[REGRESSION] rtsx_usb_sdmmc: RTS5129 (0bda:0129) no longer detects SD card

[REGRESSION] rtsx_usb_sdmmc: RTS5129 (0bda:0129) no longer detects SD card

[Philipp]

Hi,

the SD card in the internal Realtek RTS5129 USB card reader
(0bda:0129) of my Lenovo V110-15IKB is no longer detected.

Works:  6.12.116, 7.2.x
Broken: 6.18.x, 7.3.3

On broken kernels rtsx_usb and rtsx_usb_sdmmc load and the platform
devices rtsx_usb_sdmmc.0.auto / rtsx_usb_ms.1.auto exist, but no
mmc host/card appears (no mmc0, no mmcblk0) and there are no errors
in dmesg. Reloading rtsx_usb / rtsx_usb_sdmmc, re-inserting the card
and disabling USB autosuspend (power/control=on) do not help.

Since 7.2.x worked but 6.18.x does not, I suspect a change that was
backported to stable.

Distribution: NixOS (distribution kernel, no custom patches).

#regzbot introduced: v6.12..v6.18

Thanks,
Philipp

Re: [REGRESSION] rtsx_usb_sdmmc: RTS5129 (0bda:0129) no longer detects SD card

[Thorsten Leemhuis]

On 11/8/26 19:51, Philipp wrote:
> 
> the SD card in the internal Realtek RTS5129 USB card reader
> (0bda:0129) of my Lenovo V110-15IKB is no longer detected.
> 
> Works:  6.12.116, 7.2.x
> Broken: 6.18.x, 7.3.3
> 
> On broken kernels rtsx_usb and rtsx_usb_sdmmc load and the platform
> devices rtsx_usb_sdmmc.0.auto / rtsx_usb_ms.1.auto exist, but no
> mmc host/card appears (no mmc0, no mmcblk0) and there are no errors
> in dmesg. Reloading rtsx_usb / rtsx_usb_sdmmc, re-inserting the card
> and disabling USB autosuspend (power/control=on) do not help.
> 
> Since 7.2.x worked but 6.18.x does not, I suspect a change that was
> backported to stable.

Might, but might not -- and I wonder if you have two bugs, some race, or
if some .config setting is disturbing something. Overall this situations
is somewhat puzzling. The best way to bring some light into this is afaics:

Check if 7.2 is working; if it is, bisect between that and 7.3.3; of
not, bisect between 7.2 and 7.2.x to find what fixes things. And ensure
to use the same .config as a base during the bisection, as explained here:
https://docs.kernel.org/admin-guide/verify-bugs-and-bisect-regressions.html

dmesg outputs from a working and a not-that-different version that is
broken (say 7.2 and 7.3.3) would be helpful, too.

Ciao, Thorsten

Re: [REGRESSION] rtsx_usb_sdmmc: RTS5129 (0bda:0129) no longer detects SD card

[Philipp Jonas]

On Wed Nov 11, 2026 at 10:06 AM CET, Thorsten Leemhuis wrote:
> On 11/8/26 19:51, Philipp wrote:
>> 
>> the SD card in the internal Realtek RTS5129 USB card reader
>> (0bda:0129) of my Lenovo V110-15IKB is no longer detected.
>> 
>> Works:  6.12.116, 7.2.x
>> Broken: 6.18.x, 7.3.3
>> 
>> On broken kernels rtsx_usb and rtsx_usb_sdmmc load and the platform
>> devices rtsx_usb_sdmmc.0.auto / rtsx_usb_ms.1.auto exist, but no
>> mmc host/card appears (no mmc0, no mmcblk0) and there are no errors
>> in dmesg. Reloading rtsx_usb / rtsx_usb_sdmmc, re-inserting the card
>> and disabling USB autosuspend (power/control=on) do not help.
>> 
>> Since 7.2.x worked but 6.18.x does not, I suspect a change that was
>> backported to stable.
>
> Might, but might not -- and I wonder if you have two bugs, some race, or
> if some .config setting is disturbing something. Overall this situations
> is somewhat puzzling. The best way to bring some light into this is afaics:
>
> Check if 7.2 is working; if it is, bisect between that and 7.3.3; of
> not, bisect between 7.2 and 7.2.x to find what fixes things. And ensure
> to use the same .config as a base during the bisection, as explained here:
> https://docs.kernel.org/admin-guide/verify-bugs-and-bisect-regressions.html
>
> dmesg outputs from a working and a not-that-different version that is
> broken (say 7.2 and 7.3.3) would be helpful, too.
>
> Ciao, Thorsten

Thanks Thorsten, will do. I'll check 7.2 first and report back
with dmesg outputs and, if needed, bisection results. This may
take a few days, as kernel builds take a while here.

