qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: "Michael S. Tsirkin" <mst@redhat.com>
To: qemu-devel@nongnu.org
Cc: "Peter Maydell" <peter.maydell@linaro.org>,
	"Peter Crosthwaite" <peter.crostwaite@xilinx.com>,
	qemu-stable@nongnu.org, "Anthony Liguori" <aliguori@amazon.com>,
	"Paolo Bonzini" <pbonzini@redhat.com>,
	"Igor Mammedov" <imammedo@redhat.com>,
	=?UTF-8?q?Andreas=20F=C3=A4rber?= <afaerber@suse.de>
Subject: [Qemu-devel] [PULL 095/103] qdev: recursively unrealize devices when unrealizing bus
Date: Tue, 17 Jun 2014 20:42:10 +0300	[thread overview]
Message-ID: <1403021756-15960-96-git-send-email-mst@redhat.com> (raw)
In-Reply-To: <1403021756-15960-1-git-send-email-mst@redhat.com>

From: Paolo Bonzini <pbonzini@redhat.com>

When the patch was posted that became 5c21ce7 (qdev: Realize buses
on device realization, 2014-03-12), it included recursive realization
and unrealization of devices when the bus's "realized" property
was toggled.

However, due to the same old worries about recursive realization
and prerequisites not being realized yet, those hunks were dropped when
committing the patch.  Unfortunately, this causes a use-after-free bug
(easily reproduced by a PCI hot-unplug action).

Before the patch, device_unparent behaved as follows:

   for each child bus
     unparent bus ----------------------------.
     | for each child device                  |
     |   unparent device ---------------.     |
     |   | unrealize device             |     |
     |   | call dc->unparent            |     |
     |   '-------------------------------     |
     '----------------------------------------'
   unrealize device

After the patch, it behaves as follows instead:

   unrealize device --------------------.
   | for each child bus                 |
   |   unrealize bus               (A)  |
   '------------------------------------'
   for each child bus
     unparent bus ----------------------.
     | for each child device            |
     |   unrealize device          (B)  |
     |   call dc->unparent              |
     '----------------------------------'

At the step marked (B) the device might use data from the bus that is
not available anymore due to step (A).

To fix this, we need to unrealize devices before step (A).  To sidestep
concerns about recursive realization, only do recursive unrealization
and leave the "value && !bus->realized" case as it is.

The resulting flow is:

   for each child bus
     unrealize bus ---------------------.
     | for each child device            |
     |   unrealize device          (B)  |
     | call bc->unrealize          (A)  |
     '----------------------------------'
   unrealize device
   for each child bus
     unparent bus ----------------------.
     | for each child device            |
     |   unparent device                |
     '----------------------------------'

where everything is "powered down" before it is unassembled.

Cc: qemu-stable@nongnu.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Tested-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Andreas Färber <afaerber@suse.de>
---
 hw/core/qdev.c | 13 ++++++++++++-
 1 file changed, 12 insertions(+), 1 deletion(-)

diff --git a/hw/core/qdev.c b/hw/core/qdev.c
index 65aa041..b9cd4fc 100644
--- a/hw/core/qdev.c
+++ b/hw/core/qdev.c
@@ -568,14 +568,25 @@ static void bus_set_realized(Object *obj, bool value, Error **errp)
 {
     BusState *bus = BUS(obj);
     BusClass *bc = BUS_GET_CLASS(bus);
+    BusChild *kid;
     Error *local_err = NULL;
 
     if (value && !bus->realized) {
         if (bc->realize) {
             bc->realize(bus, &local_err);
         }
+
+        /* TODO: recursive realization */
     } else if (!value && bus->realized) {
-        if (bc->unrealize) {
+        QTAILQ_FOREACH(kid, &bus->children, sibling) {
+            DeviceState *dev = kid->child;
+            object_property_set_bool(OBJECT(dev), false, "realized",
+                                     &local_err);
+            if (local_err != NULL) {
+                break;
+            }
+        }
+        if (bc->unrealize && local_err == NULL) {
             bc->unrealize(bus, &local_err);
         }
     }
-- 
MST

  parent reply	other threads:[~2014-06-17 17:41 UTC|newest]

