All of lore.kernel.org
 help / color / mirror / Atom feed
From: "Marc-André Lureau" <marcandre.lureau@gmail.com>
To: Eric Blake <eblake@redhat.com>
Cc: Michael Roth <mdroth@linux.vnet.ibm.com>,
	QEMU <qemu-devel@nongnu.org>,
	Markus Armbruster <armbru@redhat.com>
Subject: Re: [Qemu-devel] [PATCH v8 23/35] qmp: Tighten output visitor rules
Date: Tue, 5 Jan 2016 15:05:12 +0100	[thread overview]
Message-ID: <CAJ+F1C+waL=8th1hxaxEMvJAMSeGjLmZiGV6qYcttONCtkd9AQ@mail.gmail.com> (raw)
In-Reply-To: <1450717720-9627-24-git-send-email-eblake@redhat.com>

Hi

On Mon, Dec 21, 2015 at 6:08 PM, Eric Blake <eblake@redhat.com> wrote:
> Add a new qmp_output_visitor_reset(), which must be called before
> reusing an exising QmpOutputVisitor on a new root object.  Tighten
> assertions to require that qmp_output_get_qobject() can only be
> called after pairing a visit_end_* for every visit_start_* (rather
> than allowing it to return a partially built object), that it must
> not be called unless at least one visit_type_* or visit_start/
> visit_end pair has occurred since creation/reset (the accidental
> return of NULL fixed by commit ab8bf1d7 would have been much
> easier to diagnose), and that it may only be called once per visit.
>
> To keep the semantics of test_visitor_out_empty, we now have to
> explicitly request a top-level visit of a NULL object, by
> implementing the just-added visitor type_null() callback.
>
> Signed-off-by: Eric Blake <eblake@redhat.com>
>
> ---
> v8: rename qmp_output_reset to qmp_output_visitor_reset
> v7: new patch, based on discussion about spapr_drc.c
> ---
>  include/qapi/qmp-output-visitor.h |  1 +
>  include/qapi/visitor-impl.h       |  2 +-
>  qapi/qmp-output-visitor.c         | 37 ++++++++++++++++++++++++-------------
>  tests/test-qmp-output-visitor.c   |  2 ++
>  4 files changed, 28 insertions(+), 14 deletions(-)
>
> diff --git a/include/qapi/qmp-output-visitor.h b/include/qapi/qmp-output-visitor.h
> index 2266770..5093f0d 100644
> --- a/include/qapi/qmp-output-visitor.h
> +++ b/include/qapi/qmp-output-visitor.h
> @@ -21,6 +21,7 @@ typedef struct QmpOutputVisitor QmpOutputVisitor;
>
>  QmpOutputVisitor *qmp_output_visitor_new(void);
>  void qmp_output_visitor_cleanup(QmpOutputVisitor *v);
> +void qmp_output_visitor_reset(QmpOutputVisitor *v);
>
>  QObject *qmp_output_get_qobject(QmpOutputVisitor *v);
>  Visitor *qmp_output_get_visitor(QmpOutputVisitor *v);
> diff --git a/include/qapi/visitor-impl.h b/include/qapi/visitor-impl.h
> index d301901..ed2934b 100644
> --- a/include/qapi/visitor-impl.h
> +++ b/include/qapi/visitor-impl.h
> @@ -76,7 +76,7 @@ struct Visitor
>      void (*type_any)(Visitor *v, const char *name, QObject **obj,
>                       Error **errp);
>      /* Must be provided to visit explicit null values (right now, only the
> -     * dealloc visitor supports this).  */
> +     * dealloc and qmp-output visitors support this).  */
>      void (*type_null)(Visitor *v, const char *name, Error **errp);
>
>      /* May be NULL; most useful for input visitors. */
> diff --git a/qapi/qmp-output-visitor.c b/qapi/qmp-output-visitor.c
> index df22999..f2c39c5 100644
> --- a/qapi/qmp-output-visitor.c
> +++ b/qapi/qmp-output-visitor.c
> @@ -1,6 +1,7 @@
>  /*
>   * Core Definitions for QAPI/QMP Command Registry
>   *
> + * Copyright (C) 2015 Red Hat, Inc.
>   * Copyright IBM, Corp. 2011
>   *
>   * Authors:
> @@ -88,9 +89,8 @@ static void qmp_output_add_obj(QmpOutputVisitor *qov, const char *name,
>      cur = qmp_output_last(qov);
>
>      if (!cur) {
> -        /* FIXME we should require the user to reset the visitor, rather
> -         * than throwing away the previous root */
> -        qobject_decref(qov->root);
> +        /* Don't allow reuse of visitor on more than one root */
> +        assert(!qov->root);
>          qov->root = value;
>      } else {
>          switch (qobject_type(cur)) {
> @@ -202,18 +202,22 @@ static void qmp_output_type_any(Visitor *v, const char *name, QObject **obj,
>      qmp_output_add_obj(qov, name, *obj);
>  }
>
> +static void qmp_output_type_null(Visitor *v, const char *name, Error **errp)
> +{
> +    QmpOutputVisitor *qov = to_qov(v);
> +    qmp_output_add_obj(qov, name, qnull());
> +}
> +
>  /* Finish building, and return the root object. Will not be NULL. */
>  QObject *qmp_output_get_qobject(QmpOutputVisitor *qov)
>  {
> -    /* FIXME: we should require that a visit occurred, and that it is
> -     * complete (no starts without a matching end) */
> -    QObject *obj = qov->root;
> -    if (obj) {
> -        qobject_incref(obj);
> -    } else {
> -        obj = qnull();
> -    }
> -    return obj;
> +    QObject *root;
> +
> +    assert(qov->root);              /* A visit must have occurred...  */
> +    assert(!qmp_output_last(qov));  /* ...with each start paired with end.  */
> +    root = qov->root;
> +    qov->root = NULL;
> +    return root;
>  }
>
>  Visitor *qmp_output_get_visitor(QmpOutputVisitor *v)
> @@ -221,7 +225,7 @@ Visitor *qmp_output_get_visitor(QmpOutputVisitor *v)
>      return &v->visitor;
>  }
>
> -void qmp_output_visitor_cleanup(QmpOutputVisitor *v)
> +void qmp_output_visitor_reset(QmpOutputVisitor *v)
>  {
>      QStackEntry *e, *tmp;
>
> @@ -231,6 +235,12 @@ void qmp_output_visitor_cleanup(QmpOutputVisitor *v)
>      }
>
>      qobject_decref(v->root);
> +    v->root = NULL;
> +}
> +
> +void qmp_output_visitor_cleanup(QmpOutputVisitor *v)

