You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
PTL: request topology + enablement for Senary SN6245 (0x6245) — MECHREVO YAOSHI, speakers silent, fallback to HDMI-only #6012
Machine: MECHREVO YAOSHI Series (Panther Lake, Core Ultra Series 3), DMI product_family PTL, product_name YAOSHI Series, board longname MECHREVO-YAOSHISeries-Standard-YAOSHISeries_XxPTL5xx
Audio: single SoundWire peripheral on link 3: link 3 mfg_id 0x0496 part_id 0x6245 version 0x3 (Senary SN624x family, SDCA version 3)
No analog HDA codec on the bus (codec mask 0x4 = iDisp only)
Problem
Kernel falls back to the HDMI-only HDA machine driver, so internal speakers/headphone jack are dead (Dummy Output in PipeWire):
sof-audio-pci-intel-ptl 0000:00:1f.3: No SoundWire machine driver found for the ACPI-reported configuration:
sof-audio-pci-intel-ptl 0000:00:1f.3: link 3 mfg_id 0x0496 part_id 0x6245 version 0x3
sof-audio-pci-intel-ptl 0000:00:1f.3: part id 0x6245 is not supported
sof-audio-pci-intel-ptl 0000:00:1f.3: using HDA machine driver skl_hda_dsp_generic now
Gaps identified (linux 7.2.15 and current master)
sound/soc/codecs/sn624x-sdca-sdw.c SDW id table lists 0x6242/0x6244/0x6247 but not 0x6245 (same version 0x3). The SN624x driver series (Qianghua Wang, merged 2026-09-01 for v7.4) covers the family but not this part id.
sound/soc/sdw_utils/soc_sdw_utils.c codec_info table likewise has no 0x6245 entry.
linux-firmware/sof-firmware ships no PTL topology mentioning sn624x at all (sof-ptl-sn624x*.tplg absent) — even with kernel entries added, a topology is required.
Request
A PTL topology for a single SN624x jack+amp+dmic codec on link 3 (sof-ptl-sn624x-l3 or similar; the SN6245 is register-compatible with the SN6244 per family documentation).
Kernel-side: SN6245 part id in the codec driver + sdw_utils codec_info (3-line additions; happy to send the patch — wanted to align topology naming with you first).
DMIC note: NHLT reports no DMICs (external mic presumably via the codec). BT link on ssp 0x4 per NHLT.
@wangxiaohu2026 Looks like a new id for the sn624x codec driver? @jfdnet for the topology issue, please apply the [PATCH] ASoC: Intel: soc-acpi: use dummy topology for SN624x patch and use the function topologies.
Hi Bard,
Yes — 0x6245 (SN6245) is a new part id in the SN624x SDCA family.
It is register-compatible with SN6244 for jack/amp/dmic, so we will add
0x6245 to the sn624x SDW id table, sdw_utils codec_info, and the PTL
match entries (alongside the dummy topology change).
We will send a follow-up patch after dummy topology, record 4channels issue merged in Linux kernel.
Thanks,
Qianghua
2. @wangxiaohu2026 Looks like a new id for the sn624x codec driver? @jfdnet for the topology issue, please apply the [PATCH] ASoC: Intel: soc-acpi: use dummy topology for SN624x patch and use the function topologies.
Please test it on top of a v7.4-rc1 (or later) tree that already contains
"[PATCH] ASoC: Intel: soc-acpi: use dummy topology for SN624x". No
sof-ptl-sn624x topology is required. With a current sof-firmware, the card
should probe through sof_sdw and the SDCA function topologies
(sof-sdca-jack, sof-sdca-1amp, sof-sdca-mic) plus HDMI.
These lines should be gone:
part id 0x6245 is not supported
using HDA machine driver skl_hda_dsp_generic now
Please reply with the sof-audio-pci-intel-ptl dmesg and whether the
internal speakers and the headset jack work.
