diff options
Diffstat (limited to 'drivers/vdpa/vdpa_sim/vdpa_sim_blk.c')
| -rw-r--r-- | drivers/vdpa/vdpa_sim/vdpa_sim_blk.c | 271 |
1 files changed, 230 insertions, 41 deletions
diff --git a/drivers/vdpa/vdpa_sim/vdpa_sim_blk.c b/drivers/vdpa/vdpa_sim/vdpa_sim_blk.c index 42d401d43911..b137f3679343 100644 --- a/drivers/vdpa/vdpa_sim/vdpa_sim_blk.c +++ b/drivers/vdpa/vdpa_sim/vdpa_sim_blk.c @@ -11,7 +11,6 @@ #include <linux/module.h> #include <linux/device.h> #include <linux/kernel.h> -#include <linux/sched.h> #include <linux/blkdev.h> #include <linux/vringh.h> #include <linux/vdpa.h> @@ -25,31 +24,80 @@ #define DRV_LICENSE "GPL v2" #define VDPASIM_BLK_FEATURES (VDPASIM_FEATURES | \ + (1ULL << VIRTIO_BLK_F_FLUSH) | \ (1ULL << VIRTIO_BLK_F_SIZE_MAX) | \ (1ULL << VIRTIO_BLK_F_SEG_MAX) | \ (1ULL << VIRTIO_BLK_F_BLK_SIZE) | \ (1ULL << VIRTIO_BLK_F_TOPOLOGY) | \ - (1ULL << VIRTIO_BLK_F_MQ)) + (1ULL << VIRTIO_BLK_F_MQ) | \ + (1ULL << VIRTIO_BLK_F_DISCARD) | \ + (1ULL << VIRTIO_BLK_F_WRITE_ZEROES)) #define VDPASIM_BLK_CAPACITY 0x40000 #define VDPASIM_BLK_SIZE_MAX 0x1000 #define VDPASIM_BLK_SEG_MAX 32 +#define VDPASIM_BLK_DWZ_MAX_SECTORS UINT_MAX + +/* 1 virtqueue, 1 address space, 1 virtqueue group */ #define VDPASIM_BLK_VQ_NUM 1 +#define VDPASIM_BLK_AS_NUM 1 +#define VDPASIM_BLK_GROUP_NUM 1 + +struct vdpasim_blk { + struct vdpasim vdpasim; + void *buffer; + bool shared_backend; +}; + +static struct vdpasim_blk *sim_to_blk(struct vdpasim *vdpasim) +{ + return container_of(vdpasim, struct vdpasim_blk, vdpasim); +} static char vdpasim_blk_id[VIRTIO_BLK_ID_BYTES] = "vdpa_blk_sim"; -static bool vdpasim_blk_check_range(u64 start_sector, size_t range_size) +static bool shared_backend; +module_param(shared_backend, bool, 0444); +MODULE_PARM_DESC(shared_backend, "Enable the shared backend between virtio-blk devices"); + +static void *shared_buffer; +/* mutex to synchronize shared_buffer access */ +static DEFINE_MUTEX(shared_buffer_mutex); + +static void vdpasim_blk_buffer_lock(struct vdpasim_blk *blk) { - u64 range_sectors = range_size >> SECTOR_SHIFT; + if (blk->shared_backend) + mutex_lock(&shared_buffer_mutex); +} - if (range_size > VDPASIM_BLK_SIZE_MAX * VDPASIM_BLK_SEG_MAX) - return false; +static void vdpasim_blk_buffer_unlock(struct vdpasim_blk *blk) +{ + if (blk->shared_backend) + mutex_unlock(&shared_buffer_mutex); +} + +static bool vdpasim_blk_check_range(struct vdpasim *vdpasim, u64 start_sector, + u64 num_sectors, u64 max_sectors) +{ + if (start_sector > VDPASIM_BLK_CAPACITY) { + dev_dbg(&vdpasim->vdpa.dev, + "starting sector exceeds the capacity - start: 0x%llx capacity: 0x%x\n", + start_sector, VDPASIM_BLK_CAPACITY); + } - if (start_sector > VDPASIM_BLK_CAPACITY) + if (num_sectors > max_sectors) { + dev_dbg(&vdpasim->vdpa.dev, + "number of sectors exceeds the max allowed in a request - num: 0x%llx max: 0x%llx\n", + num_sectors, max_sectors); return false; + } - if (range_sectors > VDPASIM_BLK_CAPACITY - start_sector) + if (num_sectors > VDPASIM_BLK_CAPACITY - start_sector) { + dev_dbg(&vdpasim->vdpa.dev, + "request exceeds the capacity - start: 0x%llx num: 0x%llx capacity: 