From mboxrd@z Thu Jan  1 00:00:00 1970
Received: from eggs.gnu.org ([2001:4830:134:3::10]:45762)
	by lists.gnu.org with esmtp (Exim 4.71)
	(envelope-from <paolo.bonzini@gmail.com>) id 1bFIV9-0004IU-F5
	for qemu-devel@nongnu.org; Tue, 21 Jun 2016 06:01:48 -0400
Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71)
	(envelope-from <paolo.bonzini@gmail.com>) id 1bFIV7-0008Qf-G6
	for qemu-devel@nongnu.org; Tue, 21 Jun 2016 06:01:46 -0400
Sender: Paolo Bonzini <paolo.bonzini@gmail.com>
References: <1466016055-31351-1-git-send-email-clord@redhat.com>
	<20160617095451.GC6994@stefanha-x1.localdomain>
	<e111e4d3-dd4a-b8c7-8169-fedd820a07d6@redhat.com>
	<20160621093252.GF32560@stefanha-x1.localdomain>
From: Paolo Bonzini <pbonzini@redhat.com>
Message-ID: <c8b669ee-b073-efb9-5a59-797f3c15071c@redhat.com>
Date: Tue, 21 Jun 2016 12:01:35 +0200
MIME-Version: 1.0
In-Reply-To: <20160621093252.GF32560@stefanha-x1.localdomain>
Content-Type: text/plain; charset=windows-1252
Content-Transfer-Encoding: 8bit
Subject: Re: [Qemu-devel] [PATCH 0/2] Dynamic module loading for block
 drivers
List-Id: <qemu-devel.nongnu.org>
List-Unsubscribe: <https://lists.nongnu.org/mailman/options/qemu-devel>,
	<mailto:qemu-devel-request@nongnu.org?subject=unsubscribe>
List-Archive: <http://lists.nongnu.org/archive/html/qemu-devel/>
List-Post: <mailto:qemu-devel@nongnu.org>
List-Help: <mailto:qemu-devel-request@nongnu.org?subject=help>
List-Subscribe: <https://lists.nongnu.org/mailman/listinfo/qemu-devel>,
	<mailto:qemu-devel-request@nongnu.org?subject=subscribe>
To: Stefan Hajnoczi <stefanha@gmail.com>, Colin Lord <clord@redhat.com>
Cc: kwolf@redhat.com, den@openvz.org, qemu-devel@nongnu.org, qemu-block@nongnu.org, mreitz@redhat.com



On 21/06/2016 11:32, Stefan Hajnoczi wrote:
> I think the issue comes from the fact that you are considering something
> like load_block_module(const char *filename) as the API instead of
> request_block_driver(const char *driver_name).  In the latter case it's
> possible to return a BlockDriver pointer.  In the former it's not.
> 
> The request_block_driver() approach requires a mapping from block driver
> names to modules.  This can be achieved using a directory layout with
> symlinks (hmm...Windows portability?):
> 
>   /usr/lib/qemu/block/
>     +--- sheepdog.so
>     +--- by-protocol/
>       +--- sheepdog+unix -> ../sheepdog.so
> 
> request_block_driver() would look at
> /usr/lib/qemu/block/by-protocol/<protocol> to find the module file.

Another possibility is to add a ".loaded" element to the 
block_driver_modules[] array and break the recursion (or infinite loop):


retry:
    QLIST_FOREACH(drv1, &bdrv_drivers, list) {
        if (!strcmp(drv1->format_name, format_name)) {
            return drv1;
        }
    }

    for (i = 0; i < ARRAY_SIZE(block_driver_modules); ++i) {
        if (!block_driver_modules[i].loaded &&
            !strcmp(block_driver_modules[i].format_name, format_name)) {
            block_driver_modules[i].loaded = true;
	    block_module_load_one(block_driver_modules[i].library_name);
            goto retry;
        }
    }

BTW, please give a name to block_driver_modules[x]'s type, so that you
can assign &block_driver_modules[i] to a pointer and use that as a
shortcut.

Thanks,

Paolo