GitLab

Also fix a calculation for u_idx. Fixes 5646860283281408 of #183.

These functions have been tuned against Cortex A53 and Snapdragon
835. The bilin functions have mainly been written with code size
in mind, as they aren't used much in practice.

Relative speedups for the actual filtering fuctions (that don't
just do a plain copy) are around 4-15x, some over 20x. This is
in comparison with GCC 5.4 with autovectorization disabled; the
actual real-world speedup against autovectorized C code is around
4-10x.

Relative speedups measured with checkasm:
                                Cortex A53   Snapdragon 835
mc_8tap_regular_w2_0_8bpc_neon:       6.96   5.28
mc_8tap_regular_w2_h_8bpc_neon:       5.16   4.35
mc_8tap_regular_w2_hv_8bpc_neon:      5.37   4.98
mc_8tap_regular_w2_v_8bpc_neon:       6.35   4.85
mc_8tap_regular_w4_0_8bpc_neon:       6.78   5.73
mc_8tap_regular_w4_h_8bpc_neon:       8.40   6.60
mc_8tap_regular_w4_hv_8bpc_neon:      7.23   7.10
mc_8tap_regular_w4_v_8bpc_neon:       9.06   7.76
mc_8tap_regular_w8_0_8bpc_neon:       6.96   5.55
mc_8tap_regular_w8_h_8bpc_neon:      10.36   6.88
mc_8tap_regular_w8_hv_8bpc_neon:      9.49   6.86
mc_8tap_regular_w8_v_8bpc_neon:      12.06   9.61
mc_8tap_regular_w16_0_8bpc_neon:      6.68   4.51
mc_8tap_regular_w16_h_8bpc_neon:     12.30   7.77
mc_8tap_regular_w16_hv_8bpc_neon:     9.50   6.68
mc_8tap_regular_w16_v_8bpc_neon:     12.93   9.68
mc_8tap_regular_w32_0_8bpc_neon:      3.91   2.93
mc_8tap_regular_w32_h_8bpc_neon:     13.06   7.89
mc_8tap_regular_w32_hv_8bpc_neon:     9.37   6.70
mc_8tap_regular_w32_v_8bpc_neon:     12.88   9.49
mc_8tap_regular_w64_0_8bpc_neon:      2.89   1.68
mc_8tap_regular_w64_h_8bpc_neon:     13.48   8.00
mc_8tap_regular_w64_hv_8bpc_neon:     9.23   6.53
mc_8tap_regular_w64_v_8bpc_neon:     13.11   9.68
mc_8tap_regular_w128_0_8bpc_neon:     1.89   1.24
mc_8tap_regular_w128_h_8bpc_neon:    13.58   7.98
mc_8tap_regular_w128_hv_8bpc_neon:    8.86   6.53
mc_8tap_regular_w128_v_8bpc_neon:    12.46   9.63
mc_bilinear_w2_0_8bpc_neon:           7.02   5.40
mc_bilinear_w2_h_8bpc_neon:           3.65   3.14
mc_bilinear_w2_hv_8bpc_neon:          4.36   4.84
mc_bilinear_w2_v_8bpc_neon:           5.22   4.28
mc_bilinear_w4_0_8bpc_neon:           6.87   5.99
mc_bilinear_w4_h_8bpc_neon:           6.50   8.61
mc_bilinear_w4_hv_8bpc_neon:          7.70   7.99
mc_bilinear_w4_v_8bpc_neon:           7.04   9.10
mc_bilinear_w8_0_8bpc_neon:           7.03   5.70
mc_bilinear_w8_h_8bpc_neon:          11.30  15.14
mc_bilinear_w8_hv_8bpc_neon:         15.74  13.50
mc_bilinear_w8_v_8bpc_neon:          13.40  17.54
mc_bilinear_w16_0_8bpc_neon:          6.75   4.48
mc_bilinear_w16_h_8bpc_neon:         17.02  13.95
mc_bilinear_w16_hv_8bpc_neon:        17.37  13.78
mc_bilinear_w16_v_8bpc_neon:         23.69  22.98
mc_bilinear_w32_0_8bpc_neon:          3.88   3.18
mc_bilinear_w32_h_8bpc_neon:         18.80  14.97
