[PATCH V3 0/6] IRQ offloading for vDPA

[PATCH V3 0/6] IRQ offloading for vDPA

[Zhu Lingshan]

This series intends to implement IRQ offloading for
vhost_vdpa.

By the feat of irq forwarding facilities like posted
interrupt on X86, irq bypass can  help deliver
interrupts to vCPU directly.

vDPA devices have dedicated hardware backends like VFIO
pass-throughed devices. So it would be possible to setup
irq offloading(irq bypass) for vDPA devices and gain
performance improvements.

In my testing, with this feature, we can save 0.1ms
in a ping between two VFs on average.

changes from V2:
(1)rename struct vhost_call_ctx to vhost_vring_call
(2)add kvm_arch_end_assignment() in del_producer()
code path
(3)rename vDPA helpers to vdpa_devm_request_irq()
and vdpa_devm_free_irq(). Add better comments
for them.
(4)better comments for setup_vq_irq() and
unsetup_vq_irq()
(5)In vDPA VHOST_SET_VRING_CALL, will call vhost_vdpa_update_vq_irq()
without checking producer.irq, move this check into
vhost_vdpa_update_vq_irq(), so that get advantage of the spinlock.
(6)Add a function vhost_vdpa_clean_irq(), this function will unregister
the producer of vqs when vhost_vdpa_release(). This is safe
for control vq.
(7) minor improvements

changes from V1:
(1)dropped vfio changes.
(3)removed KVM_HVAE_IRQ_BYPASS checks
(4)locking fixes
(5)simplified vhost_vdpa_updat

Zhu Lingshan (6):
  vhost: introduce vhost_vring_call
  kvm: detect assigned device via irqbypass manager
  vDPA: implement vq IRQ allocate/free helpers in vDPA core
  vhost_vdpa: implement IRQ offloading in vhost_vdpa
  ifcvf: replace irq_request/free with vDPA helpers
  irqbypass: do not start cons/prod when failed connect

 arch/x86/kvm/x86.c              | 11 +++++-
 drivers/vdpa/ifcvf/ifcvf_main.c | 14 ++++---
 drivers/vdpa/vdpa.c             | 49 +++++++++++++++++++++++
 drivers/vhost/Kconfig           |  1 +
 drivers/vhost/vdpa.c            | 70 +++++++++++++++++++++++++++++++--
 drivers/vhost/vhost.c           | 22 ++++++++---
 drivers/vhost/vhost.h           |  9 ++++-
 include/linux/vdpa.h            | 13 ++++++
 virt/lib/irqbypass.c            | 16 +++++---
 9 files changed, 182 insertions(+), 23 deletions(-)

-- 
2.18.4

[PATCH V3 1/6] vhost: introduce vhost_vring_call

[Zhu Lingshan]

This commit introduces struct vhost_vring_call which replaced
raw struct eventfd_ctx *call_ctx in struct vhost_virtqueue.
Besides eventfd_ctx, it contains a spin lock and an
irq_bypass_producer in its structure.

Signed-off-by: Zhu Lingshan <lingshan.zhu@intel.com>
---
 drivers/vhost/vdpa.c  |  4 ++--
 drivers/vhost/vhost.c | 22 ++++++++++++++++------
 drivers/vhost/vhost.h |  9 ++++++++-
 3 files changed, 26 insertions(+), 9 deletions(-)

diff --git a/drivers/vhost/vdpa.c b/drivers/vhost/vdpa.c
index a54b60d6623f..df3cf386b0cd 100644
--- a/drivers/vhost/vdpa.c
+++ b/drivers/vhost/vdpa.c
@@ -96,7 +96,7 @@ static void handle_vq_kick(struct vhost_work *work)
 static irqreturn_t vhost_vdpa_virtqueue_cb(void *private)
 {
 	struct vhost_virtqueue *vq = private;
-	struct eventfd_ctx *call_ctx = vq->call_ctx;
+	struct eventfd_ctx *call_ctx = vq->call_ctx.ctx;
 
