1959238 - Update opus to new version d8d7763369e2d5af79925c61bf4b4e7adccb38fa from 2025-04-08 13:40:36

Status: RESOLVED FIXED

(Core :: Audio/Video: Playback, enhancement)

Description

[Update Bot]

This update covers 11 commits, including 3 new upstream commits I've never filed a bug on before. (They're the top 3.). Here are the overall diff statistics, and then the commit information.

---

media/libopus/celt/arch.h | 2 ++
media/libopus/celt/fixed_debug.h | 2 ++
media/libopus/celt/fixed_generic.h | 2 ++
media/libopus/celt/mdct.c | 4 ++--
media/libopus/moz.build | 2 +-
media/libopus/moz.yaml | 4 ++--
media/libopus/silk/NSQ.c | 2 +-
media/libopus/src/mapping_matrix.c | 6 +++++-
media/libopus/src/opus_decoder.c | 4 ++++
9 files changed, 21 insertions(+), 7 deletions(-)

---

d8d7763369e2d5af79925c61bf4b4e7adccb38fa by Jean-Marc Valin <[e-mail]>

https://gitlab.xiph.org/xiph/opus/commit/d8d7763369e2d5af79925c61bf4b4e7adccb38fa
Authored: 2025-04-04 14:46:37 -0400
Committed: 2025-04-08 13:40:36 -0400

Fix decoder gain for res24

Files Modified:

• celt/arch.h
• src/opus_decoder.c

---

f0bf32a58c71e9324d00728e13e6d2647c023435 by Jean-Marc Valin <[e-mail]>

https://gitlab.xiph.org/xiph/opus/commit/f0bf32a58c71e9324d00728e13e6d2647c023435
Authored: 2025-04-04 14:44:14 -0400
Committed: 2025-04-08 13:40:36 -0400

Fix projection encoder integer overflow

Files Modified:

• src/mapping_matrix.c

---

60b8f5b80307c1ecb1070f602074c791444c52bc by Jean-Marc Valin <[e-mail]>

https://gitlab.xiph.org/xiph/opus/commit/60b8f5b80307c1ecb1070f602074c791444c52bc
Authored: 2025-04-03 14:45:24 -0400
Committed: 2025-04-08 13:40:35 -0400

Add SHL32_ovflw() and use it in IMDCT

Prevents integer overflow UB in the headroom left shift when
the signal blows up on bad bitstreams (if it triggers, the
signal was already unusable anyway).

Files Modified:

• celt/arch.h
• celt/fixed_debug.h
• celt/fixed_generic.h
• celt/mdct.c

---

a013aa047760ec7508aec5755c225aa3b23a2581 by Jan Buethe <[e-mail]>

https://gitlab.xiph.org/xiph/opus/commit/a013aa047760ec7508aec5755c225aa3b23a2581
Authored: 2025-04-06 17:29:00 -0700
Committed: 2025-04-06 20:42:57 -0700

deprecating _DELAY from conv1d layer dump

Files Modified:

• dnn/torch/weight-exchange/wexchange/c_export/common.py

---

df02d25f0c6334a60f8c01c5ecf63081845d6c9d by Jean-Marc Valin <[e-mail]>

https://gitlab.xiph.org/xiph/opus/commit/df02d25f0c6334a60f8c01c5ecf63081845d6c9d
Authored: 2025-03-29 11:12:24 -0400
Committed: 2025-03-29 11:12:24 -0400

Updated FARGAN model with softquant

Files Modified:

• autogen.sh
• dnn/torch/fargan/dump_fargan_weights.py
• tar_list.txt

---

af6dbd84a82b326d26ec49ebe671c2b3d3ed963b by Jean-Marc Valin <[e-mail]>

https://gitlab.xiph.org/xiph/opus/commit/af6dbd84a82b326d26ec49ebe671c2b3d3ed963b
Authored: 2025-03-28 19:28:14 -0400
Committed: 2025-03-28 20:04:51 -0400

Adds --chunks-per-offset option to train_rdovae.py

Files Modified:

