* Re: virtio scsi host draft specification, v3
From: Michael S. Tsirkin @ 2011-06-29 10:31 UTC (permalink / raw)
To: Paolo Bonzini
Cc: Christoph Hellwig, Stefan Hajnoczi, kvm@vger.kernel.org,
Linux Kernel Mailing List, qemu-devel, Christoph Hellwig,
Linux Virtualization
In-Reply-To: <4E0AF925.2050707@redhat.com>
On Wed, Jun 29, 2011 at 12:06:29PM +0200, Paolo Bonzini wrote:
> On 06/29/2011 12:03 PM, Christoph Hellwig wrote:
> >> I agree here, in fact I misread Hannes's comment as "if a driver
> >> uses more than one queue it is responsibility of the driver to
> >> ensure strict request ordering". If you send requests to different
> >> queues, you know that those requests are independent. I don't think
> >> anything else is feasible in the virtio framework.
> >
> >That doesn't really fit very well with the SAM model. If we want
> >to use multiple queues for a single LUN it has to be transparent to
> >the SCSI command stream. Then again I don't quite see the use for
> >that anyway.
>
> Agreed, I see a use for multiple queues (MSI-X), but not for
> multiple queues shared by a single LUN.
>
> Paolo
Then let's make it explicit in the spec?
--
MST
^ permalink raw reply
* Re: virtio scsi host draft specification, v3
From: Paolo Bonzini @ 2011-06-29 10:35 UTC (permalink / raw)
To: Michael S. Tsirkin
Cc: Christoph Hellwig, Stefan Hajnoczi, kvm@vger.kernel.org,
Linux Kernel Mailing List, qemu-devel, Christoph Hellwig,
Linux Virtualization
In-Reply-To: <20110629103139.GC16757@redhat.com>
On 06/29/2011 12:31 PM, Michael S. Tsirkin wrote:
> On Wed, Jun 29, 2011 at 12:06:29PM +0200, Paolo Bonzini wrote:
>> On 06/29/2011 12:03 PM, Christoph Hellwig wrote:
>>>> I agree here, in fact I misread Hannes's comment as "if a driver
>>>> uses more than one queue it is responsibility of the driver to
>>>> ensure strict request ordering". If you send requests to different
>>>> queues, you know that those requests are independent. I don't think
>>>> anything else is feasible in the virtio framework.
>>>
>>> That doesn't really fit very well with the SAM model. If we want
>>> to use multiple queues for a single LUN it has to be transparent to
>>> the SCSI command stream. Then again I don't quite see the use for
>>> that anyway.
>>
>> Agreed, I see a use for multiple queues (MSI-X), but not for
>> multiple queues shared by a single LUN.
>
> Then let's make it explicit in the spec?
What, forbid it or say ordering is not guaranteed? The latter is
already explicit with the wording suggested in the thread.
Paolo
^ permalink raw reply
* [PATCH 1/4] xen: Populate xenbus device attributes
From: Bastian Blank @ 2011-06-29 12:39 UTC (permalink / raw)
To: xen-devel, virtualization; +Cc: linux-kernel
The xenbus bus type uses device_create_file to assign all used device
attributes. However it does not remove them when the device goes away.
This patch uses the dev_attrs field of the bus type to specify default
attributes for all devices.
Signed-off-by: Bastian Blank <waldi@debian.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
---
drivers/xen/xenbus/xenbus_probe.c | 41 +++++++++------------------
drivers/xen/xenbus/xenbus_probe.h | 2 +
drivers/xen/xenbus/xenbus_probe_backend.c | 6 +---
drivers/xen/xenbus/xenbus_probe_frontend.c | 6 +---
4 files changed, 18 insertions(+), 37 deletions(-)
diff --git a/drivers/xen/xenbus/xenbus_probe.c b/drivers/xen/xenbus/xenbus_probe.c
index 7397695..2ed0b04 100644
--- a/drivers/xen/xenbus/xenbus_probe.c
+++ b/drivers/xen/xenbus/xenbus_probe.c
@@ -378,26 +378,31 @@ static void xenbus_dev_release(struct device *dev)
kfree(to_xenbus_device(dev));
}
-static ssize_t xendev_show_nodename(struct device *dev,
- struct device_attribute *attr, char *buf)
+static ssize_t nodename_show(struct device *dev,
+ struct device_attribute *attr, char *buf)
{
return sprintf(buf, "%s\n", to_xenbus_device(dev)->nodename);
}
-static DEVICE_ATTR(nodename, S_IRUSR | S_IRGRP | S_IROTH, xendev_show_nodename, NULL);
-static ssize_t xendev_show_devtype(struct device *dev,
- struct device_attribute *attr, char *buf)
+static ssize_t devtype_show(struct device *dev,
+ struct device_attribute *attr, char *buf)
{
return sprintf(buf, "%s\n", to_xenbus_device(dev)->devicetype);
}
-static DEVICE_ATTR(devtype, S_IRUSR | S_IRGRP | S_IROTH, xendev_show_devtype, NULL);
-static ssize_t xendev_show_modalias(struct device *dev,
- struct device_attribute *attr, char *buf)
+static ssize_t modalias_show(struct device *dev,
+ struct device_attribute *attr, char *buf)
{
return sprintf(buf, "xen:%s\n", to_xenbus_device(dev)->devicetype);
}
-static DEVICE_ATTR(modalias, S_IRUSR | S_IRGRP | S_IROTH, xendev_show_modalias, NULL);
+
+struct device_attribute xenbus_dev_attrs[] = {
+ __ATTR_RO(nodename),
+ __ATTR_RO(devtype),
+ __ATTR_RO(modalias),
+ __ATTR_NULL
+};
+EXPORT_SYMBOL_GPL(xenbus_dev_attrs);
int xenbus_probe_node(struct xen_bus_type *bus,
const char *type,
@@ -449,25 +454,7 @@ int xenbus_probe_node(struct xen_bus_type *bus,
if (err)
goto fail;
- err = device_create_file(&xendev->dev, &dev_attr_nodename);
- if (err)
- goto fail_unregister;
-
- err = device_create_file(&xendev->dev, &dev_attr_devtype);
- if (err)
- goto fail_remove_nodename;
-
- err = device_create_file(&xendev->dev, &dev_attr_modalias);
- if (err)
- goto fail_remove_devtype;
-
return 0;
-fail_remove_devtype:
- device_remove_file(&xendev->dev, &dev_attr_devtype);
-fail_remove_nodename:
- device_remove_file(&xendev->dev, &dev_attr_nodename);
-fail_unregister:
- device_unregister(&xendev->dev);
fail:
kfree(xendev);
return err;
diff --git a/drivers/xen/xenbus/xenbus_probe.h b/drivers/xen/xenbus/xenbus_probe.h
index 888b990..b814935 100644
--- a/drivers/xen/xenbus/xenbus_probe.h
+++ b/drivers/xen/xenbus/xenbus_probe.h
@@ -48,6 +48,8 @@ struct xen_bus_type
