how to use NAND Controller in linux kernel 5.4-stm32mp-r2?

[jandriea]

I have modified STM32MP157C-DK2 board with NAND flash connected through FMC and it works properly with linux kernel 5.4-stm32mp-r1. I notice that there are some changes to FMC in 5.4-stm32mp-r2 and tried to build my custom linux using the newer release. Unfortunately, the kernel always crashed because it can't open the MTD device. The device can boot TF-A and u-boot properly but it always stuck at the kernel.

in 5.4-stm32mp-r1 version I only modify the DK2 device tree by stripping unused nodes, removing nodes that use the same pin as FMC, and adding FMC node by copying the configuration from stm32mp15xx-evx.dtsi. I use the same approach in 5.4-stm32mp-r2 as I assume the configuration from stm32mp15xx-evx.dtsi is enough to make the device boot from NAND.

[    1.158408] UBI error: cannot open mtd UBI, error -2
[    1.162737] stm32_rtc 5c004000.rtc: setting system clock to 2000-01-16T08:50:47 UTC (948012647)
[    1.173134] VFS: Cannot open root device "(null)" or unknown-block(0,0): error -6
[    1.179147] Please append a correct "root=" boot option; here are the available partitions:
[    1.187592] 0100           65536 ram0
[    1.187596]  (driver?)
[    1.193642] 0101           65536 ram1
[    1.193645]  (driver?)
[    1.199654] 0102           65536 ram2
[    1.199657]  (driver?)
[    1.205750] 0103           65536 ram3
[    1.205754]  (driver?)
[    1.211906] 0104           65536 ram4
[    1.211910]  (driver?)
[    1.217898] 0105           65536 ram5
[    1.217901]  (driver?)
[    1.223980] 0106           65536 ram6
[    1.223983]  (driver?)
[    1.230269] 0107           65536 ram7
[    1.230273]  (driver?)
[    1.236141] 0108           65536 ram8
[    1.236144]  (driver?)
[    1.242243] 0109           65536 ram9
[    1.242246]  (driver?)
[    1.248303] 010a           65536 ram10
[    1.248306]  (driver?)
[    1.254473] 010b           65536 ram11
[    1.254476]  (driver?)
[    1.260653] 010c           65536 ram12
[    1.260656]  (driver?)
[    1.266808] 010d           65536 ram13
[    1.266811]  (driver?)
[    1.272977] 010e           65536 ram14
[    1.272980]  (driver?)
[    1.279144] 010f           65536 ram15
[    1.279147]  (driver?)
[    1.285319] Kernel panic - not syncing: VFS: Unable to mount root fs on unknown-block(0,0)
[    1.293566] CPU: 0 PID: 1 Comm: swapper Not tainted 5.4.56 #1
[    1.299293] Hardware name: STM32 (Device Tree Support)
[    1.304450] [<c010d5e0>] (unwind_backtrace) from [<c010b01c>] (show_stack+0x10/0x14)
[    1.312163] [<c010b01c>] (show_stack) from [<c0116e08>] (panic+0x114/0x2fc)
[    1.319118] [<c0116e08>] (panic) from [<c0901318>] (mount_block_root+0x248/0x2f8)
[    1.326586] [<c0901318>] (mount_block_root) from [<c0901718>] (prepare_namespace+0x180/0x1c4)
[    1.335097] [<c0901718>] (prepare_namespace) from [<c06157d8>] (kernel_init+0x8/0x110)
[    1.343003] [<c06157d8>] (kernel_init) from [<c01010e8>] (ret_from_fork+0x14/0x2c)
[    1.350551] Exception stack(0xc7033fb0 to 0xc7033ff8)
[    1.355594] 3fa0:                                     00000000 00000000 00000000 00000000
[    1.363763] 3fc0: 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000
[    1.371927] 3fe0: 00000000 00000000 00000000 00000000 00000013 00000000
[    1.378534] ---[ end Kernel panic - not syncing: VFS: Unable to mount root fs on unknown-block(0,0) ]---

I have another DK2 board without any modification and the kernel runs without any problem

[Olivier GALLIEN]

Hi @jandriea​ ,

Moving to  5.4-stm32mp-r2  did you also upgradre TF-A and U-boot with V2.1 version ?

It's recommended to be consistent and not mix element from various deliveries.

Olivier

[jandriea]

Hi @Community member​ 

Thanks for your response. I am pretty sure that I also upgrade the TF-A and U-Boot as I clone the meta-st-stm32mp layer from STMicroelectronis github site and checked out openstlinux-5.4-dunfell-mp1-20-11-12 tag.

The bootloader runs properly as it can run from NAND and I can load the zImage and kernel device tree from U-Boot. The problem only occurred when the kernel is already loaded. As I search on another site, someone mentioned that the problem is the kernel didn't detect NAND flash. Thus, the kernel failed to load the UBI volume