All of lore.kernel.org
 help / color / mirror / Atom feed
From: Ian Jackson <ian.jackson@eu.citrix.com>
To: xen-devel@lists.xensource.com
Cc: Ian Jackson <ian.jackson@eu.citrix.com>,
	Euan Harris <euan.harris@citrix.com>
Subject: [PATCH 16/29] libxl: ao: Provide manip_refcnt
Date: Tue, 10 Feb 2015 20:10:03 +0000	[thread overview]
Message-ID: <1423599016-32639-17-git-send-email-ian.jackson@eu.citrix.com> (raw)
In-Reply-To: <1423599016-32639-1-git-send-email-ian.jackson@eu.citrix.com>

Previously we used in_initiator to stop the ao being freed while we
were still in the initiator function (which would result in the
initiator's call to lixl__ao_inprogress accessing the ao after it had
been freed).

We are going to introduce a new libxl entrypoint which finds, and
operates on, ongoing aos.  This function needs the same protection,
and might even end up running on the same ao multiple times
concurrently.

So do this with reference counting instead, with a new variable
ao->manip_refcnt.

We keep ao->in_initiator because that allows us to keep some useful
asserts about the sequencing of libxl__ao_inprogress, etc.

Signed-off-by: Ian Jackson <ian.jackson@eu.citrix.com>
---
 tools/libxl/libxl_event.c    |   43 +++++++++++++++++++++++++++++++++---------
 tools/libxl/libxl_internal.h |    1 +
 2 files changed, 35 insertions(+), 9 deletions(-)

diff --git a/tools/libxl/libxl_event.c b/tools/libxl/libxl_event.c
index 37c3d2f..80677e0 100644
--- a/tools/libxl/libxl_event.c
+++ b/tools/libxl/libxl_event.c
@@ -33,6 +33,8 @@
 
 static libxl__ao *ao_nested_root(libxl__ao *ao);
 
+static void ao__check_destroy(libxl_ctx *ctx, libxl__ao *ao);
+
 
 /*
  * The counter osevent_in_hook is used to ensure that the application
@@ -1344,8 +1346,7 @@ static void egc_run_callbacks(libxl__egc *egc)
         ao->how.callback(CTX, ao->rc, ao->how.u.for_callback);
         CTX_LOCK;
         ao->notified = 1;
-        if (!ao->in_initiator)
-            libxl__ao__destroy(CTX, ao);
+        ao__check_destroy(CTX, ao);
         CTX_UNLOCK;
     }
 }
@@ -1726,6 +1727,33 @@ int libxl_event_wait(libxl_ctx *ctx, libxl_event **event_r,
  *                              - destroy the ao
  */
 
+
+/*
+ * A "manip" is a libxl public function manipulating this ao, which
+ * has a pointer to it.  We have to not destroy it while that's the
+ * case, obviously.
+ */
+static void ao__manip_enter(libxl__ao *ao)
+{
+    assert(ao->manip_refcnt < INT_MAX);
+    ao->manip_refcnt++;
+}
+
+static void ao__manip_leave(libxl_ctx *ctx, libxl__ao *ao)
+{
+    assert(ao->manip_refcnt > 0);
+    ao->manip_refcnt--;
+    ao__check_destroy(ctx, ao);
+}
+
+static void ao__check_destroy(libxl_ctx *ctx, libxl__ao *ao)
+{
+    if (!ao->manip_refcnt && ao->notified) {
+        assert(ao->complete);
+        libxl__ao__destroy(ctx,ao);
+    }
+}
+
 void libxl__ao__destroy(libxl_ctx *ctx, libxl__ao *ao)
 {
     AO_GC;
@@ -1807,8 +1835,8 @@ void libxl__ao_complete_check_progress_reports(libxl__egc *egc, libxl__ao *ao)
         }
         ao->notified = 1;
     }
-    if (!ao->in_initiator && ao->notified)
-        libxl__ao__destroy(ctx, ao);
+    
+    ao__check_destroy(ctx, ao);
 }
 
 libxl__ao *libxl__ao_create(libxl_ctx *ctx, uint32_t domid,
@@ -1823,6 +1851,7 @@ libxl__ao *libxl__ao_create(libxl_ctx *ctx, uint32_t domid,
     ao->magic = LIBXL__AO_MAGIC;
     ao->constructing = 1;
     ao->in_initiator = 1;
+    ao__manip_enter(ao);
     ao->poller = 0;
     ao->domid = domid;
     LIBXL_INIT_GC(ao->gc, ctx);
@@ -1903,11 +1932,7 @@ int libxl__ao_inprogress(libxl__ao *ao,
     }
 
