[PATCH AUTOSEL 6.18] ASoC: cs35l56: Add support for CS35L62 for SoundWire
From: Sasha Levin
Date: Mon Oct 19 2026 - 11:57:46 EST
From: Stefan Binding <sbinding@xxxxxxxxxxxxxxxxxxxxx>
[ Upstream commit 3330b5f6d8dea66df51c07362f01a39ad9845aad ]
CS35L62 uses the same control and regmap interface as CS35L63 so
support for it can be added into the CS35L56 driver.
Signed-off-by: Stefan Binding <sbinding@xxxxxxxxxxxxxxxxxxxxx>
Signed-off-by: Richard Fitzgerald <rf@xxxxxxxxxxxxxxxxxxxxx>
Link: https://patch.msgid.link/20260715120135.939280-2-rf@xxxxxxxxxxxxxxxxxxxxx
Signed-off-by: Mark Brown <broonie@xxxxxxxxxx>
Signed-off-by: Sasha Levin <sashal@xxxxxxxxxx>
---
LLM Generated explanations, may be completely bogus:
## Phase 1: Commit message forensics
**Step 1.1 — Subject**
- Record: `[ASoC/cs35l56] [add] [enable CS35L62 on SoundWire by reusing
CS35L63 paths]`
**Step 1.2 — Tags**
- Record:
- `Fixes:` none (expected)
- `Reported-by:` none
- `Tested-by:` none
- `Reviewed-by:` none on the commit
- `Acked-by:` none
- `Link: https://patch.msgid.link/20260715120135.939280-2-
rf@xxxxxxxxxxxxxxxxxxxxx`
- `Cc: stable@xxxxxxxxxxxxxxx` none (expected)
- `Signed-off-by: Stefan Binding` (author)
- `Signed-off-by: Richard Fitzgerald` (Cirrus; submitted the series)
- `Signed-off-by: Mark Brown` (ASoC maintainer)
- Ignore any Sasha Levin SOB if present — pipeline noise
**Step 1.3 — Body**
- Record: No bug, crash, leak, or regression is described. Claimed
intent is hardware enablement: CS35L62 shares CS35L63’s control/regmap
interface, so it can live in the existing cs35l56 driver. Symptom
without the patch: a CS35L62 SoundWire amp is not matched (`-ENODEV` /
unknown DEVID). Root cause: missing part ID and type cases. No version
range is given. Mark Brown applied it to `for-7.3` as merge-window
material, not as a bugfix.
**Step 1.4 — Hidden bug fix?**
- Record: No. Wording is “Add support”, not cleanup/locking/error-path
repair. This is new silicon enablement, not a disguised UAF/leak/race
fix.
## Phase 2: Diff analysis
**Step 2.1 — Inventory**
- Record: 3 files, **+7 / −0**, single-driver surgical change.
- `sound/soc/codecs/cs35l56-sdw.c` (+2): `cs35l56_sdw_probe()`,
`cs35l56_sdw_id[]`
- `sound/soc/codecs/cs35l56-shared.c` (+4): `cs35l56_set_patch()`,
`cs35l56_set_fw_reg_table()`, `cs35l56_system_reset()`,
`cs35l56_hw_init()`
- `sound/soc/codecs/cs35l56.c` (+1): `_cs35l56_component_probe()`
- Scope: device-ID addition to an existing codec driver.
**Step 2.2 — Flow**
- Before: SoundWire ID `0x01FA:0x3562` is unmatched; type `0x62` / DEVID
`0x35A620` fall through to `-ENODEV` or skip CS35L63
firmware/reset/controls.
- After: ID binds; all CS35L63-specific switches also accept CS35L62 and
reuse `cs35l63_regmap_sdw`, `cs35l63_patch_fw`, `cs35l63_fw_reg`,
`cs35l63_system_reset_seq`, `cs35l63_calibration_controls`,
`cs35l63_controls`.
