* mostly rewritten by Michael Niedermayer <michaelni@gmx.at>
*/
+/* XXX: we use explicit registers to avoid a gcc 2.95.2 register asm
+ clobber bug */
static void DEF(put_pixels_x2)(UINT8 *block, const UINT8 *pixels, int line_size, int h)
{
__asm __volatile(
"subl $4, %0 \n\t"
" jnz 1b \n\t"
:"+g"(h)
- :"r"(pixels), "r"(pixels+line_size), "r" (block), "r" (block+line_size),
- "r"(line_size<<1)
+ :"b"(pixels), "c"(pixels+line_size), "d" (block), "S" (block+line_size),
+ "D"(line_size<<1)
:"%eax", "memory");
}
+/* GL: this function does incorrect rounding if overflow */
static void DEF(put_no_rnd_pixels_x2)(UINT8 *block, const UINT8 *pixels, int line_size, int h)
{
__asm __volatile(
"subl $4, %0 \n\t"
" jnz 1b \n\t"
:"+g"(h)
- :"r"(pixels), "r"(pixels+line_size), "r" (block), "r" (block+line_size),
- "r"(line_size<<1)
+ :"b"(pixels), "c"(pixels+line_size), "d" (block), "S" (block+line_size),
+ "D"(line_size<<1)
:"%eax", "memory");
}
"subl $4, %0 \n\t"
" jnz 1b \n\t"
:"+g"(h)
- :"r"(pixels), "r"(pixels+line_size), "r"(pixels+line_size*2), "r" (block),
- "r" (block+line_size), "g"(line_size<<1)
+ :"b"(pixels), "c"(pixels+line_size), "d"(pixels+line_size*2), "S" (block),
+ "D" (block+line_size), "g"(line_size<<1)
:"%eax", "memory");
}
+/* GL: this function does incorrect rounding if overflow */
static void DEF(put_no_rnd_pixels_y2)(UINT8 *block, const UINT8 *pixels, int line_size, int h)
{
__asm __volatile(
"subl $4, %0 \n\t"
" jnz 1b \n\t"
:"+g"(h)
- :"r"(pixels), "r"(pixels+line_size), "r"(pixels+line_size*2), "r" (block),
- "r" (block+line_size), "g"(line_size<<1)
+ :"b"(pixels), "c"(pixels+line_size), "d"(pixels+line_size*2), "S" (block),
+ "D" (block+line_size), "g"(line_size<<1)
:"%eax", "memory");
}
"subl $4, %0 \n\t"
" jnz 1b \n\t"
:"+g"(h)
- :"r"(pixels), "r"(pixels+line_size), "r" (block), "r" (block+line_size),
- "r"(line_size<<1)
+ :"b"(pixels), "c"(pixels+line_size), "d" (block), "S" (block+line_size),
+ "D"(line_size<<1)
:"%eax", "memory");
}
"subl $4, %0 \n\t"
" jnz 1b \n\t"
:"+g"(h)
- :"r"(pixels), "r"(pixels+line_size), "r" (block), "r" (block+line_size),
- "r"(line_size<<1)
+ :"b"(pixels), "c"(pixels+line_size), "d" (block), "S" (block+line_size),
+ "D"(line_size<<1)
:"%eax", "memory");
}
"subl $4, %0 \n\t"
" jnz 1b \n\t"
:"+g"(h)
- :"r"(pixels), "r"(pixels+line_size), "r"(pixels+line_size*2), "r" (block),
- "r" (block+line_size), "g"(line_size<<1)
+ :"b"(pixels), "c"(pixels+line_size), "d"(pixels+line_size*2), "S" (block),
+ "D" (block+line_size), "g"(line_size<<1)
:"%eax", "memory");
}
"subl $4, %0 \n\t"
" jnz 1b \n\t"
:"+g"(h)
- :"r"(pixels), "r"(pixels+line_size), "r"(pixels+line_size*2), "r" (block),
- "r" (block+line_size), "g"(line_size<<1)
+ :"b"(pixels), "c"(pixels+line_size), "d"(pixels+line_size*2), "S" (block),
+ "D" (block+line_size), "g"(line_size<<1)
:"%eax", "memory");
}