From: Jiang Liu <jiang.liu@linux.intel.com>
To: "Rafael J. Wysocki" <rjw@rjwysocki.net>,
Thomas Gleixner <tglx@linutronix.de>,
Bjorn Helgaas <bhelgaas@google.com>,
Yinghai Lu <yinghai@kernel.org>, Borislav Petkov <bp@alien8.de>,
Lv Zheng <lv.zheng@intel.com>, Ingo Molnar <mingo@redhat.com>,
"H. Peter Anvin" <hpa@zytor.com>,
x86@kernel.org, Len Brown <lenb@kernel.org>
Cc: Jiang Liu <jiang.liu@linux.intel.com>,
Tony Luck <tony.luck@intel.com>,
linux-kernel@vger.kernel.org, linux-pci@vger.kernel.org,
linux-acpi@vger.kernel.org
Subject: [Patch v2 21/23] x86/PCI: Refine the way to release PCI IRQ resources
Date: Mon, 2 Feb 2015 10:43:06 +0800 [thread overview]
Message-ID: <1422844988-13854-22-git-send-email-jiang.liu@linux.intel.com> (raw)
In-Reply-To: <1422844988-13854-1-git-send-email-jiang.liu@linux.intel.com>
Some PCI device drivers assume that pci_dev->irq won't change after
calling pci_disable_device() and pci_enable_device() during suspend and
resume.
Commit c03b3b0738a5 ("x86, irq, mpparse: Release IOAPIC pin when
PCI device is disabled") frees PCI IRQ resources when pci_disable_device()
is called and reallocate IRQ resources when pci_enable_device() is
called again. This breaks above assumption. So commit 3eec595235c1
("x86, irq, PCI: Keep IRQ assignment for PCI devices during
suspend/hibernation") and 9eabc99a635a ("x86, irq, PCI: Keep IRQ
assignment for runtime power management") fix the issue by avoiding
freeing/reallocating IRQ resources during PCI device suspend/resume.
They achieve this by checking dev.power.is_prepared and
dev.power.runtime_status. PM maintainer, Rafael, then pointed out that
it's really an ugly fix which leaking PM internal state information to
IRQ subsystem.
Recently David Vrabel <david.vrabel@citrix.com> also reports an
regression in pciback driver caused by commit cffe0a2b5a34 ("x86, irq:
Keep balance of IOAPIC pin reference count"). Please refer to:
http://lkml.org/lkml/2015/1/14/546
So this patch refine the way to release PCI IRQ resources. Instead of
releasing PCI IRQ resources in pci_disable_device()/
pcibios_disable_device(), we now release it at driver unbinding
notification BUS_NOTIFY_UNBOUND_DRIVER. In other word, we only release
PCI IRQ resources when there's no driver bound to the PCI device, and
it keeps the assumption that pci_dev->irq won't through multiple
invocation of pci_enable_device()/pci_disable_device().
Signed-off-by: Jiang Liu <jiang.liu@linux.intel.com>
---
arch/x86/include/asm/pci_x86.h | 2 --
arch/x86/pci/common.c | 30 ++++++++++++++++++++++++++++--
arch/x86/pci/intel_mid_pci.c | 4 ++--
arch/x86/pci/irq.c | 15 +--------------
drivers/acpi/pci_irq.c | 9 +--------
5 files changed, 32 insertions(+), 28 deletions(-)
diff --git a/arch/x86/include/asm/pci_x86.h b/arch/x86/include/asm/pci_x86.h
index 164e3f8d3c3d..fa1195dae425 100644
--- a/arch/x86/include/asm/pci_x86.h
+++ b/arch/x86/include/asm/pci_x86.h
@@ -93,8 +93,6 @@ extern raw_spinlock_t pci_config_lock;
extern int (*pcibios_enable_irq)(struct pci_dev *dev);
extern void (*pcibios_disable_irq)(struct pci_dev *dev);
-extern bool mp_should_keep_irq(struct device *dev);
-
struct pci_raw_ops {
int (*read)(unsigned int domain, unsigned int bus, unsigned int devfn,
int reg, int len, u32 *val);
diff --git a/arch/x86/pci/common.c b/arch/x86/pci/common.c
index 7b20bccf3648..99f15ed19f38 100644
