On Mon, Oct 22, 2012 at 10:15:00AM -0500, Dave Kleikamp wrote:

This is the current version of the patchset I presented at the LSF-MM
Summit in San Francisco in April. I apologize for letting it go so
long before re-submitting.

This patchset was begun by Zach Brown and was originally submitted for
review in October, 2009. Feedback was positive, and I have picked up
where he left off, porting his patches to the latest mainline kernel
and adding support more file systems.

This patch series adds a kernel interface to fs/aio.c so that kernel code can
issue concurrent asynchronous IO to file systems.  It adds an aio command and
file system methods which specify io memory with pages instead of userspace
addresses.

This series was written to reduce the current overhead loop imposes by
performing synchronus buffered file system IO from a kernel thread.  These
patches turn loop into a light weight layer that translates bios into iocbs.

The downside of this is that in its current implementation, performance takes
a big hit for non-synchonous I/O, since the underlying page cache is bypassed.
The tradeoff is that all writes to the loop device make it to the underlying
media, making loop-mounted file systems recoverable.

It also seems to still not fully kill thr old aio_read/write codepath.
At least XFS isn't touched yet.  It also doesn't seem to kill the nasty
hack for in-kernel direct I/O introduced with the swap over nfs code
(grep for REQ_KERNEL / KERNEL_READ / KERNEL_WRITE)