search

armink / FlashDB

An ultra-lightweight database that supports key-value and time series data | 一款支持 KV 数据和时序数据的超轻量级数据库
Apache License 2.0
1.94k stars 436 forks source link

fdb_write_gran应该怎么理解？ #167

Open as564335sa opened 2 years ago

as564335sa commented 2 years ago

大佬，fdb_write_gran 这个宏应该怎么理解？它的作用到底是什么啊，和实际flash的页，扇区、单次写入大小有什么关联？看源码也没找到有什么关联呢，它是怎么确定的

armink commented 2 years ago

这个是 Flash 的 写粒度 ，即写 Flash 时的最小单位

benpicco commented 1 year ago

What does a setting of 1 mean? Even on RAM one can't address individual bits - is write_gran in meant to mean Bytes instead?

armink commented 1 year ago

What does a setting of 1 mean? Even on RAM one can't address individual bits - is write_gran in meant to mean Bytes instead?

Yes, Flash and RAM are addressed by byte. After the NorFlash is erased, each bit can be individually modified to 0. In FlashDB the write granularity is 1 bit. For example: address 0, write 0xFE first, then write 0xFC.

benpicco commented 1 year ago

Ok, but it looks like .write_gran is not used anywhere outside of one demo application. Is this intentional, is it only a hit to the application and not used by FlashDB at all?

armink commented 1 year ago

The write_gran must be defined in fdb_cfg.h https://github.com/armink/FlashDB/blob/master/inc/fdb_cfg.h#L32

kleinsand commented 1 year ago

这个是 Flash 的 写粒度 ，即写 Flash 时的最小单位

如果芯片没有uint8，uint8都被typedef成了uint16，每次写入都是写入16bit，这个值fdb_write_gran是设置成8吗？@armink