struct bus_type bus;
};
+extern struct device_attribute xenbus_dev_attrs[];
+
extern int xenbus_match(struct device *_dev, struct device_driver *_drv);
extern int xenbus_dev_probe(struct device *_dev);
extern int xenbus_dev_remove(struct device *_dev);
diff --git a/drivers/xen/xenbus/xenbus_probe_backend.c b/drivers/xen/xenbus/xenbus_probe_backend.c
index 6cf467b..ec510e5 100644
--- a/drivers/xen/xenbus/xenbus_probe_backend.c
+++ b/drivers/xen/xenbus/xenbus_probe_backend.c
@@ -183,10 +183,6 @@ static void frontend_changed(struct xenbus_watch *watch,
xenbus_otherend_changed(watch, vec, len, 0);
}
-static struct device_attribute xenbus_backend_dev_attrs[] = {
- __ATTR_NULL
-};
-
static struct xen_bus_type xenbus_backend = {
.root = "backend",
.levels = 3, /* backend/type/<frontend>/<id> */
@@ -200,7 +196,7 @@ static struct xen_bus_type xenbus_backend = {
.probe = xenbus_dev_probe,
.remove = xenbus_dev_remove,
.shutdown = xenbus_dev_shutdown,
- .dev_attrs = xenbus_backend_dev_attrs,
+ .dev_attrs = xenbus_dev_attrs,
},
};
diff --git a/drivers/xen/xenbus/xenbus_probe_frontend.c b/drivers/xen/xenbus/xenbus_probe_frontend.c
index b6a2690..ed2ba47 100644
--- a/drivers/xen/xenbus/xenbus_probe_frontend.c
+++ b/drivers/xen/xenbus/xenbus_probe_frontend.c
@@ -81,10 +81,6 @@ static void backend_changed(struct xenbus_watch *watch,
xenbus_otherend_changed(watch, vec, len, 1);
}
-static struct device_attribute xenbus_frontend_dev_attrs[] = {
- __ATTR_NULL
-};
-
static const struct dev_pm_ops xenbus_pm_ops = {
.suspend = xenbus_dev_suspend,
.resume = xenbus_dev_resume,
@@ -106,7 +102,7 @@ static struct xen_bus_type xenbus_frontend = {
.probe = xenbus_dev_probe,
.remove = xenbus_dev_remove,
.shutdown = xenbus_dev_shutdown,
- .dev_attrs = xenbus_frontend_dev_attrs,
+ .dev_attrs = xenbus_dev_attrs,
.pm = &xenbus_pm_ops,
},
--
1.7.5.4
^ permalink raw reply related
* [PATCH 2/4] xen: Add module alias to autoload backend drivers
From: Bastian Blank @ 2011-06-29 12:40 UTC (permalink / raw)
To: xen-devel, virtualization; +Cc: linux-kernel
All the Xen backend drivers are assigned to a special bus type
xen-backend. This patch exports xen-backend:* names through modalias and
uevent to autoload them.
Signed-off-by: Bastian Blank <waldi@debian.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
---
drivers/xen/xenbus/xenbus_probe.c | 3 ++-
drivers/xen/xenbus/xenbus_probe_backend.c | 3 +++
2 files changed, 5 insertions(+), 1 deletions(-)
diff --git a/drivers/xen/xenbus/xenbus_probe.c b/drivers/xen/xenbus/xenbus_probe.c
index 2ed0b04..bd2f90c 100644
--- a/drivers/xen/xenbus/xenbus_probe.c
+++ b/drivers/xen/xenbus/xenbus_probe.c
@@ -393,7 +393,8 @@ static ssize_t devtype_show(struct device *dev,
static ssize_t modalias_show(struct device *dev,
struct device_attribute *attr, char *buf)
{
- return sprintf(buf, "xen:%s\n", to_xenbus_device(dev)->devicetype);
+ return sprintf(buf, "%s:%s\n", dev->bus->name,
+ to_xenbus_device(dev)->devicetype);
}
struct device_attribute xenbus_dev_attrs[] = {
diff --git a/drivers/xen/xenbus/xenbus_probe_backend.c b/drivers/xen/xenbus/xenbus_probe_backend.c
index ec510e5..60adf91 100644
--- a/drivers/xen/xenbus/xenbus_probe_backend.c
+++ b/drivers/xen/xenbus/xenbus_probe_backend.c
@@ -107,6 +107,9 @@ static int xenbus_uevent_backend(struct device *dev,
if (xdev == NULL)
return -ENODEV;
+ if (add_uevent_var(env, "MODALIAS=xen-backend:%s", xdev->devicetype))
+ return -ENOMEM;
+
/* stuff we want to pass to /sbin/hotplug */
if (add_uevent_var(env, "XENBUS_TYPE=%s", xdev->devicetype))
return -ENOMEM;
--
1.7.5.4
^ permalink raw reply related
* [PATCH 3/4] xen/blkback: Add module alias for autoloading
From: Bastian Blank @ 2011-06-29 12:40 UTC (permalink / raw)
To: xen-devel, virtualization, David S. Miller; +Cc: linux-kernel
Add xen-backend:vbd module alias to the xen-blkback module. This allows
automatic loading of the module.
Signed-off-by: Bastian Blank <waldi@debian.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
---
drivers/block/xen-blkback/blkback.c | 1 +
1 files changed, 1 insertions(+), 0 deletions(-)
diff --git a/drivers/block/xen-blkback/blkback.c b/drivers/block/xen-blkback/blkback.c
index 5cf2993..ed62008 100644
--- a/drivers/block/xen-blkback/blkback.c
+++ b/drivers/block/xen-blkback/blkback.c
@@ -824,3 +824,4 @@ static int __init xen_blkif_init(void)
module_init(xen_blkif_init);
MODULE_LICENSE("Dual BSD/GPL");
+MODULE_ALIAS("xen-backend:vbd");
--
1.7.5.4
^ permalink raw reply related
* [PATCH 4/4] xen/netback: Add module alias for autoloading
From: Bastian Blank @ 2011-06-29 12:41 UTC (permalink / raw)
To: xen-devel, virtualization, Jens Axboe; +Cc: linux-kernel
Add xen-backend:vif module alias to the xen-netback module. This allows
automatic loading of the module.
Signed-off-by: Bastian Blank <waldi@debian.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
---
drivers/net/xen-netback/netback.c | 1 +
1 files changed, 1 insertions(+), 0 deletions(-)
diff --git a/drivers/net/xen-netback/netback.c b/drivers/net/xen-netback/netback.c
index 0e4851b..fd00f25 100644
--- a/drivers/net/xen-netback/netback.c
+++ b/drivers/net/xen-netback/netback.c
@@ -1743,3 +1743,4 @@ failed_init:
module_init(netback_init);
MODULE_LICENSE("Dual BSD/GPL");
+MODULE_ALIAS("xen-backend:vif");
--
1.7.5.4
^ permalink raw reply related
* [PATCH 00/40] Staging: hv: Driver cleanup
From: K. Y. Srinivasan @ 2011-06-29 14:38 UTC (permalink / raw)
To: gregkh, linux-kernel, devel, virtualization; +Cc: K. Y. Srinivasan
Further cleanup of the hv drivers:
1) Cleanup the reference counting mess for both stor and net devices.
2) Handle all block devices using the storvsc driver.
3) Accomodate some host side scsi emulation bugs.
4) In case of scsi errors off-line the device.