It would make sense to call it _free() imho..

> +{
> +    qmp_output_visitor_reset(v);
>      g_free(v);
>  }
>
> @@ -252,6 +262,7 @@ QmpOutputVisitor *qmp_output_visitor_new(void)
>      v->visitor.type_str = qmp_output_type_str;
>      v->visitor.type_number = qmp_output_type_number;
>      v->visitor.type_any = qmp_output_type_any;
> +    v->visitor.type_null = qmp_output_type_null;
>
>      QTAILQ_INIT(&v->stack);
>
> diff --git a/tests/test-qmp-output-visitor.c b/tests/test-qmp-output-visitor.c
> index 26dc752..74d0ac4 100644
> --- a/tests/test-qmp-output-visitor.c
> +++ b/tests/test-qmp-output-visitor.c
> @@ -260,6 +260,7 @@ static void test_visitor_out_struct_errors(TestOutputVisitorData *data,
>          visit_type_UserDefOne(data->ov, "unused", &pu, &err);
>          g_assert(err);
>          error_free(err);
> +        qmp_output_visitor_reset(data->qov);
>      }
>  }
>
> @@ -459,6 +460,7 @@ static void test_visitor_out_empty(TestOutputVisitorData *data,
>  {
>      QObject *arg;
>
> +    visit_type_null(data->ov, NULL, &error_abort);

I guess this is going to be used outside of just this artificial case.
Otherwise, I would have suggested to get rid of it.

>      arg = qmp_output_get_qobject(data->qov);
>      g_assert(qobject_type(arg) == QTYPE_QNULL);
>      /* Check that qnull reference counting is sane */
> --
> 2.4.3
>
>

Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>



-- 
Marc-André Lureau

  reply	other threads:[~2016-01-05 14:05 UTC|newest]

