[FFmpeg-devel] [PATCH 1/9] lavu/riscv: allow any number of extensions

[FFmpeg-devel] [PATCH 1/9] lavu/riscv: allow any number of extensions

[Rémi Denis-Courmont]

This reworks the func/endfunc macros to support any number of ISA extension
as parameters.
---
 libavutil/riscv/asm.S | 17 +++++++++--------
 1 file changed, 9 insertions(+), 8 deletions(-)

diff --git a/libavutil/riscv/asm.S b/libavutil/riscv/asm.S
index 2cf4f7b7ab..78e9defbd4 100644
--- a/libavutil/riscv/asm.S
+++ b/libavutil/riscv/asm.S
@@ -36,17 +36,18 @@
 #define HWD
 #endif
 
-        .macro func sym, ext1=, ext2=
+        .macro  archadd ext=, more:vararg
+            .ifnb   \ext
+                .option arch, +\ext
+                archadd \more
+            .endif
+        .endm
+
+        .macro func sym, exts:vararg
             .text
             .align 2
-
             .option push
-            .ifnb \ext1
-            .option arch, +\ext1
-            .ifnb \ext2
-            .option arch, +\ext2
-            .endif
-            .endif
+            archadd \exts
 
             .global \sym
             .hidden \sym
-- 
2.45.2

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

[FFmpeg-devel] [PATCH 2/9] lavu/riscv: grok B as an extension

[Rémi Denis-Courmont]

The RISC-V B bit manipulation extension was ratified only two months ago.
But it is strictly equivalent to the union of the zba, zbb and zbs
extensions which were defined almost 3 years earlier. Rather than require
new assembler, we can just match the extension name manually and translate
it into its constituent parts.
---
 libavutil/riscv/asm.S | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/libavutil/riscv/asm.S b/libavutil/riscv/asm.S
index 78e9defbd4..0c29680d84 100644
--- a/libavutil/riscv/asm.S
+++ b/libavutil/riscv/asm.S
@@ -38,7 +38,12 @@
 
         .macro  archadd ext=, more:vararg
             .ifnb   \ext
-                .option arch, +\ext
+                .ifc    \ext, b
+                # B was defined later, is known to fewer assemblers.
+                archadd zba, zbb, zbs
+                .else
+                    .option arch, +\ext
+                .endif
                 archadd \more
             .endif
         .endm
-- 
2.45.2

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

[FFmpeg-devel] [PATCH 3/9] lavc/riscv: require B or zba explicitly

[Rémi Denis-Courmont]

---
 libavcodec/riscv/aacencdsp_rvv.S   |  4 +--
 libavcodec/riscv/aacpsdsp_rvv.S    | 10 +++----
 libavcodec/riscv/ac3dsp_rvv.S      |  6 ++--
 libavcodec/riscv/ac3dsp_rvvb.S     |  2 +-
 libavcodec/riscv/alacdsp_rvv.S     |  6 ++--
 libavcodec/riscv/audiodsp_rvv.S    |  6 ++--
 libavcodec/riscv/bswapdsp_rvb.S    |  2 +-
 libavcodec/riscv/bswapdsp_rvv.S    |  2 +-
 libavcodec/riscv/exrdsp_rvv.S      |  2 +-
 libavcodec/riscv/fixed_vtype.S     | 48 ++++++++++++++++++++++++++++++
 libavcodec/riscv/flacdsp_rvv.S     | 42 +++++++++++++-------------
 libavcodec/riscv/fmtconvert_rvv.S  |  4 +--
 libavcodec/riscv/h264_mc_chroma.S  |  4 +--
 libavcodec/riscv/h264idct_rvv.S    |  2 +-
 libavcodec/riscv/huffyuvdsp_rvv.S  |  4 +--
 libavcodec/riscv/jpeg2000dsp_rvv.S |  4 +--
 libavcodec/riscv/llauddsp_rvv.S    |  4 +--
 libavcodec/riscv/lpc_rvv.S         |  4 +--
 libavcodec/riscv/opusdsp_rvv.S     |  2 +-
 libavcodec/riscv/rv40dsp_rvv.S     |  4 +--
 libavcodec/riscv/sbrdsp_rvv.S      | 16 +++++-----
 libavcodec/riscv/svqenc_rvv.S      |  2 +-
 libavcodec/riscv/takdsp_rvv.S      |  8 ++---
 libavcodec/riscv/utvideodsp_rvv.S  |  4 +--
 libavcodec/riscv/vc1dsp_rvv.S      |  6 ++--
 libavcodec/riscv/vorbisdsp_rvv.S   |  2 +-
 libavcodec/riscv/vp7dsp_rvv.S      |  2 +-
 libavcodec/riscv/vp8dsp_rvv.S      |  4 +--
 libavcodec/riscv/vp9_intra_rvi.S   |  6 ++--
 29 files changed, 129 insertions(+), 83 deletions(-)
 create mode 100644 libavcodec/riscv/fixed_vtype.S

diff --git a/libavcodec/riscv/aacencdsp_rvv.S b/libavcodec/riscv/aacencdsp_rvv.S
index 21e66a77ae..05a603b6f6 100644
--- a/libavcodec/riscv/aacencdsp_rvv.S
+++ b/libavcodec/riscv/aacencdsp_rvv.S
@@ -21,7 +21,7 @@
 
 #include "libavutil/riscv/asm.S"
 
-func ff_abs_pow34_rvv, zve32f
+func ff_abs_pow34_rvv, zve32f, zba
 1:
         vsetvli      t0, a2, e32, m8, ta, ma
         sub          a2, a2, t0
@@ -38,7 +38,7 @@ func ff_abs_pow34_rvv, zve32f
         ret
 endfunc
 
-func ff_aac_quant_bands_rvv, zve32f
+func ff_aac_quant_bands_rvv, zve32f, zba
 NOHWF   fmv.w.x     fa0, a6
 NOHWF   fmv.w.x     fa1, a7
         fcvt.s.w    ft0, a5
diff --git a/libavcodec/riscv/aacpsdsp_rvv.S b/libavcodec/riscv/aacpsdsp_rvv.S
index 2d6858688a..72e2103c22 100644
--- a/libavcodec/riscv/aacpsdsp_rvv.S
+++ b/libavcodec/riscv/aacpsdsp_rvv.S
@@ -20,7 +20,7 @@
 
 #include "libavutil/riscv/asm.S"
 
-func ff_ps_add_squares_rvv, zve64f
+func ff_ps_add_squares_rvv, zve64f, zba
         li          t1, 32
 1:
         vsetvli     t0, a2, e32, m4, ta, ma
@@ -39,7 +39,7 @@ func ff_ps_add_squares_rvv, zve64f
         ret
 endfunc
 
-func ff_ps_mul_pair_single_rvv, zve32f
+func ff_ps_mul_pair_single_rvv, zve32f, zba
 1:
         vsetvli     t0, a3, e32, m4, ta, ma
         vlseg2e32.v v24, (a1)
@@ -134,7 +134,7 @@ NOHWD   flw     fs\n, (4 * \n)(sp)
         .purgem filter
 endfunc
 
-func ff_ps_hybrid_analysis_ileave_rvv, zve32x /* no needs for zve32f here */
+func ff_ps_hybrid_analysis_ileave_rvv, zve32x /* no zve32f here */, zba
         slli        t0, a2, 5 + 1 + 2 // ctz(32 * 2 * 4)
         sh2add      a1, a2, a1
         add         a0, a0, t0
@@ -169,7 +169,7 @@ func ff_ps_hybrid_analysis_ileave_rvv, zve32x /* no needs for zve32f here */
         ret
 endfunc
 
-func ff_ps_hybrid_synthesis_deint_rvv, zve64x
+func ff_ps_hybrid_synthesis_deint_rvv, zve64x, zba
         slli    t0, a2, 5 + 1 + 2
         sh2add  a0, a2, a0
         add     a1, a1, t0
@@ -207,7 +207,7 @@ func ff_ps_hybrid_synthesis_deint_rvv, zve64x
         ret
 endfunc
 
-func ff_ps_stereo_interpolate_rvv, zve32f, zbb
+func ff_ps_stereo_interpolate_rvv, zve32f, b
         vsetvli      t0, zero, e32, m2, ta, ma
         vid.v        v24
         flw          ft0,   (a2)
diff --git a/libavcodec/riscv/ac3dsp_rvv.S b/libavcodec/riscv/ac3dsp_rvv.S
index 1b5f67a9ec..c733733286 100644
--- a/libavcodec/riscv/ac3dsp_rvv.S
+++ b/libavcodec/riscv/ac3dsp_rvv.S
@@ -43,7 +43,7 @@ func ff_ac3_exponent_min_rvv, zve32x
         ret
 endfunc
 
-func ff_float_to_fixed24_rvv, zve32f
+func ff_float_to_fixed24_rvv, zve32f, zba
         li            t1, 1 << 24
         fcvt.s.w      f0, t1
 1:
@@ -61,7 +61,7 @@ func ff_float_to_fixed24_rvv, zve32f
 endfunc
 
 #if __riscv_xlen >= 64
-func ff_sum_square_butterfly_int32_rvv, zve64x
+func ff_sum_square_butterfly_int32_rvv, zve64x, zba
         vsetvli    t0, zero, e64, m8, ta, ma
         vmv.v.x    v0, zero
         vmv.v.x    v8, zero
@@ -101,7 +101,7 @@ func ff_sum_square_butterfly_int32_rvv, zve64x
 endfunc
 #endif
 
-func ff_sum_square_butterfly_float_rvv, zve32f
+func ff_sum_square_butterfly_float_rvv, zve32f, zba
         vsetvli     t0, zero, e32, m8, ta, ma
         vmv.v.x     v0, zero
         vmv.v.x     v8, zero
diff --git a/libavcodec/riscv/ac3dsp_rvvb.S b/libavcodec/riscv/ac3dsp_rvvb.S
index 64766b56be..5bffb40bba 100644
--- a/libavcodec/riscv/ac3dsp_rvvb.S
+++ b/libavcodec/riscv/ac3dsp_rvvb.S
@@ -21,7 +21,7 @@
 #include "config.h"
 #include "libavutil/riscv/asm.S"
 