Regads,
K. Y
^ permalink raw reply
* [PATCH 01/40] Staging: hv: storvsc: Do not aquire an unnecessary reference on stor_device
From: K. Y. Srinivasan @ 2011-06-29 14:38 UTC (permalink / raw)
To: gregkh, linux-kernel, devel, virtualization
Cc: K. Y. Srinivasan, Haiyang Zhang, Abhishek Kane, Hank Janssen
In-Reply-To: <1309358301-8488-1-git-send-email-kys@microsoft.com>
On entry into storvsc_on_io_completion() we have already acquired a reference
on the stor_device; there is no need to acquire an additional reference here.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Abhishek Kane <v-abkane@microsoft.com>
Signed-off-by: Hank Janssen <hjanssen@microsoft.com>
---
drivers/staging/hv/storvsc.c | 5 +----
1 files changed, 1 insertions(+), 4 deletions(-)
diff --git a/drivers/staging/hv/storvsc.c b/drivers/staging/hv/storvsc.c
index 8c62829..cd38cd6 100644
--- a/drivers/staging/hv/storvsc.c
+++ b/drivers/staging/hv/storvsc.c
@@ -232,9 +232,7 @@ static void storvsc_on_io_completion(struct hv_device *device,
struct storvsc_device *stor_device;
struct vstor_packet *stor_pkt;
- stor_device = must_get_stor_device(device);
- if (!stor_device)
- return;
+ stor_device = (struct storvsc_device *)device->ext;
stor_pkt = &request->vstor_packet;
@@ -279,7 +277,6 @@ static void storvsc_on_io_completion(struct hv_device *device,
wake_up(&stor_device->waiting_to_drain);
- put_stor_device(device);
}
static void storvsc_on_receive(struct hv_device *device,
--
1.7.4.1
^ permalink raw reply related
* [PATCH 02/40] Staging: hv: storvsc: Rename must_get_stor_device()
From: K. Y. Srinivasan @ 2011-06-29 14:38 UTC (permalink / raw)
To: gregkh, linux-kernel, devel, virtualization; +Cc: Haiyang Zhang, Abhishek Kane
In-Reply-To: <1309358377-8537-1-git-send-email-kys@microsoft.com>
In preparation for cleaning up how we manage reference counts on the stor
device, clearly distinguish why we are attempting to acquire a reference.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Abhishek Kane <v-abkane@microsoft.com>
Signed-off-by: Hank Janssen <hjanssen@microsoft.com>
---
drivers/staging/hv/storvsc.c | 6 +++---
1 files changed, 3 insertions(+), 3 deletions(-)
diff --git a/drivers/staging/hv/storvsc.c b/drivers/staging/hv/storvsc.c
index cd38cd6..89708b1 100644
--- a/drivers/staging/hv/storvsc.c
+++ b/drivers/staging/hv/storvsc.c
@@ -41,7 +41,7 @@ static inline struct storvsc_device *alloc_stor_device(struct hv_device *device)
return NULL;
/* Set to 2 to allow both inbound and outbound traffics */
- /* (ie get_stor_device() and must_get_stor_device()) to proceed. */
+ /* (ie get_stor_device() and get_in_stor_device()) to proceed. */
atomic_cmpxchg(&stor_device->ref_count, 0, 2);
init_waitqueue_head(&stor_device->waiting_to_drain);
@@ -53,7 +53,7 @@ static inline struct storvsc_device *alloc_stor_device(struct hv_device *device)
/* Get the stordevice object iff exists and its refcount > 0 */
-static inline struct storvsc_device *must_get_stor_device(
+static inline struct storvsc_device *get_in_stor_device(
struct hv_device *device)
{
struct storvsc_device *stor_device;
@@ -305,7 +305,7 @@ static void storvsc_on_channel_callback(void *context)
int ret;
- stor_device = must_get_stor_device(device);
+ stor_device = get_in_stor_device(device);
if (!stor_device)
return;
--
1.7.4.1
^ permalink raw reply related
* [PATCH 03/40] Staging: hv: storvsc: Rename get_stor_device()
From: K. Y. Srinivasan @ 2011-06-29 14:39 UTC (permalink / raw)
To: gregkh, linux-kernel, devel, virtualization; +Cc: Haiyang Zhang, Abhishek Kane
In-Reply-To: <1309358377-8537-1-git-send-email-kys@microsoft.com>
In preparation for cleaning up how we manage reference counts on the stor
device, clearly distinguish why we are attempting to acquire a reference.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Abhishek Kane <v-abkane@microsoft.com>
Signed-off-by: Hank Janssen <hjanssen@microsoft.com>
---
drivers/staging/hv/hyperv_storage.h | 3 ++-
drivers/staging/hv/storvsc.c | 8 ++++----
drivers/staging/hv/storvsc_drv.c | 2 +-
3 files changed, 7 insertions(+), 6 deletions(-)
diff --git a/drivers/staging/hv/hyperv_storage.h b/drivers/staging/hv/hyperv_storage.h
index a01f9a0..a224413 100644
--- a/drivers/staging/hv/hyperv_storage.h
+++ b/drivers/staging/hv/hyperv_storage.h
@@ -288,7 +288,8 @@ struct storvsc_device {
/* Get the stordevice object iff exists and its refcount > 1 */
-static inline struct storvsc_device *get_stor_device(struct hv_device *device)
+static inline struct storvsc_device *get_out_stor_device(
+ struct hv_device *device)
{
struct storvsc_device *stor_device;
diff --git a/drivers/staging/hv/storvsc.c b/drivers/staging/hv/storvsc.c
index 89708b1..313a3f8 100644
--- a/drivers/staging/hv/storvsc.c
+++ b/drivers/staging/hv/storvsc.c
@@ -41,7 +41,7 @@ static inline struct storvsc_device *alloc_stor_device(struct hv_device *device)
return NULL;
/* Set to 2 to allow both inbound and outbound traffics */
- /* (ie get_stor_device() and get_in_stor_device()) to proceed. */
+ /* (ie get_out_stor_device() and get_in_stor_device()) to proceed. */
atomic_cmpxchg(&stor_device->ref_count, 0, 2);
init_waitqueue_head(&stor_device->waiting_to_drain);
@@ -67,7 +67,7 @@ static inline struct storvsc_device *get_in_stor_device(
return stor_device;
}
-/* Drop ref count to 1 to effectively disable get_stor_device() */
+/* Drop ref count to 1 to effectively disable get_out_stor_device() */
static inline struct storvsc_device *release_stor_device(
struct hv_device *device)
{
