linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: "Rafael J. Wysocki" <rjw@rjwysocki.net>
To: ACPI Devel Maling List <linux-acpi@vger.kernel.org>
Cc: Bjorn Helgaas <bhelgaas@google.com>,
	Aaron Lu <aaron.lu@intel.com>,
	Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
	Linux PCI <linux-pci@vger.kernel.org>,
	Mika Westerberg <mika.westerberg@linux.intel.com>,
	Robert Moore <robert.moore@intel.com>
Subject: Re: [PATCH v2 1/6] ACPI / hotplug: Fix theoretical race in acpi_hotplug_notify_cb()
Date: Sun, 02 Feb 2014 18:01:31 +0100	[thread overview]
Message-ID: <10544516.0W0fJK2WE2@vostro.rjw.lan> (raw)
In-Reply-To: <2304222.JefaLY6VAk@vostro.rjw.lan>

On Sunday, February 02, 2014 01:54:02 AM Rafael J. Wysocki wrote:
> From: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
> 
> There is a slight possibility for the ACPI device object pointed to
> by adev in acpi_hotplug_notify_cb() to become invalid between the
> acpi_bus_get_device() that it comes from and the subsequent get_device().
> Namely, if acpi_scan_drop_device() runs concurrently with respect to
> acpi_hotplug_notify_cb() and acpi_device_del_list is not empty,
> acpi_device_del_work_fn() may delete the device object in question
> without waiting for the ACPI events workqueue to drain, which very
> well may happen right after a successful execution of
> acpi_bus_get_device() in acpi_hotplug_notify_cb().
> 
> To prevent that from happening, run acpi_bus_get_device() and the
> subsequent get_device() in acpi_hotplug_notify_cb() under
> acpi_device_del_lock, so that the deletion of the given device
> object cannot be queued up by acpi_scan_drop_device() between the
> two.
> 
> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
> ---
>  drivers/acpi/scan.c |   14 ++++++++------
>  1 file changed, 8 insertions(+), 6 deletions(-)
> 
> Index: linux-pm/drivers/acpi/scan.c
> ===================================================================
> --- linux-pm.orig/drivers/acpi/scan.c
> +++ linux-pm/drivers/acpi/scan.c
> @@ -41,6 +41,8 @@ static DEFINE_MUTEX(acpi_scan_lock);
>  static LIST_HEAD(acpi_scan_handlers_list);
>  DEFINE_MUTEX(acpi_device_lock);
>  LIST_HEAD(acpi_wakeup_device_list);
> +static LIST_HEAD(acpi_device_del_list);
> +static DEFINE_MUTEX(acpi_device_del_lock);
>  
>  struct acpi_device_bus_id{
>  	char bus_id[15];
> @@ -488,9 +490,6 @@ static void acpi_hotplug_notify_cb(acpi_
>  	struct acpi_device *adev;
>  	acpi_status status;
>  
> -	if (acpi_bus_get_device(handle, &adev))
> -		goto err_out;
> -
>  	switch (type) {
>  	case ACPI_NOTIFY_BUS_CHECK:
>  		acpi_handle_debug(handle, "ACPI_NOTIFY_BUS_CHECK event\n");
> @@ -512,7 +511,13 @@ static void acpi_hotplug_notify_cb(acpi_
>  		/* non-hotplug event; possibly handled by other handler */
>  		return;
>  	}
> +	mutex_lock(&acpi_device_del_lock);
> +	if (acpi_bus_get_device(handle, &adev)) {
> +		mutex_unlock(&acpi_device_del_lock);
> +		goto err_out;
> +	}
>  	get_device(&adev->dev);
> +	mutex_unlock(&acpi_device_del_lock);
>  	status = acpi_hotplug_execute(acpi_device_hotplug, adev, type);
>  	if (ACPI_SUCCESS(status))
>  		return;

Well, that would have been good if it hand't been broken. :-(

acpi_scan_drop_device() which acquires acpi_device_del_lock is called under
the ACPICA's namespace mutex and acpi_bus_get_device() above acquires that
mutex, so this leads to a classical ABBA deadlock scenario.  Bummer.

