aarch64: Implement stack spilling in a consistent way.

Currently it is done in several different ways, which
might cause needless dependencies or in case of
tx_float_neon.S is incorrect.

Reviewed-by: Martin Storsjö <martin@martin.st>
Signed-off-by: Reimar Döffinger <Reimar.Doeffinger@gmx.de>

10 files changed, 109 insertions, 122 deletions

diff --git a/libavcodec/aarch64/fft_neon.S b/libavcodec/aarch64/fft_neon.S
index 9ff3f9c526..d7225511dd 100644
--- a/libavcodec/aarch64/fft_neon.S
+++ b/libavcodec/aarch64/fft_neon.S
@@ -342,8 +342,7 @@ endfunc
 function fft\n\()_neon, align=6
         AARCH64_VALID_JUMP_TARGET
         AARCH64_SIGN_LINK_REGISTER
-        sub             sp,  sp,  #16
-        stp             x28, x30, [sp]
+        stp             x28, x30, [sp, #-16]!
         add             x28, x0,  #\n4*2*8
         bl              fft\n2\()_neon
         mov             x0,  x28
diff --git a/libavcodec/aarch64/h264idct_neon.S b/libavcodec/aarch64/h264idct_neon.S
index 7d2879b0ce..375da31d65 100644
--- a/libavcodec/aarch64/h264idct_neon.S
+++ b/libavcodec/aarch64/h264idct_neon.S
@@ -157,8 +157,7 @@ function ff_h264_idct_add16intra_neon, export=1
 endfunc
 
 function ff_h264_idct_add8_neon, export=1
-        sub             sp,  sp, #0x40
-        stp             x19, x20, [sp]
+        stp             x19, x20, [sp, #-0x40]!
         mov             x12, x30
         ldp             x6,  x15, [x0]          // dest[0], dest[1]
         add             x5,  x1,  #16*4         // block_offset
@@ -187,8 +186,7 @@ function ff_h264_idct_add8_neon, export=1
         csel            x6,  x15, x6,  eq
         cmp             x10, #20
         b.lt            1b
-        ldp             x19, x20, [sp]
-        add             sp,  sp,  #0x40
+        ldp             x19, x20, [sp], #0x40
         ret             x12
 endfunc
 
diff --git a/libavcodec/aarch64/hevcdsp_sao_neon.S b/libavcodec/aarch64/hevcdsp_sao_neon.S
index d4decfde3b..30e83dda5d 100644
--- a/libavcodec/aarch64/hevcdsp_sao_neon.S
+++ b/libavcodec/aarch64/hevcdsp_sao_neon.S
@@ -33,8 +33,7 @@
 //                      int16_t *sao_offset_val, int sao_left_class,
 //                      int width, int height)
 function ff_hevc_sao_band_filter_8x8_8_neon, export=1
-        sub             sp,  sp, #64
-        stp             xzr, xzr, [sp]
+        stp             xzr, xzr, [sp, #-64]!
         stp             xzr, xzr, [sp, #16]
         stp             xzr, xzr, [sp, #32]
         stp             xzr, xzr, [sp, #48]
diff --git a/libavcodec/aarch64/mdct_neon.S b/libavcodec/aarch64/mdct_neon.S
index 6091e72022..98b09bf1ab 100644
--- a/libavcodec/aarch64/mdct_neon.S
+++ b/libavcodec/aarch64/mdct_neon.S
@@ -23,8 +23,7 @@
 #include "libavutil/aarch64/asm.S"
 
 function ff_imdct_half_neon, export=1
-        sub             sp,  sp,  #32
-        stp             x19, x20, [sp]
+        stp             x19, x20, [sp, #-32]!
         AARCH64_SIGN_LINK_REGISTER
         str             x30, [sp, #16]
         mov             x12, #1
@@ -120,17 +119,15 @@ function ff_imdct_half_neon, export=1
         st2             {v4.2s,v5.2s},  [x0]
         st2             {v6.2s,v7.2s},  [x8]
 
-        ldp             x19, x20, [sp]
         ldr             x30, [sp, #16]
         AARCH64_VALIDATE_LINK_REGISTER
-        add             sp,  sp,  #32
+        ldp             x19, x20, [sp], #32
 
         ret
 endfunc
 
 function ff_imdct_calc_neon, export=1
-        sub             sp,  sp,  #32
-        stp             x19, x20, [sp]
+        stp             x19, x20, [sp, #-32]!
         AARCH64_SIGN_LINK_REGISTER
         str             x30, [sp, #16]
         ldr             w3,  [x0, #28]          // mdct_bits
@@ -163,18 +160,16 @@ function ff_imdct_calc_neon, export=1
         subs            x19, x19,  #16
         b.gt            1b
 
-        ldp             x19, x20, [sp]
         ldr             x30, [sp, #16]
         AARCH64_VALIDATE_LINK_REGISTER
-        add             sp,  sp,  #32
+        ldp             x19, x20, [sp], #32
 
         ret
 endfunc
 
 
 function ff_mdct_calc_neon, export=1
-        sub             sp,  sp,  #32
-        stp             x19, x20, [sp]
+        stp             x19, x20, [sp, #-32]!
         AARCH64_SIGN_LINK_REGISTER
         str             x30, [sp, #16]
 
@@ -323,10 +318,9 @@ function ff_mdct_calc_neon, export=1
         st2             {v4.2s,v5.2s},  [x0]
         st2             {v6.2s,v7.2s},  [x8]
 
-        ldp             x19, x20, [sp]
         ldr             x30, [sp, #16]
         AARCH64_VALIDATE_LINK_REGISTER
-        add             sp,  sp,  #32
+        ldp             x19, x20, [sp], #32
 
         ret
 endfunc
diff --git a/libavcodec/aarch64/me_cmp_neon.S b/libavcodec/aarch64/me_cmp_neon.S
index d8a18cd4b8..cf86e5081d 100644
--- a/libavcodec/aarch64/me_cmp_neon.S
+++ b/libavcodec/aarch64/me_cmp_neon.S
@@ -1173,10 +1173,9 @@ function nsse16_neon, export=1
         bl              X(sse16_neon)
         ldr             x30, [sp, #0x30]
         mov             w9, w0                                  // here we store score1
-        ldr             x5, [sp]
         ldp             x1, x2, [sp, #0x10]
         ldp             x3, x4, [sp, #0x20]
-        add             sp, sp, #0x40
+        ldr             x5, [sp], #0x40
 
         movi            v16.8h, #0
         movi            v17.8h, #0
@@ -1295,10 +1294,9 @@ function nsse8_neon, export=1
         bl              X(sse8_neon)
         ldr             x30, [sp, #0x30]
         mov             w9, w0                                  // here we store score1
-        ldr             x5, [sp]
         ldp             x1, x2, [sp, #0x10]
         ldp             x3, x4, [sp, #0x20]
-        add             sp, sp, #0x40
+        ldr             x5, [sp], #0x40
 
         movi            v16.8h, #0
         movi            v17.8h, #0
diff --git a/libavcodec/aarch64/synth_filter_neon.S b/libavcodec/aarch64/synth_filter_neon.S
index ba79ba9686..4f8494ce30 100644
--- a/libavcodec/aarch64/synth_filter_neon.S
+++ b/libavcodec/aarch64/synth_filter_neon.S
@@ -61,11 +61,11 @@ function ff_synth_filter_float_neon, export=1
 
         blr             x9
 
-        ldp             x2,  x4,  [sp]          // synct_buf_2, window
         ldp             x13, x9,  [sp, #16]     // out, synth_buf
         ldp             x0,  x30, [sp, #32]     // *synth_buf_offset
         AARCH64_VALIDATE_LINK_REGISTER
         ldr             s0,  [sp, #48]
+        ldp             x2,  x4,  [sp], #64     // synct_buf_2, window
 
         add             x3,  x2,  #16*4         // synct_buf_2 + 16
         add             x14, x13, #16*4         // out + 16
@@ -116,6 +116,5 @@ function ff_synth_filter_float_neon, export=1
         b               1b
 
 10:
-        add             sp,  sp,  #64
         ret
 endfunc
diff --git a/libavcodec/aarch64/vp9itxfm_neon.S b/libavcodec/aarch64/vp9itxfm_neon.S
index 03272eae82..a27f7b8ae5 100644
--- a/libavcodec/aarch64/vp9itxfm_neon.S
+++ b/libavcodec/aarch64/vp9itxfm_neon.S
@@ -850,10 +850,10 @@ function ff_vp9_\txfm1\()_\txfm2\()_16x16_add_neon, export=1
         mov             x15, x30
         // iadst16 requires clobbering v8-v15, but idct16 doesn't need to.
 .ifnc \txfm1\()_\txfm2,idct_idct
-        stp             d14, d15, [sp, #-0x10]!
-        stp             d12, d13, [sp, #-0x10]!
-        stp             d10, d11, [sp, #-0x10]!
-        stp             d8,  d9,  [sp, #-0x10]!
+        stp             d8,  d9,  [sp, #-0x40]!
+        stp             d14, d15, [sp, #0x30]
+        stp             d12, d13, [sp, #0x20]
+        stp             d10, d11, [sp, #0x10]
 .endif
 
         sub             sp,  sp,  #512
@@ -920,10 +920,10 @@ function ff_vp9_\txfm1\()_\txfm2\()_16x16_add_neon, export=1
 
         add             sp,  sp,  #512
 .ifnc \txfm1\()_\txfm2,idct_idct
-        ldp             d8,  d9,  [sp], 0x10
-        ldp             d10, d11, [sp], 0x10
-        ldp             d12, d13, [sp], 0x10
-        ldp             d14, d15, [sp], 0x10
+        ldp             d10, d11, [sp, #0x10]
+        ldp             d12, d13, [sp, #0x20]
+        ldp             d14, d15, [sp, #0x30]
+        ldp             d8,  d9,  [sp], #0x40
 .endif
         ret             x15
 endfunc
@@ -1486,8 +1486,8 @@ function ff_vp9_idct_idct_32x32_add_neon, export=1
 
         mov             x15, x30
 
-        stp             d10, d11, [sp, #-0x10]!
-        stp             d8,  d9,  [sp, #-0x10]!
+        stp             d8,  d9,  [sp, #-0x20]!
+        stp             d10, d11, [sp, #0x10]
 
         sub             sp,  sp,  #2048
 
@@ -1544,8 +1544,8 @@ function ff_vp9_idct_idct_32x32_add_neon, export=1
 
         add             sp,  sp,  #2048
 
-        ldp             d8,  d9,  [sp], 0x10
-        ldp             d10, d11, [sp], 0x10
+        ldp             d10, d11, [sp, #0x10]
+        ldp             d8,  d9,  [sp], #0x20
 
         ret             x15
 endfunc
@@ -1569,8 +1569,8 @@ function idct32x32_\size\()_add_neon
 
         add             sp,  sp,  #2048
 
-        ldp             d8,  d9,  [sp], 0x10
-        ldp             d10, d11, [sp], 0x10
+        ldp             d10, d11, [sp, #0x10]
+        ldp             d8,  d9,  [sp], #0x20
 
         ret             x15
 endfunc
diff --git a/libavcodec/aarch64/vp9lpf_16bpp_neon.S b/libavcodec/aarch64/vp9lpf_16bpp_neon.S
index a092617b92..e3e70491c6 100644
--- a/libavcodec/aarch64/vp9lpf_16bpp_neon.S
+++ b/libavcodec/aarch64/vp9lpf_16bpp_neon.S
@@ -417,10 +417,10 @@ endfunc
 function ff_\func\()_\bpp\()_neon, export=1
 .if \push
         mov             x16, x30
-        stp             d14, d15, [sp, #-0x10]!
-        stp             d12, d13, [sp, #-0x10]!
-        stp             d10, d11, [sp, #-0x10]!
-        stp             d8,  d9,  [sp, #-0x10]!
+        stp             d8,  d9,  [sp, #-0x40]!
+        stp             d14, d15, [sp, #0x30]
+        stp             d12, d13, [sp, #0x20]
+        stp             d10, d11, [sp, #0x10]
 .endif
         lsl             w2,  w2,  #\bpp - 8
         lsl             w3,  w3,  #\bpp - 8
@@ -430,10 +430,10 @@ function ff_\func\()_\bpp\()_neon, export=1
         mov             x7,  #((1 << \bpp) - 1)
 .if \push
         bl              \func\()_16_neon
-        ldp             d8,  d9,  [sp], 0x10
-        ldp             d10, d11, [sp], 0x10
-        ldp             d12, d13, [sp], 0x10
-        ldp             d14, d15, [sp], 0x10
+        ldp             d10, d11, [sp, #0x10]
+        ldp             d12, d13, [sp, #0x20]
+        ldp             d14, d15, [sp, #0x30]
+        ldp             d8,  d9,  [sp], #0x40
         ret             x16
 .else
         b               \func\()_16_neon
@@ -450,10 +450,10 @@ endfunc
 function ff_\func\()_\suffix\()_\bpp\()_neon, export=1
         mov             x16, x30
 .if \push
-        stp             d14, d15, [sp, #-0x10]!
-        stp             d12, d13, [sp, #-0x10]!
-        stp             d10, d11, [sp, #-0x10]!
-        stp             d8,  d9,  [sp, #-0x10]!
+        stp             d8,  d9,  [sp, #-0x40]!
+        stp             d14, d15, [sp, #0x30]
+        stp             d12, d13, [sp, #0x20]
+        stp             d10, d11, [sp, #0x10]
 .endif
         lsl             w2,  w2,  #\bpp - 8
         lsl             w3,  w3,  #\bpp - 8
@@ -469,10 +469,10 @@ function ff_\func\()_\suffix\()_\bpp\()_neon, export=1
 .endif
         bl              \func\()_\int_suffix\()_16_neon
 .if \push
-        ldp             d8,  d9,  [sp], 0x10
-        ldp             d10, d11, [sp], 0x10
-        ldp             d12, d13, [sp], 0x10
-        ldp             d14, d15, [sp], 0x10
+        ldp             d10, d11, [sp, #0x10]
+        ldp             d12, d13, [sp, #0x20]
+        ldp             d14, d15, [sp, #0x30]
+        ldp             d8,  d9,  [sp], 0x40
 .endif
         ret             x16
 endfunc
diff --git a/libavcodec/aarch64/vp9lpf_neon.S b/libavcodec/aarch64/vp9lpf_neon.S
index 694ff8956f..9a79f48df3 100644
--- a/libavcodec/aarch64/vp9lpf_neon.S
+++ b/libavcodec/aarch64/vp9lpf_neon.S
@@ -564,10 +564,10 @@ function vp9_loop_filter_16
         loop_filter     16, .8b,  0,    v8,  v9,  v10, v11, v12, v13, v14, v15
         ret
 9:
-        ldp             d8,  d9,  [sp], 0x10
-        ldp             d10, d11, [sp], 0x10
-        ldp             d12, d13, [sp], 0x10
-        ldp             d14, d15, [sp], 0x10
+        ldp             d10, d11, [sp, #0x10]
+        ldp             d12, d13, [sp, #0x20]
+        ldp             d14, d15, [sp, #0x30]
+        ldp             d8,  d9,  [sp], #0x40
         ret             x10
 endfunc
 
@@ -575,10 +575,10 @@ function vp9_loop_filter_16_16b
         loop_filter     16, .16b, 0,    v8,  v9,  v10, v11, v12, v13, v14, v15
         ret
 9:
-        ldp             d8,  d9,  [sp], 0x10
-        ldp             d10, d11, [sp], 0x10
-        ldp             d12, d13, [sp], 0x10
-        ldp             d14, d15, [sp], 0x10
+        ldp             d10, d11, [sp, #0x10]
+        ldp             d12, d13, [sp, #0x20]
+        ldp             d14, d15, [sp, #0x30]
+        ldp             d8,  d9,  [sp], #0x40
         ret             x10
 endfunc
 
@@ -973,10 +973,10 @@ mix_h_16 88
 
 function ff_vp9_loop_filter_v_16_8_neon, export=1
         mov             x10, x30
-        stp             d14, d15, [sp, #-0x10]!
-        stp             d12, d13, [sp, #-0x10]!
-        stp             d10, d11, [sp, #-0x10]!
-        stp             d8,  d9,  [sp, #-0x10]!
+        stp             d8,  d9,  [sp, #-0x40]!
+        stp             d14, d15, [sp, #0x30]
+        stp             d12, d13, [sp, #0x20]
+        stp             d10, d11, [sp, #0x10]
         sub             x9,  x0,  x1, lsl #3
         ld1             {v16.8b}, [x9], x1 // p7
         ld1             {v24.8b}, [x0], x1 // q0
@@ -1018,10 +1018,10 @@ function ff_vp9_loop_filter_v_16_8_neon, export=1
         st1             {v9.8b},  [x9], x1
         st1             {v17.8b}, [x0], x1
 9:
-        ldp             d8,  d9,  [sp], 0x10
-        ldp             d10, d11, [sp], 0x10
-        ldp             d12, d13, [sp], 0x10
-        ldp             d14, d15, [sp], 0x10
+        ldp             d10, d11, [sp, #0x10]
+        ldp             d12, d13, [sp, #0x20]
+        ldp             d14, d15, [sp, #0x30]
+        ldp             d8,  d9,  [sp], #0x40
         ret             x10
 8:
         add             x9,  x9,  x1, lsl #2
@@ -1045,10 +1045,10 @@ endfunc
 
 function ff_vp9_loop_filter_v_16_16_neon, export=1
         mov             x10, x30
-        stp             d14, d15, [sp, #-0x10]!
-        stp             d12, d13, [sp, #-0x10]!
-        stp             d10, d11, [sp, #-0x10]!
-        stp             d8,  d9,  [sp, #-0x10]!
+        stp             d8,  d9,  [sp, #-0x40]!
+        stp             d14, d15, [sp, #0x30]
+        stp             d12, d13, [sp, #0x20]
+        stp             d10, d11, [sp, #0x10]
         sub             x9,  x0,  x1, lsl #3
         ld1             {v16.16b}, [x9], x1 // p7
         ld1             {v24.16b}, [x0], x1 // q0
@@ -1087,10 +1087,10 @@ function ff_vp9_loop_filter_v_16_16_neon, export=1
         st1             {v9.16b},  [x9], x1
         st1             {v17.16b}, [x0], x1
 9:
-        ldp             d8,  d9,  [sp], 0x10
-        ldp             d10, d11, [sp], 0x10
-        ldp             d12, d13, [sp], 0x10
-        ldp             d14, d15, [sp], 0x10
+        ldp             d10, d11, [sp, #0x10]
+        ldp             d12, d13, [sp, #0x20]
+        ldp             d14, d15, [sp, #0x30]
+        ldp             d8,  d9,  [sp], #0x40
         ret             x10
 8:
         add             x9,  x9,  x1, lsl #2
@@ -1112,10 +1112,10 @@ endfunc
 
 function ff_vp9_loop_filter_h_16_8_neon, export=1
         mov             x10, x30
-        stp             d14, d15, [sp, #-0x10]!
-        stp             d12, d13, [sp, #-0x10]!
-        stp             d10, d11, [sp, #-0x10]!
-        stp             d8,  d9,  [sp, #-0x10]!
+        stp             d8,  d9,  [sp, #-0x40]!
+        stp             d14, d15, [sp, #0x30]
+        stp             d12, d13, [sp, #0x20]
+        stp             d10, d11, [sp, #0x10]
         sub             x9,  x0,  #8
         ld1             {v16.8b}, [x9], x1
         ld1             {v24.8b}, [x0], x1
@@ -1164,10 +1164,10 @@ function ff_vp9_loop_filter_h_16_8_neon, export=1
         st1             {v9.8b},  [x9], x1
         st1             {v31.8b}, [x0], x1
 9:
-        ldp             d8,  d9,  [sp], 0x10
-        ldp             d10, d11, [sp], 0x10
-        ldp             d12, d13, [sp], 0x10
-        ldp             d14, d15, [sp], 0x10
+        ldp             d10, d11, [sp, #0x10]
+        ldp             d12, d13, [sp, #0x20]
+        ldp             d14, d15, [sp, #0x30]
+        ldp             d8,  d9,  [sp], #0x40
         ret             x10
 8:
         // The same writeback as in loop_filter_h_8_8
@@ -1202,10 +1202,10 @@ endfunc
 
 function ff_vp9_loop_filter_h_16_16_neon, export=1
         mov             x10, x30
-        stp             d14, d15, [sp, #-0x10]!
-        stp             d12, d13, [sp, #-0x10]!
-        stp             d10, d11, [sp, #-0x10]!
-        stp             d8,  d9,  [sp, #-0x10]!
+        stp             d8,  d9,  [sp, #-0x40]!
+        stp             d14, d15, [sp, #0x30]
+        stp             d12, d13, [sp, #0x20]
+        stp             d10, d11, [sp, #0x10]
         sub             x9,  x0,  #8
         ld1             {v16.8b},   [x9], x1
         ld1             {v24.8b},   [x0], x1
@@ -1283,10 +1283,10 @@ function ff_vp9_loop_filter_h_16_16_neon, export=1
         st1             {v9.d}[1],  [x9], x1
         st1             {v31.d}[1], [x0], x1
 9:
-        ldp             d8,  d9,  [sp], 0x10
-        ldp             d10, d11, [sp], 0x10
-        ldp             d12, d13, [sp], 0x10
-        ldp             d14, d15, [sp], 0x10
+        ldp             d10, d11, [sp, #0x10]
+        ldp             d12, d13, [sp, #0x20]
+        ldp             d14, d15, [sp, #0x30]
+        ldp             d8,  d9,  [sp], #0x40
         ret             x10
 8:
         sub             x9,  x0,  #4
diff --git a/libavutil/aarch64/tx_float_neon.S b/libavutil/aarch64/tx_float_neon.S
index 4126c3b812..e5531dcc7c 100644
--- a/libavutil/aarch64/tx_float_neon.S
+++ b/libavutil/aarch64/tx_float_neon.S
@@ -866,10 +866,10 @@ FFT16_FN ns_float, 1
 