@@ -105,7 +105,7 @@ static int storvsc_channel_init(struct hv_device *device)
struct vstor_packet *vstor_packet;
int ret, t;
- stor_device = get_stor_device(device);
+ stor_device = get_out_stor_device(device);
if (!stor_device)
return -ENODEV;
@@ -427,7 +427,7 @@ int storvsc_do_io(struct hv_device *device,
int ret = 0;
vstor_packet = &request->vstor_packet;
- stor_device = get_stor_device(device);
+ stor_device = get_out_stor_device(device);
if (!stor_device)
return -ENODEV;
diff --git a/drivers/staging/hv/storvsc_drv.c b/drivers/staging/hv/storvsc_drv.c
index 7db5246..8d5be51 100644
--- a/drivers/staging/hv/storvsc_drv.c
+++ b/drivers/staging/hv/storvsc_drv.c
@@ -346,7 +346,7 @@ static int storvsc_host_reset(struct hv_device *device)
int ret, t;
- stor_device = get_stor_device(device);
+ stor_device = get_out_stor_device(device);
if (!stor_device)
return -ENODEV;
--
1.7.4.1
^ permalink raw reply related
* [PATCH 04/40] Staging: hv: storvsc: Cleanup alloc_stor_device()
From: K. Y. Srinivasan @ 2011-06-29 14:39 UTC (permalink / raw)
To: gregkh, linux-kernel, devel, virtualization; +Cc: Haiyang Zhang, Abhishek Kane
In-Reply-To: <1309358377-8537-1-git-send-email-kys@microsoft.com>
Cleanup alloc_stor_device(), we can set the ref_count directly.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Abhishek Kane <v-abkane@microsoft.com>
Signed-off-by: Hank Janssen <hjanssen@microsoft.com>
---
drivers/staging/hv/storvsc.c | 2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/drivers/staging/hv/storvsc.c b/drivers/staging/hv/storvsc.c
index 313a3f8..48bd8da 100644
--- a/drivers/staging/hv/storvsc.c
+++ b/drivers/staging/hv/storvsc.c
@@ -42,7 +42,7 @@ static inline struct storvsc_device *alloc_stor_device(struct hv_device *device)
/* Set to 2 to allow both inbound and outbound traffics */
/* (ie get_out_stor_device() and get_in_stor_device()) to proceed. */
- atomic_cmpxchg(&stor_device->ref_count, 0, 2);
+ atomic_set(&stor_device->ref_count, 2);
init_waitqueue_head(&stor_device->waiting_to_drain);
stor_device->device = device;
--
1.7.4.1
^ permalink raw reply related
* [PATCH 05/40] Staging: hv: storvsc: Introduce state to manage the lifecycle of stor device
From: K. Y. Srinivasan @ 2011-06-29 14:39 UTC (permalink / raw)
To: gregkh, linux-kernel, devel, virtualization; +Cc: Haiyang Zhang, Abhishek Kane
In-Reply-To: <1309358377-8537-1-git-send-email-kys@microsoft.com>
Introduce state to manage the lifecycle of stor device. This would be the
basis for managing the references on the stor object.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Abhishek Kane <v-abkane@microsoft.com>
Signed-off-by: Hank Janssen <hjanssen@microsoft.com>
---
drivers/staging/hv/hyperv_storage.h | 2 +-
drivers/staging/hv/storvsc.c | 3 ++-
2 files changed, 3 insertions(+), 2 deletions(-)
diff --git a/drivers/staging/hv/hyperv_storage.h b/drivers/staging/hv/hyperv_storage.h
index a224413..d93bf93 100644
--- a/drivers/staging/hv/hyperv_storage.h
+++ b/drivers/staging/hv/hyperv_storage.h
@@ -266,7 +266,7 @@ struct storvsc_device {
/* 0 indicates the device is being destroyed */
atomic_t ref_count;
-
+ bool destroy;
bool drain_notify;
atomic_t num_outstanding_req;
diff --git a/drivers/staging/hv/storvsc.c b/drivers/staging/hv/storvsc.c
index 48bd8da..357b08a 100644
--- a/drivers/staging/hv/storvsc.c
+++ b/drivers/staging/hv/storvsc.c
@@ -43,7 +43,7 @@ static inline struct storvsc_device *alloc_stor_device(struct hv_device *device)
/* Set to 2 to allow both inbound and outbound traffics */
/* (ie get_out_stor_device() and get_in_stor_device()) to proceed. */
atomic_set(&stor_device->ref_count, 2);
-
+ stor_device->destroy = false;
init_waitqueue_head(&stor_device->waiting_to_drain);
stor_device->device = device;
device->ext = stor_device;
@@ -401,6 +401,7 @@ int storvsc_dev_remove(struct hv_device *device)
struct storvsc_device *stor_device;
stor_device = release_stor_device(device);
+ stor_device->destroy = true;
/*
* At this point, all outbound traffic should be disable. We
--
1.7.4.1
^ permalink raw reply related
* [PATCH 06/40] Staging: hv: vmbus: Introduce a lock to protect the ext field in hv_device
From: K. Y. Srinivasan @ 2011-06-29 14:39 UTC (permalink / raw)
To: gregkh, linux-kernel, devel, virtualization; +Cc: Haiyang Zhang, Abhishek Kane
In-Reply-To: <1309358377-8537-1-git-send-email-kys@microsoft.com>
The current mechanism for handling references in broken.
Introduce a lock to protect the ext field in hv_device.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Abhishek Kane <v-abkane@microsoft.com>
Signed-off-by: Hank Janssen <hjanssen@microsoft.com>
---
drivers/staging/hv/hyperv.h | 3 +++
drivers/staging/hv/vmbus_drv.c | 1 +
2 files changed, 4 insertions(+), 0 deletions(-)
diff --git a/drivers/staging/hv/hyperv.h b/drivers/staging/hv/hyperv.h
index 370b096..ff3c69d 100644
--- a/drivers/staging/hv/hyperv.h
+++ b/drivers/staging/hv/hyperv.h
@@ -830,6 +830,9 @@ struct hv_device {
struct vmbus_channel *channel;
+ /* This lock protects the device extension field */
+ spinlock_t ext_lock;
+
/* Device extension; */
void *ext;
};
diff --git a/drivers/staging/hv/vmbus_drv.c b/drivers/staging/hv/vmbus_drv.c
index 2e6bfc1..903362e 100644
--- a/drivers/staging/hv/vmbus_drv.c
+++ b/drivers/staging/hv/vmbus_drv.c
@@ -557,6 +557,7 @@ struct hv_device *vmbus_child_device_create(struct hv_guid *type,
return NULL;
}
+ spin_lock_init(&child_device_obj->ext_lock);
child_device_obj->channel = channel;
/*
* Get the human readable device type name and stash it away.