Root cause: sound/soc/sdw_utils/soc_sdw_senary_sdca_jack_common.c:93 calls strstr(component->name_prefix, ...) with component->name_prefix == NULL (RDI is 0) — the v2 codec_info dais entry for 0x6245 apparently never gets a name_prefix bound on the PTL single-codec layout, and unlike the rt_sdca_jack equivalent there is no NULL guard before the loop.
The codec driver also logs parse fallbacks during probe (may or may not be related):
sn624x-sdca sdw:0:3:0496:6245:01: GE 35: malformed affected controls list
sn624x-sdca sdw:0:3:0496:6245:01: sn624x: sdca_parse_function(UAJ adr=1) failed (-22), keeping ACPI init table
sn624x-sdca sdw:0:3:0496:6245:01: FU 11: control node mipi-sdca-control-0xB-subproperties not found
sn624x-sdca sdw:0:3:0496:6245:01: sn624x: sdca_parse_function(SmartMic adr=2) failed (-22), keeping ACPI init table
sn624x-sdca sdw:0:3:0496:6245:01: FU 23: control node mipi-sdca-control-0xB-subproperties not found
sn624x-sdca sdw:0:3:0496:6245:01: sn624x: sdca_parse_function(SmartAmp adr=4) failed (-22)
Speakers / headset jack: untestable — the sof_sdw card never registers, so no analog PCM device exists. aplay -l only lists the four NVidia HDMI devices.
Full dmesg | grep sof-audio-pci-intel-ptl attached below; happy to provide the complete dmesg, .config or a test run of any follow-up patch — I kept the v7.4-rc2 tree with v2 applied around for quick rebuilds.
Root cause: sound/soc/sdw_utils/soc_sdw_senary_sdca_jack_common.c:93 calls strstr(component->name_prefix, ...) with component->name_prefix == NULL (RDI is 0) — the v2 codec_info dais entry for 0x6245 apparently never gets a name_prefix bound on the PTL single-codec layout, and unlike the rt_sdca_jack equivalent there is no NULL guard before the loop.
The codec driver also logs parse fallbacks during probe (may or may not be related):
sn624x-sdca sdw:0:3:0496:6245:01: GE 35: malformed affected controls list
sn624x-sdca sdw:0:3:0496:6245:01: sn624x: sdca_parse_function(UAJ adr=1) failed (-22), keeping ACPI init table
sn624x-sdca sdw:0:3:0496:6245:01: FU 11: control node mipi-sdca-control-0xB-subproperties not found
sn624x-sdca sdw:0:3:0496:6245:01: sn624x: sdca_parse_function(SmartMic adr=2) failed (-22), keeping ACPI init table
sn624x-sdca sdw:0:3:0496:6245:01: FU 23: control node mipi-sdca-control-0xB-subproperties not found
sn624x-sdca sdw:0:3:0496:6245:01: sn624x: sdca_parse_function(SmartAmp adr=4) failed (-22)
Speakers / headset jack: untestable — the sof_sdw card never registers, so no analog PCM device exists. aplay -l only lists the four NVidia HDMI devices.
Full dmesg | grep sof-audio-pci-intel-ptl attached below; happy to provide the complete dmesg, .config or a test run of any follow-up patch — I kept the v7.4-rc2 tree with v2 applied around for quick rebuilds.
@jfdnet Thanks for the trace. The Oops is in asoc_sdw_senary_sdca_jack_rtd_init(): strstr() is called with component->name_prefix == NULL.
A NULL check by itself does not register the card. When name_prefix is missing, the codec widgets are created as HP, MIC2, SPK and DMIC, while the routes look for "sn624x HP" and so on. The follow-up keeps the existing prefixed routes when name_prefix is "sn624x", and connects the bare widget names when it is NULL. card->components is still hs:sn624x-sdca and mic:sn624x-dmic.
Please apply this on top of the v2 tree you already have (v7.4-rc2 + https://lore.kernel.org/linux-sound/20261110034736.742465-1-qianghua.wang@senarytech.com/): https://lore.kernel.org/linux-sound/20261110073420.842175-1-qianghua.wang@senarytech.com/
The sdca_parse_function() -22 lines are the existing ACPI fallback and are not the Oops. After this patch the sof_sdw card should register. Please check /proc/asound/cards and whether the speakers and the headset jack work. If you see "sn624x jack: component ... has no name_prefix", please include that line. If the card registers but the speakers stay silent, please send the new dmesg — SmartAmp did not apply an init table on this BIOS.
