc/motion_field_y_igd.g4i

xvmc/shader/mc/motion_frame_uv.g4i

xvmc/shader/mc/motion_frame_uv_igd.g4i

xvmc/shader/mc/motion_frame_y.g4i

xvmc/shader/mc/motion_frame_y_igd.g4i

xvmc/shader/mc/null.g4a

xvmc/shader/mc/null.g4b

xvmc/shader/mc/null.g4b.gen5

xvmc/shader/mc/read_field_x0y0_uv.g4i

xvmc/shader/mc/read_field_x0y0_uv_igd.g4i

xvmc/shader/mc/read_field_x0y0_y.g4i

xvmc/shader/mc/read_field_x0y0_y_igd.g4i

xvmc/shader/mc/read_field_x0y1_uv.g4i

xvmc/shader/mc/read_field_x0y1_uv_igd.g4i

xvmc/shader/mc/read_field_x0y1_y.g4i

xvmc/shader/mc/read_field_x0y1_y_igd.g4i

xvmc/shader/mc/read_field_x1y0_uv.g4i

xvmc/shader/mc/read_field_x1y0_uv_igd.g4i

xvmc/shader/mc/read_field_x1y0_y.g4i

xvmc/shader/mc/read_field_x1y0_y_igd.g4i

xvmc/shader/mc/read_field_x1y1_uv.g4i

xvmc/shader/mc/read_field_x1y1_uv_igd.g4i

xvmc/shader/mc/read_field_x1y1_y.g4i

xvmc/shader/mc/read_field_x1y1_y_igd.g4i

xvmc/shader/mc/read_frame_x0y0_uv.g4i

xvmc/shader/mc/read_frame_x0y0_uv_igd.g4i

xvmc/shader/mc/read_frame_x0y0_y.g4i

xvmc/shader/mc/read_frame_x0y0_y_igd.g4i

xvmc/shader/mc/read_frame_x0y1_uv.g4i

xvmc/shader/mc/read_frame_x0y1_uv_igd.g4i

xvmc/shader/mc/read_frame_x0y1_y.g4i

xvmc/shader/mc/read_frame_x0y1_y_igd.g4i

xvmc/shader/mc/read_frame_x1y0_uv.g4i

xvmc/shader/mc/read_frame_x1y0_uv_igd.g4i

xvmc/shader/mc/read_frame_x1y0_y.g4i

xvmc/shader/mc/read_frame_x1y0_y_igd.g4i

xvmc/shader/mc/read_frame_x1y1_uv.g4i

xvmc/shader/mc/read_frame_x1y1_uv_igd.g4i

xvmc/shader/mc/read_frame_x1y1_y.g4i

xvmc/shader/mc/read_frame_x1y1_y_igd.g4i

xvmc/shader/vld

xvmc/shader/vld/Makefile.am

xvmc/shader/vld/addidct.g4i

xvmc/shader/vld/do_iq_intra.g4i

xvmc/shader/vld/do_iq_non_intra.g4i

xvmc/shader/vld/field_backward.g4a

xvmc/shader/vld/field_backward.g4b

xvmc/shader/vld/field_backward.g4b.gen5

xvmc/shader/vld/field_f_b.g4a

xvmc/shader/vld/field_f_b.g4b

xvmc/shader/vld/field_f_b.g4b.gen5

xvmc/shader/vld/field_forward.g4a

xvmc/shader/vld/field_forward.g4b

xvmc/shader/vld/field_forward.g4b.gen5

xvmc/shader/vld/frame_backward.g4a

xvmc/shader/vld/frame_backward.g4b

xvmc/shader/vld/frame_backward.g4b.gen5

xvmc/shader/vld/frame_f_b.g4a

xvmc/shader/vld/frame_f_b.g4b

xvmc/shader/vld/frame_f_b.g4b.gen5

xvmc/shader/vld/frame_forward.g4a

xvmc/shader/vld/frame_forward.g4b

xvmc/shader/vld/frame_forward.g4b.gen5

xvmc/shader/vld/idct.g4i

xvmc/shader/vld/ipicture.g4a

xvmc/shader/vld/ipicture.g4b

xvmc/shader/vld/ipicture.g4b.gen5

xvmc/shader/vld/iq_intra.g4i

xvmc/shader/vld/iq_non_intra.g4i

xvmc/shader/vld/lib.g4a

xvmc/shader/vld/lib.g4b

xvmc/shader/vld/lib.g4b.gen5

xvmc/shader/vld/motion_field_uv.g4i

xvmc/shader/vld/motion_field_y.g4i

xvmc/shader/vld/motion_frame_uv.g4i

xvmc/shader/vld/motion_frame_y.g4i

xvmc/shader/vld/read_field_x0y0_uv.g4i

xvmc/shader/vld/read_field_x0y0_y.g4i

xvmc/shader/vld/read_field_x0y1_uv.g4i

xvmc/shader/vld/read_field_x0y1_y.g4i

xvmc/shader/vld/read_field_x1y0_uv.g4i

xvmc/shader/vld/read_field_x1y0_y.g4i

xvmc/shader/vld/read_field_x1y1_uv.g4i

xvmc/shader/vld/read_field_x1y1_y.g4i

xvmc/shader/vld/read_frame_x0y0_uv.g4i

xvmc/shader/vld/read_frame_x0y0_y.g4i

xvmc/shader/vld/read_frame_x0y1_uv.g4i

xvmc/shader/vld/read_frame_x0y1_y.g4i

xvmc/shader/vld/read_frame_x1y0_uv.g4i

xvmc/shader/vld/read_frame_x1y0_y.g4i

xvmc/shader/vld/read_frame_x1y1_uv.g4i

xvmc/shader/vld/read_frame_x1y1_y.g4i

xvmc/xvmc_vld.c

Show diffs side-by-side

added added

removed removed

src/sna/gen2_render.c

* Permission is hereby granted, free of charge, to any person obtaining a

* copy of this software and associated documentation files (the "Software"),

* to deal in the Software without restriction, including without limitation

* the rights to use, copy, modify, merge, publish, distribute, sublicense,

* and/or sell copies of the Software, and to permit persons to whom the

* Software is furnished to do so, subject to the following conditions:

* The above copyright notice and this permission notice (including the next

* paragraph) shall be included in all copies or substantial portions of the

* Software.

* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR

* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,

* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL

* THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER

* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,

* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE

* SOFTWARE.

* Authors:

* Wang Zhenyu <zhenyu.z.wang@intel.com>

* Eric Anholt <eric@anholt.net>

* Chris Wilson <chris@chris-wilson.co.uk>

#ifdef HAVE_CONFIG_H

#include "config.h"

#endif

#include "sna.h"

#include "sna_reg.h"

#include "sna_render.h"

#include "sna_render_inline.h"

#include "sna_video.h"

#include "gen2_render.h"

#define NO_COMPOSITE 0

#define NO_COMPOSITE_SPANS 0

#define NO_COPY 0

#define NO_COPY_BOXES 0

#define NO_FILL 0

#define NO_FILL_ONE 0

#define NO_FILL_BOXES 0

#define MAX_3D_SIZE 2048

#define MAX_3D_PITCH 8192

#define MAX_INLINE (1 << 18)

#define BATCH(v) batch_emit(sna, v)

#define BATCH_ALIGNED(v, a) batch_emit_aligned(sna, v, a)

#define BATCH_F(v) batch_emit_float(sna, v)

#define VERTEX(v) batch_emit_float(sna, v)

static const struct blendinfo {

bool dst_alpha;

bool src_alpha;

uint32_t src_blend;

uint32_t dst_blend;

} gen2_blend_op[] = {

/* Clear */

{0, 0, BLENDFACTOR_ZERO, BLENDFACTOR_ZERO},

/* Src */

{0, 0, BLENDFACTOR_ONE, BLENDFACTOR_ZERO},

/* Dst */

{0, 0, BLENDFACTOR_ZERO, BLENDFACTOR_ONE},

/* Over */

{0, 1, BLENDFACTOR_ONE, BLENDFACTOR_INV_SRC_ALPHA},

/* OverReverse */

{1, 0, BLENDFACTOR_INV_DST_ALPHA, BLENDFACTOR_ONE},

/* In */

{1, 0, BLENDFACTOR_DST_ALPHA, BLENDFACTOR_ZERO},

/* InReverse */

{0, 1, BLENDFACTOR_ZERO, BLENDFACTOR_SRC_ALPHA},

/* Out */

{1, 0, BLENDFACTOR_INV_DST_ALPHA, BLENDFACTOR_ZERO},

/* OutReverse */

{0, 1, BLENDFACTOR_ZERO, BLENDFACTOR_INV_SRC_ALPHA},

/* Atop */

{1, 1, BLENDFACTOR_DST_ALPHA, BLENDFACTOR_INV_SRC_ALPHA},

/* AtopReverse */

{1, 1, BLENDFACTOR_INV_DST_ALPHA, BLENDFACTOR_SRC_ALPHA},

/* Xor */

{1, 1, BLENDFACTOR_INV_DST_ALPHA, BLENDFACTOR_INV_SRC_ALPHA},

/* Add */

{0, 0, BLENDFACTOR_ONE, BLENDFACTOR_ONE},

};

static const struct formatinfo {

unsigned int fmt;

uint32_t card_fmt;

} i8xx_tex_formats[] = {

{PICT_a8, MAPSURF_8BIT | MT_8BIT_A8},

{PICT_a8r8g8b8, MAPSURF_32BIT | MT_32BIT_ARGB8888},

{PICT_a8b8g8r8, MAPSURF_32BIT | MT_32BIT_ABGR8888},

100

{PICT_r5g6b5, MAPSURF_16BIT | MT_16BIT_RGB565},

101

{PICT_a1r5g5b5, MAPSURF_16BIT | MT_16BIT_ARGB1555},

102

{PICT_a4r4g4b4, MAPSURF_16BIT | MT_16BIT_ARGB4444},

103

}, i85x_tex_formats[] = {

104

{PICT_x8r8g8b8, MAPSURF_32BIT | MT_32BIT_XRGB8888},

105

{PICT_x8b8g8r8, MAPSURF_32BIT | MT_32BIT_XBGR8888},

106

};

107

108

static inline bool

109

too_large(int width, int height)

110

{

111

return width > MAX_3D_SIZE || height > MAX_3D_SIZE;

112

}

113

114

static inline uint32_t

115

gen2_buf_tiling(uint32_t tiling)

116

{

117

uint32_t v = 0;

118

switch (tiling) {

119

default:

120

assert(0);

121

/* fall through */

122

case I915_TILING_Y:

123

v |= BUF_3D_TILE_WALK_Y;

124

/* fall through */

125

case I915_TILING_X:

126

v |= BUF_3D_TILED_SURFACE;

127

/* fall through */

128

case I915_TILING_NONE:

129

break;

130

}

131

return v;

132

}

133

134

static uint32_t

135

gen2_get_dst_format(uint32_t format)

136

{

137

#define BIAS DSTORG_HORT_BIAS(0x8) | DSTORG_VERT_BIAS(0x8)

138

switch (format) {

139

default:

140

assert(0);

141

/* fall through */

142

case PICT_a8r8g8b8:

143

case PICT_x8r8g8b8:

144

return COLR_BUF_ARGB8888 | BIAS;

145

case PICT_r5g6b5:

146

return COLR_BUF_RGB565 | BIAS;

147

case PICT_a1r5g5b5:

148

case PICT_x1r5g5b5:

149

return COLR_BUF_ARGB1555 | BIAS;

150

case PICT_a8:

151

return COLR_BUF_8BIT | BIAS;

152

case PICT_a4r4g4b4:

153

case PICT_x4r4g4b4:

154

return COLR_BUF_ARGB4444 | BIAS;

155

}

156

#undef BIAS

157

}

158

159

static bool

160

gen2_check_dst_format(uint32_t format)

161

{

162

switch (format) {

163

case PICT_a8r8g8b8:

164

case PICT_x8r8g8b8:

165

case PICT_r5g6b5:

166

case PICT_a1r5g5b5:

167

case PICT_x1r5g5b5:

168

case PICT_a8:

169

case PICT_a4r4g4b4:

170

case PICT_x4r4g4b4:

171

return true;

172

default:

173

return false;

174

}

175

}

176

177

static uint32_t

178

gen2_get_card_format(struct sna *sna, uint32_t format)

179

{

180

unsigned int i;

181

182

for (i = 0; i < ARRAY_SIZE(i8xx_tex_formats); i++)

183

if (i8xx_tex_formats[i].fmt == format)

184

return i8xx_tex_formats[i].card_fmt;

185

186

if (sna->kgem.gen < 021) {

187

/* Whilst these are not directly supported on 830/845,

188

* we only enable them when we can implicitly convert

189

* them to a supported variant through the texture

190

* combiners.

191

192

for (i = 0; i < ARRAY_SIZE(i85x_tex_formats); i++)

193

if (i85x_tex_formats[i].fmt == format)

194

return i8xx_tex_formats[1+i].card_fmt;

195

} else {

196

for (i = 0; i < ARRAY_SIZE(i85x_tex_formats); i++)

197

if (i85x_tex_formats[i].fmt == format)

198

return i85x_tex_formats[i].card_fmt;

199

}

200

201

assert(0);

202

return 0;

203

}

204

205

static uint32_t

206

gen2_check_format(struct sna *sna, PicturePtr p)

207

{

208

unsigned int i;

209

210

for (i = 0; i < ARRAY_SIZE(i8xx_tex_formats); i++)

211

if (i8xx_tex_formats[i].fmt == p->format)

212

return true;

213

214

if (sna->kgem.gen > 021) {

215

for (i = 0; i < ARRAY_SIZE(i85x_tex_formats); i++)

216

if (i85x_tex_formats[i].fmt == p->format)

217

return true;

218

}

219

220

return false;

221

}

222

223

static uint32_t

224

gen2_sampler_tiling_bits(uint32_t tiling)

225

{

226

uint32_t bits = 0;

227

switch (tiling) {

228

default:

229

assert(0);

230

/* fall through */

231

case I915_TILING_Y:

232

bits |= TM0S1_TILE_WALK;

233

/* fall through */

234

case I915_TILING_X:

235

bits |= TM0S1_TILED_SURFACE;

236

/* fall through */

237

case I915_TILING_NONE:

238

break;

239

}

240

return bits;

241

}

242

243

static bool

244

gen2_check_filter(PicturePtr picture)

245

{

246

switch (picture->filter) {

247

case PictFilterNearest:

248

case PictFilterBilinear:

249

return true;

250

default:

251

return false;

252

}

253

}

254

255

static bool

256

gen2_check_repeat(PicturePtr picture)

257

{

258

if (!picture->repeat)

259

return true;

260

261

switch (picture->repeatType) {

262

case RepeatNone:

263

case RepeatNormal:

264

case RepeatPad:

265

case RepeatReflect:

266

return true;

267

default:

268

return false;

269

}

270

}

271

272

static void

273

gen2_emit_texture(struct sna *sna,

274

const struct sna_composite_channel *channel,

275

int unit)

276

{

277

uint32_t wrap_mode_u, wrap_mode_v;

278

uint32_t texcoordtype;

279

uint32_t filter;

280

281

assert(channel->bo);

282

283

if (channel->is_affine)

284

texcoordtype = TEXCOORDTYPE_CARTESIAN;

285

else

286

texcoordtype = TEXCOORDTYPE_HOMOGENEOUS;

287

288

switch (channel->repeat) {

289

default:

290

assert(0);

291

/* fall through */

292

case RepeatNone:

293

wrap_mode_u = TEXCOORDMODE_CLAMP_BORDER;

294

break;

295

case RepeatNormal:

296

wrap_mode_u = TEXCOORDMODE_WRAP;

297

break;

298

case RepeatPad:

299

wrap_mode_u = TEXCOORDMODE_CLAMP;

300

break;

301

case RepeatReflect:

302

wrap_mode_u = TEXCOORDMODE_MIRROR;

303

break;

304

}

305

if (channel->is_linear)

306

wrap_mode_v = TEXCOORDMODE_WRAP;

307

else

308

wrap_mode_v = wrap_mode_u;

309

310

switch (channel->filter) {

311

default:

312

assert(0);

313

/* fall through */

314

case PictFilterNearest:

315

filter = (FILTER_NEAREST << TM0S3_MAG_FILTER_SHIFT |

316

FILTER_NEAREST << TM0S3_MIN_FILTER_SHIFT |

317

MIPFILTER_NONE << TM0S3_MIP_FILTER_SHIFT);

318

break;

319

case PictFilterBilinear:

320

filter = (FILTER_LINEAR << TM0S3_MAG_FILTER_SHIFT |

321

FILTER_LINEAR << TM0S3_MIN_FILTER_SHIFT |

322

MIPFILTER_NONE << TM0S3_MIP_FILTER_SHIFT);

323

break;

324

}

325

326

BATCH(_3DSTATE_LOAD_STATE_IMMEDIATE_2 | LOAD_TEXTURE_MAP(unit) | 4);

327

BATCH(kgem_add_reloc(&sna->kgem, sna->kgem.nbatch,

328

channel->bo,

329

I915_GEM_DOMAIN_SAMPLER << 16,

330

0));

331

BATCH(((channel->height - 1) << TM0S1_HEIGHT_SHIFT) |

332

((channel->width - 1) << TM0S1_WIDTH_SHIFT) |

333

gen2_get_card_format(sna, channel->pict_format) |

334

gen2_sampler_tiling_bits(channel->bo->tiling));

335

BATCH((channel->bo->pitch / 4 - 1) << TM0S2_PITCH_SHIFT | TM0S2_MAP_2D);

336

BATCH(filter);

337

BATCH(0); /* default color */

338

339

BATCH(_3DSTATE_MAP_COORD_SET_CMD | TEXCOORD_SET(unit) |

340

ENABLE_TEXCOORD_PARAMS | TEXCOORDS_ARE_NORMAL | texcoordtype |

341

ENABLE_ADDR_V_CNTL | TEXCOORD_ADDR_V_MODE(wrap_mode_v) |

342

ENABLE_ADDR_U_CNTL | TEXCOORD_ADDR_U_MODE(wrap_mode_u));

343

}

344

345

static void

346

gen2_get_blend_factors(const struct sna_composite_op *op,

347

int blend,

348

uint32_t *c_out,

349

uint32_t *a_out)

350

{

351

uint32_t cblend, ablend;

352

353

/* If component alpha is active in the mask and the blend operation

354

* uses the source alpha, then we know we don't need the source

355

* value (otherwise we would have hit a fallback earlier), so we

356

* provide the source alpha (src.A * mask.X) as output color.

357

* Conversely, if CA is set and we don't need the source alpha, then

358

* we produce the source value (src.X * mask.X) and the source alpha

359

* is unused.. Otherwise, we provide the non-CA source value

360

* (src.X * mask.A).

361

362

* The PICT_FORMAT_RGB(pict) == 0 fixups are not needed on 855+'s a8

363

* pictures, but we need to implement it for 830/845 and there's no

364

* harm done in leaving it in.

365

366

cblend = TB0C_LAST_STAGE | TB0C_RESULT_SCALE_1X | TB0C_OUTPUT_WRITE_CURRENT;

367

ablend = TB0A_RESULT_SCALE_1X | TB0A_OUTPUT_WRITE_CURRENT;

368

369

/* Get the source picture's channels into TBx_ARG1 */

370

if ((op->has_component_alpha && gen2_blend_op[blend].src_alpha) ||

371

op->dst.format == PICT_a8) {

372

/* Producing source alpha value, so the first set of channels

373

* is src.A instead of src.X. We also do this if the destination

374

* is a8, in which case src.G is what's written, and the other

375

* channels are ignored.

