[PATCH v2] pv-ops: register xen pci notifier

[PATCH v2] pv-ops: register xen pci notifier

[Weidong Han]

Register the notifier to handle hot-plug devices and SR-IOV devices
for Xen hypervisor. When a device is hot added or removed, it adds
or removes it to Xen via hypercalls.

Changes in v2:
Remove inline #ifdef and the awkward dangling else/#endif construction,
and rather than using memset, use variable declaration and initializer
to assign the elements in xen_add_device.


Signed-off-by: Weidong Han <weidong.han@intel.com>
---
 drivers/xen/Makefile            |    5 +-
 drivers/xen/pci.c               |  118 +++++++++++++++++++++++++++++++++++++++
 include/xen/interface/physdev.h |   21 +++++++
 3 files changed, 142 insertions(+), 2 deletions(-)
 create mode 100644 drivers/xen/pci.c

diff --git a/drivers/xen/Makefile b/drivers/xen/Makefile
index 007aa99..fd5c88c 100644
--- a/drivers/xen/Makefile
+++ b/drivers/xen/Makefile
@@ -1,12 +1,13 @@
 obj-y	+= grant-table.o features.o events.o manage.o biomerge.o
 obj-y	+= xenbus/
 
+obj-$(CONFIG_PCI)			+= pci.o
 obj-$(CONFIG_HOTPLUG_CPU)		+= cpu_hotplug.o
 obj-$(CONFIG_XEN_XENCOMM)		+= xencomm.o
 obj-$(CONFIG_XEN_BALLOON)		+= balloon.o
 obj-$(CONFIG_XEN_DEV_EVTCHN)		+= evtchn.o
-obj-$(CONFIG_XEN_GNTDEV)	+= gntdev.o
+obj-$(CONFIG_XEN_GNTDEV)		+= gntdev.o
 obj-$(CONFIG_XEN_BLKDEV_BACKEND)	+= blkback/
 obj-$(CONFIG_XEN_NETDEV_BACKEND)	+= netback/
 obj-$(CONFIG_XENFS)			+= xenfs/
-obj-$(CONFIG_XEN_SYS_HYPERVISOR)	+= sys-hypervisor.o
\ No newline at end of file
+obj-$(CONFIG_XEN_SYS_HYPERVISOR)	+= sys-hypervisor.o
diff --git a/drivers/xen/pci.c b/drivers/xen/pci.c
new file mode 100644
index 0000000..8df0445
--- /dev/null
+++ b/drivers/xen/pci.c
@@ -0,0 +1,118 @@
+/*
+ * Copyright (c) 2009, Intel Corporation.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License along with
+ * this program; if not, write to the Free Software Foundation, Inc., 59 Temple
+ * Place - Suite 330, Boston, MA 02111-1307 USA.
+ *
+ * Author: Weidong Han <weidong.han@intel.com>
+ */
+
+#include <linux/pci.h>
+#include <xen/interface/physdev.h>
+#include <asm/xen/hypercall.h>
+#include "../pci/pci.h"
+
+
+#ifdef CONFIG_PCI_IOV
+#define HANDLE_PCI_IOV	1
+#else
+#define HANDLE_PCI_IOV	0
+#endif
+
+static int xen_add_device(struct device *dev)
+{
+	int r;
+	struct pci_dev *pci_dev = to_pci_dev(dev);
+
+	if (HANDLE_PCI_IOV && pci_dev->is_virtfn) {
+		struct physdev_manage_pci_ext manage_pci_ext = {
+			.bus		= pci_dev->bus->number,
+			.devfn		= pci_dev->devfn,
+			.is_extfn	= 0,
+			.is_virtfn 	= 1,
+			.physfn.bus	= pci_dev->physfn->bus->number,
+			.physfn.devfn	= pci_dev->physfn->devfn,
+		};
+
+		r = HYPERVISOR_physdev_op(PHYSDEVOP_manage_pci_add_ext,
+			&manage_pci_ext);
+	} else if (pci_ari_enabled(pci_dev->bus) && PCI_SLOT(pci_dev->devfn)) {
+		struct physdev_manage_pci_ext manage_pci_ext = {
+			.bus		= pci_dev->bus->number,
+			.devfn		= pci_dev->devfn,
+			.is_extfn	= 1,
+			.is_virtfn 	= 0,
+			.physfn.bus	= 0,
+			.physfn.devfn	= 0,
+		};
+
+		r = HYPERVISOR_physdev_op(PHYSDEVOP_manage_pci_add_ext,
+			&manage_pci_ext);
+	} else {
+		struct physdev_manage_pci manage_pci = {
+			.bus 	= pci_dev->bus->number,
+			.devfn	= pci_dev->devfn,
+		};
+
+		r = HYPERVISOR_physdev_op(PHYSDEVOP_manage_pci_add,
+			&manage_pci);
+	}
+
+	return r;
+}
+
+static int xen_remove_device(struct device *dev)
+{
+	int r;
+	struct pci_dev *pci_dev = to_pci_dev(dev);
+	struct physdev_manage_pci manage_pci;
+
+	manage_pci.bus = pci_dev->bus->number;
+	manage_pci.devfn = pci_dev->devfn;
+
+	r = HYPERVISOR_physdev_op(PHYSDEVOP_manage_pci_remove,
+		&manage_pci);
+
+	return r;
+}
+
+static int xen_pci_notifier(struct notifier_block *nb,
+			    unsigned long action, void *data)
+{
+	struct device *dev = data;
+	int r = 0;
+
+	switch (action) {
+	case BUS_NOTIFY_ADD_DEVICE:
+		r = xen_add_device(dev);
+		break;
+	case BUS_NOTIFY_DEL_DEVICE:
+		r = xen_remove_device(dev);
+		break;
+	default:
+		break;
+	}
+
+	return r;
+}
+
+struct notifier_block device_nb = {
+	.notifier_call = xen_pci_notifier,
+};
+
+void __init register_xen_pci_notifier(void)
+{
+	bus_register_notifier(&pci_bus_type, &device_nb);
+}
+
+fs_initcall(register_xen_pci_notifier);
diff --git a/include/xen/interface/physdev.h b/include/xen/interface/physdev.h
index cd69391..dcae970 100644
--- a/include/xen/interface/physdev.h
+++ b/include/xen/interface/physdev.h
@@ -106,6 +106,27 @@ struct physdev_irq {
 	uint32_t vector;
 };
 
