Message-ID: <fa5a65cb-59c3-e5df-b217-0352c1f003d3@redhat.com>
Date:   Thu, 11 Jun 2020 10:56:33 +0800
From:   Jason Wang <jasowang@...hat.com>
To:     "Michael S. Tsirkin" <mst@...hat.com>
Cc:     linux-kernel@...r.kernel.org,
        virtualization@...ts.linux-foundation.org, rob.miller@...adcom.com,
        lingshan.zhu@...el.com, eperezma@...hat.com, lulu@...hat.com,
        shahafs@...lanox.com, hanand@...inx.com, mhabets@...arflare.com,
        gdawar@...inx.com, saugatm@...inx.com, vmireyno@...vell.com,
        zhangweining@...jie.com.cn, eli@...lanox.com
Subject: Re: [PATCH V3] vdpa: introduce virtio pci driver


On 2020/6/10 下午4:51, Michael S. Tsirkin wrote:
> On Wed, Jun 10, 2020 at 04:25:06PM +0800, Jason Wang wrote:
>>>> +
>>>> +#define VP_VDPA_FEATURES \
>>>> +	((1ULL << VIRTIO_F_ANY_LAYOUT)			| \
>>> This is presumably for transitional devices only.  In fact looking at
>>> code it seems that only net in legacy mode accepts VIRTIO_F_ANY_LAYOUT.
>>> Spec violation I guess ... but what should we do? Relax the spec
>>> or fix drivers?
>>
>> I don't get how it violates the spec.
> Spec says transitional drivers must ack VIRTIO_F_ANY_LAYOUT


I don't see this in the spec, maybe you can point it to me?


>
> But grep for VIRTIO_F_ANY_LAYOUT and you will see they don't.
> Only legacy virtio net does.
>
> Maybe it should have been
> transitional drivers when using the lgacy interface, but there
> you are.
>
>>>
>>>> +	 (1ULL << VIRTIO_F_VERSION_1)			| \
>>>> +	 (1ULL << VIRTIO_F_ORDER_PLATFORM)		| \
>>>> +	 (1ULL << VIRTIO_F_IOMMU_PLATFORM))
>>>> +
>>>> +struct vp_vring {
>>>> +	void __iomem *notify;
>>>> +	char msix_name[256];
>>>> +	resource_size_t notify_pa;
>>>> +	struct vdpa_callback cb;
>>>> +	int irq;
>>>> +};
>>>> +
>>>> +struct vp_vdpa {
>>>> +	struct vdpa_device vdpa;
>>>> +	struct pci_dev *pdev;
>>>> +
>>>> +	struct virtio_device_id id;
>>>> +
>>>> +	struct vp_vring vring[VP_VDPA_MAX_QUEUE];
>>>> +
>>>> +	/* The IO mapping for the PCI config space */
>>>> +	void __iomem * const *base;
>>>> +	struct virtio_pci_common_cfg __iomem *common;
>>>> +	void __iomem *device;
>>>> +	/* Base of vq notifications */
>>>> +	void __iomem *notify;
>>>> +
>>>> +	/* Multiplier for queue_notify_off. */
>>>> +	u32 notify_off_multiplier;
>>>> +
>>>> +	int modern_bars;
>>>> +	int vectors;
>>>> +};
>>>> +
>>>> +static struct vp_vdpa *vdpa_to_vp(struct vdpa_device *vdpa)
>>>> +{
>>>> +	return container_of(vdpa, struct vp_vdpa, vdpa);
>>>> +}
>>>> +
>>>> +/*
>>>> + * Type-safe wrappers for io accesses.
>>>> + * Use these to enforce at compile time the following spec requirement:
>>>> + *
>>>> + * The driver MUST access each field using the “natural” access
>>>> + * method, i.e. 32-bit accesses for 32-bit fields, 16-bit accesses
>>>> + * for 16-bit fields and 8-bit accesses for 8-bit fields.