--- a/arch/x86/pci/common.c
+++ b/arch/x86/pci/common.c
@@ -497,6 +497,25 @@ void __init pcibios_set_cache_line_size(void)
}
}
+static int pci_irq_notifier(struct notifier_block *nb, unsigned long action,
+ void *data)
+{
+ struct pci_dev *dev = to_pci_dev(data);
+
+ if (action != BUS_NOTIFY_UNBOUND_DRIVER)
+ return NOTIFY_DONE;
+
+ if (pcibios_disable_irq)
+ pcibios_disable_irq(dev);
+
+ return NOTIFY_OK;
+}
+
+static struct notifier_block pci_irq_nb = {
+ .notifier_call = pci_irq_notifier,
+ .priority = INT_MIN,
+};
+
int __init pcibios_init(void)
{
if (!raw_pci_ops) {
@@ -509,6 +528,9 @@ int __init pcibios_init(void)
if (pci_bf_sort >= pci_force_bf)
pci_sort_breadthfirst();
+
+ bus_register_notifier(&pci_bus_type, &pci_irq_nb);
+
return 0;
}
@@ -669,8 +691,12 @@ int pcibios_enable_device(struct pci_dev *dev, int mask)
void pcibios_disable_device (struct pci_dev *dev)
{
- if (!pci_dev_msi_enabled(dev) && pcibios_disable_irq)
- pcibios_disable_irq(dev);
+ /*
+ * Some device drivers assume dev->irq won't change after calling
+ * pci_disable_device(). So delay releasing of IRQ resource to driver
+ * unbinding time. Otherwise it will break PM subsystem and drivers
+ * like xen-pciback etc.
+ */
}
int pci_ext_cfg_avail(void)
diff --git a/arch/x86/pci/intel_mid_pci.c b/arch/x86/pci/intel_mid_pci.c
index 44b9271580b5..95c2471f6819 100644
--- a/arch/x86/pci/intel_mid_pci.c
+++ b/arch/x86/pci/intel_mid_pci.c
@@ -234,10 +234,10 @@ static int intel_mid_pci_irq_enable(struct pci_dev *dev)
static void intel_mid_pci_irq_disable(struct pci_dev *dev)
{
- if (!mp_should_keep_irq(&dev->dev) && dev->irq_managed &&
- dev->irq > 0) {
+ if (dev->irq_managed && dev->irq > 0) {
mp_unmap_irq(dev->irq);
dev->irq_managed = 0;
+ dev->irq = 0;
}
}
diff --git a/arch/x86/pci/irq.c b/arch/x86/pci/irq.c
index 5dc6ca5e1741..e71b3dbd87b8 100644
--- a/arch/x86/pci/irq.c
+++ b/arch/x86/pci/irq.c
@@ -1256,22 +1256,9 @@ static int pirq_enable_irq(struct pci_dev *dev)
return 0;
}
-bool mp_should_keep_irq(struct device *dev)
-{
- if (dev->power.is_prepared)
- return true;
-#ifdef CONFIG_PM
- if (dev->power.runtime_status == RPM_SUSPENDING)
- return true;
-#endif
-
- return false;
-}
-
static void pirq_disable_irq(struct pci_dev *dev)
{
- if (io_apic_assign_pci_irqs && !mp_should_keep_irq(&dev->dev) &&
- dev->irq_managed && dev->irq) {
+ if (io_apic_assign_pci_irqs && dev->irq_managed && dev->irq) {
mp_unmap_irq(dev->irq);
dev->irq = 0;
dev->irq_managed = 0;
diff --git a/drivers/acpi/pci_irq.c b/drivers/acpi/pci_irq.c
index b1def411c0b8..e7f718d6918a 100644
--- a/drivers/acpi/pci_irq.c
+++ b/drivers/acpi/pci_irq.c
@@ -485,14 +485,6 @@ void acpi_pci_irq_disable(struct pci_dev *dev)
if (!pin || !dev->irq_managed || dev->irq <= 0)
return;
- /* Keep IOAPIC pin configuration when suspending */
- if (dev->dev.power.is_prepared)
- return;
-#ifdef CONFIG_PM
- if (dev->dev.power.runtime_status == RPM_SUSPENDING)
- return;
-#endif
-
entry = acpi_pci_irq_lookup(dev, pin);
if (!entry)
return;
@@ -513,5 +505,6 @@ void acpi_pci_irq_disable(struct pci_dev *dev)
if (gsi >= 0) {
acpi_unregister_gsi(gsi);
dev->irq_managed = 0;
+ dev->irq = 0;
}
}
--
1.7.10.4
next prev parent reply other threads:[~2015-02-02 2:42 UTC|newest]
Thread overview: 31+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-02-02 2:42 [Patch v2 00/23] Improve ACPI resource parsing interfaces and enable IOAPIC hotplug Jiang Liu
