3 comments

[ 3.0 ms ] story [ 12.4 ms ] thread
Thought that number sounded crazy, here's the important bit that explains it:

> Rather than doing many one-cluster allocations, the few lines of code makes it allocate multiple clusters in one call.

Since FAT32 doesn’t journal I wonder if this affects the chances for corrupting data.
Considering the Linux block elevator, page cache, and write bundling, no more or less than any other non-journaled filesystem configuration on Linux.