>>>> + */
>>>> +static inline u8 vp_ioread8(u8 __iomem *addr)
>>>> +{
>>>> +	return ioread8(addr);
>>>> +}
>>>> +static inline u16 vp_ioread16(__le16 __iomem *addr)
>>>> +{
>>>> +	return ioread16(addr);
>>>> +}
>>>> +
>>>> +static inline u32 vp_ioread32(__le32 __iomem *addr)
>>>> +{
>>>> +	return ioread32(addr);
>>>> +}
>>>> +
>>>> +static inline void vp_iowrite8(u8 value, u8 __iomem *addr)
>>>> +{
>>>> +	iowrite8(value, addr);
>>>> +}
>>>> +
>>>> +static inline void vp_iowrite16(u16 value, __le16 __iomem *addr)
>>>> +{
>>>> +	iowrite16(value, addr);
>>>> +}
>>>> +
>>>> +static inline void vp_iowrite32(u32 value, __le32 __iomem *addr)
>>>> +{
>>>> +	iowrite32(value, addr);
>>>> +}
>>>> +
>>>> +static void vp_iowrite64_twopart(u64 val,
>>>> +				 __le32 __iomem *lo, __le32 __iomem *hi)
>>>> +{
>>>> +	vp_iowrite32((u32)val, lo);
>>>> +	vp_iowrite32(val >> 32, hi);
>>>> +}
>>>> +
>>>> +static int find_capability(struct pci_dev *dev, u8 cfg_type,
>>>> +			   u32 ioresource_types, int *bars)
>>>> +{
>>>> +	int pos;
>>>> +
>>>> +	for (pos = pci_find_capability(dev, PCI_CAP_ID_VNDR);
>>>> +	     pos > 0;
>>>> +	     pos = pci_find_next_capability(dev, pos, PCI_CAP_ID_VNDR)) {
>>>> +		u8 type, bar;
>>>> +
>>>> +		pci_read_config_byte(dev, pos + offsetof(struct virtio_pci_cap,
>>>> +							 cfg_type),
>>>> +				     &type);
>>>> +		pci_read_config_byte(dev, pos + offsetof(struct virtio_pci_cap,
>>>> +							 bar),
>>>> +				     &bar);
>>>> +
>>>> +		/* Ignore structures with reserved BAR values */
>>>> +		if (bar > 0x5)
>>>> +			continue;
>>>> +
>>>> +		if (type == cfg_type) {
>>>> +			if (pci_resource_len(dev, bar) &&
>>>> +			    pci_resource_flags(dev, bar) & ioresource_types) {
>>>> +				*bars |= (1 << bar);
>>>> +				return pos;
>>>> +			}
>>>> +		}
>>>> +	}
>>>> +	return 0;
>>>> +}
>>>> +
>>>> +static void __iomem *map_capability(struct vp_vdpa *vp_vdpa, int off,
>>>> +				    resource_size_t *pa)
>>>> +{
>>>> +	struct pci_dev *pdev = vp_vdpa->pdev;
>>>> +	u32 offset;
>>>> +	u8 bar;
>>>> +
>>>> +	pci_read_config_byte(pdev,
>>>> +			     off + offsetof(struct virtio_pci_cap, bar),
>>>> +			     &bar);
>>>> +	pci_read_config_dword(pdev,
>>>> +			      off + offsetof(struct virtio_pci_cap, offset),
>>>> +			      &offset);
>>>> +
>>>> +	if (pa)
>>>> +		*pa = pci_resource_start(pdev, bar) + offset;
>>>> +
>>>> +	return vp_vdpa->base[bar] + offset;
>>>> +}
>>>> +