And I haven't been able to convince myself that what we're doing in
acpi_hotplug_notify_cb() is actually safe without any locking.  Not to
mention acpi_bus_notify() for that matter.  Moreover, the *only* safe
way to do that I'm seeing at the moment is to call the get_device()
under the ACPICA's namespace mutex, before it is released in
acpi_get_data().

Of course, ACPICA will need to be modified slightly for that to be
possible (sorry, Bob), but at least that *should* work, so I have a
new version of this patchset doing just that.  I'll send it out
shortly.

Thanks!

-- 
I speak only for myself.
Rafael J. Wysocki, Intel Open Source Technology Center.

  reply	other threads:[~2014-02-02 16:47 UTC|newest]

Thread overview: 100+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2014-01-27  0:37 [PATCH 0/11] ACPI / hotplug / PCI: Updates on top of changes merged recently Rafael J. Wysocki
2014-01-27  0:38 ` [PATCH 1/11] ACPI / hotplug / PCI: Proper kerneldoc comments for enumeration/removal Rafael J. Wysocki
2014-01-27  0:38 ` [PATCH 2/11] ACPI / hotplug / PCI: Simplify register_slot() Rafael J. Wysocki
2014-01-27  0:39 ` [PATCH 3/11] ACPI / hotplug / PCI: Drop acpiphp_bus_trim() Rafael J. Wysocki
2014-01-27  0:40 ` [PATCH 4/11] ACPI / hotplug / PCI: Rework acpiphp_no_hotplug() Rafael J. Wysocki
2014-01-27  0:41 ` [PATCH 5/11] ACPI / hotplug / PCI: Store acpi_device pointer in acpiphp_context Rafael J. Wysocki
2014-01-27  0:41 ` [PATCH 6/11] ACPI / hotplug / PCI: Drop acpiphp_bus_add() Rafael J. Wysocki
2014-01-27  0:42 ` [PATCH 7/11] ACPI / hotplug / PCI: Drop crit_sect locking Rafael J. Wysocki
2014-01-27  0:43 ` [PATCH 8/11] ACPI / hotplug / PCI: Simplify hotplug_event() Rafael J. Wysocki
2014-01-27  0:44 ` [PATCH 9/11] ACPI / hotplug / PCI: Simplify disable_slot() Rafael J. Wysocki
2014-01-27  0:45 ` [PATCH 10/11] ACPI / hotplug / PCI: Use acpi_handle_debug() in hotplug_event() Rafael J. Wysocki
2014-01-27  0:46 ` [PATCH 11/11] ACPI / hotplug: Do not pass ACPI handles to ACPI dock operations Rafael J. Wysocki
2014-01-28 22:10 ` [PATCH 0/5][RFT] ACPI / hotplug / PCI: Consolidation of ACPIPHP with ACPI core device hotplug Rafael J. Wysocki
2014-01-28 22:12   ` [PATCH 1/5][RFT] ACPI / hotplug / PCI: Attach hotplug contexts to struct acpi_device Rafael J. Wysocki
2014-01-28 22:13   ` [PATCH 2/5][RFT] ACPI / hotplug: Introduce acpi_install_hotplug_notify_handler() Rafael J. Wysocki
2014-01-28 22:14   ` [PATCH 3/5][RFT] ACPI / hotplug / PCI: Consolidate ACPIPHP with ACPI core hotplug Rafael J. Wysocki
2014-01-28 22:14   ` [PATCH 4/5][RFT] ACPI / hotplug / PCI: Simplify acpi_install_hotplug_notify_handler() Rafael J. Wysocki
2014-01-28 22:16   ` [PATCH 5/5][RFT] ACPI / hotplug: Dispach hotplug notifications from acpi_bus_notify() Rafael J. Wysocki
2014-01-28 23:57   ` [PATCH 0/5][RFT] ACPI / hotplug / PCI: Consolidation of ACPIPHP with ACPI core device hotplug Rafael J. Wysocki
2014-01-28 23:58     ` [Resend][PATCH 1/5][RFT] ACPI / hotplug / PCI: Attach hotplug contexts to struct acpi_device Rafael J. Wysocki