 	if (call_ctx)
 		eventfd_signal(call_ctx, 1);
@@ -382,7 +382,7 @@ static long vhost_vdpa_vring_ioctl(struct vhost_vdpa *v, unsigned int cmd,
 		break;
 
 	case VHOST_SET_VRING_CALL:
-		if (vq->call_ctx) {
+		if (vq->call_ctx.ctx) {
 			cb.callback = vhost_vdpa_virtqueue_cb;
 			cb.private = vq;
 		} else {
diff --git a/drivers/vhost/vhost.c b/drivers/vhost/vhost.c
index d7b8df3edffc..9f1a845a9302 100644
--- a/drivers/vhost/vhost.c
+++ b/drivers/vhost/vhost.c
@@ -298,6 +298,13 @@ static void vhost_vq_meta_reset(struct vhost_dev *d)
 		__vhost_vq_meta_reset(d->vqs[i]);
 }
 
+static void vhost_vring_call_reset(struct vhost_vring_call *call_ctx)
+{
+	call_ctx->ctx = NULL;
+	memset(&call_ctx->producer, 0x0, sizeof(struct irq_bypass_producer));
+	spin_lock_init(&call_ctx->ctx_lock);
+}
+
 static void vhost_vq_reset(struct vhost_dev *dev,
 			   struct vhost_virtqueue *vq)
 {
@@ -319,13 +326,13 @@ static void vhost_vq_reset(struct vhost_dev *dev,
 	vq->log_base = NULL;
 	vq->error_ctx = NULL;
 	vq->kick = NULL;
-	vq->call_ctx = NULL;
 	vq->log_ctx = NULL;
 	vhost_reset_is_le(vq);
 	vhost_disable_cross_endian(vq);
 	vq->busyloop_timeout = 0;
 	vq->umem = NULL;
 	vq->iotlb = NULL;
+	vhost_vring_call_reset(&vq->call_ctx);
 	__vhost_vq_meta_reset(vq);
 }
 
@@ -685,8 +692,8 @@ void vhost_dev_cleanup(struct vhost_dev *dev)
 			eventfd_ctx_put(dev->vqs[i]->error_ctx);
 		if (dev->vqs[i]->kick)
 			fput(dev->vqs[i]->kick);
-		if (dev->vqs[i]->call_ctx)
-			eventfd_ctx_put(dev->vqs[i]->call_ctx);
+		if (dev->vqs[i]->call_ctx.ctx)
+			eventfd_ctx_put(dev->vqs[i]->call_ctx.ctx);
 		vhost_vq_reset(dev, dev->vqs[i]);
 	}
 	vhost_dev_free_iovecs(dev);
@@ -1629,7 +1636,10 @@ long vhost_vring_ioctl(struct vhost_dev *d, unsigned int ioctl, void __user *arg
 			r = PTR_ERR(ctx);
 			break;
 		}
-		swap(ctx, vq->call_ctx);
+
+		spin_lock(&vq->call_ctx.ctx_lock);
+		swap(ctx, vq->call_ctx.ctx);
+		spin_unlock(&vq->call_ctx.ctx_lock);
 		break;
 	case VHOST_SET_VRING_ERR:
 		if (copy_from_user(&f, argp, sizeof f)) {
@@ -2440,8 +2450,8 @@ static bool vhost_notify(struct vhost_dev *dev, struct vhost_virtqueue *vq)
 void vhost_signal(struct vhost_dev *dev, struct vhost_virtqueue *vq)
 {
 	/* Signal the Guest tell them we used something up. */
-	if (vq->call_ctx && vhost_notify(dev, vq))
-		eventfd_signal(vq->call_ctx, 1);
+	if (vq->call_ctx.ctx && vhost_notify(dev, vq))
+		eventfd_signal(vq->call_ctx.ctx, 1);
 }
 EXPORT_SYMBOL_GPL(vhost_signal);
 
diff --git a/drivers/vhost/vhost.h b/drivers/vhost/vhost.h
index c8e96a095d3b..38eb1aa3b68d 100644
--- a/drivers/vhost/vhost.h
+++ b/drivers/vhost/vhost.h
@@ -13,6 +13,7 @@
 #include <linux/virtio_ring.h>
 #include <linux/atomic.h>
 #include <linux/vhost_iotlb.h>
+#include <linux/irqbypass.h>
 
 struct vhost_work;
 typedef void (*vhost_work_fn_t)(struct vhost_work *work);
@@ -60,6 +61,12 @@ enum vhost_uaddr_type {
 	VHOST_NUM_ADDRS = 3,
 };
 
