I've noticed that master is broken when performance testing by writing to a large file. The following problem is also existing in v1.0.0 but happens after a much longer time.
My scenario is the following:
16GB SD-card
MBR with one partition 1GB
Formatted on Ubuntu using fdisk and mkfs.ext4
Insert card into board running an ST MCU using SDIO interface.
calling ext4_mbr_scan, ext4_mkfs_read_info, ext4_device_register, ext4_mount, ext4_recover, ext4_journal_start to mount it
calling ext4_fopen2 and continuously ext4_fwrite with a buffer of 32768 bytes.
Hi and thank you for an amazing project!
I've noticed that master is broken when performance testing by writing to a large file. The following problem is also existing in v1.0.0 but happens after a much longer time.
My scenario is the following:
fdiskandmkfs.ext4ext4_mbr_scan,ext4_mkfs_read_info,ext4_device_register,ext4_mount,ext4_recover,ext4_journal_startto mount itext4_fopen2and continuouslyext4_fwritewith a buffer of 32768 bytes.It works fine for a while but after about 20MB written, the block device write function gets called with an address outside of the SRAM of the MCU. After a little digging, size wraps around here because
block_size * fblock_countbecomes larger than size. https://github.com/gkostka/lwext4/blob/58bcf89a121b72d4fb66334f1693d3b30e4cb9c5/src/ext4.c#L1952I've tried with the provided
ext4_mbr_writeandext4_mkfsbut it's the same problem.