[Libguestfs] [PATCH libnbd 0/2] Avoid lock and error overhead on some calls.

[Richard W.M. Jones]

On Thu, May 30, 2019 at 04:35:04PM +0100, Richard W.M. Jones wrote:
> This works.  I'm in the middle of testing whether there is any
> noticable benefit.

Is it faster?  It's not very clear, but here are some timings.  This
is with examples/threaded-reads-and-writes with the number of cycles
multipled by 100.

Before:

real	0m27.616s
real	0m27.417s
real	0m28.396s
real	0m28.772s
real	0m26.622s

After:

real	0m23.984s
real	0m24.357s
real	0m28.294s
real	0m27.036s
real	0m24.862s

The test only calls the nbd_aio_is_* functions in 4 places, but they
are called quite often.

I think this patch series is good, even if it's not going to fix your
10x slowdown :-/

Rich.

-- 
Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones
Read my programming and virtualization blog: http://rwmj.wordpress.com
virt-top is 'top' for virtual machines.  Tiny program with many
powerful monitoring features, net stats, disk stats, logging, etc.
http://people.redhat.com/~rjones/virt-top