vhost: support inflight share memory protocol feature
A message that is not in the archive (sent off-list or not yet received): 20190920120102.29828
v8 cover vhost: support inflight share memory protocol feature
This patches introduces two new messages VHOST_USER_GET_INFLIGHT_FD and VHOST_USER_SET_INFLIGHT_FD to support transferring a shared buffer between qemu and backend.
Now It can both support split and packed ring. The example code show how these API work. The test has passed.
How to test the example:
1, Qemu need two patches.
https://patchwork.kernel.org/patch/10766813/
https://patchwork.kernel.org/patch/10861411/(QEMU merged) it also needs some manual modifications:
we should confirm that before we send get inflight we have already sent the set features but it seems Qemu didn't do like this. So we manually revise this, we can add below code in vhost_dev_get_inflight
int r;
r = vhost_dev_set_features(dev, dev->log_enabled);
if (r < 0) {
return -errno;
}before get_inflight_fd.
2, Guest OS version >= 5.0
3, run the example
4, run the qemu with vhost-user-blk-pci.
eg:
-chardev socket,id=spdk_vhost_blk0,reconnect=1,path=xxxx\
-device vhost-user-blk-pci,ring_packed=on,chardev=spdk_vhost_blk0,num-queues=1\5, run fio in the guest
6, kill the example and run again.
7, the fio in the guest should continue run without errors.
Jin Yu (10):
vhost: add the inflight description
vhost: add packed ring
vhost: add the inflight structure
vhost: add two new messages to support a shared buffer
vhost: checkout the resubmit inflight information
vhost: add the APIs to operate inflight ring
vhost: add APIs for user getting inflight ring
vhost: add vring functions packed ring support
vhost: add an API for judging vq format
vhost: add vhost-user-blk example which support inflight
examples/vhost_blk/Makefile | 68 ++ examples/vhost_blk/blk.c | 125 +++ examples/vhost_blk/blk_spec.h | 95 +++ examples/vhost_blk/meson.build | 21 + examples/vhost_blk/vhost_blk.c | 1086 ++++++++++++++++++++++++ examples/vhost_blk/vhost_blk.h | 127 +++ examples/vhost_blk/vhost_blk_compat.c | 195 +++++ lib/librte_vhost/rte_vhost.h | 250 +++++- lib/librte_vhost/rte_vhost_version.map | 9 + lib/librte_vhost/vhost.c | 416 ++++++++- lib/librte_vhost/vhost.h | 16 + lib/librte_vhost/vhost_user.c | 468 +++++++++- lib/librte_vhost/vhost_user.h | 12 +- 13 files changed, 2875 insertions(+), 13 deletions(-) create mode 100644 examples/vhost_blk/Makefile create mode 100644 examples/vhost_blk/blk.c create mode 100644 examples/vhost_blk/blk_spec.h create mode 100644 examples/vhost_blk/meson.build create mode 100644 examples/vhost_blk/vhost_blk.c create mode 100644 examples/vhost_blk/vhost_blk.h create mode 100644 examples/vhost_blk/vhost_blk_compat.c
Signature · 1 line
2.17.2
PATCH v8 1/10 vhost: add the inflight description
This patch add the inflight message description and the inflight share fd protocol feature flag.
Signed-off-by Lin Li <[email protected]>
Signed-off-by Xun Ni <[email protected]>
Signed-off-by Yu Zhang <[email protected]>
Signed-off-by Jin Yu <[email protected]>
Reviewed-by Maxime Coquelin <[email protected]>
---
v1 - specify the APIs are split-ring only
v2 - fix APIs and judge split or packed
v3 - add rte_vhost_ prefix and fix one issue
v4 - add the packed ring support
v5 - revise get_vring_base func depend on Tiwei's suggestion v6 - divide patch into small patches
v7 - updated base on Maxime's comments
--- lib/librte_vhost/rte_vhost.h | 4 ++++ lib/librte_vhost/vhost_user.h | 8 ++++++++ 2 files changed, 12 insertions(+)
lib/librte_vhost/rte_vhost.h +4 -0
@@ -71,6 +71,10 @@ extern "C" {#define VHOST_USER_PROTOCOL_F_HOST_NOTIFIER 11#endif +#ifndef VHOST_USER_PROTOCOL_F_INFLIGHT_SHMFD +#define VHOST_USER_PROTOCOL_F_INFLIGHT_SHMFD 12 +#endif +/** Indicate whether protocol features negotiation is supported. */ #ifndef VHOST_USER_F_PROTOCOL_FEATURES #define VHOST_USER_F_PROTOCOL_FEATURES 30
lib/librte_vhost/vhost_user.h +8 -0
@@ -112,6 +112,13 @@ typedef struct VhostUserVringArea {uint64_t offset; } VhostUserVringArea; +typedef struct VhostUserInflight { +uint64_t mmap_size; +uint64_t mmap_offset; +uint16_t num_queues; +uint16_t queue_size; +}VhostUserInflight ; + typedef struct VhostUserMsg { union { uint32_t master; /* a VhostUserRequest value */
@@ -134,6 +141,7 @@ typedef struct VhostUserMsg {struct vhost_iotlb_msg iotlb;VhostUserCryptoSessionParam crypto_session;VhostUserVringArea area; +VhostUserInflight inflight; } payload;int fds[VHOST_MEMORY_MAX_NREGIONS ];int fd_num;
Signature · 1 line
2.17.2
PATCH v8 2/10 vhost: add packed ring
In reply to Jin YuThis patch add the packed ring in the rte_vhost_vring.
