[FFmpeg-cvslog] x86: dsputil: Move inline assembly macros to a separate header
Diego Biurrun
git at videolan.org
Sat Mar 22 22:16:37 CET 2014
ffmpeg | branch: master | Diego Biurrun <diego at biurrun.de> | Mon Jan 27 15:06:54 2014 +0100| [1a8d0cf77ed2611e542ae98f341d4c43a04467bd] | committer: Diego Biurrun
x86: dsputil: Move inline assembly macros to a separate header
> http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=1a8d0cf77ed2611e542ae98f341d4c43a04467bd
---
libavcodec/x86/dsputil_mmx.c | 1 +
libavcodec/x86/dsputil_qns_template.c | 2 +
libavcodec/x86/dsputil_x86.h | 76 -------------------------
libavcodec/x86/fpel_mmx.c | 1 +
libavcodec/x86/hpeldsp_mmx.c | 1 +
libavcodec/x86/inline_asm.h | 100 +++++++++++++++++++++++++++++++++
libavcodec/x86/rnd_mmx.c | 1 +
libavcodec/x86/rnd_template.c | 2 +
8 files changed, 108 insertions(+), 76 deletions(-)
diff --git a/libavcodec/x86/dsputil_mmx.c b/libavcodec/x86/dsputil_mmx.c
index d6136f6..c0b3edd 100644
--- a/libavcodec/x86/dsputil_mmx.c
+++ b/libavcodec/x86/dsputil_mmx.c
@@ -27,6 +27,7 @@
#include "libavutil/x86/asm.h"
#include "constants.h"
#include "dsputil_x86.h"
+#include "inline_asm.h"
#if HAVE_INLINE_ASM
diff --git a/libavcodec/x86/dsputil_qns_template.c b/libavcodec/x86/dsputil_qns_template.c
index 20a40a1..14ab425 100644
--- a/libavcodec/x86/dsputil_qns_template.c
+++ b/libavcodec/x86/dsputil_qns_template.c
@@ -22,6 +22,8 @@
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
+#include "inline_asm.h"
+
#define MAX_ABS (512 >> (SCALE_OFFSET>0 ? SCALE_OFFSET : 0))
static int DEF(try_8x8basis)(int16_t rem[64], int16_t weight[64], int16_t basis[64], int scale)
diff --git a/libavcodec/x86/dsputil_x86.h b/libavcodec/x86/dsputil_x86.h
index 8f8ea05..4b61523 100644
--- a/libavcodec/x86/dsputil_x86.h
+++ b/libavcodec/x86/dsputil_x86.h
@@ -27,82 +27,6 @@
#include "libavcodec/avcodec.h"
#include "libavcodec/dsputil.h"
-#include "libavutil/x86/asm.h"
-#include "constants.h"
-
-#define MOVQ_WONE(regd) \
- __asm__ volatile ( \
- "pcmpeqd %%" #regd ", %%" #regd " \n\t" \
- "psrlw $15, %%" #regd ::)
-
-#define JUMPALIGN() __asm__ volatile (".p2align 3"::)
-#define MOVQ_ZERO(regd) __asm__ volatile ("pxor %%"#regd", %%"#regd ::)
-
-#define MOVQ_BFE(regd) \
- __asm__ volatile ( \
- "pcmpeqd %%"#regd", %%"#regd" \n\t" \
- "paddb %%"#regd", %%"#regd" \n\t" ::)
-
-#ifndef PIC
-#define MOVQ_WTWO(regd) __asm__ volatile ("movq %0, %%"#regd" \n\t" :: "m"(ff_wtwo))
-#else
-// for shared library it's better to use this way for accessing constants
-// pcmpeqd -> -1
-#define MOVQ_WTWO(regd) \
- __asm__ volatile ( \