- Path: probe/init only for the new part. Existing 0x54/56/57/63 paths
are unchanged.
**Step 2.3 — Bug mechanism**
- Record: `[Hardware enablement / SoundWire device ID]` not a memory-
safety or locking bug. Mechanism: add `SDW_SLAVE_ENTRY(0x01FA, 0x3562,
0x3562)` and sibling `case` labels so CS35L62 is treated as CS35L63.
**Step 2.4 — Fix quality**
- Record: Obviously correct given the stated aliasing. Minimal. No
public API, no new driver. Regression risk for existing chips is very
low (new cases only). Completeness: other `type` switches in this
driver already use `default:` for CS35L63-class silicon, so they pick
up 0x62 without extra cases. I2C/SPI tables are intentionally
untouched (series is SoundWire-only). KUnit still special-cases only
0x63 in one test; that is test-only.
## Phase 3: Git history
**Step 3.1 — Blame**
- Record: CS35L63 cases/IDs came from `978858791ceda` (“Add initial
support for CS35L63…”, 2025-04-07). That commit is an ancestor of HEAD
and is in v6.16. cs35l56 itself dates to `e496112529006` (2023-03-20).
The CS35L62 lines are this commit (`3330b5f6d8dea`).
**Step 3.2 — Fixes: tag**
- Record: None. There is no “buggy commit”; support was simply never
added. N/A.
**Step 3.3 — File history / series**
- Record: Patch **1/2**. Companion `0306d21155819` (“ASoC: sdw_utils:
Add codec info for CS35L62”) adds the SOF/generic machine-driver table
entry. Cover letter: chip IDs in the codec driver **and** sdw_utils.
Follow-up `998b8a6c8aff2` sorted the SDW ID table (addresses Pierre-
Louis Bossart’s nit). Later cs35l56 changes are independent (IRQ,
probe deadlock, pm_runtime).
**Step 3.4 — Author**
- Record: Stefan Binding is a regular Cirrus cs35l56/cs35l41 contributor
(including the original CS35L63 support). Submitted via Richard
Fitzgerald; applied by Mark Brown. High confidence in the ID mapping.
**Step 3.5 — Prerequisites**
- Record: Needs CS35L63 symbols (`cs35l63_regmap_sdw`,
patch/fw/reset/calibration/controls). Those exist in this checkout.
Standalone for the codec driver; SOF auto-wiring wants patch 2/2 as
well.
## Phase 4: Mailing list / external
**Step 4.1 — b4 dig**
- Record: Lore/thread: `https://patch.msgid.link/20260715120135.939280-
2-rf@xxxxxxxxxxxxxxxxxxxxx`
- Series: **v1 only** (no v2/v3). Applied as submitted.
- Review: Pierre-Louis Bossart (SoundWire) only nitted sort order (62
after 63). No NAK, no stable nomination, no functional concern.
- Mark Brown: applied to `for-7.3`, with the standard note that merge-
window patches go in the next window “or sooner **if it is a bug
fix**” — i.e. this was **not** treated as a bugfix.
**Step 4.2 — Recipients (`b4 dig -w`)**
- Record: `broonie@xxxxxxxxxx`, `linux-sound@xxxxxxxxxxxxxxx`, `linux-
kernel@xxxxxxxxxxxxxxx`, `patches@xxxxxxxxxxxxxxxxxxxxx`. Maintainer
applied it.
**Step 4.3 — Bug report**
- Record: None. No syzbot, bugzilla, or user crash report.
**Step 4.4 — Related patches**
- Record: `[PATCH 0/2]` + `[PATCH 2/2]` sdw_utils codec_info. Both
applied together (`3330b5f6d8dea`, `0306d21155819`).
**Step 4.5 — stable@**
- Record: lore.kernel.org/stable WebFetch returned 403. **UNVERIFIED**
whether stable@ discussed it. Unlikely: 7.3 is still rc4 and this was
merge-window work.
