Device loses barrier support

[Eric Sandeen]

Mikulas Patocka wrote:

> Another thing:
> 
> I'm wondering, where in fsync() does Linux wait for hardware disk cache to 
> be flushed? Isn't there a bug that fsync() will return before the cache is 
> flushed? I couldn't really find it. The last thing do_fsync calls is 
> filemap_fdatawait and it doesn't do cache flush (blkdev_issue_flush).

ext4, reiserfs, and xfs all call blkdev_issue_flush() in their ->fsync
file operations (or down that path).

-Eric