>>>> +static u64 vp_vdpa_get_features(struct vdpa_device *vdpa)
>>>> +{
>>>> +	struct vp_vdpa *vp_vdpa = vdpa_to_vp(vdpa);
>>>> +	u64 features;
>>>> +
>>>> +	vp_iowrite32(0, &vp_vdpa->common->device_feature_select);
>>>> +	features = vp_ioread32(&vp_vdpa->common->device_feature);
>>>> +	vp_iowrite32(1, &vp_vdpa->common->device_feature_select);
>>>> +	features |= ((u64)vp_ioread32(&vp_vdpa->common->device_feature) << 32);
>>>> +	features &= VP_VDPA_FEATURES;
>>>> +
>>>> +	return features;
>>>> +}
>>>> +
>>>> +static int vp_vdpa_set_features(struct vdpa_device *vdpa, u64 features)
>>>> +{
>>>> +	struct vp_vdpa *vp_vdpa = vdpa_to_vp(vdpa);
>>>> +
>>>> +	vp_iowrite32(0, &vp_vdpa->common->guest_feature_select);
>>>> +	vp_iowrite32((u32)features, &vp_vdpa->common->guest_feature);
>>>> +	vp_iowrite32(1, &vp_vdpa->common->guest_feature_select);
>>>> +	vp_iowrite32(features >> 32, &vp_vdpa->common->guest_feature);
>>>> +
>>>> +	return 0;
>>>> +}
>>>> +
>>>> +static u8 vp_vdpa_get_status(struct vdpa_device *vdpa)
>>>> +{
>>>> +	struct vp_vdpa *vp_vdpa = vdpa_to_vp(vdpa);
>>>> +
>>>> +	return vp_ioread8(&vp_vdpa->common->device_status);
>>>> +}
>>>> +
>>>> +static void vp_vdpa_free_irq(struct vp_vdpa *vp_vdpa)
>>>> +{
>>>> +	struct pci_dev *pdev = vp_vdpa->pdev;
>>>> +	int i;
>>>> +
>>>> +	for (i = 0; i < VP_VDPA_MAX_QUEUE; i++) {
>>>> +		if (vp_vdpa->vring[i].irq != -1) {
>>>> +			vp_iowrite16(i, &vp_vdpa->common->queue_select);
>>>> +			vp_iowrite16(VIRTIO_MSI_NO_VECTOR,
>>>> +				     &vp_vdpa->common->queue_msix_vector);
>>>> +			devm_free_irq(&pdev->dev, vp_vdpa->vring[i].irq,
>>>> +				      &vp_vdpa->vring[i]);
>>>> +			vp_vdpa->vring[i].irq = -1;
>>>> +		}
>>>> +	}
>>>> +
>>>> +	if (vp_vdpa->vectors) {
>>>> +		pci_free_irq_vectors(pdev);
>>>> +		vp_vdpa->vectors = 0;
>>>> +	}
>>>> +}
>>>> +
>>>> +static irqreturn_t vp_vdpa_intr_handler(int irq, void *arg)
>>>> +{
>>>> +	struct vp_vring *vring = arg;
>>>> +
>>>> +	if (vring->cb.callback)
>>>> +		return vring->cb.callback(vring->cb.private);
>>>> +
>>>> +	return IRQ_HANDLED;
>>>> +}
>>>> +
>>>> +static int vp_vdpa_request_irq(struct vp_vdpa *vp_vdpa)
>>>> +{
>>>> +	struct pci_dev *pdev = vp_vdpa->pdev;
>>>> +	int i, ret, irq;
>>>> +
>>>> +	ret = pci_alloc_irq_vectors(pdev, VP_VDPA_MAX_QUEUE,
>>>> +				    VP_VDPA_MAX_QUEUE, PCI_IRQ_MSIX);
>>>> +	if (ret != VP_VDPA_MAX_QUEUE) {
>>>> +		dev_err(&pdev->dev, "vp_vdpa: fail to allocate irq vectors\n");