376

377

if (op->src.is_opaque) {

378

ablend |= TB0C_ARG1_SEL_ONE;

379

cblend |= TB0C_ARG1_SEL_ONE;

380

} else if (op->src.is_solid) {

381

ablend |= TB0C_ARG1_SEL_DIFFUSE;

382

cblend |= TB0C_ARG1_SEL_DIFFUSE | TB0C_ARG1_REPLICATE_ALPHA;

383

} else {

384

ablend |= TB0C_ARG1_SEL_TEXEL0;

385

cblend |= TB0C_ARG1_SEL_TEXEL0 | TB0C_ARG1_REPLICATE_ALPHA;

386

}

387

} else {

388

if (op->src.is_solid)

389

cblend |= TB0C_ARG1_SEL_DIFFUSE;

390

else if (PICT_FORMAT_RGB(op->src.pict_format) != 0)

391

cblend |= TB0C_ARG1_SEL_TEXEL0;

392

else

393

cblend |= TB0C_ARG1_SEL_ONE | TB0C_ARG1_INVERT; /* 0.0 */

394

395

if (op->src.is_opaque)

396

ablend |= TB0A_ARG1_SEL_ONE;

397

else if (op->src.is_solid)

398

ablend |= TB0A_ARG1_SEL_DIFFUSE;

399

else

400

ablend |= TB0A_ARG1_SEL_TEXEL0;

401

}

402

403

if (op->mask.bo) {

404

if (op->src.is_solid) {

405

cblend |= TB0C_ARG2_SEL_TEXEL0;

406

ablend |= TB0A_ARG2_SEL_TEXEL0;

407

} else {

408

cblend |= TB0C_ARG2_SEL_TEXEL1;

409

ablend |= TB0A_ARG2_SEL_TEXEL1;

410

}

411

412

if (op->dst.format == PICT_a8 || !op->has_component_alpha)

413

cblend |= TB0C_ARG2_REPLICATE_ALPHA;

414

415

cblend |= TB0C_OP_MODULATE;

416

ablend |= TB0A_OP_MODULATE;

417

} else if (op->mask.is_solid) {

418

cblend |= TB0C_ARG2_SEL_DIFFUSE;

419

ablend |= TB0A_ARG2_SEL_DIFFUSE;

420

421

if (op->dst.format == PICT_a8 || !op->has_component_alpha)

422

cblend |= TB0C_ARG2_REPLICATE_ALPHA;

423

424

cblend |= TB0C_OP_MODULATE;

425

ablend |= TB0A_OP_MODULATE;

426

} else {

427

cblend |= TB0C_OP_ARG1;

428

ablend |= TB0A_OP_ARG1;

429

}

430

431

*c_out = cblend;

432

*a_out = ablend;

433

}

434

435

static uint32_t gen2_get_blend_cntl(int op,

436

bool has_component_alpha,

437

uint32_t dst_format)

438

{

439

uint32_t sblend, dblend;

440

441

if (op <= PictOpSrc)

442

return S8_ENABLE_COLOR_BUFFER_WRITE;

443

444

sblend = gen2_blend_op[op].src_blend;

445

dblend = gen2_blend_op[op].dst_blend;

446

447

if (gen2_blend_op[op].dst_alpha) {

448

/* If there's no dst alpha channel, adjust the blend op so that

449

* we'll treat it as always 1.

450

451

if (PICT_FORMAT_A(dst_format) == 0) {

452

if (sblend == BLENDFACTOR_DST_ALPHA)

453

sblend = BLENDFACTOR_ONE;

454

else if (sblend == BLENDFACTOR_INV_DST_ALPHA)

455

sblend = BLENDFACTOR_ZERO;

456

}

457

458

/* gen2 engine reads 8bit color buffer into green channel

459

* in cases like color buffer blending etc., and also writes

460

* back green channel. So with dst_alpha blend we should use

461

* color factor.

462

463

if (dst_format == PICT_a8) {

464

if (sblend == BLENDFACTOR_DST_ALPHA)

465

sblend = BLENDFACTOR_DST_COLR;

466

else if (sblend == BLENDFACTOR_INV_DST_ALPHA)

467

sblend = BLENDFACTOR_INV_DST_COLR;

468

}

469

}

470

471

/* If the source alpha is being used, then we should only be in a case

472

* where the source blend factor is 0, and the source blend value is

473

* the mask channels multiplied by the source picture's alpha.

474

475

if (has_component_alpha && gen2_blend_op[op].src_alpha) {

476

if (dblend == BLENDFACTOR_SRC_ALPHA)

477

dblend = BLENDFACTOR_SRC_COLR;

478

else if (dblend == BLENDFACTOR_INV_SRC_ALPHA)

479

dblend = BLENDFACTOR_INV_SRC_COLR;

480

}

481

482

return (sblend << S8_SRC_BLEND_FACTOR_SHIFT |

483

dblend << S8_DST_BLEND_FACTOR_SHIFT |

484

S8_ENABLE_COLOR_BLEND | S8_BLENDFUNC_ADD |

485

S8_ENABLE_COLOR_BUFFER_WRITE);

486

}

487

488

static void gen2_emit_invariant(struct sna *sna)

489

{

490

int i;

491

492

for (i = 0; i < 4; i++) {

493

BATCH(_3DSTATE_MAP_CUBE | MAP_UNIT(i));

494

BATCH(_3DSTATE_MAP_TEX_STREAM_CMD | MAP_UNIT(i) |

495

DISABLE_TEX_STREAM_BUMP |

496

ENABLE_TEX_STREAM_COORD_SET | TEX_STREAM_COORD_SET(i) |

497

ENABLE_TEX_STREAM_MAP_IDX | TEX_STREAM_MAP_IDX(i));

498

BATCH(_3DSTATE_MAP_COORD_TRANSFORM);

499

BATCH(DISABLE_TEX_TRANSFORM | TEXTURE_SET(i));

500

}

501

502

BATCH(_3DSTATE_MAP_COORD_SETBIND_CMD);

503

BATCH(TEXBIND_SET3(TEXCOORDSRC_VTXSET_3) |

504

TEXBIND_SET2(TEXCOORDSRC_VTXSET_2) |

505

TEXBIND_SET1(TEXCOORDSRC_VTXSET_1) |

506

TEXBIND_SET0(TEXCOORDSRC_VTXSET_0));

507

508

BATCH(_3DSTATE_SCISSOR_ENABLE_CMD | DISABLE_SCISSOR_RECT);

509

510

BATCH(_3DSTATE_VERTEX_TRANSFORM);

511

BATCH(DISABLE_VIEWPORT_TRANSFORM | DISABLE_PERSPECTIVE_DIVIDE);

512

513

BATCH(_3DSTATE_W_STATE_CMD);

514

BATCH(MAGIC_W_STATE_DWORD1);

515

BATCH_F(1.0);

516

517

BATCH(_3DSTATE_INDPT_ALPHA_BLEND_CMD |

518

DISABLE_INDPT_ALPHA_BLEND |

519

ENABLE_ALPHA_BLENDFUNC | ABLENDFUNC_ADD);

520

521

BATCH(_3DSTATE_CONST_BLEND_COLOR_CMD);

522

BATCH(0);

523

524

BATCH(_3DSTATE_MODES_1_CMD |

525

ENABLE_COLR_BLND_FUNC | BLENDFUNC_ADD |

526

ENABLE_SRC_BLND_FACTOR | SRC_BLND_FACT(BLENDFACTOR_ONE) |

527

ENABLE_DST_BLND_FACTOR | DST_BLND_FACT(BLENDFACTOR_ZERO));

528

529

BATCH(_3DSTATE_ENABLES_1_CMD |

530

DISABLE_LOGIC_OP |

531

DISABLE_STENCIL_TEST |

532

DISABLE_DEPTH_BIAS |

533

DISABLE_SPEC_ADD |

534

DISABLE_FOG |

535

DISABLE_ALPHA_TEST |

536

DISABLE_DEPTH_TEST |

537

ENABLE_COLOR_BLEND);

538

539

BATCH(_3DSTATE_ENABLES_2_CMD |

540

DISABLE_STENCIL_WRITE |

541

DISABLE_DITHER |

542

DISABLE_DEPTH_WRITE |

543

ENABLE_COLOR_MASK |

544

ENABLE_COLOR_WRITE |

545

ENABLE_TEX_CACHE);

546

547

BATCH(_3DSTATE_STIPPLE);

548

BATCH(0);

549

550

BATCH(_3DSTATE_MAP_BLEND_OP_CMD(0) |

551

TEXPIPE_COLOR |

552

ENABLE_TEXOUTPUT_WRT_SEL |

553

TEXOP_OUTPUT_CURRENT |

554

DISABLE_TEX_CNTRL_STAGE |

555

TEXOP_SCALE_1X |

556

TEXOP_MODIFY_PARMS | TEXOP_LAST_STAGE |

557

TEXBLENDOP_ARG1);

558

BATCH(_3DSTATE_MAP_BLEND_OP_CMD(0) |

559

TEXPIPE_ALPHA |

560

ENABLE_TEXOUTPUT_WRT_SEL |

561

TEXOP_OUTPUT_CURRENT |

562

TEXOP_SCALE_1X | TEXOP_MODIFY_PARMS |

563

TEXBLENDOP_ARG1);

564

BATCH(_3DSTATE_MAP_BLEND_ARG_CMD(0) |

565

TEXPIPE_COLOR |

566

TEXBLEND_ARG1 |

567

TEXBLENDARG_MODIFY_PARMS |

568

TEXBLENDARG_DIFFUSE);

569

BATCH(_3DSTATE_MAP_BLEND_ARG_CMD(0) |

570

TEXPIPE_ALPHA |

571

TEXBLEND_ARG1 |

572

TEXBLENDARG_MODIFY_PARMS |

573

TEXBLENDARG_DIFFUSE);

574

575

#define INVARIANT_SIZE 35

576

577

sna->render_state.gen2.need_invariant = false;

578

}

579

580

static void

581

gen2_get_batch(struct sna *sna, const struct sna_composite_op *op)

582

{

583

kgem_set_mode(&sna->kgem, KGEM_RENDER, op->dst.bo);

584

585

/* +7 for i830 3DSTATE_BUFFER_INFO w/a */

586

if (!kgem_check_batch(&sna->kgem, INVARIANT_SIZE+40+7)) {

587

DBG(("%s: flushing batch: size %d > %d\n",

588

__FUNCTION__, INVARIANT_SIZE+40,

589

sna->kgem.surface-sna->kgem.nbatch));

590

kgem_submit(&sna->kgem);

591

_kgem_set_mode(&sna->kgem, KGEM_RENDER);

592

}

593

594

if (!kgem_check_reloc(&sna->kgem, 3)) {

595

DBG(("%s: flushing batch: reloc %d >= %d\n",

596

__FUNCTION__,

597

sna->kgem.nreloc + 3,

598

(int)KGEM_RELOC_SIZE(&sna->kgem)));

599

kgem_submit(&sna->kgem);

600

_kgem_set_mode(&sna->kgem, KGEM_RENDER);

601

}

602

603

if (!kgem_check_exec(&sna->kgem, 3)) {

604

DBG(("%s: flushing batch: exec %d >= %d\n",

605

__FUNCTION__,

606

sna->kgem.nexec + 1,

607

(int)KGEM_EXEC_SIZE(&sna->kgem)));

608

kgem_submit(&sna->kgem);

609

_kgem_set_mode(&sna->kgem, KGEM_RENDER);

610

}

611

612

if (sna->render_state.gen2.need_invariant)

613

gen2_emit_invariant(sna);

614

}

615

616

static void gen2_emit_target(struct sna *sna,

617

struct kgem_bo *bo,

618

int width,

619

int height,

620

int format)

621

{

622

assert(!too_large(width, height));

623

assert(bo->pitch >= 8 && bo->pitch <= MAX_3D_PITCH);

624

assert(sna->render.vertex_offset == 0);

625

626

assert(bo->unique_id);

627

if (sna->render_state.gen2.target == bo->unique_id) {

628

kgem_bo_mark_dirty(bo);

629

return;

630

}

631

632

633

* i830 w/a: 3DSTATE_BUFFER_INFO

634

* must not straddle two cachelines.

635

636

if (intel_get_device_id(sna->dev) == 0x3577)

637

BATCH_ALIGNED(_3DSTATE_BUF_INFO_CMD, 8);

638

else

639

BATCH(_3DSTATE_BUF_INFO_CMD);

640

BATCH(BUF_3D_ID_COLOR_BACK |

641

gen2_buf_tiling(bo->tiling) |

642

BUF_3D_PITCH(bo->pitch));

643

BATCH(kgem_add_reloc(&sna->kgem, sna->kgem.nbatch,

644

bo,

645

I915_GEM_DOMAIN_RENDER << 16 |

646

I915_GEM_DOMAIN_RENDER,

647

0));

648

649

BATCH(_3DSTATE_DST_BUF_VARS_CMD);

650

BATCH(gen2_get_dst_format(format));

651

652

BATCH(_3DSTATE_DRAW_RECT_CMD);

653

BATCH(0);

654

BATCH(0); /* ymin, xmin */

655

BATCH(DRAW_YMAX(height - 1) |

656

DRAW_XMAX(width - 1));

657

BATCH(0); /* yorig, xorig */

658

659

sna->render_state.gen2.target = bo->unique_id;

660

}

661

662

static void gen2_disable_logic_op(struct sna *sna)

663

{

664

if (!sna->render_state.gen2.logic_op_enabled)

665

return;

666

667

DBG(("%s\n", __FUNCTION__));

668

669

BATCH(_3DSTATE_ENABLES_1_CMD |

670

DISABLE_LOGIC_OP | ENABLE_COLOR_BLEND);

671

672

sna->render_state.gen2.logic_op_enabled = 0;

673

}

674

675

static void gen2_enable_logic_op(struct sna *sna, int op)

676

{

677

static const uint8_t logic_op[] = {

678

LOGICOP_CLEAR, /* GXclear */

679

LOGICOP_AND, /* GXand */

680

LOGICOP_AND_RVRSE, /* GXandReverse */

681

LOGICOP_COPY, /* GXcopy */

682

LOGICOP_AND_INV, /* GXandInverted */

683

LOGICOP_NOOP, /* GXnoop */

684

LOGICOP_XOR, /* GXxor */

685

LOGICOP_OR, /* GXor */

686

LOGICOP_NOR, /* GXnor */

687

LOGICOP_EQUIV, /* GXequiv */

688

LOGICOP_INV, /* GXinvert */

689

LOGICOP_OR_RVRSE, /* GXorReverse */

690

LOGICOP_COPY_INV, /* GXcopyInverted */

691

LOGICOP_OR_INV, /* GXorInverted */

692

LOGICOP_NAND, /* GXnand */

693

LOGICOP_SET /* GXset */

694

};

695

696

if (sna->render_state.gen2.logic_op_enabled != op+1) {

697

if (!sna->render_state.gen2.logic_op_enabled) {

698

if (op == GXclear || op == GXcopy)

699

return;

700

701

DBG(("%s\n", __FUNCTION__));

702

703

BATCH(_3DSTATE_ENABLES_1_CMD |

704

ENABLE_LOGIC_OP | DISABLE_COLOR_BLEND);

705

}

706

707

BATCH(_3DSTATE_MODES_4_CMD |

708

ENABLE_LOGIC_OP_FUNC | LOGIC_OP_FUNC(logic_op[op]));

709

sna->render_state.gen2.logic_op_enabled = op+1;

710

}

711

}

712

713

static void gen2_emit_composite_state(struct sna *sna,

714

const struct sna_composite_op *op)

715

{

716

uint32_t texcoordfmt, v, unwind;

717

uint32_t cblend, ablend;

718

int tex;

719

720

gen2_get_batch(sna, op);

721

722

if (kgem_bo_is_dirty(op->src.bo) || kgem_bo_is_dirty(op->mask.bo)) {

723

if (op->src.bo == op->dst.bo || op->mask.bo == op->dst.bo)

724

BATCH(MI_FLUSH | MI_INVALIDATE_MAP_CACHE);

725

else

726

BATCH(_3DSTATE_MODES_5_CMD |

727

PIPELINE_FLUSH_RENDER_CACHE |

728

PIPELINE_FLUSH_TEXTURE_CACHE);

729

kgem_clear_dirty(&sna->kgem);

730

}

731

732

gen2_emit_target(sna,

733

op->dst.bo,

734

op->dst.width,

735

op->dst.height,

736

op->dst.format);

737

738

unwind = sna->kgem.nbatch;

739

BATCH(_3DSTATE_LOAD_STATE_IMMEDIATE_1 |

740

I1_LOAD_S(2) | I1_LOAD_S(3) | I1_LOAD_S(8) | 2);

741

BATCH((!op->src.is_solid + (op->mask.bo != NULL)) << 12);

742

BATCH(S3_CULLMODE_NONE | S3_VERTEXHAS_XY);

743

BATCH(gen2_get_blend_cntl(op->op,

744

op->has_component_alpha,

745

op->dst.format));

746

if (memcmp(sna->kgem.batch + sna->render_state.gen2.ls1 + 1,

747

sna->kgem.batch + unwind + 1,

748

3 * sizeof(uint32_t)) == 0)

749

sna->kgem.nbatch = unwind;

750

else

751

sna->render_state.gen2.ls1 = unwind;

752

753

gen2_disable_logic_op(sna);

754

755

gen2_get_blend_factors(op, op->op, &cblend, &ablend);

756

unwind = sna->kgem.nbatch;

757

BATCH(_3DSTATE_LOAD_STATE_IMMEDIATE_2 |

758

LOAD_TEXTURE_BLEND_STAGE(0) | 1);

759

BATCH(cblend);

760

BATCH(ablend);

761

if (memcmp(sna->kgem.batch + sna->render_state.gen2.ls2 + 1,

762

sna->kgem.batch + unwind + 1,

763

2 * sizeof(uint32_t)) == 0)

764

sna->kgem.nbatch = unwind;

765

else

766

sna->render_state.gen2.ls2 = unwind;

767

768

tex = texcoordfmt = 0;

769

if (!op->src.is_solid) {

770

if (op->src.is_affine)

771

texcoordfmt |= TEXCOORDFMT_2D << (2*tex);

772

else

773

texcoordfmt |= TEXCOORDFMT_3D << (2*tex);

774

gen2_emit_texture(sna, &op->src, tex++);

775

} else {

776

if (op->src.u.gen2.pixel != sna->render_state.gen2.diffuse) {

777

BATCH(_3DSTATE_DFLT_DIFFUSE_CMD);

778

BATCH(op->src.u.gen2.pixel);

779

sna->render_state.gen2.diffuse = op->src.u.gen2.pixel;

780

}

781

}

782

if (op->mask.bo) {

783

if (op->mask.is_affine)

784

texcoordfmt |= TEXCOORDFMT_2D << (2*tex);

785

else

786

texcoordfmt |= TEXCOORDFMT_3D << (2*tex);

787

gen2_emit_texture(sna, &op->mask, tex++);

788

} else if (op->mask.is_solid) {

789

if (op->mask.u.gen2.pixel != sna->render_state.gen2.diffuse) {

790

BATCH(_3DSTATE_DFLT_DIFFUSE_CMD);

791

BATCH(op->mask.u.gen2.pixel);

792

sna->render_state.gen2.diffuse = op->mask.u.gen2.pixel;

793

}

794

}

795

796

v = _3DSTATE_VERTEX_FORMAT_2_CMD | texcoordfmt;

797

if (sna->render_state.gen2.vft != v) {

798

BATCH(v);

799

sna->render_state.gen2.vft = v;

800

}

801

}

802

803

static inline void

804

gen2_emit_composite_dstcoord(struct sna *sna, int dstX, int dstY)

805

{

806

VERTEX(dstX);

807

VERTEX(dstY);

808

}

809

810

inline static void

811

gen2_emit_composite_linear(struct sna *sna,

812

const struct sna_composite_channel *channel,

813

int16_t x, int16_t y)

814

{

815

float v;

816

817

v = (x * channel->u.linear.dx +

818

y * channel->u.linear.dy +

819

channel->u.linear.offset);

820

DBG(("%s: (%d, %d) -> %f\n", __FUNCTION__, x, y, v));

821

VERTEX(v);

822

VERTEX(v);

823

}

824

825

static void

826

gen2_emit_composite_texcoord(struct sna *sna,

827

const struct sna_composite_channel *channel,

828

int16_t x, int16_t y)

829

{

830

float s = 0, t = 0, w = 1;

831

832

x += channel->offset[0];

833

y += channel->offset[1];

834

835

if (channel->is_affine) {

836

sna_get_transformed_coordinates(x, y,

837

channel->transform,

838

&s, &t);

839

VERTEX(s * channel->scale[0]);

840

VERTEX(t * channel->scale[1]);

841

} else {

842

sna_get_transformed_coordinates_3d(x, y,

843

channel->transform,

844

&s, &t, &w);

845

VERTEX(s * channel->scale[0]);

846

VERTEX(t * channel->scale[1]);

847

VERTEX(w);

848

}

849

}

850

851

static void

852

gen2_emit_composite_vertex(struct sna *sna,

853

const struct sna_composite_op *op,

854

int16_t srcX, int16_t srcY,

855

int16_t mskX, int16_t mskY,

856

int16_t dstX, int16_t dstY)