--
1.7.4.1
^ permalink raw reply related
* [PATCH 07/40] Staging: hv: storvsc: Use the newly introduced lock in accessing ext field
From: K. Y. Srinivasan @ 2011-06-29 14:39 UTC (permalink / raw)
To: gregkh, linux-kernel, devel, virtualization
Cc: K. Y. Srinivasan, Haiyang Zhang, Abhishek Kane, Hank Janssen
In-Reply-To: <1309358377-8537-1-git-send-email-kys@microsoft.com>
Use the newly introduced lock in accessing ext field.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Abhishek Kane <v-abkane@microsoft.com>
Signed-off-by: Hank Janssen <hjanssen@microsoft.com>
---
drivers/staging/hv/hyperv_storage.h | 6 ++++++
drivers/staging/hv/storvsc.c | 6 ++++++
2 files changed, 12 insertions(+), 0 deletions(-)
diff --git a/drivers/staging/hv/hyperv_storage.h b/drivers/staging/hv/hyperv_storage.h
index d93bf93..6b20f1d 100644
--- a/drivers/staging/hv/hyperv_storage.h
+++ b/drivers/staging/hv/hyperv_storage.h
@@ -292,12 +292,15 @@ static inline struct storvsc_device *get_out_stor_device(
struct hv_device *device)
{
struct storvsc_device *stor_device;
+ unsigned long flags;
+ spin_lock_irqsave(&device->ext_lock, flags);
stor_device = (struct storvsc_device *)device->ext;
if (stor_device && atomic_read(&stor_device->ref_count) > 1)
atomic_inc(&stor_device->ref_count);
else
stor_device = NULL;
+ spin_unlock_irqrestore(&device->ext_lock, flags);
return stor_device;
}
@@ -306,10 +309,13 @@ static inline struct storvsc_device *get_out_stor_device(
static inline void put_stor_device(struct hv_device *device)
{
struct storvsc_device *stor_device;
+ unsigned long flags;
+ spin_lock_irqsave(&device->ext_lock, flags);
stor_device = (struct storvsc_device *)device->ext;
atomic_dec(&stor_device->ref_count);
+ spin_unlock_irqrestore(&device->ext_lock, flags);
}
static inline void storvsc_wait_to_drain(struct storvsc_device *dev)
diff --git a/drivers/staging/hv/storvsc.c b/drivers/staging/hv/storvsc.c
index 357b08a..d1b6c4e 100644
--- a/drivers/staging/hv/storvsc.c
+++ b/drivers/staging/hv/storvsc.c
@@ -57,12 +57,15 @@ static inline struct storvsc_device *get_in_stor_device(
struct hv_device *device)
{
struct storvsc_device *stor_device;
+ unsigned long flags;
+ spin_lock_irqsave(&device->ext_lock, flags);
stor_device = (struct storvsc_device *)device->ext;
if (stor_device && atomic_read(&stor_device->ref_count))
atomic_inc(&stor_device->ref_count);
else
stor_device = NULL;
+ spin_unlock_irqrestore(&device->ext_lock, flags);
return stor_device;
}
@@ -87,6 +90,7 @@ static inline struct storvsc_device *final_release_stor_device(
struct hv_device *device)
{
struct storvsc_device *stor_device;
+ unsigned long flags;
stor_device = (struct storvsc_device *)device->ext;
@@ -94,7 +98,9 @@ static inline struct storvsc_device *final_release_stor_device(
while (atomic_cmpxchg(&stor_device->ref_count, 1, 0) != 1)
udelay(100);
+ spin_lock_irqsave(&device->ext_lock, flags);
device->ext = NULL;
+ spin_unlock_irqrestore(&device->ext_lock, flags);
return stor_device;
}
--
1.7.4.1
^ permalink raw reply related
* [PATCH 08/40] Staging: hv: storvsc: Prevent outgoing traffic when stor dev is destroyed
From: K. Y. Srinivasan @ 2011-06-29 14:39 UTC (permalink / raw)
To: gregkh, linux-kernel, devel, virtualization; +Cc: Haiyang Zhang, Abhishek Kane
In-Reply-To: <1309358377-8537-1-git-send-email-kys@microsoft.com>
Prevent outgoing traffic when stor dev is destroyed.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Abhishek Kane <v-abkane@microsoft.com>
Signed-off-by: Hank Janssen <hjanssen@microsoft.com>
---
drivers/staging/hv/hyperv_storage.h | 3 ++-
1 files changed, 2 insertions(+), 1 deletions(-)
diff --git a/drivers/staging/hv/hyperv_storage.h b/drivers/staging/hv/hyperv_storage.h
index 6b20f1d..53b65be 100644
--- a/drivers/staging/hv/hyperv_storage.h
+++ b/drivers/staging/hv/hyperv_storage.h
@@ -296,7 +296,8 @@ static inline struct storvsc_device *get_out_stor_device(
spin_lock_irqsave(&device->ext_lock, flags);
stor_device = (struct storvsc_device *)device->ext;
- if (stor_device && atomic_read(&stor_device->ref_count) > 1)
+ if (stor_device && (atomic_read(&stor_device->ref_count) > 1) &&
+ !stor_device->destroy)
atomic_inc(&stor_device->ref_count);
else
stor_device = NULL;
--
1.7.4.1
^ permalink raw reply related
* [PATCH 09/40] Staging: hv: storvsc: Get rid of release_stor_device() by inlining the code
From: K. Y. Srinivasan @ 2011-06-29 14:39 UTC (permalink / raw)
To: gregkh, linux-kernel, devel, virtualization; +Cc: Haiyang Zhang, Abhishek Kane
In-Reply-To: <1309358377-8537-1-git-send-email-kys@microsoft.com>
Get rid of release_stor_device() by inlining the code.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Abhishek Kane <v-abkane@microsoft.com>
Signed-off-by: Hank Janssen <hjanssen@microsoft.com>
---
drivers/staging/hv/storvsc.c | 23 +++++++----------------
1 files changed, 7 insertions(+), 16 deletions(-)
diff --git a/drivers/staging/hv/storvsc.c b/drivers/staging/hv/storvsc.c
index d1b6c4e..f52e610 100644
--- a/drivers/staging/hv/storvsc.c
+++ b/drivers/staging/hv/storvsc.c
@@ -70,21 +70,6 @@ static inline struct storvsc_device *get_in_stor_device(
return stor_device;
}
-/* Drop ref count to 1 to effectively disable get_out_stor_device() */
-static inline struct storvsc_device *release_stor_device(
- struct hv_device *device)
-{
- struct storvsc_device *stor_device;
-
- stor_device = (struct storvsc_device *)device->ext;
-
- /* Busy wait until the ref drop to 2, then set it to 1 */
- while (atomic_cmpxchg(&stor_device->ref_count, 2, 1) != 2)
- udelay(100);
-
- return stor_device;
-}
-
/* Drop ref count to 0. No one can use stor_device object. */
static inline struct storvsc_device *final_release_stor_device(
struct hv_device *device)
@@ -406,7 +391,13 @@ int storvsc_dev_remove(struct hv_device *device)
{
struct storvsc_device *stor_device;
- stor_device = release_stor_device(device);
+ /*
+ * Since we currently hold a reference on the stor
+ * device, it is safe to dereference the ext
+ * pointer.
+ */
+ stor_device = (struct storvsc_device *)device->ext;
+ atomic_dec(&stor_device->ref_count);
stor_device->destroy = true;
/*
--
1.7.4.1
^ permalink raw reply related
* [PATCH 10/40] Staging: hv: storvsc: Get rid of final_release_stor_device() by inlining code
From: K. Y. Srinivasan @ 2011-06-29 14:39 UTC (permalink / raw)
To: gregkh, linux-kernel, devel, virtualization; +Cc: Haiyang Zhang, Abhishek Kane
In-Reply-To: <1309358377-8537-1-git-send-email-kys@microsoft.com>
Get rid of final_release_stor_device() by inlining code.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Abhishek Kane <v-abkane@microsoft.com>
Signed-off-by: Hank Janssen <hjanssen@microsoft.com>
---
drivers/staging/hv/storvsc.c | 25 +++++--------------------
1 files changed, 5 insertions(+), 20 deletions(-)
diff --git a/drivers/staging/hv/storvsc.c b/drivers/staging/hv/storvsc.c
index f52e610..c4cb170 100644
--- a/drivers/staging/hv/storvsc.c
+++ b/drivers/staging/hv/storvsc.c
@@ -70,25 +70,6 @@ static inline struct storvsc_device *get_in_stor_device(
return stor_device;
}
-/* Drop ref count to 0. No one can use stor_device object. */
-static inline struct storvsc_device *final_release_stor_device(
- struct hv_device *device)
-{
- struct storvsc_device *stor_device;
- unsigned long flags;
-
