Littlefs Crash - reports volume full despite free space reported

[txf-]

I'm getting disk full errors in a littlefs partition despite the fact that the fact that reading esp_err_t ret = esp_littlefs_info("storagefs", &bytes_total, &bytes_used); does not indicate that the disk is full.

This is what I get when printing the results and the file with its size in the partition:

I (381) storage: flash usage
I (392) storage: Partition size: total: 1638400, used: 602112
I (392) storage: Partition size: total: 1638400, used: 602112
I (395) storage: list files
I (400) storage: scan file: MLOG0.db time:1757322637 size:581632

This is the crash stack trace:

--- 0x4208e6fe: _fwalk_sglue at /builds/idf/crosstool-NG/.build/riscv32-esp-elf/src/newlib/newlib/libc/stdio/fwalk.c:32
E (32797) task_wdt: Task watchdog got triggered. The following tasks/users did not reset the watchdog in time:
E (32797) task_wdt:  - IDLE (CPU 0)
E (32797) task_wdt: Tasks currently running:
E (32797) task_wdt: CPU 0: storage
E (32797) task_wdt: Print CPU 0 (current core) backtrace
Backtrace: 0x42079c80:0x4081a2b0 0x42079fc8:0x4081a340 0x4207a31e:0x4081a3c0 0x4207b6e6:0x4081a400 0x420766b2:0x4081a410 0x4207677c:0x4081a440 0x4208e6fe:0x4081d624 0x00000000:0x00000000
--- 0x42079c80: lfs_bd_prog at /home/tiago/Documents/transmitter/managed_components/joltwallet__littlefs/src/littlefs/lfs.c:251
--- 0x42079fc8: lfs_ctz_extend at /home/tiago/Documents/transmitter/managed_components/joltwallet__littlefs/src/littlefs/lfs.c:2955
--- (inlined by) lfs_file_flushedwrite at /home/tiago/Documents/transmitter/managed_components/joltwallet__littlefs/src/littlefs/lfs.c:3572
--- 0x4207a31e: lfs_file_write_ at /home/tiago/Documents/transmitter/managed_components/joltwallet__littlefs/src/littlefs/lfs.c:3622
--- 0x4207b6e6: lfs_file_sync_ at /home/tiago/Documents/transmitter/managed_components/joltwallet__littlefs/src/littlefs/lfs.c:3440
--- 0x420766b2: vfs_littlefs_write at /home/tiago/Documents/transmitter/managed_components/joltwallet__littlefs/src/esp_littlefs.c:1503
--- 0x4207677c: vfs_littlefs_ftruncate at /home/tiago/Documents/transmitter/managed_components/joltwallet__littlefs/src/esp_littlefs.c:2221
--- 0x4208e6fe: _fwalk_sglue at /builds/idf/crosstool-NG/.build/riscv32-esp-elf/src/newlib/newlib/libc/stdio/fwalk.c:32

I'm using ESP-IDF 5.5 and version 1.20.1 of the library on and ESP32-C6 with the following partition table:

# Name,   Type, SubType, Offset,  Size, Flags
# Note: if you change the phy_init or app partition offset, make sure to change the offset in Kconfig.projbuild
esp_secure_cert, 0x3F,    ,     0xD000,   0x2000,     
nvs,             data, nvs,     0x13000,  0x6000,
otadata,         data, ota,     0x19000,  0x2000,     
phy_init,        data, phy,     0x1b000,  0x1000,     
ota_0,           app,  ota_0,   0x20000,  0x190000,   
storagefs,       data, littlefs,  0x1b0000, 0x190000,   
storage,         data, nvs,     ,         0x10000,    
nvs_key,         data, nvs_keys,,         0x1000,     

[BrianPugh]

Your watchdog is triggering, which is indirectly related to the filesystem filling up. We opened up this upstream issue earlier this year.

Lemme see if I can whip up the solution i mentioned in the comments. In the meantime, try reseting your watchdog timer in your storage task.

[txf-]

Oh the watchdog is creating the crash, of course.

However this doesn't quite tackle the underlying condition. I'm handling the situation where the filesystem is getting full, by limiting additions to the file in case the used space is >95%. However I can't really handle it if it reports plenty of free space, but I get filesystem full errors. I'm getting a filesystem full with MLOG0.DB being just over a 3rd of the total file system size.

My current workaround is to report the maximum usable space as 600000 bytes, despite the filesystem size being a lot larger.

[BrianPugh]

what filesystem errors are you seeing?

[txf-]

what filesystem errors are you seeing?

E (5373201) esp_littlefs: ./managed_components/joltwallet__littlefs/src/littlefs/lfs.c:689:error: No more free space 0x7b

(I should note that this was logged in version 1.14, but the error is also triggered in 1.20.1)

[BrianPugh]

hmmm, sorry i don't have any ideas off the top of my head to what could be causing this. Are you 100% sure that such a large amount of space is being reported as free before the write? how big is the write?

[txf-]

Yes,

I'm pretty sure. If you see there is only one file in the filesystem. The used space is just the size of the file+overhead.

The file is an sqlite database, and the operation being performed is just adding a new entry to the database, I'd say 100 bytes at most (however if a new page is being created its a few KB).