2014-01-28 23:59     ` [Resend][PATCH 2/5][RFT] ACPI / hotplug: Introduce acpi_install_hotplug_notify_handler() Rafael J. Wysocki
2014-01-28 23:59     ` [Update][PATCH 3/5][RFT] ACPI / hotplug / PCI: Consolidate ACPIPHP with ACPI core hotplug Rafael J. Wysocki
2014-01-29  0:00     ` [Update][PATCH 4/5][RFT] ACPI / hotplug / PCI: Simplify acpi_install_hotplug_notify_handler() Rafael J. Wysocki
2014-01-31 15:28       ` Mika Westerberg
2014-01-31 15:42         ` Rafael J. Wysocki
2014-01-31 15:40           ` Mika Westerberg
2014-01-31 16:01             ` Mika Westerberg
2014-01-31 16:16               ` Rafael J. Wysocki
2014-01-31 17:09                 ` Mika Westerberg
2014-01-31 17:34                   ` Rafael J. Wysocki
2014-02-02  0:12                     ` Rafael J. Wysocki
2014-01-29  0:02     ` [Update][PATCH 5/5][RFT] ACPI / hotplug / PCI: Hotplug notifications from acpi_bus_notify() Rafael J. Wysocki
2014-01-30 15:28       ` [Update 2x][PATCH " Rafael J. Wysocki
2014-02-02  0:52     ` [PATCH v2 0/6] ACPI / hotplug / PCI: Consolidation of ACPIPHP with ACPI core device hotplug Rafael J. Wysocki
2014-02-02  0:54       ` [PATCH v2 1/6] ACPI / hotplug: Fix theoretical race in acpi_hotplug_notify_cb() Rafael J. Wysocki
2014-02-02 17:01         ` Rafael J. Wysocki [this message]
2014-02-02  0:55       ` [PATCH v2 2/6] ACPI / hotplug / PCI: Define hotplug context lock in the core Rafael J. Wysocki
2014-02-02  0:56       ` [PATCH v2 3/6] ACPI / hotplug / PCI: Consolidate ACPIPHP with ACPI core hotplug Rafael J. Wysocki
2014-02-02  0:57       ` [PATCH v2 4/6] ACPI / hotplug / PCI: Rework the handling of eject requests Rafael J. Wysocki
2014-02-02  0:58       ` [PATCH v2 5/6] ACPI / hotplug / PCI: Simplify acpi_install_hotplug_notify_handler() Rafael J. Wysocki
2014-02-02  0:58       ` [PATCH v2 6/6] ACPI / hotplug / PCI: Hotplug notifications from acpi_bus_notify() Rafael J. Wysocki
2014-02-02 17:11       ` [PATCH v3 0/7] ACPI / hotplug / PCI: Consolidation of ACPIPHP with ACPI core device hotplug Rafael J. Wysocki
2014-02-02 17:12         ` [PATCH v3 1/7] ACPICA: Introduce acpi_get_data_full() and rework acpi_get_data() Rafael J. Wysocki
2014-02-02 17:15         ` [PATCH v3 2/7] ACPI / hotplug: Fix potential races in notify handlers Rafael J. Wysocki
2014-02-02 17:16         ` [PATCH v3 3/7] ACPI / hotplug / PCI: Define hotplug context lock in the core Rafael J. Wysocki
2014-02-02 17:17         ` [PATCH v3 4/7] ACPI / hotplug / PCI: Consolidate ACPIPHP with ACPI core hotplug Rafael J. Wysocki
2014-02-02 17:18         ` [PATCH v3 5/7] ACPI / hotplug / PCI: Rework the handling of eject requests Rafael J. Wysocki
2014-02-02 17:19         ` [PATCH v3 6/7] ACPI / hotplug / PCI: Simplify acpi_install_hotplug_notify_handler() Rafael J. Wysocki
2014-02-02 17:20         ` [PATCH v3 7/7] ACPI / hotplug / PCI: Hotplug notifications from acpi_bus_notify() Rafael J. Wysocki
2014-02-03 10:45         ` [PATCH v3 0/7] ACPI / hotplug / PCI: Consolidation of ACPIPHP with ACPI core device hotplug Mika Westerberg
