Hello, On Wed, Sep 16, 2026 at 10:57:33PM +0200, gianfranco giorgianni wrote: > Reopening per the maintainer's request in the closing message — this > reproduces exactly on the current bookworm-backports kernel. > > Hardware: Samsung Galaxy Book 4 (DMI: SAMSUNG ELECTRONICS CO., LTD. > 750XGK/NP750XGJ-KG5IT) > Reader: Realtek RTS5129 (usb 0bda:0129) > Kernel: 6.12.95+deb12-amd64 (bookworm-backports) > > Same failure signature as the original report, continuous on every card > insertion: > > mmc0: Card stuck being busy! __mmc_poll_for_busy > mmc0: error -110 whilst initialising SD card
110 is a timeout error. Can you confirm that happens with several different cards? > Regarding the commit mentioned earlier in this thread > (4dad599b8b5d1ffc5ef12a2edb13d15d537202ba): I checked it, and it's a > runtime-PM/polling-efficiency change (avoid polling when no card is > present), not a fix for card initialization. It's already present in > 6.12.95 and the failure still occurs, so that commit does not resolve this > bug. > > This confirms the RTS5129 incompatibility in > rtsx_usb_sdmmc/rtsx_usb_ms/rtsx_usb (originally introduced in the > 3.15->3.16 transition, see the 2015 LKML thread [1] and Ubuntu #1579748 > [2]) is still present, and that OEMs are still shipping this exact > controller in new laptops as of 2023-2024 — so this will keep affecting new > hardware, not just legacy devices. > > Workaround: reverting to the old staging-era driver still works with zero > source changes required against 6.12. Built and confirmed working via > https://github.com/aurorafossorg/rts5139 (DKMS-packaged, survives kernel > updates automatically). I've opened a PR there documenting this exact > configuration: https://github.com/ljmf00/rts5139/pull/3 (pending merge). > Blacklisting rtsx_usb_sdmmc/rtsx_usb_ms/rtsx_usb and loading that driver > instead restores full read/write access. > > Given this is an eleven-year-old regression that appears to be a genuine > architectural gap (rtsx_usb_sdmmc is missing logic like rtsx_usb_get_rsp() > that exists in the old driver), I don't expect an upstream fix soon, but > wanted to leave this here for the next person who searches this error on > newer hardware. I think upstream should hear about that. To prepare a mail, can you please verify that the issue also happens on a backport kernel (7.2.x)? I guess yes, but upstream (understandably) doesn't like reports about old kernels. Best regards Uwe
signature.asc
Description: PGP signature