+struct vhost_vring_call {
+	struct eventfd_ctx *ctx;
+	struct irq_bypass_producer producer;
+	spinlock_t ctx_lock;
+};
+
 /* The virtqueue structure describes a queue attached to a device. */
 struct vhost_virtqueue {
 	struct vhost_dev *dev;
@@ -72,7 +79,7 @@ struct vhost_virtqueue {
 	vring_used_t __user *used;
 	const struct vhost_iotlb_map *meta_iotlb[VHOST_NUM_ADDRS];
 	struct file *kick;
-	struct eventfd_ctx *call_ctx;
+	struct vhost_vring_call call_ctx;
 	struct eventfd_ctx *error_ctx;
 	struct eventfd_ctx *log_ctx;
 
-- 
2.18.4

[PATCH V3 2/6] kvm: detect assigned device via irqbypass manager

[Zhu Lingshan]

vDPA devices has dedicated backed hardware like
passthrough-ed devices. Then it is possible to setup irq
offloading to vCPU for vDPA devices. Thus this patch tries to
manipulated assigned device counters by
kvm_arch_start/end_assignment() in irqbypass manager, so that
assigned devices could be detected in update_pi_irte()

We will increase/decrease the assigned device counter in kvm/x86.
Both vDPA and VFIO would go through this code path.

Only X86 uses these counters and kvm_arch_start/end_assignment(),
so this code path only affect x86 for now.

Signed-off-by: Zhu Lingshan <lingshan.zhu@intel.com>
---
 arch/x86/kvm/x86.c | 11 +++++++++--
 1 file changed, 9 insertions(+), 2 deletions(-)

diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 88c593f83b28..b2bf17f808b6 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -10630,11 +10630,17 @@ int kvm_arch_irq_bypass_add_producer(struct irq_bypass_consumer *cons,
 {
 	struct kvm_kernel_irqfd *irqfd =
 		container_of(cons, struct kvm_kernel_irqfd, consumer);
+	int ret;
 
 	irqfd->producer = prod;
+	kvm_arch_start_assignment(irqfd->kvm);
+	ret = kvm_x86_ops.update_pi_irte(irqfd->kvm,
+					 prod->irq, irqfd->gsi, 1);
+
+	if (ret)
+		kvm_arch_end_assignment(irqfd->kvm);
 
-	return kvm_x86_ops.update_pi_irte(irqfd->kvm,
-					   prod->irq, irqfd->gsi, 1);
+	return ret;
 }
 
 void kvm_arch_irq_bypass_del_producer(struct irq_bypass_consumer *cons,
@@ -10645,6 +10651,7 @@ void kvm_arch_irq_bypass_del_producer(struct irq_bypass_consumer *cons,
 		container_of(cons, struct kvm_kernel_irqfd, consumer);
 
 	WARN_ON(irqfd->producer != prod);
+	kvm_arch_end_assignment(irqfd->kvm);
 	irqfd->producer = NULL;
 
 	/*
-- 
2.18.4

[PATCH V3 3/6] vDPA: implement vq IRQ allocate/free helpers in vDPA core

[Zhu Lingshan]

This commit implements vq IRQ allocate/free helpers in vDPA
core by introducing two couple of functions:

vdpa_devm_request_Irq() and vdpa_devm_free_irq(): request irq and free
irq, and setup irq offloading.

vdpa_setup_irq() and vdpa_unsetup_irq(): supportive functions,
will call vhost_vdpa helpers.