Thread overview: 120+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2014-06-17 17:36 [Qemu-devel] [PULL 000/103] pc, pci, virtio, hotplug fixes, enhancements for 2.1 Michael S. Tsirkin
2014-06-17 17:36 ` [Qemu-devel] [PULL 001/103] pc: create custom generic PC machine type Michael S. Tsirkin
2014-06-17 17:36 ` [Qemu-devel] [PULL 002/103] pc: ACPI BIOS: use enum for defining memory affinity flags Michael S. Tsirkin
2014-06-17 17:36 ` [Qemu-devel] [PULL 003/103] object_add: allow completion handler to get canonical path Michael S. Tsirkin
2014-06-17 17:36 ` [Qemu-devel] [PULL 004/103] vl.c: daemonize before guest memory allocation Michael S. Tsirkin
2014-06-17 17:36 ` [Qemu-devel] [PULL 005/103] add memdev backend infrastructure Michael S. Tsirkin
2014-06-17 17:36 ` [Qemu-devel] [PULL 006/103] vl.c: extend -m option to support options for memory hotplug Michael S. Tsirkin
2014-06-17 17:36 ` [Qemu-devel] [PULL 007/103] qdev: hotplug for buss-less devices Michael S. Tsirkin
2014-06-18 14:11   ` Eric Blake
2014-06-18 14:36     ` Michael S. Tsirkin
2014-06-17 17:36 ` [Qemu-devel] [PULL 008/103] qdev: expose DeviceState.hotplugged field as a property Michael S. Tsirkin
2014-06-17 17:36 ` [Qemu-devel] [PULL 009/103] pc: implement pc-dimm device abstraction Michael S. Tsirkin
2014-06-17 17:36 ` [Qemu-devel] [PULL 010/103] memory: add memory_region_is_mapped() API Michael S. Tsirkin
2014-06-17 17:36 ` [Qemu-devel] [PULL 011/103] pc-dimm: do not allow to set already used memdev Michael S. Tsirkin
2014-06-17 18:39   ` Eric Blake
2014-06-17 18:42     ` Michael S. Tsirkin
2014-06-17 18:45     ` Peter Maydell
2014-06-17 18:53       ` Eric Blake
2014-06-17 19:49       ` Michael S. Tsirkin
2014-06-17 17:37 ` [Qemu-devel] [PULL 012/103] pc: initialize memory hotplug address space Michael S. Tsirkin
2014-06-17 17:37 ` [Qemu-devel] [PULL 013/103] pc: exit QEMU if number of slots more than supported 256 Michael S. Tsirkin
2014-06-17 18:42   ` Eric Blake
2014-06-17 19:08     ` Michael S. Tsirkin
2014-06-17 17:37 ` [Qemu-devel] [PULL 014/103] pc: add 'etc/reserved-memory-end' fw_cfg interface for SeaBIOS Michael S. Tsirkin
2014-06-17 17:37 ` [Qemu-devel] [PULL 015/103] pc: exit QEMU if compat machine doesn't support memory hotlpug Michael S. Tsirkin
2014-06-17 17:37 ` [Qemu-devel] [PULL 016/103] pc: add memory hotplug handler to PC_MACHINE Michael S. Tsirkin
2014-06-17 17:37 ` [Qemu-devel] [PULL 017/103] pc-dimm: add busy address check and address auto-allocation Michael S. Tsirkin
2014-06-17 17:37 ` [Qemu-devel] [PULL 018/103] pc-dimm: add busy slot check and slot auto-allocation Michael S. Tsirkin
2014-06-17 17:37 ` [Qemu-devel] [PULL 019/103] acpi: rename cpu_hotplug_defs.h to pc-hotplug.h Michael S. Tsirkin
2014-06-17 17:37 ` [Qemu-devel] [PULL 020/103] acpi: memory hotplug ACPI hardware implementation Michael S. Tsirkin
2014-06-17 17:37 ` [Qemu-devel] [PULL 021/103] trace: add acpi memory hotplug IO region events Michael S. Tsirkin
2014-06-17 17:37 ` [Qemu-devel] [PULL 022/103] trace: pc: add PC_DIMM slot & address allocation Michael S. Tsirkin
2014-06-17 17:37 ` [Qemu-devel] [PULL 023/103] acpi:piix4: allow plug/unlug callbacks handle not only PCI devices Michael S. Tsirkin
