[FFmpeg-cvslog] x86: cavs: Remove an unneeded scratch buffer

[Michael Niedermayer]

ffmpeg | branch: release/2.4 | Michael Niedermayer <michaelni at gmx.at> | Thu May 28 12:38:35 2015 +0200| [4dc0fbb13c33b4e5bdb766652f4daf900ccc952f] | committer: Luca Barbato

x86: cavs: Remove an unneeded scratch buffer

Simplifies the code and makes it build on certain compilers
running out of registers on x86.

CC: libav-stable at libav.org
Reported-By: mudler
(cherry picked from commit e4610300de6869bd6b3b00e76cfeabb6d7653dcd)
Signed-off-by: Luca Barbato <lu_zero at gentoo.org>

> http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=4dc0fbb13c33b4e5bdb766652f4daf900ccc952f
---

 libavcodec/x86/cavsdsp.c |   28 +++++++++++++---------------
 1 file changed, 13 insertions(+), 15 deletions(-)

diff --git a/libavcodec/x86/cavsdsp.c b/libavcodec/x86/cavsdsp.c
index b323a10..39eec4b 100644
--- a/libavcodec/x86/cavsdsp.c
+++ b/libavcodec/x86/cavsdsp.c
@@ -142,9 +142,7 @@ static void cavs_idct8_add_mmx(uint8_t *dst, int16_t *block, int stride)
     DECLARE_ALIGNED(8, int16_t, b2)[64];
 
     for(i=0; i<2; i++){
-        DECLARE_ALIGNED(8, uint64_t, tmp);
-
-        cavs_idct8_1d(block+4*i, ff_pw_4.a);
+        cavs_idct8_1d(block + 4 * i, ff_pw_4.a);
 
         __asm__ volatile(
             "psraw     $3, %%mm7  \n\t"
@@ -155,20 +153,20 @@ static void cavs_idct8_add_mmx(uint8_t *dst, int16_t *block, int stride)
             "psraw     $3, %%mm2  \n\t"
             "psraw     $3, %%mm1  \n\t"
             "psraw     $3, %%mm0  \n\t"
-            "movq   %%mm7,    %0   \n\t"
+            "movq   %%mm7,  (%0)  \n\t"
             TRANSPOSE4( %%mm0, %%mm2, %%mm4, %%mm6, %%mm7 )
-            "movq   %%mm0,  8(%1)  \n\t"
-            "movq   %%mm6, 24(%1)  \n\t"
-            "movq   %%mm7, 40(%1)  \n\t"
-            "movq   %%mm4, 56(%1)  \n\t"
-            "movq    %0,    %%mm7  \n\t"
+            "movq   %%mm0,  8(%0)  \n\t"
+            "movq   %%mm6, 24(%0)  \n\t"
+            "movq   %%mm7, 40(%0)  \n\t"
+            "movq   %%mm4, 56(%0)  \n\t"
+            "movq    (%0),  %%mm7  \n\t"
             TRANSPOSE4( %%mm7, %%mm5, %%mm3, %%mm1, %%mm0 )
-            "movq   %%mm7,   (%1)  \n\t"
-            "movq   %%mm1, 16(%1)  \n\t"
-            "movq   %%mm0, 32(%1)  \n\t"
-            "movq   %%mm3, 48(%1)  \n\t"
-            : "=m"(tmp)
-            : "r"(b2+32*i)
+            "movq   %%mm7,   (%0)  \n\t"
+            "movq   %%mm1, 16(%0)  \n\t"
+            "movq   %%mm0, 32(%0)  \n\t"
+            "movq   %%mm3, 48(%0)  \n\t"
+            :
+            : "r"(b2 + 32 * i)
             : "memory"
         );
     }