- stor_device = (struct storvsc_device *)device->ext;
-
- /* Busy wait until the ref drop to 1, then set it to 0 */
- while (atomic_cmpxchg(&stor_device->ref_count, 1, 0) != 1)
- udelay(100);
-
- spin_lock_irqsave(&device->ext_lock, flags);
- device->ext = NULL;
- spin_unlock_irqrestore(&device->ext_lock, flags);
- return stor_device;
-}
-
static int storvsc_channel_init(struct hv_device *device)
{
struct storvsc_device *stor_device;
@@ -390,6 +371,7 @@ int storvsc_dev_add(struct hv_device *device,
int storvsc_dev_remove(struct hv_device *device)
{
struct storvsc_device *stor_device;
+ unsigned long flags;
/*
* Since we currently hold a reference on the stor
@@ -408,7 +390,10 @@ int storvsc_dev_remove(struct hv_device *device)
storvsc_wait_to_drain(stor_device);
- stor_device = final_release_stor_device(device);
+ spin_lock_irqsave(&device->ext_lock, flags);
+ atomic_set(&stor_device->ref_count, 0);
+ device->ext = NULL;
+ spin_unlock_irqrestore(&device->ext_lock, flags);
/* Close the channel */
vmbus_close(device->channel);
--
1.7.4.1
^ permalink raw reply related
* [PATCH 11/40] Staging: hv: storvsc: Leverage the spinlock to manage ref_cnt
From: K. Y. Srinivasan @ 2011-06-29 14:39 UTC (permalink / raw)
To: gregkh, linux-kernel, devel, virtualization; +Cc: Haiyang Zhang, Abhishek Kane
In-Reply-To: <1309358377-8537-1-git-send-email-kys@microsoft.com>
Now that we have a spin lock protecting access to the stor device pointer,
use it manage the reference count as well.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Abhishek Kane <v-abkane@microsoft.com>
Signed-off-by: Hank Janssen <hjanssen@microsoft.com>
---
drivers/staging/hv/hyperv_storage.h | 8 ++++----
drivers/staging/hv/storvsc.c | 10 +++++-----
2 files changed, 9 insertions(+), 9 deletions(-)
diff --git a/drivers/staging/hv/hyperv_storage.h b/drivers/staging/hv/hyperv_storage.h
index 53b65be..d946211 100644
--- a/drivers/staging/hv/hyperv_storage.h
+++ b/drivers/staging/hv/hyperv_storage.h
@@ -265,7 +265,7 @@ struct storvsc_device {
struct hv_device *device;
/* 0 indicates the device is being destroyed */
- atomic_t ref_count;
+ int ref_count;
bool destroy;
bool drain_notify;
atomic_t num_outstanding_req;
@@ -296,9 +296,9 @@ static inline struct storvsc_device *get_out_stor_device(
spin_lock_irqsave(&device->ext_lock, flags);
stor_device = (struct storvsc_device *)device->ext;
- if (stor_device && (atomic_read(&stor_device->ref_count) > 1) &&
+ if (stor_device && (stor_device->ref_count > 1) &&
!stor_device->destroy)
- atomic_inc(&stor_device->ref_count);
+ stor_device->ref_count++;
else
stor_device = NULL;
spin_unlock_irqrestore(&device->ext_lock, flags);
@@ -315,7 +315,7 @@ static inline void put_stor_device(struct hv_device *device)
spin_lock_irqsave(&device->ext_lock, flags);
stor_device = (struct storvsc_device *)device->ext;
- atomic_dec(&stor_device->ref_count);
+ stor_device->ref_count--;
spin_unlock_irqrestore(&device->ext_lock, flags);
}
diff --git a/drivers/staging/hv/storvsc.c b/drivers/staging/hv/storvsc.c
index c4cb170..a41be2a 100644
--- a/drivers/staging/hv/storvsc.c
+++ b/drivers/staging/hv/storvsc.c
@@ -42,7 +42,7 @@ static inline struct storvsc_device *alloc_stor_device(struct hv_device *device)
/* Set to 2 to allow both inbound and outbound traffics */
/* (ie get_out_stor_device() and get_in_stor_device()) to proceed. */
- atomic_set(&stor_device->ref_count, 2);
+ stor_device->ref_count = 2;
stor_device->destroy = false;
init_waitqueue_head(&stor_device->waiting_to_drain);
stor_device->device = device;
@@ -61,8 +61,8 @@ static inline struct storvsc_device *get_in_stor_device(
spin_lock_irqsave(&device->ext_lock, flags);
stor_device = (struct storvsc_device *)device->ext;
- if (stor_device && atomic_read(&stor_device->ref_count))
- atomic_inc(&stor_device->ref_count);
+ if (stor_device && stor_device->ref_count)
+ stor_device->ref_count++;
else
stor_device = NULL;
spin_unlock_irqrestore(&device->ext_lock, flags);
@@ -379,7 +379,7 @@ int storvsc_dev_remove(struct hv_device *device)
* pointer.
*/
stor_device = (struct storvsc_device *)device->ext;
- atomic_dec(&stor_device->ref_count);
+
stor_device->destroy = true;
/*
@@ -391,7 +391,7 @@ int storvsc_dev_remove(struct hv_device *device)
storvsc_wait_to_drain(stor_device);
spin_lock_irqsave(&device->ext_lock, flags);
- atomic_set(&stor_device->ref_count, 0);
+ stor_device->ref_count = 0;
device->ext = NULL;
spin_unlock_irqrestore(&device->ext_lock, flags);
--
1.7.4.1
^ permalink raw reply related
* [PATCH 12/40] Staging: hv: storvsc: Further cleanup reference counting of stor_device
From: K. Y. Srinivasan @ 2011-06-29 14:39 UTC (permalink / raw)
To: gregkh, linux-kernel, devel, virtualization; +Cc: Haiyang Zhang, Abhishek Kane
In-Reply-To: <1309358377-8537-1-git-send-email-kys@microsoft.com>
Further cleanup reference counting of stor_device - when the device is being
destroyed, we will permit incoming traffic only to drain outstanding
requests.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Abhishek Kane <v-abkane@microsoft.com>
Signed-off-by: Hank Janssen <hjanssen@microsoft.com>
---
drivers/staging/hv/hyperv_storage.h | 3 +--
drivers/staging/hv/storvsc.c | 22 +++++++++++++---------
2 files changed, 14 insertions(+), 11 deletions(-)
diff --git a/drivers/staging/hv/hyperv_storage.h b/drivers/staging/hv/hyperv_storage.h
index d946211..a1f3e27 100644
--- a/drivers/staging/hv/hyperv_storage.h
+++ b/drivers/staging/hv/hyperv_storage.h
@@ -287,7 +287,6 @@ struct storvsc_device {
};
-/* Get the stordevice object iff exists and its refcount > 1 */
static inline struct storvsc_device *get_out_stor_device(
struct hv_device *device)
{
@@ -296,7 +295,7 @@ static inline struct storvsc_device *get_out_stor_device(
spin_lock_irqsave(&device->ext_lock, flags);
stor_device = (struct storvsc_device *)device->ext;
- if (stor_device && (stor_device->ref_count > 1) &&
+ if (stor_device && (stor_device->ref_count) &&
!stor_device->destroy)
stor_device->ref_count++;
else
diff --git a/drivers/staging/hv/storvsc.c b/drivers/staging/hv/storvsc.c
index a41be2a..4d13044 100644
--- a/drivers/staging/hv/storvsc.c
+++ b/drivers/staging/hv/storvsc.c
@@ -40,9 +40,7 @@ static inline struct storvsc_device *alloc_stor_device(struct hv_device *device)
if (!stor_device)
return NULL;
- /* Set to 2 to allow both inbound and outbound traffics */
- /* (ie get_out_stor_device() and get_in_stor_device()) to proceed. */
- stor_device->ref_count = 2;
+ stor_device->ref_count = 1;
stor_device->destroy = false;
init_waitqueue_head(&stor_device->waiting_to_drain);
stor_device->device = device;
@@ -51,8 +49,6 @@ static inline struct storvsc_device *alloc_stor_device(struct hv_device *device)
return stor_device;
}
-
-/* Get the stordevice object iff exists and its refcount > 0 */
static inline struct storvsc_device *get_in_stor_device(
struct hv_device *device)
{
@@ -61,10 +57,18 @@ static inline struct storvsc_device *get_in_stor_device(
spin_lock_irqsave(&device->ext_lock, flags);
stor_device = (struct storvsc_device *)device->ext;
- if (stor_device && stor_device->ref_count)
- stor_device->ref_count++;
- else
- stor_device = NULL;
+ if (!stor_device)
+ goto cleanup;
+
+ /*
+ * If the device is being destroyed; allow incoming
+ * traffic only to cleanup outstanding requests.