     ao->in_initiator = 0;
-
-    if (ao->notified) {
-        assert(ao->complete);
-        libxl__ao__destroy(CTX,ao);
-    }
+    ao__manip_leave(CTX, ao);
 
     return rc;
 }
diff --git a/tools/libxl/libxl_internal.h b/tools/libxl/libxl_internal.h
index e29db43..d2c2637 100644
--- a/tools/libxl/libxl_internal.h
+++ b/tools/libxl/libxl_internal.h
@@ -449,6 +449,7 @@ struct libxl__ao {
      */
     uint32_t magic;
     unsigned constructing:1, in_initiator:1, complete:1, notified:1;
+    int manip_refcnt;
     libxl__ao *nested_root;
     int nested_progeny;
     int progress_reports_outstanding;
-- 
1.7.10.4

  parent reply	other threads:[~2015-02-10 20:10 UTC|newest]

Thread overview: 98+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2015-02-10 20:09 [RFC PATCH v2 00/29] libxl: Cancelling asynchronous operations Ian Jackson
2015-02-10 20:09 ` [PATCH 01/29] libxl: Further fix exit paths from libxl_device_events_handler Ian Jackson
2015-02-18 16:23   ` Roger Pau Monné
2015-03-24 10:49     ` Ian Campbell
2015-02-10 20:09 ` [PATCH 02/29] libxl: Comment cleanups Ian Jackson
2015-02-11  1:46   ` Hongyang Yang
2015-02-11 12:05     ` Ian Jackson
2015-03-30  9:08     ` Ian Campbell
2015-03-24 10:50   ` Ian Campbell
2015-02-10 20:09 ` [PATCH 03/29] libxl: suspend: switch_logdirty_done takes rc Ian Jackson
2015-03-24 10:53   ` Ian Campbell
2015-03-31 18:02     ` Ian Jackson
2015-02-10 20:09 ` [PATCH 04/29] libxl: suspend: common suspend callbacks take rc Ian Jackson
2015-03-24 10:55   ` Ian Campbell
2015-03-31 18:03     ` Ian Jackson
2015-02-10 20:09 ` [PATCH 05/29] libxl: suspend: Return correct error from callbacks Ian Jackson
2015-03-24 10:58   ` Ian Campbell
2015-02-10 20:09 ` [PATCH 06/29] libxl: Use libxl__xswait* in libxl__ao_device Ian Jackson
2015-03-24 11:04   ` Ian Campbell
2015-03-31 18:06     ` Ian Jackson
2015-04-01  9:25       ` Ian Campbell
2015-02-10 20:09 ` [PATCH 07/29] libxl: xswait/devstate: Move xswait to before devstate Ian Jackson
2015-03-24 11:05   ` Ian Campbell
2015-02-10 20:09 ` [PATCH 08/29] libxl: devstate: Use libxl__xswait* Ian Jackson
2015-03-24 11:07   ` Ian Campbell
2015-02-10 20:09 ` [PATCH 09/29] libxl: New error codes CANCELLED etc Ian Jackson
2015-03-24 11:08   ` Ian Campbell
2015-02-10 20:09 ` [PATCH 10/29] libxl: events: Make timeout and async exec setup take an ao, not a gc Ian Jackson
2015-02-11  1:04   ` Wen Congyang
2015-02-11 12:04     ` Ian Jackson
2015-03-31 18:09     ` Ian Jackson
2015-03-24 11:09   ` Ian Campbell
2015-02-10 20:09 ` [PATCH 11/29] libxl: events: Make libxl__async_exec_* pass caller an rc Ian Jackson
2015-03-24 11:20   ` Ian Campbell
2015-03-31 18:12     ` Ian Jackson
2015-04-01  9:29       ` Ian Campbell
2015-02-10 20:09 ` [PATCH 12/29] libxl: events: Permit timeouts to signal cancellation Ian Jackson
2015-03-24 11:22   ` Ian Campbell
2015-02-10 20:10 ` [PATCH 13/29] libxl: domain create: Do not destroy on cancellation Ian Jackson
2015-03-24 11:24   ` Ian Campbell
2015-03-31 18:14     ` Ian Jackson
2015-02-10 20:10 ` [PATCH 14/29] libxl: ao: Record ultimate parent of a nested ao Ian Jackson
2015-03-24 11:26   ` Ian Campbell
