On Tue, May 3, 2011 at 2:26 PM, Kevin Wolf <kw...@redhat.com> wrote: > A thread should only be counted as idle when it really is waiting for new > requests. Without this patch, sometimes too few threads are started as busy > threads are counted as idle. > > Not sure if it makes a difference in practice outside some artificial > qemu-io/qemu-img tests, but I think the change makes sense in any case. > > Signed-off-by: Kevin Wolf <kw...@redhat.com> > --- > posix-aio-compat.c | 6 ++---- > 1 files changed, 2 insertions(+), 4 deletions(-)
I think the critical change here is that idle_threads is not being incremented by spawn_thread(). This means that we will keep spawning threads as new requests come in and until the first thread goes idle. Previously you could imagine a scenario where we spawn a thread but don't schedule it yet. Then we immediately submit more I/O and since idle_threads was incremented we don't spawn additional threads to handle the requests. Are these the cases you were thinking about? I like your patch because it reduces the number of places where we mess with idle_threads. I'd move the increment/decrement out of the loop since the mutex is held here anyway so no one will test idle_threads between loop iterations, but it's up to you if you want to send a v2 or not. Stefan
