search

zephyrproject-rtos / zephyr

Primary Git Repository for the Zephyr Project. Zephyr is a new generation, scalable, optimized, secure RTOS for multiple hardware architectures.
https://docs.zephyrproject.org
Apache License 2.0
10.68k stars 6.53k forks source link

LittleFS samples silently fail on the NXP MIMXRT1020-EVK board #78729

Open luispimo opened 3 weeks ago

luispimo commented 3 weeks ago

UART samples work correctly, but all the LittleFS samples fail on this board (eg samples/subsys/shell/fs).

However, a few months ago, everything worked correctly.

To Reproduce

west build -p always -b mimxrt1020_evk samples/subsys/shell/fs
...
[182/182] Linking C executable zephyr/zephyr.elf
Memory region         Used Size  Region Size  %age Used
           FLASH:       97992 B         8 MB      1.17%
             RAM:       31200 B        32 MB      0.09%
            ITCM:        6232 B        64 KB      9.51%
            DTCM:          0 GB        64 KB      0.00%
           OCRAM:          0 GB       128 KB      0.00%
        IDT_LIST:          0 GB        32 KB      0.00%
west flash -r jlink
-- west flash: rebuilding
ninja: no work to do.
-- west flash: using runner jlink
-- runners.jlink: reset after flashing requested
-- runners.jlink: JLink version: 7.96f
-- runners.jlink: Flashing file: /Users/luis/Dev/zephyr-workspace/zephyr/build/zephyr/zephyr.bin

Nothing appears on the serial console, not even the boot message.

The debugger exists due to a signal.

nixward commented 3 weeks ago

You could try git bisecting this @luispimo to find the feature breaking commit.

luispimo commented 3 weeks ago

Apparently, the problem is not in Zephyr itself, as I have tried versions 3.5, 3.6, and 3.7, obtaining the same result. It must be in the NXP hal module or in the littlefs module.

nakato commented 1 week ago

I wonder if I've been hitting the same bug. I had created a discussion about it at https://github.com/zephyrproject-rtos/zephyr/discussions/79033

I trigger it a bit differently, using fstab automount, and I have managed to trigger the issue on both a RP2040 and NRF52840 based board. But, it does present the same, as the lockup during mkfs happens so quickly there's no output unless you start putting delays in.

I've found I can get logs out of the console on boot by inserting a healthy amount of k_sleep(K_MSEC(1000)) into the littlefs format functions, but I've yet to work out what is going on. I can put a log message and sleep on line 1991, and get it, but a log message and sleep at the very top of the lfs_dir_traverse never shows up, which is called at 1993 during mkfs, which I don't get how that happens.

Edit: I downgraded zephyrproject-rtos/littlefs to commit 652f2c5, which has resulted in a working littlefs.

luispimo commented 1 week ago

Downgrading the LittleFS module to 652f2c5 does not work for the mimxrt1020_evk board when running LittleFS examples.

Edit: littlefs examples works on v3.6-branch (v3.6.0-118-g6937dabf26f2).

luispimo commented 1 week ago

Update: Downgrading hal_nxp to d45b14c1 solves the problem and littlefs examples work again.