Brent Fulgham <[EMAIL PROTECTED]> writes: > For instance, while editing files in vim, I notice that each > keystroke I make (while writing -- not just moving the cursor) > is mirrored by a "tick" noise from my hard disk (i.e., a write > action is occuring).
I'm not sure exactly why a disk I/O needs to be happening on each such read; basically: each character is a completed read call each completed read call means the atime must be updated the filesystem ends up getting a file_utimes call on the terminal's "underlying node" which holds the atime information for the tty but hurd/libdiskfs/file-utimes.c does not do any sync at all, so there's no reason disk I/O should get queued for any of this. (Though it is still dirty, and the next time the filesystem gets synced you will certainly see a disk I/O.) What are the options to your root filesystem? Do `fsysopts /' and maybe we can see something odd. Thomas