- "pcmpeqd %%"#regd", %%"#regd" \n\t" \
- "psrlw $15, %%"#regd" \n\t" \
- "psllw $1, %%"#regd" \n\t"::)
-
-#endif
-
-// using regr as temporary and for the output result
-// first argument is unmodifed and second is trashed
-// regfe is supposed to contain 0xfefefefefefefefe
-#define PAVGB_MMX_NO_RND(rega, regb, regr, regfe) \
- "movq "#rega", "#regr" \n\t" \
- "pand "#regb", "#regr" \n\t" \
- "pxor "#rega", "#regb" \n\t" \
- "pand "#regfe", "#regb" \n\t" \
- "psrlq $1, "#regb" \n\t" \
- "paddb "#regb", "#regr" \n\t"
-
-#define PAVGB_MMX(rega, regb, regr, regfe) \
- "movq "#rega", "#regr" \n\t" \
- "por "#regb", "#regr" \n\t" \
- "pxor "#rega", "#regb" \n\t" \
- "pand "#regfe", "#regb" \n\t" \
- "psrlq $1, "#regb" \n\t" \
- "psubb "#regb", "#regr" \n\t"
-
-// mm6 is supposed to contain 0xfefefefefefefefe
-#define PAVGBP_MMX_NO_RND(rega, regb, regr, regc, regd, regp) \
- "movq "#rega", "#regr" \n\t" \
- "movq "#regc", "#regp" \n\t" \
- "pand "#regb", "#regr" \n\t" \
- "pand "#regd", "#regp" \n\t" \
- "pxor "#rega", "#regb" \n\t" \
- "pxor "#regc", "#regd" \n\t" \
- "pand %%mm6, "#regb" \n\t" \
- "pand %%mm6, "#regd" \n\t" \
- "psrlq $1, "#regb" \n\t" \
- "psrlq $1, "#regd" \n\t" \
- "paddb "#regb", "#regr" \n\t" \
- "paddb "#regd", "#regp" \n\t"
-
-#define PAVGBP_MMX(rega, regb, regr, regc, regd, regp) \
- "movq "#rega", "#regr" \n\t" \
- "movq "#regc", "#regp" \n\t" \
- "por "#regb", "#regr" \n\t" \
- "por "#regd", "#regp" \n\t" \
- "pxor "#rega", "#regb" \n\t" \
- "pxor "#regc", "#regd" \n\t" \
- "pand %%mm6, "#regb" \n\t" \
- "pand %%mm6, "#regd" \n\t" \
- "psrlq $1, "#regd" \n\t" \
- "psrlq $1, "#regb" \n\t" \
- "psubb "#regb", "#regr" \n\t" \
- "psubb "#regd", "#regp" \n\t"
void ff_dsputilenc_init_mmx(DSPContext *c, AVCodecContext *avctx,
unsigned high_bit_depth);
diff --git a/libavcodec/x86/fpel_mmx.c b/libavcodec/x86/fpel_mmx.c
index 1ae8f86..9c88ccf 100644
--- a/libavcodec/x86/fpel_mmx.c
+++ b/libavcodec/x86/fpel_mmx.c
@@ -26,6 +26,7 @@
#include "config.h"
#include "dsputil_x86.h"
+#include "inline_asm.h"
#if HAVE_MMX_INLINE
diff --git a/libavcodec/x86/hpeldsp_mmx.c b/libavcodec/x86/hpeldsp_mmx.c
index fece265..b78a0c8 100644
--- a/libavcodec/x86/hpeldsp_mmx.c
+++ b/libavcodec/x86/hpeldsp_mmx.c
@@ -25,6 +25,7 @@
#include "config.h"
#include "dsputil_x86.h"
+#include "inline_asm.h"
#if HAVE_MMX_INLINE
diff --git a/libavcodec/x86/inline_asm.h b/libavcodec/x86/inline_asm.h
new file mode 100644
index 0000000..e4affab
--- /dev/null
+++ b/libavcodec/x86/inline_asm.h
@@ -0,0 +1,100 @@
+/*
+ * inline assembly helper macros
+ *
+ * This file is part of Libav.