Philipp

Re: [REGRESSION] rtsx_usb_sdmmc: RTS5129 (0bda:0129) no longer detects SD card

[Philipp Jonas]

On Wed Nov 11, 2026 at 10:06 AM CET, Thorsten Leemhuis wrote:
> On 11/8/26 19:51, Philipp wrote:
>> 
>> the SD card in the internal Realtek RTS5129 USB card reader
>> (0bda:0129) of my Lenovo V110-15IKB is no longer detected.
>> 
>> Works:  6.12.116, 7.2.x
>> Broken: 6.18.x, 7.3.3
>> 
>> On broken kernels rtsx_usb and rtsx_usb_sdmmc load and the platform
>> devices rtsx_usb_sdmmc.0.auto / rtsx_usb_ms.1.auto exist, but no
>> mmc host/card appears (no mmc0, no mmcblk0) and there are no errors
>> in dmesg. Reloading rtsx_usb / rtsx_usb_sdmmc, re-inserting the card
>> and disabling USB autosuspend (power/control=on) do not help.
>> 
>> Since 7.2.x worked but 6.18.x does not, I suspect a change that was
>> backported to stable.
>
> Might, but might not -- and I wonder if you have two bugs, some race, or
> if some .config setting is disturbing something. Overall this situations
> is somewhat puzzling. The best way to bring some light into this is afaics:
>
> Check if 7.2 is working; if it is, bisect between that and 7.3.3; of
> not, bisect between 7.2 and 7.2.x to find what fixes things. And ensure
> to use the same .config as a base during the bisection, as explained here:
> https://docs.kernel.org/admin-guide/verify-bugs-and-bisect-regressions.html
>
> dmesg outputs from a working and a not-that-different version that is
> broken (say 7.2 and 7.3.3) would be helpful, too.
>
> Ciao, Thorsten

Here are the relevant parts. Both kernels are NixOS builds from
nearly the same nixpkgs revision (Nov 5 vs. Nov 7), same config
generator, same hardware, card inserted during boot.

7.2.15 (works):

[    4.127442] usb 1-9: New USB device found, idVendor=0bda, idProduct=0129, bcdDevice=39.60
[    4.142705] usbcore: registered new interface driver rtsx_usb
[    5.580725] mmc0: new high speed SD card at address b368
[    5.622983] mmcblk0: mmc0:b368 SD 1.90 GiB

7.3.3 (broken):

[    4.128601] usb 1-9: New USB device found, idVendor=0bda, idProduct=0129, bcdDevice=39.60
[    4.143802] usbcore: registered new interface driver rtsx_usb

After that, nothing card related shows up at all on 7.3.3 -- no
mmc0, no errors, no timeouts. The rest of both logs is essentially
identical. Full logs available if needed.

To sum up what I tested:

  works:  6.12.116, 6.12.117, 7.2.11 - 7.2.15
  broken: 6.18.62, 7.3.1, 7.3.3

Since 7.2.15 works but 6.18.62 does not, my guess is that the
culprit went into 7.3 and was then backported to 6.18.y, but never
made it to 7.2.y before it went EOL.

Next I'll check vanilla 7.2 and then bisect v7.2..v7.3 as you
suggested. Builds take a while here, so this may take some days.

Philipp

Re: [REGRESSION] rtsx_usb_sdmmc: RTS5129 (0bda:0129) no longer detects SD card

[Philipp Jonas]

On Wed Nov 11, 2026 at 10:06 AM CET, Thorsten Leemhuis wrote:
> On 11/8/26 19:51, Philipp wrote:
>> 
>> the SD card in the internal Realtek RTS5129 USB card reader
>> (0bda:0129) of my Lenovo V110-15IKB is no longer detected.
>> 
>> Works:  6.12.116, 7.2.x
>> Broken: 6.18.x, 7.3.3
>> 
>> On broken kernels rtsx_usb and rtsx_usb_sdmmc load and the platform
>> devices rtsx_usb_sdmmc.0.auto / rtsx_usb_ms.1.auto exist, but no
>> mmc host/card appears (no mmc0, no mmcblk0) and there are no errors
>> in dmesg. Reloading rtsx_usb / rtsx_usb_sdmmc, re-inserting the card
>> and disabling USB autosuspend (power/control=on) do not help.
>> 
>> Since 7.2.x worked but 6.18.x does not, I suspect a change that was
>> backported to stable.
>
> Might, but might not -- and I wonder if you have two bugs, some race, or
> if some .config setting is disturbing something. Overall this situations
> is somewhat puzzling. The best way to bring some light into this is afaics:
>
> Check if 7.2 is working; if it is, bisect between that and 7.3.3; of
> not, bisect between 7.2 and 7.2.x to find what fixes things. And ensure
> to use the same .config as a base during the bisection, as explained here:
> https://docs.kernel.org/admin-guide/verify-bugs-and-bisect-regressions.html
>
> dmesg outputs from a working and a not-that-different version that is
> broken (say 7.2 and 7.3.3) would be helpful, too.
>
> Ciao, Thorsten

I found the culprit:

  7185b6d33b17 ("mmc: rtsx_usb_sdmmc: suppress false CD after init timeout")

Reverting just that commit on top of 7.3.4 fixes it. Same kernel
version and config, the revert being the only difference:

  7.3.4:           card not detected
  7.3.4 + revert:  card detected again

dmesg with the revert:

[    4.133620] usbcore: registered new interface driver rtsx_usb
[    5.582812] mmc0: new high speed SD card at address b368
[    5.638774] mmcblk0: mmc0:b368 SD 1.90 GiB

Without the revert, nothing card related shows up after the
rtsx_usb line.

Hardware: Lenovo V110-15IKB, internal Realtek RTS5129 (0bda:0129),
no tray, plain SD slot. The card is permanently inserted and holds
a LUKS key, so it has to be detected during boot.

Judging from the commit message, I assume an init command times out
before the card is attached, suppress_cd gets set, and as the card
is never removed, raw SD_CD never drops and the suppression is never
cleared. Note that physically removing and re-inserting the card on
7.3.3 did not bring it back either, so the clearing path does not
seem to work here as intended.

Sean, Greg, Ulf: CCing you as author/acker/committer.
Happy to test patches.

#regzbot introduced: 7185b6d33b17

Philipp