2015-02-10 20:10 ` [PATCH 15/29] libxl: ao: Count the nested progeny of an ao Ian Jackson
2015-03-24 11:27   ` Ian Campbell
2015-02-10 20:10 ` Ian Jackson [this message]
2015-03-24 11:32   ` [PATCH 16/29] libxl: ao: Provide manip_refcnt Ian Campbell
2015-03-31 18:23     ` Ian Jackson
2015-04-01  9:34       ` Ian Campbell
2015-02-10 20:10 ` [PATCH 17/29] libxl: cancellation: Provide public ao cancellation API Ian Jackson
2015-03-24 11:45   ` Ian Campbell
2015-03-31 18:26     ` Ian Jackson
2015-03-24 11:48   ` Ian Campbell
2015-03-31 18:33     ` Ian Jackson
2015-04-01  9:38       ` Ian Campbell
2015-02-10 20:10 ` [PATCH 18/29] libxl: cancellation: Provide explicit internal cancel check API Ian Jackson
2015-03-24 11:45   ` Ian Campbell
2015-02-10 20:10 ` [PATCH 19/29] libxl: cancellation: Make timeouts cancellable Ian Jackson
2015-03-24 11:50   ` Ian Campbell
2015-02-10 20:10 ` [PATCH 20/29] libxl: cancellation: Note that driver domain task cannot be usefully cancelled Ian Jackson
2015-02-18 16:24   ` Roger Pau Monné
2015-03-24 11:51   ` Ian Campbell
2015-03-31 18:37     ` Ian Jackson
2015-02-10 20:10 ` [PATCH 21/29] libxl: cancellation: Make spawns cancellable Ian Jackson
2015-03-24 11:53   ` Ian Campbell
2015-03-31 18:45     ` Ian Jackson
2015-04-01  9:40       ` Ian Campbell
2015-02-10 20:10 ` [PATCH 22/29] libxl: Introduce DOMAIN_DESTROYED error code Ian Jackson
2015-03-24 11:56   ` Ian Campbell
2015-03-31 18:47     ` Ian Jackson
2015-04-01  9:41       ` Ian Campbell
2015-02-10 20:10 ` [PATCH 23/29] libxl: cancellation: Support cancellation where we spot domain death Ian Jackson
2015-03-24 11:58   ` Ian Campbell
2015-02-10 20:10 ` [PATCH 24/29] libxl: Introduce FILLZERO Ian Jackson
2015-03-24 12:03   ` Ian Campbell
2015-03-31 18:51     ` Ian Jackson
2015-04-01  9:42       ` Ian Campbell
2015-02-10 20:10 ` [PATCH 25/29] libxl: cancellation: Preparations for save/restore cancellation Ian Jackson
2015-03-24 12:04   ` Ian Campbell
2015-02-10 20:10 ` [PATCH 26/29] libxl: cancellation: Handle SIGTERM in save/restore helper Ian Jackson
2015-03-24 12:07   ` Ian Campbell
2015-02-10 20:10 ` [PATCH 27/29] libxl: cancellation: Cancel libxc save/restore Ian Jackson
2015-03-24 12:08   ` Ian Campbell
2015-02-10 20:10 ` [PATCH 28/29] libxl: ao: datacopier callback gets an rc Ian Jackson
2015-03-24 12:10   ` Ian Campbell
2015-02-10 20:10 ` [PATCH 29/29] libxl: cancellation: Make datacopiers cancellable Ian Jackson
2015-03-24 12:11   ` Ian Campbell
2015-02-10 20:13 ` [RFC PATCH v2 00/29] libxl: Cancelling asynchronous operations Ian Jackson
2015-02-18 16:10 ` Euan Harris
2015-02-18 16:12   ` Ian Jackson
2015-04-07 17:08   ` Euan Harris
2015-04-07 17:19     ` Ian Jackson
2015-04-09 16:44       ` Euan Harris
2015-04-09 16:50         ` Ian Jackson
2015-04-14  9:43       ` Euan Harris
2015-03-03 12:08 ` Ian Campbell
2015-03-20 10:39   ` Euan Harris
2015-04-07 17:22   ` Ian Jackson

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=1423599016-32639-17-git-send-email-ian.jackson@eu.citrix.com \
    --to=ian.jackson@eu.citrix.com \
    --cc=euan.harris@citrix.com \
    --cc=xen-devel@lists.xensource.com \
    /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.