Re: [PATCH 0/4] qemu: fix block jobs when copy_on_read is enabled

On 23.01.20 16:57, Peter Krempa wrote: It isn’t. block-stream says: block-commit says: blockdev-mirror says: Note the subtle difference: “top” != “root”. “top” is meant just as part of the job: There’s a base image, and a top image. The top image is the target (for stream, that is), the base is the first image in the chain that will be kept. “root node” means actually a root node in the graph. Hence, stream doesn’t require @device to point to a root node. (Before qemu commit 554b6147650 it did, but that commit relaxed block-stream.) The plans are to fix it. The latest version of the series is here: https://lists.nongnu.org/archive/html/qemu-block/2019-08/msg00350.html It’s from August because it’s an extreme churn to work on new versions. It touches basically everything in the block layer, so rebasing it alone takes time. And then there are facts like reviews leading to new prerequesite series, namely: https://lists.nongnu.org/archive/html/qemu-block/2019-11/msg00370.html https://lists.nongnu.org/archive/html/qemu-block/2019-11/msg00908.html So in total, there are 96 patches. The churn for each version is hard. Finding willing reviewers is even harder. (I actually have basically no reviews for one of those prerequisite series (“block: Introduce real BdrvChildRole”) yet, so there’s little point in working on v7 of the main series.) Max