Re: invalidate: busy-buffer

[Andreas Dilger]

On Feb 22, 2002  06:17 +1100, Andrew Clausen wrote:
> On Sat, Feb 09, 2002 at 08:38:42PM -0700, Andreas Dilger wrote:
> > No, this is a recent "addition" to the kernel.  Basically, it happens
> > when you try BLKFSLBUF on a device that has locked and/or dirty buffers
> > that cannot be flushed.  On any ext2/ext3 (or probably other filesystems)
> > the kernel is holding buffers busy that cannot be dropped.
> 
> Hmmm, so it looks like _flush_cache() (in libparted/linux.c) shouldn't
> flush on mounted devices.  (It's probably pointless anyway)

Probably not.  Given that the filesystem is mounted, it probably won't
change much w.r.t. superblock data, excluding free blocks and such.
For online ext2 resizing, it will sync the data to disk anyways (I
assume resierfs does the same) to ensure consistency, so the flush
is really only for fs-internals (data, inodes, etc) which parted should
not really care about.

Cheers, Andreas
--
Andreas Dilger
http://sourceforge.net/projects/ext2resize/
http://www-mddsp.enel.ucalgary.ca/People/adilger/