Thread overview: 90+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2015-12-21 17:08 [Qemu-devel] [PATCH v8 00/35] qapi visitor cleanups (post-introspection cleanups subset E) Eric Blake
2015-12-21 17:08 ` [Qemu-devel] [PATCH v8 01/35] qobject: Document more shortcomings in our number handling Eric Blake
2016-01-05 14:07   ` Marc-André Lureau
2015-12-21 17:08 ` [Qemu-devel] [PATCH v8 02/35] qapi: Avoid use of misnamed DO_UPCAST() Eric Blake
2016-01-05 14:08   ` Marc-André Lureau
2015-12-21 17:08 ` [Qemu-devel] [PATCH v8 03/35] qapi: Drop dead dealloc visitor variable Eric Blake
2016-01-05 14:07   ` Marc-André Lureau
2015-12-21 17:08 ` [Qemu-devel] [PATCH v8 04/35] hmp: Improve use of qapi visitor Eric Blake
2016-01-05 14:06   ` Marc-André Lureau
2015-12-21 17:08 ` [Qemu-devel] [PATCH v8 05/35] vl: " Eric Blake
2016-01-05 14:06   ` Marc-André Lureau
2015-12-21 17:08 ` [Qemu-devel] [PATCH v8 06/35] balloon: " Eric Blake
2016-01-05 14:08   ` Marc-André Lureau
2015-12-21 17:08 ` [Qemu-devel] [PATCH v8 07/35] qapi: Improve generated event " Eric Blake
2016-01-05 14:07   ` Marc-André Lureau
2016-01-05 15:21     ` Eric Blake
2015-12-21 17:08 ` [Qemu-devel] [PATCH v8 08/35] qapi: Track all failures between visit_start/stop Eric Blake
2016-01-05 14:06   ` Marc-André Lureau
2015-12-21 17:08 ` [Qemu-devel] [PATCH v8 09/35] qapi: Prefer type_int64 over type_int in visitors Eric Blake
2016-01-05 14:07   ` Marc-André Lureau
2015-12-21 17:08 ` [Qemu-devel] [PATCH v8 10/35] qapi: Make all visitors supply uint64 callbacks Eric Blake
2016-01-05 14:07   ` Marc-André Lureau
2015-12-21 17:08 ` [Qemu-devel] [PATCH v8 11/35] qapi: Consolidate visitor small integer callbacks Eric Blake
2016-01-05 14:07   ` Marc-André Lureau
2015-12-21 17:08 ` [Qemu-devel] [PATCH v8 12/35] qapi: Don't cast Enum* to int* Eric Blake
2016-01-05 14:06   ` Marc-André Lureau
2016-01-05 15:23     ` Eric Blake
2015-12-21 17:08 ` [Qemu-devel] [PATCH v8 13/35] qom: Use typedef for Visitor Eric Blake
2016-01-05 14:07   ` Marc-André Lureau
2015-12-21 17:08 ` [Qemu-devel] [PATCH v8 14/35] qapi: Swap visit_* arguments for consistent 'name' placement Eric Blake
2016-01-05 14:06   ` Marc-André Lureau
2016-01-05 15:32     ` Eric Blake
2016-01-05 22:47       ` Eric Blake
2016-01-06  0:01   ` [Qemu-devel] [PATCH v8 14.5/35] qapi: Update docs to match recent generator changes Eric Blake
2016-01-06  0:16     ` Eric Blake
2015-12-21 17:08 ` [Qemu-devel] [PATCH v8 15/35] qom: Swap 'name' next to visitor in ObjectPropertyAccessor Eric Blake
2015-12-23 16:30   ` Eric Blake
2016-01-05 14:06     ` Marc-André Lureau
2015-12-21 17:08 ` [Qemu-devel] [PATCH v8 16/35] qapi: Swap 'name' in visit_* callbacks to match public API Eric Blake
2016-01-05 14:05   ` Marc-André Lureau
2015-12-21 17:08 ` [Qemu-devel] [PATCH v8 17/35] qapi: Drop unused 'kind' for struct/enum visit Eric Blake
2016-01-05 14:05   ` Marc-André Lureau
2016-01-06  0:26   ` Eric Blake
2015-12-21 17:08 ` [Qemu-devel] [PATCH v8 18/35] qapi: Drop unused error argument for list and implicit struct Eric Blake
2016-01-05 14:05   ` Marc-André Lureau
2016-01-05 15:58     ` Eric Blake
2015-12-21 17:08 ` [Qemu-devel] [PATCH v8 19/35] qmp: Fix reference-counting of qnull on empty output visit Eric Blake
2016-01-05 14:05   ` Marc-André Lureau
2016-01-06 17:42     ` Eric Blake