857

{

858

gen2_emit_composite_dstcoord(sna, dstX, dstY);

859

if (op->src.is_linear)

860

gen2_emit_composite_linear(sna, &op->src, srcX, srcY);

861

else if (!op->src.is_solid)

862

gen2_emit_composite_texcoord(sna, &op->src, srcX, srcY);

863

864

if (op->mask.is_linear)

865

gen2_emit_composite_linear(sna, &op->mask, mskX, mskY);

866

else if (op->mask.bo)

867

gen2_emit_composite_texcoord(sna, &op->mask, mskX, mskY);

868

}

869

870

fastcall static void

871

gen2_emit_composite_primitive(struct sna *sna,

872

const struct sna_composite_op *op,

873

const struct sna_composite_rectangles *r)

874

{

875

gen2_emit_composite_vertex(sna, op,

876

r->src.x + r->width,

877

r->src.y + r->height,

878

r->mask.x + r->width,

879

r->mask.y + r->height,

880

op->dst.x + r->dst.x + r->width,

881

op->dst.y + r->dst.y + r->height);

882

gen2_emit_composite_vertex(sna, op,

883

r->src.x,

884

r->src.y + r->height,

885

r->mask.x,

886

r->mask.y + r->height,

887

op->dst.x + r->dst.x,

888

op->dst.y + r->dst.y + r->height);

889

gen2_emit_composite_vertex(sna, op,

890

r->src.x,

891

r->src.y,

892

r->mask.x,

893

r->mask.y,

894

op->dst.x + r->dst.x,

895

op->dst.y + r->dst.y);

896

}

897

898

fastcall static void

899

gen2_emit_composite_primitive_constant(struct sna *sna,

900

const struct sna_composite_op *op,

901

const struct sna_composite_rectangles *r)

902

{

903

int16_t dst_x = r->dst.x + op->dst.x;

904

int16_t dst_y = r->dst.y + op->dst.y;

905

906

gen2_emit_composite_dstcoord(sna, dst_x + r->width, dst_y + r->height);

907

gen2_emit_composite_dstcoord(sna, dst_x, dst_y + r->height);

908

gen2_emit_composite_dstcoord(sna, dst_x, dst_y);

909

}

910

911

fastcall static void

912

gen2_emit_composite_primitive_linear(struct sna *sna,

913

const struct sna_composite_op *op,

914

const struct sna_composite_rectangles *r)

915

{

916

int16_t dst_x = r->dst.x + op->dst.x;

917

int16_t dst_y = r->dst.y + op->dst.y;

918

919

gen2_emit_composite_dstcoord(sna, dst_x + r->width, dst_y + r->height);

920

gen2_emit_composite_linear(sna, &op->src,

921

r->src.x + r->width, r->src.y + r->height);

922

923

gen2_emit_composite_dstcoord(sna, dst_x, dst_y + r->height);

924

gen2_emit_composite_linear(sna, &op->src,

925

r->src.x, r->src.y + r->height);

926

927

gen2_emit_composite_dstcoord(sna, dst_x, dst_y);

928

gen2_emit_composite_linear(sna, &op->src,

929

r->src.x, r->src.y);

930

}

931

932

fastcall static void

933

gen2_emit_composite_primitive_identity(struct sna *sna,

934

const struct sna_composite_op *op,

935

const struct sna_composite_rectangles *r)

936

{

937

float w = r->width;

938

float h = r->height;

939

float *v;

940

941

v = (float *)sna->kgem.batch + sna->kgem.nbatch;

942

sna->kgem.nbatch += 12;

943

944

v[8] = v[4] = r->dst.x + op->dst.x;

945

v[0] = v[4] + w;

946

947

v[9] = r->dst.y + op->dst.y;

948

v[5] = v[1] = v[9] + h;

949

950

v[10] = v[6] = (r->src.x + op->src.offset[0]) * op->src.scale[0];

951

v[2] = v[6] + w * op->src.scale[0];

952

953

v[11] = (r->src.y + op->src.offset[1]) * op->src.scale[1];

954

v[7] = v[3] = v[11] + h * op->src.scale[1];

955

}

956

957

fastcall static void

958

gen2_emit_composite_primitive_affine(struct sna *sna,

959

const struct sna_composite_op *op,

960

const struct sna_composite_rectangles *r)

961

{

962

PictTransform *transform = op->src.transform;

963

int src_x = r->src.x + (int)op->src.offset[0];

964

int src_y = r->src.y + (int)op->src.offset[1];

965

float *v;

966

967

v = (float *)sna->kgem.batch + sna->kgem.nbatch;

968

sna->kgem.nbatch += 12;

969

970

v[8] = v[4] = r->dst.x + op->dst.x;

971

v[0] = v[4] + r->width;

972

973

v[9] = r->dst.y + op->dst.y;

974

v[5] = v[1] = v[9] + r->height;

975

976

_sna_get_transformed_scaled(src_x + r->width, src_y + r->height,

977

transform, op->src.scale,

978

&v[2], &v[3]);

979

980

_sna_get_transformed_scaled(src_x, src_y + r->height,

981

transform, op->src.scale,

982

&v[6], &v[7]);

983

984

_sna_get_transformed_scaled(src_x, src_y,

985

transform, op->src.scale,

986

&v[10], &v[11]);

987

}

988

989

fastcall static void

990

gen2_emit_composite_primitive_constant_identity_mask(struct sna *sna,

991

const struct sna_composite_op *op,

992

const struct sna_composite_rectangles *r)

993

{

994

float w = r->width;

995

float h = r->height;

996

float *v;

997

998

v = (float *)sna->kgem.batch + sna->kgem.nbatch;

999

sna->kgem.nbatch += 12;

1000

1001

v[8] = v[4] = r->dst.x + op->dst.x;

1002

v[0] = v[4] + w;

1003

1004

v[9] = r->dst.y + op->dst.y;

1005

v[5] = v[1] = v[9] + h;

1006

1007

v[10] = v[6] = (r->mask.x + op->mask.offset[0]) * op->mask.scale[0];

1008

v[2] = v[6] + w * op->mask.scale[0];

1009

1010

v[11] = (r->mask.y + op->mask.offset[1]) * op->mask.scale[1];

1011

v[7] = v[3] = v[11] + h * op->mask.scale[1];

1012

}

1013

1014

#if defined(sse2) && !defined(__x86_64__)

1015

sse2 fastcall static void

1016

gen2_emit_composite_primitive_constant__sse2(struct sna *sna,

1017

const struct sna_composite_op *op,

1018

const struct sna_composite_rectangles *r)

1019

{

1020

int16_t dst_x = r->dst.x + op->dst.x;

1021

int16_t dst_y = r->dst.y + op->dst.y;

1022

1023

gen2_emit_composite_dstcoord(sna, dst_x + r->width, dst_y + r->height);

1024

gen2_emit_composite_dstcoord(sna, dst_x, dst_y + r->height);

1025

gen2_emit_composite_dstcoord(sna, dst_x, dst_y);

1026

}

1027

1028

sse2 fastcall static void

1029

gen2_emit_composite_primitive_linear__sse2(struct sna *sna,

1030

const struct sna_composite_op *op,

1031

const struct sna_composite_rectangles *r)

1032

{

1033

int16_t dst_x = r->dst.x + op->dst.x;

1034

int16_t dst_y = r->dst.y + op->dst.y;

1035

1036

gen2_emit_composite_dstcoord(sna, dst_x + r->width, dst_y + r->height);

1037

gen2_emit_composite_linear(sna, &op->src,

1038

r->src.x + r->width, r->src.y + r->height);

1039

1040

gen2_emit_composite_dstcoord(sna, dst_x, dst_y + r->height);

1041

gen2_emit_composite_linear(sna, &op->src,

1042

r->src.x, r->src.y + r->height);

1043

1044

gen2_emit_composite_dstcoord(sna, dst_x, dst_y);

1045

gen2_emit_composite_linear(sna, &op->src,

1046

r->src.x, r->src.y);

1047

}

1048

1049

sse2 fastcall static void

1050

gen2_emit_composite_primitive_identity__sse2(struct sna *sna,

1051

const struct sna_composite_op *op,

1052

const struct sna_composite_rectangles *r)

1053

{

1054

float w = r->width;

1055

float h = r->height;

1056

float *v;

1057

1058

v = (float *)sna->kgem.batch + sna->kgem.nbatch;

1059

sna->kgem.nbatch += 12;

1060

1061

v[8] = v[4] = r->dst.x + op->dst.x;

1062

v[0] = v[4] + w;

1063

1064

v[9] = r->dst.y + op->dst.y;

1065

v[5] = v[1] = v[9] + h;

1066

1067

v[10] = v[6] = (r->src.x + op->src.offset[0]) * op->src.scale[0];

1068

v[2] = v[6] + w * op->src.scale[0];

1069

1070

v[11] = (r->src.y + op->src.offset[1]) * op->src.scale[1];

1071

v[7] = v[3] = v[11] + h * op->src.scale[1];

1072

}

1073

1074

sse2 fastcall static void

1075

gen2_emit_composite_primitive_affine__sse2(struct sna *sna,

1076

const struct sna_composite_op *op,

1077

const struct sna_composite_rectangles *r)

1078

{

1079

PictTransform *transform = op->src.transform;

1080

int src_x = r->src.x + (int)op->src.offset[0];

1081

int src_y = r->src.y + (int)op->src.offset[1];

1082

float *v;

1083

1084

v = (float *)sna->kgem.batch + sna->kgem.nbatch;

1085

sna->kgem.nbatch += 12;

1086

1087

v[8] = v[4] = r->dst.x + op->dst.x;

1088

v[0] = v[4] + r->width;

1089

1090

v[9] = r->dst.y + op->dst.y;

1091

v[5] = v[1] = v[9] + r->height;

1092

1093

_sna_get_transformed_scaled(src_x + r->width, src_y + r->height,

1094

transform, op->src.scale,

1095

&v[2], &v[3]);

1096

1097

_sna_get_transformed_scaled(src_x, src_y + r->height,

1098

transform, op->src.scale,

1099

&v[6], &v[7]);

1100

1101

_sna_get_transformed_scaled(src_x, src_y,

1102

transform, op->src.scale,

1103

&v[10], &v[11]);

1104

}

1105

1106

sse2 fastcall static void

1107

gen2_emit_composite_primitive_constant_identity_mask__sse2(struct sna *sna,

1108

const struct sna_composite_op *op,

1109

const struct sna_composite_rectangles *r)

1110

{

1111

float w = r->width;

1112

float h = r->height;

1113

float *v;

1114

1115

v = (float *)sna->kgem.batch + sna->kgem.nbatch;

1116

sna->kgem.nbatch += 12;

1117

1118

v[8] = v[4] = r->dst.x + op->dst.x;

1119

v[0] = v[4] + w;

1120

1121

v[9] = r->dst.y + op->dst.y;

1122

v[5] = v[1] = v[9] + h;

1123

1124

v[10] = v[6] = (r->mask.x + op->mask.offset[0]) * op->mask.scale[0];

1125

v[2] = v[6] + w * op->mask.scale[0];

1126

1127

v[11] = (r->mask.y + op->mask.offset[1]) * op->mask.scale[1];

1128

v[7] = v[3] = v[11] + h * op->mask.scale[1];

1129

}

1130

#endif

1131

1132

static void gen2_magic_ca_pass(struct sna *sna,

1133

const struct sna_composite_op *op)

1134

{

1135

uint32_t ablend, cblend, *src, *dst;

1136

int n;

1137

1138

if (!op->need_magic_ca_pass)

1139

return;

1140

1141

DBG(("%s: batch=%x, vertex=%x\n", __FUNCTION__,

1142

sna->kgem.nbatch, sna->render.vertex_offset));

1143

1144

assert(op->mask.bo);

1145

assert(op->has_component_alpha);

1146

1147

BATCH(_3DSTATE_LOAD_STATE_IMMEDIATE_1 | I1_LOAD_S(8) | 0);

1148

BATCH(BLENDFACTOR_ONE << S8_SRC_BLEND_FACTOR_SHIFT |

1149

BLENDFACTOR_ONE << S8_DST_BLEND_FACTOR_SHIFT |

1150

S8_ENABLE_COLOR_BLEND | S8_BLENDFUNC_ADD |

1151

S8_ENABLE_COLOR_BUFFER_WRITE);

1152

sna->render_state.gen2.ls1 = 0;

1153

1154

gen2_get_blend_factors(op, PictOpAdd, &cblend, &ablend);

1155

BATCH(_3DSTATE_LOAD_STATE_IMMEDIATE_2 |

1156

LOAD_TEXTURE_BLEND_STAGE(0) | 1);

1157

BATCH(cblend);

1158

BATCH(ablend);

1159

sna->render_state.gen2.ls2 = 0;

1160

1161

src = sna->kgem.batch + sna->render.vertex_offset;

1162

dst = sna->kgem.batch + sna->kgem.nbatch;

1163

n = 1 + sna->render.vertex_index;

1164

sna->kgem.nbatch += n;

1165

assert(sna->kgem.nbatch <= KGEM_BATCH_SIZE(&sna->kgem));

1166

while (n--)

1167

*dst++ = *src++;

1168

}

1169

1170

static void gen2_vertex_flush(struct sna *sna,

1171

const struct sna_composite_op *op)

1172

{

1173

if (sna->render.vertex_index == 0)

1174

return;

1175

1176

sna->kgem.batch[sna->render.vertex_offset] |=

1177

sna->render.vertex_index - 1;

1178

1179

gen2_magic_ca_pass(sna, op);

1180

1181

sna->render.vertex_offset = 0;

1182

sna->render.vertex_index = 0;

1183

}

1184

1185

inline static int gen2_get_rectangles(struct sna *sna,

1186

const struct sna_composite_op *op,

1187

int want)

1188

{

1189

int rem = batch_space(sna), size, need;

1190

1191

DBG(("%s: want=%d, floats_per_vertex=%d, rem=%d\n",

1192

__FUNCTION__, want, op->floats_per_vertex, rem));

1193

1194

assert(op->floats_per_vertex);

1195

assert(op->floats_per_rect == 3 * op->floats_per_vertex);

1196

1197

need = 1;

1198

size = op->floats_per_rect;

1199

if (op->need_magic_ca_pass)

1200

need += 6 + size*sna->render.vertex_index, size *= 2;

1201

1202

DBG(("%s: want=%d, need=%d,size=%d, rem=%d\n",

1203

__FUNCTION__, want, need, size, rem));

1204

if (rem < need + size) {

1205

gen2_vertex_flush(sna, op);

1206

kgem_submit(&sna->kgem);

1207

_kgem_set_mode(&sna->kgem, KGEM_RENDER);

1208

return 0;

1209

}

1210

1211

rem -= need;

1212

if (sna->render.vertex_offset == 0) {

1213

if ((sna->kgem.batch[sna->kgem.nbatch-1] & ~0xffff) ==

1214

(PRIM3D_INLINE | PRIM3D_RECTLIST)) {

1215

uint32_t *b = &sna->kgem.batch[sna->kgem.nbatch-1];

1216

assert(*b & 0xffff);

1217

sna->render.vertex_index = 1 + (*b & 0xffff);

1218

*b = PRIM3D_INLINE | PRIM3D_RECTLIST;

1219

sna->render.vertex_offset = sna->kgem.nbatch - 1;

1220

assert(!op->need_magic_ca_pass);

1221

} else {

1222

sna->render.vertex_offset = sna->kgem.nbatch;

1223

BATCH(PRIM3D_INLINE | PRIM3D_RECTLIST);

1224

}

1225

1226

need = 0;

1227

} else

1228

need = sna->kgem.nbatch - sna->render.vertex_offset;

1229

1230

if (rem > MAX_INLINE - need)

1231

rem = MAX_INLINE -need;

1232

1233

if (want > 1 && want * size > rem)

1234

want = rem / size;

1235

1236

assert(want);

1237

sna->render.vertex_index += want*op->floats_per_rect;

1238

return want;

1239

}

1240

1241

fastcall static void

1242

gen2_render_composite_blt(struct sna *sna,

1243

const struct sna_composite_op *op,

1244

const struct sna_composite_rectangles *r)

1245

{

1246

if (!gen2_get_rectangles(sna, op, 1)) {

1247

gen2_emit_composite_state(sna, op);

1248

gen2_get_rectangles(sna, op, 1);

1249

}

1250

1251

op->prim_emit(sna, op, r);

1252

}

1253

1254

fastcall static void

1255

gen2_render_composite_box(struct sna *sna,

1256

const struct sna_composite_op *op,

1257

const BoxRec *box)

1258

{

1259

struct sna_composite_rectangles r;

1260

1261

if (!gen2_get_rectangles(sna, op, 1)) {

1262

gen2_emit_composite_state(sna, op);

1263

gen2_get_rectangles(sna, op, 1);

1264

}

1265

1266

DBG((" %s: (%d, %d) x (%d, %d)\n", __FUNCTION__,

1267

box->x1, box->y1,

1268

box->x2 - box->x1,

1269

box->y2 - box->y1));

1270

1271

r.dst.x = box->x1; r.dst.y = box->y1;

1272

r.width = box->x2 - box->x1;

1273

r.height = box->y2 - box->y1;

1274

r.src = r.mask = r.dst;

1275

1276

op->prim_emit(sna, op, &r);

1277

}

1278

1279

static void

1280

gen2_render_composite_boxes(struct sna *sna,

1281

const struct sna_composite_op *op,

1282

const BoxRec *box, int nbox)

1283

{

1284

do {

1285

int nbox_this_time;

1286

1287

nbox_this_time = gen2_get_rectangles(sna, op, nbox);

1288

if (nbox_this_time == 0) {

1289

gen2_emit_composite_state(sna, op);

1290

nbox_this_time = gen2_get_rectangles(sna, op, nbox);

1291

}

1292

nbox -= nbox_this_time;

1293

1294

do {

1295

struct sna_composite_rectangles r;

1296

1297

DBG((" %s: (%d, %d) x (%d, %d)\n", __FUNCTION__,

1298

box->x1, box->y1,

1299

box->x2 - box->x1,

1300

box->y2 - box->y1));

1301

1302

r.dst.x = box->x1; r.dst.y = box->y1;

1303

r.width = box->x2 - box->x1;

1304

r.height = box->y2 - box->y1;

1305

r.src = r.mask = r.dst;

1306

1307

op->prim_emit(sna, op, &r);

1308

box++;

1309

} while (--nbox_this_time);

1310

} while (nbox);

1311

}

1312

1313

static void gen2_render_composite_done(struct sna *sna,

1314

const struct sna_composite_op *op)

1315

{

1316

gen2_vertex_flush(sna, op);

1317

1318

if (op->mask.bo)

1319

kgem_bo_destroy(&sna->kgem, op->mask.bo);

1320

if (op->src.bo)

1321

kgem_bo_destroy(&sna->kgem, op->src.bo);

1322

sna_render_composite_redirect_done(sna, op);

1323

}

1324

1325

static bool

1326

gen2_composite_solid_init(struct sna *sna,

1327

struct sna_composite_channel *channel,

1328

uint32_t color)

1329

{

1330

channel->filter = PictFilterNearest;

1331

channel->repeat = RepeatNormal;

1332

channel->is_solid = true;

1333

channel->is_affine = true;

1334

channel->width = 1;

1335

channel->height = 1;

1336

channel->pict_format = PICT_a8r8g8b8;

1337

1338

channel->bo = NULL;

1339

channel->u.gen2.pixel = color;

1340

1341

channel->scale[0] = channel->scale[1] = 1;

1342

channel->offset[0] = channel->offset[1] = 0;

1343

return true;

1344

}

1345

1346

#define xFixedToDouble(f) pixman_fixed_to_double(f)

1347

1348

static bool

1349

gen2_composite_linear_init(struct sna *sna,

1350

PicturePtr picture,

1351

struct sna_composite_channel *channel,

1352

int x, int y,

1353

int w, int h,

1354

int dst_x, int dst_y)

1355

{

1356

PictLinearGradient *linear =

1357

(PictLinearGradient *)picture->pSourcePict;

1358

pixman_fixed_t tx, ty;

1359

float x0, y0, sf;

1360

float dx, dy;

1361

1362

DBG(("%s: p1=(%f, %f), p2=(%f, %f)\n",

1363

__FUNCTION__,

1364

xFixedToDouble(linear->p1.x), xFixedToDouble(linear->p1.y),

1365

xFixedToDouble(linear->p2.x), xFixedToDouble(linear->p2.y)));