>>>> +		return ret;
>>>> +	}
>>>> +
>>>> +	vp_vdpa->vectors = VP_VDPA_MAX_QUEUE;
>>>> +
>>>> +	for (i = 0; i < VP_VDPA_MAX_QUEUE; i++) {
>>>> +		snprintf(vp_vdpa->vring[i].msix_name, 256,
>>>> +			"vp-vdpa[%s]-%d\n", pci_name(pdev), i);
>>>> +		irq = pci_irq_vector(pdev, i);
>>>> +		ret = devm_request_irq(&pdev->dev, irq,
>>>> +				       vp_vdpa_intr_handler,
>>>> +				       0, vp_vdpa->vring[i].msix_name,
>>>> +				       &vp_vdpa->vring[i]);
>>>> +		if (ret) {
>>>> +			dev_err(&pdev->dev, "vp_vdpa: fail to request irq for vq %d\n",
>>>> +				i);
>>>> +			goto err;
>>>> +		}
>>>> +		vp_iowrite16(i, &vp_vdpa->common->queue_select);
>>>> +		vp_iowrite16(i, &vp_vdpa->common->queue_msix_vector);
>>>> +		vp_vdpa->vring[i].irq = irq;
>>>> +	}
>>>> +
>>>> +	return 0;
>>>> +err:
>>>> +	vp_vdpa_free_irq(vp_vdpa);
>>>> +	return ret;
>>>> +}
>>>> +
>>>> +static void vp_vdpa_set_status(struct vdpa_device *vdpa, u8 status)
>>>> +{
>>>> +	struct vp_vdpa *vp_vdpa = vdpa_to_vp(vdpa);
>>>> +	u8 s = vp_vdpa_get_status(vdpa);
>>>> +
>>>> +	if (status & VIRTIO_CONFIG_S_DRIVER_OK &&
>>>> +	    !(s & VIRTIO_CONFIG_S_DRIVER_OK)) {
>>>> +		vp_vdpa_request_irq(vp_vdpa);
>>>> +	}
>>>> +
>>>> +	vp_iowrite8(status, &vp_vdpa->common->device_status);
>>>> +
>>>> +	if (!(status & VIRTIO_CONFIG_S_DRIVER_OK) &&
>>>> +	    (s & VIRTIO_CONFIG_S_DRIVER_OK))
>>>> +		vp_vdpa_free_irq(vp_vdpa);
>>>> +}
>>>> +
>>>> +static u16 vp_vdpa_get_vq_num_max(struct vdpa_device *vdpa)
>>>> +{
>>>> +	struct vp_vdpa *vp_vdpa = vdpa_to_vp(vdpa);
>>>> +
>>>> +	return vp_ioread16(&vp_vdpa->common->queue_size);
>>>> +}
>>>> +
>>>> +static u64 vp_vdpa_get_vq_state(struct vdpa_device *vdpa, u16 qid)
>>>> +{
>>>> +	return 0;
>>>> +}
>>>> +
>>>> +static struct vdpa_notification_area
>>>> +vp_vdpa_get_vq_notification(struct vdpa_device *vdpa, u16 qid)
>>>> +{
>>>> +	struct vp_vdpa *vp_vdpa = vdpa_to_vp(vdpa);
>>>> +	struct vdpa_notification_area notify;
>>>> +
>>>> +	notify.addr = vp_vdpa->vring[qid].notify_pa;
>>>> +	notify.size = vp_vdpa->notify_off_multiplier;
>>>> +
>>>> +	return notify;
>>>> +}
>>>> +
>>>> +static int vp_vdpa_set_vq_state(struct vdpa_device *vdpa, u16 qid,
>>>> +				u64 num)
>>>> +{
>>>> +	/* Note that this is not supported by virtio specification, so
>>>> +	 * we return -ENOTSUPP here. This means we can't support live
>>>> +	 * migration, vhost device start/stop.