-func ff_extract_exponents_rvvb, zve32x, zvbb
+func ff_extract_exponents_rvvb, zve32x, zvbb, zba
 1:
         vsetvli     t0, a2, e32, m8, ta, ma
         vle32.v     v8, (a1)
diff --git a/libavcodec/riscv/alacdsp_rvv.S b/libavcodec/riscv/alacdsp_rvv.S
index 8efb04e0c8..19714bd6e3 100644
--- a/libavcodec/riscv/alacdsp_rvv.S
+++ b/libavcodec/riscv/alacdsp_rvv.S
@@ -21,7 +21,7 @@
 #include "libavutil/riscv/asm.S"
 
 #if (__riscv_xlen == 64)
-func ff_alac_decorrelate_stereo_rvv, zve32x
+func ff_alac_decorrelate_stereo_rvv, zve32x, zba
         ld          a4, 8(a0)
         ld          a0, 0(a0)
 1:
@@ -43,7 +43,7 @@ func ff_alac_decorrelate_stereo_rvv, zve32x
         ret
 endfunc
 
-func ff_alac_append_extra_bits_mono_rvv, zve32x
+func ff_alac_append_extra_bits_mono_rvv, zve32x, zba
         ld      a0, (a0)
         ld      a1, (a1)
 1:
@@ -61,7 +61,7 @@ func ff_alac_append_extra_bits_mono_rvv, zve32x
         ret
 endfunc
 
-func ff_alac_append_extra_bits_stereo_rvv, zve32x
+func ff_alac_append_extra_bits_stereo_rvv, zve32x, zba
         ld      a6, 8(a0)
         ld      a0,  (a0)
         ld      a7, 8(a1)
diff --git a/libavcodec/riscv/audiodsp_rvv.S b/libavcodec/riscv/audiodsp_rvv.S
index f0b23bab5e..b7134de523 100644
--- a/libavcodec/riscv/audiodsp_rvv.S
+++ b/libavcodec/riscv/audiodsp_rvv.S
@@ -20,7 +20,7 @@
 
 #include "libavutil/riscv/asm.S"
 
-func ff_scalarproduct_int16_rvv, zve32x
+func ff_scalarproduct_int16_rvv, zve32x, zba
         vsetvli     t0, zero, e32, m8, ta, ma
         vmv.v.x     v8, zero
         vmv.s.x     v0, zero
@@ -40,7 +40,7 @@ func ff_scalarproduct_int16_rvv, zve32x
         ret
 endfunc
 
-func ff_vector_clip_int32_rvv, zve32x
+func ff_vector_clip_int32_rvv, zve32x, zba
 1:
         vsetvli t0, a4, e32, m8, ta, ma
         vle32.v v8, (a1)
@@ -55,7 +55,7 @@ func ff_vector_clip_int32_rvv, zve32x
         ret
 endfunc
 
-func ff_vector_clipf_rvv, zve32f
+func ff_vector_clipf_rvv, zve32f, zba
 NOHWF   fmv.w.x  fa0, a3
 NOHWF   fmv.w.x  fa1, a4
 1:
diff --git a/libavcodec/riscv/bswapdsp_rvb.S b/libavcodec/riscv/bswapdsp_rvb.S
index 0786bd3f36..17cfd5d7ef 100644
--- a/libavcodec/riscv/bswapdsp_rvb.S
+++ b/libavcodec/riscv/bswapdsp_rvb.S
@@ -23,7 +23,7 @@
 #include "libavutil/riscv/bswap_rvb.S"
 
 #if (__riscv_xlen >= 64)
-func ff_bswap32_buf_rvb, zbb
+func ff_bswap32_buf_rvb, zba, zbb
         bswap32_rvb a0, a1, a2
 endfunc
 #endif
diff --git a/libavcodec/riscv/bswapdsp_rvv.S b/libavcodec/riscv/bswapdsp_rvv.S
index b37fe26255..14484a772d 100644
--- a/libavcodec/riscv/bswapdsp_rvv.S
+++ b/libavcodec/riscv/bswapdsp_rvv.S
@@ -21,7 +21,7 @@
 #include "config.h"
 #include "libavutil/riscv/asm.S"
 
-func ff_bswap16_buf_rvv, zve32x
+func ff_bswap16_buf_rvv, zve32x, zba
 1:
         vsetvli t0, a2, e16, m8, ta, ma
         vle16.v v8, (a1)
diff --git a/libavcodec/riscv/exrdsp_rvv.S b/libavcodec/riscv/exrdsp_rvv.S
index f4a35f58ff..c1d7dfcb86 100644
--- a/libavcodec/riscv/exrdsp_rvv.S
+++ b/libavcodec/riscv/exrdsp_rvv.S
@@ -20,7 +20,7 @@
 
 #include "libavutil/riscv/asm.S"
 
-func ff_reorder_pixels_rvv, zve32x
+func ff_reorder_pixels_rvv, zve32x, zba
         srai    a2, a2, 1
         add     t1, a1, a2
 1:
diff --git a/libavcodec/riscv/fixed_vtype.S b/libavcodec/riscv/fixed_vtype.S
new file mode 100644
index 0000000000..7aac70bda8
--- /dev/null
+++ b/libavcodec/riscv/fixed_vtype.S
@@ -0,0 +1,48 @@
+/*
+ * Copyright © 2024 Rémi Denis-Courmont.
+ *
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg 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.
+ *
+ * FFmpeg 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 FFmpeg; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#define VILL -0x8000
+
+        .data
+ff_vtype_e8:
+ff_vtype_e8_1:
+        .half   VILL
+ff_vtype_e8_2:
+        .half   VILL
+ff_vtype_e8_4:
+        .half   VILL
+ff_vtype_e8_8:
+        .half   VILL
+ff_vtype_e8_16:
+        .half   VILL
+ff_vtype_e8_32:
+        .half   VILL
+
+ff_vtype_e16:
+ff_vtype_e16_1:
+        .half   VILL
+ff_vtype_e16_2:
+        .half   VILL
+ff_vtype_e16_4:
+        .half   VILL
+ff_vtype_e16_8:
+        .half   VILL
+ff_vtype_e16_16:
+        .half   VILL
diff --git a/libavcodec/riscv/flacdsp_rvv.S b/libavcodec/riscv/flacdsp_rvv.S
index 1724aee9d7..69505c694a 100644
--- a/libavcodec/riscv/flacdsp_rvv.S
+++ b/libavcodec/riscv/flacdsp_rvv.S
@@ -20,7 +20,7 @@
 
 #include "libavutil/riscv/asm.S"
 
-func ff_flac_lpc16_rvv, zve32x, zbb
+func ff_flac_lpc16_rvv, zve32x, b
         vtype_vli t0, a2, t2, e32, ta, ma
         vsetvl  zero, a2, t0
         vle32.v v8, (a1)
@@ -45,7 +45,7 @@ func ff_flac_lpc16_rvv, zve32x, zbb
 endfunc
 
 #if (__riscv_xlen == 64)
-func ff_flac_lpc32_rvv, zve64x
+func ff_flac_lpc32_rvv, zve64x, zba
         addi    t2, a2, -16
         ble     t2, zero, ff_flac_lpc32_rvv_simple
         vsetivli zero, 1, e64, m1, ta, ma
@@ -76,7 +76,7 @@ func ff_flac_lpc32_rvv, zve64x
         ret
 endfunc
 
-func ff_flac_lpc32_rvv_simple, zve64x, zbb
+func ff_flac_lpc32_rvv_simple, zve64x, b
         vtype_vli t3, a2, t1, e64, ta, ma
         vntypei t2, t3
         vsetvl  zero, a2, t3 // e64
@@ -104,7 +104,7 @@ func ff_flac_lpc32_rvv_simple, zve64x, zbb
         ret
 endfunc
 
-func ff_flac_lpc33_rvv, zve64x, zbb
+func ff_flac_lpc33_rvv, zve64x, b
         vtype_vli t0, a3, t1, e64, ta, ma
         vsetvl  zero, a3, t0
         vmv.s.x v0, zero
@@ -132,7 +132,7 @@ func ff_flac_lpc33_rvv, zve64x, zbb
 endfunc
 #endif
 
-func ff_flac_wasted32_rvv, zve32x
+func ff_flac_wasted32_rvv, zve32x, zba
 1:
         vsetvli t0, a2, e32, m8, ta, ma
         vle32.v v8, (a0)
@@ -145,7 +145,7 @@ func ff_flac_wasted32_rvv, zve32x
         ret
 endfunc
 
-func ff_flac_wasted33_rvv, zve64x
+func ff_flac_wasted33_rvv, zve64x, zba
         srli         t0, a2, 5
         li           t1, 1
         bnez         t0, 2f
@@ -177,7 +177,7 @@ func ff_flac_wasted33_rvv, zve64x
 endfunc
 
 #if (__riscv_xlen == 64)
-func ff_flac_decorrelate_indep2_16_rvv, zve32x
+func ff_flac_decorrelate_indep2_16_rvv, zve32x, zba
         ld      a0,  (a0)
         ld      a2, 8(a1)
         ld      a1,  (a1)
@@ -200,7 +200,7 @@ func ff_flac_decorrelate_indep2_16_rvv, zve32x
         ret
 endfunc
 
-func ff_flac_decorrelate_indep4_16_rvv, zve32x
+func ff_flac_decorrelate_indep4_16_rvv, zve32x, zba
         ld      a0,   (a0)
         ld      a2,  8(a1)
         ld      t1, 16(a1)
@@ -233,7 +233,7 @@ func ff_flac_decorrelate_indep4_16_rvv, zve32x
         ret
 endfunc
 