1366

1367

if (linear->p2.x == linear->p1.x && linear->p2.y == linear->p1.y)

1368

return 0;

1369

1370

if (!sna_transform_is_affine(picture->transform)) {

1371

DBG(("%s: fallback due to projective transform\n",

1372

__FUNCTION__));

1373

return sna_render_picture_fixup(sna, picture, channel,

1374

x, y, w, h, dst_x, dst_y);

1375

}

1376

1377

channel->bo = sna_render_get_gradient(sna, (PictGradient *)linear);

1378

if (!channel->bo)

1379

return 0;

1380

1381

channel->filter = PictFilterNearest;

1382

channel->repeat = picture->repeat ? picture->repeatType : RepeatNone;

1383

channel->is_linear = true;

1384

channel->width = channel->bo->pitch / 4;

1385

channel->height = 1;

1386

channel->pict_format = PICT_a8r8g8b8;

1387

1388

channel->scale[0] = channel->scale[1] = 1;

1389

channel->offset[0] = channel->offset[1] = 0;

1390

1391

if (sna_transform_is_translation(picture->transform, &tx, &ty)) {

1392

dx = xFixedToDouble(linear->p2.x - linear->p1.x);

1393

dy = xFixedToDouble(linear->p2.y - linear->p1.y);

1394

1395

x0 = xFixedToDouble(linear->p1.x);

1396

y0 = xFixedToDouble(linear->p1.y);

1397

1398

if (tx | ty) {

1399

x0 -= pixman_fixed_to_double(tx);

1400

y0 -= pixman_fixed_to_double(ty);

1401

}

1402

} else {

1403

struct pixman_f_vector p1, p2;

1404

struct pixman_f_transform m, inv;

1405

1406

pixman_f_transform_from_pixman_transform(&m, picture->transform);

1407

DBG(("%s: transform = [%f %f %f, %f %f %f, %f %f %f]\n",

1408

__FUNCTION__,

1409

m.m[0][0], m.m[0][1], m.m[0][2],

1410

m.m[1][0], m.m[1][1], m.m[1][2],

1411

m.m[2][0], m.m[2][1], m.m[2][2]));

1412

if (!pixman_f_transform_invert(&inv, &m))

1413

return 0;

1414

1415

p1.v[0] = pixman_fixed_to_double(linear->p1.x);

1416

p1.v[1] = pixman_fixed_to_double(linear->p1.y);

1417

p1.v[2] = 1.;

1418

pixman_f_transform_point(&inv, &p1);

1419

1420

p2.v[0] = pixman_fixed_to_double(linear->p2.x);

1421

p2.v[1] = pixman_fixed_to_double(linear->p2.y);

1422

p2.v[2] = 1.;

1423

pixman_f_transform_point(&inv, &p2);

1424

1425

DBG(("%s: untransformed: p1=(%f, %f, %f), p2=(%f, %f, %f)\n",

1426

__FUNCTION__,

1427

p1.v[0], p1.v[1], p1.v[2],

1428

p2.v[0], p2.v[1], p2.v[2]));

1429

1430

dx = p2.v[0] - p1.v[0];

1431

dy = p2.v[1] - p1.v[1];

1432

1433

x0 = p1.v[0];

1434

y0 = p1.v[1];

1435

}

1436

1437

sf = dx*dx + dy*dy;

1438

dx /= sf;

1439

dy /= sf;

1440

1441

channel->u.linear.dx = dx;

1442

channel->u.linear.dy = dy;

1443

channel->u.linear.offset = -dx*(x0+dst_x-x) + -dy*(y0+dst_y-y);

1444

1445

DBG(("%s: dx=%f, dy=%f, offset=%f\n",

1446

__FUNCTION__, dx, dy, channel->u.linear.offset));

1447

1448

return channel->bo != NULL;

1449

}

1450

1451

static bool source_is_covered(PicturePtr picture,

1452

int x, int y,

1453

int width, int height)

1454

{

1455

int x1, y1, x2, y2;

1456

1457

if (picture->repeat && picture->repeatType != RepeatNone)

1458

return true;

1459

1460

if (picture->pDrawable == NULL)

1461

return false;

1462

1463

if (picture->transform) {

1464

pixman_box16_t sample;

1465

1466

sample.x1 = x;

1467

sample.y1 = y;

1468

sample.x2 = x + width;

1469

sample.y2 = y + height;

1470

1471

pixman_transform_bounds(picture->transform, &sample);

1472

1473

x1 = sample.x1;

1474

x2 = sample.x2;

1475

y1 = sample.y1;

1476

y2 = sample.y2;

1477

} else {

1478

x1 = x;

1479

y1 = y;

1480

x2 = x + width;

1481

y2 = y + height;

1482

}

1483

1484

return

1485

x1 >= 0 && y1 >= 0 &&

1486

x2 <= picture->pDrawable->width &&

1487

y2 <= picture->pDrawable->height;

1488

}

1489

1490

static bool

1491

gen2_check_card_format(struct sna *sna,

1492

PicturePtr picture,

1493

struct sna_composite_channel *channel,

1494

int x, int y, int w, int h,

1495

bool *fixup_alpha)

1496

{

1497

uint32_t format = picture->format;

1498

unsigned int i;

1499

1500

for (i = 0; i < ARRAY_SIZE(i8xx_tex_formats); i++) {

1501

if (i8xx_tex_formats[i].fmt == format)

1502

return true;

1503

}

1504

1505

for (i = 0; i < ARRAY_SIZE(i85x_tex_formats); i++) {

1506

if (i85x_tex_formats[i].fmt == format) {

1507

if (sna->kgem.gen >= 021)

1508

return true;

1509

1510

if (source_is_covered(picture, x, y, w,h)) {

1511

channel->is_opaque = true;

1512

return true;

1513

}

1514

1515

*fixup_alpha = true;

1516

return false;

1517

}

1518

}

1519

1520

*fixup_alpha = false;

1521

return false;

1522

}

1523

1524

static int

1525

gen2_composite_picture(struct sna *sna,

1526

PicturePtr picture,

1527

struct sna_composite_channel *channel,

1528

int x, int y,

1529

int w, int h,

1530

int dst_x, int dst_y,

1531

bool precise)

1532

{

1533

PixmapPtr pixmap;

1534

uint32_t color;

1535

int16_t dx, dy;

1536

bool fixup_alpha;

1537

1538

DBG(("%s: (%d, %d)x(%d, %d), dst=(%d, %d)\n",

1539

__FUNCTION__, x, y, w, h, dst_x, dst_y));

1540

1541

channel->is_solid = false;

1542

channel->is_linear = false;

1543

channel->is_opaque = false;

1544

channel->is_affine = true;

1545

channel->transform = NULL;

1546

channel->card_format = -1;

1547

1548

if (sna_picture_is_solid(picture, &color))

1549

return gen2_composite_solid_init(sna, channel, color);

1550

1551

if (!gen2_check_repeat(picture)) {

1552

DBG(("%s -- fallback, unhandled repeat %d\n",

1553

__FUNCTION__, picture->repeat));

1554

return sna_render_picture_fixup(sna, picture, channel,

1555

x, y, w, h, dst_x, dst_y);

1556

}

1557

1558

if (!gen2_check_filter(picture)) {

1559

DBG(("%s -- fallback, unhandled filter %d\n",

1560

__FUNCTION__, picture->filter));

1561

return sna_render_picture_fixup(sna, picture, channel,

1562

x, y, w, h, dst_x, dst_y);

1563

}

1564

1565

if (picture->pDrawable == NULL) {

1566

int ret;

1567

1568

if (picture->pSourcePict->type == SourcePictTypeLinear)

1569

return gen2_composite_linear_init(sna, picture, channel,

1570

x, y,

1571

w, h,

1572

dst_x, dst_y);

1573

1574

DBG(("%s -- fallback, unhandled source %d\n",

1575

__FUNCTION__, picture->pSourcePict->type));

1576

ret = -1;

1577

if (!precise)

1578

ret = sna_render_picture_approximate_gradient(sna, picture, channel,

1579

x, y, w, h, dst_x, dst_y);

1580

if (ret == -1)

1581

ret = sna_render_picture_fixup(sna, picture, channel,

1582

x, y, w, h, dst_x, dst_y);

1583

return ret;

1584

}

1585

1586

if (picture->alphaMap) {

1587

DBG(("%s -- fallback, alphamap\n", __FUNCTION__));

1588

return sna_render_picture_fixup(sna, picture, channel,

1589

x, y, w, h, dst_x, dst_y);

1590

}

1591

1592

channel->repeat = picture->repeat ? picture->repeatType : RepeatNone;

1593

channel->filter = picture->filter;

1594

1595

pixmap = get_drawable_pixmap(picture->pDrawable);

1596

get_drawable_deltas(picture->pDrawable, pixmap, &dx, &dy);

1597

1598

x += dx + picture->pDrawable->x;

1599

y += dy + picture->pDrawable->y;

1600

1601

channel->is_affine = sna_transform_is_affine(picture->transform);

1602

if (sna_transform_is_imprecise_integer_translation(picture->transform, picture->filter, precise, &dx, &dy)) {

1603

DBG(("%s: integer translation (%d, %d), removing\n",

1604

__FUNCTION__, dx, dy));

1605

x += dx;

1606

y += dy;

1607

channel->transform = NULL;

1608

channel->filter = PictFilterNearest;

1609

1610

if (channel->repeat &&

1611

(x >= 0 &&

1612

y >= 0 &&

1613

x + w <= pixmap->drawable.width &&

1614

y + h <= pixmap->drawable.height)) {

1615

struct sna_pixmap *priv = sna_pixmap(pixmap);

1616

if (priv && priv->clear) {

1617

DBG(("%s: converting large pixmap source into solid [%08x]\n", __FUNCTION__, priv->clear_color));

1618

return gen2_composite_solid_init(sna, channel, solid_color(picture->format, priv->clear_color));

1619

}

1620

}

1621

} else

1622

channel->transform = picture->transform;

1623

1624

if (!gen2_check_card_format(sna, picture, channel, x, y, w ,h, &fixup_alpha))

1625

return sna_render_picture_convert(sna, picture, channel, pixmap,

1626

x, y, w, h, dst_x, dst_y, fixup_alpha);

1627

1628

channel->pict_format = picture->format;

1629

if (too_large(pixmap->drawable.width, pixmap->drawable.height))

1630

return sna_render_picture_extract(sna, picture, channel,

1631

x, y, w, h, dst_x, dst_y);

1632

1633

return sna_render_pixmap_bo(sna, channel, pixmap,

1634

x, y, w, h, dst_x, dst_y);

1635

}

1636

1637

static bool

1638

gen2_composite_set_target(struct sna *sna,

1639

struct sna_composite_op *op,

1640

PicturePtr dst,

1641

int x, int y, int w, int h,

1642

bool partial)

1643

{

1644

BoxRec box;

1645

unsigned hint;

1646

1647

op->dst.pixmap = get_drawable_pixmap(dst->pDrawable);

1648

op->dst.format = dst->format;

1649

op->dst.width = op->dst.pixmap->drawable.width;

1650

op->dst.height = op->dst.pixmap->drawable.height;

1651

1652

if (w && h) {

1653

box.x1 = x;

1654

box.y1 = y;

1655

box.x2 = x + w;

1656

box.y2 = y + h;

1657

} else

1658

sna_render_picture_extents(dst, &box);

1659

1660

hint = PREFER_GPU | RENDER_GPU;

1661

if (!need_tiling(sna, op->dst.width, op->dst.height))

1662

hint |= FORCE_GPU;

1663

if (!partial) {

1664

hint |= IGNORE_DAMAGE;

1665

if (w == op->dst.width && h == op->dst.height)

1666

hint |= REPLACES;

1667

}

1668

1669

op->dst.bo = sna_drawable_use_bo(dst->pDrawable, hint, &box, &op->damage);

1670

if (op->dst.bo == NULL)

1671

return false;

1672

1673

if (hint & REPLACES) {

1674

struct sna_pixmap *priv = sna_pixmap(op->dst.pixmap);

1675

kgem_bo_pair_undo(&sna->kgem, priv->gpu_bo, priv->cpu_bo);

1676

}

1677

1678

assert((op->dst.bo->pitch & 7) == 0);

1679

1680

get_drawable_deltas(dst->pDrawable, op->dst.pixmap,

1681

&op->dst.x, &op->dst.y);

1682

1683

DBG(("%s: pixmap=%ld, format=%08x, size=%dx%d, pitch=%d, delta=(%d,%d),damage=%p\n",

1684

__FUNCTION__,

1685

op->dst.pixmap->drawable.serialNumber, (int)op->dst.format,

1686

op->dst.width, op->dst.height,

1687

op->dst.bo->pitch,

1688

op->dst.x, op->dst.y,

1689

op->damage ? *op->damage : (void *)-1));

1690

1691

assert(op->dst.bo->proxy == NULL);

1692

1693

if (((too_large(op->dst.width, op->dst.height) ||

1694

op->dst.bo->pitch > MAX_3D_PITCH)) &&

1695

!sna_render_composite_redirect(sna, op, x, y, w, h, partial))

1696

return false;

1697

1698

return true;

1699

}

1700

1701

static bool

1702

is_unhandled_gradient(PicturePtr picture, bool precise)

1703

{

1704

if (picture->pDrawable)

1705

return false;

1706

1707

switch (picture->pSourcePict->type) {

1708

case SourcePictTypeSolidFill:

1709

case SourcePictTypeLinear:

1710

return false;

1711

default:

1712

return precise;

1713

}

1714

}

1715

1716

static bool

1717

has_alphamap(PicturePtr p)

1718

{

1719

return p->alphaMap != NULL;

1720

}

1721

1722

static bool

1723

need_upload(PicturePtr p)

1724

{

1725

return p->pDrawable && unattached(p->pDrawable) && untransformed(p);

1726

}

1727

1728

static bool

1729

source_is_busy(PixmapPtr pixmap)

1730

{

1731

struct sna_pixmap *priv = sna_pixmap(pixmap);

1732

if (priv == NULL)

1733

return false;

1734

1735

if (priv->clear)

1736

return false;

1737

1738

if (priv->gpu_bo && kgem_bo_is_busy(priv->gpu_bo))

1739

return true;

1740

1741

if (priv->cpu_bo && kgem_bo_is_busy(priv->cpu_bo))

1742

return true;

1743

1744

return priv->gpu_damage && !priv->cpu_damage;

1745

}

1746

1747

static bool

1748

source_fallback(PicturePtr p, PixmapPtr pixmap, bool precise)

1749

{

1750

if (sna_picture_is_solid(p, NULL))

1751

return false;

1752

1753

if (is_unhandled_gradient(p, precise) || !gen2_check_repeat(p))

1754

return true;

1755

1756

if (pixmap && source_is_busy(pixmap))

1757

return false;

1758

1759

return has_alphamap(p) || !gen2_check_filter(p) || need_upload(p);

1760

}

1761

1762

static bool

1763

gen2_composite_fallback(struct sna *sna,

1764

PicturePtr src,

1765

PicturePtr mask,

1766

PicturePtr dst)

1767

{

1768

PixmapPtr src_pixmap;

1769

PixmapPtr mask_pixmap;

1770

PixmapPtr dst_pixmap;

1771

bool src_fallback, mask_fallback;

1772

1773

if (!gen2_check_dst_format(dst->format)) {

1774

DBG(("%s: unknown destination format: %d\n",

1775

__FUNCTION__, dst->format));

1776

return true;

1777

}

1778

1779

dst_pixmap = get_drawable_pixmap(dst->pDrawable);

1780

1781

src_pixmap = src->pDrawable ? get_drawable_pixmap(src->pDrawable) : NULL;

1782

src_fallback = source_fallback(src, src_pixmap,

1783

dst->polyMode == PolyModePrecise);

1784

1785

if (mask) {

1786

mask_pixmap = mask->pDrawable ? get_drawable_pixmap(mask->pDrawable) : NULL;

1787

mask_fallback = source_fallback(mask, mask_pixmap,

1788

dst->polyMode == PolyModePrecise);

1789

} else {

1790

mask_pixmap = NULL;

1791

mask_fallback = NULL;

1792

}

1793

1794

/* If we are using the destination as a source and need to

1795

* readback in order to upload the source, do it all

1796

* on the cpu.

1797

1798

if (src_pixmap == dst_pixmap && src_fallback) {

1799

DBG(("%s: src is dst and will fallback\n",__FUNCTION__));

1800

return true;

1801

}

1802

if (mask_pixmap == dst_pixmap && mask_fallback) {

1803

DBG(("%s: mask is dst and will fallback\n",__FUNCTION__));

1804

return true;

1805

}

1806

1807

/* If anything is on the GPU, push everything out to the GPU */

1808

if (dst_use_gpu(dst_pixmap)) {

1809

DBG(("%s: dst is already on the GPU, try to use GPU\n",

1810

__FUNCTION__));

1811

return false;

1812

}

1813

1814

if (src_pixmap && !src_fallback) {

1815

DBG(("%s: src is already on the GPU, try to use GPU\n",

1816

__FUNCTION__));

1817

return false;

1818

}

1819

if (mask_pixmap && !mask_fallback) {

1820

DBG(("%s: mask is already on the GPU, try to use GPU\n",

1821

__FUNCTION__));

1822

return false;

1823

}

1824

1825

/* However if the dst is not on the GPU and we need to

1826

* render one of the sources using the CPU, we may

1827

* as well do the entire operation in place onthe CPU.

1828

1829

if (src_fallback) {

1830

DBG(("%s: dst is on the CPU and src will fallback\n",

1831

__FUNCTION__));

1832

return true;

1833

}

1834

1835

if (mask && mask_fallback) {

1836

DBG(("%s: dst is on the CPU and mask will fallback\n",

1837

__FUNCTION__));

1838

return true;

1839

}

1840

1841

if (too_large(dst_pixmap->drawable.width,

1842

dst_pixmap->drawable.height) &&

1843

dst_is_cpu(dst_pixmap)) {

1844

DBG(("%s: dst is on the CPU and too large\n", __FUNCTION__));

1845

return true;

1846

}

1847

1848

DBG(("%s: dst is not on the GPU and the operation should not fallback\n",

1849

__FUNCTION__));

1850

return dst_use_cpu(dst_pixmap);

1851

}

1852

1853

static int

1854

reuse_source(struct sna *sna,

1855

PicturePtr src, struct sna_composite_channel *sc, int src_x, int src_y,

1856

PicturePtr mask, struct sna_composite_channel *mc, int msk_x, int msk_y)

1857

{

1858

uint32_t color;

1859

1860

if (src_x != msk_x || src_y != msk_y)

1861

return false;

1862

1863

if (sna_picture_is_solid(mask, &color))

1864

return gen2_composite_solid_init(sna, mc, color);

1865

1866

if (sc->is_solid)

1867

return false;

1868

1869

if (src == mask) {

1870

DBG(("%s: mask is source\n", __FUNCTION__));

1871

*mc = *sc;

1872

mc->bo = kgem_bo_reference(mc->bo);

1873

return true;

1874

}

1875

1876

if (src->pDrawable == NULL || mask->pDrawable != src->pDrawable)

1877

return false;

1878

1879

DBG(("%s: mask reuses source drawable\n", __FUNCTION__));

1880

1881

if (!sna_transform_equal(src->transform, mask->transform))

1882

return false;

1883

1884

if (!sna_picture_alphamap_equal(src, mask))

1885

return false;

1886

1887

if (!gen2_check_repeat(mask))

1888

return false;

1889

1890

if (!gen2_check_filter(mask))

1891

return false;

1892

1893

if (!gen2_check_format(sna, mask))

1894

return false;

1895

1896

DBG(("%s: reusing source channel for mask with a twist\n",

1897

__FUNCTION__));

1898

1899

*mc = *sc;

1900

mc->repeat = mask->repeat ? mask->repeatType : RepeatNone;

1901

mc->filter = mask->filter;

1902

mc->pict_format = mask->format;

1903

mc->bo = kgem_bo_reference(mc->bo);

1904

return true;

1905

}

1906

1907

static bool

1908

gen2_render_composite(struct sna *sna,

1909

uint8_t op,

1910

PicturePtr src,

1911

PicturePtr mask,

1912

PicturePtr dst,

1913

int16_t src_x, int16_t src_y,

1914

int16_t mask_x, int16_t mask_y,

1915

int16_t dst_x, int16_t dst_y,

1916

int16_t width, int16_t height,

1917

unsigned flags,

1918

struct sna_composite_op *tmp)

1919

{

1920

DBG(("%s()\n", __FUNCTION__));

1921

1922

if (op >= ARRAY_SIZE(gen2_blend_op)) {

1923

DBG(("%s: fallback due to unhandled blend op: %d\n",

1924

__FUNCTION__, op));

1925

return false;

1926

}

1927

1928

if (mask == NULL &&

1929

sna_blt_composite(sna, op, src, dst,

1930

src_x, src_y,

1931

dst_x, dst_y,

1932

width, height,

1933

flags, tmp))

1934

return true;

1935

1936

if (gen2_composite_fallback(sna, src, mask, dst))

1937

goto fallback;

1938

1939

if (need_tiling(sna, width, height))

1940

return sna_tiling_composite(op, src, mask, dst,

1941

src_x, src_y,

1942

mask_x, mask_y,

1943

dst_x, dst_y,

1944

width, height,

1945

tmp);

1946

1947

tmp->op = op;

1948

sna_render_composite_redirect_init(tmp);

1949

1950

if (!gen2_composite_set_target(sna, tmp, dst,

1951

dst_x, dst_y, width, height,

1952

flags & COMPOSITE_PARTIAL || op > PictOpSrc)) {

1953

DBG(("%s: unable to set render target\n",

1954

__FUNCTION__));

1955

goto fallback;

1956

}

1957

1958

switch (gen2_composite_picture(sna, src, &tmp->src,

1959

src_x, src_y,

1960

width, height,

1961

dst_x, dst_y,

1962

dst->polyMode == PolyModePrecise)) {

1963

case -1:

1964

DBG(("%s: fallback -- unable to prepare source\n",

1965

__FUNCTION__));

1966

goto cleanup_dst;

1967

case 0:

1968

gen2_composite_solid_init(sna, &tmp->src, 0);

1969

break;

1970

case 1:

1971

if (mask == NULL && tmp->src.bo &&

1972

sna_blt_composite__convert(sna,

1973

dst_x, dst_y, width, height,

1974

tmp))

1975

return true;

1976

break;

1977

}

1978

1979

if (mask) {

1980

if (!reuse_source(sna,

1981

src, &tmp->src, src_x, src_y,

1982

mask, &tmp->mask, mask_x, mask_y)) {

1983

switch (gen2_composite_picture(sna, mask, &tmp->mask,

1984

mask_x, mask_y,

1985

width, height,

1986

dst_x, dst_y,

1987

dst->polyMode == PolyModePrecise)) {

1988

case -1:

1989

DBG(("%s: fallback -- unable to prepare mask\n",

1990

__FUNCTION__));

1991

goto cleanup_src;

1992

case 0:

1993

gen2_composite_solid_init(sna, &tmp->mask, 0);

1994

case 1:

1995

break;

1996

}

1997

}

1998

1999

if (mask->componentAlpha && PICT_FORMAT_RGB(mask->format)) {

2000

/* Check if it's component alpha that relies on a source alpha

2001

* and on the source value. We can only get one of those

2002

* into the single source value that we get to blend with.

