X-Git-Url: https://notaz.gp2x.de/cgi-bin/gitweb.cgi?p=pcsx_rearmed.git;a=blobdiff_plain;f=plugins%2Fgpu_neon%2Fpsx_gpu%2Fpsx_gpu_arm_neon.S;h=6108bc35b26d12951b457f9c55776b0bd17f2015;hp=fd9979808f81814a44c6e56e58dd846eb48b71fb;hb=ad38f92fe406f2f0c9008e5a85d7e02a6410f9e5;hpb=3867c6efed8d1cd6cd40f07cd46876f59da8912f

diff --git a/plugins/gpu_neon/psx_gpu/psx_gpu_arm_neon.S b/plugins/gpu_neon/psx_gpu/psx_gpu_arm_neon.S
index fd997980..6108bc35 100644
--- a/plugins/gpu_neon/psx_gpu/psx_gpu_arm_neon.S
+++ b/plugins/gpu_neon/psx_gpu/psx_gpu_arm_neon.S
@@ -241,6 +241,18 @@
 
 .align 4
 
+/* FIXME: users of this should be in psx_gpu instead */
+#ifndef __PIC__
+#define load_pointer(register, pointer)                                        \
+  movw register, :lower16:pointer;                                             \
+  movt register, :upper16:pointer;                                             \
+
+#else
+#define load_pointer(register, pointer)                                        \
+  ldr  register, =pointer                                                      \
+
+#endif
+
 #define function(name)                                                         \
   .global name;                                                                \
   name:                                                                        \
@@ -609,8 +621,7 @@ function(compute_all_gradients)
   vld1.32 { uvrg }, [ temp ];                                                  \
   add temp, psx_gpu, #psx_gpu_uvrg_dy_offset;                                  \
   vld1.32 { uvrg_dy }, [ temp ];                                               \
-  movw reciprocal_table_ptr, :lower16:reciprocal_table;                        \
-  movt reciprocal_table_ptr, :upper16:reciprocal_table;                        \
+  load_pointer(reciprocal_table_ptr, reciprocal_table);                        \
                                                                                \
   vmov.u32 c_0x01, #0x01                                                       \
 
@@ -1016,6 +1027,7 @@ function(setup_spans_up_left)
 function(setup_spans_up_right)
   setup_spans_up_up(right, left)
 
+.pool
 
 #define setup_spans_down_down(minor, major)                                    \
   setup_spans_prologue();                                                      \
@@ -1224,6 +1236,7 @@ function(setup_spans_up_down)
   setup_spans_prologue_b()
   bal 4b
 
+.pool
 
 #undef span_uvrg_offset
 #undef span_edge_data
@@ -2002,8 +2015,6 @@ function(setup_blocks_unshaded_untextured_undithered_unswizzled_direct)
  1:
   add span_edge_data, span_edge_data, #8
   subs num_spans, num_spans, #1
-
-  strh num_blocks, [ psx_gpu, #psx_gpu_num_blocks_offset ]
   bne 0b
 
   ldmia sp!, { r4 - r11, pc }
@@ -4339,102 +4350,52 @@ function(warmup)
 
   bx lr
 
+#undef vram_ptr
 #undef color
-#undef y
+#undef width
 #undef height
-
-#define psx_gpu                                           r0
-#define color                                             r1
-#define x                                                 r2
-#define y                                                 r3
+#undef pitch
 
 #define vram_ptr                                          r0
-#define width                                             r3
-#define height                                            r12
-
-#define parameter_width_offset                            0
-#define parameter_height_offset                           4
+#define color                                             r1
+#define width                                             r2
+#define height                                            r3
 
-#define color_r                                           r14
-#define color_g                                           r4
-#define color_b                                           r5  
+#define pitch                                             r1
 
-#define left_unaligned                                    r14
-#define right_unaligned                                   r4
-#define pitch                                             r5
-#define num_unaligned                                     r2
-#define num_width                                         r6
+#define num_width                                         r12
 
-#undef colors
+#undef colors_a
+#undef colors_b
 
-#define colors                                            q0
+#define colors_a                                          q0
+#define colors_b                                          q1
 
 .align 3
 
 function(render_block_fill_body)
-  ldr vram_ptr, [ psx_gpu, #psx_gpu_vram_ptr_offset ]
-  ldr height, [ sp, #parameter_height_offset ]
-
-  add vram_ptr, vram_ptr, y, lsl #11
-  ldr width, [ sp, #parameter_width_offset ]
-
-  add vram_ptr, vram_ptr, x, lsl #1
-  stmdb sp!, { r4 - r6, r14 }
-
-  ubfx color_r, color, #3, #5
-  ubfx color_g, color, #11, #5
-
-  ubfx color_b, color, #19, #5
-  orr color, color_r, color_g, lsl #5
-
-  orr color, color, color_b, lsl #10
-  add left_unaligned, x, #0x7
-
-  bic left_unaligned, left_unaligned, #0x7
-  vdup.u16 colors, color
-
-  sub left_unaligned, left_unaligned, x
+  vdup.u16 colors_a, color
   mov pitch, #2048
 
+  vmov colors_b, colors_a
   sub pitch, pitch, width, lsl #1
-  sub width, width, left_unaligned
-
-  and right_unaligned, width, #0x7
-  bic width, width, #0x7
 
- 0:
-  mov num_width, width, lsr #3
+  mov num_width, width
 
-  movs num_unaligned, left_unaligned
-  beq 2f
+ 0:  
+  vst1.u32 { colors_a, colors_b }, [ vram_ptr, :256 ]!
 
- 1:
-  strh color, [ vram_ptr ], #2
-
-  subs num_unaligned, num_unaligned, #1
-  bne 1b
-
- 2:
-  vst1.u32 { colors }, [ vram_ptr, :128 ]!
-  subs num_width, num_width, #1
-  bne 2b
-
-  movs num_unaligned, right_unaligned
-  beq 4f
-
- 3:
-  strh color, [ vram_ptr ], #2
-  
-  subs num_unaligned, num_unaligned, #1
-  bne 3b
+  subs num_width, num_width, #2
+  bne 0b
 
- 4:
   add vram_ptr, vram_ptr, pitch
+  mov num_width, width
+
   subs height, height, #1
   bne 0b
-  
-  ldmia sp!, { r4 - r6, pc }
 
+  bx lr
+ 
 
 #undef x
 #undef y