search

pellepl / spiffs

Wear-leveled SPI flash file system for embedded devices
MIT License
1.5k stars 402 forks source link

File missing #146

Open BelMoretto opened 7 years ago

BelMoretto commented 7 years ago

Hi all, I've experienced a file unrecognizing during a "power loss test".

The program flow was the following:

  1. power up
  2. perform SPIFFS_check() function
  3. check the presence of files
  4. check the content of files
  5. Rewrite all files (deleting the previous one)
  6. return to 3

Meanwhile intentional power_loss/restart occurs. I noticed that when power loss occurs during the execution of SPIFFS_checks() it can happen that a file written in the previous cycle is unrecognized. Is this a well known issue? Thanks in advance. M.

MCU Kinetis KV5x Family

pellepl commented 7 years ago

Hmm, what does unrecognized mean in this context? It disappears totally?

I guess it could happen during stage 5 if the powerloss comes precisely between the deletion of the old version and the creation of the new version. Rewrite via O_CREAT | O_TRUNC first deletes, and then creates new in a non-atomic way.

clmnin commented 6 years ago

Hey, I'm facing a similar issue. When I reset the MCU the file system (occasionally) outputs an error -10004 (ERR_DELETED). This is not seen every time I tried out this test. Also seen when I tried the "power loss test". Here's what I'm doing:

  1. Mount
  2. Open file (previously made)
  3. Seek to the beginning of file and read - - - > Fails here
  4. Seek to the beginning and write new data into flash
  5. Flush
  6. Go to step 3