2003

2004

tmp->has_component_alpha = true;

2005

if (gen2_blend_op[op].src_alpha &&

2006

(gen2_blend_op[op].src_blend != BLENDFACTOR_ZERO)) {

2007

if (op != PictOpOver) {

2008

DBG(("%s: fallback -- unsupported CA blend (src_blend=%d)\n",

2009

__FUNCTION__,

2010

gen2_blend_op[op].src_blend));

2011

goto cleanup_src;

2012

}

2013

2014

tmp->need_magic_ca_pass = true;

2015

tmp->op = PictOpOutReverse;

2016

}

2017

}

2018

2019

/* convert solid to a texture (pure convenience) */

2020

if (tmp->mask.is_solid && tmp->src.is_solid) {

2021

assert(tmp->mask.is_affine);

2022

tmp->mask.bo = sna_render_get_solid(sna, tmp->mask.u.gen2.pixel);

2023

if (!tmp->mask.bo)

2024

goto cleanup_src;

2025

}

2026

}

2027

2028

tmp->floats_per_vertex = 2;

2029

if (!tmp->src.is_solid)

2030

tmp->floats_per_vertex += tmp->src.is_affine ? 2 : 3;

2031

if (tmp->mask.bo)

2032

tmp->floats_per_vertex += tmp->mask.is_affine ? 2 : 3;

2033

tmp->floats_per_rect = 3*tmp->floats_per_vertex;

2034

2035

tmp->prim_emit = gen2_emit_composite_primitive;

2036

if (tmp->mask.bo) {

2037

if (tmp->mask.transform == NULL) {

2038

if (tmp->src.is_solid) {

2039

assert(tmp->floats_per_rect == 12);

2040

#if defined(sse2) && !defined(__x86_64__)

2041

if (sna->cpu_features & SSE2) {

2042

tmp->prim_emit = gen2_emit_composite_primitive_constant_identity_mask__sse2;

2043

} else

2044

#endif

2045

{

2046

tmp->prim_emit = gen2_emit_composite_primitive_constant_identity_mask;

2047

}

2048

}

2049

}

2050

} else {

2051

if (tmp->src.is_solid) {

2052

assert(tmp->floats_per_rect == 6);

2053

#if defined(sse2) && !defined(__x86_64__)

2054

if (sna->cpu_features & SSE2) {

2055

tmp->prim_emit = gen2_emit_composite_primitive_constant__sse2;

2056

} else

2057

#endif

2058

{

2059

tmp->prim_emit = gen2_emit_composite_primitive_constant;

2060

}

2061

} else if (tmp->src.is_linear) {

2062

assert(tmp->floats_per_rect == 12);

2063

#if defined(sse2) && !defined(__x86_64__)

2064

if (sna->cpu_features & SSE2) {

2065

tmp->prim_emit = gen2_emit_composite_primitive_linear__sse2;

2066

} else

2067

#endif

2068

{

2069

tmp->prim_emit = gen2_emit_composite_primitive_linear;

2070

}

2071

} else if (tmp->src.transform == NULL) {

2072

assert(tmp->floats_per_rect == 12);

2073

#if defined(sse2) && !defined(__x86_64__)

2074

if (sna->cpu_features & SSE2) {

2075

tmp->prim_emit = gen2_emit_composite_primitive_identity__sse2;

2076

} else

2077

#endif

2078

{

2079

tmp->prim_emit = gen2_emit_composite_primitive_identity;

2080

}

2081

} else if (tmp->src.is_affine) {

2082

assert(tmp->floats_per_rect == 12);

2083

tmp->src.scale[0] /= tmp->src.transform->matrix[2][2];

2084

tmp->src.scale[1] /= tmp->src.transform->matrix[2][2];

2085

#if defined(sse2) && !defined(__x86_64__)

2086

if (sna->cpu_features & SSE2) {

2087

tmp->prim_emit = gen2_emit_composite_primitive_affine__sse2;

2088

} else

2089

#endif

2090

{

2091

tmp->prim_emit = gen2_emit_composite_primitive_affine;

2092

}

2093

}

2094

}

2095

2096

tmp->blt = gen2_render_composite_blt;

2097

tmp->box = gen2_render_composite_box;

2098

tmp->boxes = gen2_render_composite_boxes;

2099

tmp->done = gen2_render_composite_done;

2100

2101

if (!kgem_check_bo(&sna->kgem,

2102

tmp->dst.bo, tmp->src.bo, tmp->mask.bo,

2103

NULL)) {

2104

kgem_submit(&sna->kgem);

2105

if (!kgem_check_bo(&sna->kgem,

2106

tmp->dst.bo, tmp->src.bo, tmp->mask.bo,

2107

NULL)) {

2108

DBG(("%s: fallback, operation does not fit into GTT\n",

2109

__FUNCTION__));

2110

goto cleanup_mask;

2111

}

2112

}

2113

2114

gen2_emit_composite_state(sna, tmp);

2115

return true;

2116

2117

cleanup_mask:

2118

if (tmp->mask.bo) {

2119

kgem_bo_destroy(&sna->kgem, tmp->mask.bo);

2120

tmp->mask.bo = NULL;

2121

}

2122

cleanup_src:

2123

if (tmp->src.bo) {

2124

kgem_bo_destroy(&sna->kgem, tmp->src.bo);

2125

tmp->src.bo = NULL;

2126

}

2127

cleanup_dst:

2128

if (tmp->redirect.real_bo) {

2129

kgem_bo_destroy(&sna->kgem, tmp->dst.bo);

2130

tmp->redirect.real_bo = NULL;

2131

}

2132

fallback:

2133

return (mask == NULL &&

2134

sna_blt_composite(sna, op, src, dst,

2135

src_x, src_y,

2136

dst_x, dst_y,

2137

width, height,

2138

flags | COMPOSITE_FALLBACK, tmp));

2139

}

2140

2141

fastcall static void

2142

gen2_emit_composite_spans_primitive_constant(struct sna *sna,

2143

const struct sna_composite_spans_op *op,

2144

const BoxRec *box,

2145

float opacity)

2146

{

2147

float *v = (float *)sna->kgem.batch + sna->kgem.nbatch;

2148

uint32_t alpha = (uint8_t)(255 * opacity) << 24;

2149

sna->kgem.nbatch += 9;

2150

2151

v[0] = op->base.dst.x + box->x2;

2152

v[1] = op->base.dst.y + box->y2;

2153

*((uint32_t *)v + 2) = alpha;

2154

2155

v[3] = op->base.dst.x + box->x1;

2156

v[4] = v[1];

2157

*((uint32_t *)v + 5) = alpha;

2158

2159

v[6] = v[3];

2160

v[7] = op->base.dst.y + box->y1;

2161

*((uint32_t *)v + 8) = alpha;

2162

}

2163

2164

fastcall static void

2165

gen2_emit_composite_spans_primitive_linear(struct sna *sna,

2166

const struct sna_composite_spans_op *op,

2167

const BoxRec *box,

2168

float opacity)

2169

{

2170

union {

2171

float f;

2172

uint32_t u;

2173

} alpha;

2174

2175

alpha.u = (uint8_t)(255 * opacity) << 24;

2176

2177

gen2_emit_composite_dstcoord(sna,

2178

op->base.dst.x + box->x2,

2179

op->base.dst.y + box->y2);

2180

VERTEX(alpha.f);

2181

gen2_emit_composite_linear(sna, &op->base.src, box->x2, box->y2);

2182

2183

gen2_emit_composite_dstcoord(sna,

2184

op->base.dst.x + box->x1,

2185

op->base.dst.y + box->y2);

2186

VERTEX(alpha.f);

2187

gen2_emit_composite_linear(sna, &op->base.src, box->x1, box->y2);

2188

2189

gen2_emit_composite_dstcoord(sna,

2190

op->base.dst.x + box->x1,

2191

op->base.dst.y + box->y1);

2192

VERTEX(alpha.f);

2193

gen2_emit_composite_linear(sna, &op->base.src, box->x1, box->y1);

2194

}

2195

2196

fastcall static void

2197

gen2_emit_composite_spans_primitive_identity_source(struct sna *sna,

2198

const struct sna_composite_spans_op *op,

2199

const BoxRec *box,

2200

float opacity)

2201

{

2202

float *v = (float *)sna->kgem.batch + sna->kgem.nbatch;

2203

uint32_t alpha = (uint8_t)(255 * opacity) << 24;

2204

sna->kgem.nbatch += 15;

2205

2206

v[0] = op->base.dst.x + box->x2;

2207

v[1] = op->base.dst.y + box->y2;

2208

*((uint32_t *)v + 2) = alpha;

2209

v[3] = (op->base.src.offset[0] + box->x2) * op->base.src.scale[0];

2210

v[4] = (op->base.src.offset[1] + box->y2) * op->base.src.scale[1];

2211

2212

v[5] = op->base.dst.x + box->x1;

2213

v[6] = v[1];

2214

*((uint32_t *)v + 7) = alpha;

2215

v[8] = (op->base.src.offset[0] + box->x1) * op->base.src.scale[0];

2216

v[9] = v[4];

2217

2218

v[10] = v[5];

2219

v[11] = op->base.dst.y + box->y1;

2220

*((uint32_t *)v + 12) = alpha;

2221

v[13] = v[8];

2222

v[14] = (op->base.src.offset[1] + box->y1) * op->base.src.scale[1];

2223

}

2224

2225

fastcall static void

2226

gen2_emit_composite_spans_primitive_affine_source(struct sna *sna,

2227

const struct sna_composite_spans_op *op,

2228

const BoxRec *box,

2229

float opacity)

2230

{

2231

PictTransform *transform = op->base.src.transform;

2232

uint32_t alpha = (uint8_t)(255 * opacity) << 24;

2233

float *v;

2234

2235

v = (float *)sna->kgem.batch + sna->kgem.nbatch;

2236

sna->kgem.nbatch += 15;

2237

2238

v[0] = op->base.dst.x + box->x2;

2239

v[6] = v[1] = op->base.dst.y + box->y2;

2240

v[10] = v[5] = op->base.dst.x + box->x1;

2241

v[11] = op->base.dst.y + box->y1;

2242

*((uint32_t *)v + 2) = alpha;

2243

*((uint32_t *)v + 7) = alpha;

2244

*((uint32_t *)v + 12) = alpha;

2245

2246

_sna_get_transformed_scaled((int)op->base.src.offset[0] + box->x2,

2247

(int)op->base.src.offset[1] + box->y2,

2248

transform, op->base.src.scale,

2249

&v[3], &v[4]);

2250

2251

_sna_get_transformed_scaled((int)op->base.src.offset[0] + box->x1,

2252

(int)op->base.src.offset[1] + box->y2,

2253

transform, op->base.src.scale,

2254

&v[8], &v[9]);

2255

2256

_sna_get_transformed_scaled((int)op->base.src.offset[0] + box->x1,

2257

(int)op->base.src.offset[1] + box->y1,

2258

transform, op->base.src.scale,

2259

&v[13], &v[14]);

2260

}

2261

2262

#if defined(sse2) && !defined(__x86_64__)

2263

sse2 fastcall static void

2264

gen2_emit_composite_spans_primitive_constant__sse2(struct sna *sna,

2265

const struct sna_composite_spans_op *op,

2266

const BoxRec *box,

2267

float opacity)

2268

{

2269

float *v = (float *)sna->kgem.batch + sna->kgem.nbatch;

2270

uint32_t alpha = (uint8_t)(255 * opacity) << 24;

2271

sna->kgem.nbatch += 9;

2272

2273

v[0] = op->base.dst.x + box->x2;

2274

v[1] = op->base.dst.y + box->y2;

2275

*((uint32_t *)v + 2) = alpha;

2276

2277

v[3] = op->base.dst.x + box->x1;

2278

v[4] = v[1];

2279

*((uint32_t *)v + 5) = alpha;

2280

2281

v[6] = v[3];

2282

v[7] = op->base.dst.y + box->y1;

2283

*((uint32_t *)v + 8) = alpha;

2284

}

2285

2286

sse2 fastcall static void

2287

gen2_emit_composite_spans_primitive_linear__sse2(struct sna *sna,

2288

const struct sna_composite_spans_op *op,

2289

const BoxRec *box,

2290

float opacity)

2291

{

2292

union {

2293

float f;

2294

uint32_t u;

2295

} alpha;

2296

2297

alpha.u = (uint8_t)(255 * opacity) << 24;

2298

2299

gen2_emit_composite_dstcoord(sna,

2300

op->base.dst.x + box->x2,

2301

op->base.dst.y + box->y2);

2302

VERTEX(alpha.f);

2303

gen2_emit_composite_linear(sna, &op->base.src, box->x2, box->y2);

2304

2305

gen2_emit_composite_dstcoord(sna,

2306

op->base.dst.x + box->x1,

2307

op->base.dst.y + box->y2);

2308

VERTEX(alpha.f);

2309

gen2_emit_composite_linear(sna, &op->base.src, box->x1, box->y2);

2310

2311

gen2_emit_composite_dstcoord(sna,

2312

op->base.dst.x + box->x1,

2313

op->base.dst.y + box->y1);

2314

VERTEX(alpha.f);

2315

gen2_emit_composite_linear(sna, &op->base.src, box->x1, box->y1);

2316

}

2317

2318

sse2 fastcall static void

2319

gen2_emit_composite_spans_primitive_identity_source__sse2(struct sna *sna,

2320

const struct sna_composite_spans_op *op,

2321

const BoxRec *box,

2322

float opacity)

2323

{

2324

float *v = (float *)sna->kgem.batch + sna->kgem.nbatch;

2325

uint32_t alpha = (uint8_t)(255 * opacity) << 24;

2326

sna->kgem.nbatch += 15;

2327

2328

v[0] = op->base.dst.x + box->x2;

2329

v[1] = op->base.dst.y + box->y2;

2330

*((uint32_t *)v + 2) = alpha;

2331

v[3] = (op->base.src.offset[0] + box->x2) * op->base.src.scale[0];

2332

v[4] = (op->base.src.offset[1] + box->y2) * op->base.src.scale[1];

2333

2334

v[5] = op->base.dst.x + box->x1;

2335

v[6] = v[1];

2336

*((uint32_t *)v + 7) = alpha;

2337

v[8] = (op->base.src.offset[0] + box->x1) * op->base.src.scale[0];

2338

v[9] = v[4];

2339

2340

v[10] = v[5];

2341

v[11] = op->base.dst.y + box->y1;

2342

*((uint32_t *)v + 12) = alpha;

2343

v[13] = v[8];

2344

v[14] = (op->base.src.offset[1] + box->y1) * op->base.src.scale[1];

2345

}

2346

2347

sse2 fastcall static void

2348

gen2_emit_composite_spans_primitive_affine_source__sse2(struct sna *sna,

2349

const struct sna_composite_spans_op *op,

2350

const BoxRec *box,

2351

float opacity)

2352

{

2353

PictTransform *transform = op->base.src.transform;

2354

uint32_t alpha = (uint8_t)(255 * opacity) << 24;

2355

float *v;

2356

2357

v = (float *)sna->kgem.batch + sna->kgem.nbatch;

2358

sna->kgem.nbatch += 15;

2359

2360

v[0] = op->base.dst.x + box->x2;

2361

v[6] = v[1] = op->base.dst.y + box->y2;

2362

v[10] = v[5] = op->base.dst.x + box->x1;

2363

v[11] = op->base.dst.y + box->y1;

2364

*((uint32_t *)v + 2) = alpha;

2365

*((uint32_t *)v + 7) = alpha;

2366

*((uint32_t *)v + 12) = alpha;

2367

2368

_sna_get_transformed_scaled((int)op->base.src.offset[0] + box->x2,

2369

(int)op->base.src.offset[1] + box->y2,

2370

transform, op->base.src.scale,

2371

&v[3], &v[4]);

2372

2373

_sna_get_transformed_scaled((int)op->base.src.offset[0] + box->x1,

2374

(int)op->base.src.offset[1] + box->y2,

2375

transform, op->base.src.scale,

2376

&v[8], &v[9]);

2377

2378

_sna_get_transformed_scaled((int)op->base.src.offset[0] + box->x1,

2379

(int)op->base.src.offset[1] + box->y1,

2380

transform, op->base.src.scale,

2381

&v[13], &v[14]);

2382

}

2383

#endif

2384

2385

static void

2386

gen2_emit_composite_spans_vertex(struct sna *sna,

2387

const struct sna_composite_spans_op *op,

2388

int16_t x, int16_t y,

2389

float opacity)

2390

{

2391

gen2_emit_composite_dstcoord(sna, x + op->base.dst.x, y + op->base.dst.y);

2392

BATCH((uint8_t)(opacity * 255) << 24);

2393

assert(!op->base.src.is_solid);

2394

if (op->base.src.is_linear)

2395

gen2_emit_composite_linear(sna, &op->base.src, x, y);

2396

else

2397

gen2_emit_composite_texcoord(sna, &op->base.src, x, y);

2398

}

2399

2400

fastcall static void

2401

gen2_emit_composite_spans_primitive(struct sna *sna,

2402

const struct sna_composite_spans_op *op,

2403

const BoxRec *box,

2404

float opacity)

2405

{

2406

gen2_emit_composite_spans_vertex(sna, op, box->x2, box->y2, opacity);

2407

gen2_emit_composite_spans_vertex(sna, op, box->x1, box->y2, opacity);

2408

gen2_emit_composite_spans_vertex(sna, op, box->x1, box->y1, opacity);

2409

}

2410

2411

static void

2412

gen2_emit_spans_pipeline(struct sna *sna,

2413

const struct sna_composite_spans_op *op)

