Mikulas Patocka wrote: > I looked at the patches. Thanks - the patchset was meant more like impulse to start discussion:-) This implementation uses just simple idea to reuse dm suspend mechanism for barrier processing. > I have thought about it and written an implementation of asynchronous > barriers --- it works by passing the requests to kdmflush and it returns > immediatelly from dm_request. I'll post it in a few days. Great. I'll wait for the patches then:) The internal thread was added to DM devices mainly because of implementing barriers but this intention was never finished... Milan -- mbroz@xxxxxxxxxx -- dm-devel mailing list dm-devel@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/dm-devel