+#define PHYSDEVOP_manage_pci_add         15
+#define PHYSDEVOP_manage_pci_remove      16
+struct physdev_manage_pci {
+	/* IN */
+	uint8_t bus;
+	uint8_t devfn;
+};
+
+#define PHYSDEVOP_manage_pci_add_ext     20
+struct physdev_manage_pci_ext {
+	/* IN */
+	uint8_t bus;
+	uint8_t devfn;
+	unsigned is_extfn;
+	unsigned is_virtfn;
+	struct {
+		uint8_t bus;
+		uint8_t devfn;
+	} physfn;
+};
+
 /*
  * Argument to physdev_op_compat() hypercall. Superceded by new physdev_op()
  * hypercall since 0x00030202.
-- 
1.6.0.4

Re: [PATCH v2] pv-ops: register xen pci notifier

[Jeremy Fitzhardinge]

On 07/28/09 19:38, Weidong Han wrote:
> Register the notifier to handle hot-plug devices and SR-IOV devices
> for Xen hypervisor. When a device is hot added or removed, it adds
> or removes it to Xen via hypercalls.
>
> Changes in v2:
> Remove inline #ifdef and the awkward dangling else/#endif construction,
> and rather than using memset, use variable declaration and initializer
> to assign the elements in xen_add_device.
>   

That looks much better.  Just one tiny nit:

> +
> +	if (HANDLE_PCI_IOV && pci_dev->is_virtfn) {
> +		struct physdev_manage_pci_ext manage_pci_ext = {
> +			.bus		= pci_dev->bus->number,
> +			.devfn		= pci_dev->devfn,
> +			.is_extfn	= 0,
> +			.is_virtfn 	= 1,
> +			.physfn.bus	= pci_dev->physfn->bus->number,
> +			.physfn.devfn	= pci_dev->physfn->devfn,
> +		};
> +
> +		r = HYPERVISOR_physdev_op(PHYSDEVOP_manage_pci_add_ext,
> +			&manage_pci_ext);
> +	} else if (pci_ari_enabled(pci_dev->bus) && PCI_SLOT(pci_dev->devfn)) {
> +		struct physdev_manage_pci_ext manage_pci_ext = {
> +			.bus		= pci_dev->bus->number,
> +			.devfn		= pci_dev->devfn,
> +			.is_extfn	= 1,
> +			.is_virtfn 	= 0,
> +			.physfn.bus	= 0,
> +			.physfn.devfn	= 0,
>   

It isn't necessarily to explicitly initialize elements to 0; that will
happen implicitly as a result of using an initializer.

Thanks,
    J

RE: [PATCH v2] pv-ops: register xen pci notifier

[Han, Weidong]

Jeremy Fitzhardinge wrote:
> On 07/28/09 19:38, Weidong Han wrote:
>> Register the notifier to handle hot-plug devices and SR-IOV devices
>> for Xen hypervisor. When a device is hot added or removed, it adds
>> or removes it to Xen via hypercalls.
>> 
>> Changes in v2:
>> Remove inline #ifdef and the awkward dangling else/#endif
>> construction, and rather than using memset, use variable declaration
>> and initializer to assign the elements in xen_add_device.
>> 
> 
> That looks much better.  Just one tiny nit:
> 
>> +
>> +	if (HANDLE_PCI_IOV && pci_dev->is_virtfn) {
>> +		struct physdev_manage_pci_ext manage_pci_ext = {
>> +			.bus		= pci_dev->bus->number,
>> +			.devfn		= pci_dev->devfn,
>> +			.is_extfn	= 0,
>> +			.is_virtfn 	= 1,
>> +			.physfn.bus	= pci_dev->physfn->bus->number,
>> +			.physfn.devfn	= pci_dev->physfn->devfn,
>> +		};
>> +
>> +		r = HYPERVISOR_physdev_op(PHYSDEVOP_manage_pci_add_ext,
>> +			&manage_pci_ext); +	} else if (pci_ari_enabled(pci_dev->bus) &&
>> PCI_SLOT(pci_dev->devfn)) { +		struct physdev_manage_pci_ext
>> manage_pci_ext = { +			.bus		= pci_dev->bus->number,
>> +			.devfn		= pci_dev->devfn,
>> +			.is_extfn	= 1,
>> +			.is_virtfn 	= 0,
>> +			.physfn.bus	= 0,
>> +			.physfn.devfn	= 0,
>> 
> 
> It isn't necessarily to explicitly initialize elements to 0; that will
> happen implicitly as a result of using an initializer.
> 

Will remove the unnecessary initialization, and send out the updated patch soon. Thanks.

Regards,
Weidong