2414

{

2415

uint32_t cblend, ablend;

2416

uint32_t unwind;

2417

2418

cblend =

2419

TB0C_LAST_STAGE | TB0C_RESULT_SCALE_1X | TB0C_OP_MODULATE |

2420

TB0C_ARG1_SEL_DIFFUSE | TB0C_ARG1_REPLICATE_ALPHA |

2421

TB0C_OUTPUT_WRITE_CURRENT;

2422

ablend =

2423

TB0A_RESULT_SCALE_1X | TB0A_OP_MODULATE |

2424

TB0A_ARG1_SEL_DIFFUSE |

2425

TB0A_OUTPUT_WRITE_CURRENT;

2426

2427

if (op->base.src.is_solid) {

2428

ablend |= TB0A_ARG2_SEL_SPECULAR;

2429

cblend |= TB0C_ARG2_SEL_SPECULAR;

2430

if (op->base.dst.format == PICT_a8)

2431

cblend |= TB0C_ARG2_REPLICATE_ALPHA;

2432

} else if (op->base.dst.format == PICT_a8) {

2433

ablend |= TB0A_ARG2_SEL_TEXEL0;

2434

cblend |= TB0C_ARG2_SEL_TEXEL0 | TB0C_ARG2_REPLICATE_ALPHA;

2435

} else {

2436

if (PICT_FORMAT_RGB(op->base.src.pict_format) != 0)

2437

cblend |= TB0C_ARG2_SEL_TEXEL0;

2438

else

2439

cblend |= TB0C_ARG2_SEL_ONE | TB0C_ARG2_INVERT;

2440

2441

if (op->base.src.is_opaque)

2442

ablend |= TB0A_ARG2_SEL_ONE;

2443

else

2444

ablend |= TB0A_ARG2_SEL_TEXEL0;

2445

}

2446

2447

unwind = sna->kgem.nbatch;

2448

BATCH(_3DSTATE_LOAD_STATE_IMMEDIATE_2 |

2449

LOAD_TEXTURE_BLEND_STAGE(0) | 1);

2450

BATCH(cblend);

2451

BATCH(ablend);

2452

if (memcmp(sna->kgem.batch + sna->render_state.gen2.ls2 + 1,

2453

sna->kgem.batch + unwind + 1,

2454

2 * sizeof(uint32_t)) == 0)

2455

sna->kgem.nbatch = unwind;

2456

else

2457

sna->render_state.gen2.ls2 = unwind;

2458

}

2459

2460

static void gen2_emit_composite_spans_state(struct sna *sna,

2461

const struct sna_composite_spans_op *op)

2462

{

2463

uint32_t unwind;

2464

2465

gen2_get_batch(sna, &op->base);

2466

gen2_emit_target(sna,

2467

op->base.dst.bo,

2468

op->base.dst.width,

2469

op->base.dst.height,

2470

op->base.dst.format);

2471

2472

unwind = sna->kgem.nbatch;

2473

BATCH(_3DSTATE_LOAD_STATE_IMMEDIATE_1 |

2474

I1_LOAD_S(2) | I1_LOAD_S(3) | I1_LOAD_S(8) | 2);

2475

BATCH(!op->base.src.is_solid << 12);

2476

BATCH(S3_CULLMODE_NONE | S3_VERTEXHAS_XY | S3_DIFFUSE_PRESENT);

2477

BATCH(gen2_get_blend_cntl(op->base.op, false, op->base.dst.format));

2478

if (memcmp(sna->kgem.batch + sna->render_state.gen2.ls1 + 1,

2479

sna->kgem.batch + unwind + 1,

2480

3 * sizeof(uint32_t)) == 0)

2481

sna->kgem.nbatch = unwind;

2482

else

2483

sna->render_state.gen2.ls1 = unwind;

2484

2485

gen2_disable_logic_op(sna);

2486

gen2_emit_spans_pipeline(sna, op);

2487

2488

if (op->base.src.is_solid) {

2489

if (op->base.src.u.gen2.pixel != sna->render_state.gen2.specular) {

2490

BATCH(_3DSTATE_DFLT_SPECULAR_CMD);

2491

BATCH(op->base.src.u.gen2.pixel);

2492

sna->render_state.gen2.specular = op->base.src.u.gen2.pixel;

2493

}

2494

} else {

2495

uint32_t v =_3DSTATE_VERTEX_FORMAT_2_CMD |

2496

(op->base.src.is_affine ? TEXCOORDFMT_2D : TEXCOORDFMT_3D);

2497

if (sna->render_state.gen2.vft != v) {

2498

BATCH(v);

2499

sna->render_state.gen2.vft = v;

2500

}

2501

gen2_emit_texture(sna, &op->base.src, 0);

2502

}

2503

}

2504

2505

fastcall static void

2506

gen2_render_composite_spans_box(struct sna *sna,

2507

const struct sna_composite_spans_op *op,

2508

const BoxRec *box, float opacity)

2509

{

2510

DBG(("%s: src=+(%d, %d), opacity=%f, dst=+(%d, %d), box=(%d, %d) x (%d, %d)\n",

2511

__FUNCTION__,

2512

op->base.src.offset[0], op->base.src.offset[1],

2513

opacity,

2514

op->base.dst.x, op->base.dst.y,

2515

box->x1, box->y1,

2516

box->x2 - box->x1,

2517

box->y2 - box->y1));

2518

2519

if (gen2_get_rectangles(sna, &op->base, 1) == 0) {

2520

gen2_emit_composite_spans_state(sna, op);

2521

gen2_get_rectangles(sna, &op->base, 1);

2522

}

2523

2524

op->prim_emit(sna, op, box, opacity);

2525

}

2526

2527

static void

2528

gen2_render_composite_spans_boxes(struct sna *sna,

2529

const struct sna_composite_spans_op *op,

2530

const BoxRec *box, int nbox,

2531

float opacity)

2532

{

2533

DBG(("%s: nbox=%d, src=+(%d, %d), opacity=%f, dst=+(%d, %d)\n",

2534

__FUNCTION__, nbox,

2535

op->base.src.offset[0], op->base.src.offset[1],

2536

opacity,

2537

op->base.dst.x, op->base.dst.y));

2538

2539

do {

2540

int nbox_this_time;

2541

2542

nbox_this_time = gen2_get_rectangles(sna, &op->base, nbox);

2543

if (nbox_this_time == 0) {

2544

gen2_emit_composite_spans_state(sna, op);

2545

nbox_this_time = gen2_get_rectangles(sna, &op->base, nbox);

2546

}

2547

nbox -= nbox_this_time;

2548

2549

do {

2550

DBG((" %s: (%d, %d) x (%d, %d)\n", __FUNCTION__,

2551

box->x1, box->y1,

2552

box->x2 - box->x1,

2553

box->y2 - box->y1));

2554

2555

op->prim_emit(sna, op, box++, opacity);

2556

} while (--nbox_this_time);

2557

} while (nbox);

2558

}

2559

2560

fastcall static void

2561

gen2_render_composite_spans_done(struct sna *sna,

2562

const struct sna_composite_spans_op *op)

2563

{

2564

DBG(("%s()\n", __FUNCTION__));

2565

2566

gen2_vertex_flush(sna, &op->base);

2567

2568

if (op->base.src.bo)

2569

kgem_bo_destroy(&sna->kgem, op->base.src.bo);

2570

2571

sna_render_composite_redirect_done(sna, &op->base);

2572

}

2573

2574

static bool

2575

gen2_check_composite_spans(struct sna *sna,

2576

uint8_t op, PicturePtr src, PicturePtr dst,

2577

int16_t width, int16_t height, unsigned flags)

2578

{

2579

if (op >= ARRAY_SIZE(gen2_blend_op))

2580

return false;

2581

2582

if (gen2_composite_fallback(sna, src, NULL, dst))

2583

return false;

2584

2585

if (need_tiling(sna, width, height)) {

2586

if (!is_gpu(sna, dst->pDrawable, PREFER_GPU_SPANS)) {

2587

DBG(("%s: fallback, tiled operation not on GPU\n",

2588

__FUNCTION__));

2589

return false;

2590

}

2591

}

2592

2593

return true;

2594

}

2595

2596

static bool

2597

gen2_render_composite_spans(struct sna *sna,

2598

uint8_t op,

2599

PicturePtr src,

2600

PicturePtr dst,

2601

int16_t src_x, int16_t src_y,

2602

int16_t dst_x, int16_t dst_y,

2603

int16_t width, int16_t height,

2604

unsigned flags,

2605

struct sna_composite_spans_op *tmp)

2606

{

2607

DBG(("%s(src=(%d, %d), dst=(%d, %d), size=(%d, %d))\n", __FUNCTION__,

2608

src_x, src_y, dst_x, dst_y, width, height));

2609

2610

assert(gen2_check_composite_spans(sna, op, src, dst, width, height, flags));

2611

if (need_tiling(sna, width, height)) {

2612

DBG(("%s: tiling, operation (%dx%d) too wide for pipeline\n",

2613

__FUNCTION__, width, height));

2614

return sna_tiling_composite_spans(op, src, dst,

2615

src_x, src_y, dst_x, dst_y,

2616

width, height, flags, tmp);

2617

}

2618

2619

tmp->base.op = op;

2620

sna_render_composite_redirect_init(&tmp->base);

2621

if (!gen2_composite_set_target(sna, &tmp->base, dst,

2622

dst_x, dst_y, width, height,

2623

true)) {

2624

DBG(("%s: unable to set render target\n",

2625

__FUNCTION__));

2626

return false;

2627

}

2628

2629

switch (gen2_composite_picture(sna, src, &tmp->base.src,

2630

src_x, src_y,

2631

width, height,

2632

dst_x, dst_y,

2633

dst->polyMode == PolyModePrecise)) {

2634

case -1:

2635

goto cleanup_dst;

2636

case 0:

2637

gen2_composite_solid_init(sna, &tmp->base.src, 0);

2638

case 1:

2639

break;

2640

}

2641

assert(tmp->base.src.bo || tmp->base.src.is_solid);

2642

2643

tmp->prim_emit = gen2_emit_composite_spans_primitive;

2644

tmp->base.floats_per_vertex = 3;

2645

if (tmp->base.src.is_solid) {

2646

#if defined(sse2) && !defined(__x86_64__)

2647

if (sna->cpu_features & SSE2) {

2648

tmp->prim_emit = gen2_emit_composite_spans_primitive_constant__sse2;

2649

} else

2650

#endif

2651

{

2652

tmp->prim_emit = gen2_emit_composite_spans_primitive_constant;

2653

}

2654

} else if (tmp->base.src.is_linear) {

2655

tmp->base.floats_per_vertex += 2;

2656

#if defined(sse2) && !defined(__x86_64__)

2657

if (sna->cpu_features & SSE2) {

2658

tmp->prim_emit = gen2_emit_composite_spans_primitive_linear__sse2;

2659

} else

2660

#endif

2661

{

2662

tmp->prim_emit = gen2_emit_composite_spans_primitive_linear;

2663

}

2664

} else {

2665

assert(tmp->base.src.bo);

2666

tmp->base.floats_per_vertex += tmp->base.src.is_affine ? 2 : 3;

2667

if (tmp->base.src.transform == NULL) {

2668

#if defined(sse2) && !defined(__x86_64__)

2669

if (sna->cpu_features & SSE2) {

2670

tmp->prim_emit = gen2_emit_composite_spans_primitive_identity_source__sse2;

2671

} else

2672

#endif

2673

{

2674

tmp->prim_emit = gen2_emit_composite_spans_primitive_identity_source;

2675

}

2676

} else if (tmp->base.src.is_affine) {

2677

tmp->base.src.scale[0] /= tmp->base.src.transform->matrix[2][2];

2678

tmp->base.src.scale[1] /= tmp->base.src.transform->matrix[2][2];

2679

#if defined(sse2) && !defined(__x86_64__)

2680

if (sna->cpu_features & SSE2) {

2681

tmp->prim_emit = gen2_emit_composite_spans_primitive_affine_source__sse2;

2682

} else

2683

#endif

2684

{

2685

tmp->prim_emit = gen2_emit_composite_spans_primitive_affine_source;

2686

}

2687

}

2688

}

2689

tmp->base.mask.bo = NULL;

2690

tmp->base.floats_per_rect = 3*tmp->base.floats_per_vertex;

2691

2692

tmp->box = gen2_render_composite_spans_box;

2693

tmp->boxes = gen2_render_composite_spans_boxes;

2694

tmp->done = gen2_render_composite_spans_done;

2695

2696

if (!kgem_check_bo(&sna->kgem,

2697

tmp->base.dst.bo, tmp->base.src.bo,

2698

NULL)) {

2699

kgem_submit(&sna->kgem);

2700

if (!kgem_check_bo(&sna->kgem,

2701

tmp->base.dst.bo, tmp->base.src.bo,

2702

NULL))

2703

goto cleanup_src;

2704

}

2705

2706

gen2_emit_composite_spans_state(sna, tmp);

2707

return true;

2708

2709

cleanup_src:

2710

if (tmp->base.src.bo)

2711

kgem_bo_destroy(&sna->kgem, tmp->base.src.bo);

2712

cleanup_dst:

2713

if (tmp->base.redirect.real_bo)

2714

kgem_bo_destroy(&sna->kgem, tmp->base.dst.bo);

2715

return false;

2716

}

2717

2718

static void

2719

gen2_emit_fill_pipeline(struct sna *sna, const struct sna_composite_op *op)

2720

{

2721

uint32_t blend, unwind;

2722

2723

unwind = sna->kgem.nbatch;

2724

BATCH(_3DSTATE_LOAD_STATE_IMMEDIATE_2 |

2725

LOAD_TEXTURE_BLEND_STAGE(0) | 1);

2726

2727

blend = TB0C_LAST_STAGE | TB0C_RESULT_SCALE_1X | TB0C_OP_ARG1 |

2728

TB0C_ARG1_SEL_DIFFUSE |

2729

TB0C_OUTPUT_WRITE_CURRENT;

2730

if (op->dst.format == PICT_a8)

2731

blend |= TB0C_ARG1_REPLICATE_ALPHA;

2732

BATCH(blend);

2733

2734

BATCH(TB0A_RESULT_SCALE_1X | TB0A_OP_ARG1 |

2735

TB0A_ARG1_SEL_DIFFUSE |

2736

TB0A_OUTPUT_WRITE_CURRENT);

2737

2738

if (memcmp(sna->kgem.batch + sna->render_state.gen2.ls2 + 1,

2739

sna->kgem.batch + unwind + 1,

2740

2 * sizeof(uint32_t)) == 0)

2741

sna->kgem.nbatch = unwind;

2742

else

2743

sna->render_state.gen2.ls2 = unwind;

2744

}

2745

2746

static void gen2_emit_fill_composite_state(struct sna *sna,

2747

const struct sna_composite_op *op,

2748

uint32_t pixel)

2749

{

2750

uint32_t ls1;

2751

2752

gen2_get_batch(sna, op);

2753

gen2_emit_target(sna,

2754

op->dst.bo,

2755

op->dst.width,

2756

op->dst.height,

2757

op->dst.format);

2758

2759

ls1 = sna->kgem.nbatch;

2760

BATCH(_3DSTATE_LOAD_STATE_IMMEDIATE_1 |

2761

I1_LOAD_S(2) | I1_LOAD_S(3) | I1_LOAD_S(8) | 2);

2762

BATCH(0);

2763

BATCH(S3_CULLMODE_NONE | S3_VERTEXHAS_XY);

2764

BATCH(gen2_get_blend_cntl(op->op, false, op->dst.format));

2765

if (memcmp(sna->kgem.batch + sna->render_state.gen2.ls1 + 1,

2766

sna->kgem.batch + ls1 + 1,

2767

3 * sizeof(uint32_t)) == 0)

2768

sna->kgem.nbatch = ls1;

2769

else

2770

sna->render_state.gen2.ls1 = ls1;

2771

2772

gen2_emit_fill_pipeline(sna, op);

2773

2774

if (pixel != sna->render_state.gen2.diffuse) {

2775

BATCH(_3DSTATE_DFLT_DIFFUSE_CMD);

2776

BATCH(pixel);

2777

sna->render_state.gen2.diffuse = pixel;

2778

}

2779

}

2780

2781

static bool

2782

gen2_render_fill_boxes_try_blt(struct sna *sna,

2783

CARD8 op, PictFormat format,

2784

const xRenderColor *color,

2785

const DrawableRec *dst, struct kgem_bo *dst_bo,

2786

const BoxRec *box, int n)

2787

{

2788

uint8_t alu;

2789

uint32_t pixel;

2790

2791

if (op > PictOpSrc)

2792

return false;

2793

2794

if (op == PictOpClear) {

2795

alu = GXclear;

2796

pixel = 0;

2797

} else if (!sna_get_pixel_from_rgba(&pixel,

2798

color->red,

2799

color->green,

2800

color->blue,

2801

color->alpha,

2802

format))

2803

return false;

2804

else

2805

alu = GXcopy;

2806

2807

return sna_blt_fill_boxes(sna, alu,

2808

dst_bo, dst->bitsPerPixel,

2809

pixel, box, n);

2810

}

2811

2812

static bool

2813

gen2_render_fill_boxes(struct sna *sna,

2814

CARD8 op,

2815

PictFormat format,

2816

const xRenderColor *color,

2817

const DrawableRec *dst, struct kgem_bo *dst_bo,

2818

const BoxRec *box, int n)

2819

{

2820

struct sna_composite_op tmp;

2821

uint32_t pixel;

2822

2823

if (op >= ARRAY_SIZE(gen2_blend_op)) {

2824

DBG(("%s: fallback due to unhandled blend op: %d\n",

2825

__FUNCTION__, op));

2826

return false;

2827

}

2828

2829

#if NO_FILL_BOXES

2830

return gen2_render_fill_boxes_try_blt(sna, op, format, color,

2831

dst, dst_bo,

2832

box, n);

2833

#endif

2834

if (gen2_render_fill_boxes_try_blt(sna, op, format, color,

2835

dst, dst_bo,

2836

box, n))

2837

return true;

2838

2839

2840

DBG(("%s (op=%d, format=%x, color=(%04x,%04x,%04x, %04x))\n",

2841

__FUNCTION__, op, (int)format,

2842

color->red, color->green, color->blue, color->alpha));

2843

2844

if (too_large(dst->width, dst->height) ||

2845

dst_bo->pitch < 8 || dst_bo->pitch > MAX_3D_PITCH ||

2846

!gen2_check_dst_format(format)) {

2847

DBG(("%s: try blt, too large or incompatible destination\n",

2848

__FUNCTION__));

2849

if (!gen2_check_dst_format(format))

2850

return false;

2851

2852

assert(dst_bo->pitch >= 8);

2853

return sna_tiling_fill_boxes(sna, op, format, color,

2854

dst, dst_bo, box, n);

2855

}

2856

2857

if (op == PictOpClear)

2858

pixel = 0;

2859

else if (!sna_get_pixel_from_rgba(&pixel,

2860

color->red,

2861

color->green,

2862

color->blue,

2863

color->alpha,

2864

PICT_a8r8g8b8))

2865

return false;

2866

2867

DBG(("%s: using shader for op=%d, format=%x, pixel=%x\n",

2868

__FUNCTION__, op, (int)format, pixel));

2869

2870

memset(&tmp, 0, sizeof(tmp));

2871

tmp.op = op;

2872

tmp.dst.pixmap = (PixmapPtr)dst;

2873

tmp.dst.width = dst->width;

2874

tmp.dst.height = dst->height;

2875

tmp.dst.format = format;

2876

tmp.dst.bo = dst_bo;

2877

tmp.floats_per_vertex = 2;

2878

tmp.floats_per_rect = 6;

2879

2880

if (!kgem_check_bo(&sna->kgem, dst_bo, NULL)) {

2881

kgem_submit(&sna->kgem);

2882

if (!kgem_check_bo(&sna->kgem, dst_bo, NULL))

2883

return false;

2884

}

2885

2886

gen2_emit_fill_composite_state(sna, &tmp, pixel);

2887

2888

do {

2889

int n_this_time = gen2_get_rectangles(sna, &tmp, n);

2890

if (n_this_time == 0) {

2891

gen2_emit_fill_composite_state(sna, &tmp, pixel);

2892

n_this_time = gen2_get_rectangles(sna, &tmp, n);

2893

}

2894

n -= n_this_time;

2895

2896

do {

2897

DBG((" (%d, %d), (%d, %d): %x\n",

2898

box->x1, box->y1, box->x2, box->y2, pixel));