2014-02-03 21:51           ` Rafael J. Wysocki
2014-02-02  0:19 ` [PATCH v2 0/13] ACPI / hotplug / PCI: Updates on top of changes merged recently Rafael J. Wysocki
2014-02-02  0:21   ` [PATCH v2 1/13] ACPI / hotplug / PCI: Remove entries from bus->devices in reverse order Rafael J. Wysocki
2014-02-02  0:22   ` [PATCH v2 2/13] ACPI / hotplug / PCI: Move PCI rescan-remove locking to hotplug_event() Rafael J. Wysocki
2014-02-02  0:23   ` [PATCH v2 3/13] ACPI / hotplug / PCI: Simplify disable_slot() Rafael J. Wysocki
2014-02-02  0:24   ` [PATCH v2 4/13] ACPI / hotplug / PCI: Proper kerneldoc comments for enumeration/removal Rafael J. Wysocki
2014-02-02  0:25   ` [PATCH v2 5/13] ACPI / hotplug / PCI: Simplify register_slot() Rafael J. Wysocki
2014-02-02  0:26   ` [PATCH v2 6/13] ACPI / hotplug / PCI: Drop acpiphp_bus_trim() Rafael J. Wysocki
2014-02-02  0:27   ` [PATCH v2 7/13] ACPI / hotplug / PCI: Rework acpiphp_no_hotplug() Rafael J. Wysocki
2014-02-02  0:27   ` [PATCH v2 8/13] ACPI / hotplug / PCI: Store acpi_device pointer in acpiphp_context Rafael J. Wysocki
2014-02-02  0:28   ` [PATCH v2 9/13] ACPI / hotplug / PCI: Drop acpiphp_bus_add() Rafael J. Wysocki
2014-02-02  0:29   ` [PATCH v2 10/13] ACPI / hotplug / PCI: Drop crit_sect locking Rafael J. Wysocki
2014-02-02  0:30   ` [PATCH v2 11/13] ACPI / hotplug / PCI: Simplify hotplug_event() Rafael J. Wysocki
2014-02-02  0:31   ` [PATCH v2 12/13] ACPI / hotplug / PCI: Use acpi_handle_debug() in hotplug_event() Rafael J. Wysocki
2014-02-02  0:31   ` [PATCH v2 13/13] ACPI / hotplug: Do not pass ACPI handles to ACPI dock operations Rafael J. Wysocki
2014-02-03  1:47   ` [PATCH] ACPI / hotplug / PCI: Scan root bus under the PCI rescan-remove lock Rafael J. Wysocki
2014-02-03 10:44   ` [PATCH v2 0/13] ACPI / hotplug / PCI: Updates on top of changes merged recently Mika Westerberg
2014-02-03 23:12     ` [PATCH v2 0/24] ACPI / hotplug / PCI: ACPIPHP updates and consolidation with ACPI core Rafael J. Wysocki
2014-02-03 23:14       ` [PATCH 1/24][Resend] ACPI / hotplug / PCI: Remove entries from bus->devices in reverse order Rafael J. Wysocki
2014-02-03 23:15       ` [PATCH 2/24][Resend] ACPI / hotplug / PCI: Move PCI rescan-remove locking to hotplug_event() Rafael J. Wysocki
2014-02-03 23:16       ` [PATCH 3/24][Resend] ACPI / hotplug / PCI: Scan root bus under the PCI rescan-remove lock Rafael J. Wysocki
2014-02-03 23:18       ` [PATCH 4/24][New] ACPI / hotplug / PCI: Fix bridge removal race in handle_hotplug_event() Rafael J. Wysocki
2014-02-03 23:18       ` [PATCH 5/24][New] ACPI / hotplug / PCI: Fix bridge removal race vs dock events Rafael J. Wysocki
2014-02-03 23:20       ` [PATCH 6/24][Resend] ACPI / hotplug / PCI: Simplify disable_slot() Rafael J. Wysocki
2014-02-03 23:21       ` [PATCH 7/24][Resend] ACPI / hotplug / PCI: Proper kerneldoc comments for enumeration/removal Rafael J. Wysocki
2014-02-03 23:22       ` [PATCH 8/24][Resend] ACPI / hotplug / PCI: Simplify register_slot() Rafael J. Wysocki
2014-02-03 23:23       ` [PATCH 9/24][Resend] ACPI / hotplug / PCI: Drop acpiphp_bus_trim() Rafael J. Wysocki
2014-02-03 23:24       ` [PATCH 10/24][Resend] ACPI / hotplug / PCI: Rework acpiphp_no_hotplug() Rafael J. Wysocki
2014-02-03 23:25       ` [PATCH 11/24][Update] ACPI / hotplug / PCI: Store acpi_device pointer in acpiphp_context Rafael J. Wysocki
2014-02-03 23:26       ` [PATCH 12/24][Resend] ACPI / hotplug / PCI: Drop acpiphp_bus_add() Rafael J. Wysocki
2014-02-03 23:27       ` [PATCH 13/24][Resend] ACPI / hotplug / PCI: Drop crit_sect locking Rafael J. Wysocki
2014-02-03 23:33       ` [PATCH 14/24][Update] ACPI / hotplug / PCI: Simplify hotplug_event() Rafael J. Wysocki
2014-02-03 23:33       ` [PATCH 15/24][Resend] ACPI / hotplug / PCI: Use acpi_handle_debug() in hotplug_event() Rafael J. Wysocki
2014-02-03 23:34       ` [PATCH 16/24][New] ACPI / hotplug / PCI: Do not pass ACPI handle to hotplug_event() Rafael J. Wysocki
2014-02-03 23:34       ` [PATCH 17/24][Resend] ACPICA: Introduce acpi_get_data_full() and rework acpi_get_data() Rafael J. Wysocki
2014-02-03 23:36       ` [PATCH 18/24][Update] ACPI / hotplug: Fix potential race in acpi_bus_notify() Rafael J. Wysocki
2014-02-03 23:36       ` [PATCH 19/24][Update] ACPI / hotplug / PCI: Define hotplug context lock in the core Rafael J. Wysocki
2014-02-03 23:37       ` [PATCH 20/24][Update] ACPI / hotplug / PCI: Consolidate ACPIPHP with ACPI core hotplug Rafael J. Wysocki
2014-02-06 13:07         ` [Update][PATCH 20/24] " Rafael J. Wysocki
2014-02-06 23:35           ` [Update 2x][PATCH " Rafael J. Wysocki
2014-02-03 23:38       ` [PATCH 21/24][Resend] ACPI / hotplug / PCI: Rework the handling of eject requests Rafael J. Wysocki
2014-02-03 23:39       ` [PATCH 22/24][Resend] ACPI / hotplug / PCI: Simplify acpi_install_hotplug_notify_handler() Rafael J. Wysocki
2014-02-06 13:09         ` [Update][PATCH 22/24] " Rafael J. Wysocki
2014-02-03 23:40       ` [PATCH 23/24][Update] ACPI / hotplug / PCI: Hotplug notifications from acpi_bus_notify() Rafael J. Wysocki
2014-02-06 13:10         ` [Update][PATCH 23/24] " Rafael J. Wysocki
2014-02-03 23:41       ` [PATCH 24/24][New] ACPI / hotplug / PCI: Rework acpiphp_check_host_bridge() Rafael J. Wysocki
2014-02-05 10:14       ` [PATCH v2 0/24] ACPI / hotplug / PCI: ACPIPHP updates and consolidation with ACPI core Mika Westerberg
2014-02-05 10:59         ` Rafael J. Wysocki
2014-02-11  0:19 ` [PATCH 0/11] ACPI / hotplug / PCI: Updates on top of changes merged recently Bjorn Helgaas
2014-02-11 12:53   ` Rafael J. Wysocki

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=10544516.0W0fJK2WE2@vostro.rjw.lan \
    --to=rjw@rjwysocki.net \
    --cc=aaron.lu@intel.com \
    --cc=bhelgaas@google.com \
    --cc=linux-acpi@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-pci@vger.kernel.org \
    --cc=mika.westerberg@linux.intel.com \
    --cc=robert.moore@intel.com \
    --subject='Re: [PATCH v2 1/6] ACPI / hotplug: Fix theoretical race in acpi_hotplug_notify_cb()' \
    /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

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).