+ *
+ * Libav is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * Libav is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with Libav; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#ifndef AVCODEC_X86_INLINE_ASM_H
+#define AVCODEC_X86_INLINE_ASM_H
+
+#include "constants.h"
+
+#define MOVQ_WONE(regd) \
+ __asm__ volatile ( \
+ "pcmpeqd %%" #regd ", %%" #regd " \n\t" \
+ "psrlw $15, %%" #regd ::)
+
+#define JUMPALIGN() __asm__ volatile (".p2align 3"::)
+#define MOVQ_ZERO(regd) __asm__ volatile ("pxor %%"#regd", %%"#regd ::)
+
+#define MOVQ_BFE(regd) \
+ __asm__ volatile ( \
+ "pcmpeqd %%"#regd", %%"#regd" \n\t" \
+ "paddb %%"#regd", %%"#regd" \n\t" ::)
+
+#ifndef PIC
+#define MOVQ_WTWO(regd) __asm__ volatile ("movq %0, %%"#regd" \n\t" :: "m"(ff_wtwo))
+#else
+// for shared library it's better to use this way for accessing constants
+// pcmpeqd -> -1
+#define MOVQ_WTWO(regd) \
+ __asm__ volatile ( \
+ "pcmpeqd %%"#regd", %%"#regd" \n\t" \
+ "psrlw $15, %%"#regd" \n\t" \
+ "psllw $1, %%"#regd" \n\t"::)
+
+#endif
+
+// using regr as temporary and for the output result
+// first argument is unmodifed and second is trashed
+// regfe is supposed to contain 0xfefefefefefefefe
+#define PAVGB_MMX_NO_RND(rega, regb, regr, regfe) \
+ "movq "#rega", "#regr" \n\t" \
+ "pand "#regb", "#regr" \n\t" \
+ "pxor "#rega", "#regb" \n\t" \
+ "pand "#regfe", "#regb" \n\t" \
+ "psrlq $1, "#regb" \n\t" \
+ "paddb "#regb", "#regr" \n\t"
+
+#define PAVGB_MMX(rega, regb, regr, regfe) \
+ "movq "#rega", "#regr" \n\t" \
+ "por "#regb", "#regr" \n\t" \
+ "pxor "#rega", "#regb" \n\t" \
+ "pand "#regfe", "#regb" \n\t" \
+ "psrlq $1, "#regb" \n\t" \
+ "psubb "#regb", "#regr" \n\t"
+
+// mm6 is supposed to contain 0xfefefefefefefefe
+#define PAVGBP_MMX_NO_RND(rega, regb, regr, regc, regd, regp) \
+ "movq "#rega", "#regr" \n\t" \
+ "movq "#regc", "#regp" \n\t" \
+ "pand "#regb", "#regr" \n\t" \
+ "pand "#regd", "#regp" \n\t" \
+ "pxor "#rega", "#regb" \n\t" \
+ "pxor "#regc", "#regd" \n\t" \
+ "pand %%mm6, "#regb" \n\t" \
+ "pand %%mm6, "#regd" \n\t" \
+ "psrlq $1, "#regb" \n\t" \
+ "psrlq $1, "#regd" \n\t" \
+ "paddb "#regb", "#regr" \n\t" \
+ "paddb "#regd", "#regp" \n\t"
+
+#define PAVGBP_MMX(rega, regb, regr, regc, regd, regp) \
+ "movq "#rega", "#regr" \n\t" \
+ "movq "#regc", "#regp" \n\t" \
+ "por "#regb", "#regr" \n\t" \
+ "por "#regd", "#regp" \n\t" \
+ "pxor "#rega", "#regb" \n\t" \
+ "pxor "#regc", "#regd" \n\t" \
+ "pand %%mm6, "#regb" \n\t" \
+ "pand %%mm6, "#regd" \n\t" \
+ "psrlq $1, "#regd" \n\t" \
+ "psrlq $1, "#regb" \n\t" \
+ "psubb "#regb", "#regr" \n\t" \
+ "psubb "#regd", "#regp" \n\t"
+
+#endif /* AVCODEC_X86_INLINE_ASM_H */
diff --git a/libavcodec/x86/rnd_mmx.c b/libavcodec/x86/rnd_mmx.c
index db4515a..f5239fc 100644
--- a/libavcodec/x86/rnd_mmx.c
+++ b/libavcodec/x86/rnd_mmx.c
@@ -18,6 +18,7 @@
#include "config.h"
#include "dsputil_x86.h"
+#include "inline_asm.h"
#if HAVE_INLINE_ASM
diff --git a/libavcodec/x86/rnd_template.c b/libavcodec/x86/rnd_template.c
index 144f111..a9fb132 100644
--- a/libavcodec/x86/rnd_template.c
+++ b/libavcodec/x86/rnd_template.c
@@ -27,6 +27,8 @@
#include <stddef.h>
#include <stdint.h>
+#include "inline_asm.h"
+
// put_pixels
STATIC void DEF(put, pixels8_xy2)(uint8_t *block, const uint8_t *pixels,
ptrdiff_t line_size, int h)
More information about the ffmpeg-cvslog
mailing list