Signed-off-by: Zhu Lingshan <lingshan.zhu@intel.com>
---
 drivers/vdpa/vdpa.c  | 49 ++++++++++++++++++++++++++++++++++++++++++++
 include/linux/vdpa.h | 13 ++++++++++++
 2 files changed, 62 insertions(+)

diff --git a/drivers/vdpa/vdpa.c b/drivers/vdpa/vdpa.c
index de211ef3738c..4dfc86eba0f6 100644
--- a/drivers/vdpa/vdpa.c
+++ b/drivers/vdpa/vdpa.c
@@ -163,6 +163,55 @@ void vdpa_unregister_driver(struct vdpa_driver *drv)
 }
 EXPORT_SYMBOL_GPL(vdpa_unregister_driver);
 
+static void vdpa_setup_irq(struct vdpa_device *vdev, int qid, int irq)
+{
+	struct vdpa_driver *drv = drv_to_vdpa(vdev->dev.driver);
+
+	if (drv->setup_vq_irq)
+		drv->setup_vq_irq(vdev, qid, irq);
+}
+
+static void vdpa_unsetup_irq(struct vdpa_device *vdev, int qid)
+{
+	struct vdpa_driver *drv = drv_to_vdpa(vdev->dev.driver);
+
+	if (drv && drv->unsetup_vq_irq)
+		drv->unsetup_vq_irq(vdev, qid);
+}
+
+/*
+ * Request irq for a vq, setup irq offloading if its a vhost_vdpa vq.
+ * This function should be only called through setting virtio DRIVER_OK.
+ * If you want to request irq during probe, you should use raw APIs
+ * like request_irq() or devm_request_irq().
+ */
+int vdpa_devm_request_irq(struct device *dev, struct vdpa_device *vdev,
+			  unsigned int irq, irq_handler_t handler,
+			  unsigned long irqflags, const char *devname, void *dev_id,
+			  int qid)
+{
+	int ret;
+
+	ret = devm_request_irq(dev, irq, handler, irqflags, devname, dev_id);
+	if (ret)
+		dev_err(dev, "Failed to request irq for vq %d\n", qid);
+	else
+		vdpa_setup_irq(vdev, qid, irq);
+
+	return ret;
+
+}
+EXPORT_SYMBOL_GPL(vdpa_devm_request_irq);
+
+/* free irq for a vq, unsetup irq offloading if its a vhost_vdpa vq */
+void vdpa_devm_free_irq(struct device *dev, struct vdpa_device *vdev, int irq,
+			int qid, void *dev_id)
+{
+	devm_free_irq(dev, irq, dev_id);
+	vdpa_unsetup_irq(vdev, qid);
+}
+EXPORT_SYMBOL_GPL(vdpa_devm_free_irq);
+
 static int vdpa_init(void)
 {
 	return bus_register(&vdpa_bus);
diff --git a/include/linux/vdpa.h b/include/linux/vdpa.h
index 239db794357c..9412e3b56589 100644
--- a/include/linux/vdpa.h
+++ b/include/linux/vdpa.h
@@ -220,17 +220,30 @@ struct vdpa_device *__vdpa_alloc_device(struct device *parent,
 
 int vdpa_register_device(struct vdpa_device *vdev);
 void vdpa_unregister_device(struct vdpa_device *vdev);
+int vdpa_devm_request_irq(struct device *dev, struct vdpa_device *vdev,
+			  unsigned int irq, irq_handler_t handler,
+			  unsigned long irqflags, const char *devname, void *dev_id,
+			  int qid);
+void vdpa_devm_free_irq(struct device *dev, struct vdpa_device *vdev, int irq,
+			int qid, void *dev_id);
+
 