## Phase 5: Semantic / call graph
**Step 5.1 — Functions**
- Record: `cs35l56_sdw_probe`, `cs35l56_sdw_id[]`, `cs35l56_set_patch`,
`cs35l56_set_fw_reg_table`, `cs35l56_system_reset`, `cs35l56_hw_init`,
`_cs35l56_component_probe`.
**Step 5.2 — Callers**
- Record: `cs35l56_sdw_probe` is the `sdw_driver.probe` (SoundWire
enumeration). `cs35l56_common_probe()` is called from SDW/I2C/SPI
probe. `cs35l56_hw_init` / `cs35l56_set_patch` are called from the
codec init path in `cs35l56.c`. `_cs35l56_component_probe` is the ALSA
component probe. Not a syscall hot path; probe/runtime of this amp.
**Step 5.3 — Callees**
- Record: `devm_regmap_init_sdw`, `cs35l56_common_probe`,
`regmap_register_patch`, `regmap_multi_reg_write_bypassed`,
`snd_soc_add_component_controls`. No new helpers.
**Step 5.4 — Reachability**
- Record: Reachable when a CS35L62 SoundWire slave enumerates
(laptops/SOF). Unprivileged userspace cannot trigger a crash path
here; without the ID the device simply does not bind.
**Step 5.5 — Similar patterns**
- Record: Same pattern as CS35L63 (`978858791ceda`) and CS35L57 IDs.
Remaining `case 0x57` blocks use `default:` for CS35L63-class parts,
so CS35L62 is covered there.
## Phase 6: This tree
**Step 6.1 — Local tree and whether the code exists**
- Record: `git describe HEAD` = **v7.3-rc4-27-gf0100363d8c37**; `make
kernelversion` = **7.3.0-rc4**. This is mainline 7.3-rc, not an LTS
branch.
- Driver, CS35L63 paths, and the new IDs all exist here. `git merge-base
--is-ancestor 978858791ceda HEAD` succeeds. `git merge-base --is-
ancestor 3330b5f6d8dea HEAD` succeeds. CS35L62 first appears in
**v7.3-rc1**, not v7.2.
**Step 6.2 — Backport difficulty**
- Record: Context here already includes later IRQ/ID-sort changes, but
the hunks are trivial case/ID inserts. Expected apply: clean or
trivial reorder. **Low difficulty.**
**Step 6.3 — Related fixes already here**
- Record: No alternate CS35L62 fix. Support is this series. ID table
sort came after.
## Phase 7: Subsystem
**Step 7.1**
- Record: `[ASoC codec driver, Cirrus CS35L56 family] [IMPORTANT —
common laptop speaker amps, not core mm/fs; not staging]`
**Step 7.2**
- Record: Actively maintained (many 2026 probe/IRQ/pm fixes). CS35L56 is
mature; CS35L62 is a small ID alias on top of CS35L63.
## Phase 8: Impact and risk
**Step 8.1 — Who is affected**
- Record: Driver-specific / config-specific
(`CONFIG_SND_SOC_CS35L56_SDW`). Only CS35L62 SoundWire hardware.
Existing CS35L54/56/57/63 behavior unchanged.
**Step 8.2 — Trigger**
- Record: Device enumeration of part `0x3562`. Not a race. Not
unprivileged exploit surface.
**Step 8.3 — Failure mode without the patch**
- Record: Device does not probe; no speaker amp. Severity **LOW** as a
kernel bug (no oops/corruption), **HIGH** as hardware enablement for
those machines.
**Step 8.4 — Benefit vs risk**
- Record: Benefit: CS35L62 SoundWire amps work on this kernel. Risk:
very low (7 lines, additive cases). Ratio: favorable **if** treating
this as a device-ID exception; poor **if** requiring a crash/security
bug.
## Phase 9: Synthesis
**Step 9.1 — Evidence**
For:
- Matches the documented **new device ID** exception (SoundWire ID +
cases routing to existing CS35L63 code).
- Small (7 lines), obviously correct, maintainer-applied.