This patch works well on the PTL platform PC with SN6244.
Thanks,
Qianghua
Follow-up patch tested (v7.4-rc2 + v2 + 842175). The Oops is gone and the diagnostic line fires, but card registration still fails with -ENODEV — and this time I have the exact reason.
sof_sdw sof_sdw: sn624x jack: component sdw:0:3:0496:6245:01 has no name_prefix
sof_sdw sof_sdw: ASoC: Failed to add route HP(*) -> Headphone
sof_sdw sof_sdw: ASoC: Failed to add route Headset Mic -> MIC2(*)
sof_sdw sof_sdw: senary sdca jack map addition failed: -19
SDW3-Playback-SimpleJack: ASoC error (-19): at snd_soc_link_init()
sof_sdw sof_sdw: error -ENODEV: snd_soc_register_card failed -19
Per the soc-dapm.c error format the (*) marks the missing endpoint — so the codec-side widgets "HP"/"MIC2" are not found while the jack pins are.
To make that unambiguous I instrumented asoc_sdw_senary_sdca_jack_rtd_init() with a for_each_card_widgets() dump right before snd_soc_dapm_add_routes():
At the moment SDW3-Playback-SimpleJack's rtd_init runs, the card widget list contains only the two jack pins — zero widgets from the codec component. The component itself exists (dai->component is sdw:0:3:0496:6245:01, prefix NULL), and the codec driver did probe far enough to log its init tables:
…yet none of sn624x_sdca_dapm_widgets[] (RX/TX/SPK RX/DMIC TX, PDE *, FU *, HP/SPK/MIC2/DMIC) is present in card->widgets when the first link inits. That is also consistent with the SmartAmp probe log on this BIOS, where the function was dropped without a fallback table:
sn624x-sdca sdw:0:3:0496:6245:01: FU 23: control node mipi-sdca-control-0xB-subproperties not found
sn624x-sdca sdw:0:3:0496:6245:01: sn624x: sdca_parse_function(SmartAmp adr=4) failed (-22)
Two questions this raises on my side (you'll know the answer immediately):
Is component DAPM widget creation expected to have happened before the first rtd_init on your SN6244 box — i.e. is this a probe/link-init ordering problem specific to the single-component multi-DAI SN6245 layout, or does the missing codec_conf (the same thing that leaves name_prefix NULL) also skip whatever attaches the codec's widgets to the card here?
Should sn624x_sdca_sdw_probe() be failing (or at least leaving widgets unregistered) when amp_func ends up NULL after the SmartAmp drop, instead of registering a widget-less component?
I still have the instrumented tree booted — tell me what else to dump (component probe order, w->dapm contexts, the dmic/amp dai rtd_inits) and I'll have it in minutes.
Full instrumentation dump as promised — three modules instrumented (codec driver, machine driver sof_sdw.c, sdw_utils), fresh boot. This one pretty much pins the failure: the codec_conf is correct and present, but the codec component is never initialized before the first rtd_init runs.
1. Codec driver registers its component early and cleanly (t=12.32)
Note the codec_conf entries exist with a dlc name that is character-for-character identical to the registered component name, prefix=sn624x. (Also note three conf entries for the same component — one per SDW function endpoint — in case that matters for soc_bind_component/prefix application.)
