qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Eric Blake <eblake@redhat.com>
To: Max Reitz <mreitz@redhat.com>, qemu-block@nongnu.org
Cc: qemu-devel@nongnu.org, Kevin Wolf <kwolf@redhat.com>,
	Paolo Bonzini <pbonzini@redhat.com>,
	"Daniel P . Berrange" <berrange@redhat.com>,
	Markus Armbruster <armbru@redhat.com>,
	Luiz Capitulino <lcapitulino@redhat.com>
Subject: Re: [Qemu-devel] [PATCH v3 07/14] block/nbd: "address" in nbd_refresh_filename()
Date: Tue, 14 Jun 2016 17:03:47 -0600	[thread overview]
Message-ID: <57608D53.9040808@redhat.com> (raw)
In-Reply-To: <1459967330-4573-8-git-send-email-mreitz@redhat.com>

[-- Attachment #1: Type: text/plain, Size: 4551 bytes --]

On 04/06/2016 12:28 PM, Max Reitz wrote:
> As of a future patch, the NBD block driver will accept a SocketAddress
> structure for a new "address" option. In order to support this,
> nbd_refresh_filename() needs some changes.
> 
> The two TODOs introduced by this patch will be removed in the very next
> one. They exist to explain that it is currently impossible for
> nbd_refresh_filename() to emit an "address.*" option (which the NBD
> block driver does not handle yet). The next patch will arm these code
> paths, but it will also enable handling of these options.
> 
> Signed-off-by: Max Reitz <mreitz@redhat.com>
> ---
>  block/nbd.c | 84 ++++++++++++++++++++++++++++++++++++++++++++-----------------
>  1 file changed, 61 insertions(+), 23 deletions(-)
> 
> diff --git a/block/nbd.c b/block/nbd.c
> index 1736f68..3adf302 100644
> --- a/block/nbd.c
> +++ b/block/nbd.c
> @@ -441,37 +441,75 @@ static void nbd_attach_aio_context(BlockDriverState *bs,
>  static void nbd_refresh_filename(BlockDriverState *bs, QDict *options)
>  {
>      QDict *opts = qdict_new();
> -    const char *path   = qdict_get_try_str(options, "path");
> -    const char *host   = qdict_get_try_str(options, "host");
> -    const char *port   = qdict_get_try_str(options, "port");
> +    bool can_generate_filename = true;
> +    const char *path = NULL, *host = NULL, *port = NULL;
>      const char *export = qdict_get_try_str(options, "export");
>      const char *tlscreds = qdict_get_try_str(options, "tls-creds");
>  
> -    if (host && !port) {
> -        port = stringify(NBD_DEFAULT_PORT);
> +    if (qdict_get_try_str(options, "address.type")) {
> +        /* This path will only be possible as of a future patch;
> +         * TODO: Remove this note once it is */
> +
> +        const char *type = qdict_get_str(options, "address.type");
> +

Oh, I'm sooooo tempted to teach the QAPI generator how to make a
discriminated union have a default 'type' value (thus making the
discriminator optional), so that we don't need a layer of nesting behind
'address.'.

> +        if (!strcmp(type, "unix")) {
> +            path = qdict_get_str(options, "address.data.path");
> +        } else if (!strcmp(type, "inet")) {
> +            host = qdict_get_str(options, "address.data.host");
> +            port = qdict_get_str(options, "address.data.port");

It's especially annoying that because SocketAddress is not flat, we have
to expose the 'data.' layer of nesting, even if we could avoid the
'address.' layer.

> +
> +            can_generate_filename = !qdict_haskey(options, "address.data.to")
> +                                 && !qdict_haskey(options, "address.data.ipv4")
> +                                 && !qdict_haskey(options, "address.data.ipv6");
> +        } else {
> +            can_generate_filename = false;
> +        }
> +    } else {
> +        path = qdict_get_try_str(options, "path");
> +        host = qdict_get_try_str(options, "host");
> +        port = qdict_get_try_str(options, "port");
> +
> +        if (host && !port) {
> +            port = stringify(NBD_DEFAULT_PORT);
> +        }
>      }

Looks clean given the constraints of what you are able to use from QAPI.

> +
> +    if (qdict_get_try_str(options, "address.type")) {
> +        /* This path will only be possible as of a future patch;
> +         * TODO: Remove this note once it is */
> +
> +        const QDictEntry *e;
> +        for (e = qdict_first(options); e; e = qdict_next(options, e)) {
> +            if (!strncmp(e->key, "address.", 8)) {
> +                qobject_incref(e->value);
> +                qdict_put_obj(opts, e->key, e->value);
> +            }
> +        }

This part makes me wonder if we want Dan's qdict_crumple() working first.