Signed-off-by Lin Li <[email protected]>
Signed-off-by Xun Ni <[email protected]>
Signed-off-by Yu Zhang <[email protected]>
Signed-off-by Jin Yu <[email protected]>
Reviewed-by Maxime Coquelin <[email protected]>
--- lib/librte_vhost/rte_vhost.h | 15 ++++++++++++--- 1 file changed, 12 insertions(+), 3 deletions(-)
lib/librte_vhost/rte_vhost.h +12 -3
@@ -103,9 +103,18 @@ struct rte_vhost_memory { };struct rte_vhost_vring { -struct vring_desc *desc; -struct vring_avail *avail; -struct vring_used *used; +union { +struct vring_desc *desc; +struct vring_packed_desc *desc_packed; + }; +union { +struct vring_avail *avail; +struct vring_packed_desc_event *driver_event; + }; +union { +struct vring_used *used; +struct vring_packed_desc_event *device_event; + };uint64_t log_guest_addr;/** Deprecated, use rte_vhost_vring_call() instead. */
Signature · 1 line
2.17.2
PATCH v8 3/10 vhost: add the inflight structure
In reply to Jin YuThis patch adds the inflight queue region structure include the split and packed.
Signed-off-by Lin Li <[email protected]>
Signed-off-by Xun Ni <[email protected]>
Signed-off-by Yu Zhang <[email protected]>
Signed-off-by Jin Yu <[email protected]>
--- lib/librte_vhost/rte_vhost.h | 43 ++++++++++++++++++++++++++++++++++++ 1 file changed, 43 insertions(+)
lib/librte_vhost/rte_vhost.h +43 -0
@@ -102,6 +102,49 @@ struct rte_vhost_memory {struct rte_vhost_mem_region regions[]; }; +struct rte_vhost_inflight_desc_split { +uint8_t inflight; +uint8_t padding[5 ]; +uint16_t next; +uint64_t counter; +}; + +struct rte_vhost_inflight_info_split { +uint64_t features; +uint16_t version; +uint16_t desc_num; +uint16_t last_inflight_io; +uint16_t used_idx; +struct rte_vhost_inflight_desc_split desc[0 ]; +}; + +struct rte_vhost_inflight_desc_packed { +uint8_t inflight; +uint8_t padding; +uint16_t next; +uint16_t last; +uint16_t num; +uint64_t counter; +uint16_t id; +uint16_t flags; +uint32_t len; +uint64_t addr; + }; + +struct rte_vhost_inflight_info_packed { +uint64_t features; +uint16_t version; +uint16_t desc_num; +uint16_t free_head; +uint16_t old_free_head; +uint16_t used_idx; +uint16_t old_used_idx; +uint8_t used_wrap_counter; +uint8_t old_used_wrap_counter; +uint8_t padding[7 ]; +struct rte_vhost_inflight_desc_packed desc[0 ]; +}; + struct rte_vhost_vring { union { struct vring_desc *desc;
Signature · 1 line
2.17.2
PATCH v8 4/10 vhost: add two new messages to support a shared buffer
In reply to Jin YuThis patch introduces two new messages VHOST_USER_GET_INFLIGHT_FD and VHOST_USER_SET_INFLIGHT_FD to support transferring a shared buffer between qemu and backend.
Signed-off-by Lin Li <[email protected]>
Signed-off-by Xun Ni <[email protected]>
Signed-off-by Yu Zhang <[email protected]>
Signed-off-by Jin Yu <[email protected]>
--- lib/librte_vhost/vhost.h | 7 + lib/librte_vhost/vhost_user.c | 255 +++++++++++++++++++++++++++++++++- lib/librte_vhost/vhost_user.h | 4 +- 3 files changed, 264 insertions(+), 2 deletions(-)
lib/librte_vhost/vhost.h +7 -0
@@ -286,6 +286,12 @@ struct guest_page {uint64_t size; }; +struct inflight_mem_info { +int fd; +void *addr; +uint64_t size; +}; +/** * Device structure contains all configuration information relating * to the device.
@@ -303,6 +309,7 @@ struct virtio_net {uint32_t nr_vring;int dequeue_zero_copy;struct vhost_virtqueue *virtqueue[VHOST_MAX_QUEUE_PAIRS *2 ]; +struct inflight_mem_info *inflight_info;#define IF_NAME_SZ (PATH_MAX > IFNAMSIZ ? PATH_MAX : IFNAMSIZ)char ifname[IF_NAME_SZ ];uint64_t log_size;
lib/librte_vhost/vhost_user.c +254 -1
293 lines not on this page · raw message
lib/librte_vhost/vhost_user.h +3 -1
@@ -54,7 +54,9 @@ typedef enum VhostUserRequest {VHOST_USER_POSTCOPY_ADVISE =28 ,VHOST_USER_POSTCOPY_LISTEN =29 ,VHOST_USER_POSTCOPY_END =30 , - VHOST_USER_MAX = 31 +VHOST_USER_GET_INFLIGHT_FD =31 , +VHOST_USER_SET_INFLIGHT_FD =32 , + VHOST_USER_MAX = 33 } VhostUserRequest; typedef enum VhostUserSlaveRequest {
Signature · 1 line
2.17.2
Re: [dpdk-dev] [PATCH v8 04/10] vhost: add two new messages to support a shared buffer
Quoted 56 lines · Jin Yu
This patch introduces two new messages
VHOST_USER_GET_INFLIGHT_FDandVHOST_USER_SET_INFLIGHT_FDto support transferring a shared buffer between qemu and backend.Signed-off-by Lin Li <[email protected]>Signed-off-by Xun Ni <[email protected]>Signed-off-by Yu Zhang <[email protected]>Signed-off-by Jin Yu <[email protected]>--- lib/librte_vhost/vhost.h | 7 + lib/librte_vhost/vhost_user.c | 255 +++++++++++++++++++++++++++++++++- lib/librte_vhost/vhost_user.h | 4 +- 3 files changed, 264 insertions(+), 2 deletions(-)lib/librte_vhost/vhost.h +7
-0@@ -286,6 +286,12 @@ struct guest_page {uint64_t size; }; +struct inflight_mem_info { +int fd; +void *addr; +uint64_t size; +}; +/** * Device structure contains all configuration information relating * to the device. @@ -303,6 +309,7 @@ struct virtio_net {uint32_t nr_vring;int dequeue_zero_copy;struct vhost_virtqueue *virtqueue[VHOST_MAX_QUEUE_PAIRS *2 ]; +struct inflight_mem_info *inflight_info;#define IF_NAME_SZ (PATH_MAX > IFNAMSIZ ? PATH_MAX : IFNAMSIZ)char ifname[IF_NAME_SZ ];uint64_t log_size;lib/librte_vhost/vhost_user.c +4
-0@@ -37,6 +37,10 @@#ifdef RTE_LIBRTE_VHOST_POSTCOPY #include <linux/userfaultfd.h> #endif +#ifdef F_ADD_SEALS /* if file sealing is supported, so is memfd */ +#include <linux/memfd.h> +#define MEMFD_SUPPORTED +#endif #include <rte_common.h> #include <rte_malloc.h> @@ -49,6 +53,15 @@#define VIRTIO_MIN_MTU 68
lib/librte_vhost/vhost_user.c line 63
Quoted 10 lines
lib/librte_vhost/vhost_user.c +8
-0#define VIRTIO_MAX_MTU 65535 +#define INFLIGHT_ALIGNMENT 64 +#define INFLIGHT_VERSION 0x1 +#define VIRTQUEUE_MAX_SIZE 1024 + +#define CLOEXEC 0x0001U + +#define ALIGN_DOWN (n, m) ((n) / (m) * (m)) +#define ALIGN_UP (n, m) ALIGN_DOWN((n) + (m) - 1, (m))
There are equivalent macros in lib/librte_eal/common/include/rte_common.h You might want to consider using those instead of adding new ones.