• dnn/torch/rdovae/rdovae/rdovae.py
• dnn/torch/rdovae/train_rdovae.py

---

1ca6933ac480c8584e5e47b4fea5ad2d857a336d by Jan Buethe <[e-mail]>

https://gitlab.xiph.org/xiph/opus/commit/1ca6933ac480c8584e5e47b4fea5ad2d857a336d
Authored: 2025-03-08 13:03:02 -0800
Committed: 2025-03-28 20:04:51 -0400

added soft quantization to RDOVAE and FARGAN

Files Modified:

• dnn/torch/fargan/adv_train_fargan.py
• dnn/torch/fargan/fargan.py
• dnn/torch/fargan/train_fargan.py
• dnn/torch/rdovae/rdovae/rdovae.py
• dnn/torch/rdovae/train_rdovae.py

---

ebccedd918f5803dd1bf3459ad3afd23c4956ea0 by Jean-Marc Valin <[e-mail]>

https://gitlab.xiph.org/xiph/opus/commit/ebccedd918f5803dd1bf3459ad3afd23c4956ea0
Authored: 2025-03-28 19:25:03 -0400
Committed: 2025-03-28 20:04:50 -0400

Fix rdovae exporting

Files Modified:

• dnn/torch/weight-exchange/wexchange/c_export/common.py

---

bb564fab954511def5eff2846cc4b2f00bf04e13 by Jean-Marc Valin <[e-mail]>

https://gitlab.xiph.org/xiph/opus/commit/bb564fab954511def5eff2846cc4b2f00bf04e13
Authored: 2025-03-28 19:19:05 -0400
Committed: 2025-03-28 19:19:05 -0400

UBSan fixes for opus_demo

Files Modified:

• src/opus_demo.c

---

1c370855aa0e7fe3f583d74a2ffd48ca1432a479 by Jean-Marc Valin <[e-mail]>

https://gitlab.xiph.org/xiph/opus/commit/1c370855aa0e7fe3f583d74a2ffd48ca1432a479
Authored: 2025-03-26 15:04:37 -0400
Committed: 2025-03-26 16:34:34 -0400

Forgot one wraparound to ignore

Files Modified:

• silk/NSQ.c

---

e6dd7a3de4237e2bf5762e537c57696110a5fbcf by Jean-Marc Valin <[e-mail]>

https://gitlab.xiph.org/xiph/opus/commit/e6dd7a3de4237e2bf5762e537c57696110a5fbcf
Authored: 2025-03-26 14:59:02 -0400
Committed: 2025-03-26 16:34:31 -0400

Do not run valgrind on -ffast-math builds

valgrind does not properly handle the Denormals-Are-Zero (DAZ) flag,
which can be set by -ffast-math. That would not generally be a problem
except when the compiler implements sqrt(x) as x*rsqrt(x) with an
exception for x=0. In that case, when x is a non-zero denorm value,
the exception does not trigger with valgrind (but does on real
hardware), which causes the sqrt to evaluate to -inf instead of 0.

Files Modified:

• tests/random_config.sh

---

Comment 2

[Update Bot]

I've submitted a try run for this commit: https://treeherder.mozilla.org/jobs?repo=try&revision=46606bdc91291a3e2bc37ac2f2fe3093c19a1058

Comment 3

[Update Bot]

Attachment: Bug 1959238 - Update opus to d8d7763369e2d5af79925c61bf4b4e7adccb38fa

Comment 4

[Update Bot]

All the jobs in the try run succeeded. Like literally all of them, there weren't
even any intermittents. That is pretty surprising to me, so maybe you should double
check to make sure I didn't misinterpret things and that the correct tests ran...

Anyway, I've done all I can, so I'm passing to you to review and land the patch.
When reviewing, please note that this is external code, which needs a full and
careful inspection - not a rubberstamp.

Comment 5

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/556eb766ded1
Update opus to d8d7763369e2d5af79925c61bf4b4e7adccb38fa r=kinetik

Comment 6

[amarc]

https://hg.mozilla.org/mozilla-central/rev/556eb766ded1