>>>> +	 */
>>>> +
>>>> +	return -ENOTSUPP;
>>>> +}
>>>> +
>>>> +static void vp_vdpa_set_vq_cb(struct vdpa_device *vdpa, u16 qid,
>>>> +			      struct vdpa_callback *cb)
>>>> +{
>>>> +	struct vp_vdpa *vp_vdpa = vdpa_to_vp(vdpa);
>>>> +
>>>> +	vp_vdpa->vring[qid].cb = *cb;
>>>> +}
>>>> +
>>>> +static void vp_vdpa_set_vq_ready(struct vdpa_device *vdpa,
>>>> +				 u16 qid, bool ready)
>>>> +{
>>>> +	struct vp_vdpa *vp_vdpa = vdpa_to_vp(vdpa);
>>>> +
>>>> +	vp_iowrite16(qid, &vp_vdpa->common->queue_select);
>>>> +	vp_iowrite16(ready, &vp_vdpa->common->queue_enable);
>>>> +}
>>>> +
>>>> +static bool vp_vdpa_get_vq_ready(struct vdpa_device *vdpa, u16 qid)
>>>> +{
>>>> +	struct vp_vdpa *vp_vdpa = vdpa_to_vp(vdpa);
>>>> +
>>>> +	vp_iowrite16(qid, &vp_vdpa->common->queue_select);
>>>> +
>>>> +	return vp_ioread16(&vp_vdpa->common->queue_enable);
>>>> +}
>>>> +
>>>> +static void vp_vdpa_set_vq_num(struct vdpa_device *vdpa, u16 qid,
>>>> +			       u32 num)
>>>> +{
>>>> +	struct vp_vdpa *vp_vdpa = vdpa_to_vp(vdpa);
>>>> +
>>>> +	vp_iowrite16(num, &vp_vdpa->common->queue_size);
>>>> +}
>>>> +
>>>> +static int vp_vdpa_set_vq_address(struct vdpa_device *vdpa, u16 qid,
>>>> +				  u64 desc_area, u64 driver_area,
>>>> +				  u64 device_area)
>>>> +{
>>>> +	struct vp_vdpa *vp_vdpa = vdpa_to_vp(vdpa);
>>>> +	struct virtio_pci_common_cfg __iomem *cfg = vp_vdpa->common;
>>>> +
>>>> +	vp_iowrite16(qid, &cfg->queue_select);
>>>> +	vp_iowrite64_twopart(desc_area,
>>>> +			     &cfg->queue_desc_lo, &cfg->queue_desc_hi);
>>>> +	vp_iowrite64_twopart(driver_area,
>>>> +			     &cfg->queue_avail_lo, &cfg->queue_avail_hi);
>>>> +	vp_iowrite64_twopart(device_area,
>>>> +			     &cfg->queue_used_lo, &cfg->queue_used_hi);
>>>> +
>>>> +	return 0;
>>>> +}
>>>> +
>>>> +static void vp_vdpa_kick_vq(struct vdpa_device *vdpa, u16 qid)
>>>> +{
>>>> +	struct vp_vdpa *vp_vdpa = vdpa_to_vp(vdpa);
>>>> +
>>>> +	vp_iowrite16(qid, vp_vdpa->vring[qid].notify);
>>>> +}
>>>> +
>>>> +static u32 vp_vdpa_get_generation(struct vdpa_device *vdpa)
>>>> +{
>>>> +	struct vp_vdpa *vp_vdpa = vdpa_to_vp(vdpa);
>>>> +
>>>> +	return vp_ioread8(&vp_vdpa->common->config_generation);
>>>> +}
>>>> +
>>>> +static u32 vp_vdpa_get_device_id(struct vdpa_device *vdpa)
>>>> +{
>>>> +	struct vp_vdpa *vp_vdpa = vdpa_to_vp(vdpa);
>>>> +
>>>> +	return vp_vdpa->id.device;
>>>> +}
>>>> +
>>>> +static u32 vp_vdpa_get_vendor_id(struct vdpa_device *vdpa)
>>>> +{
>>>> +	struct vp_vdpa *vp_vdpa = vdpa_to_vp(vdpa);
>>>> +
>>>> +	return vp_vdpa->id.vendor;
>>>> +}
>>>> +
>>>> +static u32 vp_vdpa_get_vq_align(struct vdpa_device *vdpa)
>>>> +{
>>>> +	return PAGE_SIZE;
>>>> +}
>>>> +
>>>> +static void vp_vdpa_get_config(struct vdpa_device *vdpa,
>>>> +			       unsigned int offset,