+ */
+ if (stor_device->destroy &&
+ (atomic_read(&stor_device->num_outstanding_req) == 0))
+ goto cleanup;
+ stor_device->ref_count++;
+cleanup:
spin_unlock_irqrestore(&device->ext_lock, flags);
return stor_device;
--
1.7.4.1
^ permalink raw reply related
* [PATCH 13/40] Staging: hv: netvsc: Introduce code to properly manage outstanding sends
From: K. Y. Srinivasan @ 2011-06-29 14:39 UTC (permalink / raw)
To: gregkh, linux-kernel, devel, virtualization; +Cc: Haiyang Zhang, Abhishek Kane
In-Reply-To: <1309358377-8537-1-git-send-email-kys@microsoft.com>
Introduce code to properly manage outstanding sends.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Abhishek Kane <v-abkane@microsoft.com>
Signed-off-by: Hank Janssen <hjanssen@microsoft.com>
---
drivers/staging/hv/hyperv_net.h | 10 ++++++++++
drivers/staging/hv/netvsc.c | 6 +++++-
2 files changed, 15 insertions(+), 1 deletions(-)
diff --git a/drivers/staging/hv/hyperv_net.h b/drivers/staging/hv/hyperv_net.h
index 5782fea..b5bee9e 100644
--- a/drivers/staging/hv/hyperv_net.h
+++ b/drivers/staging/hv/hyperv_net.h
@@ -371,6 +371,8 @@ struct netvsc_device {
atomic_t refcnt;
atomic_t num_outstanding_sends;
+ bool drain_notify;
+ wait_queue_head_t waiting_to_drain;
/*
* List of free preallocated hv_netvsc_packet to represent receive
* packet
@@ -1051,6 +1053,14 @@ struct rndis_filter_packet {
#define NDIS_PACKET_TYPE_FUNCTIONAL 0x00000400
#define NDIS_PACKET_TYPE_MAC_FRAME 0x00000800
+static inline void netvsc_wait_to_drain(struct netvsc_device *dev)
+{
+ dev->drain_notify = true;
+ wait_event(dev->waiting_to_drain,
+ atomic_read(&dev->num_outstanding_sends) == 0);
+ dev->drain_notify = false;
+}
+
#endif /* _HYPERV_NET_H */
diff --git a/drivers/staging/hv/netvsc.c b/drivers/staging/hv/netvsc.c
index 551537a..bdd5c2b 100644
--- a/drivers/staging/hv/netvsc.c
+++ b/drivers/staging/hv/netvsc.c
@@ -42,6 +42,8 @@ static struct netvsc_device *alloc_net_device(struct hv_device *device)
/* Set to 2 to allow both inbound and outbound traffic */
atomic_cmpxchg(&net_device->refcnt, 0, 2);
+ net_device->drain_notify = false;
+ init_waitqueue_head(&net_device->waiting_to_drain);
net_device->dev = device;
device->ext = net_device;
@@ -483,7 +485,9 @@ static void netvsc_send_completion(struct hv_device *device,
nvsc_packet->completion.send.send_completion(
nvsc_packet->completion.send.send_completion_ctx);
- atomic_dec(&net_device->num_outstanding_sends);
+ if (atomic_dec_and_test(&net_device->num_outstanding_sends) &&
+ net_device->drain_notify)
+ wake_up(&net_device->waiting_to_drain);
} else {
netdev_err(ndev, "Unknown send completion packet type- "
"%d received!!\n", nvsp_packet->hdr.msg_type);
--
1.7.4.1
^ permalink raw reply related
* [PATCH 14/40] Staging: hv: netvsc: Inline the code for free_net_device()
From: K. Y. Srinivasan @ 2011-06-29 14:39 UTC (permalink / raw)
To: gregkh, linux-kernel, devel, virtualization; +Cc: Haiyang Zhang, Abhishek Kane
In-Reply-To: <1309358377-8537-1-git-send-email-kys@microsoft.com>
Inline the code for free_net_device().
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Abhishek Kane <v-abkane@microsoft.com>
Signed-off-by: Hank Janssen <hjanssen@microsoft.com>
---
drivers/staging/hv/netvsc.c | 12 ++----------
1 files changed, 2 insertions(+), 10 deletions(-)
diff --git a/drivers/staging/hv/netvsc.c b/drivers/staging/hv/netvsc.c
index bdd5c2b..c3c934c 100644
--- a/drivers/staging/hv/netvsc.c
+++ b/drivers/staging/hv/netvsc.c
@@ -51,14 +51,6 @@ static struct netvsc_device *alloc_net_device(struct hv_device *device)
return net_device;
}
-static void free_net_device(struct netvsc_device *device)
-{
- WARN_ON(atomic_read(&device->refcnt) != 0);
- device->dev->ext = NULL;
- kfree(device);
-}
-
-
/* Get the net device object iff exists and its refcount > 1 */
static struct netvsc_device *get_outbound_net_device(struct hv_device *device)
{
@@ -444,7 +436,7 @@ int netvsc_device_remove(struct hv_device *device)
kfree(netvsc_packet);
}
- free_net_device(net_device);
+ kfree(net_device);
return 0;
}
@@ -994,7 +986,7 @@ cleanup:
release_outbound_net_device(device);
release_inbound_net_device(device);
- free_net_device(net_device);
+ kfree(net_device);
}
return ret;
--
1.7.4.1
^ permalink raw reply related
* [PATCH 15/40] Staging: hv: netvsc: Cleanup alloc_net_device()
From: K. Y. Srinivasan @ 2011-06-29 14:39 UTC (permalink / raw)
To: gregkh, linux-kernel, devel, virtualization; +Cc: Haiyang Zhang, Abhishek Kane
In-Reply-To: <1309358377-8537-1-git-send-email-kys@microsoft.com>
Cleanup alloc_net_device(); we can directly set the refcnt.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Abhishek Kane <v-abkane@microsoft.com>
Signed-off-by: Hank Janssen <hjanssen@microsoft.com>
---
drivers/staging/hv/netvsc.c | 2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/drivers/staging/hv/netvsc.c b/drivers/staging/hv/netvsc.c
index c3c934c..56749ca 100644
--- a/drivers/staging/hv/netvsc.c
+++ b/drivers/staging/hv/netvsc.c
@@ -41,7 +41,7 @@ static struct netvsc_device *alloc_net_device(struct hv_device *device)
return NULL;
/* Set to 2 to allow both inbound and outbound traffic */
- atomic_cmpxchg(&net_device->refcnt, 0, 2);
+ atomic_set(&net_device->refcnt, 2);
net_device->drain_notify = false;
init_waitqueue_head(&net_device->waiting_to_drain);
--
1.7.4.1
^ permalink raw reply related
* [PATCH 16/40] Staging: hv: netvsc: Introduce state to manage the lifecycle of net device
From: K. Y. Srinivasan @ 2011-06-29 14:39 UTC (permalink / raw)
To: gregkh, linux-kernel, devel, virtualization; +Cc: Haiyang Zhang, Abhishek Kane
In-Reply-To: <1309358377-8537-1-git-send-email-kys@microsoft.com>
In preparation for cleaning up reference counts, introduce state to manage
the lifecycle of net device.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Abhishek Kane <v-abkane@microsoft.com>
Signed-off-by: Hank Janssen <hjanssen@microsoft.com>