Quoted 112 lines
lib/librte_vhost/vhost_user.c +91
-0+static const char *vhost_message_str[VHOST_USER_MAX ] = { [VHOST_USER_NONE ] ="VHOST_USER_NONE" , [VHOST_USER_GET_FEATURES ] ="VHOST_USER_GET_FEATURES" ,@@ -78,6 +91,8 @@ static const char *vhost_message_str[VHOST_USER_MAX] = { [VHOST_USER_POSTCOPY_ADVISE] = "VHOST_USER_POSTCOPY_ADVISE", [VHOST_USER_POSTCOPY_LISTEN] = "VHOST_USER_POSTCOPY_LISTEN", [VHOST_USER_POSTCOPY_END] = "VHOST_USER_POSTCOPY_END", + [VHOST_USER_GET_INFLIGHT_FD] = "VHOST_USER_GET_INFLIGHT_FD", + [VHOST_USER_SET_INFLIGHT_FD] = "VHOST_USER_SET_INFLIGHT_FD", };static int send_vhost_reply (int sockfd,struct VhostUserMsg *msg);@@ -160,6 +175,22 @@ vhost_backend_cleanup(struct virtio_net *dev) dev->log_addr =0 ; } +if (dev->inflight_info) { +if (dev->inflight_info->addr) { +munmap (dev->inflight_info->addr, + dev->inflight_info->size); + dev->inflight_info->addr =NULL ; + } + +if (dev->inflight_info->fd >0 ) { +close (dev->inflight_info->fd); + dev->inflight_info->fd =-1 ; + } + +free (dev->inflight_info); + dev->inflight_info =NULL ; + } +if (dev->slave_req_fd >=0 ) {close (dev->slave_req_fd); dev->slave_req_fd =-1 ;@@ -1165,6 +1196,227 @@ virtio_is_ready(struct virtio_net *dev)return 1 ; } +static void * +inflight_mem_alloc (const char *name,size_t size,int *fd) +{ +void *ptr; +int mfd =-1 ; +char fname[20 ] ="/tmp/memfd-XXXXXX" ; + + *fd =-1 ; +#ifdef MEMFD_SUPPORTED + mfd =memfd_create (name,MFD_CLOEXEC ); +#else +RTE_SET_USED (name); +#endif +if (mfd !=-1 ) { +if (ftruncate (mfd, size) ==-1 ) { +RTE_LOG (ERR ,VHOST_CONFIG , +"ftruncate fail for alloc inflight buffer\n" ); +close (mfd); +return NULL ; + } + }else { + mfd =mkstemp (fname); +unlink (fname); + +if (mfd ==-1 ) { +RTE_LOG (ERR ,VHOST_CONFIG , +"mkstemp fail for alloc inflight buffer\n" ); +return NULL ; + } + +if (ftruncate (mfd, size) ==-1 ) { +RTE_LOG (ERR ,VHOST_CONFIG , +"ftruncate fail for alloc inflight buffer\n" ); +close (mfd); +return NULL ; + } + } + + ptr =mmap (0 , size,PROT_READ |PROT_WRITE ,MAP_SHARED , mfd,0 ); +if (ptr ==MAP_FAILED ) { +RTE_LOG (ERR ,VHOST_CONFIG , +"mmap fail for alloc inflight buffer\n" ); +close (mfd); +return NULL ; + } + + *fd = mfd; +return ptr; +} + +static uint32_t +get_pervq_shm_size_split (uint16_t queue_size) +{ +return ALIGN_UP (sizeof (struct rte_vhost_inflight_desc_split ) * + queue_size +sizeof (uint64_t ) +sizeof (uint16_t ) *4 , +INFLIGHT_ALIGNMENT ); +} + +static uint32_t +get_pervq_shm_size_packed (uint16_t queue_size) +{ +return ALIGN_UP (sizeof (struct rte_vhost_inflight_desc_packed ) * + queue_size +sizeof (uint64_t ) +sizeof (uint16_t ) *6 + +sizeof (uint8_t ) *9 ,INFLIGHT_ALIGNMENT ); +} + +static int +vhost_user_get_inflight_fd (struct virtio_net **pdev, +VhostUserMsg *msg, + int main_fd __rte_unused) +{ +int fd, i, j;
lib/librte_vhost/vhost_user.c virtio_is_ready(struct virtio_net *dev) · line 1280
Quoted 10 lines
lib/librte_vhost/vhost_user.c +10
-0+uint64_t pervq_inflight_size, mmap_size; +void *addr; +uint16_t num_queues, queue_size; +struct virtio_net *dev = *pdev; +struct rte_vhost_inflight_info_packed *inflight_packed =NULL ; + +if (msg->size !=sizeof (msg->payload.inflight)) { +RTE_LOG (ERR ,VHOST_CONFIG , +"Invalid get_inflight_fd message size is %d" , + msg->size);
Did you forget to end the error message with a '\n'?
lib/librte_vhost/vhost_user.c virtio_is_ready(struct virtio_net *dev) · line 1289
Quoted 9 lines
lib/librte_vhost/vhost_user.c +9
-0+return RTE_VHOST_MSG_RESULT_ERR ; + } + +if (dev->inflight_info ==NULL ) { + dev->inflight_info = +calloc (1 ,sizeof (struct inflight_mem_info )); +if (!dev->inflight_info) { +RTE_LOG (ERR ,VHOST_CONFIG , +"Failed to alloc dev inflight area" );
Same here: maybe missing '\n'?
Quoted 151 lines
lib/librte_vhost/vhost_user.c +132
-1+return RTE_VHOST_MSG_RESULT_ERR ; + } + } + + num_queues = msg->payload.inflight.num_queues; + queue_size = msg->payload.inflight.queue_size; + +RTE_LOG (INFO ,VHOST_CONFIG ,"get_inflight_fd num_queues: %u\n" , + msg->payload.inflight.num_queues); +RTE_LOG (INFO ,VHOST_CONFIG ,"get_inflight_fd queue_size: %u\n" , + msg->payload.inflight.queue_size); + +if (vq_is_packed (dev)) + pervq_inflight_size =get_pervq_shm_size_packed (queue_size); +else + pervq_inflight_size =get_pervq_shm_size_split (queue_size); + + mmap_size = num_queues * pervq_inflight_size; + addr =inflight_mem_alloc ("vhost-inflight" , mmap_size, &fd); +if (!addr) { +RTE_LOG (ERR ,VHOST_CONFIG , +"Failed to alloc vhost inflight area" ); + msg->payload.inflight.mmap_size =0 ; +return RTE_VHOST_MSG_RESULT_ERR ; + } +memset (addr,0 , mmap_size); + + dev->inflight_info->addr = addr; + dev->inflight_info->size = msg->payload.inflight.mmap_size = mmap_size; + dev->inflight_info->fd = msg->fds[0 ] = fd; + msg->payload.inflight.mmap_offset =0 ; + msg->fd_num =1 ; + +if (vq_is_packed (dev)) { +for (i =0 ; i < num_queues; i++) { + inflight_packed = + (struct rte_vhost_inflight_info_packed *)addr; + inflight_packed->used_wrap_counter =1 ; + inflight_packed->old_used_wrap_counter =1 ; +for (j =0 ; j < queue_size; j++) + inflight_packed->desc[j].next = j +1 ; + addr = (void *)((char *)addr + pervq_inflight_size); + } + } + +RTE_LOG (INFO ,VHOST_CONFIG , +"send inflight mmap_size: %" PRIu64"\n" , + msg->payload.inflight.mmap_size); +RTE_LOG (INFO ,VHOST_CONFIG , +"send inflight mmap_offset: %" PRIu64"\n" , + msg->payload.inflight.mmap_offset); +RTE_LOG (INFO ,VHOST_CONFIG , +"send inflight fd: %d\n" , msg->fds[0 ]); + +return RTE_VHOST_MSG_RESULT_REPLY ; +} + +static int +vhost_user_set_inflight_fd (struct virtio_net **pdev,VhostUserMsg *msg, + int main_fd __rte_unused) +{ +int fd; +uint64_t mmap_size, mmap_offset; +uint16_t num_queues, queue_size; +uint32_t pervq_inflight_size; +void *addr; +struct virtio_net *dev = *pdev; + + fd = msg->fds[0 ]; +if (msg->size !=sizeof (msg->payload.inflight) || fd <0 ) { +RTE_LOG (ERR ,VHOST_CONFIG , +"Invalid set_inflight_fd message size is %d,fd is %d\n" , + msg->size, fd); +return RTE_VHOST_MSG_RESULT_ERR ; + } + + mmap_size = msg->payload.inflight.mmap_size; + mmap_offset = msg->payload.inflight.mmap_offset; + num_queues = msg->payload.inflight.num_queues; + queue_size = msg->payload.inflight.queue_size; + +if (vq_is_packed (dev)) + pervq_inflight_size =get_pervq_shm_size_packed (queue_size); +else + pervq_inflight_size =get_pervq_shm_size_split (queue_size); + +RTE_LOG (INFO ,VHOST_CONFIG , +"set_inflight_fd mmap_size: %" PRIu64"\n" , mmap_size); +RTE_LOG (INFO ,VHOST_CONFIG , +"set_inflight_fd mmap_offset: %" PRIu64"\n" , mmap_offset); +RTE_LOG (INFO ,VHOST_CONFIG , +"set_inflight_fd num_queues: %u\n" , num_queues); +RTE_LOG (INFO ,VHOST_CONFIG , +"set_inflight_fd queue_size: %u\n" , queue_size); +RTE_LOG (INFO ,VHOST_CONFIG , +"set_inflight_fd fd: %d\n" , fd); +RTE_LOG (INFO ,VHOST_CONFIG , +"set_inflight_fd pervq_inflight_size: %d\n" , + pervq_inflight_size); + +if (!dev->inflight_info) { + dev->inflight_info =calloc (1 , +sizeof (struct inflight_mem_info )); +if (dev->inflight_info ==NULL ) { +RTE_LOG (ERR ,VHOST_CONFIG , +"Failed to alloc dev inflight area" ); +return RTE_VHOST_MSG_RESULT_ERR ; + } + } + +if (dev->inflight_info->addr) +munmap (dev->inflight_info->addr, dev->inflight_info->size); + + addr =mmap (0 , mmap_size,PROT_READ |PROT_WRITE ,MAP_SHARED , + fd, mmap_offset); +if (addr ==MAP_FAILED ) { +RTE_LOG (ERR ,VHOST_CONFIG ,"failed to mmap share memory.\n" ); +return RTE_VHOST_MSG_RESULT_ERR ; + } + +if (dev->inflight_info->fd) +close (dev->inflight_info->fd); + + dev->inflight_info->fd = fd; + dev->inflight_info->addr = addr; + dev->inflight_info->size = mmap_size; + +return RTE_VHOST_MSG_RESULT_OK ; +} +static int vhost_user_set_vring_call (struct virtio_net **pdev,struct VhostUserMsg *msg, int main_fd __rte_unused)@@ -1762,9 +2014,10 @@ static vhost_message_handler_t vhost_message_handlers[VHOST_USER_MAX] = { [VHOST_USER_POSTCOPY_ADVISE] = vhost_user_set_postcopy_advise, [VHOST_USER_POSTCOPY_LISTEN] = vhost_user_set_postcopy_listen, [VHOST_USER_POSTCOPY_END] = vhost_user_postcopy_end, + [VHOST_USER_GET_INFLIGHT_FD] = vhost_user_get_inflight_fd, + [VHOST_USER_SET_INFLIGHT_FD] = vhost_user_set_inflight_fd, }; -/* return bytes# of read on success or negative val on failure. */ static int read_vhost_message (int sockfd,struct VhostUserMsg *msg)lib/librte_vhost/vhost_user.h +0
-0
lib/librte_vhost/vhost_user.h typedef enum VhostUserRequest { · line 62
Quoted 10 lines
lib/librte_vhost/vhost_user.h +3
-1@@ -54,7 +54,9 @@ typedef enum VhostUserRequest {VHOST_USER_POSTCOPY_ADVISE =28 ,VHOST_USER_POSTCOPY_LISTEN =29 ,VHOST_USER_POSTCOPY_END =30 , - VHOST_USER_MAX = 31 +VHOST_USER_GET_INFLIGHT_FD =31 , +VHOST_USER_SET_INFLIGHT_FD =32 , + VHOST_USER_MAX = 33 } VhostUserRequest; typedef enum VhostUserSlaveRequest {
Thanks.
Adrián
PATCH v8 5/10 vhost: checkout the resubmit inflight information
In reply to Jin YuThis patch shows how to checkout the inflight ring and construct the resubmit information also include destroying resubmit info.
Signed-off-by Lin Li <[email protected]>
Signed-off-by Xun Ni <[email protected]>
Signed-off-by Yu Zhang <[email protected]>
Signed-off-by Jin Yu <[email protected]>
--- lib/librte_vhost/rte_vhost.h | 19 +++ lib/librte_vhost/vhost.c | 29 ++++- lib/librte_vhost/vhost.h | 9 ++ lib/librte_vhost/vhost_user.c | 215 +++++++++++++++++++++++++++++++++- 4 files changed, 270 insertions(+), 2 deletions(-)
lib/librte_vhost/rte_vhost.h +19 -0
@@ -145,6 +145,25 @@ struct rte_vhost_inflight_info_packed {struct rte_vhost_inflight_desc_packed desc[0 ]; }; +struct rte_vhost_resubmit_desc { +uint16_t index; +uint64_t counter; +}; + +struct rte_vhost_resubmit_info { +struct rte_vhost_resubmit_desc *resubmit_list; +uint16_t resubmit_num; +}; + +struct rte_vhost_ring_inflight { +union { +struct rte_vhost_inflight_info_split *inflight_split; +struct rte_vhost_inflight_info_packed *inflight_packed; + }; + +struct rte_vhost_resubmit_info *resubmit_inflight; +}; + struct rte_vhost_vring { union { struct vring_desc *desc;
lib/librte_vhost/vhost.c +28 -1
@@ -242,6 +242,31 @@ cleanup_vq(struct vhost_virtqueue *vq, int destroy)close (vq->kickfd); } +void +cleanup_vq_inflight (struct virtio_net *dev,struct vhost_virtqueue *vq) +{ +if (!(dev->protocol_features & + (1ULL <<VHOST_USER_PROTOCOL_F_INFLIGHT_SHMFD ))) +return ; + +if (vq_is_packed (dev)) { +if (vq->inflight_packed) + vq->inflight_packed =NULL ; + }else { +if (vq->inflight_split) + vq->inflight_split =NULL ; + } + +if (vq->resubmit_inflight) { +if (vq->resubmit_inflight->resubmit_list) { +free (vq->resubmit_inflight->resubmit_list); + vq->resubmit_inflight->resubmit_list =NULL ; + } +free (vq->resubmit_inflight); + vq->resubmit_inflight =NULL ; + } +} +/* * Unmap any memory, close any file descriptors and * free any memory owned by a device.
@@ -253,8 +278,10 @@ cleanup_device(struct virtio_net *dev, int destroy)vhost_backend_cleanup (dev); -for (i =0 ; i < dev->nr_vring; i++) +for (i =0 ; i < dev->nr_vring; i++) {cleanup_vq (dev->virtqueue[i], destroy); +cleanup_vq_inflight (dev, dev->virtqueue[i]); + } }void
lib/librte_vhost/vhost.h +9 -0
@@ -128,6 +128,14 @@ struct vhost_virtqueue {/* Physical address of used ring, for logging */ uint64_t log_guest_addr; +/* inflight share memory info */ +union { +struct rte_vhost_inflight_info_split *inflight_split; +struct rte_vhost_inflight_info_packed *inflight_packed; + }; +struct rte_vhost_resubmit_info *resubmit_inflight; +uint64_t global_counter; +uint16_t nr_zmbuf;uint16_t zmbuf_size;uint16_t last_zmbuf_idx;
@@ -474,6 +482,7 @@ void vhost_destroy_device(int);void vhost_destroy_device_notify (struct virtio_net *dev);void cleanup_vq (struct vhost_virtqueue *vq,int destroy); +void cleanup_vq_inflight (struct virtio_net *dev,struct vhost_virtqueue *vq);void free_vq (struct virtio_net *dev,struct vhost_virtqueue *vq);int alloc_vring_queue (struct virtio_net *dev,uint32_t vring_idx);
lib/librte_vhost/vhost_user.c +214 -1
249 lines not on this page · raw message
Signature · 1 line
2.17.2
PATCH v8 6/10 vhost: add the APIs to operate inflight ring
In reply to Jin YuThis patch introduces three APIs to operate the inflight ring. Three APIs are set, set last and clear. It includes split and packed ring.
Signed-off-by Lin Li <[email protected]>
Signed-off-by Xun Ni <[email protected]>
Signed-off-by Yu Zhang <[email protected]>
Signed-off-by Jin Yu <[email protected]>
--- lib/librte_vhost/rte_vhost.h | 116 +++++++++++ lib/librte_vhost/rte_vhost_version.map | 6 + lib/librte_vhost/vhost.c | 272 +++++++++++++++++++++++++ 3 files changed, 394 insertions(+)
lib/librte_vhost/rte_vhost.h +116 -0
@@ -693,6 +693,122 @@ int rte_vhost_get_mem_table(int vid, struct rte_vhost_memory **mem);int rte_vhost_get_vhost_vring (int vid,uint16_t vring_idx,struct rte_vhost_vring *vring); +/** +* Set split inflight descriptor. +* +* This function save descriptors that has been comsumed in available +* ring +* +* @param vid +* vhost device ID +* @param vring_idx +* vring index +* @param idx +* inflight entry index +* @return +* 0 on success, -1 on failure +*/ +__rte_experimental +int +rte_vhost_set_inflight_desc_split (int vid,uint16_t vring_idx, +uint16_t idx); + +/** +* Set packed inflight descriptor and get corresponding inflight entry +* +* This function save descriptors that has been comsumed +* +* @param vid +* vhost device ID +* @param vring_idx +* vring index +* @param head +* head of descriptors +* @param last +* last of descriptors +* @param inflight_entry +* corresponding inflight entry +* @return +* 0 on success, -1 on failure +*/ +__rte_experimental +int +rte_vhost_set_inflight_desc_packed (int vid,uint16_t vring_idx, +uint16_t head,uint16_t last,uint16_t *inflight_entry); + +/** +* Save the head of list that the last batch of used descriptors. +* +* @param vid +* vhost device ID +* @param vring_idx +* vring index +* @param idx +* descriptor entry index +* @return +* 0 on success, -1 on failure +*/ +__rte_experimental +int +rte_vhost_set_last_inflight_io_split (int vid, +uint16_t vring_idx,uint16_t idx); + +/** +* Update the inflight free_head, used_idx and used_wrap_counter. +* +* This function will update status first before updating descriptors +* to used +* +* @param vid +* vhost device ID +* @param vring_idx +* vring index +* @param head +* head of descriptors +* @return +* 0 on success, -1 on failure +*/ +__rte_experimental +int +rte_vhost_set_last_inflight_io_packed (int vid, +uint16_t vring_idx,uint16_t head); + +/** +* Clear the split inflight status. +* +* @param vid +* vhost device ID +* @param vring_idx +* vring index +* @param last_used_idx +* last used idx of used ring +* @param idx +* inflight entry index +* @return +* 0 on success, -1 on failure +*/ +__rte_experimental +int +rte_vhost_clr_inflight_desc_split (int vid,uint16_t vring_idx, +uint16_t last_used_idx,uint16_t idx); + +/** +* Clear the packed inflight status. +* +* @param vid +* vhost device ID +* @param vring_idx +* vring index +* @param head +* inflight entry index +* @return +* 0 on success, -1 on failure +*/ +__rte_experimental +int +rte_vhost_clr_inflight_desc_packed (int vid,uint16_t vring_idx, +uint16_t head); +/** * Notify the guest that used descriptors have been added to the vring. This * function acts as a memory barrier.
lib/librte_vhost/rte_vhost_version.map +6 -0
@@ -87,4 +87,10 @@ EXPERIMENTAL { rte_vdpa_relay_vring_used; rte_vhost_extern_callback_register; rte_vhost_driver_set_protocol_features; + rte_vhost_set_inflight_desc_split; + rte_vhost_set_inflight_desc_packed; + rte_vhost_set_last_inflight_io_split; + rte_vhost_set_last_inflight_io_packed; + rte_vhost_clr_inflight_desc_split; + rte_vhost_clr_inflight_desc_packed; };
lib/librte_vhost/vhost.c +272 -0
278 lines not on this page · raw message
Signature · 1 line
2.17.2
PATCH v8 7/10 vhost: add APIs for user getting inflight ring
In reply to Jin YuThis patch introduces two APIs. one is for getting inflgiht ring and the other is for getting base.
Signed-off-by Lin Li <[email protected]>
Signed-off-by Xun Ni <[email protected]>
Signed-off-by Yu Zhang <[email protected]>
Signed-off-by Jin Yu <[email protected]>
--- lib/librte_vhost/rte_vhost.h | 40 +++++++++++++++++ lib/librte_vhost/rte_vhost_version.map | 2 + lib/librte_vhost/vhost.c | 59 ++++++++++++++++++++++++++ 3 files changed, 101 insertions(+)
lib/librte_vhost/rte_vhost.h +40 -0
@@ -693,6 +693,23 @@ int rte_vhost_get_mem_table(int vid, struct rte_vhost_memory **mem);int rte_vhost_get_vhost_vring (int vid,uint16_t vring_idx,struct rte_vhost_vring *vring); +/** +* Get guest inflight vring info, including inflight ring and resubmit list. +* +* @param vid +* vhost device ID +* @param vring_idx +* vring index +* @param vring +* the structure to hold the requested inflight vring info +* @return +* 0 on success, -1 on failure +*/ +__rte_experimental +int +rte_vhost_get_vhost_ring_inflight (int vid,uint16_t vring_idx, +struct rte_vhost_ring_inflight *vring); +/** * Set split inflight descriptor. *
@@ -867,6 +884,29 @@ int __rte_experimental rte_vhost_get_vring_base(int vid, uint16_t queue_id, uint16_t *last_avail_idx, uint16_t *last_used_idx); +/** +* Get last_avail/last_used of the vhost virtqueue +* +* This function is designed for the reconnection and it's specific for +* the packed ring as we can get the two parameters from the inflight +* queueregion +* +* @param vid +* vhost device ID +* @param queue_id +* vhost queue index +* @param last_avail_idx +* vhost last_avail_idx to get +* @param last_used_idx +* vhost last_used_idx to get +* @return +* 0 on success, -1 on failure +*/ +__rte_experimental +int +rte_vhost_get_vring_base_from_inflight (int vid, +uint16_t queue_id,uint16_t *last_avail_idx,uint16_t *last_used_idx); +/** * Set last_avail/used_idx of the vhost virtqueue *
lib/librte_vhost/rte_vhost_version.map +2 -0
@@ -93,4 +93,6 @@ EXPERIMENTAL { rte_vhost_set_last_inflight_io_packed; rte_vhost_clr_inflight_desc_split; rte_vhost_clr_inflight_desc_packed; + rte_vhost_get_vhost_ring_inflight; + rte_vhost_get_vring_base_from_inflight; };
lib/librte_vhost/vhost.c +59 -0
71 lines not on this page · raw message
Signature · 1 line
2.17.2
PATCH v8 8/10 vhost: add vring functions packed ring support
In reply to Jin YuThis patch add packed ring support in two APIs so user can get the packed ring`.
Signed-off-by Lin Li <[email protected]>
Signed-off-by Xun Ni <[email protected]>
Signed-off-by Yu Zhang <[email protected]>
Signed-off-by Jin Yu <[email protected]>
--- lib/librte_vhost/vhost.c | 68 +++++++++++++++++++++++++++++----------- 1 file changed, 49 insertions(+), 19 deletions(-)
lib/librte_vhost/vhost.c +49 -19
@@ -771,9 +771,15 @@ rte_vhost_get_vhost_vring(int vid, uint16_t vring_idx,if (!vq)return -1 ; - vring->desc = vq->desc; - vring->avail = vq->avail; - vring->used = vq->used; +if (vq_is_packed (dev)) { + vring->desc_packed = vq->desc_packed; + vring->driver_event = vq->driver_event; + vring->device_event = vq->device_event; + }else { + vring->desc = vq->desc; + vring->avail = vq->avail; + vring->used = vq->used; + } vring->log_guest_addr = vq->log_guest_addr; vring->callfd = vq->callfd;
@@ -1273,13 +1279,51 @@ int rte_vhost_get_log_base(int vid, uint64_t *log_base,int rte_vhost_get_vring_base (int vid,uint16_t queue_id,uint16_t *last_avail_idx,uint16_t *last_used_idx) { +struct vhost_virtqueue *vq;struct virtio_net *dev =get_device (vid);if (dev ==NULL || last_avail_idx ==NULL || last_used_idx ==NULL )return -1 ; - *last_avail_idx = dev->virtqueue[queue_id]->last_avail_idx; - *last_used_idx = dev->virtqueue[queue_id]->last_used_idx; + vq = dev->virtqueue[queue_id]; +if (!vq) +return -1 ; + +if (vq_is_packed (dev)) { + *last_avail_idx = (vq->avail_wrap_counter <<15 ) | + vq->last_avail_idx; + *last_used_idx = (vq->used_wrap_counter <<15 ) | + vq->last_used_idx; + }else { + *last_avail_idx = vq->last_avail_idx; + *last_used_idx = vq->last_used_idx; + } + +return 0 ; +} + +int rte_vhost_set_vring_base (int vid,uint16_t queue_id, +uint16_t last_avail_idx,uint16_t last_used_idx) +{ +struct vhost_virtqueue *vq; +struct virtio_net *dev =get_device (vid); + +if (!dev) +return -1 ; + + vq = dev->virtqueue[queue_id]; +if (!vq) +return -1 ; + +if (vq_is_packed (dev)) { + vq->last_avail_idx = last_avail_idx &0x7fff ; + vq->avail_wrap_counter = !!(last_avail_idx & (1 <<15 )); + vq->last_used_idx = last_used_idx &0x7fff ; + vq->used_wrap_counter = !!(last_used_idx & (1 <<15 )); + }else { + vq->last_avail_idx = last_avail_idx; + vq->last_used_idx = last_used_idx; + }return 0 ; }
@@ -1308,20 +1352,6 @@ rte_vhost_get_vring_base_from_inflight(int vid,return 0 ; } -int rte_vhost_set_vring_base (int vid,uint16_t queue_id, -uint16_t last_avail_idx,uint16_t last_used_idx) -{ -struct virtio_net *dev =get_device (vid); - -if (!dev) -return -1 ; - - dev->virtqueue[queue_id]->last_avail_idx = last_avail_idx; - dev->virtqueue[queue_id]->last_used_idx = last_used_idx; - -return 0 ; -} -int rte_vhost_extern_callback_register (int vid,struct rte_vhost_user_extern_ops const *const ops,void *ctx) {
Signature · 1 line
2.17.2
PATCH v8 9/10 vhost: add an API for judging vq format
In reply to Jin YuThis patch introduces an API for getting virtqueue format.
Signed-off-by Lin Li <[email protected]>
Signed-off-by Xun Ni <[email protected]>
Signed-off-by Yu Zhang <[email protected]>
Signed-off-by Jin Yu <[email protected]>
--- lib/librte_vhost/rte_vhost.h | 13 +++++++++++++ lib/librte_vhost/rte_vhost_version.map | 1 + lib/librte_vhost/vhost.c | 12 ++++++++++++ 3 files changed, 26 insertions(+)
lib/librte_vhost/rte_vhost.h +13 -0
@@ -11,6 +11,7 @@*/ #include <stdint.h> +#include <stdbool.h> #include <sys/eventfd.h> #include <rte_memory.h>
@@ -678,6 +679,18 @@ uint16_t rte_vhost_dequeue_burst(int vid, uint16_t queue_id,*/ int rte_vhost_get_mem_table (int vid,struct rte_vhost_memory **mem); +/** +* Get vq is packed +* +* @param vid +* vhost device ID +* @return +* true on packed, false on split +*/ +__rte_experimental +bool +rte_vhost_vq_is_packed (int vid); +/** * Get guest vring info, including the vring address, vring size, etc. *
lib/librte_vhost/rte_vhost_version.map +1 -0
@@ -95,4 +95,5 @@ EXPERIMENTAL { rte_vhost_clr_inflight_desc_packed; rte_vhost_get_vhost_ring_inflight; rte_vhost_get_vring_base_from_inflight; + rte_vhost_vq_is_packed; };
lib/librte_vhost/vhost.c +12 -0
@@ -753,6 +753,18 @@ rte_vhost_get_mem_table(int vid, struct rte_vhost_memory **mem)return 0 ; } +bool +rte_vhost_vq_is_packed (int vid) +{ +struct virtio_net *dev; + + dev =get_device (vid); +if (unlikely (!dev)) +return -1 ; + +return vq_is_packed (dev); +} +int rte_vhost_get_vhost_vring (int vid,uint16_t vring_idx,struct rte_vhost_vring *vring)
Signature · 1 line
2.17.2
PATCH v8 10/10 vhost: add vhost-user-blk example which support inflight
In reply to Jin YuA vhost-user-blk example that support inflight feature. It uses the new APIs that introduced in the first patch, so it can show how these APIs work to support inflight feature.
Signed-off-by Jin Yu <[email protected]>
---
V1 - add the case.
V2 - add the rte_vhost prefix.
V3 - add packed ring support
--- examples/vhost_blk/Makefile | 68 ++ examples/vhost_blk/blk.c | 125 +++ examples/vhost_blk/blk_spec.h | 95 +++ examples/vhost_blk/meson.build | 21 + examples/vhost_blk/vhost_blk.c | 1086 +++++++++++++++++++++++++ examples/vhost_blk/vhost_blk.h | 127 +++ examples/vhost_blk/vhost_blk_compat.c | 195 +++++ 7 files changed, 1717 insertions(+) create mode 100644 examples/vhost_blk/Makefile create mode 100644 examples/vhost_blk/blk.c create mode 100644 examples/vhost_blk/blk_spec.h create mode 100644 examples/vhost_blk/meson.build create mode 100644 examples/vhost_blk/vhost_blk.c create mode 100644 examples/vhost_blk/vhost_blk.h create mode 100644 examples/vhost_blk/vhost_blk_compat.c
examples/vhost_blk/Makefile new file +68 -0
@@ -0,0 +1,68 @@ +# SPDX-License-Identifier: BSD-3-Clause +# Copyright(c) 2010-2014 Intel Corporation + +# binary name +APP = vhost-blk + +# all source are stored in SRCS-y +SRCS-y := blk.c vhost_blk.c vhost_blk_compat.c + +# Build using pkg-config variables if possible +$(shell pkg-config --exists libdpdk) +ifeq ($(.SHELLSTATUS),0) + +all: shared +.PHONY: shared static +shared: build/$(APP)-shared + ln -sf $(APP)-shared build/$(APP) +static: build/$(APP)-static + ln -sf $(APP)-static build/$(APP) + +LDFLAGS += -pthread + +PC_FILE := $(shell pkg-config --path libdpdk) +CFLAGS += -O3 $(shell pkg-config --cflags libdpdk) +LDFLAGS_SHARED = $(shell pkg-config --libs libdpdk) +LDFLAGS_STATIC = -Wl,-Bstatic $(shell pkg-config --static --libs libdpdk) + +CFLAGS += -DALLOW_EXPERIMENTAL_API + +build/$(APP)-shared: $(SRCS-y) Makefile $(PC_FILE) | build + $(CC) $(CFLAGS) $(SRCS-y) -o $@ $(LDFLAGS) $(LDFLAGS_SHARED) + +build/$(APP)-static: $(SRCS-y) Makefile $(PC_FILE) | build + $(CC) $(CFLAGS) $(SRCS-y) -o $@ $(LDFLAGS) $(LDFLAGS_STATIC) + +build: + @mkdir -p $@ + +.PHONY: clean +clean: + rm -f build/$(APP) build/$(APP)-static build/$(APP)-shared + test -d build && rmdir -p build || true + +else # Build using legacy build system + +ifeq ($(RTE_SDK),) +$(error "Please define RTE_SDK environment variable") +endif + +# Default target, detect a build directory, by looking for a path with a .config +RTE_TARGET ?= $(notdir $(abspath $(dir $(firstword $(wildcard $(RTE_SDK)/*/.config))))) + +include $(RTE_SDK)/mk/rte.vars.mk + +ifneq ($(CONFIG_RTE_EXEC_ENV_LINUX),y) +$(info This application can only operate in a linux environment, \ +please change the definition of the RTE_TARGET environment variable) +all: +else + +CFLAGS += -DALLOW_EXPERIMENTAL_API +CFLAGS += -O2 -D_FILE_OFFSET_BITS=64 +CFLAGS += $(WERROR_FLAGS) + +include $(RTE_SDK)/mk/rte.extapp.mk + +endif +endif
examples/vhost_blk/blk.c new file +125 -0
125 lines not on this page · raw message
examples/vhost_blk/blk_spec.h new file +95 -0
95 lines not on this page · raw message
examples/vhost_blk/meson.build new file +21 -0
@@ -0,0 +1,21 @@ +# SPDX-License-Identifier: BSD-3-Clause +# Copyright(c) 2017 Intel Corporation + +# meson file, for building this example as part of a main DPDK build. +# +# To build this example as a standalone application with an already-installed +# DPDK instance, use 'make' + +if not is_linux + build = false +endif + +if not cc.has_header('linux/virtio_blk.h') + build = false +endif + +deps += 'vhost' +allow_experimental_apis = true +sources = files( + 'blk.c', 'vhost_blk.c', 'vhost_blk_compat.c' +)
examples/vhost_blk/vhost_blk.c new file +1086 -0
1,086 lines not on this page · raw message
examples/vhost_blk/vhost_blk.h new file +127 -0
127 lines not on this page · raw message
examples/vhost_blk/vhost_blk_compat.c new file +195 -0
195 lines not on this page · raw message
Signature · 1 line
2.17.2