2014-06-17 17:37 ` [Qemu-devel] [PULL 024/103] acpi:piix4: add memory hotplug handling Michael S. Tsirkin
2014-06-17 17:37 ` [Qemu-devel] [PULL 025/103] pc: ich9 lpc: make it work with global/compat properties Michael S. Tsirkin
2014-06-17 17:37 ` [Qemu-devel] [PULL 026/103] acpi:ich9: add memory hotplug handling Michael S. Tsirkin
2014-06-17 17:37 ` [Qemu-devel] [PULL 027/103] pc: migrate piix4 & ich9 MemHotplugState Michael S. Tsirkin
2014-06-17 17:37 ` [Qemu-devel] [PULL 028/103] pc: add acpi-device link to PCMachineState Michael S. Tsirkin
2014-06-17 17:37 ` [Qemu-devel] [PULL 029/103] pc: propagate memory hotplug event to ACPI device Michael S. Tsirkin
2014-06-17 17:38 ` [Qemu-devel] [PULL 030/103] pc: ACPI BIOS: implement memory hotplug interface Michael S. Tsirkin
2014-06-17 17:38 ` [Qemu-devel] [PULL 031/103] pc: add "hotplug-memory-region-size" property to PC_MACHINE Michael S. Tsirkin
2014-06-17 17:38 ` [Qemu-devel] [PULL 032/103] pc: ACPI BIOS: reserve SRAT entry for hotplug mem hole Michael S. Tsirkin
2014-06-17 17:38 ` [Qemu-devel] [PULL 033/103] pc: ACPI BIOS: make GPE.3 handle memory hotplug event on PIIX and Q35 machines Michael S. Tsirkin
2014-06-17 17:38 ` [Qemu-devel] [PULL 034/103] acpi: update generated files Michael S. Tsirkin
2014-06-17 17:38 ` [Qemu-devel] [PULL 035/103] acpi-test: update expected tables Michael S. Tsirkin
2014-06-17 17:38 ` [Qemu-devel] [PULL 036/103] virtio: Drop superfluous conditionals around g_free() Michael S. Tsirkin
2014-06-17 17:38 ` [Qemu-devel] [PULL 037/103] virtio: Drop superfluous conditionals around g_strdup() Michael S. Tsirkin
2014-06-17 17:38 ` [Qemu-devel] [PULL 038/103] qtest: fix hex2nib for capital characters Michael S. Tsirkin
2014-06-17 17:38 ` [Qemu-devel] [PULL 039/103] ich: get rid of spaces in type name Michael S. Tsirkin
2014-06-17 17:38 ` [Qemu-devel] [PULL 040/103] pc: q35: acpi: report error to user on unsupported unplug request Michael S. Tsirkin
2014-06-17 17:38 ` [Qemu-devel] [PULL 041/103] migration: export SELF_ANNOUNCE_ROUNDS Michael S. Tsirkin
2014-06-17 17:38 ` [Qemu-devel] [PULL 042/103] migration: introduce self_announce_delay() Michael S. Tsirkin
2014-06-17 17:38 ` [Qemu-devel] [PULL 043/103] virtio-net: announce self by guest Michael S. Tsirkin
2014-06-17 17:38 ` [Qemu-devel] [PULL 044/103] Add kvm_eventfds_enabled function Michael S. Tsirkin
2014-06-17 17:38 ` [Qemu-devel] [PULL 045/103] Add chardev API qemu_chr_fe_read_all Michael S. Tsirkin
2014-06-17 17:38 ` [Qemu-devel] [PULL 046/103] Add chardev API qemu_chr_fe_set_msgfds Michael S. Tsirkin
2014-06-17 17:39 ` [Qemu-devel] [PULL 047/103] Add chardev API qemu_chr_fe_get_msgfds Michael S. Tsirkin
2014-06-17 17:39 ` [Qemu-devel] [PULL 048/103] Add G_IO_HUP handler for socket chardev Michael S. Tsirkin
2014-06-17 17:39 ` [Qemu-devel] [PULL 049/103] vhost: add vhost_get_features and vhost_ack_features Michael S. Tsirkin
2014-06-17 17:39 ` [Qemu-devel] [PULL 050/103] vhost_net should call the poll callback only when it is set Michael S. Tsirkin
2014-06-17 17:39 ` [Qemu-devel] [PULL 051/103] Refactor virtio-net to use generic get_vhost_net Michael S. Tsirkin