- Driver and CS35L63 prerequisites exist in this 7.3 tree.
- No change to existing IDs’ control flow.
- Author is the Cirrus driver owner.
Against:
- Subject/body are **new feature**, not a bugfix. No crash, leak, UAF,
deadlock, or CVE.
- Mark Brown applied it to **for-7.3 merge window**, not as a stable
bugfix.
- Patch 1/2; SOF machine auto-config also wants `0306d21155819`.
- No `Reported-by` / no in-field bug.
- Stable rules 2 and 3 fail unless the ID exception is used.
Unresolved:
- stable@ lore search could not be fetched (403).
**Step 9.2 — Stable rules checklist**
1. Obviously correct and tested? **PASS** — aliasing is local and
reviewed; applied by broonie. Only review comment was sort order.
2. Fixes a real bug that affects users? **FAIL** as a defect; **PASS**
only under the device-ID exception (unsupported CS35L62).
3. Important issue (crash/security/corruption/deadlock)? **FAIL**
4. Small and contained? **PASS** (7 lines, one driver)
5. No new features or APIs? **FAIL** as a feature; **PASS** under
device-ID exception (no new driver/API, only an ID)
6. Can apply to the local tree? **PASS** — CS35L63 paths present on
7.3-rc4
**Step 9.3 — Exception**
- Record: **NEW DEVICE IDs** on an existing driver. Same class as adding
a PCI/USB ID. Not a new driver, not a new uAPI.
**Step 9.4 — Decision**
This tree is **7.3.0-rc4**. The cs35l56 driver and CS35L63 handling are
present, so the patch is applicable here. It does not fix a stability
bug, but it is the standard “add ID to existing driver” pattern the
stable rules explicitly allow: one SoundWire ID and a handful of `case`
labels so CS35L62 reuses CS35L63. Risk to current users is negligible.
Companion sdw_utils entry is separate and not required for the codec to
bind.
### What it solves for users of this tree
Without the ID, a CS35L62 SoundWire amp never matches the driver. With
it, the existing CS35L63 path runs. That is hardware enablement, not a
crash fix.
### Rules / risk
Meets “small, obviously correct, no new API”. Importance is hardware
support, covered by the device-ID exception. Backport/apply risk is very
low.
---
Verification:
- [Phase 1] Parsed tags: SOB Binding/Fitzgerald/Brown; Link to patch
1/2; no Fixes/Reported-by/stable Cc
- [Phase 2] Diff: +7 lines, ID `0x3562` + cases `0x62`/`0x35A620`
reusing CS35L63
- [Phase 3] `git blame`: CS35L63 from `978858791ceda` (v6.16); CS35L62
from `3330b5f6d8dea`
- [Phase 3] `git merge-base --is-ancestor 978858791ceda HEAD`: CS35L63
in this tree
- [Phase 3] `git merge-base --is-ancestor 3330b5f6d8dea v7.3-rc1`: first
in 7.3-rc1; not in v7.2
- [Phase 3] Series: 1/2 codec IDs, 2/2 `0306d21155819` sdw_utils
- [Phase 4] `b4 dig -c 3330b5f6d8dea`: v1 only; URL
patch.msgid.link/20260715120135.939280-2-…
- [Phase 4] `b4 dig -a`: no later revision
- [Phase 4] `b4 dig -w`: broonie + linux-sound
- [Phase 4] mbox: Bossart sort-order nit only; Brown applied to for-7.3