2015-02-02 2:42 ` [Patch v2 01/23] ACPICA: Resources: Provide common part for struct acpi_resource_address structures Jiang Liu
2015-02-02 2:42 ` [Patch v2 02/23] ACPI: Remove redundant check in function acpi_dev_resource_address_space() Jiang Liu
2015-02-02 2:42 ` [Patch v2 03/23] ACPI: Implement proper length checks for mem resources Jiang Liu
2015-02-02 2:42 ` [Patch v2 04/23] ACPI: Use the length check for io resources as well Jiang Liu
2015-02-02 2:42 ` [Patch v2 05/23] ACPI: Let the parser return false for disabled resources Jiang Liu
2015-02-02 2:42 ` [Patch v2 06/23] ACPI: Unify the parsing of address_space and ext_address_space Jiang Liu
2015-02-02 2:42 ` [Patch v2 07/23] ACPI: Move the window flag logic to the combined parser Jiang Liu
2015-02-02 2:42 ` [Patch v2 08/23] ACPI: Add prefetch decoding to the address space parser Jiang Liu
2015-02-02 2:42 ` [Patch v2 09/23] ACPI: Fix a bug in parsing ACPI Memory24 resource Jiang Liu
2015-02-02 2:42 ` [Patch v2 10/23] ACPI: Normalize return value of resource parser functions Jiang Liu
2015-02-02 2:42 ` [Patch v2 11/23] ACPI: Set flag IORESOURCE_UNSET for unassigned resources Jiang Liu
2015-02-02 2:42 ` [Patch v2 12/23] ACPI: Enforce stricter checks for address space descriptors Jiang Liu
2015-02-02 2:42 ` [Patch v2 13/23] ACPI: Return translation offset when parsing ACPI address space resources Jiang Liu
2015-02-02 2:42 ` [Patch v2 14/23] ACPI: Translate resource into master side address for bridge window resources Jiang Liu
2015-02-02 2:43 ` [Patch v2 15/23] ACPI: Add field offset to struct resource_list_entry Jiang Liu
2015-02-02 2:43 ` [Patch v2 16/23] ACPI: Introduce helper function acpi_dev_filter_resource_type() Jiang Liu
2015-02-02 2:43 ` [Patch v2 17/23] resources: Move struct resource_list_entry from ACPI into resource core Jiang Liu
2015-02-02 2:43 ` [Patch v2 18/23] PCI: Use common resource list management code instead of private implementation Jiang Liu
2015-02-02 2:43 ` [Patch v2 19/23] x86/PCI: Fix the range check for IO resources Jiang Liu
2015-02-02 2:43 ` [Patch v2 20/23] x86/PCI/ACPI: Use common ACPI resource interfaces to simplify implementation Jiang Liu
2015-02-02 2:43 ` Jiang Liu [this message]
2015-02-03 21:57 ` [Patch v2 21/23] x86/PCI: Refine the way to release PCI IRQ resources Rafael J. Wysocki
2015-02-04 8:35 ` [Patch v3] " Jiang Liu
2015-02-04 13:26 ` Rafael J. Wysocki
2015-02-02 2:43 ` [Patch v2 22/23] ACPI: Add interfaces to parse IOAPIC ID for IOAPIC hotplug Jiang Liu
2015-02-02 2:43 ` [Patch v2 23/23] x86/irq, ACPI: Implement ACPI driver to support " Jiang Liu
2015-02-02 3:15 ` [Patch v2 00/23] Improve ACPI resource parsing interfaces and enable " Jiang Liu
2015-02-03 21:54 ` Rafael J. Wysocki
2015-02-04 5:32 ` Jiang Liu
2015-02-04 13:30 ` 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=1422844988-13854-22-git-send-email-jiang.liu@linux.intel.com \
--to=jiang.liu@linux.intel.com \
--cc=bhelgaas@google.com \
--cc=bp@alien8.de \
--cc=hpa@zytor.com \
--cc=lenb@kernel.org \
--cc=linux-acpi@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-pci@vger.kernel.org \
--cc=lv.zheng@intel.com \
--cc=mingo@redhat.com \
--cc=rjw@rjwysocki.net \
--cc=tglx@linutronix.de \
--cc=tony.luck@intel.com \
--cc=x86@kernel.org \
--cc=yinghai@kernel.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).