 /**
  * vdpa_driver - operations for a vDPA driver
  * @driver: underlying device driver
  * @probe: the function to call when a device is found.  Returns 0 or -errno.
  * @remove: the function to call when a device is removed.
+ * @setup_vq_irq: the function to call after request irq for a vhost_vdpa vq,
+ * do setup works e.g irq offloading.
+ * @unsetup_vq_irq: the function to call after free irq for a vhost_vdpa vq,
+ * do unsetup works e.g relieve irq offloading.
  */
 struct vdpa_driver {
 	struct device_driver driver;
 	int (*probe)(struct vdpa_device *vdev);
 	void (*remove)(struct vdpa_device *vdev);
+	void (*setup_vq_irq)(struct vdpa_device *vdev, int qid, int irq);
+	void (*unsetup_vq_irq)(struct vdpa_device *vdev, int qid);
 };
 
 #define vdpa_register_driver(drv) \
-- 
2.18.4

[PATCH V3 4/6] vhost_vdpa: implement IRQ offloading in vhost_vdpa

[Zhu Lingshan]

This patch introduce a set of functions for setup/unsetup
and update irq offloading respectively by register/unregister
and re-register the irq_bypass_producer.

Signed-off-by: Zhu Lingshan <lingshan.zhu@intel.com>
---
 drivers/vhost/Kconfig |  1 +
 drivers/vhost/vdpa.c  | 66 ++++++++++++++++++++++++++++++++++++++++++-
 2 files changed, 66 insertions(+), 1 deletion(-)

diff --git a/drivers/vhost/Kconfig b/drivers/vhost/Kconfig
index d3688c6afb87..587fbae06182 100644
--- a/drivers/vhost/Kconfig
+++ b/drivers/vhost/Kconfig
@@ -65,6 +65,7 @@ config VHOST_VDPA
 	tristate "Vhost driver for vDPA-based backend"
 	depends on EVENTFD
 	select VHOST
+	select IRQ_BYPASS_MANAGER
 	depends on VDPA
 	help
 	  This kernel module can be loaded in host kernel to accelerate
diff --git a/drivers/vhost/vdpa.c b/drivers/vhost/vdpa.c
index df3cf386b0cd..558cf0863126 100644
--- a/drivers/vhost/vdpa.c
+++ b/drivers/vhost/vdpa.c
@@ -115,6 +115,54 @@ static irqreturn_t vhost_vdpa_config_cb(void *private)
 	return IRQ_HANDLED;
 }
 
+static void vhost_vdpa_setup_vq_irq(struct vdpa_device *dev, int qid, int irq)
+{
+	struct vhost_vdpa *v = vdpa_get_drvdata(dev);
+	struct vhost_virtqueue *vq = &v->vqs[qid];
+	int ret;
+
+	spin_lock(&vq->call_ctx.ctx_lock);
+	if (!vq->call_ctx.ctx) {
+		spin_unlock(&vq->call_ctx.ctx_lock);
+		return;
+	}
+
+	vq->call_ctx.producer.token = vq->call_ctx.ctx;
+	vq->call_ctx.producer.irq = irq;
+	ret = irq_bypass_register_producer(&vq->call_ctx.producer);
+	spin_unlock(&vq->call_ctx.ctx_lock);
+}
+
+static void vhost_vdpa_unsetup_vq_irq(struct vdpa_device *dev, int qid)
+{
+	struct vhost_vdpa *v = vdpa_get_drvdata(dev);
+	struct vhost_virtqueue *vq = &v->vqs[qid];
+
+	spin_lock(&vq->call_ctx.ctx_lock);
+	irq_bypass_unregister_producer(&vq->call_ctx.producer);
+	spin_unlock(&vq->call_ctx.ctx_lock);
+}
+
+static void vhost_vdpa_update_vq_irq(struct vhost_virtqueue *vq)
+{
+	spin_lock(&vq->call_ctx.ctx_lock);
+	/*
+	 * if it has a non-zero irq, means there is a
+	 * previsouly registered irq_bypass_producer,
+	 * we should update it when ctx (its token)
+	 * changes.
+	 */
+	if (!vq->call_ctx.producer.irq) {
+		spin_unlock(&vq->call_ctx.ctx_lock);
+		return;
+	}
+
+	irq_bypass_unregister_producer(&vq->call_ctx.producer);
+	vq->call_ctx.producer.token = vq->call_ctx.ctx;
+	irq_bypass_register_producer(&vq->call_ctx.producer);
+	spin_unlock(&vq->call_ctx.ctx_lock);
+}
+
 static void vhost_vdpa_reset(struct vhost_vdpa *v)
 {
 	struct vdpa_device *vdpa = v->vdpa;
@@ -122,7 +170,6 @@ static void vhost_vdpa_reset(struct vhost_vdpa *v)
 