3. Yet at the first rtd_init the component is uninitialized (t=12.65, +130ms later)
sof_sdw sof_sdw: DBG rtd_init link=SDW3-Playback-SimpleJack dai=sn624x-sdca-aif(0) comp=sdw:0:3:0496:6245:01 prefix=0000000000000000
sof_sdw sof_sdw: DBG widget: Headphone (dapm 0xffff89e6894f3338)
sof_sdw sof_sdw: DBG widget: Headset Mic (dapm 0xffff89e6894f3338)
sof_sdw sof_sdw: DBG total widgets 2
sof_sdw sof_sdw: sn624x jack: component sdw:0:3:0496:6245:01 has no name_prefix
sof_sdw sof_sdw: ASoC: Failed to add route HP(*) -> Headphone
sof_sdw sof_sdw: ASoC: Failed to add route Headset Mic -> MIC2(*)
sof_sdw sof_sdw: senary sdca jack map addition failed: -19
sof_sdw sof_sdw: error -ENODEV: snd_soc_register_card failed -19
The only two widgets on the card at that point are the two jack pins, both on the same DAPM context (0xffff89e6894f3338, the card's). The sn624x component — registered 300 ms earlier with 16 declared widgets and three DAIs — has contributed zero widgets, and its name_prefix is still NULL despite three matching codec_conf entries.
What this says
The conf never gets applied and the component's DAPM widgets are never instantiated before snd_soc_link_init() of the first SDW3 link runs — i.e. whatever step of snd_soc_bind_card() is supposed to initialize this component and apply the prefix (component association / soc_bind_component / component init path) is not reaching the SDW component on this layout, while it evidently does on your SN6244 test box. Since the names all match exactly, a name-format mismatch is ruled out; the remaining suspects are the card's component-list association for the multi-DAI single-component SDW case, or ordering between component init and link init for it.
Everything is still instrumented and booted — happy to add a dump inside soc_bind_component/component-init next round if you want to see which lookup drops it.
Instrumentation used (on top of v7.4-rc2 + v2 + 842175)
sound/soc/intel/boards/sof_sdw.c (sof_card_dai_links_create()): dump all built dai_links and codec_confs
sound/soc/sdw_utils/soc_sdw_senary_sdca_jack_common.c (asoc_sdw_senary_sdca_jack_rtd_init()): dump link/dai/component/prefix, per-card-component widget counts (via w->dapm == comp->dapm pointer identity — snd_soc_dapm_context is private in 7.4-rc2), and the full card widget list
@jfdnet Thanks for testing v2 + the follow-up patch and for the detailed logs.
The Oops fix was only meant to handle a NULL name_prefix; your traces point to
codec widgets and prefix not being applied before the first rtd_init, which I
cannot reproduce on my SN6244 setup.
I do not have a MECHREVO / SN6245 unit here right now. I will verify and follow
up on this bind/init path once I have the same hardware. Until then I may not be
able to iterate quickly on your instrumented tree.
Retest update (2026-11-10): I repeated the test from a clean v7.4-rc2 tree with the SN6245 v2 part-ID patch and the missing-name_prefix follow-up (commits a15f57d56 and e10ee3314), using a full modules_install and no diagnostic instrumentation. The kernel boots to userspace and /boot mounts as vfat, so the earlier incomplete test-module installation is no longer a confound.
The SN6245 component binds at sdw:0:3:0496:6245:01 and the SOF function topologies load. This run does not Oops, but the card still fails registration. The follow-up sees the NULL name_prefix and selects the bare routes, which still fail:
sof_sdw: sn624x jack: component sdw:0:3:0496:6245:01 has no name_prefix
ASoC: Failed to add route HP(*) -> Headphone
ASoC: Failed to add route Headset Mic -> MIC2(*)
sof_sdw: senary sdca jack map addition failed: -19
sof_sdw: error -ENODEV: snd_soc_register_card failed -19
/proc/asound/cards and aplay -l show only NVIDIA HDMI; no analog or capture device. Probe also reports malformed GE 35 controls and UAJ/SmartMic/SmartAmp sdca_parse_function() failures (-22), keeping or lacking the ACPI init-table fallback as logged.
This confirms the full-module-install fix corrected the vfat failure, but the codec-widget/DAPM route failure persists with the clean upstream-patched build. Speakers, headset and DMIC remain untestable until the SOF card registers. Latest detailed tracker results: #5885 (comment)
@jfdnet Can you share the full dmesg log? I will try to figure out the no name_prefix issue. Where does the sof_sdw: sn624x jack: component sdw:0:3:0496:6245:01 has no name_prefix log from? I can't find the log from my kernel tree if I grep "has no name_prefix".