-func ff_flac_decorrelate_indep6_16_rvv, zve32x
+func ff_flac_decorrelate_indep6_16_rvv, zve32x, zba
         ld      a0,   (a0)
         ld      a2,  8(a1)
         ld      t1, 16(a1)
@@ -278,7 +278,7 @@ func ff_flac_decorrelate_indep6_16_rvv, zve32x
         ret
 endfunc
 
-func ff_flac_decorrelate_indep8_16_rvv, zve32x
+func ff_flac_decorrelate_indep8_16_rvv, zve32x, zba
         ld      a0,   (a0)
         ld      a2,  8(a1)
         ld      t1, 16(a1)
@@ -332,9 +332,7 @@ func ff_flac_decorrelate_indep8_16_rvv, zve32x
         ret
 endfunc
 
-
-
-func ff_flac_decorrelate_ls_16_rvv, zve32x
+func ff_flac_decorrelate_ls_16_rvv, zve32x, zba
         ld      a0,  (a0)
         ld      a2, 8(a1)
         ld      a1,  (a1)
@@ -358,7 +356,7 @@ func ff_flac_decorrelate_ls_16_rvv, zve32x
         ret
 endfunc
 
-func ff_flac_decorrelate_rs_16_rvv, zve32x
+func ff_flac_decorrelate_rs_16_rvv, zve32x, zba
         ld      a0,  (a0)
         ld      a2, 8(a1)
         ld      a1,  (a1)
@@ -382,7 +380,7 @@ func ff_flac_decorrelate_rs_16_rvv, zve32x
         ret
 endfunc
 
-func ff_flac_decorrelate_ms_16_rvv, zve32x
+func ff_flac_decorrelate_ms_16_rvv, zve32x, zba
         ld      a0,  (a0)
         ld      a2, 8(a1)
         ld      a1,  (a1)
@@ -408,7 +406,7 @@ func ff_flac_decorrelate_ms_16_rvv, zve32x
         ret
 endfunc
 
-func ff_flac_decorrelate_indep2_32_rvv, zve32x
+func ff_flac_decorrelate_indep2_32_rvv, zve32x, zba
         ld      a0,  (a0)
         ld      a2, 8(a1)
         ld      a1,  (a1)
@@ -428,7 +426,7 @@ func ff_flac_decorrelate_indep2_32_rvv, zve32x
         ret
 endfunc
 
-func ff_flac_decorrelate_indep4_32_rvv, zve32x
+func ff_flac_decorrelate_indep4_32_rvv, zve32x, zba
         ld      a0,   (a0)
         ld      a2,  8(a1)
         ld      t1, 16(a1)
@@ -457,7 +455,7 @@ func ff_flac_decorrelate_indep4_32_rvv, zve32x
         ret
 endfunc
 
-func ff_flac_decorrelate_indep6_32_rvv, zve32x
+func ff_flac_decorrelate_indep6_32_rvv, zve32x, zba
         ld      a0,   (a0)
         ld      a2,  8(a1)
         ld      t1, 16(a1)
@@ -495,7 +493,7 @@ func ff_flac_decorrelate_indep6_32_rvv, zve32x
         ret
 endfunc
 
-func ff_flac_decorrelate_indep8_32_rvv, zve32x
+func ff_flac_decorrelate_indep8_32_rvv, zve32x, zba
         ld      a0,   (a0)
         ld      a2,  8(a1)
         ld      t1, 16(a1)
@@ -540,7 +538,7 @@ func ff_flac_decorrelate_indep8_32_rvv, zve32x
         ret
 endfunc
 
-func ff_flac_decorrelate_ls_32_rvv, zve32x
+func ff_flac_decorrelate_ls_32_rvv, zve32x, zba
         ld      a0,  (a0)
         ld      a2, 8(a1)
         ld      a1,  (a1)
@@ -561,7 +559,7 @@ func ff_flac_decorrelate_ls_32_rvv, zve32x
         ret
 endfunc
 
-func ff_flac_decorrelate_rs_32_rvv, zve32x
+func ff_flac_decorrelate_rs_32_rvv, zve32x, zba
         ld      a0,  (a0)
         ld      a2, 8(a1)
         ld      a1,  (a1)
@@ -582,7 +580,7 @@ func ff_flac_decorrelate_rs_32_rvv, zve32x
         ret
 endfunc
 
-func ff_flac_decorrelate_ms_32_rvv, zve32x
+func ff_flac_decorrelate_ms_32_rvv, zve32x, zba
         ld      a0,  (a0)
         ld      a2, 8(a1)
         ld      a1,  (a1)
diff --git a/libavcodec/riscv/fmtconvert_rvv.S b/libavcodec/riscv/fmtconvert_rvv.S
index d0e2f106d5..05cd3b38a5 100644
--- a/libavcodec/riscv/fmtconvert_rvv.S
+++ b/libavcodec/riscv/fmtconvert_rvv.S
@@ -20,7 +20,7 @@
 
 #include "libavutil/riscv/asm.S"
 
-func ff_int32_to_float_fmul_scalar_rvv, zve32f
+func ff_int32_to_float_fmul_scalar_rvv, zve32f, zba
 NOHWF   fmv.w.x     fa0, a2
 NOHWF   mv          a2, a3
 1:
@@ -37,7 +37,7 @@ NOHWF   mv          a2, a3
         ret
 endfunc
 
-func ff_int32_to_float_fmul_array8_rvv, zve32f
+func ff_int32_to_float_fmul_array8_rvv, zve32f, zba
         srai        a4, a4, 3
 
 1:      vsetvli     t0, a4, e32, m1, ta, ma
diff --git a/libavcodec/riscv/h264_mc_chroma.S b/libavcodec/riscv/h264_mc_chroma.S
index ce99bda44d..b6c0e1c635 100644
--- a/libavcodec/riscv/h264_mc_chroma.S
+++ b/libavcodec/riscv/h264_mc_chroma.S
@@ -325,7 +325,7 @@
         ret
 .endm
 
-func h264_put_chroma_mc_rvv, zve32x
+func h264_put_chroma_mc_rvv, zve32x, zba
 11:
         li      a7, 3
         blt     a3, a7, 12f
@@ -334,7 +334,7 @@ func h264_put_chroma_mc_rvv, zve32x
         do_chroma_mc put 0
 endfunc
 
-func h264_avg_chroma_mc_rvv, zve32x
+func h264_avg_chroma_mc_rvv, zve32x, zba
 21:
         li      a7, 3
         blt     a3, a7, 22f
diff --git a/libavcodec/riscv/h264idct_rvv.S b/libavcodec/riscv/h264idct_rvv.S
index c74ea18c19..514c849bce 100644
--- a/libavcodec/riscv/h264idct_rvv.S
+++ b/libavcodec/riscv/h264idct_rvv.S
@@ -298,7 +298,7 @@ func ff_h264_idct8_add_8_rvv, zve32x
         ret
 endfunc
 
-func ff_h264_idct8_add_16_rvv, zve32x
+func ff_h264_idct8_add_16_rvv, zve32x, zba
         csrwi   vxrm, 0
 .Lidct8_add_16_rvv:
         li      a4, 8
diff --git a/libavcodec/riscv/huffyuvdsp_rvv.S b/libavcodec/riscv/huffyuvdsp_rvv.S
index d334f5c6d0..54d1d94059 100644
--- a/libavcodec/riscv/huffyuvdsp_rvv.S
+++ b/libavcodec/riscv/huffyuvdsp_rvv.S
@@ -20,7 +20,7 @@
 
 #include "libavutil/riscv/asm.S"
 
-func ff_add_int16_rvv, zve32x
+func ff_add_int16_rvv, zve32x, zba
 1:
         vsetvli t0, a3, e16, m8, ta, ma
         vle16.v v16, (a0)
@@ -36,7 +36,7 @@ func ff_add_int16_rvv, zve32x
         ret
 endfunc
 
-func ff_add_hfyu_left_pred_bgr32_rvv, zve32x, zbb
+func ff_add_hfyu_left_pred_bgr32_rvv, zve32x, b
         vtype_ivli t1, 4, e8, ta, ma
         li      t0, 4
         vsetvl  zero, t0, t1
diff --git a/libavcodec/riscv/jpeg2000dsp_rvv.S b/libavcodec/riscv/jpeg2000dsp_rvv.S
index 10efe6b0db..77c6fd2d32 100644
--- a/libavcodec/riscv/jpeg2000dsp_rvv.S
+++ b/libavcodec/riscv/jpeg2000dsp_rvv.S
@@ -20,7 +20,7 @@
 
 #include "libavutil/riscv/asm.S"
 
-func ff_ict_float_rvv, zve32f
+func ff_ict_float_rvv, zve32f, zba
         lla     t0, ff_jpeg2000_f_ict_params
         flw     ft0,  0(t0)
         flw     ft1,  4(t0)
@@ -48,7 +48,7 @@ func ff_ict_float_rvv, zve32f
         ret
 endfunc
 
-func ff_rct_int_rvv, zve32x
+func ff_rct_int_rvv, zve32x, zba
 1:
         vsetvli t0, a3, e32, m8, ta, ma
         vle32.v v16, (a1)
diff --git a/libavcodec/riscv/llauddsp_rvv.S b/libavcodec/riscv/llauddsp_rvv.S
index 5569864832..6af2e6a882 100644
--- a/libavcodec/riscv/llauddsp_rvv.S
+++ b/libavcodec/riscv/llauddsp_rvv.S
@@ -20,7 +20,7 @@
 
 #include "libavutil/riscv/asm.S"
 
-func ff_scalarproduct_and_madd_int16_rvv, zve32x
+func ff_scalarproduct_and_madd_int16_rvv, zve32x, zba
         vsetvli t0, zero, e32, m8, ta, ma
         vmv.v.x v0, zero
 1:
@@ -44,7 +44,7 @@ func ff_scalarproduct_and_madd_int16_rvv, zve32x
         ret
 endfunc
 
-func ff_scalarproduct_and_madd_int32_rvv, zve32x
+func ff_scalarproduct_and_madd_int32_rvv, zve32x, zba
         vsetvli t0, zero, e32, m8, ta, ma
         vmv.v.x v0, zero
 1:
diff --git a/libavcodec/riscv/lpc_rvv.S b/libavcodec/riscv/lpc_rvv.S
index fe80305d9a..8fd1ddbbf0 100644
--- a/libavcodec/riscv/lpc_rvv.S
+++ b/libavcodec/riscv/lpc_rvv.S
@@ -21,7 +21,7 @@
 #include "libavutil/riscv/asm.S"
 
 #if __riscv_xlen >= 64
-func ff_lpc_apply_welch_window_rvv, zve64d
+func ff_lpc_apply_welch_window_rvv, zve64d, zba
         vsetvli   t0, zero, e64, m8, ta, ma
         vid.v     v0
         addi      t2, a1, -1
@@ -86,7 +86,7 @@ func ff_lpc_apply_welch_window_rvv, zve64d
         ret
 endfunc
 
-func ff_lpc_compute_autocorr_rvv, zve64d, zbb
+func ff_lpc_compute_autocorr_rvv, zve64d, b
         vtype_vli t1, a2, t2, e64, ta, ma, 1
         addi      a2, a2, 1
         li        t0, 1
diff --git a/libavcodec/riscv/opusdsp_rvv.S b/libavcodec/riscv/opusdsp_rvv.S
index 42d845a370..389dc744f5 100644
--- a/libavcodec/riscv/opusdsp_rvv.S
+++ b/libavcodec/riscv/opusdsp_rvv.S
@@ -20,7 +20,7 @@
 
 #include "libavutil/riscv/asm.S"
 
-func ff_opus_postfilter_rvv, zve32f, zbb
+func ff_opus_postfilter_rvv, zve32f, b
         flw     fa0, 0(a2) // g0
         slli    t1, a1, 2
         flw     fa1, 4(a2) // g1
diff --git a/libavcodec/riscv/rv40dsp_rvv.S b/libavcodec/riscv/rv40dsp_rvv.S
index e49345ef70..53d3d1d6f9 100644
--- a/libavcodec/riscv/rv40dsp_rvv.S
+++ b/libavcodec/riscv/rv40dsp_rvv.S
@@ -332,7 +332,7 @@
         ret
 .endm
 
-func ff_put_rv40_chroma_mc_rvv, zve32x
+func ff_put_rv40_chroma_mc_rvv, zve32x, zba
 11:
         li      a7, 3
         blt     a3, a7, 12f
@@ -341,7 +341,7 @@ func ff_put_rv40_chroma_mc_rvv, zve32x
         do_chroma_mc put 0
 endfunc
 
-func ff_avg_rv40_chroma_mc_rvv, zve32x
+func ff_avg_rv40_chroma_mc_rvv, zve32x, zba
 21:
         li      a7, 3
         blt     a3, a7, 22f
diff --git a/libavcodec/riscv/sbrdsp_rvv.S b/libavcodec/riscv/sbrdsp_rvv.S
index 331b88022c..7c90a8addf 100644
--- a/libavcodec/riscv/sbrdsp_rvv.S
+++ b/libavcodec/riscv/sbrdsp_rvv.S
@@ -20,7 +20,7 @@
 
 #include "libavutil/riscv/asm.S"
 
-func ff_sbr_sum64x5_rvv, zve32f
+func ff_sbr_sum64x5_rvv, zve32f, zba
         li      a5, 64
         addi    a1, a0, 64 * 4
         addi    a2, a0, 128 * 4
@@ -49,7 +49,7 @@ func ff_sbr_sum64x5_rvv, zve32f
         ret
 endfunc
 
-func ff_sbr_sum_square_rvv, zve32f
+func ff_sbr_sum_square_rvv, zve32f, zba
         vsetvli t0, zero, e32, m8, ta, ma
         slli    a1, a1, 1
         vmv.v.x v8, zero
@@ -157,7 +157,7 @@ func ff_sbr_autocorrelate_rvv, zve32f
         ret
 endfunc
 
-func ff_sbr_hf_gen_rvv, zve32f
+func ff_sbr_hf_gen_rvv, zve32f, zba
 NOHWF   fmv.w.x fa0, a4
 NOHWF   mv      a4, a5
 NOHWF   mv      a5, a6
@@ -207,7 +207,7 @@ NOHWF   mv      a5, a6
         ret
 endfunc
 
-func ff_sbr_hf_g_filt_rvv, zve32f
+func ff_sbr_hf_g_filt_rvv, zve32f, zba
         li      t1, 40 * 2 * 4
         sh3add  a1, a4, a1
 1:
@@ -272,16 +272,16 @@ endfunc
         ret
 .endm
 
-func ff_sbr_hf_apply_noise_0_rvv, zve32f, zbb
+func ff_sbr_hf_apply_noise_0_rvv, zve32f, b
         hf_apply_noise 0
 endfunc
 
-func ff_sbr_hf_apply_noise_3_rvv, zve32f, zbb
+func ff_sbr_hf_apply_noise_3_rvv, zve32f, b
        not     a4, a4 // invert parity of kx
        // fall through
 endfunc
 
-func ff_sbr_hf_apply_noise_1_rvv, zve32f, zbb
+func ff_sbr_hf_apply_noise_1_rvv, zve32f, b
         vsetvli t0, zero, e32, m4, ta, ma
         vid.v   v4
         vxor.vx v4, v4, a4
@@ -289,6 +289,6 @@ func ff_sbr_hf_apply_noise_1_rvv, zve32f, zbb
         hf_apply_noise 1
 endfunc
 
-func ff_sbr_hf_apply_noise_2_rvv, zve32f, zbb
+func ff_sbr_hf_apply_noise_2_rvv, zve32f, b
         hf_apply_noise 2
 endfunc
diff --git a/libavcodec/riscv/svqenc_rvv.S b/libavcodec/riscv/svqenc_rvv.S
index cfc27154dd..8b7a8b0400 100644
--- a/libavcodec/riscv/svqenc_rvv.S
+++ b/libavcodec/riscv/svqenc_rvv.S
@@ -20,7 +20,7 @@
 
 #include "libavutil/riscv/asm.S"
 
-func ff_ssd_int8_vs_int16_rvv, zve32x
+func ff_ssd_int8_vs_int16_rvv, zve32x, zba
         vsetvli      t0, zero, e32, m8, ta, ma
         vmv.v.x      v24, zero
 1:
diff --git a/libavcodec/riscv/takdsp_rvv.S b/libavcodec/riscv/takdsp_rvv.S
index fa942a3be6..f3a230ccec 100644
--- a/libavcodec/riscv/takdsp_rvv.S
+++ b/libavcodec/riscv/takdsp_rvv.S
@@ -21,7 +21,7 @@
 
 #include "libavutil/riscv/asm.S"
 
-func ff_decorrelate_ls_rvv, zve32x
+func ff_decorrelate_ls_rvv, zve32x, zba
 1:
         vsetvli  t0, a2, e32, m8, ta, ma
         sub      a2, a2, t0
@@ -35,7 +35,7 @@ func ff_decorrelate_ls_rvv, zve32x
         ret
 endfunc
 
-func ff_decorrelate_sr_rvv, zve32x
+func ff_decorrelate_sr_rvv, zve32x, zba
 1:
         vsetvli  t0, a2, e32, m8, ta, ma
         vle32.v  v0, (a0)
@@ -49,7 +49,7 @@ func ff_decorrelate_sr_rvv, zve32x
         ret
 endfunc
 
-func ff_decorrelate_sm_rvv, zve32x
+func ff_decorrelate_sm_rvv, zve32x, zba
 1:
         vsetvli  t0, a2, e32, m8, ta, ma
         vle32.v  v8, (a1)
@@ -67,7 +67,7 @@ func ff_decorrelate_sm_rvv, zve32x
         ret
 endfunc
 
-func ff_decorrelate_sf_rvv, zve32x
+func ff_decorrelate_sf_rvv, zve32x, zba
         csrwi    vxrm, 0
 1:
         vsetvli  t0, a2, e32, m8, ta, ma
diff --git a/libavcodec/riscv/utvideodsp_rvv.S b/libavcodec/riscv/utvideodsp_rvv.S
index fa70d0eb34..5e833eeb3c 100644
--- a/libavcodec/riscv/utvideodsp_rvv.S
+++ b/libavcodec/riscv/utvideodsp_rvv.S
@@ -20,7 +20,7 @@
 
 #include "libavutil/riscv/asm.S"
 
-func ff_restore_rgb_planes_rvv, zve32x
+func ff_restore_rgb_planes_rvv, zve32x, zba
         li      t1, -0x80
         sub     a3, a3, a6
         sub     a4, a4, a6
@@ -52,7 +52,7 @@ func ff_restore_rgb_planes_rvv, zve32x
         ret
 endfunc
 
-func ff_restore_rgb_planes10_rvv, zve32x
+func ff_restore_rgb_planes10_rvv, zve32x, zba
         li      t1, -0x200
         li      t2, 0x3FF
         sub     a3, a3, a6
diff --git a/libavcodec/riscv/vc1dsp_rvv.S b/libavcodec/riscv/vc1dsp_rvv.S
index d8b62579aa..aede87ccc0 100644
--- a/libavcodec/riscv/vc1dsp_rvv.S
+++ b/libavcodec/riscv/vc1dsp_rvv.S
@@ -21,7 +21,7 @@
 
 #include "libavutil/riscv/asm.S"
 
-func ff_vc1_inv_trans_8x8_dc_rvv, zve64x
+func ff_vc1_inv_trans_8x8_dc_rvv, zve64x, zba
         lh            t2, (a2)
         vsetivli      zero, 8, e8, mf2, ta, ma
         vlse64.v      v0, (a0), a1
@@ -43,7 +43,7 @@ func ff_vc1_inv_trans_8x8_dc_rvv, zve64x
         ret
 endfunc
 
-func ff_vc1_inv_trans_4x8_dc_rvv, zve32x
+func ff_vc1_inv_trans_4x8_dc_rvv, zve32x, zba
         lh            t2, (a2)
         vsetivli      zero, 8, e8, mf2, ta, ma
         vlse32.v      v0, (a0), a1
@@ -67,7 +67,7 @@ func ff_vc1_inv_trans_4x8_dc_rvv, zve32x
         ret
 endfunc
 
-func ff_vc1_inv_trans_8x4_dc_rvv, zve64x
+func ff_vc1_inv_trans_8x4_dc_rvv, zve64x, zba
         lh            t2, (a2)
         vsetivli      zero, 4, e8, mf4, ta, ma
         vlse64.v      v0, (a0), a1
diff --git a/libavcodec/riscv/vorbisdsp_rvv.S b/libavcodec/riscv/vorbisdsp_rvv.S
index 81a6c62a65..d136188d2e 100644
--- a/libavcodec/riscv/vorbisdsp_rvv.S
+++ b/libavcodec/riscv/vorbisdsp_rvv.S
@@ -20,7 +20,7 @@
 
 #include "libavutil/riscv/asm.S"
 
-func ff_vorbis_inverse_coupling_rvv, zve32f
+func ff_vorbis_inverse_coupling_rvv, zve32f, zba
         fmv.w.x   ft0, zero
 1:
         vsetvli   t0, a2, e32, m4, ta, ma
diff --git a/libavcodec/riscv/vp7dsp_rvv.S b/libavcodec/riscv/vp7dsp_rvv.S
index 856b0e8c96..bfcc220273 100644
--- a/libavcodec/riscv/vp7dsp_rvv.S
+++ b/libavcodec/riscv/vp7dsp_rvv.S
@@ -21,7 +21,7 @@
 #include "libavutil/riscv/asm.S"
 
 #if __riscv_xlen >= 64
-func ff_vp7_luma_dc_wht_rvv, zve32x
+func ff_vp7_luma_dc_wht_rvv, zve32x, zba
         li          a2, 4 * 16 * 2
         li          a7, 16 * 2
         jal         t0, 1f
diff --git a/libavcodec/riscv/vp8dsp_rvv.S b/libavcodec/riscv/vp8dsp_rvv.S
index d366748a0a..4675f4c76b 100644
--- a/libavcodec/riscv/vp8dsp_rvv.S
+++ b/libavcodec/riscv/vp8dsp_rvv.S
@@ -440,7 +440,7 @@ endconst
 .endm
 
 .macro epel len size type
-func ff_put_vp8_epel\len\()_\type\()\size\()_rvv, zve32x
+func ff_put_vp8_epel\len\()_\type\()\size\()_rvv, zve32x, zba
         epel_filter     \size \type t
         vsetvlstatic8   \len
 1:
@@ -455,7 +455,7 @@ endfunc
 .endm
 
 .macro epel_hv len hsize vsize
-func ff_put_vp8_epel\len\()_h\hsize\()v\vsize\()_rvv, zve32x
+func ff_put_vp8_epel\len\()_h\hsize\()v\vsize\()_rvv, zve32x, zba
 #if __riscv_xlen == 64
         addi            sp, sp, -48
         .irp n,0,1,2,3,4,5
diff --git a/libavcodec/riscv/vp9_intra_rvi.S b/libavcodec/riscv/vp9_intra_rvi.S
index 16b6bdb25a..dadd4be194 100644
--- a/libavcodec/riscv/vp9_intra_rvi.S
+++ b/libavcodec/riscv/vp9_intra_rvi.S
@@ -21,7 +21,7 @@
 #include "libavutil/riscv/asm.S"
 
 #if __riscv_xlen >= 64
-func ff_v_32x32_rvi
+func ff_v_32x32_rvi, zba
         ld           t0, (a3)
         ld           t1, 8(a3)
         ld           t2, 16(a3)
@@ -42,7 +42,7 @@ func ff_v_32x32_rvi
         ret
 endfunc
 
-func ff_v_16x16_rvi
+func ff_v_16x16_rvi, zba
         ld           t0, (a3)
         ld           t1, 8(a3)
         .rept 8
@@ -57,7 +57,7 @@ func ff_v_16x16_rvi
         ret
 endfunc
 
-func ff_v_8x8_rvi
+func ff_v_8x8_rvi, zba
         ld           t0, (a3)
         .rept 4
         add          a7, a0, a1
-- 
2.45.2

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

[FFmpeg-devel] [PATCH 4/9] lavfi/riscv: require B or zba explicitly

[Rémi Denis-Courmont]

---
 libavfilter/riscv/af_afir_rvv.S | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/libavfilter/riscv/af_afir_rvv.S b/libavfilter/riscv/af_afir_rvv.S
index 04ec2e50d8..2107d97166 100644
--- a/libavfilter/riscv/af_afir_rvv.S
+++ b/libavfilter/riscv/af_afir_rvv.S
@@ -21,7 +21,7 @@
 #include "libavutil/riscv/asm.S"
 
 //  void ff_fcmul_add(float *sum, const float *t, const float *c, int len)
-func ff_fcmul_add_rvv, zve64f
+func ff_fcmul_add_rvv, zve64f, zba
         li          t1, 32
 1:
         vsetvli     t0, a3, e32, m4, ta, ma
-- 
2.45.2

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

[FFmpeg-devel] [PATCH 5/9] sws/riscv: require B or zba explicitly

[Rémi Denis-Courmont]

---
 libswscale/riscv/input_rvv.S   | 12 ++++++------
 libswscale/riscv/range_rvv.S   |  8 ++++----
 libswscale/riscv/rgb2rgb_rvb.S |  2 +-
 libswscale/riscv/rgb2rgb_rvv.S | 12 ++++++------
 4 files changed, 17 insertions(+), 17 deletions(-)

diff --git a/libswscale/riscv/input_rvv.S b/libswscale/riscv/input_rvv.S
index 1d7de59c66..d07db43b55 100644
--- a/libswscale/riscv/input_rvv.S
+++ b/libswscale/riscv/input_rvv.S
@@ -26,7 +26,7 @@ func ff_bgr24ToY_rvv, zve32x
         j       1f
 endfunc
 
-func ff_rgb24ToY_rvv, zve32x
+func ff_rgb24ToY_rvv, zve32x, zba
         lw      t1, 0(a5) # RY
         lw      t3, 8(a5) # BY
 1:
@@ -62,7 +62,7 @@ func ff_bgr24ToUV_rvv, zve32x
         j       1f
 endfunc
 
-func ff_rgb24ToUV_rvv, zve32x
+func ff_rgb24ToUV_rvv, zve32x, zba
         lw      t1, 12(a6) # RU
         lw      t4, 24(a6) # RV
         lw      t3, 20(a6) # BU
@@ -108,7 +108,7 @@ func ff_bgr24ToUV_half_rvv, zve32x
         j       1f
 endfunc
 
-func ff_rgb24ToUV_half_rvv, zve32x
+func ff_rgb24ToUV_half_rvv, zve32x, zba
         lw      t1, 12(a6) # RU
         lw      t4, 24(a6) # RV
         lw      t3, 20(a6) # BU
@@ -157,7 +157,7 @@ func ff_\chr1\()ToY_rvv, zve32x
         j       1f
 endfunc
 
-func ff_\chr0\()ToY_rvv, zve32x
+func ff_\chr0\()ToY_rvv, zve32x, zba
         lw      t1, 0(a5) # RY
         lw      t3, 8(a5) # BY
 1:
@@ -199,7 +199,7 @@ func ff_\chr1\()ToUV_rvv, zve32x
         j       1f
 endfunc
 
-func ff_\chr0\()ToUV_rvv, zve32x
+func ff_\chr0\()ToUV_rvv, zve32x, zba
         lw      t1, 12(a6) # RU
         lw      t4, 24(a6) # RV
         lw      t3, 20(a6) # BU
@@ -251,7 +251,7 @@ func ff_\chr1\()ToUV_half_rvv, zve32x
         j       1f
 endfunc
 
-func ff_\chr0\()ToUV_half_rvv, zve32x
+func ff_\chr0\()ToUV_half_rvv, zve32x, zba
         lw      t1, 12(a6) # RU
         lw      t4, 24(a6) # RV
         lw      t3, 20(a6) # BU
diff --git a/libswscale/riscv/range_rvv.S b/libswscale/riscv/range_rvv.S
index 9da80e6199..19a74eba79 100644
--- a/libswscale/riscv/range_rvv.S
+++ b/libswscale/riscv/range_rvv.S
@@ -20,7 +20,7 @@
 
 #include "libavutil/riscv/asm.S"
 
-func ff_range_lum_to_jpeg_16_rvv, zve32x
+func ff_range_lum_to_jpeg_16_rvv, zve32x, zba
         li       t1, 30189
         li       t2, 19077
         li       t3, -39057361
@@ -41,7 +41,7 @@ func ff_range_lum_to_jpeg_16_rvv, zve32x
         ret
 endfunc
 
-func ff_range_lum_from_jpeg_16_rvv, zve32x
+func ff_range_lum_from_jpeg_16_rvv, zve32x, zba
         li       t1, 14071
         li       t2, 33561947
 1:
@@ -60,7 +60,7 @@ func ff_range_lum_from_jpeg_16_rvv, zve32x
         ret
 endfunc
 
-func ff_range_chr_to_jpeg_16_rvv, zve32x
+func ff_range_chr_to_jpeg_16_rvv, zve32x, zba
         li      t1, 30775
         li      t2, 4663
         li      t3, -9289992
@@ -88,7 +88,7 @@ func ff_range_chr_to_jpeg_16_rvv, zve32x
         ret
 endfunc
 
-func ff_range_chr_from_jpeg_16_rvv, zve32x
+func ff_range_chr_from_jpeg_16_rvv, zve32x, zba
         li      t1, 1799
         li      t2, 4081085
 1:
diff --git a/libswscale/riscv/rgb2rgb_rvb.S b/libswscale/riscv/rgb2rgb_rvb.S
index af127b32ed..d18e5ba01b 100644
--- a/libswscale/riscv/rgb2rgb_rvb.S
+++ b/libswscale/riscv/rgb2rgb_rvb.S
@@ -23,7 +23,7 @@
 #include "libavutil/riscv/bswap_rvb.S"
 