 	ops->set_status(vdpa, 0);
 }
-
 static long vhost_vdpa_get_device_id(struct vhost_vdpa *v, u8 __user *argp)
 {
 	struct vdpa_device *vdpa = v->vdpa;
@@ -332,6 +379,7 @@ static long vhost_vdpa_set_config_call(struct vhost_vdpa *v, u32 __user *argp)
 
 	return 0;
 }
+
 static long vhost_vdpa_vring_ioctl(struct vhost_vdpa *v, unsigned int cmd,
 				   void __user *argp)
 {
@@ -390,6 +438,7 @@ static long vhost_vdpa_vring_ioctl(struct vhost_vdpa *v, unsigned int cmd,
 			cb.private = NULL;
 		}
 		ops->set_vq_cb(vdpa, idx, &cb);
+		vhost_vdpa_update_vq_irq(vq);
 		break;
 
 	case VHOST_SET_VRING_NUM:
@@ -765,6 +814,18 @@ static int vhost_vdpa_open(struct inode *inode, struct file *filep)
 	return r;
 }
 
+static void vhost_vdpa_clean_irq(struct vhost_vdpa *v)
+{
+	struct vhost_virtqueue *vq;
+	int i;
+
+	for (i = 0; i < v->nvqs; i++) {
+		vq = &v->vqs[i];
+		if (vq->call_ctx.producer.irq)
+			irq_bypass_unregister_producer(&vq->call_ctx.producer);
+	}
+}
+
 static int vhost_vdpa_release(struct inode *inode, struct file *filep)
 {
 	struct vhost_vdpa *v = filep->private_data;
@@ -777,6 +838,7 @@ static int vhost_vdpa_release(struct inode *inode, struct file *filep)
 	vhost_vdpa_iotlb_free(v);
 	vhost_vdpa_free_domain(v);
 	vhost_vdpa_config_put(v);
+	vhost_vdpa_clean_irq(v);
 	vhost_dev_cleanup(&v->vdev);
 	kfree(v->vdev.vqs);
 	mutex_unlock(&d->mutex);
@@ -951,6 +1013,8 @@ static struct vdpa_driver vhost_vdpa_driver = {
 	},
 	.probe	= vhost_vdpa_probe,
 	.remove	= vhost_vdpa_remove,
+	.setup_vq_irq = vhost_vdpa_setup_vq_irq,
+	.unsetup_vq_irq = vhost_vdpa_unsetup_vq_irq,
 };
 
 static int __init vhost_vdpa_init(void)
-- 
2.18.4

[PATCH V3 5/6] ifcvf: replace irq_request/free with vDPA helpers

[Zhu Lingshan]

This commit replaced irq_request/free() with vDPA helpers
vdpa_devm_request/free_irq() so that it can request/free
irq and setup irq offloading on order.

Signed-off-by: Zhu Lingshan <lingshan.zhu@intel.com>
---
 drivers/vdpa/ifcvf/ifcvf_main.c | 14 +++++++++-----
 1 file changed, 9 insertions(+), 5 deletions(-)

diff --git a/drivers/vdpa/ifcvf/ifcvf_main.c b/drivers/vdpa/ifcvf/ifcvf_main.c
index f5a60c14b979..65aab6bcaccb 100644
--- a/drivers/vdpa/ifcvf/ifcvf_main.c
+++ b/drivers/vdpa/ifcvf/ifcvf_main.c
@@ -47,11 +47,12 @@ static void ifcvf_free_irq(struct ifcvf_adapter *adapter, int queues)
 {
 	struct pci_dev *pdev = adapter->pdev;
 	struct ifcvf_hw *vf = &adapter->vf;
+	struct vdpa_device *vdpa = &adapter->vdpa;
 	int i;
 