>>>> +			       void *buf, unsigned int len)
>>>> +{
>>>> +	struct vp_vdpa *vp_vdpa = vdpa_to_vp(vdpa);
>>>> +	u8 old, new;
>>>> +	u8 *p;
>>>> +	int i;
>>>> +
>>>> +	do {
>>>> +		old = vp_ioread8(&vp_vdpa->common->config_generation);
>>>> +		p = buf;
>>>> +		for (i = 0; i < len; i++)
>>>> +			*p++ = vp_ioread8(vp_vdpa->device + offset + i);
>>>> +
>>>> +		new = vp_ioread8(&vp_vdpa->common->config_generation);
>>>> +	} while (old != new);
>>>> +}
>>>> +
>>>> +static void vp_vdpa_set_config(struct vdpa_device *vdpa,
>>>> +			       unsigned int offset, const void *buf,
>>>> +			       unsigned int len)
>>>> +{
>>>> +	struct vp_vdpa *vp_vdpa = vdpa_to_vp(vdpa);
>>>> +	const u8 *p = buf;
>>>> +	int i;
>>>> +
>>>> +	for (i = 0; i < len; i++)
>>>> +		vp_iowrite8(*p++, vp_vdpa->device + offset + i);
>>>> +}
>>>> +
>>>> +static void vp_vdpa_set_config_cb(struct vdpa_device *vdpa,
>>>> +				  struct vdpa_callback *cb)
>>>> +{
>>>> +	/* We don't support config interrupt */
>>> Breaks things like balloon or migration notifications with virtio net, doesn't it?
>>
>> Right, will fix.
>>
>>
>>>> +}
>>>> +
>>>> +static const struct vdpa_config_ops vp_vdpa_ops = {
>>>> +	.get_features	= vp_vdpa_get_features,
>>>> +	.set_features	= vp_vdpa_set_features,
>>>> +	.get_status	= vp_vdpa_get_status,
>>>> +	.set_status	= vp_vdpa_set_status,
>>>> +	.get_vq_num_max	= vp_vdpa_get_vq_num_max,
>>>> +	.get_vq_state	= vp_vdpa_get_vq_state,
>>>> +	.get_vq_notification = vp_vdpa_get_vq_notification,
>>>> +	.set_vq_state	= vp_vdpa_set_vq_state,
>>>> +	.set_vq_cb	= vp_vdpa_set_vq_cb,
>>>> +	.set_vq_ready	= vp_vdpa_set_vq_ready,
>>>> +	.get_vq_ready	= vp_vdpa_get_vq_ready,
>>>> +	.set_vq_num	= vp_vdpa_set_vq_num,
>>>> +	.set_vq_address	= vp_vdpa_set_vq_address,
>>>> +	.kick_vq	= vp_vdpa_kick_vq,
>>>> +	.get_generation	= vp_vdpa_get_generation,
>>>> +	.get_device_id	= vp_vdpa_get_device_id,
>>>> +	.get_vendor_id	= vp_vdpa_get_vendor_id,
>>>> +	.get_vq_align	= vp_vdpa_get_vq_align,
>>>> +	.get_config	= vp_vdpa_get_config,
>>>> +	.set_config	= vp_vdpa_set_config,
>>>> +	.set_config_cb  = vp_vdpa_set_config_cb,
>>>> +};
>>>> +
>>>> +static int vp_vdpa_probe(struct pci_dev *pdev, const struct pci_device_id *id)
>>>> +{
>>>> +	struct device *dev = &pdev->dev;
>>>> +	struct vp_vdpa *vp_vdpa;
>>>> +	int common, notify, device, ret, i;
>>>> +	struct virtio_device_id virtio_id;
>>>> +	resource_size_t notify_pa;
>>>> +	u16 notify_off;
>>>> +
>>>> +	/* We only own devices >= 0x1000 and <= 0x107f: leave the rest. */
>>>> +	if (pdev->device < 0x1000 || pdev->device > 0x107f)
>>>> +		return -ENODEV;
>>>> +
>>>> +	if (pdev->device < 0x1040) {
>>>> +		/* Transitional devices: use the PCI subsystem device id as
>>>> +		 * virtio device id, same as legacy driver always did.