 .macro FFT32_FN name, no_perm
 function ff_tx_fft32_\name\()_neon, export=1
-        stp             d8,  d9,  [sp, #-16]
-        stp             d10, d11, [sp, #-32]
-        stp             d12, d13, [sp, #-48]
-        stp             d14, d15, [sp, #-64]
+        stp             d14, d15, [sp, #-16*4]!
+        stp             d8,  d9,  [sp, #16*3]
+        stp             d10, d11, [sp, #16*2]
+        stp             d12, d13, [sp, #16]
 
         LOAD_SUBADD
         SETUP_SR_RECOMB 32, x7, x8, x9
@@ -911,10 +911,10 @@ function ff_tx_fft32_\name\()_neon, export=1
         zip2            v31.2d, v11.2d, v15.2d
         st1             { v28.4s, v29.4s, v30.4s, v31.4s }, [x1]
 
-        ldp             d14, d15, [sp, #-64]
-        ldp             d12, d13, [sp, #-48]
-        ldp             d10, d11, [sp, #-32]
-        ldp             d8,  d9,  [sp, #-16]
+        ldp             d12, d13, [sp, #16]
+        ldp             d10, d11, [sp, #16*2]
+        ldp             d8,  d9,  [sp, #16*3]
+        ldp             d14, d15, [sp], #16*4
 
         ret
 endfunc
@@ -966,12 +966,12 @@ FFT32_FN ns_float, 1
 
 .macro FFT_SPLIT_RADIX_FN name, no_perm
 function ff_tx_fft_sr_\name\()_neon, export=1
-        stp             d8,  d9,  [sp, #-16]!
-        stp             d10, d11, [sp, #-16]!
-        stp             d12, d13, [sp, #-16]!
-        stp             d14, d15, [sp, #-16]!
-        stp             x19, x20, [sp, #-16]!
-        stp             x21, x22, [sp, #-16]!
+        stp             x21, x22, [sp, #-16*6]!
+        stp             d8,  d9,  [sp, #16*5]
+        stp             d10, d11, [sp, #16*4]
+        stp             d12, d13, [sp, #16*3]
+        stp             d14, d15, [sp, #16*2]
+        stp             x19, x20, [sp, #16]
 
         ldr             w19, [x0, #0] // global target
         mov             w20, w19      // local length
@@ -1185,12 +1185,12 @@ SR_TRANSFORM_DEF        131072
         subs            w19, w19, #32*4
         b.gt            0b
 
-        ldp             x21, x22, [sp], #16
-        ldp             x19, x20, [sp], #16
-        ldp             d14, d15, [sp], #16
-        ldp             d12, d13, [sp], #16
-        ldp             d10, d11, [sp], #16
-        ldp             d8,  d9,  [sp], #16
+        ldp             x19, x20, [sp, #16]
+        ldp             d14, d15, [sp, #16*2]
+        ldp             d12, d13, [sp, #16*3]
+        ldp             d10, d11, [sp, #16*4]
+        ldp             d8,  d9,  [sp, #16*5]
+        ldp             x21, x22, [sp], #16*6
 
         ret
 
@@ -1279,12 +1279,12 @@ SR_TRANSFORM_DEF        131072
         zip2            v7.2d, v15.2d, v23.2d
         st1             {  v4.4s,  v5.4s,  v6.4s,  v7.4s }, [x15]
 
-        ldp             x21, x22, [sp], #16
-        ldp             x19, x20, [sp], #16
-        ldp             d14, d15, [sp], #16
-        ldp             d12, d13, [sp], #16
-        ldp             d10, d11, [sp], #16
-        ldp             d8,  d9,  [sp], #16
+        ldp             x19, x20, [sp, #16]
+        ldp             d14, d15, [sp, #16*2]
+        ldp             d12, d13, [sp, #16*3]
+        ldp             d10, d11, [sp, #16*4]
+        ldp             d8,  d9,  [sp, #16*5]
+        ldp             x21, x22, [sp], #16*6
 
         ret
 endfunc