2899

VERTEX(box->x2);

2900

VERTEX(box->y2);

2901

VERTEX(box->x1);

2902

VERTEX(box->y2);

2903

VERTEX(box->x1);

2904

VERTEX(box->y1);

2905

box++;

2906

} while (--n_this_time);

2907

} while (n);

2908

2909

gen2_vertex_flush(sna, &tmp);

2910

return true;

2911

}

2912

2913

static void gen2_emit_fill_state(struct sna *sna,

2914

const struct sna_composite_op *op)

2915

{

2916

uint32_t ls1;

2917

2918

gen2_get_batch(sna, op);

2919

gen2_emit_target(sna,

2920

op->dst.bo,

2921

op->dst.width,

2922

op->dst.height,

2923

op->dst.format);

2924

2925

ls1 = sna->kgem.nbatch;

2926

BATCH(_3DSTATE_LOAD_STATE_IMMEDIATE_1 |

2927

I1_LOAD_S(2) | I1_LOAD_S(3) | I1_LOAD_S(8) | 2);

2928

BATCH(0);

2929

BATCH(S3_CULLMODE_NONE | S3_VERTEXHAS_XY);

2930

BATCH(S8_ENABLE_COLOR_BUFFER_WRITE);

2931

if (memcmp(sna->kgem.batch + sna->render_state.gen2.ls1 + 1,

2932

sna->kgem.batch + ls1 + 1,

2933

3 * sizeof(uint32_t)) == 0)

2934

sna->kgem.nbatch = ls1;

2935

else

2936

sna->render_state.gen2.ls1 = ls1;

2937

2938

gen2_enable_logic_op(sna, op->op);

2939

gen2_emit_fill_pipeline(sna, op);

2940

2941

if (op->src.u.gen2.pixel != sna->render_state.gen2.diffuse) {

2942

BATCH(_3DSTATE_DFLT_DIFFUSE_CMD);

2943

BATCH(op->src.u.gen2.pixel);

2944

sna->render_state.gen2.diffuse = op->src.u.gen2.pixel;

2945

}

2946

}

2947

2948

static void

2949

gen2_render_fill_op_blt(struct sna *sna,

2950

const struct sna_fill_op *op,

2951

int16_t x, int16_t y, int16_t w, int16_t h)

2952

{

2953

if (!gen2_get_rectangles(sna, &op->base, 1)) {

2954

gen2_emit_fill_state(sna, &op->base);

2955

gen2_get_rectangles(sna, &op->base, 1);

2956

}

2957

2958

VERTEX(x+w);

2959

VERTEX(y+h);

2960

VERTEX(x);

2961

VERTEX(y+h);

2962

VERTEX(x);

2963

VERTEX(y);

2964

}

2965

2966

fastcall static void

2967

gen2_render_fill_op_box(struct sna *sna,

2968

const struct sna_fill_op *op,

2969

const BoxRec *box)

2970

{

2971

if (!gen2_get_rectangles(sna, &op->base, 1)) {

2972

gen2_emit_fill_state(sna, &op->base);

2973

gen2_get_rectangles(sna, &op->base, 1);

2974

}

2975

2976

VERTEX(box->x2);

2977

VERTEX(box->y2);

2978

VERTEX(box->x1);

2979

VERTEX(box->y2);

2980

VERTEX(box->x1);

2981

VERTEX(box->y1);

2982

}

2983

2984

fastcall static void

2985

gen2_render_fill_op_boxes(struct sna *sna,

2986

const struct sna_fill_op *op,

2987

const BoxRec *box,

2988

int nbox)

2989

{

2990

DBG(("%s: (%d, %d),(%d, %d)... x %d\n", __FUNCTION__,

2991

box->x1, box->y1, box->x2, box->y2, nbox));

2992

2993

do {

2994

int nbox_this_time = gen2_get_rectangles(sna, &op->base, nbox);

2995

if (nbox_this_time == 0) {

2996

gen2_emit_fill_state(sna, &op->base);

2997

nbox_this_time = gen2_get_rectangles(sna, &op->base, nbox);

2998

}

2999

nbox -= nbox_this_time;

3000

3001

do {

3002

VERTEX(box->x2);

3003

VERTEX(box->y2);

3004

VERTEX(box->x1);

3005

VERTEX(box->y2);

3006

VERTEX(box->x1);

3007

VERTEX(box->y1);

3008

box++;

3009

} while (--nbox_this_time);

3010

} while (nbox);

3011

}

3012

3013

static void

3014

gen2_render_fill_op_done(struct sna *sna, const struct sna_fill_op *op)

3015

{

3016

gen2_vertex_flush(sna, &op->base);

3017

}

3018

3019

static bool

3020

gen2_render_fill(struct sna *sna, uint8_t alu,

3021

PixmapPtr dst, struct kgem_bo *dst_bo,

3022

uint32_t color, unsigned flags,

3023

struct sna_fill_op *tmp)

3024

{

3025

#if NO_FILL

3026

return sna_blt_fill(sna, alu,

3027

dst_bo, dst->drawable.bitsPerPixel,

3028

color,

3029

tmp);

3030

#endif

3031

3032

/* Prefer to use the BLT if already engaged */

3033

if (sna_blt_fill(sna, alu,

3034

dst_bo, dst->drawable.bitsPerPixel,

3035

color,

3036

tmp))

3037

return true;

3038

3039

/* Must use the BLT if we can't RENDER... */

3040

if (too_large(dst->drawable.width, dst->drawable.height) ||

3041

dst_bo->pitch < 8 || dst_bo->pitch > MAX_3D_PITCH)

3042

return false;

3043

3044

tmp->base.op = alu;

3045

tmp->base.dst.pixmap = dst;

3046

tmp->base.dst.width = dst->drawable.width;

3047

tmp->base.dst.height = dst->drawable.height;

3048

tmp->base.dst.format = sna_format_for_depth(dst->drawable.depth);

3049

tmp->base.dst.bo = dst_bo;

3050

tmp->base.dst.x = tmp->base.dst.y = 0;

3051

tmp->base.floats_per_vertex = 2;

3052

tmp->base.floats_per_rect = 6;

3053

3054

tmp->base.src.u.gen2.pixel =

3055

sna_rgba_for_color(color, dst->drawable.depth);

3056

3057

if (!kgem_check_bo(&sna->kgem, dst_bo, NULL)) {

3058

kgem_submit(&sna->kgem);

3059

return sna_blt_fill(sna, alu,

3060

dst_bo, dst->drawable.bitsPerPixel,

3061

color,

3062

tmp);

3063

}

3064

3065

tmp->blt = gen2_render_fill_op_blt;

3066

tmp->box = gen2_render_fill_op_box;

3067

tmp->boxes = gen2_render_fill_op_boxes;

3068

tmp->points = NULL;

3069

tmp->done = gen2_render_fill_op_done;

3070

3071

gen2_emit_fill_state(sna, &tmp->base);

3072

return true;

3073

}

3074

3075

static bool

3076

gen2_render_fill_one_try_blt(struct sna *sna, PixmapPtr dst, struct kgem_bo *bo,

3077

uint32_t color,

3078

int16_t x1, int16_t y1, int16_t x2, int16_t y2,

3079

uint8_t alu)

3080

{

3081

BoxRec box;

3082

3083

box.x1 = x1;

3084

box.y1 = y1;

3085

box.x2 = x2;

3086

box.y2 = y2;

3087

3088

return sna_blt_fill_boxes(sna, alu,

3089

bo, dst->drawable.bitsPerPixel,

3090

color, &box, 1);

3091

}

3092

3093

static bool

3094

gen2_render_fill_one(struct sna *sna, PixmapPtr dst, struct kgem_bo *bo,

3095

uint32_t color,

3096

int16_t x1, int16_t y1,

3097

int16_t x2, int16_t y2,

3098

uint8_t alu)

3099

{

3100

struct sna_composite_op tmp;

3101

3102

#if NO_FILL_ONE

3103

return gen2_render_fill_one_try_blt(sna, dst, bo, color,

3104

x1, y1, x2, y2, alu);

3105

#endif

3106

3107

/* Prefer to use the BLT if already engaged */

3108

if (gen2_render_fill_one_try_blt(sna, dst, bo, color,

3109

x1, y1, x2, y2, alu))

3110

return true;

3111

3112

/* Must use the BLT if we can't RENDER... */

3113

if (too_large(dst->drawable.width, dst->drawable.height) ||

3114

bo->pitch < 8 || bo->pitch > MAX_3D_PITCH)

3115

return false;

3116

3117

if (!kgem_check_bo(&sna->kgem, bo, NULL)) {

3118

kgem_submit(&sna->kgem);

3119

3120

if (gen2_render_fill_one_try_blt(sna, dst, bo, color,

3121

x1, y1, x2, y2, alu))

3122

return true;

3123

3124

if (!kgem_check_bo(&sna->kgem, bo, NULL))

3125

return false;

3126

}

3127

3128

tmp.op = alu;

3129

tmp.dst.pixmap = dst;

3130

tmp.dst.width = dst->drawable.width;

3131

tmp.dst.height = dst->drawable.height;

3132

tmp.dst.format = sna_format_for_depth(dst->drawable.depth);

3133

tmp.dst.bo = bo;

3134

tmp.floats_per_vertex = 2;

3135

tmp.floats_per_rect = 6;

3136

tmp.need_magic_ca_pass = false;

3137

3138

tmp.src.u.gen2.pixel =

3139

sna_rgba_for_color(color, dst->drawable.depth);

3140

3141

gen2_emit_fill_state(sna, &tmp);

3142

gen2_get_rectangles(sna, &tmp, 1);

3143

DBG(("%s: (%d, %d), (%d, %d): %x\n", __FUNCTION__,

3144

x1, y1, x2, y2, tmp.src.u.gen2.pixel));

3145

VERTEX(x2);

3146

VERTEX(y2);

3147

VERTEX(x1);

3148

VERTEX(y2);

3149

VERTEX(x1);

3150

VERTEX(y1);

3151

gen2_vertex_flush(sna, &tmp);

3152

3153

return true;

3154

}

3155

3156

static void

3157

gen2_emit_video_state(struct sna *sna,

3158

struct sna_video *video,

3159

struct sna_video_frame *frame,

3160

PixmapPtr pixmap,

3161

struct kgem_bo *dst_bo,

3162

int width, int height,

3163

bool bilinear)

3164

{

3165

uint32_t ms1, v, unwind;

3166

3167

gen2_emit_target(sna, dst_bo, width, height,

3168

sna_format_for_depth(pixmap->drawable.depth));

3169

3170

unwind = sna->kgem.nbatch;

3171

BATCH(_3DSTATE_LOAD_STATE_IMMEDIATE_1 |

3172

I1_LOAD_S(2) | I1_LOAD_S(3) | I1_LOAD_S(8) | 2);

3173

BATCH(1 << 12);

3174

BATCH(S3_CULLMODE_NONE | S3_VERTEXHAS_XY);

3175

BATCH(S8_ENABLE_COLOR_BUFFER_WRITE);

3176

if (memcmp(sna->kgem.batch + sna->render_state.gen2.ls1 + 1,

3177

sna->kgem.batch + unwind + 1,

3178

3 * sizeof(uint32_t)) == 0)

3179

sna->kgem.nbatch = unwind;

3180

else

3181

sna->render_state.gen2.ls1 = unwind;

3182

3183

gen2_disable_logic_op(sna);

3184

3185

unwind = sna->kgem.nbatch;

3186

BATCH(_3DSTATE_LOAD_STATE_IMMEDIATE_2 |

3187

LOAD_TEXTURE_BLEND_STAGE(0) | 1);

3188

BATCH(TB0C_LAST_STAGE | TB0C_RESULT_SCALE_1X | TB0C_OUTPUT_WRITE_CURRENT |

3189

TB0C_OP_ARG1 | TB0C_ARG1_SEL_TEXEL0);

3190

BATCH(TB0A_RESULT_SCALE_1X | TB0A_OUTPUT_WRITE_CURRENT |

3191

TB0A_OP_ARG1 | TB0A_ARG1_SEL_ONE);

3192

if (memcmp(sna->kgem.batch + sna->render_state.gen2.ls2 + 1,

3193

sna->kgem.batch + unwind + 1,

3194

2 * sizeof(uint32_t)) == 0)

3195

sna->kgem.nbatch = unwind;

3196

else

3197

sna->render_state.gen2.ls2 = unwind;

3198

3199

BATCH(_3DSTATE_LOAD_STATE_IMMEDIATE_2 | LOAD_TEXTURE_MAP(0) | 4);

3200

BATCH(kgem_add_reloc(&sna->kgem, sna->kgem.nbatch,

3201

frame->bo,

3202

I915_GEM_DOMAIN_SAMPLER << 16,

3203

0));

3204

ms1 = MAPSURF_422 | TM0S1_COLORSPACE_CONVERSION;

3205

switch (frame->id) {

3206

case FOURCC_YUY2:

3207

ms1 |= MT_422_YCRCB_NORMAL;

3208

break;

3209

case FOURCC_UYVY:

3210

ms1 |= MT_422_YCRCB_SWAPY;

3211

break;

3212

}

3213

BATCH(((frame->height - 1) << TM0S1_HEIGHT_SHIFT) |

3214

((frame->width - 1) << TM0S1_WIDTH_SHIFT) |

3215

ms1 |

3216

gen2_sampler_tiling_bits(frame->bo->tiling));

3217

BATCH((frame->pitch[0] / 4 - 1) << TM0S2_PITCH_SHIFT | TM0S2_MAP_2D);

3218

if (bilinear)

3219

BATCH(FILTER_LINEAR << TM0S3_MAG_FILTER_SHIFT |

3220

FILTER_LINEAR << TM0S3_MIN_FILTER_SHIFT |

3221

MIPFILTER_NONE << TM0S3_MIP_FILTER_SHIFT);

3222

else

3223

BATCH(FILTER_NEAREST << TM0S3_MAG_FILTER_SHIFT |

3224

FILTER_NEAREST << TM0S3_MIN_FILTER_SHIFT |

3225

MIPFILTER_NONE << TM0S3_MIP_FILTER_SHIFT);

3226

BATCH(0); /* default color */

3227

3228

BATCH(_3DSTATE_MAP_COORD_SET_CMD | TEXCOORD_SET(0) |

3229

ENABLE_TEXCOORD_PARAMS | TEXCOORDS_ARE_NORMAL | TEXCOORDTYPE_CARTESIAN |

3230

ENABLE_ADDR_V_CNTL | TEXCOORD_ADDR_V_MODE(TEXCOORDMODE_CLAMP) |

3231

ENABLE_ADDR_U_CNTL | TEXCOORD_ADDR_U_MODE(TEXCOORDMODE_CLAMP));

3232

3233

v = _3DSTATE_VERTEX_FORMAT_2_CMD | TEXCOORDFMT_2D;

3234

if (sna->render_state.gen2.vft != v) {

3235

BATCH(v);

3236

sna->render_state.gen2.vft = v;

3237

}

3238

}

3239

3240

static void

3241

gen2_video_get_batch(struct sna *sna, struct kgem_bo *bo)

3242

{

3243

kgem_set_mode(&sna->kgem, KGEM_RENDER, bo);

3244

3245

if (!kgem_check_batch(&sna->kgem, 120) ||

3246

!kgem_check_reloc(&sna->kgem, 4) ||

3247

!kgem_check_exec(&sna->kgem, 2)) {

3248

_kgem_submit(&sna->kgem);

3249

_kgem_set_mode(&sna->kgem, KGEM_RENDER);

3250

}

3251

3252

if (sna->render_state.gen2.need_invariant)

3253

gen2_emit_invariant(sna);

3254

}

3255

3256

static int

3257

gen2_get_inline_rectangles(struct sna *sna, int want, int floats_per_vertex)

3258

{

3259

int size = floats_per_vertex * 3;

3260

int rem = batch_space(sna) - 1;

3261

3262

if (rem > MAX_INLINE)

3263

rem = MAX_INLINE;

3264

3265

if (size * want > rem)

3266

want = rem / size;

3267

3268

return want;

3269

}

3270

3271

static bool

3272

gen2_render_video(struct sna *sna,

3273

struct sna_video *video,

3274

struct sna_video_frame *frame,

3275

RegionPtr dstRegion,

3276

PixmapPtr pixmap)

3277

{

3278

struct sna_pixmap *priv = sna_pixmap(pixmap);

3279

const BoxRec *pbox = region_rects(dstRegion);

3280

int nbox = region_num_rects(dstRegion);

3281

int dst_width = dstRegion->extents.x2 - dstRegion->extents.x1;

3282

int dst_height = dstRegion->extents.y2 - dstRegion->extents.y1;

3283

int src_width = frame->src.x2 - frame->src.x1;

3284

int src_height = frame->src.y2 - frame->src.y1;

3285

float src_offset_x, src_offset_y;

3286

float src_scale_x, src_scale_y;

3287

int pix_xoff, pix_yoff;

3288

struct kgem_bo *dst_bo;

3289

bool bilinear;

3290

int copy = 0;

3291

3292

DBG(("%s: src:%dx%d (frame:%dx%d) -> dst:%dx%d\n", __FUNCTION__,

3293

src_width, src_height, frame->width, frame->height, dst_width, dst_height));

3294

3295

assert(priv->gpu_bo);

3296

dst_bo = priv->gpu_bo;

3297

3298

bilinear = src_width != dst_width || src_height != dst_height;

3299

3300

src_scale_x = (float)src_width / dst_width / frame->width;

3301

src_offset_x = (float)frame->src.x1 / frame->width - dstRegion->extents.x1 * src_scale_x;

3302

3303

src_scale_y = (float)src_height / dst_height / frame->height;

3304

src_offset_y = (float)frame->src.y1 / frame->height - dstRegion->extents.y1 * src_scale_y;

3305

DBG(("%s: src offset (%f, %f), scale (%f, %f)\n",

3306

__FUNCTION__, src_offset_x, src_offset_y, src_scale_x, src_scale_y));

3307

3308

if (too_large(pixmap->drawable.width, pixmap->drawable.height) ||

3309

dst_bo->pitch > MAX_3D_PITCH) {

3310

int bpp = pixmap->drawable.bitsPerPixel;

3311

3312

if (too_large(dst_width, dst_height))

3313

return false;

3314

3315

dst_bo = kgem_create_2d(&sna->kgem,

3316

dst_width, dst_height, bpp,

3317

kgem_choose_tiling(&sna->kgem,

3318

I915_TILING_X,

3319

dst_width, dst_height, bpp),

3320

0);

3321

if (!dst_bo)

3322

return false;

3323

3324

pix_xoff = -dstRegion->extents.x1;

3325

pix_yoff = -dstRegion->extents.y1;

3326

copy = 1;

3327

} else {

3328

/* Set up the offset for translating from the given region

3329

* (in screen coordinates) to the backing pixmap.