>      } else {
> -        qdict_put(opts, "host", qstring_from_str(host));
> -        qdict_put(opts, "port", qstring_from_str(port));
> +        if (path) {
> +            qdict_put(opts, "path", qstring_from_str(path));
> +        } else {
> +            qdict_put(opts, "host", qstring_from_str(host));
> +            qdict_put(opts, "port", qstring_from_str(port));
> +        }
>      }
>      if (export) {
>          qdict_put(opts, "export", qstring_from_str(export));
> 

At this point, I'll reserve giving R-b until I've seen the whole series
(it may need rebasing anyways...)

-- 
Eric Blake   eblake redhat com    +1-919-301-3266
Libvirt virtualization library http://libvirt.org


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 604 bytes --]

  reply	other threads:[~2016-06-14 23:03 UTC|newest]

Thread overview: 26+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-04-06 18:28 [Qemu-devel] [PATCH v3 for-2.7 00/14] qapi: Allow blockdev-add for NBD Max Reitz
2016-04-06 18:28 ` [Qemu-devel] [PATCH v3 01/14] qdict: Add qdict_change_key() Max Reitz
2016-06-14 22:03   ` Eric Blake
2016-04-06 18:28 ` [Qemu-devel] [PATCH v3 02/14] block/nbd: Drop trailing "." in error messages Max Reitz
2016-06-14 22:04   ` Eric Blake
2016-04-06 18:28 ` [Qemu-devel] [PATCH v3 03/14] block/nbd: Reject port parameter without host Max Reitz
2016-04-06 18:28 ` [Qemu-devel] [PATCH v3 04/14] block/nbd: Default port in nbd_refresh_filename() Max Reitz
2016-06-14 22:39   ` Eric Blake
2016-04-06 18:28 ` [Qemu-devel] [PATCH v3 05/14] block/nbd: Use qdict_put() Max Reitz
2016-06-14 22:40   ` Eric Blake
2016-04-06 18:28 ` [Qemu-devel] [PATCH v3 06/14] block/nbd: Add nbd_has_filename_options_conflict() Max Reitz
2016-06-14 22:54   ` Eric Blake
2016-04-06 18:28 ` [Qemu-devel] [PATCH v3 07/14] block/nbd: "address" in nbd_refresh_filename() Max Reitz
2016-06-14 23:03   ` Eric Blake [this message]
2016-04-06 18:28 ` [Qemu-devel] [PATCH v3 08/14] block/nbd: Accept SocketAddress Max Reitz
2016-06-14 23:14   ` Eric Blake
2016-06-15 14:40     ` Max Reitz
2016-04-06 18:28 ` [Qemu-devel] [PATCH v3 09/14] block/nbd: Use SocketAddress options Max Reitz
2016-04-06 18:28 ` [Qemu-devel] [PATCH v3 10/14] qapi: Allow blockdev-add for NBD Max Reitz
2016-04-06 18:28 ` [Qemu-devel] [PATCH v3 11/14] iotests.py: Add qemu_nbd function Max Reitz
2016-04-06 18:28 ` [Qemu-devel] [PATCH v3 12/14] iotests.py: Allow concurrent qemu instances Max Reitz
2016-04-06 18:28 ` [Qemu-devel] [PATCH v3 13/14] socket_scm_helper: Accept fd directly Max Reitz
2016-04-06 18:28 ` [Qemu-devel] [PATCH v3 14/14] iotests: Add test for NBD's blockdev-add interface Max Reitz
2016-05-03 13:51 ` [Qemu-devel] [PATCH v3 for-2.7 00/14] qapi: Allow blockdev-add for NBD Max Reitz
2016-05-03 15:23 ` Kevin Wolf
2016-06-14 22:42   ` Eric Blake

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=57608D53.9040808@redhat.com \
    --to=eblake@redhat.com \
    --cc=armbru@redhat.com \
    --cc=berrange@redhat.com \
    --cc=kwolf@redhat.com \
    --cc=lcapitulino@redhat.com \
    --cc=mreitz@redhat.com \
    --cc=pbonzini@redhat.com \
    --cc=qemu-block@nongnu.org \
    --cc=qemu-devel@nongnu.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).