>>>> +		 */
>>>> +		virtio_id.device = pdev->subsystem_device;
>>>> +	} else {
>>>> +		/* Modern devices: simply use PCI device id,
>>>> +		 * but start from 0x1040.
>>>> +		 */
>>>> +		virtio_id.device = pdev->device - 0x1040;
>>>> +	}
>>>> +	virtio_id.vendor = pdev->subsystem_vendor;
>>>> +
>>>> +	ret = pcim_enable_device(pdev);
>>>> +	if (ret) {
>>>> +		dev_err(dev, "vp_vdpa: Fail to enable PCI device\n");
>>>> +		return ret;
>>>> +	}
>>>> +
>>>> +	vp_vdpa = vdpa_alloc_device(struct vp_vdpa, vdpa,
>>>> +				    dev, &vp_vdpa_ops);
>>>> +	if (vp_vdpa == NULL) {
>>>> +		dev_err(dev, "vp_vdpa: Failed to allocate vDPA structure\n");
>>>> +		return -ENOMEM;
>>>> +	}
>>>> +
>>>> +	pci_set_master(pdev);
>>>> +	pci_set_drvdata(pdev, vp_vdpa);
>>>> +
>>>> +	vp_vdpa->pdev = pdev;
>>>> +	vp_vdpa->vdpa.dma_dev = &pdev->dev;
>>>> +
>>>> +	common = find_capability(pdev, VIRTIO_PCI_CAP_COMMON_CFG,
>>>> +				 IORESOURCE_IO | IORESOURCE_MEM,
>>>> +				 &vp_vdpa->modern_bars);
>>>> +	if (!common) {
>>>> +		dev_err(&pdev->dev,
>>>> +			"vp_vdpa: legacy device is not supported\n");
>>>> +		ret = -ENODEV;
>>>> +		goto err;
>>>> +	}
>>>> +
>>>> +	notify = find_capability(pdev, VIRTIO_PCI_CAP_NOTIFY_CFG,
>>>> +				 IORESOURCE_IO | IORESOURCE_MEM,
>>>> +				 &vp_vdpa->modern_bars);
>>>> +	if (!notify) {
>>>> +		dev_err(&pdev->dev,
>>>> +			"vp_vdpa: missing notification capabilities\n");
>>>> +		ret = -EINVAL;
>>>> +		goto err;
>>>> +	}
>>>> +
>>>> +	device = find_capability(pdev, VIRTIO_PCI_CAP_DEVICE_CFG,
>>>> +				 IORESOURCE_IO | IORESOURCE_MEM,
>>>> +				 &vp_vdpa->modern_bars);
>>>> +	if (!device) {
>>>> +		dev_err(&pdev->dev,
>>>> +			"vp_vdpa: missing device capabilities\n");
>>>> +		ret = -EINVAL;
>>>> +		goto err;
>>>> +	}
>>>> +
>>>> +	ret = pcim_iomap_regions(pdev, vp_vdpa->modern_bars,
>>>> +				 VP_VDPA_DRIVER_NAME);
>>>> +	if (ret)
>>>> +		goto err;
>>>> +
>>>> +	vp_vdpa->base = pcim_iomap_table(pdev);
>>>> +
>>>> +	ret = dma_set_mask_and_coherent(&pdev->dev, DMA_BIT_MASK(64));
>>>> +	if (ret)
>>>> +		ret = dma_set_mask_and_coherent(&pdev->dev,
>>>> +						DMA_BIT_MASK(32));
>>>> +	if (ret)
>>>> +		dev_warn(&pdev->dev, "Failed to enable 64-bit or 32-bit DMA.  Trying to continue, but this might not work.\n");
>>>> +
>>>> +	vp_vdpa->device = map_capability(vp_vdpa, device, NULL);