3330

3331

#ifdef COMPOSITE

3332

pix_xoff = -pixmap->screen_x + pixmap->drawable.x;

3333

pix_yoff = -pixmap->screen_y + pixmap->drawable.y;

3334

#else

3335

pix_xoff = 0;

3336

pix_yoff = 0;

3337

#endif

3338

3339

dst_width = pixmap->drawable.width;

3340

dst_height = pixmap->drawable.height;

3341

}

3342

3343

gen2_video_get_batch(sna, dst_bo);

3344

gen2_emit_video_state(sna, video, frame, pixmap,

3345

dst_bo, dst_width, dst_height, bilinear);

3346

do {

3347

int nbox_this_time = gen2_get_inline_rectangles(sna, nbox, 4);

3348

if (nbox_this_time == 0) {

3349

gen2_video_get_batch(sna, dst_bo);

3350

gen2_emit_video_state(sna, video, frame, pixmap,

3351

dst_bo, dst_width, dst_height, bilinear);

3352

nbox_this_time = gen2_get_inline_rectangles(sna, nbox, 4);

3353

assert(nbox_this_time);

3354

}

3355

nbox -= nbox_this_time;

3356

3357

BATCH(PRIM3D_INLINE | PRIM3D_RECTLIST |

3358

((12 * nbox_this_time) - 1));

3359

do {

3360

int box_x1 = pbox->x1;

3361

int box_y1 = pbox->y1;

3362

int box_x2 = pbox->x2;

3363

int box_y2 = pbox->y2;

3364

3365

pbox++;

3366

3367

DBG(("%s: dst (%d, %d), (%d, %d) + (%d, %d); src (%f, %f), (%f, %f)\n",

3368

__FUNCTION__, box_x1, box_y1, box_x2, box_y2, pix_xoff, pix_yoff,

3369

box_x1 * src_scale_x + src_offset_x,

3370

box_y1 * src_scale_y + src_offset_y,

3371

box_x2 * src_scale_x + src_offset_x,

3372

box_y2 * src_scale_y + src_offset_y));

3373

3374

/* bottom right */

3375

BATCH_F(box_x2 + pix_xoff);

3376

BATCH_F(box_y2 + pix_yoff);

3377

BATCH_F(box_x2 * src_scale_x + src_offset_x);

3378

BATCH_F(box_y2 * src_scale_y + src_offset_y);

3379

3380

/* bottom left */

3381

BATCH_F(box_x1 + pix_xoff);

3382

BATCH_F(box_y2 + pix_yoff);

3383

BATCH_F(box_x1 * src_scale_x + src_offset_x);

3384

BATCH_F(box_y2 * src_scale_y + src_offset_y);

3385

3386

/* top left */

3387

BATCH_F(box_x1 + pix_xoff);

3388

BATCH_F(box_y1 + pix_yoff);

3389

BATCH_F(box_x1 * src_scale_x + src_offset_x);

3390

BATCH_F(box_y1 * src_scale_y + src_offset_y);

3391

} while (--nbox_this_time);

3392

} while (nbox);

3393

3394

if (copy) {

3395

#ifdef COMPOSITE

3396

pix_xoff = -pixmap->screen_x + pixmap->drawable.x;

3397

pix_yoff = -pixmap->screen_y + pixmap->drawable.y;

3398

#else

3399

pix_xoff = 0;

3400

pix_yoff = 0;

3401

#endif

3402

sna_blt_copy_boxes(sna, GXcopy,

3403

dst_bo, -dstRegion->extents.x1, -dstRegion->extents.y1,

3404

priv->gpu_bo, pix_xoff, pix_yoff,

3405

pixmap->drawable.bitsPerPixel,

3406

region_rects(dstRegion),

3407

region_num_rects(dstRegion));

3408

3409

kgem_bo_destroy(&sna->kgem, dst_bo);

3410

}

3411

3412

if (!DAMAGE_IS_ALL(priv->gpu_damage)) {

3413

if ((pix_xoff | pix_yoff) == 0) {

3414

sna_damage_add(&priv->gpu_damage, dstRegion);

3415

} else {

3416

sna_damage_add_boxes(&priv->gpu_damage,

3417

region_rects(dstRegion),

3418

region_num_rects(dstRegion),

3419

pix_xoff, pix_yoff);

3420

}

3421

}

3422

3423

return true;

3424

}

3425

3426

static void

3427

gen2_render_copy_setup_source(struct sna_composite_channel *channel,

3428

const DrawableRec *draw,

3429

struct kgem_bo *bo)

3430

{

3431

assert(draw->width && draw->height);

3432

3433

channel->filter = PictFilterNearest;

3434

channel->repeat = RepeatNone;

3435

channel->width = draw->width;

3436

channel->height = draw->height;

3437

channel->scale[0] = 1.f/draw->width;

3438

channel->scale[1] = 1.f/draw->height;

3439

channel->offset[0] = 0;

3440

channel->offset[1] = 0;

3441

channel->pict_format = sna_format_for_depth(draw->depth);

3442

channel->bo = bo;

3443

channel->is_affine = 1;

3444

3445

DBG(("%s: source=%d, (%dx%d), format=%08x\n",

3446

__FUNCTION__, bo->handle,

3447

channel->width, channel->height,

3448

channel->pict_format));

3449

}

3450

3451

static void

3452

gen2_emit_copy_pipeline(struct sna *sna, const struct sna_composite_op *op)

3453

{

3454

uint32_t blend, unwind;

3455

3456

unwind = sna->kgem.nbatch;

3457

BATCH(_3DSTATE_LOAD_STATE_IMMEDIATE_2 |

3458

LOAD_TEXTURE_BLEND_STAGE(0) | 1);

3459

3460

blend = TB0C_LAST_STAGE | TB0C_RESULT_SCALE_1X | TB0C_OP_ARG1 |

3461

TB0C_OUTPUT_WRITE_CURRENT;

3462

if (op->dst.format == PICT_a8)

3463

blend |= TB0C_ARG1_REPLICATE_ALPHA | TB0C_ARG1_SEL_TEXEL0;

3464

else if (PICT_FORMAT_RGB(op->src.pict_format) != 0)

3465

blend |= TB0C_ARG1_SEL_TEXEL0;

3466

else

3467

blend |= TB0C_ARG1_SEL_ONE | TB0C_ARG1_INVERT; /* 0.0 */

3468

BATCH(blend);

3469

3470

blend = TB0A_RESULT_SCALE_1X | TB0A_OP_ARG1 |

3471

TB0A_OUTPUT_WRITE_CURRENT;

3472

if (PICT_FORMAT_A(op->src.pict_format) == 0)

3473

blend |= TB0A_ARG1_SEL_ONE;

3474

else

3475

blend |= TB0A_ARG1_SEL_TEXEL0;

3476

BATCH(blend);

3477

3478

if (memcmp(sna->kgem.batch + sna->render_state.gen2.ls2 + 1,

3479

sna->kgem.batch + unwind + 1,

3480

2 * sizeof(uint32_t)) == 0)

3481

sna->kgem.nbatch = unwind;

3482

else

3483

sna->render_state.gen2.ls2 = unwind;

3484

}

3485

3486

static void gen2_emit_copy_state(struct sna *sna, const struct sna_composite_op *op)

3487

{

3488

uint32_t ls1, v;

3489

3490

gen2_get_batch(sna, op);

3491

3492

if (kgem_bo_is_dirty(op->src.bo)) {

3493

if (op->src.bo == op->dst.bo)

3494

BATCH(MI_FLUSH | MI_INVALIDATE_MAP_CACHE);

3495

else

3496

BATCH(_3DSTATE_MODES_5_CMD |

3497

PIPELINE_FLUSH_RENDER_CACHE |

3498

PIPELINE_FLUSH_TEXTURE_CACHE);

3499

kgem_clear_dirty(&sna->kgem);

3500

}

3501

gen2_emit_target(sna,

3502

op->dst.bo,

3503

op->dst.width,

3504

op->dst.height,

3505

op->dst.format);

3506

3507

ls1 = sna->kgem.nbatch;

3508

BATCH(_3DSTATE_LOAD_STATE_IMMEDIATE_1 |

3509

I1_LOAD_S(2) | I1_LOAD_S(3) | I1_LOAD_S(8) | 2);

3510

BATCH(1<<12);

3511

BATCH(S3_CULLMODE_NONE | S3_VERTEXHAS_XY);

3512

BATCH(S8_ENABLE_COLOR_BUFFER_WRITE);

3513

if (memcmp(sna->kgem.batch + sna->render_state.gen2.ls1 + 1,

3514

sna->kgem.batch + ls1 + 1,

3515

3 * sizeof(uint32_t)) == 0)

3516

sna->kgem.nbatch = ls1;

3517

else

3518

sna->render_state.gen2.ls1 = ls1;

3519

3520

gen2_enable_logic_op(sna, op->op);

3521

gen2_emit_copy_pipeline(sna, op);

3522

3523

v = _3DSTATE_VERTEX_FORMAT_2_CMD | TEXCOORDFMT_2D;

3524

if (sna->render_state.gen2.vft != v) {

3525

BATCH(v);

3526

sna->render_state.gen2.vft = v;

3527

}

3528

3529

gen2_emit_texture(sna, &op->src, 0);

3530

}

3531

3532

static bool

3533

gen2_render_copy_boxes(struct sna *sna, uint8_t alu,

3534

const DrawableRec *src, struct kgem_bo *src_bo, int16_t src_dx, int16_t src_dy,

3535

const DrawableRec *dst, struct kgem_bo *dst_bo, int16_t dst_dx, int16_t dst_dy,

3536

const BoxRec *box, int n, unsigned flags)

3537

{

3538

struct sna_composite_op tmp;

3539

3540

#if NO_COPY_BOXES

3541

if (!sna_blt_compare_depth(&src->drawable, &dst->drawable))

3542

return false;

3543

3544

return sna_blt_copy_boxes(sna, alu,

3545

src_bo, src_dx, src_dy,

3546

dst_bo, dst_dx, dst_dy,

3547

dst->drawable.bitsPerPixel,

3548

box, n);

3549

#endif

3550

3551

DBG(("%s (%d, %d)->(%d, %d) x %d\n",

3552

__FUNCTION__, src_dx, src_dy, dst_dx, dst_dy, n));

3553

3554

if (sna_blt_compare_depth(src, dst) &&

3555

sna_blt_copy_boxes(sna, alu,

3556

src_bo, src_dx, src_dy,

3557

dst_bo, dst_dx, dst_dy,

3558

dst->bitsPerPixel,

3559

box, n))

3560

return true;

3561

3562

if (src_bo == dst_bo || /* XXX handle overlap using 3D ? */

3563

too_large(src->width, src->height) ||

3564

src_bo->pitch > MAX_3D_PITCH || dst_bo->pitch < 8) {

3565

fallback:

3566

return sna_blt_copy_boxes_fallback(sna, alu,

3567

src, src_bo, src_dx, src_dy,

3568

dst, dst_bo, dst_dx, dst_dy,

3569

box, n);

3570

}

3571

3572

if (!kgem_check_bo(&sna->kgem, dst_bo, src_bo, NULL)) {

3573

kgem_submit(&sna->kgem);

3574

if (!kgem_check_bo(&sna->kgem, dst_bo, src_bo, NULL))

3575

goto fallback;

3576

}

3577

3578

assert(dst_bo->pitch >= 8);

3579

3580

memset(&tmp, 0, sizeof(tmp));

3581

tmp.op = alu;

3582

3583

tmp.dst.pixmap = (PixmapPtr)dst;

3584

tmp.dst.width = dst->width;

3585

tmp.dst.height = dst->height;

3586

tmp.dst.format = sna_format_for_depth(dst->depth);

3587

tmp.dst.bo = dst_bo;

3588

tmp.dst.x = tmp.dst.y = 0;

3589

tmp.damage = NULL;

3590

3591

DBG(("%s: target=%d, format=%08x, size=%dx%d\n",

3592

__FUNCTION__, dst_bo->handle,

3593

(unsigned)tmp.dst.format,

3594

tmp.dst.width,

3595

tmp.dst.height));

3596

3597

sna_render_composite_redirect_init(&tmp);

3598

if (too_large(tmp.dst.width, tmp.dst.height) ||

3599

dst_bo->pitch > MAX_3D_PITCH) {

3600

BoxRec extents = box[0];

3601

int i;

3602

3603

for (i = 1; i < n; i++) {

3604

if (box[i].x1 < extents.x1)

3605

extents.x1 = box[i].x1;

3606

if (box[i].y1 < extents.y1)

3607

extents.y1 = box[i].y1;

3608

3609

if (box[i].x2 > extents.x2)

3610

extents.x2 = box[i].x2;

3611

if (box[i].y2 > extents.y2)

3612

extents.y2 = box[i].y2;

3613

}

3614

if (!sna_render_composite_redirect(sna, &tmp,

3615

extents.x1 + dst_dx,

3616

extents.y1 + dst_dy,

3617

extents.x2 - extents.x1,

3618

extents.y2 - extents.y1,

3619

alu != GXcopy || n > 1))

3620

goto fallback_tiled;

3621

}

3622

3623

tmp.floats_per_vertex = 4;

3624

tmp.floats_per_rect = 12;

3625

3626

dst_dx += tmp.dst.x;

3627

dst_dy += tmp.dst.y;

3628

tmp.dst.x = tmp.dst.y = 0;

3629

3630

gen2_render_copy_setup_source(&tmp.src, src, src_bo);

3631

gen2_emit_copy_state(sna, &tmp);

3632

do {

3633

int n_this_time;

3634

3635

n_this_time = gen2_get_rectangles(sna, &tmp, n);

3636

if (n_this_time == 0) {

3637

gen2_emit_copy_state(sna, &tmp);

3638

n_this_time = gen2_get_rectangles(sna, &tmp, n);

3639

}

3640

n -= n_this_time;

3641

3642

do {

3643

DBG((" (%d, %d) -> (%d, %d) + (%d, %d)\n",

3644

box->x1 + src_dx, box->y1 + src_dy,

3645

box->x1 + dst_dx, box->y1 + dst_dy,

3646

box->x2 - box->x1, box->y2 - box->y1));

3647

VERTEX(box->x2 + dst_dx);

3648

VERTEX(box->y2 + dst_dy);

3649

VERTEX((box->x2 + src_dx) * tmp.src.scale[0]);

3650

VERTEX((box->y2 + src_dy) * tmp.src.scale[1]);

3651

3652

VERTEX(box->x1 + dst_dx);

3653

VERTEX(box->y2 + dst_dy);

3654

VERTEX((box->x1 + src_dx) * tmp.src.scale[0]);

3655

VERTEX((box->y2 + src_dy) * tmp.src.scale[1]);

3656

3657

VERTEX(box->x1 + dst_dx);

3658

VERTEX(box->y1 + dst_dy);

3659

VERTEX((box->x1 + src_dx) * tmp.src.scale[0]);

3660

VERTEX((box->y1 + src_dy) * tmp.src.scale[1]);

3661

3662

box++;

3663

} while (--n_this_time);

3664

} while (n);

3665

3666

gen2_vertex_flush(sna, &tmp);

3667

sna_render_composite_redirect_done(sna, &tmp);

3668

return true;

3669

3670

fallback_tiled:

3671

return sna_tiling_copy_boxes(sna, alu,

3672

src, src_bo, src_dx, src_dy,

3673

dst, dst_bo, dst_dx, dst_dy,

3674

box, n);

3675

}

3676

3677

static void

3678

gen2_render_copy_blt(struct sna *sna,

3679

const struct sna_copy_op *op,

3680

int16_t sx, int16_t sy,

3681

int16_t w, int16_t h,

3682

int16_t dx, int16_t dy)

3683

{

3684

if (!gen2_get_rectangles(sna, &op->base, 1)) {

3685

gen2_emit_copy_state(sna, &op->base);

3686

gen2_get_rectangles(sna, &op->base, 1);

3687

}

3688

3689

VERTEX(dx+w);

3690

VERTEX(dy+h);

3691

VERTEX((sx+w)*op->base.src.scale[0]);

3692

VERTEX((sy+h)*op->base.src.scale[1]);

3693

3694

VERTEX(dx);

3695

VERTEX(dy+h);

3696

VERTEX(sx*op->base.src.scale[0]);

3697

VERTEX((sy+h)*op->base.src.scale[1]);

3698

3699

VERTEX(dx);

3700

VERTEX(dy);

3701

VERTEX(sx*op->base.src.scale[0]);

3702

VERTEX(sy*op->base.src.scale[1]);

3703

}

3704

3705

static void

3706

gen2_render_copy_done(struct sna *sna, const struct sna_copy_op *op)

3707

{

3708

gen2_vertex_flush(sna, &op->base);

3709

}

3710

3711

static bool

3712

gen2_render_copy(struct sna *sna, uint8_t alu,

3713

PixmapPtr src, struct kgem_bo *src_bo,

3714

PixmapPtr dst, struct kgem_bo *dst_bo,

3715

struct sna_copy_op *tmp)

3716

{

3717

#if NO_COPY

3718

if (!sna_blt_compare_depth(&src->drawable, &dst->drawable))

3719

return false;

3720

3721

return sna_blt_copy(sna, alu,

3722

src_bo, dst_bo,

3723

dst->drawable.bitsPerPixel,

3724

tmp);

3725

#endif

3726

3727

/* Prefer to use the BLT */

3728

if (sna_blt_compare_depth(&src->drawable, &dst->drawable) &&

3729

sna_blt_copy(sna, alu,

3730

src_bo, dst_bo,

3731

dst->drawable.bitsPerPixel,

3732

tmp))

3733

return true;

3734

3735

/* Must use the BLT if we can't RENDER... */

3736

if (too_large(src->drawable.width, src->drawable.height) ||

3737

too_large(dst->drawable.width, dst->drawable.height) ||

3738

src_bo->pitch > MAX_3D_PITCH ||

3739

dst_bo->pitch < 8 || dst_bo->pitch > MAX_3D_PITCH) {

3740

fallback:

3741

if (!sna_blt_compare_depth(&src->drawable, &dst->drawable))

3742

return false;

3743

3744

return sna_blt_copy(sna, alu, src_bo, dst_bo,

3745

dst->drawable.bitsPerPixel,

3746

tmp);

3747

}

3748

3749

tmp->base.op = alu;

3750

3751

tmp->base.dst.pixmap = dst;

3752

tmp->base.dst.width = dst->drawable.width;

3753

tmp->base.dst.height = dst->drawable.height;

3754

tmp->base.dst.format = sna_format_for_depth(dst->drawable.depth);

3755

tmp->base.dst.bo = dst_bo;

3756

3757

gen2_render_copy_setup_source(&tmp->base.src, &src->drawable, src_bo);

3758

tmp->base.mask.bo = NULL;

3759

3760

tmp->base.floats_per_vertex = 4;

3761

tmp->base.floats_per_rect = 12;

3762

3763

if (!kgem_check_bo(&sna->kgem, dst_bo, src_bo, NULL)) {

3764

kgem_submit(&sna->kgem);

3765

if (!kgem_check_bo(&sna->kgem, dst_bo, src_bo, NULL))

3766

goto fallback;

3767

}

3768

3769

tmp->blt = gen2_render_copy_blt;

3770

tmp->done = gen2_render_copy_done;

3771

3772

gen2_emit_composite_state(sna, &tmp->base);

3773

return true;

3774

}

3775

3776

static void

3777

gen2_render_reset(struct sna *sna)

3778

{

3779

sna->render_state.gen2.need_invariant = true;

3780

sna->render_state.gen2.logic_op_enabled = 0;

3781

sna->render_state.gen2.target = 0;

3782

3783

sna->render_state.gen2.ls1 = 0;

3784

sna->render_state.gen2.ls2 = 0;

3785

sna->render_state.gen2.vft = 0;

3786

3787

sna->render_state.gen2.diffuse = 0x0c0ffee0;

3788

sna->render_state.gen2.specular = 0x0c0ffee0;

3789

}

3790

3791

static void

3792

gen2_render_flush(struct sna *sna)

3793

{

3794

assert(sna->render.vertex_index == 0);

3795

assert(sna->render.vertex_offset == 0);

3796

}

3797

3798

static void

3799

gen2_render_context_switch(struct kgem *kgem,

3800

int new_mode)

3801

{

3802

struct sna *sna = container_of(kgem, struct sna, kgem);

3803

3804

if (!kgem->nbatch)

3805

return;

3806

3807

/* Reload BLT registers following a lost context */

3808

sna->blt_state.fill_bo = 0;

3809

3810

if (kgem_ring_is_idle(kgem, kgem->ring)) {

3811

DBG(("%s: GPU idle, flushing\n", __FUNCTION__));

3812

_kgem_submit(kgem);

3813

}

3814

}

3815

3816

const char *gen2_render_init(struct sna *sna, const char *backend)

3817

{

3818

struct sna_render *render = &sna->render;

3819

3820

sna->kgem.context_switch = gen2_render_context_switch;

3821

3822

/* Use the BLT (and overlay) for everything except when forced to

3823

* use the texture combiners.

3824

3825

#if !NO_COMPOSITE

3826

render->composite = gen2_render_composite;

3827

render->prefer_gpu |= PREFER_GPU_RENDER;

3828

#endif

3829

#if !NO_COMPOSITE_SPANS

3830

render->check_composite_spans = gen2_check_composite_spans;

3831

render->composite_spans = gen2_render_composite_spans;

3832

render->prefer_gpu |= PREFER_GPU_SPANS;

3833

#endif

3834

render->fill_boxes = gen2_render_fill_boxes;

3835

render->fill = gen2_render_fill;

3836

render->fill_one = gen2_render_fill_one;

3837

render->copy = gen2_render_copy;

3838

render->copy_boxes = gen2_render_copy_boxes;

3839

3840

render->video = gen2_render_video;

3841

3842

render->reset = gen2_render_reset;

3843

render->flush = gen2_render_flush;

3844

3845

render->max_3d_size = MAX_3D_SIZE;

3846

render->max_3d_pitch = MAX_3D_PITCH;

3847

return "Almador (gen2)";

3848

}

Older »