 #if (__riscv_xlen >= 64)
-func ff_shuffle_bytes_3210_rvb, zbb
+func ff_shuffle_bytes_3210_rvb, zba, zbb
         srli    a2, a2, 2
         bswap32_rvb a1, a0, a2
 endfunc
diff --git a/libswscale/riscv/rgb2rgb_rvv.S b/libswscale/riscv/rgb2rgb_rvv.S
index 19f7aaf67d..e1270ac0df 100644
--- a/libswscale/riscv/rgb2rgb_rvv.S
+++ b/libswscale/riscv/rgb2rgb_rvv.S
@@ -25,7 +25,7 @@ func ff_shuffle_bytes_0321_rvv, zve32x
         j       1f
 endfunc
 
-func ff_shuffle_bytes_2103_rvv, zve32x
+func ff_shuffle_bytes_2103_rvv, zve32x, zba
         li      t1, ~0x00ff00ff
 1:
         not     t2, t1
@@ -54,7 +54,7 @@ func ff_shuffle_bytes_1230_rvv, zve32x
         j       3f
 endfunc
 
-func ff_shuffle_bytes_3012_rvv, zve32x
+func ff_shuffle_bytes_3012_rvv, zve32x, zba
         li      t1, 8
         li      t2, 24
 3:
@@ -74,7 +74,7 @@ func ff_shuffle_bytes_3012_rvv, zve32x
         ret
 endfunc
 
-func ff_interleave_bytes_rvv, zve32x
+func ff_interleave_bytes_rvv, zve32x, zba
 1:
         mv      t0, a0
         mv      t1, a1
@@ -100,7 +100,7 @@ func ff_interleave_bytes_rvv, zve32x
         ret
 endfunc
 
-func ff_deinterleave_bytes_rvv, zve32x
+func ff_deinterleave_bytes_rvv, zve32x, zba
 1:
         mv      t0, a0
         mv      t1, a1
@@ -165,10 +165,10 @@ endfunc
         ret
 .endm
 
-func ff_uyvytoyuv422_rvv, zve32x, zbb
+func ff_uyvytoyuv422_rvv, zve32x, b
         yuy2_to_i422p v20, v16
 endfunc
 
-func ff_yuyvtoyuv422_rvv, zve32x, zbb
+func ff_yuyvtoyuv422_rvv, zve32x, b
         yuy2_to_i422p v16, v20
 endfunc
-- 
2.45.2

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

[FFmpeg-devel] [PATCH 6/9] lavu/riscv: require B or zba explicitly

[Rémi Denis-Courmont]

---
 libavutil/riscv/fixed_dsp_rvv.S | 14 +++++++-------
 libavutil/riscv/float_dsp_rvv.S | 24 ++++++++++++------------
 2 files changed, 19 insertions(+), 19 deletions(-)

diff --git a/libavutil/riscv/fixed_dsp_rvv.S b/libavutil/riscv/fixed_dsp_rvv.S
index 6bac5813b8..0fa6aab3d4 100644
--- a/libavutil/riscv/fixed_dsp_rvv.S
+++ b/libavutil/riscv/fixed_dsp_rvv.S
@@ -20,7 +20,7 @@
 
 #include "asm.S"
 
-func ff_vector_fmul_window_scaled_rvv, zve64x
+func ff_vector_fmul_window_scaled_rvv, zve64x, zba
         csrwi   vxrm, 0
         vsetvli t0, zero, e16, m1, ta, ma
         sh2add  a2, a4, a2
@@ -68,7 +68,7 @@ func ff_vector_fmul_window_scaled_rvv, zve64x
         ret
 endfunc
 
-func ff_vector_fmul_window_fixed_rvv, zve64x
+func ff_vector_fmul_window_fixed_rvv, zve64x, zba
         csrwi   vxrm, 0
         vsetvli t0, zero, e16, m1, ta, ma
         sh2add  a2, a4, a2
@@ -112,7 +112,7 @@ func ff_vector_fmul_window_fixed_rvv, zve64x
         ret
 endfunc
 
-func ff_vector_fmul_fixed_rvv, zve32x
+func ff_vector_fmul_fixed_rvv, zve32x, zba
         csrwi   vxrm, 0
 1:
         vsetvli t0, a3, e32, m4, ta, ma
@@ -129,7 +129,7 @@ func ff_vector_fmul_fixed_rvv, zve32x
         ret
 endfunc
 
-func ff_vector_fmul_reverse_fixed_rvv, zve32x
+func ff_vector_fmul_reverse_fixed_rvv, zve32x, zba
         csrwi   vxrm, 0
         // e16/m4 and e32/m8 are possible but slow the gathers down.
         vsetvli t0, zero, e16, m1, ta, ma
@@ -155,7 +155,7 @@ func ff_vector_fmul_reverse_fixed_rvv, zve32x
         ret
 endfunc
 
-func ff_vector_fmul_add_fixed_rvv, zve32x
+func ff_vector_fmul_add_fixed_rvv, zve32x, zba
         csrwi   vxrm, 0
 1:
         vsetvli t0, a4, e32, m8, ta, ma
@@ -175,7 +175,7 @@ func ff_vector_fmul_add_fixed_rvv, zve32x
         ret
 endfunc
 
-func ff_scalarproduct_fixed_rvv, zve64x
+func ff_scalarproduct_fixed_rvv, zve64x, zba
         li      t1, 1 << 30
         vsetvli t0, zero, e64, m8, ta, ma
         vmv.v.x v8, zero
@@ -198,7 +198,7 @@ func ff_scalarproduct_fixed_rvv, zve64x
 endfunc
 
 // (a0) = (a0) + (a1), (a1) = (a0) - (a1) [0..a2-1]
-func ff_butterflies_fixed_rvv, zve32x
+func ff_butterflies_fixed_rvv, zve32x, zba
 1:
         vsetvli t0, a2, e32, m4, ta, ma
         vle32.v v16, (a0)
diff --git a/libavutil/riscv/float_dsp_rvv.S b/libavutil/riscv/float_dsp_rvv.S
index 2f0ade6db6..c7744cf0e8 100644
--- a/libavutil/riscv/float_dsp_rvv.S
+++ b/libavutil/riscv/float_dsp_rvv.S
@@ -21,7 +21,7 @@
 #include "asm.S"
 
 // (a0) = (a1) * (a2) [0..a3-1]
-func ff_vector_fmul_rvv, zve32f
+func ff_vector_fmul_rvv, zve32f, zba
 1:
         vsetvli  t0, a3, e32, m8, ta, ma
         vle32.v  v16, (a1)
@@ -38,7 +38,7 @@ func ff_vector_fmul_rvv, zve32f
 endfunc
 
 // (a0) += (a1) * fa0 [0..a2-1]
-func ff_vector_fmac_scalar_rvv, zve32f
+func ff_vector_fmac_scalar_rvv, zve32f, zba
 NOHWF   fmv.w.x   fa0, a2
 NOHWF   mv        a2, a3
 1:
@@ -57,7 +57,7 @@ NOHWF   mv        a2, a3
 endfunc
 
 // (a0) = (a1) * fa0 [0..a2-1]
-func ff_vector_fmul_scalar_rvv, zve32f
+func ff_vector_fmul_scalar_rvv, zve32f, zba
 NOHWF   fmv.w.x  fa0, a2
 NOHWF   mv       a2, a3
 1:
@@ -73,7 +73,7 @@ NOHWF   mv       a2, a3
         ret
 endfunc
 
-func ff_vector_fmul_window_rvv, zve32f
+func ff_vector_fmul_window_rvv, zve32f, zba
         // a0: dst, a1: src0, a2: src1, a3: window, a4: length
         // e16/m2 and e32/m4 are possible but slower due to gather.
         vsetvli    t0, zero, e16, m1, ta, ma
@@ -113,7 +113,7 @@ func ff_vector_fmul_window_rvv, zve32f
 endfunc
 
 // (a0) = (a1) * (a2) + (a3) [0..a4-1]
-func ff_vector_fmul_add_rvv, zve32f
+func ff_vector_fmul_add_rvv, zve32f, zba
 1:
         vsetvli   t0, a4, e32, m8, ta, ma
         vle32.v   v8, (a1)
@@ -133,7 +133,7 @@ endfunc
 
 // TODO factor vrsub, separate last iteration?
 // (a0) = (a1) * reverse(a2) [0..a3-1]
-func ff_vector_fmul_reverse_rvv, zve32f
+func ff_vector_fmul_reverse_rvv, zve32f, zba
         // e16/m4 and e32/m8 are possible but slower due to gather.
         vsetvli  t0, zero, e16, m1, ta, ma
         sh2add   a2, a3, a2
@@ -159,7 +159,7 @@ func ff_vector_fmul_reverse_rvv, zve32f
 endfunc
 
 // (a0) = (a0) + (a1), (a1) = (a0) - (a1) [0..a2-1]
-func ff_butterflies_float_rvv, zve32f
+func ff_butterflies_float_rvv, zve32f, zba
 1:
         vsetvli  t0, a2, e32, m8, ta, ma
         vle32.v  v16, (a0)
@@ -177,7 +177,7 @@ func ff_butterflies_float_rvv, zve32f
 endfunc
 
 // a0 = (a0).(a1) [0..a2-1]
-func ff_scalarproduct_float_rvv, zve32f
+func ff_scalarproduct_float_rvv, zve32f, zba
         vsetvli      t0, zero, e32, m8, ta, ma
         vmv.v.x      v8, zero
         vmv.s.x      v0, zero
@@ -199,7 +199,7 @@ NOHWF   fmv.x.w  a0, fa0
 endfunc
 