 
 	for (i = 0; i < queues; i++)
-		devm_free_irq(&pdev->dev, vf->vring[i].irq, &vf->vring[i]);
+		vdpa_devm_free_irq(&pdev->dev, vdpa, vf->vring[i].irq, i, &vf->vring[i]);
 
 	ifcvf_free_irq_vectors(pdev);
 }
@@ -60,6 +61,7 @@ static int ifcvf_request_irq(struct ifcvf_adapter *adapter)
 {
 	struct pci_dev *pdev = adapter->pdev;
 	struct ifcvf_hw *vf = &adapter->vf;
+	struct vdpa_device *vdpa = &adapter->vdpa;
 	int vector, i, ret, irq;
 
 	ret = pci_alloc_irq_vectors(pdev, IFCVF_MAX_INTR,
@@ -73,6 +75,10 @@ static int ifcvf_request_irq(struct ifcvf_adapter *adapter)
 		 pci_name(pdev));
 	vector = 0;
 	irq = pci_irq_vector(pdev, vector);
+	/* This is config interrupt, config accesses all go
+	 * through userspace, so no need to setup
+	 * config interrupt offloading.
+	 */
 	ret = devm_request_irq(&pdev->dev, irq,
 			       ifcvf_config_changed, 0,
 			       vf->config_msix_name, vf);
@@ -82,13 +88,11 @@ static int ifcvf_request_irq(struct ifcvf_adapter *adapter)
 			 pci_name(pdev), i);
 		vector = i + IFCVF_MSI_QUEUE_OFF;
 		irq = pci_irq_vector(pdev, vector);
-		ret = devm_request_irq(&pdev->dev, irq,
+		ret = vdpa_devm_request_irq(&pdev->dev, vdpa, irq,
 				       ifcvf_intr_handler, 0,
 				       vf->vring[i].msix_name,
-				       &vf->vring[i]);
+				       &vf->vring[i], i);
 		if (ret) {
-			IFCVF_ERR(pdev,
-				  "Failed to request irq for vq %d\n", i);
 			ifcvf_free_irq(adapter, i);
 
 			return ret;
-- 
2.18.4

[PATCH V3 6/6] irqbypass: do not start cons/prod when failed connect

[Zhu Lingshan]

If failed to connect, there is no need to start consumer nor
producer.

Signed-off-by: Zhu Lingshan <lingshan.zhu@intel.com>
---
 virt/lib/irqbypass.c | 16 ++++++++++------
 1 file changed, 10 insertions(+), 6 deletions(-)

diff --git a/virt/lib/irqbypass.c b/virt/lib/irqbypass.c
index 28fda42e471b..c9bb3957f58a 100644
--- a/virt/lib/irqbypass.c
+++ b/virt/lib/irqbypass.c
@@ -40,17 +40,21 @@ static int __connect(struct irq_bypass_producer *prod,
 	if (prod->add_consumer)
 		ret = prod->add_consumer(prod, cons);
 
-	if (!ret) {
-		ret = cons->add_producer(cons, prod);
-		if (ret && prod->del_consumer)
-			prod->del_consumer(prod, cons);
-	}
+	if (ret)
+		goto err_add_consumer;
+
+	ret = cons->add_producer(cons, prod);
+	if (ret)
+		goto err_add_producer;
 
 	if (cons->start)
 		cons->start(cons);
 	if (prod->start)
 		prod->start(prod);
-
+err_add_producer:
+	if (prod->del_consumer)
+		prod->del_consumer(prod, cons);
+err_add_consumer:
 	return ret;
 }
 
-- 
2.18.4

Re: [PATCH V3 3/6] vDPA: implement vq IRQ allocate/free helpers in vDPA core

[Jason Wang]

On 2020/7/22 下午6:08, Zhu Lingshan wrote:
> +/*
> + * Request irq for a vq, setup irq offloading if its a vhost_vdpa vq.
> + * This function should be only called through setting virtio DRIVER_OK.
> + * If you want to request irq during probe, you should use raw APIs
> + * like request_irq() or devm_request_irq().


