From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:43941) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1aqOkP-0001xE-JP for qemu-devel@nongnu.org; Wed, 13 Apr 2016 13:38:41 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1aqOkI-00014R-RG for qemu-devel@nongnu.org; Wed, 13 Apr 2016 13:38:37 -0400 Received: from mx1.redhat.com ([209.132.183.28]:11717) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1aqOkI-00014L-Jp for qemu-devel@nongnu.org; Wed, 13 Apr 2016 13:38:30 -0400 From: Markus Armbruster References: <1460131992-32278-1-git-send-email-eblake@redhat.com> <1460131992-32278-8-git-send-email-eblake@redhat.com> Date: Wed, 13 Apr 2016 19:38:27 +0200 In-Reply-To: <1460131992-32278-8-git-send-email-eblake@redhat.com> (Eric Blake's message of "Fri, 8 Apr 2016 10:13:00 -0600") Message-ID: <87a8kxflyk.fsf@dusky.pond.sub.org> MIME-Version: 1.0 Content-Type: text/plain Subject: Re: [Qemu-devel] [PATCH v14 07/19] qmp-input: Refactor when list is advanced List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Eric Blake Cc: qemu-devel@nongnu.org, Michael Roth Eric Blake writes: > Refactor the code to advance the QListEntry pointer at the point > where visit_type_FOO() is called, rather than visit_next_list(). > This will allow a future patch to move the visit of the list head > into visit_start_list(), and get rid of the 'first' flag. > > Signed-off-by: Eric Blake After having read the patch, I think the change is from this sequence of states start_list next_list type_ELT ... next_list type_ELT end_list entry NULL 1st elt 1st elt last elt last elt gone where type_ELT() returns (entry ? entry : 1st elt) and next_list() steps entry to this one start_list next_list type_ELT ... next_list type_ELT end_list entry 1st elt 1nd elt 2nd elt last elt NULL gone where type_ELT() steps entry and returns the old entry, and next_list() leaves entry alone. Correct? Now have a look at the typical generated visit: void visit_type_FOOList(Visitor *v, const char *name, FOOList **obj, Error **errp) { Error *err = NULL; FOOList *tail; size_t size = sizeof(**obj); visit_start_list(v, name, (GenericList **)obj, size, &err); if (err) { goto out; } for (tail = *obj; tail; tail = (FOOList *)visit_next_list(v, (GenericList *)tail, size)) { visit_type_FOO(v, NULL, &tail->value, &err); if (err) { break; } } if (visit_end_list(v) && err) { qapi_free_FOOList(*obj); *obj = NULL; } out: error_propagate(errp, err); } While the loop variable is still stepped in the for's third expression as it should, the actual stepping below the hood now happens in the loop body. I find this mildly confusing. Perhaps a more natural loop would be void visit_type_FOOList(Visitor *v, const char *name, FOOList **obj, Error **errp) { Error *err = NULL; GenericList *tail; size_t size = sizeof(**obj); visit_start_list(v, name, (GenericList **)obj, size, &err); if (err) { goto out; } while ((tail = visit_next_list(v, tail, size))) { visit_type_FOO(v, NULL, &((FOOList *)tail)->value, &err); if (err) { break; } } if (visit_end_list(v) && err) { qapi_free_FOOList(*obj); *obj = NULL; } out: error_propagate(errp, err); } Might also permit de-duplicating the g_malloc0(size), since we now call next_list() *before* each iteration instead of *between* iterations. > --- > v14: no change > v13: no change > v12: new patch > --- > qapi/qmp-input-visitor.c | 40 +++++++++++++++++++++------------------- > 1 file changed, 21 insertions(+), 19 deletions(-) > > diff --git a/qapi/qmp-input-visitor.c b/qapi/qmp-input-visitor.c > index cfaf378..1820360 100644 > --- a/qapi/qmp-input-visitor.c > +++ b/qapi/qmp-input-visitor.c > @@ -27,9 +27,10 @@ typedef struct StackObject > { > QObject *obj; /* Object being visited */ > > - /* If obj is list: NULL if list is at head, otherwise tail of list > - * still needing visits */ > + /* If obj is list: tail of list still needing visits */ > const QListEntry *entry; > + /* If obj is list: true if head is not visited yet */ > + bool first; Temporary, PATCH 18 will get rid of it. Good, because the less state, the happier I am :) > > GHashTable *h; /* If obj is dict: remaining keys needing visits */ > } StackObject; > @@ -77,7 +78,12 @@ static QObject *qmp_input_get_object(QmpInputVisitor *qiv, /* If we are in the middle of a list, then return the next element > * of the list. */ > if (tos->entry) { Before the patch, the condition matches the comment: tos->entry implies this is a list and we're not at its head. After the patch, the condition only implies it's a list, but ... > assert(qobject_type(qobj) == QTYPE_QLIST); > - return qlist_entry_obj(tos->entry); > + assert(!tos->first); ... you assert that "not at its head" can't happen. Why is that the case? > + qobj = qlist_entry_obj(tos->entry); > + if (consume) { > + tos->entry = qlist_next(tos->entry); > + } > + return qobj; The remainder of the hunk adds the qlist_next() call moved from qmp_input_next_list(), as advertized by the commit message. Good. > } > > /* Otherwise, we are at the root of the visit or the start of a > @@ -91,7 +97,8 @@ static void qdict_add_key(const char *key, QObject *obj, void *opaque) > g_hash_table_insert(h, (gpointer) key, NULL); > } > > -static void qmp_input_push(QmpInputVisitor *qiv, QObject *obj, Error **errp) > +static void qmp_input_push(QmpInputVisitor *qiv, QObject *obj, > + const QListEntry *entry, Error **errp) > { > GHashTable *h; > StackObject *tos = &qiv->stack[qiv->nb_stack]; > @@ -103,7 +110,8 @@ static void qmp_input_push(QmpInputVisitor *qiv, QObject *obj, Error **errp) > } > > tos->obj = obj; > - tos->entry = NULL; > + tos->entry = entry; > + tos->first = true; > tos->h = NULL; > > if (qiv->strict && qobject_type(obj) == QTYPE_QDICT) { Here, you change the initial state. Not explicitly advertized in the commit message, unlike the change to qlist_next(). I guess that's okay, but perhaps my ramblings give you ideas on improving the commit message. I wonder whether we really need the new argument. Could we do something like if (qobject_type(obj) == QTYPE_QLIST) { tos->entry = qlist_first(qobject_to_qlist(obj)); tos->first = true; } ? > @@ -153,7 +161,7 @@ static void qmp_input_start_struct(Visitor *v, const char *name, void **obj, > return; > } > > - qmp_input_push(qiv, qobj, &err); > + qmp_input_push(qiv, qobj, NULL, &err); Since @qobj is a dictionary, the StackObject's entry will be unused, so pass null. Okay. > if (err) { > error_propagate(errp, err); > return; > @@ -175,6 +183,7 @@ static void qmp_input_start_list(Visitor *v, const char *name, Error **errp) > { > QmpInputVisitor *qiv = to_qiv(v); > QObject *qobj = qmp_input_get_object(qiv, name, true); > + const QListEntry *entry; > > if (!qobj || qobject_type(qobj) != QTYPE_QLIST) { > error_setg(errp, QERR_INVALID_PARAMETER_TYPE, name ? name : "null", > @@ -182,7 +191,8 @@ static void qmp_input_start_list(Visitor *v, const char *name, Error **errp) > return; > } > > - qmp_input_push(qiv, qobj, errp); > + entry = qlist_first(qobject_to_qlist(qobj)); > + qmp_input_push(qiv, qobj, entry, errp); @qobj is a list, the StackObject's entry must be initialized to its first member, so pass that. > } > > static GenericList *qmp_input_next_list(Visitor *v, GenericList **list, > @@ -191,23 +201,15 @@ static GenericList *qmp_input_next_list(Visitor *v, GenericList **list, > QmpInputVisitor *qiv = to_qiv(v); > GenericList *entry; > StackObject *so = &qiv->stack[qiv->nb_stack - 1]; > - bool first; > > - if (so->entry == NULL) { > - so->entry = qlist_first(qobject_to_qlist(so->obj)); > - first = true; > - } else { > - so->entry = qlist_next(so->entry); > - first = false; > - } > - > - if (so->entry == NULL) { > + if (!so->entry) { > return NULL; > } > > entry = g_malloc0(size); > - if (first) { > + if (so->first) { > *list = entry; > + so->first = false; > } else { > (*list)->next = entry; > } Here, you drop the qlist_next() call. It moves to qmp_input_get_object(). Good. > @@ -382,7 +384,7 @@ QmpInputVisitor *qmp_input_visitor_new(QObject *obj) > v->visitor.type_any = qmp_input_type_any; > v->visitor.optional = qmp_input_optional; > > - qmp_input_push(v, obj, NULL); > + qmp_input_push(v, obj, NULL, NULL); @obj is the root of the visit. Can it be a list? If yes, then StackObject's entry is not its first element. Why is that correct? > qobject_incref(obj); > > return v;