- [Phase 4] WebFetch lore.kernel.org: 403; used b4 mbox + openwall
instead
- [Phase 5] grep callers: SDW probe → `cs35l56_common_probe`;
hw_init/set_patch from codec init
- [Phase 5] Other `case 0x63` sites updated; remaining type switches use
`default:` for L63-class
- [Phase 6] `git describe HEAD` / `make kernelversion`: v7.3-rc4 /
7.3.0-rc4; files present
- [Phase 8] Failure mode: no bind for CS35L62, not an oops; existing IDs
unchanged
- UNVERIFIED: stable@ mailing list (lore 403)
**YES**
sound/soc/codecs/cs35l56-sdw.c | 2 ++
sound/soc/codecs/cs35l56-shared.c | 4 ++++
sound/soc/codecs/cs35l56.c | 1 +
3 files changed, 7 insertions(+)
diff --git a/sound/soc/codecs/cs35l56-sdw.c b/sound/soc/codecs/cs35l56-sdw.c
index a513a15d7e5a7..a7b4e2a923205 100644
--- a/sound/soc/codecs/cs35l56-sdw.c
+++ b/sound/soc/codecs/cs35l56-sdw.c
@@ -529,6 +529,7 @@ static int cs35l56_sdw_probe(struct sdw_slave *peripheral, const struct sdw_devi
regmap_config = &cs35l56_regmap_sdw;
break;
case 0x3563:
+ case 0x3562:
regmap_config = &cs35l63_regmap_sdw;
break;
default:
@@ -582,6 +583,7 @@ static const struct sdw_device_id cs35l56_sdw_id[] = {
SDW_SLAVE_ENTRY(0x01FA, 0x3556, 0x3556),
SDW_SLAVE_ENTRY(0x01FA, 0x3557, 0x3557),
SDW_SLAVE_ENTRY(0x01FA, 0x3563, 0x3563),
+ SDW_SLAVE_ENTRY(0x01FA, 0x3562, 0x3562),
{},
};
MODULE_DEVICE_TABLE(sdw, cs35l56_sdw_id);
diff --git a/sound/soc/codecs/cs35l56-shared.c b/sound/soc/codecs/cs35l56-shared.c
index 646bb4fbca025..3001ec45a72de 100644
--- a/sound/soc/codecs/cs35l56-shared.c
+++ b/sound/soc/codecs/cs35l56-shared.c
@@ -86,6 +86,7 @@ int cs35l56_set_patch(struct cs35l56_base *cs35l56_base)
ARRAY_SIZE(cs35l56_patch_fw));
break;
case 0x63:
+ case 0x62:
ret = regmap_register_patch(cs35l56_base->regmap, cs35l63_patch_fw,
ARRAY_SIZE(cs35l63_patch_fw));
break;
@@ -370,6 +371,7 @@ static void cs35l56_set_fw_reg_table(struct cs35l56_base *cs35l56_base)
}
break;
case 0x63:
+ case 0x62:
cs35l56_base->fw_reg = &cs35l63_fw_reg;
break;
}
@@ -576,6 +578,7 @@ void cs35l56_system_reset(struct cs35l56_base *cs35l56_base, bool is_soundwire)
}
break;
case 0x63:
+ case 0x62:
regmap_multi_reg_write_bypassed(cs35l56_base->regmap,
cs35l63_system_reset_seq,
ARRAY_SIZE(cs35l63_system_reset_seq));
@@ -1080,6 +1083,7 @@ int cs35l56_hw_init(struct cs35l56_base *cs35l56_base)
cs35l56_base->calibration_controls = &cs35l56_calibration_controls;
break;
case 0x35A630:
+ case 0x35A620:
cs35l56_base->calibration_controls = &cs35l63_calibration_controls;
devid = devid >> 4;
break;
diff --git a/sound/soc/codecs/cs35l56.c b/sound/soc/codecs/cs35l56.c
index e3334b2d84532..69976761ffa50 100644
--- a/sound/soc/codecs/cs35l56.c
+++ b/sound/soc/codecs/cs35l56.c
@@ -968,6 +968,7 @@ static int cs35l56_component_probe(struct snd_soc_component *component)
ARRAY_SIZE(cs35l56_controls));
break;
case 0x63:
+ case 0x62:
ret = snd_soc_add_component_controls(component, cs35l63_controls,
ARRAY_SIZE(cs35l63_controls));
break;
--
2.53.0