This makes the API less flexibile. The reason is we store the irq in 
vhost-vdpa not vDPA.

I wonder whether the following looks better:

1) store irq in vdpa device
2) register producer when DRIVER_OK and unregister producer when 
!DRIVER_OK in vhost-vDPA
3) deal with the synchronization with SET_VRING_CALL
4) document that irq is not expected to be changed during DRIVER_OK

This can make sure the API works during driver probe, and we don't need 
the setup_irq and unsetup_irq method in vdpa_driver

Thanks


> + */
> +int vdpa_devm_request_irq(struct device *dev, struct vdpa_device *vdev,
> +			  unsigned int irq, irq_handler_t handler,
> +			  unsigned long irqflags, const char *devname, void *dev_id,
> +			  int qid)
> +{
> +	int ret;
> +
> +	ret = devm_request_irq(dev, irq, handler, irqflags, devname, dev_id);
> +	if (ret)
> +		dev_err(dev, "Failed to request irq for vq %d\n", qid);
> +	else
> +		vdpa_setup_irq(vdev, qid, irq);
> +
> +	return ret;
> +
> +}
> +EXPORT_SYMBOL_GPL(vdpa_devm_request_irq);

[PATCH V3 0/6] IRQ offloading for vDPA

[Zhu Lingshan]

This series intends to implement IRQ offloading for
vhost_vdpa.

By the feat of irq forwarding facilities like posted
interrupt on X86, irq bypass can  help deliver
interrupts to vCPU directly.

vDPA devices have dedicated hardware backends like VFIO
pass-throughed devices. So it would be possible to setup
irq offloading(irq bypass) for vDPA devices and gain
performance improvements.

In my testing, with this feature, we can save 0.1ms
in a ping between two VFs on average.

changes from V2:
(1)rename struct vhost_call_ctx to vhost_vring_call
(2)add kvm_arch_end_assignment() in del_producer()
code path
(3)rename vDPA helpers to vdpa_devm_request_irq()
and vdpa_devm_free_irq(). Add better comments
for them.
(4)better comments for setup_vq_irq() and
unsetup_vq_irq()
(5)In vDPA VHOST_SET_VRING_CALL, will call vhost_vdpa_update_vq_irq()
without checking producer.irq, move this check into
vhost_vdpa_update_vq_irq(), so that get advantage of the spinlock.
(6)Add a function vhost_vdpa_clean_irq(), this function will unregister
the producer of vqs when vhost_vdpa_release(). This is safe
for control vq.
(7) minor improvements

changes from V1:
(1)dropped vfio changes.
(3)removed KVM_HVAE_IRQ_BYPASS checks
(4)locking fixes
(5)simplified vhost_vdpa_update_vq_irq()
(6)minor improvements

Zhu Lingshan (6):
  vhost: introduce vhost_vring_call
  kvm: detect assigned device via irqbypass manager
  vDPA: implement vq IRQ allocate/free helpers in vDPA core
  vhost_vdpa: implement IRQ offloading in vhost_vdpa
  ifcvf: replace irq_request/free with vDPA helpers
  irqbypass: do not start cons/prod when failed connect

 arch/x86/kvm/x86.c              | 11 +++++-
 drivers/vdpa/ifcvf/ifcvf_main.c | 14 ++++---
 drivers/vdpa/vdpa.c             | 49 +++++++++++++++++++++++
 drivers/vhost/Kconfig           |  1 +
 drivers/vhost/vdpa.c            | 70 +++++++++++++++++++++++++++++++--
 drivers/vhost/vhost.c           | 22 ++++++++---
 drivers/vhost/vhost.h           |  9 ++++-
 include/linux/vdpa.h            | 13 ++++++
 virt/lib/irqbypass.c            | 16 +++++---
 9 files changed, 182 insertions(+), 23 deletions(-)

-- 
2.18.4