mc_bilinear_w32_hv_8bpc_neon:        17.74  14.02
mc_bilinear_w32_v_8bpc_neon:         24.46  23.04
mc_bilinear_w64_0_8bpc_neon:          2.87   1.66
mc_bilinear_w64_h_8bpc_neon:         19.54  16.02
mc_bilinear_w64_hv_8bpc_neon:        17.80  14.32
mc_bilinear_w64_v_8bpc_neon:         24.79  23.63
mc_bilinear_w128_0_8bpc_neon:         2.13   1.23
mc_bilinear_w128_h_8bpc_neon:        19.89  16.24
mc_bilinear_w128_hv_8bpc_neon:       17.55  14.15
mc_bilinear_w128_v_8bpc_neon:        24.45  23.54
mct_8tap_regular_w4_0_8bpc_neon:      5.56   5.51
mct_8tap_regular_w4_h_8bpc_neon:      7.48   5.80
mct_8tap_regular_w4_hv_8bpc_neon:     7.27   7.09
mct_8tap_regular_w4_v_8bpc_neon:      7.80   6.84
mct_8tap_regular_w8_0_8bpc_neon:      9.54   9.25
mct_8tap_regular_w8_h_8bpc_neon:      9.08   6.55
mct_8tap_regular_w8_hv_8bpc_neon:     9.16   6.30
mct_8tap_regular_w8_v_8bpc_neon:     10.79   8.66
mct_8tap_regular_w16_0_8bpc_neon:    15.35  10.50
mct_8tap_regular_w16_h_8bpc_neon:    10.18   6.76
mct_8tap_regular_w16_hv_8bpc_neon:    9.17   6.11
mct_8tap_regular_w16_v_8bpc_neon:    11.52   8.72
mct_8tap_regular_w32_0_8bpc_neon:    15.82  10.09
mct_8tap_regular_w32_h_8bpc_neon:    10.75   6.85
mct_8tap_regular_w32_hv_8bpc_neon:    9.00   6.22
mct_8tap_regular_w32_v_8bpc_neon:    11.58   8.67
mct_8tap_regular_w64_0_8bpc_neon:    15.28   9.68
mct_8tap_regular_w64_h_8bpc_neon:    10.93   6.96
mct_8tap_regular_w64_hv_8bpc_neon:    8.81   6.53
mct_8tap_regular_w64_v_8bpc_neon:    11.42   8.73
mct_8tap_regular_w128_0_8bpc_neon:   14.41   7.67
mct_8tap_regular_w128_h_8bpc_neon:   10.92   6.96
mct_8tap_regular_w128_hv_8bpc_neon:   8.56   6.51
mct_8tap_regular_w128_v_8bpc_neon:   11.16   8.70
mct_bilinear_w4_0_8bpc_neon:          5.66   5.77
mct_bilinear_w4_h_8bpc_neon:          5.16   6.40
mct_bilinear_w4_hv_8bpc_neon:         6.86   6.82
mct_bilinear_w4_v_8bpc_neon:          4.75   6.09
mct_bilinear_w8_0_8bpc_neon:          9.78  10.00
mct_bilinear_w8_h_8bpc_neon:          8.98  11.37
mct_bilinear_w8_hv_8bpc_neon:        14.42  10.83
mct_bilinear_w8_v_8bpc_neon:          9.12  11.62
mct_bilinear_w16_0_8bpc_neon:        15.59  10.76
mct_bilinear_w16_h_8bpc_neon:        11.98   8.77
mct_bilinear_w16_hv_8bpc_neon:       15.83  10.73
mct_bilinear_w16_v_8bpc_neon:        14.70  14.60
mct_bilinear_w32_0_8bpc_neon:        15.89  10.32
mct_bilinear_w32_h_8bpc_neon:        13.47   9.07
mct_bilinear_w32_hv_8bpc_neon:       16.01  10.95
mct_bilinear_w32_v_8bpc_neon:        14.85  14.16
mct_bilinear_w64_0_8bpc_neon:        15.36  10.51
mct_bilinear_w64_h_8bpc_neon:        14.00   9.61
mct_bilinear_w64_hv_8bpc_neon:       15.82  11.27
mct_bilinear_w64_v_8bpc_neon:        14.61  14.76
mct_bilinear_w128_0_8bpc_neon:       14.41   7.92
mct_bilinear_w128_h_8bpc_neon:       13.31   9.58
mct_bilinear_w128_hv_8bpc_neon:      14.07  11.18
mct_bilinear_w128_v_8bpc_neon:       11.57  14.42

Fixes #149.

