Date: Fri, 28 Aug 1998 14:18:22 -0700 (PDT) From: Julian Elischer <julian@whistle.com> To: current@FreeBSD.ORG Subject: FS gurus please.. Message-ID: <Pine.BSF.3.95.980828140739.4962A-100000@current1.whistle.com>
next in thread | raw e-mail | index | archive | help
Here's a question for those who have their heads more around the filesystems.. Why does MFS unconditionally call spec_fsync() on all vnodes? One presumes that since it's a memory filesysytem, fsync should do nothing.. I'm also curious why it doesn't even check that it's a vnode for a special device. What do we gain by calling spec_fsync on a normal file in an MFS? I can understand that possibly it should be trying to sync the superblocks etc from mounted devices where the device is on an MFS /dev. In that case shouldn't it be checking the vnode types? This has been worked on by several people so hopefully there is someone out there who can give me answers tothese questions.. also: MFS also calls spec_getpages in the same way, which eventually comes back through a VOP_STRATEGY to mfs_strategey(). does this make sense? I'm trying to make systems that don't HAVE specfs and also to fix an interaction between specfs and devfs/slice. It appears that the fsync() interaction could easily be solved by makinf mfs_fsync() a nop (why not?) and the mfs_getpages dependency should be fixed by importing the relevant code from spec_getpages into mfs_getpages and optimising it.. Can I have comments from the gallery please? I'm a mu limits of understanding on this and probably need assistance. What am I missing....? julian To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-current" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.3.95.980828140739.4962A-100000>