2014-06-17 17:39 ` [Qemu-devel] [PULL 052/103] vhost_net_init will use VhostNetOptions to get all its arguments Michael S. Tsirkin
2014-06-17 17:39 ` [Qemu-devel] [PULL 053/103] Add vhost_ops to vhost_dev struct and replace all relevant ioctls Michael S. Tsirkin
2014-06-17 17:39 ` [Qemu-devel] [PULL 054/103] Add vhost-backend and VhostBackendType Michael S. Tsirkin
2014-06-17 17:39 ` [Qemu-devel] [PULL 055/103] Add vhost-user as a vhost backend Michael S. Tsirkin
2014-06-17 17:39 ` [Qemu-devel] [PULL 056/103] vhost-net: vhost-user feature bits support Michael S. Tsirkin
2014-06-17 17:39 ` [Qemu-devel] [PULL 057/103] Add new vhost-user netdev backend Michael S. Tsirkin
2014-06-17 17:39 ` [Qemu-devel] [PULL 058/103] Add the vhost-user netdev backend to the command line Michael S. Tsirkin
2014-06-17 17:39 ` [Qemu-devel] [PULL 059/103] Add vhost-user protocol documentation Michael S. Tsirkin
2014-06-17 17:39 ` [Qemu-devel] [PULL 060/103] libqemustub: add stubs to be able to use qemu-char.c Michael S. Tsirkin
2014-06-17 17:39 ` [Qemu-devel] [PULL 061/103] Add qtest for vhost-user Michael S. Tsirkin
2014-06-17 17:39 ` [Qemu-devel] [PULL 062/103] NUMA: move numa related code to new file numa.c Michael S. Tsirkin
2014-06-17 17:40 ` [Qemu-devel] [PULL 063/103] NUMA: check if the total numa memory size is equal to ram_size Michael S. Tsirkin
2014-06-17 17:40 ` [Qemu-devel] [PULL 064/103] NUMA: Add numa_info structure to contain numa nodes info Michael S. Tsirkin
2014-06-17 17:40 ` [Qemu-devel] [PULL 065/103] NUMA: convert -numa option to use OptsVisitor Michael S. Tsirkin
2014-06-17 17:40 ` [Qemu-devel] [PULL 066/103] NUMA: expand MAX_NODES from 64 to 128 Michael S. Tsirkin
2014-06-17 17:40 ` [Qemu-devel] [PULL 067/103] man: improve -numa doc Michael S. Tsirkin
2014-06-17 17:40 ` [Qemu-devel] [PULL 068/103] qmp: improve error reporting for -object and object-add Michael S. Tsirkin
2014-06-17 17:40 ` [Qemu-devel] [PULL 069/103] numa: introduce memory_region_allocate_system_memory Michael S. Tsirkin
2014-06-17 17:40 ` [Qemu-devel] [PULL 070/103] memory: reorganize file-based allocation Michael S. Tsirkin
2014-06-17 17:40 ` [Qemu-devel] [PULL 071/103] memory: move preallocation code out of exec.c Michael S. Tsirkin
2014-06-17 17:40 ` [Qemu-devel] [PULL 072/103] memory: move RAM_PREALLOC_MASK to exec.c, rename Michael S. Tsirkin
2014-06-17 17:40 ` [Qemu-devel] [PULL 073/103] configure: add Linux libnuma detection Michael S. Tsirkin
2014-06-17 17:40 ` [Qemu-devel] [PULL 074/103] Introduce signed range Michael S. Tsirkin
2014-06-17 17:40 ` [Qemu-devel] [PULL 075/103] qom: introduce object_property_get_enum and object_property_get_uint16List Michael S. Tsirkin
2014-06-17 17:40 ` [Qemu-devel] [PULL 076/103] numa: add -numa node,memdev= option Michael S. Tsirkin
2014-06-17 17:40 ` [Qemu-devel] [PULL 077/103] memory: move mem_path handling to memory_region_allocate_system_memory Michael S. Tsirkin
2014-06-17 17:41 ` [Qemu-devel] [PULL 078/103] memory: add error propagation to file-based RAM allocation Michael S. Tsirkin
2014-06-17 20:49   ` Eric Blake
2014-06-17 17:41 ` [Qemu-devel] [PULL 079/103] vl: redo -object parsing Michael S. Tsirkin
2014-06-17 17:41 ` [Qemu-devel] [PULL 080/103] pc: pass MachineState to pc_memory_init Michael S. Tsirkin