>>>> +	vp_vdpa->notify = map_capability(vp_vdpa, notify, &notify_pa);
>>>> +	vp_vdpa->common = map_capability(vp_vdpa, common, NULL);
>>>> +	vp_vdpa->id = virtio_id;
>>>> +
>>>> +	ret = vdpa_register_device(&vp_vdpa->vdpa);
>>>> +	if (ret) {
>>>> +		dev_err(&pdev->dev, "Failed to register to vdpa bus\n");
>>>> +		goto err;
>>>> +	}
>>>> +
>>>> +	pci_read_config_dword(pdev, notify + sizeof(struct virtio_pci_cap),
>>>> +			      &vp_vdpa->notify_off_multiplier);
>>>> +
>>>> +	for (i = 0; i < VP_VDPA_MAX_QUEUE; i++) {
>>>> +		vp_iowrite16(i, &vp_vdpa->common->queue_select);
>>>> +		notify_off = vp_ioread16(&vp_vdpa->common->queue_notify_off);
>>>> +		vp_vdpa->vring[i].irq = -1;
>>>> +		vp_vdpa->vring[i].notify = vp_vdpa->notify +
>>>> +			notify_off * vp_vdpa->notify_off_multiplier;
>>>> +		vp_vdpa->vring[i].notify_pa = notify_pa +
>>>> +			notify_off * vp_vdpa->notify_off_multiplier;
>>>> +	}
>>>> +
>>>> +	return 0;
>>>> +
>>>> +err:
>>>> +	put_device(&vp_vdpa->vdpa.dev);
>>>> +	return ret;
>>>> +}
>>>> +
>>>> +static void vp_vdpa_remove(struct pci_dev *pdev)
>>>> +{
>>>> +	struct vp_vdpa *vp_vdpa = pci_get_drvdata(pdev);
>>>> +
>>>> +	vdpa_unregister_device(&vp_vdpa->vdpa);
>>>> +}
>>>> +
>>>> +static struct pci_driver vp_vdpa_driver = {
>>>> +	.name		= "vp-vdpa",
>>>> +	.id_table	= NULL, /* only dynamic ids */
>>>> +	.probe		= vp_vdpa_probe,
>>>> +	.remove		= vp_vdpa_remove,
>>>> +};
>>>> +
>>>> +module_pci_driver(vp_vdpa_driver);
>>>> +
>>>> +MODULE_AUTHOR("Jason Wang <jasowang@...hat.com>");
>>>> +MODULE_DESCRIPTION("vp-vdpa");
>>>> +MODULE_LICENSE("GPL");
>>>> +MODULE_VERSION("1");
>>> Isn't there something we can do to reduce the amount of code
>>> duplication? virtio, ifcvf and now this share a ton of code ...
>>> Let's make a library?
>>
>> I do think about this since IFCVF driver is posted.
>>
>> It depends on what level we want to share. Do we have to have a common
>> parent structure for modern pci device? If yes, it probably requires
>> non-trivial refactoring on the existed modern virtio-pci driver. If we just
>> want to share some helpers, it would be easy.
> Share some helpers I'd say.


Ok, I can do that.


>
>> I think we can do those stuffs on top.
> Well with a 3rd copy in-tree I'm inclined to say no, let's refactor first.


Yes.

Thanks


>
>> Thanks
>>
>>
>>>> -- 
>>>> 2.20.1

Powered by blists - more mailing lists