 // (a0) = (a1) * (a2) [0..a3-1]
-func ff_vector_dmul_rvv, zve64d
+func ff_vector_dmul_rvv, zve64d, zba
 1:
         vsetvli  t0, a3, e64, m8, ta, ma
         vle64.v  v16, (a1)
@@ -216,7 +216,7 @@ func ff_vector_dmul_rvv, zve64d
 endfunc
 
 // (a0) += (a1) * fa0 [0..a2-1]
-func ff_vector_dmac_scalar_rvv, zve64d
+func ff_vector_dmac_scalar_rvv, zve64d, zba
 NOHWD   fmv.d.x   fa0, a2
 NOHWD   mv        a2, a3
 1:
@@ -234,7 +234,7 @@ NOHWD   mv        a2, a3
 endfunc
 
 // (a0) = (a1) * fa0 [0..a2-1]
-func ff_vector_dmul_scalar_rvv, zve64d
+func ff_vector_dmul_scalar_rvv, zve64d, zba
 NOHWD   fmv.d.x  fa0, a2
 NOHWD   mv       a2, a3
 1:
@@ -250,7 +250,7 @@ NOHWD   mv       a2, a3
         ret
 endfunc
 
-func ff_scalarproduct_double_rvv, zve64f
+func ff_scalarproduct_double_rvv, zve64f, zba
         vsetvli      t0, zero, e64, m8, ta, ma
         vmv.v.x      v8, zero
         vmv.s.x      v0, zero
-- 
2.45.2

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

[FFmpeg-devel] [PATCH 7/9] lavu/riscv: remove bespoke SH{1, 2, 3}ADD assembler

[Rémi Denis-Courmont]

configure checks that the assembler supports the B extension (or rather
its constituents) anyway. These macros were dodging sanity checks for
unsupported instructions and nothing else.
---
 libavutil/riscv/asm.S | 19 -------------------
 1 file changed, 19 deletions(-)

diff --git a/libavutil/riscv/asm.S b/libavutil/riscv/asm.S
index 0c29680d84..8b96e07b75 100644
--- a/libavutil/riscv/asm.S
+++ b/libavutil/riscv/asm.S
@@ -83,25 +83,6 @@
             .endm
         .endm
 
-#if !defined (__riscv_zba)
-        /* SH{1,2,3}ADD definitions for pre-Zba assemblers */
-        .macro  shnadd n, rd, rs1, rs2
-        .insn r OP, 2 * \n, 16, \rd, \rs1, \rs2
-        .endm
-
-        .macro  sh1add rd, rs1, rs2
-        shnadd  1, \rd, \rs1, \rs2
-        .endm
-
-        .macro  sh2add rd, rs1, rs2
-        shnadd  2, \rd, \rs1, \rs2
-        .endm
-
-        .macro  sh3add rd, rs1, rs2
-        shnadd  3, \rd, \rs1, \rs2
-        .endm
-#endif
-
 #if defined (__riscv_v_elen)
 # define RV_V_ELEN __riscv_v_elen
 #else
-- 
2.45.2

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

[FFmpeg-devel] [PATCH 8/9] lavu/riscv: add CPU flag for B bit manipulations

[Rémi Denis-Courmont]

The B extension was finally ratified in May 2024, encompassing:
- Zba (addresses),
- Zbb (basics) and
- Zbs (single bits).
It does not include Zbc (base-2 polynomials).
---
 doc/APIchanges            |  3 +++
 libavutil/cpu.c           |  1 +
 libavutil/cpu.h           |  1 +
 libavutil/riscv/cpu.c     | 13 +++++++++++++
 libavutil/tests/cpu.c     |  1 +
 tests/checkasm/checkasm.c |  1 +
 6 files changed, 20 insertions(+)

diff --git a/doc/APIchanges b/doc/APIchanges
index 5751216b24..0061b084b8 100644
--- a/doc/APIchanges
+++ b/doc/APIchanges
@@ -2,6 +2,9 @@ The last version increases of all libraries were on 2024-03-07
 
 API changes, most recent first:
 
+2024-07-22 - xxxxxxxxx - lavu 59.18.100 - cpu.h
+  Add AV_CPU_FLAG_RVB.
+
 2024-07-xx - xxxxxxxxxx - lavf 61 - avformat.h
   Deprecate avformat_transfer_internal_stream_timing_info()
   and av_stream_get_codec_timebase() without replacement.
diff --git a/libavutil/cpu.c b/libavutil/cpu.c
index 9ac2f01c20..17afe8858a 100644
--- a/libavutil/cpu.c
+++ b/libavutil/cpu.c
@@ -186,6 +186,7 @@ int av_parse_cpu_caps(unsigned *flags, const char *s)
         { "rvi",      NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_RVI      },    .unit = "flags" },
         { "rvf",      NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_RVF      },    .unit = "flags" },
         { "rvd",      NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_RVD      },    .unit = "flags" },
+        { "rvb",      NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_RVB      },    .unit = "flags" },
         { "zve32x",   NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_RVV_I32  },    .unit = "flags" },
         { "zve32f",   NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_RVV_F32  },    .unit = "flags" },
         { "zve64x",   NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_RVV_I64  },    .unit = "flags" },
diff --git a/libavutil/cpu.h b/libavutil/cpu.h
index a25901433e..9f419aae02 100644
--- a/libavutil/cpu.h
+++ b/libavutil/cpu.h
@@ -92,6 +92,7 @@
 #define AV_CPU_FLAG_RVB_ADDR     (1 << 8) ///< Address bit-manipulations
 #define AV_CPU_FLAG_RV_ZVBB      (1 << 9) ///< Vector basic bit-manipulations
 #define AV_CPU_FLAG_RV_MISALIGNED (1 <<10) ///< Fast misaligned accesses
+#define AV_CPU_FLAG_RVB          (1 <<11) ///< B (bit manipulations)
 
 /**
  * Return the flags which specify extensions supported by the CPU.
diff --git a/libavutil/riscv/cpu.c b/libavutil/riscv/cpu.c
index 04ac404bbf..e035f4b024 100644
--- a/libavutil/riscv/cpu.c
+++ b/libavutil/riscv/cpu.c
@@ -72,6 +72,12 @@ int ff_get_cpu_flags_riscv(void)
 #ifdef RISCV_HWPROBE_EXT_ZBB
         if (pairs[1].value & RISCV_HWPROBE_EXT_ZBB)
             ret |= AV_CPU_FLAG_RVB_BASIC;
+#if defined (RISCV_HWPROBE_EXT_ZBA) && defined (RISCV_HWPROBE_EXT_ZBS)
+        if ((pairs[1].value & RISCV_HWPROBE_EXT_ZBA) &&
+            (pairs[1].value & RISCV_HWPROBE_EXT_ZBB) &&
+            (pairs[1].value & RISCV_HWPROBE_EXT_ZBS))
+            ret |= AV_CPU_FLAG_RVB;
+#endif
 #endif
 #ifdef RISCV_HWPROBE_EXT_ZVBB
         if (pairs[1].value & RISCV_HWPROBE_EXT_ZVBB)
@@ -94,6 +100,9 @@ int ff_get_cpu_flags_riscv(void)
             ret |= AV_CPU_FLAG_RVF;
         if (hwcap & HWCAP_RV('D'))
             ret |= AV_CPU_FLAG_RVD;
+        if (hwcap & HWCAP_RV('B'))
+            ret |= AV_CPU_FLAG_RVB_ADDR | AV_CPU_FLAG_RVB_BASIC |
+                   AV_CPU_FLAG_RVB;
 
         /* The V extension implies all Zve* functional subsets */
         if (hwcap & HWCAP_RV('V'))
@@ -118,6 +127,10 @@ int ff_get_cpu_flags_riscv(void)
 #ifdef __riscv_zbb
     ret |= AV_CPU_FLAG_RVB_BASIC;
 #endif
+#if defined (__riscv_b) || \
+    (defined (__riscv_zba) && defined (__riscv_zbb) && defined (__riscv_zbs))
+    ret |= AV_CPU_FLAG_RVB;
+#endif
 
     /* If RV-V is enabled statically at compile-time, check the details. */
 #ifdef __riscv_vector
diff --git a/libavutil/tests/cpu.c b/libavutil/tests/cpu.c
index 02b98682e3..b4b11775d8 100644
--- a/libavutil/tests/cpu.c
+++ b/libavutil/tests/cpu.c
@@ -90,6 +90,7 @@ static const struct {
     { AV_CPU_FLAG_RVD,       "rvd"        },
     { AV_CPU_FLAG_RVB_ADDR,  "zba"        },
     { AV_CPU_FLAG_RVB_BASIC, "zbb"        },
+    { AV_CPU_FLAG_RVB,       "rvb"        },
     { AV_CPU_FLAG_RVV_I32,   "zve32x"     },
     { AV_CPU_FLAG_RVV_F32,   "zve32f"     },
     { AV_CPU_FLAG_RVV_I64,   "zve64x"     },
diff --git a/tests/checkasm/checkasm.c b/tests/checkasm/checkasm.c
index de0024099a..016f2329b0 100644
--- a/tests/checkasm/checkasm.c
+++ b/tests/checkasm/checkasm.c
@@ -295,6 +295,7 @@ static const struct {
     { "RVD",      "rvd",      AV_CPU_FLAG_RVD },
     { "RVBaddr",  "rvb_a",    AV_CPU_FLAG_RVB_ADDR },
     { "RVBbasic", "rvb_b",    AV_CPU_FLAG_RVB_BASIC },
+    { "RVB",      "rvb",      AV_CPU_FLAG_RVB },
     { "RVVi32",   "rvv_i32",  AV_CPU_FLAG_RVV_I32 },
     { "RVVf32",   "rvv_f32",  AV_CPU_FLAG_RVV_F32 },
     { "RVVi64",   "rvv_i64",  AV_CPU_FLAG_RVV_I64 },
-- 
2.45.2

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

[FFmpeg-devel] [PATCH 9/9] lavc/h264dsp: use RISC-V B extension

[Rémi Denis-Courmont]