This reverts commit 597a6eb9. It leads to
assertion failures in oss-fuzz.

Fixes unaligned writes while splatting coefs for skip blocks with
clusterfuzz-testcase-minimized-dav1d_fuzzer_mt-5684725352497152 and
clusterfuzz-testcase-minimized-dav1d_fuzzer_mt-5728508249112576.

Catches warnings in assert statements.

ruy and rux are in unit_size dimensions, whereas lr_mask are in sb128
dimensions, and unit_idx is in sb64 dimensions, so one can't be derived
from the other. Instead, remove ruy/rux and derive unit_idx and sb_idx
directly from the block positions aligned to the unit_size.

This is consistent with what libaom does. Should fix #175.

Fixes #172.

With the decoupled decoding data there might be remaining input data during
draining which can cause bitstream parsing errors.

The race is exposed by not draining the decoder correctly after
02606969 (decoupled decoding api). Fixes a memleak with
clusterfuzz-testcase-minimized-dav1d_fuzzer_mt-5728508249112576. Credits
to oss-fuzz.

But don't abort decoding; instead, simply force translational motion.

See section 7.11.2.4 in AV1 spec. Because frame contexts are not
passed into the ipred_z*_c functions the flag is set as a bit inside
the 'angle' function argument.

A new coded video sequence (see page 193; section 7.5 of the spec)
begins when we see a sequence header that isn't bit identical to
previous ones. This is the point at which we can throw away previous
frames etc.

The number of read bits can be equal to the size of the packet. Fixes a
triggered assert in
clusterfuzz-testcase-minimized-dav1d_fuzzer-5746175664193536. Credits to
oss-fuzz.

This version now includes clang.

This code originally looked like "assert (init_bit_pos % 8 == 0)" and
I changed it to use "& 7" to match the prevaling style. Unfortunately,
"&" binds more weakly than "==". Oops!

This fixes failures when an OBU has more than a byte's worth of
trailing zeros.

As part of this work, it also rejigs the dav1d_flush_get_bits function
slightly. This worked before, but it wasn't very obvious why (it
worked because bits_left was never more than 7). This patch renames it
to dav1d_bytealign_get_bits, which makes it clearer what it does and
adds a comment explaining why it works properly.

The new dav1d_bytealign_get_bits is also now void (rather than
returning the next byte to read). The patch defines
dav1d_get_bits_pos, which returns the current bit position. This feels
a little easier to reason about.

We also add a new check to make sure that we haven't fallen off the
end of the OBU. This can happen when a byte buffer contains more than
one OBU: the GetBits might not have got to EOF, but we might now be
half-way through the next OBU.

See section 5.9.12 of the AV1 spec. The flag controlling U and V share
a quantization level wasn't being read.

The previous code raised an error if !segmentation.update_map but the
reference frame didn't yield any segmentation data. (The first "goto
error" that the patch removes happens if the reference frame was the
right size but had no segmentation data; the second happens if the
reference frame was the wrong size).

This doesn't match the logic in the description of
load_previous_segment_ids in section 6.8.2 of the spec.

This patch allows such streams, allocating and zeroing cur_segmap in
this case. It is still an error for a stream to signal a temporal
update but not to have valid segmentation data from the ref frame -
that's the error case that the patch puts back in.

The first memset is dead code: if primary_ref_frame is
PRIMARY_REF_NONE then segmentation.update_data is always true. The
patch removes this memset and explains why the copy in the other
branch is correct.

The second memset should always fire: if segmentation is not enabled
for this frame, the seg_data structure should be set to zero rather
than copied from a reference frame (see section 5.9.14 of the AV1
spec).

Fixes #166.

Fixes a heap buffer overflow in emu_edge_c with
clusterfuzz-testcase-minimized-dav1d_fuzzer_mt-5089954858795008 if the
reference frame is smaller than the current frame. Credits to oss-fuzz.

Fixes undefined shifts in put_bilin_scaled_c with
clusterfuzz-testcase-minimized-dav1d_fuzzer_mt-5732654503165952. Credits
to oss-fuzz.

Fixes a heap buffer overflow with high bit depth scaled reference frames
in clusterfuzz-testcase-minimized-dav1d_fuzzer_mt-5652139771166720.
Credits to oss-fuzz.

Fixes #121.

Fixes decoding of keyframe in #121.