Messages in this thread | | | Date | Thu, 23 Apr 2009 10:10:06 -0400 | From | Theodore Tso <> | Subject | Re: [RFC PATCH] fpathconf() for fsync() behavior |
| |
On Thu, Apr 23, 2009 at 08:48:01AM -0400, Jeff Garzik wrote: > Theodore Tso wrote: >> So we can create a more finer-grained controlled system call --- >> although I would suggest that we just add some extra flags to >> sync_file_range() --- but it's doubtful that many application >> programmers will use it. > > sync_file_range() seems the obvious avenue for new fsync flags. > > I even explored what it would take to add a "flush storage dev writeback > cache, for this file" flag to sync_file_range(), rather unfortunately > non-trivial given the current implementation's close ties to MM.
What I had roughly in mind was some (optional) calls to the filesystem before and after the current implementations MM magic, but I haven't thought very deeply on the subject yet, mainly because...
> But yeah... how many people will use these fancy new flags and features? >
Yeah. That issue.
It would be nice to have some additional semantics, but in terms of priorities, it's not the highest thing on my list in terms of itches to scratch.
- Ted
| |