Re: PROBLEM: mmc clock reduction error on NanoPi-R5S (rk3568)

+ Ulf's new mail

在 2026/07/21 星期二 3:56, Peter Taylor 写道:
> After the changes in 3009738a855c (Prevent illegal clock reduction in
> HS200/HS400 mode), when trying to change the drive's partition table
> using fdisk, it fails when I go to write. Here is the relevant dmesg 
> output.
> [ 19.509615] dwmmc_rockchip fe2b0000.mmc: IDMAC supports 32-bit address 
> mode.
> [ 19.510330] dwmmc_rockchip fe2b0000.mmc: Using internal DMA controller.
> [ 19.511727] dwmmc_rockchip fe2b0000.mmc: Version ID is 270a
> [ 19.512302] dwmmc_rockchip fe2b0000.mmc: DW MMC controller at irq 94,32 
> bit host data width,256 deep fifo
> [ 19.541960] mmc_host mmc1: Bus speed (slot 0) = 375000Hz (slot req 
> 400000Hz, actual 375000HZ div = 0)
> [ 19.563121] mmc0: SDHCI controller on fe310000.mmc [fe310000.mmc] using 
> ADMA
> [ 19.602217] mmc_host mmc1: Bus speed (slot 0) = 50000000Hz (slot req 
> 50000000Hz, actual 50000000HZ div = 0)
> [ 19.603277] mmc1: new high speed SDHC card at address 59b4
> [ 19.647750] mmcblk1: mmc1:59b4 SMI 7.50 GiB
> [ 19.652366] mmcblk1: p1
> [ 19.655584] sdhci-dwcmshc fe310000.mmc: Can't reduce the clock below 
> 52MHz in HS200/HS400 mode
> [ 19.656917] mmc0: new HS400 Enhanced strobe MMC card at address 0001

May I ask which dts file you're using now?
The RK3566/RK3568 SoCs only claim to support HS200.

> [ 19.658749] mmcblk0: mmc0:0001 hC8aP> 58.2 GiB
> [ 19.665252] mmcblk0:
> [ 19.665969] mmcblk0boot0: mmc0:0001 hC8aP> 4.00 MiB
> [ 19.669411] mmcblk0boot1: mmc0:0001 hC8aP> 4.00 MiB
> [ 19.672446] mmcblk0rpmb: mmc0:0001 hC8aP> 16.0 MiB, chardev (505:0)
> [ 462.450194] sdhci-dwcmshc fe310000.mmc: Can't reduce the clock below 
> 52MHz in HS200/HS400 mode
> [ 462.452168] I/O error, dev mmcblk0, sector 122142719 op 0x1:(WRITE) 
> flags 0x800800 phys_seg 1 prio class 2
> [ 462.453338] Buffer I/O error on dev mmcblk0, logical block 15267839, 
> lost async page write
> I have been testing on v7.1.3, and it is known good on v6.18.38.
> As the error also occurs at startup, my thought is it could be an 
> invalid value in
> the device tree, possibly the 24MHz value in rk356x-base.dtsi:1085 ?
>