@jfdnet Thanks again for the testing and instrumentation on your MECHREVO YAOSHI.
We still do not have an SN6245 unit in our lab, so we cannot reproduce the
bind/name_prefix path locally yet. Could you please escalate with the brand /
vendor who sold you the machine? Once they provide us with a sample unit for
joint debugging, we can verify against your logs and follow up with a proper
fix. Until then we may not be able to iterate quickly on your instrumented tree.
Activity
bardliao commentedon Nov 3, 2026
@wangxiaohu2026 Looks like a new id for the sn624x codec driver?
@jfdnet for the topology issue, please apply the
[PATCH] ASoC: Intel: soc-acpi: use dummy topology for SN624xpatch and use the function topologies.wangxiaohu2026 commentedon Nov 3, 2026
Hi Bard,
Yes — 0x6245 (SN6245) is a new part id in the SN624x SDCA family.
It is register-compatible with SN6244 for jack/amp/dmic, so we will add
0x6245 to the sn624x SDW id table, sdw_utils codec_info, and the PTL
match entries (alongside the dummy topology change).
We will send a follow-up patch after dummy topology, record 4channels issue merged in Linux kernel.
Thanks,
Qianghua
wangxiaohu2026 commentedon Nov 10, 2026
@jfdnet The SN6245 part-id patch has been sent:
ASoC: sn624x: add support for the SN6245 part ID
https://lore.kernel.org/linux-sound/20261110032012.711191-1-qianghua.wang@senarytech.com
Please test it on top of a v7.4-rc1 (or later) tree that already contains
"[PATCH] ASoC: Intel: soc-acpi: use dummy topology for SN624x". No
sof-ptl-sn624x topology is required. With a current sof-firmware, the card
should probe through sof_sdw and the SDCA function topologies
(sof-sdca-jack, sof-sdca-1amp, sof-sdca-mic) plus HDMI.
These lines should be gone:
Please reply with the sof-audio-pci-intel-ptl dmesg and whether the
internal speakers and the headset jack work.
Thanks,
Qianghua
wangxiaohu2026 commentedon Nov 10, 2026
on ARL/LNL/MTL
Updated V2 patch for SN6245:
https://lore.kernel.org/linux-sound/20261110034736.742465-1-qianghua.wang@senarytech.com/
jfdnet commentedon Nov 10, 2026
Tested v2 on top of v7.4-rc2 (MECHREVO YAOSHI, PTL, BIOS N.1.05MRO13, sof-firmware 2026.09.2 / fw 2.15.0.1).
Part-ID matching works now — both old failure lines are gone, the SN6245 codec binds and the init tables are applied:
But card registration dies in a kernel Oops, so analog audio is still absent (
/proc/asound/cardsshows only HDA NVidia HDMI):Root cause:
sound/soc/sdw_utils/soc_sdw_senary_sdca_jack_common.c:93callsstrstr(component->name_prefix, ...)withcomponent->name_prefix == NULL(RDI is 0) — the v2 codec_info dais entry for 0x6245 apparently never gets a name_prefix bound on the PTL single-codec layout, and unlike thert_sdca_jackequivalent there is no NULL guard before the loop.The codec driver also logs parse fallbacks during probe (may or may not be related):
Speakers / headset jack: untestable — the sof_sdw card never registers, so no analog PCM device exists.
aplay -lonly lists the four NVidia HDMI devices.Full
dmesg | grep sof-audio-pci-intel-ptlattached below; happy to provide the complete dmesg, .config or a test run of any follow-up patch — I kept the v7.4-rc2 tree with v2 applied around for quick rebuilds.sof-audio-pci-intel-ptl dmesg
wangxiaohu2026 commentedon Nov 10, 2026
@jfdnet Thanks for the trace. The Oops is in asoc_sdw_senary_sdca_jack_rtd_init(): strstr() is called with component->name_prefix == NULL.
A NULL check by itself does not register the card. When name_prefix is missing, the codec widgets are created as HP, MIC2, SPK and DMIC, while the routes look for "sn624x HP" and so on. The follow-up keeps the existing prefixed routes when name_prefix is "sn624x", and connects the bare widget names when it is NULL. card->components is still hs:sn624x-sdca and mic:sn624x-dmic.
Please apply this on top of the v2 tree you already have (v7.4-rc2 + https://lore.kernel.org/linux-sound/20261110034736.742465-1-qianghua.wang@senarytech.com/):
https://lore.kernel.org/linux-sound/20261110073420.842175-1-qianghua.wang@senarytech.com/
The sdca_parse_function() -22 lines are the existing ACPI fallback and are not the Oops. After this patch the sof_sdw card should register. Please check /proc/asound/cards and whether the speakers and the headset jack work. If you see "sn624x jack: component ... has no name_prefix", please include that line. If the card registers but the speakers stay silent, please send the new dmesg — SmartAmp did not apply an init table on this BIOS.
This patch works well on the PTL platform PC with SN6244.
Thanks,
Qianghua
jfdnet commentedon Nov 10, 2026
Follow-up patch tested (v7.4-rc2 + v2 + 842175). The Oops is gone and the diagnostic line fires, but card registration still fails with -ENODEV — and this time I have the exact reason.
Per the
soc-dapm.cerror format the(*)marks the missing endpoint — so the codec-side widgets "HP"/"MIC2" are not found while the jack pins are.To make that unambiguous I instrumented
asoc_sdw_senary_sdca_jack_rtd_init()with afor_each_card_widgets()dump right beforesnd_soc_dapm_add_routes():At the moment SDW3-Playback-SimpleJack's rtd_init runs, the card widget list contains only the two jack pins — zero widgets from the codec component. The component itself exists (
dai->componentissdw:0:3:0496:6245:01, prefix NULL), and the codec driver did probe far enough to log its init tables:…yet none of
sn624x_sdca_dapm_widgets[](RX/TX/SPK RX/DMIC TX, PDE *, FU *, HP/SPK/MIC2/DMIC) is present incard->widgetswhen the first link inits. That is also consistent with the SmartAmp probe log on this BIOS, where the function was dropped without a fallback table:Two questions this raises on my side (you'll know the answer immediately):
rtd_initon your SN6244 box — i.e. is this a probe/link-init ordering problem specific to the single-component multi-DAI SN6245 layout, or does the missingcodec_conf(the same thing that leavesname_prefixNULL) also skip whatever attaches the codec's widgets to the card here?sn624x_sdca_sdw_probe()be failing (or at least leaving widgets unregistered) whenamp_funcends up NULL after the SmartAmp drop, instead of registering a widget-less component?I still have the instrumented tree booted — tell me what else to dump (component probe order,
w->dapmcontexts, the dmic/amp dai rtd_inits) and I'll have it in minutes.jfdnet commentedon Nov 10, 2026
Full instrumentation dump as promised — three modules instrumented (codec driver, machine driver
sof_sdw.c,sdw_utils), fresh boot. This one pretty much pins the failure: the codec_conf is correct and present, but the codec component is never initialized before the firstrtd_initruns.1. Codec driver registers its component early and cleanly (t=12.32)
2. Machine driver builds correct dai_links and codec_confs (t=12.65)
Note the codec_conf entries exist with a dlc name that is character-for-character identical to the registered component name,
prefix=sn624x. (Also note three conf entries for the same component — one per SDW function endpoint — in case that matters forsoc_bind_component/prefix application.)3. Yet at the first rtd_init the component is uninitialized (t=12.65, +130ms later)
The only two widgets on the card at that point are the two jack pins, both on the same DAPM context (
0xffff89e6894f3338, the card's). The sn624x component — registered 300 ms earlier with 16 declared widgets and three DAIs — has contributed zero widgets, and itsname_prefixis still NULL despite three matchingcodec_confentries.What this says
The conf never gets applied and the component's DAPM widgets are never instantiated before