---
drivers/staging/hv/hyperv_net.h | 1 +
drivers/staging/hv/netvsc.c | 2 ++
2 files changed, 3 insertions(+), 0 deletions(-)
diff --git a/drivers/staging/hv/hyperv_net.h b/drivers/staging/hv/hyperv_net.h
index b5bee9e..c6836be 100644
--- a/drivers/staging/hv/hyperv_net.h
+++ b/drivers/staging/hv/hyperv_net.h
@@ -371,6 +371,7 @@ struct netvsc_device {
atomic_t refcnt;
atomic_t num_outstanding_sends;
+ bool destroy;
bool drain_notify;
wait_queue_head_t waiting_to_drain;
/*
diff --git a/drivers/staging/hv/netvsc.c b/drivers/staging/hv/netvsc.c
index 56749ca..f03018c 100644
--- a/drivers/staging/hv/netvsc.c
+++ b/drivers/staging/hv/netvsc.c
@@ -43,6 +43,7 @@ static struct netvsc_device *alloc_net_device(struct hv_device *device)
/* Set to 2 to allow both inbound and outbound traffic */
atomic_set(&net_device->refcnt, 2);
net_device->drain_notify = false;
+ net_device->destroy = false;
init_waitqueue_head(&net_device->waiting_to_drain);
net_device->dev = device;
@@ -409,6 +410,7 @@ int netvsc_device_remove(struct hv_device *device)
netdev_err(ndev, "No net device present!!\n");
return -ENODEV;
}
+ net_device->destroy = true;
/* Wait for all send completions */
while (atomic_read(&net_device->num_outstanding_sends)) {
--
1.7.4.1
^ permalink raw reply related
* [PATCH 17/40] Staging: hv: netvsc: Use the newly introduced lock in accessing ext field
From: K. Y. Srinivasan @ 2011-06-29 14:39 UTC (permalink / raw)
To: gregkh, linux-kernel, devel, virtualization; +Cc: Haiyang Zhang, Abhishek Kane
In-Reply-To: <1309358377-8537-1-git-send-email-kys@microsoft.com>
The current reference counting mechanism is broken; use the newly introduced
lock to access the net_device pointer in struct hv_device.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Abhishek Kane <v-abkane@microsoft.com>
Signed-off-by: Hank Janssen <hjanssen@microsoft.com>
---
drivers/staging/hv/netvsc.c | 24 ++++++++++++++++++++++--
1 files changed, 22 insertions(+), 2 deletions(-)
diff --git a/drivers/staging/hv/netvsc.c b/drivers/staging/hv/netvsc.c
index f03018c..531de63 100644
--- a/drivers/staging/hv/netvsc.c
+++ b/drivers/staging/hv/netvsc.c
@@ -56,12 +56,15 @@ static struct netvsc_device *alloc_net_device(struct hv_device *device)
static struct netvsc_device *get_outbound_net_device(struct hv_device *device)
{
struct netvsc_device *net_device;
+ unsigned long flags;
+ spin_lock_irqsave(&device->ext_lock, flags);
net_device = device->ext;
if (net_device && atomic_read(&net_device->refcnt) > 1)
atomic_inc(&net_device->refcnt);
else
net_device = NULL;
+ spin_unlock_irqrestore(&device->ext_lock, flags);
return net_device;
}
@@ -70,38 +73,50 @@ static struct netvsc_device *get_outbound_net_device(struct hv_device *device)
static struct netvsc_device *get_inbound_net_device(struct hv_device *device)
{
struct netvsc_device *net_device;
+ unsigned long flags;
+ spin_lock_irqsave(&device->ext_lock, flags);
net_device = device->ext;
if (net_device && atomic_read(&net_device->refcnt))
atomic_inc(&net_device->refcnt);
else
net_device = NULL;
+ spin_unlock_irqrestore(&device->ext_lock, flags);
return net_device;
}
static void put_net_device(struct hv_device *device)
{
struct netvsc_device *net_device;
+ unsigned long flags;
+
+ spin_lock_irqsave(&device->ext_lock, flags);
net_device = device->ext;
atomic_dec(&net_device->refcnt);
+ spin_unlock_irqrestore(&device->ext_lock, flags);
}
static struct netvsc_device *release_outbound_net_device(
struct hv_device *device)
{
struct netvsc_device *net_device;
+ unsigned long flags;
+ spin_lock_irqsave(&device->ext_lock, flags);
net_device = device->ext;
- if (net_device == NULL)
+ if (net_device == NULL) {
+ spin_unlock_irqrestore(&device->ext_lock, flags);
return NULL;
+ }
/* Busy wait until the ref drop to 2, then set it to 1 */
while (atomic_cmpxchg(&net_device->refcnt, 2, 1) != 2)
udelay(100);
+ spin_unlock_irqrestore(&device->ext_lock, flags);
return net_device;
}
@@ -109,16 +124,21 @@ static struct netvsc_device *release_inbound_net_device(
struct hv_device *device)
{
struct netvsc_device *net_device;
+ unsigned long flags;
+ spin_lock_irqsave(&device->ext_lock, flags);
net_device = device->ext;
- if (net_device == NULL)
+ if (net_device == NULL) {
+ spin_unlock_irqrestore(&device->ext_lock, flags);
return NULL;
+ }
/* Busy wait until the ref drop to 1, then set it to 0 */
while (atomic_cmpxchg(&net_device->refcnt, 1, 0) != 1)
udelay(100);
device->ext = NULL;
+ spin_unlock_irqrestore(&device->ext_lock, flags);
return net_device;
}
--
1.7.4.1
^ permalink raw reply related
* [PATCH 18/40] Staging: hv: netvsc: Prevent outgoing traffic when netvsc dev is destroyed
From: K. Y. Srinivasan @ 2011-06-29 14:39 UTC (permalink / raw)
To: gregkh, linux-kernel, devel, virtualization; +Cc: Haiyang Zhang, Abhishek Kane
In-Reply-To: <1309358377-8537-1-git-send-email-kys@microsoft.com>
Prevent outgoing traffic when netvsc dev is destroyed.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Abhishek Kane <v-abkane@microsoft.com>
Signed-off-by: Hank Janssen <hjanssen@microsoft.com>
---
drivers/staging/hv/netvsc.c | 3 ++-
1 files changed, 2 insertions(+), 1 deletions(-)
diff --git a/drivers/staging/hv/netvsc.c b/drivers/staging/hv/netvsc.c
index 531de63..3ac0e17 100644
--- a/drivers/staging/hv/netvsc.c
+++ b/drivers/staging/hv/netvsc.c
@@ -60,7 +60,8 @@ static struct netvsc_device *get_outbound_net_device(struct hv_device *device)
spin_lock_irqsave(&device->ext_lock, flags);
net_device = device->ext;
- if (net_device && atomic_read(&net_device->refcnt) > 1)
+ if (net_device && (atomic_read(&net_device->refcnt) > 1) &&
+ !net_device->destroy)
atomic_inc(&net_device->refcnt);
else
net_device = NULL;
--
1.7.4.1
^ permalink raw reply related
page: next (older) | prev (newer) | latest
- recent:[subjects (threaded)|topics (new)|topics (active)]
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox