vDPA support
Subsystems (19)
v8 cover vDPA support
Hi all:
This is an update version of vDPA support in kernel.
vDPA device is a device that uses a datapath which complies with the virtio specifications with vendor specific control path. vDPA devices can be both physically located on the hardware or emulated by software. vDPA hardware devices are usually implemented through PCIE with the following types:
- PF (Physical Function) - A single Physical Function
- VF (Virtual Function) - Device that supports single root I/O virtualization (SR-IOV). Its Virtual Function (VF) represents a virtualized instance of the device that can be assigned to different partitions
- ADI (Assignable Device Interface) and its equivalents - With technologies such as Intel Scalable IOV, a virtual device (VDEV) composed by host OS utilizing one or more ADIs. Or its equivalent like SF (Sub function) from Mellanox.
From a driver's perspective, depends on how and where the DMA translation is done, vDPA devices are split into two types:
- Platform specific DMA translation - From the driver's perspective, the device can be used on a platform where device access to data in memory is limited and/or translated. An example is a PCIE vDPA whose DMA request was tagged via a bus (e.g PCIE) specific way. DMA translation and protection are done at PCIE bus IOMMU level.
- Device specific DMA translation - The device implements DMA isolation and protection through its own logic. An example is a vDPA device which uses on-chip IOMMU.
To hide the differences and complexity of the above types for a vDPA device/IOMMU options and in order to present a generic virtio device to the upper layer, a device agnostic framework is required.
This series introduces a software vDPA bus which abstracts the common attributes of vDPA device, vDPA bus driver and the communication method, the bus operations (vdpa_config_ops) between the vDPA device abstraction and the vDPA bus driver. This allows multiple types of drivers to be used for vDPA device like the virtio_vdpa and vhost_vdpa driver to operate on the bus and allow vDPA device could be used by either kernel virtio driver or userspace vhost drivers as:
virtio drivers vhost drivers
| |
[virtio bus] [vhost uAPI]
| |
virtio device vhost device
virtio_vdpa drv vhost_vdpa drv
\ /
[vDPA bus]
|
vDPA device
hardware drv
|
[hardware bus]
|
vDPA hardwarevirtio_vdpa driver is a transport implementation for kernel virtio drivers on top of vDPA bus operations. An alternative is to refactor virtio bus which is sub-optimal since the bus and drivers are designed to be use by kernel subsystem, a non-trivial major refactoring is needed which may impact a brunches of drivers and devices implementation inside the kernel. Using a new transport may grealy simply both the design and changes.
vhost_vdpa driver is a new type of vhost device which allows userspace vhost drivers to use vDPA devices via vhost uAPI (with minor extension). This help to minimize the changes of existed vhost drivers for using vDPA devices.
With the abstraction of vDPA bus and vDPA bus operations, the difference and complexity of the under layer hardware is hidden from upper layer. The vDPA bus drivers on top can use a unified vdpa_config_ops to control different types of vDPA device.
Two drivers were implemented with the framework introduced in this series:
- Intel IFC VF driver which depends on the platform IOMMU for DMA translation
- VDPA simulator which is a software test device with an emulated onchip IOMMU
Future work:
- direct doorbell mapping support
- control virtqueue support
- dirty page tracking support
- direct interrupt support
- management API (devlink)
Please review.
Thanks
Changes from V7:
- refine kconfig to solve the dependency issues on archs that lacks of CONFIG_VIRTUALIZATION (kbuild)
Changes from V6:
- vdpa_alloc_device() will allocate parent strcture (Jason)
- remove the vdpa core dev info in IFCVF patch (Jason)
- provide a free method in the vdpa bus operations for drivet to free private data
- switch to use new layout in vdapsim and IFCVF
- make IFCVF depends on PCI_MSI (kbuild)
- some tweaks on the IFCVF driver
Changes from V5:
- include Intel IFCVF driver and vhost-vdpa drivers
- add the platform IOMMU support for vhost-vdpa
- check the return value of dev_set_name() (Jason)
- various tweaks and fixes
Changes from V4:
- use put_device() instead of kfree when fail to register virtio device (Jason)
- simplify the error handling when allocating vdpasim device (Jason)
- don't use device_for_each_child() during module exit (Jason)
- correct the error checking for vdpa_alloc_device() (Harpreet, Lingshan)
Changes from V3:
- various Kconfig fixes (Randy)
Changes from V2:
- release idr in the release function for put_device() unwind (Jason)
- don't panic when fail to register vdpa bus (Jason)
- use unsigned int instead of int for ida (Jason)
- fix the wrong commit log in virito_vdpa patches (Jason)
- make vdpa_sim depends on RUNTIME_TESTING_MENU (Michael)
- provide a bus release function for vDPA device (Jason)
- fix the wrong unwind when creating devices for vDPA simulator (Jason)
- move vDPA simulator to a dedicated directory (Lingshan)
- cancel the work before release vDPA simulator
Changes from V1:
- drop sysfs API, leave the management interface to future development (Michael)
- introduce incremental DMA ops (dma_map/dma_unmap) (Michael)
- introduce dma_device and use it instead of parent device for doing IOMMU or DMA from bus driver (Michael, Jason, Ling Shan, Tiwei)
- accept parent device and dma device when register vdpa device
- coding style and compile fixes (Randy)
- using vdpa_xxx instead of xxx_vdpa (Jason)
- ove vDPA accessors to header and make it static inline (Jason)
- split vdp_register_device() into two helpers vdpa_init_device() and vdpa_register_device() which allows intermediate step to be done (Jason)
- warn on invalidate queue state when fail to creating virtqueue (Jason)
- make to_virtio_vdpa_device() static (Jason)
- use kmalloc/kfree instead of devres for virtio vdpa device (Jason)
- avoid using cast in vdpa bus function (Jason)
- introduce module_vdpa_driver and fix module refcnt (Jason)
- fix returning freed address in vdapsim coherent DMA addr allocation (Dan)
- various other fixes and tweaks
V7: https://lkml.org/lkml/2020/3/24/21
V6: https://lkml.org/lkml/2020/3/18/88
V5: https://lkml.org/lkml/2020/2/26/58
V4: https://lkml.org/lkml/2020/2/20/59
V3: https://lkml.org/lkml/2020/2/19/1347
V2: https://lkml.org/lkml/2020/2/9/275
V1: https://lkml.org/lkml/2020/1/16/353
Jason Wang (7):
vhost: refine vhost and vringh kconfig
vhost: allow per device message handler
vhost: factor out IOTLB
vringh: IOTLB support
vDPA: introduce vDPA bus
virtio: introduce a vDPA based transport
vdpasim: vDPA device simulator
Tiwei Bie (1):
vhost: introduce vDPA-based backend
Zhu Lingshan (1):
virtio: Intel IFC VF driver for VDPA
MAINTAINERS | 2 + arch/arm/kvm/Kconfig | 2 - arch/arm64/kvm/Kconfig | 2 - arch/mips/kvm/Kconfig | 2 - arch/powerpc/kvm/Kconfig | 2 - arch/s390/kvm/Kconfig | 4 - arch/x86/kvm/Kconfig | 4 - drivers/Kconfig | 2 + drivers/misc/mic/Kconfig | 4 - drivers/net/caif/Kconfig | 4 - drivers/vhost/Kconfig | 42 +- drivers/vhost/Kconfig.vringh | 6 - drivers/vhost/Makefile | 6 + drivers/vhost/iotlb.c | 177 +++++ drivers/vhost/net.c | 5 +- drivers/vhost/scsi.c | 2 +- drivers/vhost/vdpa.c | 883 ++++++++++++++++++++++++ drivers/vhost/vhost.c | 233 +++---- drivers/vhost/vhost.h | 45 +- drivers/vhost/vringh.c | 421 ++++++++++- drivers/vhost/vsock.c | 2 +- drivers/virtio/Kconfig | 15 + drivers/virtio/Makefile | 2 + drivers/virtio/vdpa/Kconfig | 37 + drivers/virtio/vdpa/Makefile | 4 + drivers/virtio/vdpa/ifcvf/Makefile | 3 + drivers/virtio/vdpa/ifcvf/ifcvf_base.c | 386 +++++++++++ drivers/virtio/vdpa/ifcvf/ifcvf_base.h | 132 ++++ drivers/virtio/vdpa/ifcvf/ifcvf_main.c | 474 +++++++++++++ drivers/virtio/vdpa/vdpa.c | 180 +++++ drivers/virtio/vdpa/vdpa_sim/Makefile | 2 + drivers/virtio/vdpa/vdpa_sim/vdpa_sim.c | 629 +++++++++++++++++ drivers/virtio/virtio_vdpa.c | 396 +++++++++++ include/linux/vdpa.h | 247 +++++++ include/linux/vhost_iotlb.h | 47 ++ include/linux/vringh.h | 36 + include/uapi/linux/vhost.h | 24 + include/uapi/linux/vhost_types.h | 8 + 38 files changed, 4224 insertions(+), 248 deletions(-) delete mode 100644 drivers/vhost/Kconfig.vringh create mode 100644 drivers/vhost/iotlb.c create mode 100644 drivers/vhost/vdpa.c create mode 100644 drivers/virtio/vdpa/Kconfig create mode 100644 drivers/virtio/vdpa/Makefile create mode 100644 drivers/virtio/vdpa/ifcvf/Makefile create mode 100644 drivers/virtio/vdpa/ifcvf/ifcvf_base.c create mode 100644 drivers/virtio/vdpa/ifcvf/ifcvf_base.h create mode 100644 drivers/virtio/vdpa/ifcvf/ifcvf_main.c create mode 100644 drivers/virtio/vdpa/vdpa.c create mode 100644 drivers/virtio/vdpa/vdpa_sim/Makefile create mode 100644 drivers/virtio/vdpa/vdpa_sim/vdpa_sim.c create mode 100644 drivers/virtio/virtio_vdpa.c create mode 100644 include/linux/vdpa.h create mode 100644 include/linux/vhost_iotlb.h
Signature · 1 line
2.20.1
PATCH v8 1/9 vhost: refine vhost and vringh kconfig
Currently, CONFIG_VHOST depends on CONFIG_VIRTUALIZATION. But vhost is not necessarily for VM since it's a generic userspace and kernel communication protocol. Such dependency may prevent archs without virtualization support from using vhost.
To solve this, a dedicated vhost menu is created under drivers so CONIFG_VHOST can be decoupled out of CONFIG_VIRTUALIZATION.
While at it, also squash Kconfig.vringh into vhost Kconfig file. This avoids the trick of conditional inclusion from VOP or CAIF. Then it will be easier to introduce new vringh users and common dependency for both vringh and vhost.
--- arch/arm/kvm/Kconfig | 2 -- arch/arm64/kvm/Kconfig | 2 -- arch/mips/kvm/Kconfig | 2 -- arch/powerpc/kvm/Kconfig | 2 -- arch/s390/kvm/Kconfig | 4 ---- arch/x86/kvm/Kconfig | 4 ---- drivers/Kconfig | 2 ++ drivers/misc/mic/Kconfig | 4 ---- drivers/net/caif/Kconfig | 4 ---- drivers/vhost/Kconfig | 23 ++++++++++++++--------- drivers/vhost/Kconfig.vringh | 6 ------ 11 files changed, 16 insertions(+), 39 deletions(-) delete mode 100644 drivers/vhost/Kconfig.vringh
arch/arm/kvm/Kconfig +0 -2
@@ -54,6 +54,4 @@ config KVM_ARM_HOST ---help--- Provides host support for ARM processors. -source "drivers/vhost/Kconfig" - endif # VIRTUALIZATION
arch/arm64/kvm/Kconfig +0 -2
@@ -64,6 +64,4 @@ config KVM_ARM_PMU config KVM_INDIRECT_VECTORS def_bool KVM && (HARDEN_BRANCH_PREDICTOR || HARDEN_EL2_VECTORS) -source "drivers/vhost/Kconfig" - endif # VIRTUALIZATION
arch/mips/kvm/Kconfig +0 -2
@@ -72,6 +72,4 @@ config KVM_MIPS_DEBUG_COP0_COUNTERS If unsure, say N. -source "drivers/vhost/Kconfig" - endif # VIRTUALIZATION
arch/powerpc/kvm/Kconfig +0 -2
@@ -204,6 +204,4 @@ config KVM_XIVE default y depends on KVM_XICS && PPC_XIVE_NATIVE && KVM_BOOK3S_HV_POSSIBLE -source "drivers/vhost/Kconfig" - endif # VIRTUALIZATION
arch/s390/kvm/Kconfig +0 -4
@@ -55,8 +55,4 @@ config KVM_S390_UCONTROL If unsure, say N. -# OK, it's a little counter-intuitive to do this, but it puts it neatly under -# the virtualization menu. -source "drivers/vhost/Kconfig" - endif # VIRTUALIZATION
arch/x86/kvm/Kconfig +0 -4
@@ -94,8 +94,4 @@ config KVM_MMU_AUDIT This option adds a R/W kVM module parameter 'mmu_audit', which allows auditing of KVM MMU events at runtime. -# OK, it's a little counter-intuitive to do this, but it puts it neatly under -# the virtualization menu. -source "drivers/vhost/Kconfig" - endif # VIRTUALIZATION
drivers/Kconfig +2 -0
@@ -138,6 +138,8 @@ source "drivers/virt/Kconfig" source "drivers/virtio/Kconfig" +source "drivers/vhost/Kconfig" + source "drivers/hv/Kconfig" source "drivers/xen/Kconfig"
drivers/misc/mic/Kconfig +0 -4
@@ -133,8 +133,4 @@ config VOP OS and tools for MIC to use with this driver are available from <http://software.intel.com/en-us/mic-developer>. -if VOP -source "drivers/vhost/Kconfig.vringh" -endif - endmenu
drivers/net/caif/Kconfig +0 -4
@@ -58,8 +58,4 @@ config CAIF_VIRTIO ---help--- The CAIF driver for CAIF over Virtio. -if CAIF_VIRTIO -source "drivers/vhost/Kconfig.vringh" -endif - endif # CAIF_DRIVERS
drivers/vhost/Kconfig +14 -9
@@ -1,8 +1,20 @@ # SPDX-License-Identifier: GPL-2.0-only +config VHOST_RING + tristate + help + This option is selected by any driver which needs to access + the host side of a virtio ring. + +menuconfig VHOST + tristate "Host kernel accelerator for virtio (VHOST)" + help + This option is selected by any driver which needs to access + the core of vhost. +if VHOST + config VHOST_NET tristate "Host kernel accelerator for virtio net" depends on NET && EVENTFD && (TUN || !TUN) && (TAP || !TAP) - select VHOST ---help--- This kernel module can be loaded in host kernel to accelerate guest networking with virtio_net. Not to be confused with virtio_net
@@ -14,7 +26,6 @@ config VHOST_NET config VHOST_SCSI tristate "VHOST_SCSI TCM fabric driver" depends on TARGET_CORE && EVENTFD - select VHOST default n ---help--- Say M here to enable the vhost_scsi TCM fabric module
@@ -24,7 +35,6 @@ config VHOST_VSOCK tristate "vhost virtio-vsock driver" depends on VSOCKETS && EVENTFD select VIRTIO_VSOCKETS_COMMON - select VHOST default n ---help--- This kernel module can be loaded in the host kernel to provide AF_VSOCK
@@ -34,12 +44,6 @@ config VHOST_VSOCK To compile this driver as a module, choose M here: the module will be called vhost_vsock. -config VHOST - tristate - ---help--- - This option is selected by any driver which needs to access - the core of vhost. - config VHOST_CROSS_ENDIAN_LEGACY bool "Cross-endian support for vhost" default n
@@ -54,3 +58,4 @@ config VHOST_CROSS_ENDIAN_LEGACY adds some overhead, it is disabled by default. If unsure, say "N". +endif
drivers/vhost/Kconfig.vringh deleted +0 -6
@@ -1,6 +0,0 @@ -# SPDX-License-Identifier: GPL-2.0-only -config VHOST_RING - tristate - ---help--- - This option is selected by any driver which needs to access - the host side of a virtio ring.
Signature · 1 line
2.20.1
PATCH v8 2/9 vhost: allow per device message handler
In reply to Jason WangThis patch allow device to register its own message handler during vhost_dev_init(). vDPA device will use it to implement its own DMA mapping logic.
--- drivers/vhost/net.c | 3 ++- drivers/vhost/scsi.c | 2 +- drivers/vhost/vhost.c | 12 ++++++++++-- drivers/vhost/vhost.h | 6 +++++- drivers/vhost/vsock.c | 2 +- 5 files changed, 19 insertions(+), 6 deletions(-)
drivers/vhost/net.c +2 -1
@@ -1324,7 +1324,8 @@ static int vhost_net_open(struct inode *inode, struct file *f) }vhost_dev_init (dev, vqs,VHOST_NET_VQ_MAX ,UIO_MAXIOV +VHOST_NET_BATCH , -VHOST_NET_PKT_WEIGHT ,VHOST_NET_WEIGHT ); +VHOST_NET_PKT_WEIGHT ,VHOST_NET_WEIGHT , +NULL );vhost_poll_init (n->poll +VHOST_NET_VQ_TX , handle_tx_net,EPOLLOUT , dev);vhost_poll_init (n->poll +VHOST_NET_VQ_RX , handle_rx_net,EPOLLIN , dev);
drivers/vhost/scsi.c +1 -1
@@ -1628,7 +1628,7 @@ static int vhost_scsi_open(struct inode *inode, struct file *f) vs->vqs[i].vq.handle_kick = vhost_scsi_handle_kick; }vhost_dev_init (&vs->dev, vqs,VHOST_SCSI_MAX_VQ ,UIO_MAXIOV , -VHOST_SCSI_WEIGHT ,0 ); +VHOST_SCSI_WEIGHT ,0 ,NULL );vhost_scsi_init_inflight (vs,NULL );
drivers/vhost/vhost.c +10 -2
@@ -457,7 +457,9 @@ static size_t vhost_get_desc_size(struct vhost_virtqueue *vq,void vhost_dev_init (struct vhost_dev *dev,struct vhost_virtqueue **vqs,int nvqs, -int iov_limit,int weight,int byte_weight) +int iov_limit,int weight,int byte_weight, +int (*msg_handler)(struct vhost_dev *dev, +struct vhost_iotlb_msg *msg)) {struct vhost_virtqueue *vq;int i;
@@ -473,6 +475,7 @@ void vhost_dev_init(struct vhost_dev *dev, dev->iov_limit = iov_limit; dev->weight = weight; dev->byte_weight = byte_weight; + dev->msg_handler = msg_handler;init_llist_head (&dev->work_list);init_waitqueue_head (&dev->wait);INIT_LIST_HEAD (&dev->read_list);
@@ -1178,7 +1181,12 @@ ssize_t vhost_chr_write_iter(struct vhost_dev *dev, ret = -EINVAL ; goto done; } -if (vhost_process_iotlb_msg (dev, &msg)) { + +if (dev->msg_handler) + ret = dev->msg_handler (dev, &msg); +else + ret =vhost_process_iotlb_msg (dev, &msg); +if (ret) { ret = -EFAULT ; goto done; }
drivers/vhost/vhost.h +5 -1
@@ -174,11 +174,15 @@ struct vhost_dev {int weight;int byte_weight;u64 kcov_handle; +int (*msg_handler)(struct vhost_dev *dev, +struct vhost_iotlb_msg *msg); };bool vhost_exceeds_weight (struct vhost_virtqueue *vq,int pkts,int total_len);void vhost_dev_init (struct vhost_dev *,struct vhost_virtqueue **vqs, -int nvqs,int iov_limit,int weight,int byte_weight); +int nvqs,int iov_limit,int weight,int byte_weight, +int (*msg_handler)(struct vhost_dev *dev, +struct vhost_iotlb_msg *msg));long vhost_dev_set_owner (struct vhost_dev *dev);bool vhost_dev_has_owner (struct vhost_dev *dev);long vhost_dev_check_owner (struct vhost_dev *);
drivers/vhost/vsock.c +1 -1
@@ -621,7 +621,7 @@ static int vhost_vsock_dev_open(struct inode *inode, struct file *file)vhost_dev_init (&vsock->dev, vqs,ARRAY_SIZE (vsock->vqs),UIO_MAXIOV ,VHOST_VSOCK_PKT_WEIGHT , -VHOST_VSOCK_WEIGHT ); +VHOST_VSOCK_WEIGHT ,NULL ); file->private_data = vsock;spin_lock_init (&vsock->send_pkt_list_lock);
Signature · 1 line
2.20.1
PATCH v8 3/9 vhost: factor out IOTLB
In reply to Jason WangThis patch factors out IOTLB into a dedicated module in order to be reused by other modules like vringh. User may choose to enable the automatic retiring by specifying VHOST_IOTLB_FLAG_RETIRE flag to fit for the case of vhost device IOTLB implementation.
--- MAINTAINERS | 1 + drivers/vhost/Kconfig | 9 ++ drivers/vhost/Makefile | 3 + drivers/vhost/iotlb.c | 177 +++++++++++++++++++++++++++++ drivers/vhost/net.c | 2 +- drivers/vhost/vhost.c | 221 +++++++++++------------------------- drivers/vhost/vhost.h | 39 ++----- include/linux/vhost_iotlb.h | 47 ++++++++ 8 files changed, 318 insertions(+), 181 deletions(-) create mode 100644 drivers/vhost/iotlb.c create mode 100644 include/linux/vhost_iotlb.h
MAINTAINERS +1 -0
@@ -17768,6 +17768,7 @@ T: git git://git.kernel.org/pub/scm/linux/kernel/git/mst/vhost.git S: Maintained F: drivers/vhost/ F: include/uapi/linux/vhost.h +F: include/linux/vhost_iotlb.h VIRTIO INPUT DRIVER M: Gerd Hoffmann <[email protected]>
drivers/vhost/Kconfig +9 -0
@@ -1,4 +1,9 @@ # SPDX-License-Identifier: GPL-2.0-only +config VHOST_IOTLB + tristate + help + Generic IOTLB implementation for vhost and vringh. + config VHOST_RING tristate help
@@ -7,9 +12,11 @@ config VHOST_RING menuconfig VHOST tristate "Host kernel accelerator for virtio (VHOST)" + select VHOST_IOTLB help This option is selected by any driver which needs to access the core of vhost. + if VHOST config VHOST_NET
@@ -58,4 +65,6 @@ config VHOST_CROSS_ENDIAN_LEGACY adds some overhead, it is disabled by default. If unsure, say "N". + endif +
drivers/vhost/Makefile +3 -0
@@ -11,3 +11,6 @@ vhost_vsock-y := vsock.o obj-$(CONFIG_VHOST_RING) += vringh.o obj-$(CONFIG_VHOST) += vhost.o + +obj-$(CONFIG_VHOST_IOTLB) += vhost_iotlb.o +vhost_iotlb-y := iotlb.o
drivers/vhost/iotlb.c new file +177 -0
177 lines not on this page · raw message
drivers/vhost/net.c +1 -1
@@ -1595,7 +1595,7 @@ static long vhost_net_reset_owner(struct vhost_net *n)struct socket *tx_sock =NULL ;struct socket *rx_sock =NULL ;long err; -struct vhost_umem *umem; +struct vhost_iotlb *umem;mutex_lock (&n->dev.mutex); err =vhost_dev_check_owner (&n->dev);
drivers/vhost/vhost.c +69 -152
432 lines not on this page · raw message
drivers/vhost/vhost.h +11 -28
@@ -12,6 +12,7 @@#include <linux/virtio_config.h> #include <linux/virtio_ring.h> #include <linux/atomic.h> +#include <linux/vhost_iotlb.h> struct vhost_work ;typedef void (*vhost_work_fn_t )(struct vhost_work *work);
@@ -52,27 +53,6 @@ struct vhost_log {u64 len; }; -#define START (node) ((node)->start) -#define LAST (node) ((node)->last) - -struct vhost_umem_node { -struct rb_node rb; -struct list_head link; -__u64 start; -__u64 last; -__u64 size; -__u64 userspace_addr; -__u32 perm; -__u32 flags_padding; -__u64 __subtree_last; -}; - -struct vhost_umem { -struct rb_root_cached umem_tree; -struct list_head umem_list; -int numem; -}; -enum vhost_uaddr_type {VHOST_ADDR_DESC =0 ,VHOST_ADDR_AVAIL =1 ,
@@ -90,7 +70,7 @@ struct vhost_virtqueue {struct vring_desc __user *desc;struct vring_avail __user *avail;struct vring_used __user *used; -const struct vhost_umem_node *meta_iotlb[VHOST_NUM_ADDRS ]; +const struct vhost_iotlb_map *meta_iotlb[VHOST_NUM_ADDRS ];struct file *kick;struct eventfd_ctx *call_ctx;struct eventfd_ctx *error_ctx;
@@ -128,8 +108,8 @@ struct vhost_virtqueue {struct iovec *indirect;struct vring_used_elem *heads;/* Protected by virtqueue mutex. */ -struct vhost_umem *umem; -struct vhost_umem *iotlb; +struct vhost_iotlb *umem; +struct vhost_iotlb *iotlb;void *private_data;u64 acked_features;u64 acked_backend_features;
@@ -164,8 +144,8 @@ struct vhost_dev {struct eventfd_ctx *log_ctx;struct llist_head work_list;struct task_struct *worker; -struct vhost_umem *umem; -struct vhost_umem *iotlb; +struct vhost_iotlb *umem; +struct vhost_iotlb *iotlb;spinlock_t iotlb_lock;struct list_head read_list;struct list_head pending_list;
@@ -186,8 +166,8 @@ void vhost_dev_init(struct vhost_dev *, struct vhost_virtqueue **vqs,long vhost_dev_set_owner (struct vhost_dev *dev);bool vhost_dev_has_owner (struct vhost_dev *dev);long vhost_dev_check_owner (struct vhost_dev *); -struct vhost_umem *vhost_dev_reset_owner_prepare (void ); -void vhost_dev_reset_owner (struct vhost_dev *,struct vhost_umem *); +struct vhost_iotlb *vhost_dev_reset_owner_prepare (void ); +void vhost_dev_reset_owner (struct vhost_dev *dev,struct vhost_iotlb *iotlb);void vhost_dev_cleanup (struct vhost_dev *);void vhost_dev_stop (struct vhost_dev *);long vhost_dev_ioctl (struct vhost_dev *,unsigned int ioctl,void __user *argp);
@@ -233,6 +213,9 @@ ssize_t vhost_chr_write_iter(struct vhost_dev *dev, struct iov_iter *from);int vhost_init_device_iotlb (struct vhost_dev *d,bool enabled); +void vhost_iotlb_map_free (struct vhost_iotlb *iotlb, +struct vhost_iotlb_map *map); +#define vq_err (vq, fmt, ...) do { \ pr_debug(pr_fmt(fmt), ##__VA_ARGS__); \ if ((vq)->error_ctx) \
include/linux/vhost_iotlb.h new file +47 -0
47 lines not on this page · raw message
Signature · 1 line
2.20.1
PATCH v8 4/9 vringh: IOTLB support
In reply to Jason WangThis patch implements the third memory accessor for vringh besides current kernel and userspace accessors. This idea is to allow vringh to do the address translation through an IOTLB which is implemented via vhost_map interval tree. Users should setup and IOVA to PA mapping in this IOTLB.
This allows us to:
- Use vringh to access virtqueues with vIOMMU
- Use vringh to implement software virtqueues for vDPA devices
--- drivers/vhost/Kconfig | 1 + drivers/vhost/vringh.c | 421 ++++++++++++++++++++++++++++++++++++++--- include/linux/vringh.h | 36 ++++ 3 files changed, 435 insertions(+), 23 deletions(-)
drivers/vhost/Kconfig +1 -0
@@ -6,6 +6,7 @@ config VHOST_IOTLB config VHOST_RING tristate + select VHOST_IOTLB help This option is selected by any driver which needs to access the host side of a virtio ring.
drivers/vhost/vringh.c +398 -23
557 lines not on this page · raw message
include/linux/vringh.h +36 -0
@@ -14,6 +14,8 @@#include <linux/virtio_byteorder.h> #include <linux/uio.h> #include <linux/slab.h> +#include <linux/dma-direction.h> +#include <linux/vhost_iotlb.h> #include <asm/barrier.h> /* virtio_ring with information needed for host access. */
@@ -39,6 +41,9 @@ struct vringh {/* The vring (note: it may contain user pointers!) */ struct vring vring; +/* IOTLB for this vring */ +struct vhost_iotlb *iotlb; +/* The function to call to notify the guest about added buffers */ void (*notify)(struct vringh *); };
@@ -248,4 +253,35 @@ static inline __virtio64 cpu_to_vringh64(const struct vringh *vrh, u64 val) {return __cpu_to_virtio64 (vringh_is_little_endian (vrh), val); } + +void vringh_set_iotlb (struct vringh *vrh,struct vhost_iotlb *iotlb); + +int vringh_init_iotlb (struct vringh *vrh,u64 features, +unsigned int num,bool weak_barriers, +struct vring_desc *desc, +struct vring_avail *avail, +struct vring_used *used); + +int vringh_getdesc_iotlb (struct vringh *vrh, +struct vringh_kiov *riov, +struct vringh_kiov *wiov, +u16 *head, +gfp_t gfp); + +ssize_t vringh_iov_pull_iotlb (struct vringh *vrh, +struct vringh_kiov *riov, +void *dst,size_t len); +ssize_t vringh_iov_push_iotlb (struct vringh *vrh, +struct vringh_kiov *wiov, +const void *src,size_t len); + +void vringh_abandon_iotlb (struct vringh *vrh,unsigned int num); + +int vringh_complete_iotlb (struct vringh *vrh,u16 head,u32 len); + +bool vringh_notify_enable_iotlb (struct vringh *vrh); +void vringh_notify_disable_iotlb (struct vringh *vrh); + +int vringh_need_notify_iotlb (struct vringh *vrh); +#endif /* _LINUX_VRINGH_H */
Signature · 1 line
2.20.1
PATCH v8 5/9 vDPA: introduce vDPA bus
In reply to Jason WangvDPA device is a device that uses a datapath which complies with the virtio specifications with vendor specific control path. vDPA devices can be both physically located on the hardware or emulated by software. vDPA hardware devices are usually implemented through PCIE with the following types:
- PF (Physical Function) - A single Physical Function
- VF (Virtual Function) - Device that supports single root I/O virtualization (SR-IOV). Its Virtual Function (VF) represents a virtualized instance of the device that can be assigned to different partitions
- ADI (Assignable Device Interface) and its equivalents - With technologies such as Intel Scalable IOV, a virtual device (VDEV) composed by host OS utilizing one or more ADIs. Or its equivalent like SF (Sub function) from Mellanox.
From a driver's perspective, depends on how and where the DMA translation is done, vDPA devices are split into two types:
- Platform specific DMA translation - From the driver's perspective, the device can be used on a platform where device access to data in memory is limited and/or translated. An example is a PCIE vDPA whose DMA request was tagged via a bus (e.g PCIE) specific way. DMA translation and protection are done at PCIE bus IOMMU level.
- Device specific DMA translation - The device implements DMA isolation and protection through its own logic. An example is a vDPA device which uses on-chip IOMMU.
To hide the differences and complexity of the above types for a vDPA device/IOMMU options and in order to present a generic virtio device to the upper layer, a device agnostic framework is required.
This patch introduces a software vDPA bus which abstracts the common attributes of vDPA device, vDPA bus driver and the communication method (vdpa_config_ops) between the vDPA device abstraction and the vDPA bus driver. This allows multiple types of drivers to be used for vDPA device like the virtio_vdpa and vhost_vdpa driver to operate on the bus and allow vDPA device could be used by either kernel virtio driver or userspace vhost drivers as:
virtio drivers vhost drivers
| |
[virtio bus] [vhost uAPI]
| |
virtio device vhost device
virtio_vdpa drv vhost_vdpa drv
\ /
[vDPA bus]
|
vDPA device
hardware drv
|
[hardware bus]
|
vDPA hardwareWith the abstraction of vDPA bus and vDPA bus operations, the difference and complexity of the under layer hardware is hidden from upper layer. The vDPA bus drivers on top can use a unified vdpa_config_ops to control different types of vDPA device.
--- MAINTAINERS | 1 + drivers/virtio/Kconfig | 2 + drivers/virtio/Makefile | 1 + drivers/virtio/vdpa/Kconfig | 7 + drivers/virtio/vdpa/Makefile | 2 + drivers/virtio/vdpa/vdpa.c | 180 +++++++++++++++++++++++++ include/linux/vdpa.h | 247 +++++++++++++++++++++++++++++++++++ 7 files changed, 440 insertions(+) create mode 100644 drivers/virtio/vdpa/Kconfig create mode 100644 drivers/virtio/vdpa/Makefile create mode 100644 drivers/virtio/vdpa/vdpa.c create mode 100644 include/linux/vdpa.h
MAINTAINERS +1 -0
@@ -17701,6 +17701,7 @@ F: tools/virtio/ F: drivers/net/virtio_net.c F: drivers/block/virtio_blk.c F: include/linux/virtio*.h +F: include/linux/vdpa.h F: include/uapi/linux/virtio_*.h F: drivers/crypto/virtio/ F: mm/balloon_compaction.c
drivers/virtio/Kconfig +2 -0
@@ -96,3 +96,5 @@ config VIRTIO_MMIO_CMDLINE_DEVICES If unsure, say 'N'. endif # VIRTIO_MENU + +source "drivers/virtio/vdpa/Kconfig"
drivers/virtio/Makefile +1 -0
@@ -6,3 +6,4 @@ virtio_pci-y := virtio_pci_modern.o virtio_pci_common.o virtio_pci-$(CONFIG_VIRTIO_PCI_LEGACY) += virtio_pci_legacy.o obj-$(CONFIG_VIRTIO_BALLOON) += virtio_balloon.o obj-$(CONFIG_VIRTIO_INPUT) += virtio_input.o +obj-$(CONFIG_VDPA) += vdpa/
drivers/virtio/vdpa/Kconfig new file +7 -0
@@ -0,0 +1,7 @@ +# SPDX-License-Identifier: GPL-2.0-only +config VDPA + tristate + help + Enable this module to support vDPA device that uses a + datapath which complies with virtio specifications with + vendor specific control path.
drivers/virtio/vdpa/Makefile new file +2 -0
@@ -0,0 +1,2 @@ +# SPDX-License-Identifier: GPL-2.0 +obj-$(CONFIG_VDPA) += vdpa.o
drivers/virtio/vdpa/vdpa.c new file +180 -0
180 lines not on this page · raw message
include/linux/vdpa.h new file +247 -0
247 lines not on this page · raw message
Signature · 1 line
2.20.1
Re: [PATCH V8 5/9] vDPA: introduce vDPA bus
include/linux/vdpa.h line 197
Quoted 11 lines · Jason Wang
include/linux/vdpa.h new file +11
-0+struct vdpa_device *__vdpa_alloc_device (struct device *parent, +const struct vdpa_config_ops *config, +size_t size); + +#define vdpa_alloc_device (dev_struct, member, parent, config) \ + container_of(__vdpa_alloc_device( \ + parent, config, \ + sizeof(struct dev_struct) + \ + BUILD_BUG_ON_ZERO(offsetof( \ + struct dev_struct, member))), \ + struct dev_struct, member)
This all looks robust now, one minor remark is to not do 'struct dev_struct' here so the caller has to do
vdpa_alloc_device(struct foo, vpda, ...)
Which suggests to the reader something unusual is happening
Jason
Re: [PATCH V8 5/9] vDPA: introduce vDPA bus
Quoted 15 lines · Jason Gunthorpe
include/linux/vdpa.h line 197
Quoted 11 lines · Jason Wang
include/linux/vdpa.h new file +11
-0+struct vdpa_device *__vdpa_alloc_device (struct device *parent, +const struct vdpa_config_ops *config, +size_t size); + +#define vdpa_alloc_device (dev_struct, member, parent, config) \ + container_of(__vdpa_alloc_device( \ + parent, config, \ + sizeof(struct dev_struct) + \ + BUILD_BUG_ON_ZERO(offsetof( \ + struct dev_struct, member))), \ + struct dev_struct, member)This all looks robust now, one minor remark is to not do 'struct
dev_struct' here so the caller has to do
Quoted 5 lines
vdpa_alloc_device(struct foo, vpda, ...)Which suggests to the reader something unusual is happening
Jason
Yes, that's better.
But my understanding is that this can be done on top.
Thanks
PATCH v8 6/9 virtio: introduce a vDPA based transport
In reply to Jason WangThis patch introduces a vDPA transport for virtio. This is used to use kernel virtio driver to drive the vDPA device that is capable of populating virtqueue directly.
A new virtio-vdpa driver will be registered to the vDPA bus, when a new virtio-vdpa device is probed, it will register the device with vdpa based config ops. This means it is a software transport between vDPA driver and vDPA device. The transport was implemented through bus_ops of vDPA parent.
--- drivers/virtio/Kconfig | 13 ++ drivers/virtio/Makefile | 1 + drivers/virtio/virtio_vdpa.c | 396 +++++++++++++++++++++++++++++++++++ 3 files changed, 410 insertions(+) create mode 100644 drivers/virtio/virtio_vdpa.c
drivers/virtio/Kconfig +13 -0
@@ -43,6 +43,19 @@ config VIRTIO_PCI_LEGACY If unsure, say Y. +config VIRTIO_VDPA + tristate "vDPA driver for virtio devices" + select VDPA + select VIRTIO + help + This driver provides support for virtio based paravirtual + device driver over vDPA bus. For this to be useful, you need + an appropriate vDPA device implementation that operates on a + physical device to allow the datapath of virtio to be + offloaded to hardware. + + If unsure, say M. + config VIRTIO_PMEM tristate "Support for virtio pmem driver" depends on VIRTIO
drivers/virtio/Makefile +1 -0
@@ -6,4 +6,5 @@ virtio_pci-y := virtio_pci_modern.o virtio_pci_common.o virtio_pci-$(CONFIG_VIRTIO_PCI_LEGACY) += virtio_pci_legacy.o obj-$(CONFIG_VIRTIO_BALLOON) += virtio_balloon.o obj-$(CONFIG_VIRTIO_INPUT) += virtio_input.o +obj-$(CONFIG_VIRTIO_VDPA) += virtio_vdpa.o obj-$(CONFIG_VDPA) += vdpa/
drivers/virtio/virtio_vdpa.c new file +396 -0
396 lines not on this page · raw message
Signature · 1 line
2.20.1
PATCH v8 7/9 vhost: introduce vDPA-based backend
In reply to Jason WangFrom: Tiwei Bie <[email protected]>
This patch introduces a vDPA-based vhost backend. This backend is built on top of the same interface defined in virtio-vDPA and provides a generic vhost interface for userspace to accelerate the virtio devices in guest.
This backend is implemented as a vDPA device driver on top of the same ops used in virtio-vDPA. It will create char device entry named vhost-vdpa-$index for userspace to use. Userspace can use vhost ioctls on top of this char device to setup the backend.
Vhost ioctls are extended to make it type agnostic and behave like a virtio device, this help to eliminate type specific API like what vhost_net/scsi/vsock did:
- VHOST_VDPA_GET_DEVICE_ID: get the virtio device ID which is defined by virtio specification to differ from different type of devices
- VHOST_VDPA_GET_VRING_NUM: get the maximum size of virtqueue supported by the vDPA device
- VHSOT_VDPA_SET/GET_STATUS: set and get virtio status of vDPA device
- VHOST_VDPA_SET/GET_CONFIG: access virtio config space
- VHOST_VDPA_SET_VRING_ENABLE: enable a specific virtqueue
For memory mapping, IOTLB API is mandated for vhost-vDPA which means userspace drivers are required to use VHOST_IOTLB_UPDATE/VHOST_IOTLB_INVALIDATE to add or remove mapping for a specific userspace memory region.
The vhost-vDPA API is designed to be type agnostic, but it allows net device only in current stage. Due to the lacking of control virtqueue support, some features were filter out by vhost-vdpa.
We will enable more features and devices in the near future.
--- drivers/vhost/Kconfig | 11 + drivers/vhost/Makefile | 3 + drivers/vhost/vdpa.c | 883 +++++++++++++++++++++++++++++++ include/uapi/linux/vhost.h | 24 + include/uapi/linux/vhost_types.h | 8 + 5 files changed, 929 insertions(+) create mode 100644 drivers/vhost/vdpa.c
drivers/vhost/Kconfig +11 -0
@@ -52,6 +52,17 @@ config VHOST_VSOCK To compile this driver as a module, choose M here: the module will be called vhost_vsock. +config VHOST_VDPA + tristate "Vhost driver for vDPA-based backend" + depends on EVENTFD + select VDPA + help + This kernel module can be loaded in host kernel to accelerate + guest virtio devices with the vDPA-based backends. + + To compile this driver as a module, choose M here: the module + will be called vhost_vdpa. + config VHOST_CROSS_ENDIAN_LEGACY bool "Cross-endian support for vhost" default n
drivers/vhost/Makefile +3 -0
@@ -10,6 +10,9 @@ vhost_vsock-y := vsock.o obj-$(CONFIG_VHOST_RING) += vringh.o +obj-$(CONFIG_VHOST_VDPA) += vhost_vdpa.o +vhost_vdpa-y := vdpa.o + obj-$(CONFIG_VHOST) += vhost.o obj-$(CONFIG_VHOST_IOTLB) += vhost_iotlb.o
drivers/vhost/vdpa.c new file +883 -0
883 lines not on this page · raw message
include/uapi/linux/vhost.h +24 -0
@@ -116,4 +116,28 @@#define VHOST_VSOCK_SET_GUEST_CID _IOW(VHOST_VIRTIO, 0x60, __u64)#define VHOST_VSOCK_SET_RUNNING _IOW(VHOST_VIRTIO, 0x61, int) +/* VHOST_VDPA specific defines */ + +/* Get the device id. The device ids follow the same definition of +* the device id defined in virtio-spec. +*/ +#define VHOST_VDPA_GET_DEVICE_ID _IOR(VHOST_VIRTIO, 0x70, __u32) +/* Get and set the status. The status bits follow the same definition +* of the device status defined in virtio-spec. +*/ +#define VHOST_VDPA_GET_STATUS _IOR(VHOST_VIRTIO, 0x71, __u8) +#define VHOST_VDPA_SET_STATUS _IOW(VHOST_VIRTIO, 0x72, __u8) +/* Get and set the device config. The device config follows the same +* definition of the device config defined in virtio-spec. +*/ +#define VHOST_VDPA_GET_CONFIG _IOR(VHOST_VIRTIO, 0x73, \ + struct vhost_vdpa_config) +#define VHOST_VDPA_SET_CONFIG _IOW(VHOST_VIRTIO, 0x74, \ + struct vhost_vdpa_config) +/* Enable/disable the ring. */ +#define VHOST_VDPA_SET_VRING_ENABLE _IOW(VHOST_VIRTIO, 0x75, \ + struct vhost_vring_state) +/* Get the max ring size. */ +#define VHOST_VDPA_GET_VRING_NUM _IOR(VHOST_VIRTIO, 0x76, __u16) +#endif
include/uapi/linux/vhost_types.h +8 -0
@@ -119,6 +119,14 @@ struct vhost_scsi_target {unsigned short reserved; }; +/* VHOST_VDPA specific definitions */ + +struct vhost_vdpa_config { +__u32 off; +__u32 len; +__u8 buf[0 ]; +}; +/* Feature bits */ /* Log all write descriptors. Can be changed while device is active. */ #define VHOST_F_LOG_ALL 26
Signature · 1 line
2.20.1
PATCH v8 8/9 vdpasim: vDPA device simulator
In reply to Jason WangThis patch implements a software vDPA networking device. The datapath is implemented through vringh and workqueue. The device has an on-chip IOMMU which translates IOVA to PA. For kernel virtio drivers, vDPA simulator driver provides dma_ops. For vhost driers, set_map() methods of vdpa_config_ops is implemented to accept mappings from vhost.
Currently, vDPA device simulator will loopback TX traffic to RX. So the main use case for the device is vDPA feature testing, prototyping and development.
Note, there's no management API implemented, a vDPA device will be registered once the module is probed. We need to handle this in the future development.
--- drivers/virtio/vdpa/Kconfig | 19 + drivers/virtio/vdpa/Makefile | 1 + drivers/virtio/vdpa/vdpa_sim/Makefile | 2 + drivers/virtio/vdpa/vdpa_sim/vdpa_sim.c | 629 ++++++++++++++++++++++++ 4 files changed, 651 insertions(+) create mode 100644 drivers/virtio/vdpa/vdpa_sim/Makefile create mode 100644 drivers/virtio/vdpa/vdpa_sim/vdpa_sim.c
drivers/virtio/vdpa/Kconfig +19 -0
@@ -5,3 +5,22 @@ config VDPA Enable this module to support vDPA device that uses a datapath which complies with virtio specifications with vendor specific control path. + +menuconfig VDPA_MENU + bool "VDPA drivers" + default n + +if VDPA_MENU + +config VDPA_SIM + tristate "vDPA device simulator" + depends on RUNTIME_TESTING_MENU + select VDPA + select VHOST_RING + default n + help + vDPA networking device simulator which loop TX traffic back + to RX. This device is used for testing, prototyping and + development of vDPA. + +endif # VDPA_MENU
drivers/virtio/vdpa/Makefile +1 -0
@@ -1,2 +1,3 @@ # SPDX-License-Identifier: GPL-2.0 obj-$(CONFIG_VDPA) += vdpa.o +obj-$(CONFIG_VDPA_SIM) += vdpa_sim/
drivers/virtio/vdpa/vdpa_sim/Makefile new file +2 -0
@@ -0,0 +1,2 @@ +# SPDX-License-Identifier: GPL-2.0 +obj-$(CONFIG_VDPA_SIM) += vdpa_sim.o
drivers/virtio/vdpa/vdpa_sim/vdpa_sim.c new file +629 -0
629 lines not on this page · raw message
Signature · 1 line
2.20.1
PATCH v8 9/9 virtio: Intel IFC VF driver for VDPA
In reply to Jason WangFrom: Zhu Lingshan <[email protected]>
This commit introduced two layers to drive IFC VF:
(1) ifcvf_base layer, which handles IFC VF NIC hardware operations and configurations.
(2) ifcvf_main layer, which complies to VDPA bus framework, implemented device operations for VDPA bus, handles device probe, bus attaching, vring operations, etc.
--- drivers/virtio/vdpa/Kconfig | 11 + drivers/virtio/vdpa/Makefile | 1 + drivers/virtio/vdpa/ifcvf/Makefile | 3 + drivers/virtio/vdpa/ifcvf/ifcvf_base.c | 386 ++++++++++++++++++++ drivers/virtio/vdpa/ifcvf/ifcvf_base.h | 132 +++++++ drivers/virtio/vdpa/ifcvf/ifcvf_main.c | 474 +++++++++++++++++++++++++ 6 files changed, 1007 insertions(+) create mode 100644 drivers/virtio/vdpa/ifcvf/Makefile create mode 100644 drivers/virtio/vdpa/ifcvf/ifcvf_base.c create mode 100644 drivers/virtio/vdpa/ifcvf/ifcvf_base.h create mode 100644 drivers/virtio/vdpa/ifcvf/ifcvf_main.c
drivers/virtio/vdpa/Kconfig +11 -0
@@ -23,4 +23,15 @@ config VDPA_SIM to RX. This device is used for testing, prototyping and development of vDPA. +config IFCVF + tristate "Intel IFC VF VDPA driver" + depends on PCI_MSI + select VDPA + default n + help + This kernel module can drive Intel IFC VF NIC to offload + virtio dataplane traffic to hardware. + To compile this driver as a module, choose M here: the module will + be called ifcvf. + endif # VDPA_MENU
drivers/virtio/vdpa/Makefile +1 -0
@@ -1,3 +1,4 @@ # SPDX-License-Identifier: GPL-2.0 obj-$(CONFIG_VDPA) += vdpa.o obj-$(CONFIG_VDPA_SIM) += vdpa_sim/ +obj-$(CONFIG_IFCVF) += ifcvf/
drivers/virtio/vdpa/ifcvf/Makefile new file +3 -0
@@ -0,0 +1,3 @@ +# SPDX-License-Identifier: GPL-2.0 +obj-$(CONFIG_IFCVF) += ifcvf.o +ifcvf-$(CONFIG_IFCVF) += ifcvf_main.o ifcvf_base.o
drivers/virtio/vdpa/ifcvf/ifcvf_base.c new file +386 -0
386 lines not on this page · raw message
drivers/virtio/vdpa/ifcvf/ifcvf_base.h new file +132 -0
132 lines not on this page · raw message
drivers/virtio/vdpa/ifcvf/ifcvf_main.c new file +474 -0
474 lines not on this page · raw message
Signature · 1 line
2.20.1
Re: [PATCH V8 9/9] virtio: Intel IFC VF driver for VDPA
Quoted 86 lines · Jason Wang
drivers/virtio/vdpa/ifcvf/ifcvf_main.c new file +86
-0+static int ifcvf_probe (struct pci_dev *pdev,const struct pci_device_id *id) +{ +struct device *dev = &pdev->dev; +struct ifcvf_adapter *adapter; +struct ifcvf_hw *vf; +int ret, i; + + ret =pci_enable_device (pdev); +if (ret) { +IFCVF_ERR (&pdev->dev,"Failed to enable device\n" ); + goto err_enable; + } + + ret =pci_request_regions (pdev,IFCVF_DRIVER_NAME ); +if (ret) { +IFCVF_ERR (&pdev->dev,"Failed to request MMIO region\n" ); + goto err_regions; + } + + ret =pci_alloc_irq_vectors (pdev,IFCVF_MAX_INTR , +IFCVF_MAX_INTR ,PCI_IRQ_MSIX ); +if (ret <0 ) { +IFCVF_ERR (&pdev->dev,"Failed to alloc irq vectors\n" ); + goto err_vectors; + } + + adapter =vdpa_alloc_device (ifcvf_adapter, vdpa, dev, &ifc_vdpa_ops); +if (adapter ==NULL ) { +IFCVF_ERR (&pdev->dev,"Failed to allocate vDPA structure" ); + ret = -ENOMEM ; + goto err_alloc; + } + + adapter->dev = dev; +pci_set_master (pdev); +pci_set_drvdata (pdev, adapter); + + ret =ifcvf_request_irq (adapter); +if (ret) { +IFCVF_ERR (&pdev->dev,"Failed to request MSI-X irq\n" ); + goto err_msix; + } + + vf = &adapter->vf; +for (i =0 ; i <IFCVF_PCI_MAX_RESOURCE ; i++) { + vf->mem_resource[i].phys_addr =pci_resource_start (pdev, i); + vf->mem_resource[i].len =pci_resource_len (pdev, i); +if (!vf->mem_resource[i].len) +continue ; + + vf->mem_resource[i].addr =pci_iomap_range (pdev, i,0 , + vf->mem_resource[i].len); +if (!vf->mem_resource[i].addr) { +IFCVF_ERR (&pdev->dev, +"Failed to map IO resource %d\n" , i); + ret = -EINVAL ; + goto err_msix; + } + } + +if (ifcvf_init_hw (vf, pdev) <0 ) { + ret = -EINVAL ; + goto err_msix; + } + + 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) { +IFCVF_ERR (adapter->dev,"No usable DMA confiugration\n" ); + ret = -EINVAL ; + goto err_msix; + } + + adapter->vdpa.dma_dev = dev; + ret =vdpa_register_device (&adapter->vdpa); +if (ret) { +IFCVF_ERR (adapter->dev,"Failed to register ifcvf to vdpa bus" ); + goto err_msix; + } + +return 0 ; + +err_msix: +put_device (&adapter->vdpa.dev);
drivers/virtio/vdpa/ifcvf/ifcvf_main.c line 446
Quoted 10 lines
drivers/virtio/vdpa/ifcvf/ifcvf_main.c new file +10
-0+return ret; +err_alloc: +pci_free_irq_vectors (pdev); +err_vectors: +pci_release_regions (pdev); +err_regions: +pci_disable_device (pdev); +err_enable: +return ret; +}
I personally don't like seeing goto unwinds with multiple returns, and here I think it is actually a tiny bug.
All touches to the PCI device must stop before the driver core remove() returns - so these pci function cannot be in the kref put release function anyhow.
Suggest to use the devm versions of the above instead, and then you can reorder things so the allocation is done first.
Jason
Re: [PATCH V8 9/9] virtio: Intel IFC VF driver for VDPA
Quoted 100 lines · Jason Gunthorpe
Quoted 86 lines · Jason Wang
drivers/virtio/vdpa/ifcvf/ifcvf_main.c new file +86
-0+static int ifcvf_probe (struct pci_dev *pdev,const struct pci_device_id *id) +{ +struct device *dev = &pdev->dev; +struct ifcvf_adapter *adapter; +struct ifcvf_hw *vf; +int ret, i; + + ret =pci_enable_device (pdev); +if (ret) { +IFCVF_ERR (&pdev->dev,"Failed to enable device\n" ); + goto err_enable; + } + + ret =pci_request_regions (pdev,IFCVF_DRIVER_NAME ); +if (ret) { +IFCVF_ERR (&pdev->dev,"Failed to request MMIO region\n" ); + goto err_regions; + } + + ret =pci_alloc_irq_vectors (pdev,IFCVF_MAX_INTR , +IFCVF_MAX_INTR ,PCI_IRQ_MSIX ); +if (ret <0 ) { +IFCVF_ERR (&pdev->dev,"Failed to alloc irq vectors\n" ); + goto err_vectors; + } + + adapter =vdpa_alloc_device (ifcvf_adapter, vdpa, dev, &ifc_vdpa_ops); +if (adapter ==NULL ) { +IFCVF_ERR (&pdev->dev,"Failed to allocate vDPA structure" ); + ret = -ENOMEM ; + goto err_alloc; + } + + adapter->dev = dev; +pci_set_master (pdev); +pci_set_drvdata (pdev, adapter); + + ret =ifcvf_request_irq (adapter); +if (ret) { +IFCVF_ERR (&pdev->dev,"Failed to request MSI-X irq\n" ); + goto err_msix; + } + + vf = &adapter->vf; +for (i =0 ; i <IFCVF_PCI_MAX_RESOURCE ; i++) { + vf->mem_resource[i].phys_addr =pci_resource_start (pdev, i); + vf->mem_resource[i].len =pci_resource_len (pdev, i); +if (!vf->mem_resource[i].len) +continue ; + + vf->mem_resource[i].addr =pci_iomap_range (pdev, i,0 , + vf->mem_resource[i].len); +if (!vf->mem_resource[i].addr) { +IFCVF_ERR (&pdev->dev, +"Failed to map IO resource %d\n" , i); + ret = -EINVAL ; + goto err_msix; + } + } + +if (ifcvf_init_hw (vf, pdev) <0 ) { + ret = -EINVAL ; + goto err_msix; + } + + 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) { +IFCVF_ERR (adapter->dev,"No usable DMA confiugration\n" ); + ret = -EINVAL ; + goto err_msix; + } + + adapter->vdpa.dma_dev = dev; + ret =vdpa_register_device (&adapter->vdpa); +if (ret) { +IFCVF_ERR (adapter->dev,"Failed to register ifcvf to vdpa bus" ); + goto err_msix; + } + +return 0 ; + +err_msix: +put_device (&adapter->vdpa.dev);drivers/virtio/vdpa/ifcvf/ifcvf_main.c line 446
Quoted 10 lines
drivers/virtio/vdpa/ifcvf/ifcvf_main.c new file +10
-0+return ret; +err_alloc: +pci_free_irq_vectors (pdev); +err_vectors: +pci_release_regions (pdev); +err_regions: +pci_disable_device (pdev); +err_enable: +return ret; +}I personally don't like seeing goto unwinds with multiple returns, and here I think it is actually a tiny bug.
Quoted 3 lines
All touches to the PCI device must stop before the driver core
remove()returns - so these pci function cannot be in the kref put release function anyhow.
I'm not sure I get here. IFCVF held refcnt of its PCI parent, so it looks to me it's safe to free PCI resources in vDPA free callback?
Quoted 5 lines
Suggest to use the devm versions of the above instead, and then you can reorder things so the allocation is done first.
Jason
Using devm looks nice, but if it's possible I prefer to tweak on top.
Thanks
Quoted 1 line
Re: [PATCH V8 9/9] virtio: Intel IFC VF driver for VDPA
Quoted 24 lines · Jason Wang
Quoted 24 lines
Quoted 11 lines
drivers/virtio/vdpa/ifcvf/ifcvf_main.c new file +11
-0+ adapter->vdpa.dma_dev = dev; + ret =vdpa_register_device (&adapter->vdpa); +if (ret) { +IFCVF_ERR (adapter->dev,"Failed to register ifcvf to vdpa bus" ); + goto err_msix; + } + +return 0 ; + +err_msix: +put_device (&adapter->vdpa.dev);drivers/virtio/vdpa/ifcvf/ifcvf_main.c line 446
Quoted 10 lines
drivers/virtio/vdpa/ifcvf/ifcvf_main.c new file +10
-0+return ret; +err_alloc: +pci_free_irq_vectors (pdev); +err_vectors: +pci_release_regions (pdev); +err_regions: +pci_disable_device (pdev); +err_enable: +return ret; +}I personally don't like seeing goto unwinds with multiple returns, and here I think it is actually a tiny bug.
Quoted 7 lines
Quoted 3 lines
All touches to the PCI device must stop before the driver core
remove()returns - so these pci function cannot be in the kref put release function anyhow.I'm not sure I get here. IFCVF held refcnt of its PCI parent, so it looks to me it's safe to free PCI resources in vDPA free callback?
The refcnt doesn't prevent the driver core from re-binding the pci_device to another driver. Then the refcount put would do a pci_disable_device() after another driver has started
For this reason all touches to a struct pci_device must stop before remove returns.
Jason
Re: [PATCH V8 9/9] virtio: Intel IFC VF driver for VDPA
Quoted 32 lines · Jason Gunthorpe
Quoted 24 lines · Jason Wang
Quoted 24 lines
Quoted 11 lines
drivers/virtio/vdpa/ifcvf/ifcvf_main.c new file +11
-0+ adapter->vdpa.dma_dev = dev; + ret =vdpa_register_device (&adapter->vdpa); +if (ret) { +IFCVF_ERR (adapter->dev,"Failed to register ifcvf to vdpa bus" ); + goto err_msix; + } + +return 0 ; + +err_msix: +put_device (&adapter->vdpa.dev);drivers/virtio/vdpa/ifcvf/ifcvf_main.c line 446
Quoted 10 lines
drivers/virtio/vdpa/ifcvf/ifcvf_main.c new file +10
-0+return ret; +err_alloc: +pci_free_irq_vectors (pdev); +err_vectors: +pci_release_regions (pdev); +err_regions: +pci_disable_device (pdev); +err_enable: +return ret; +}I personally don't like seeing goto unwinds with multiple returns, and here I think it is actually a tiny bug.
Quoted 6 lines
Quoted 3 lines
All touches to the PCI device must stop before the driver core
remove()returns - so these pci function cannot be in the kref put release function anyhow.I'm not sure I get here. IFCVF held refcnt of its PCI parent, so it looks to me it's safe to free PCI resources in vDPA free callback?
Quoted 8 lines
The refcnt doesn't prevent the driver core from re-binding the
pci_deviceto another driver. Then the refcount put would do apci_disable_device()after another driver has startedFor this reason all touches to a struct
pci_devicemust stop before remove returns.Jason
Ok, will send a new version shortly.
Thanks