2015-12-21 17:08 ` [Qemu-devel] [PATCH v8 20/35] qmp: Don't abuse stack to track qmp-output root Eric Blake
2016-01-05 14:05   ` Marc-André Lureau
2015-12-21 17:08 ` [Qemu-devel] [PATCH v8 21/35] qapi: Document visitor interfaces, add assertions Eric Blake
2016-01-05 14:05   ` Marc-André Lureau
2015-12-21 17:08 ` [Qemu-devel] [PATCH v8 22/35] qapi: Add visit_type_null() visitor Eric Blake
2016-01-05 14:05   ` Marc-André Lureau
2016-01-05 16:08     ` Eric Blake
2016-01-06 22:15   ` [Qemu-devel] [PATCH v8 22.5/35] qmp: Support explicit null on input visit Eric Blake
2015-12-21 17:08 ` [Qemu-devel] [PATCH v8 23/35] qmp: Tighten output visitor rules Eric Blake
2016-01-05 14:05   ` Marc-André Lureau [this message]
2016-01-06 22:18     ` Eric Blake
2016-01-06 22:40   ` [Qemu-devel] [PATCH v8 23.5/35] qmp: Tighten output visitor rules, part 2 Eric Blake
2015-12-21 17:08 ` [Qemu-devel] [PATCH v8 24/35] spapr_drc: Expose 'null' in qom-get when there is no fdt Eric Blake
2015-12-21 17:08 ` [Qemu-devel] [PATCH v8 25/35] qapi: Simplify excess input reporting in input visitors Eric Blake
2016-01-05 14:05   ` Marc-André Lureau
2015-12-21 17:08 ` [Qemu-devel] [PATCH v8 26/35] qapi: Add type.is_empty() helper Eric Blake
2016-01-05 14:04   ` Marc-André Lureau
2016-01-05 16:10     ` Eric Blake
2015-12-21 17:08 ` [Qemu-devel] [PATCH v8 27/35] qapi: Fix command with named empty argument type Eric Blake
2016-01-05 14:04   ` Marc-André Lureau
2015-12-21 17:08 ` [Qemu-devel] [PATCH v8 28/35] qapi: Eliminate empty visit_type_FOO_fields Eric Blake
2016-01-05 14:04   ` Marc-André Lureau
2015-12-21 17:08 ` [Qemu-devel] [PATCH v8 29/35] qapi: Canonicalize missing object to :empty Eric Blake
2015-12-23 17:54   ` [Qemu-devel] [PATCH v8 29.5/35] fixup! " Eric Blake
2016-01-05 14:03   ` [Qemu-devel] [PATCH v8 29/35] " Marc-André Lureau
2015-12-21 17:08 ` [Qemu-devel] [PATCH v8 30/35] qapi-visit: Unify struct and union visit Eric Blake
2016-01-05 14:03   ` Marc-André Lureau
2015-12-21 17:08 ` [Qemu-devel] [PATCH v8 31/35] qapi: Rework deallocation of partial struct Eric Blake
2016-01-05 13:58   ` Marc-André Lureau
2015-12-21 17:08 ` [Qemu-devel] [PATCH v8 32/35] qapi: Split visit_end_struct() into pieces Eric Blake
2016-01-05 17:22   ` Marc-André Lureau
2015-12-21 17:08 ` [Qemu-devel] [PATCH v8 33/35] qapi: Simplify semantics of visit_next_list() Eric Blake
2016-01-05 17:22   ` Marc-André Lureau
2015-12-21 17:08 ` [Qemu-devel] [PATCH v8 34/35] qapi: Change visit_type_FOO() to no longer return partial objects Eric Blake
2016-01-05 17:22   ` Marc-André Lureau
2016-01-05 18:02     ` Eric Blake
2016-01-07 23:02   ` [Qemu-devel] [PATCH v8 34/35] fixup! " Eric Blake
2015-12-21 17:08 ` [Qemu-devel] [PATCH v8 35/35] RFC: qapi: Adjust layout of FooList types Eric Blake
2016-01-05 17:22   ` Marc-André Lureau
2016-01-08 16:45   ` [Qemu-devel] [PATCH] qapi: Update docs to match recent generated changes, part 2 Eric Blake
2016-01-19  9:10 ` [Qemu-devel] [PATCH v8 00/35] qapi visitor cleanups (post-introspection cleanups subset E) Markus Armbruster

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='CAJ+F1C+waL=8th1hxaxEMvJAMSeGjLmZiGV6qYcttONCtkd9AQ@mail.gmail.com' \
    --to=marcandre.lureau@gmail.com \
    --cc=armbru@redhat.com \
    --cc=eblake@redhat.com \
    --cc=mdroth@linux.vnet.ibm.com \
    --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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.