2014-06-17 17:41 ` [Qemu-devel] [PULL 081/103] backend:hostmem: replace hostmemory with host_memory Michael S. Tsirkin
2014-06-17 17:41 ` [Qemu-devel] [PULL 082/103] hostmem: separate allocation from UserCreatable complete method Michael S. Tsirkin
2014-06-17 17:41 ` [Qemu-devel] [PULL 083/103] hostmem: add file-based HostMemoryBackend Michael S. Tsirkin
2014-06-17 17:41 ` [Qemu-devel] [PULL 084/103] hostmem: add merge and dump properties Michael S. Tsirkin
2014-06-17 17:41 ` [Qemu-devel] [PULL 085/103] hostmem: allow preallocation of any memory region Michael S. Tsirkin
2014-06-17 17:41 ` [Qemu-devel] [PULL 086/103] hostmem: add property to map memory with MAP_SHARED Michael S. Tsirkin
2014-06-17 17:41 ` [Qemu-devel] [PULL 087/103] hostmem: add properties for NUMA memory policy Michael S. Tsirkin
2014-06-17 17:41 ` [Qemu-devel] [PULL 088/103] qmp: add query-memdev Michael S. Tsirkin
2014-06-17 17:41 ` [Qemu-devel] [PULL 089/103] hmp: add info memdev Michael S. Tsirkin
2014-06-17 17:41 ` [Qemu-devel] [PULL 090/103] tests: fix memory leak in test of string input visitor Michael S. Tsirkin
2014-06-17 17:41 ` [Qemu-devel] [PULL 091/103] qapi: make string input visitor parse int list Michael S. Tsirkin
2014-06-17 21:36   ` Eric Blake
2014-06-18  3:13     ` Michael S. Tsirkin
2014-06-17 17:41 ` [Qemu-devel] [PULL 092/103] qapi: make string output " Michael S. Tsirkin
2014-06-17 21:45   ` Eric Blake
2014-06-18 15:02     ` Michael S. Tsirkin
2014-06-19  0:43       ` Hu Tao
2014-06-17 17:42 ` [Qemu-devel] [PULL 093/103] qapi: fix build on glib < 2.28 Michael S. Tsirkin
2014-06-17 17:42 ` [Qemu-devel] [PULL 094/103] qdev: reorganize error reporting in bus_set_realized Michael S. Tsirkin
2014-06-17 17:42 ` Michael S. Tsirkin [this message]
2014-06-17 17:42 ` [Qemu-devel] [PULL 096/103] qmp: clean out whitespace Michael S. Tsirkin
2014-06-17 17:42 ` [Qemu-devel] [PULL 097/103] pc: acpi: do not hardcode preprocessor Michael S. Tsirkin
2014-06-17 17:42 ` [Qemu-devel] [PULL 098/103] numa: handle mmaped memory allocation failure correctly Michael S. Tsirkin
2014-06-17 17:42 ` [Qemu-devel] [PULL 099/103] qmp: add query-memory-devices command Michael S. Tsirkin
2014-06-17 17:42 ` [Qemu-devel] [PULL 100/103] acpi: introduce TYPE_ACPI_DEVICE_IF interface Michael S. Tsirkin
2014-06-17 17:42 ` [Qemu-devel] [PULL 101/103] acpi: implement ospm_status() method for PIIX4/ICH9_LPC devices Michael S. Tsirkin
2014-06-17 17:42 ` [Qemu-devel] [PULL 102/103] qmp: add query-acpi-ospm-status command Michael S. Tsirkin
2014-06-17 17:42 ` [Qemu-devel] [PULL 103/103] qmp: add ACPI_DEVICE_OST event handling Michael S. Tsirkin
2014-06-18 14:07 ` [Qemu-devel] [PULL 000/103] pc, pci, virtio, hotplug fixes, enhancements for 2.1 Peter Maydell

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=1403021756-15960-96-git-send-email-mst@redhat.com \
    --to=mst@redhat.com \
    --cc=afaerber@suse.de \
    --cc=aliguori@amazon.com \
    --cc=imammedo@redhat.com \
    --cc=pbonzini@redhat.com \
    --cc=peter.crostwaite@xilinx.com \
    --cc=peter.maydell@linaro.org \
    --cc=qemu-devel@nongnu.org \
    --cc=qemu-stable@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).