On 10/02/16 23:06, Junichi Nomura wrote:
So I don't think the patch makes sense...
Hello Junichi,I had overlooked that pg_init_done() holds m->lock while manipulating m->pg_init_in_progress. So now I agree with you that this patch should not change the behavior of the dm-mpath driver. I will continue my search for what is causing the high number of multipath_busy() calls.
Bart. -- dm-devel mailing list dm-devel@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/dm-devel
