On Wed, Sep 08, 2021 at 09:10:20AM -0400, Emanuele Giuseppe Esposito wrote: > diff --git a/include/sysemu/block-backend-graph.h > b/include/sysemu/block-backend-graph.h > new file mode 100644 > index 0000000000..3310987b16 > --- /dev/null > +++ b/include/sysemu/block-backend-graph.h > @@ -0,0 +1,132 @@ > +/* > + * QEMU Block backends > + * > + * Copyright (C) 2014-2016 Red Hat, Inc. > + * > + * Authors: > + * Markus Armbruster <arm...@redhat.com>, > + * > + * This work is licensed under the terms of the GNU LGPL, version 2.1 > + * or later. See the COPYING.LIB file in the top-level directory. > + */ > + > +#ifndef BLOCK_BACKEND_GRAPH_H > +#define BLOCK_BACKEND_GRAPH_H > + > +#include "block-backend-common.h" > + > +/* > + * Graph API. These functions run under the BQL lock. > + * > + * If a function modifies the graph, it uses drain and/or > + * aio_context_acquire/release to be sure it has unique access.
It's not obvious why additional locking is needed if the BQL is already held. Please refer to the thread-safe I/O API functions that can be running concurrently without the BQL. > + * > + * All functions in this header must use this assertion: > + * g_assert(qemu_in_main_thread()); > + * to be sure they belong here. I suggest "to catch when they are accidentally called without the BQL". It explains the rationale whereas "to be sure they belong here" doesn't explain anything. > diff --git a/include/sysemu/block-backend-io.h > b/include/sysemu/block-backend-io.h > new file mode 100644 > index 0000000000..66a7bed9f0 > --- /dev/null > +++ b/include/sysemu/block-backend-io.h > @@ -0,0 +1,123 @@ > +/* > + * QEMU Block backends > + * > + * Copyright (C) 2014-2016 Red Hat, Inc. > + * > + * Authors: > + * Markus Armbruster <arm...@redhat.com>, > + * > + * This work is licensed under the terms of the GNU LGPL, version 2.1 > + * or later. See the COPYING.LIB file in the top-level directory. > + */ > + > +#ifndef BLOCK_BACKEND_IO_H > +#define BLOCK_BACKEND_IO_H > + > +#include "block-backend-common.h" > + > +/* > + * I/O API functions. These functions are thread-safe, and therefore > + * can run in any AioContext. "can run in any AioContext" makes me wonder what the exact requirements are. Can they run in any *thread* (regardless of whether an AioContext even exists for that thread) or do they need to run in a thread that has called qemu_set_current_aio_context()?
signature.asc
Description: PGP signature