This saves one register and one instruction per transform.
add16 and add16intra thus become stack-less.
---
 libavcodec/riscv/h264dsp_init.c | 25 ++++++++--------
 libavcodec/riscv/h264idct_rvv.S | 51 ++++++++++++++++-----------------
 2 files changed, 38 insertions(+), 38 deletions(-)

diff --git a/libavcodec/riscv/h264dsp_init.c b/libavcodec/riscv/h264dsp_init.c
index 9ae182151c..836c073559 100644
--- a/libavcodec/riscv/h264dsp_init.c
+++ b/libavcodec/riscv/h264dsp_init.c
@@ -98,13 +98,14 @@ av_cold void ff_h264dsp_init_riscv(H264DSPContext *dsp, const int bit_depth,
 
             dsp->h264_idct_add  = ff_h264_idct_add_8_rvv;
             dsp->h264_idct8_add = ff_h264_idct8_add_8_rvv;
+            dsp->h264_idct_dc_add = ff_h264_idct4_dc_add_8_rvv;
+            if (flags & AV_CPU_FLAG_RVB) {
+                dsp->h264_idct_add16      = ff_h264_idct_add16_8_rvv;
+                dsp->h264_idct_add16intra = ff_h264_idct_add16intra_8_rvv;
 #  if __riscv_xlen == 64
-            dsp->h264_idct_add16      = ff_h264_idct_add16_8_rvv;
-            dsp->h264_idct_add16intra = ff_h264_idct_add16intra_8_rvv;
-            dsp->h264_idct8_add4      = ff_h264_idct8_add4_8_rvv;
+                dsp->h264_idct8_add4      = ff_h264_idct8_add4_8_rvv;
 #  endif
-            if (flags & AV_CPU_FLAG_RVV_I32)
-                dsp->h264_idct_dc_add = ff_h264_idct4_dc_add_8_rvv;
+            }
             if (flags & AV_CPU_FLAG_RVV_I64) {
                 dsp->h264_add_pixels8_clear = ff_h264_add_pixels8_8_rvv;
                 dsp->h264_idct8_dc_add      = ff_h264_idct8_dc_add_8_rvv;
@@ -118,16 +119,16 @@ av_cold void ff_h264dsp_init_riscv(H264DSPContext *dsp, const int bit_depth,
                 dsp->h264_idct_add = ff_h264_idct_add_##depth##_rvv; \
             if (flags & AV_CPU_FLAG_RVB_ADDR) \
                 dsp->h264_idct8_add = ff_h264_idct8_add_##depth##_rvv; \
-            if (zvl128b && (flags & AV_CPU_FLAG_RVB_ADDR)) { \
+            if (zvl128b && (flags & AV_CPU_FLAG_RVB)) { \
                 dsp->h264_idct_dc_add  = ff_h264_idct4_dc_add_##depth##_rvv; \
                 dsp->h264_idct8_dc_add = ff_h264_idct8_dc_add_##depth##_rvv; \
+                if (__riscv_xlen == 64) { \
+                    dsp->h264_idct_add16 = ff_h264_idct_add16_##depth##_rvv; \
+                    dsp->h264_idct_add16intra = \
+                        ff_h264_idct_add16intra_##depth##_rvv; \
+                } \
             } \
-            if (__riscv_xlen == 64 && zvl128b) { \
-                dsp->h264_idct_add16 = ff_h264_idct_add16_##depth##_rvv; \
-                dsp->h264_idct_add16intra = \
-                    ff_h264_idct_add16intra_##depth##_rvv; \
-            } \
-            if (__riscv_xlen == 64 && (flags & AV_CPU_FLAG_RVB_ADDR)) \
+            if (__riscv_xlen == 64 && (flags & AV_CPU_FLAG_RVB)) \
                 dsp->h264_idct8_add4 = ff_h264_idct8_add4_##depth##_rvv; \
         }
 
diff --git a/libavcodec/riscv/h264idct_rvv.S b/libavcodec/riscv/h264idct_rvv.S
index 514c849bce..a49a32c47e 100644
--- a/libavcodec/riscv/h264idct_rvv.S
+++ b/libavcodec/riscv/h264idct_rvv.S
@@ -532,16 +532,11 @@ const ff_h264_scan8
         .byte   034, 035, 044, 045, 036, 037, 046, 047
 endconst
 
-#if (__riscv_xlen == 64)
 .macro  idct4_adds type, depth
-func ff_h264_idct_add\type\()_\depth\()_rvv, zve32x
+func ff_h264_idct_add\type\()_\depth\()_rvv, zve32x, b
         csrwi   vxrm, 0
-        addi    sp, sp, -16
         lla     t0, ff_h264_scan8
-        sd      s0,   (sp)
         li      t1, 32 * (\depth / 8)
-        mv      s0, sp
-        sd      ra,  8(sp)
         vsetivli  zero, 16, e8, m1, ta, ma
         vle8.v    v8, (t0)
 .if \depth == 8
@@ -567,20 +562,23 @@ func ff_h264_idct_add\type\()_\depth\()_rvv, zve32x
         vsetvli   zero, zero, e16, m2, ta, ma
         vmv.x.s   a4, v0
         vmv.x.s   a7, v1
+        zext.h  a4, a4
+        slli    a7, a7, 16
         mv      t4, a0
+        or      a4, a4, a7
         mv      t5, a1
         mv      a1, a2
         mv      a2, a3
         li      a3, 16
+        mv      a7, ra
 1:
         andi    t0, a4, 1
         addi    a3, a3, -1
-        srli    a4, a4, 1
 .ifc \type, 16
         beqz    t0, 3f     # if (nnz)
 .endif
         lw      t2, (t5)   # block_offset[i]
-        andi    t1, a7, 1
+        bexti   t1, a4, 16
         add     a0, t4, t2
 .ifc \type, 16
         bnez    t1, 2f     # if (nnz == 1 && block[i * 16])
@@ -595,14 +593,12 @@ func ff_h264_idct_add\type\()_\depth\()_rvv, zve32x
 .endif
         jal     ff_h264_idct4_dc_add_\depth\()_rvv
 3:
-        srli    a7, a7, 1
+        srli    a4, a4, 1
         addi    t5, t5, 4
         addi    a1, a1, 16 * 2 * (\depth / 8)
         bnez    a3, 1b
 
-        ld      ra,  8(sp)
-        ld      s0,  0(sp)
-        addi    sp, sp, 16
+        mv      ra, a7
         ret
 endfunc
 .endm
@@ -611,9 +607,10 @@ endfunc
 idct4_adds 16, \depth
 idct4_adds 16intra, \depth
 
-func ff_h264_idct8_add4_\depth\()_rvv, zve32x
+#if (__riscv_xlen == 64)
+func ff_h264_idct8_add4_\depth\()_rvv, zve32x, b
         csrwi       vxrm, 0
-        addi    sp, sp, -64
+        addi    sp, sp, -48
         lla     t0, ff_h264_scan8
         sd      s0,   (sp)
         li      t1, 4 * 32 * (\depth / 8)
@@ -622,9 +619,8 @@ func ff_h264_idct8_add4_\depth\()_rvv, zve32x
         sd      ra,  8(sp)
         sd      s1, 16(sp)
         sd      s2, 24(sp)
-        sd      s3, 32(sp)
-        sd      s4, 40(sp)
-        sd      s5, 48(sp)
+        sd      s4, 32(sp)
+        sd      s5, 40(sp)
         vsetivli  zero, 4, e8, mf4, ta, ma
         vlse8.v   v8, (t0), t2
 .if \depth == 8
@@ -644,8 +640,11 @@ func ff_h264_idct8_add4_\depth\()_rvv, zve32x
         vmsne.vi  v0, v12, 0
         vmand.mm  v1, v1, v2
         vmv.x.s   s2, v0
-        vmv.x.s   s3, v1
+        vmv.x.s   a7, v1
+        zext.h  s2, s2
+        slli    a7, a7, 16
         li      s1, 4
+        or      s2, s2, a7
         mv      s4, a0
         mv      s5, a1
         mv      a1, a2
@@ -653,10 +652,9 @@ func ff_h264_idct8_add4_\depth\()_rvv, zve32x
 1:
         andi    t0, s2, 1
         addi    s1, s1, -1
-        srli    s2, s2, 1
         beqz    t0, 3f     # if (nnz)
         lw      t2, (s5)   # block_offset[i]
-        andi    t1, s3, 1
+        bexti   t1, s2, 16
         add     a0, s4, t2
         bnez    t1, 2f    # if (nnz == 1 && block[i * 16])
         jal     .Lidct8_add_\depth\()_rvv
@@ -670,20 +668,20 @@ func ff_h264_idct8_add4_\depth\()_rvv, zve32x
 3:
         addi    a1, a1, 4 * 16 * 2 * (\depth / 8)
 4:
-        srli    s3, s3, 1
+        srli    s2, s2, 1
         addi    s5, s5, 4 * 4
         bnez    s1, 1b
 
-        ld      s5, 48(sp)
-        ld      s4, 40(sp)
-        ld      s3, 32(sp)
+        ld      s5, 40(sp)
+        ld      s4, 32(sp)
         ld      s2, 24(sp)
         ld      s1, 16(sp)
         ld      ra,  8(sp)
         ld      s0,  0(sp)
-        addi    sp, sp, 64
+        addi    sp, sp, 48
         ret
 endfunc
+#endif
 .endr
 
 .irp    depth, 9, 10, 12, 14
@@ -697,9 +695,10 @@ func ff_h264_idct_add16intra_\depth\()_rvv, zve32x
         j       ff_h264_idct_add16intra_16_rvv
 endfunc
 
+#if (__riscv_xlen == 64)
 func ff_h264_idct8_add4_\depth\()_rvv, zve32x
         li      a5, (1 << \depth) - 1
         j       ff_h264_idct8_add4_16_rvv
 endfunc
-.endr
 #endif
+.endr
-- 
2.45.2

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".