snd_soc_link_init()of the first SDW3 link runs — i.e. whatever step ofsnd_soc_bind_card()is supposed to initialize this component and apply the prefix (component association /soc_bind_component/ component init path) is not reaching the SDW component on this layout, while it evidently does on your SN6244 test box. Since the names all match exactly, a name-format mismatch is ruled out; the remaining suspects are the card's component-list association for the multi-DAI single-component SDW case, or ordering between component init and link init for it.Everything is still instrumented and booted — happy to add a dump inside
soc_bind_component/component-init next round if you want to see which lookup drops it.Instrumentation used (on top of v7.4-rc2 + v2 + 842175)
sound/soc/codecs/sn624x-sdca-sdw.c: dev_warn before/afterdevm_snd_soc_register_component()sound/soc/intel/boards/sof_sdw.c(sof_card_dai_links_create()): dump all built dai_links and codec_confssound/soc/sdw_utils/soc_sdw_senary_sdca_jack_common.c(asoc_sdw_senary_sdca_jack_rtd_init()): dump link/dai/component/prefix, per-card-component widget counts (viaw->dapm == comp->dapmpointer identity —snd_soc_dapm_contextis private in 7.4-rc2), and the full card widget listwangxiaohu2026 commentedon Nov 10, 2026
@jfdnet Thanks for testing v2 + the follow-up patch and for the detailed logs.
The Oops fix was only meant to handle a NULL name_prefix; your traces point to
codec widgets and prefix not being applied before the first rtd_init, which I
cannot reproduce on my SN6244 setup.
I do not have a MECHREVO / SN6245 unit here right now. I will verify and follow
up on this bind/init path once I have the same hardware. Until then I may not be
able to iterate quickly on your instrumented tree.
Thanks again for your help on this.
Qianghua
jfdnet commentedon Nov 10, 2026
Retest update (2026-11-10): I repeated the test from a clean v7.4-rc2 tree with the SN6245 v2 part-ID patch and the missing-name_prefix follow-up (commits a15f57d56 and e10ee3314), using a full modules_install and no diagnostic instrumentation. The kernel boots to userspace and /boot mounts as vfat, so the earlier incomplete test-module installation is no longer a confound.
The SN6245 component binds at sdw:0:3:0496:6245:01 and the SOF function topologies load. This run does not Oops, but the card still fails registration. The follow-up sees the NULL name_prefix and selects the bare routes, which still fail:
/proc/asound/cards and aplay -l show only NVIDIA HDMI; no analog or capture device. Probe also reports malformed GE 35 controls and UAJ/SmartMic/SmartAmp sdca_parse_function() failures (-22), keeping or lacking the ACPI init-table fallback as logged.
This confirms the full-module-install fix corrected the vfat failure, but the codec-widget/DAPM route failure persists with the clean upstream-patched build. Speakers, headset and DMIC remain untestable until the SOF card registers. Latest detailed tracker results: #5885 (comment)
bardliao commentedon Nov 11, 2026
@jfdnet Can you share the full dmesg log? I will try to figure out the no name_prefix issue. Where does the
sof_sdw: sn624x jack: component sdw:0:3:0496:6245:01 has no name_prefixlog from? I can't find the log from my kernel tree if I grep "has no name_prefix".bardliao commentedon Nov 11, 2026
@jfdnet Please make sure CONFIG_SND_SOC_SN624X_SDCA_SDW is selected in your kconfig. Also, please add
to one .conf file in the
etc/modprobe.d/directory and share the dmesg log again.wangxiaohu2026 commentedon Nov 12, 2026
@jfdnet Thanks again for the testing and instrumentation on your MECHREVO YAOSHI.
We still do not have an SN6245 unit in our lab, so we cannot reproduce the
bind/name_prefix path locally yet. Could you please escalate with the brand /
vendor who sold you the machine? Once they provide us with a sample unit for
joint debugging, we can verify against your logs and follow up with a proper
fix. Until then we may not be able to iterate quickly on your instrumented tree.
Thanks for your patience.
Qianghua
ASoC: sn624x: add support for the SN6245 part ID