0x%x\n", + start_sector, num_sectors, VDPASIM_BLK_CAPACITY); return false; + } return true; } @@ -61,8 +109,10 @@ static bool vdpasim_blk_check_range(u64 start_sector, size_t range_size) static bool vdpasim_blk_handle_req(struct vdpasim *vdpasim, struct vdpasim_virtqueue *vq) { + struct vdpasim_blk *blk = sim_to_blk(vdpasim); size_t pushed = 0, to_pull, to_push; struct virtio_blk_outhdr hdr; + bool handled = false; ssize_t bytes; loff_t offset; u64 sector; @@ -76,14 +126,14 @@ static bool vdpasim_blk_handle_req(struct vdpasim *vdpasim, return false; if (vq->out_iov.used < 1 || vq->in_iov.used < 1) { - dev_err(&vdpasim->vdpa.dev, "missing headers - out_iov: %u in_iov %u\n", + dev_dbg(&vdpasim->vdpa.dev, "missing headers - out_iov: %u in_iov %u\n", vq->out_iov.used, vq->in_iov.used); - return false; + goto err; } if (vq->in_iov.iov[vq->in_iov.used - 1].iov_len < 1) { - dev_err(&vdpasim->vdpa.dev, "request in header too short\n"); - return false; + dev_dbg(&vdpasim->vdpa.dev, "request in header too short\n"); + goto err; } /* The last byte is the status and we checked if the last iov has @@ -96,8 +146,8 @@ static bool vdpasim_blk_handle_req(struct vdpasim *vdpasim, bytes = vringh_iov_pull_iotlb(&vq->vring, &vq->out_iov, &hdr, sizeof(hdr)); if (bytes != sizeof(hdr)) { - dev_err(&vdpasim->vdpa.dev, "request out header too short\n"); - return false; + dev_dbg(&vdpasim->vdpa.dev, "request out header too short\n"); + goto err; } to_pull -= bytes; @@ -107,21 +157,30 @@ static bool vdpasim_blk_handle_req(struct vdpasim *vdpasim, offset = sector << SECTOR_SHIFT; status = VIRTIO_BLK_S_OK; + if (type != VIRTIO_BLK_T_IN && type != VIRTIO_BLK_T_OUT && + sector != 0) { + dev_dbg(&vdpasim->vdpa.dev, + "sector must be 0 for %u request - sector: 0x%llx\n", + type, sector); + status = VIRTIO_BLK_S_IOERR; + goto err_status; + } + switch (type) { case VIRTIO_BLK_T_IN: - if (!vdpasim_blk_check_range(sector, to_push)) { - dev_err(&vdpasim->vdpa.dev, - "reading over the capacity - offset: 0x%llx len: 0x%zx\n", - offset, to_push); + if (!vdpasim_blk_check_range(vdpasim, sector, + to_push >> SECTOR_SHIFT, + VDPASIM_BLK_SIZE_MAX * VDPASIM_BLK_SEG_MAX)) { status = VIRTIO_BLK_S_IOERR; break; } + vdpasim_blk_buffer_lock(blk); bytes = vringh_iov_push_iotlb(&vq->vring, &vq->in_iov, - vdpasim->buffer + offset, - to_push); + blk->buffer + offset, to_push); + vdpasim_blk_buffer_unlock(blk); if (bytes < 0) { - dev_err(&vdpasim->vdpa.dev, + dev_dbg(&vdpasim->vdpa.dev, "vringh_iov_push_iotlb() error: %zd offset: 0x%llx len: 0x%zx\n", bytes, offset, to_push); status = VIRTIO_BLK_S_IOERR; @@ -132,19 +191,19 @@ static bool vdpasim_blk_handle_req(struct vdpasim *vdpasim, break; case VIRTIO_BLK_T_OUT: - if (!vdpasim_blk_check_range(sector, to_pull)) { - dev_err(&vdpasim->vdpa.dev, - "writing over the capacity - offset: 0x%llx len: 0x%zx\n", - offset, to_pull); + if (!vdpasim_blk_check_range(vdpasim, sector, + to_pull >> SECTOR_SHIFT, + VDPASIM_BLK_SIZE_MAX * VDPASIM_BLK_SEG_MAX)) { status = VIRTIO_BLK_S_IOERR; break; } + vdpasim_blk_buffer_lock(blk); bytes = vringh_iov_pull_iotlb(&vq->vring, &vq->out_iov, - vdpasim->buffer + offset, - to_pull); + blk->buffer + offset, to_pull); + vdpasim_blk_buffer_unlock(blk); if (bytes < 0) { - dev_err(&vdpasim->vdpa.dev, + dev_dbg(&vdpasim->vdpa.dev, "vringh_iov_pull_iotlb() error: %zd offset: 0x%llx len: 0x%zx\n", bytes, offset, to_pull); status = VIRTIO_BLK_S_IOERR; @@ -157,7 +216,7 @@ static bool vdpasim_blk_handle_req(struct vdpasim *vdpasim, vdpasim_blk_id, VIRTIO_BLK_ID_BYTES); if (bytes < 0) { - dev_err(&vdpasim->vdpa.dev, + dev_dbg(&vdpasim->vdpa.dev, "vringh_iov_push_iotlb() error: %zd\n", bytes); status = VIRTIO_BLK_S_IOERR; break; @@ -166,13 +225,78 @@ static bool vdpasim_blk_handle_req(struct vdpasim *vdpasim, pushed += bytes; break; + case VIRTIO_BLK_T_FLUSH: + /* nothing to do */ + break; + + case VIRTIO_BLK_T_DISCARD: + case VIRTIO_BLK_T_WRITE_ZEROES: { + struct virtio_blk_discard_write_zeroes range; + u32 num_sectors, flags; + + if (to_pull != sizeof(range)) { + dev_dbg(&vdpasim->vdpa.dev, + "discard/write_zeroes header len: 0x%zx [expected: 0x%zx]\n", + to_pull, sizeof(range)); + status = VIRTIO_BLK_S_IOERR; + break; + } + + bytes = vringh_iov_pull_iotlb(&vq->vring, &vq->out_iov, &range, + to_pull); + if (bytes < 0) { + dev_dbg(&vdpasim->vdpa.dev, + "vringh_iov_pull_iotlb() error: %zd offset: 0x%llx len: 0x%zx\n", + bytes, offset, to_pull); + status = VIRTIO_BLK_S_IOERR; + break; + } + + sector = le64_to_cpu(range.sector); + offset = sector << SECTOR_SHIFT; + num_sectors = le32_to_cpu(range.num_sectors); + flags = le32_to_cpu(range.flags); + + if (type == VIRTIO_BLK_T_DISCARD && flags != 0) { + dev_dbg(&vdpasim->vdpa.dev, + "discard unexpected flags set - flags: 0x%x\n", + flags); + status = VIRTIO_BLK_S_UNSUPP; + break; + } + + if (type == VIRTIO_BLK_T_WRITE_ZEROES && + flags & ~VIRTIO_BLK_WRITE_ZEROES_FLAG_UNMAP) { + dev_dbg(&vdpasim->vdpa.dev, + "write_zeroes unexpected flags set - flags: 0x%x\n", + flags); + status = VIRTIO_BLK_S_UNSUPP; + break; + } + + if (!vdpasim_blk_check_range(vdpasim, sector, num_sectors, + VDPASIM_BLK_DWZ_MAX_SECTORS)) { + status = VIRTIO_BLK_S_IOERR; + break; + } + + if (type == VIRTIO_BLK_T_WRITE_ZEROES) { + vdpasim_blk_buffer_lock(blk); + memset(blk->buffer + offset, 0, + num_sectors << SECTOR_SHIFT); + vdpasim_blk_buffer_unlock(blk); + } + + break; + } default: - dev_warn(&vdpasim->vdpa.dev, - "Unsupported request type %d\n", type); + dev_dbg(&vdpasim->vdpa.dev, + "Unsupported request type %d\n", type); status = VIRTIO_BLK_S_IOERR; break; } +err_status: /* If some operations fail, we need to skip the remaining bytes * to put the status in the last byte */ @@ -182,30 +306,37 @@ static bool vdpasim_blk_handle_req(struct vdpasim *vdpasim, /* Last byte is the status */ bytes = vringh_iov_push_iotlb(&vq->vring, &vq->in_iov, &status, 1); if (bytes != 1) - return false; + goto err; pushed += bytes; /* Make sure data is wrote before advancing index */ smp_wmb(); + handled = true; + +err: vringh_complete_iotlb(&vq->vring, vq->head, pushed); - return true; + return handled; } -static void vdpasim_blk_work(struct work_struct *work) +static void vdpasim_blk_work(struct vdpasim *vdpasim) { - struct vdpasim *vdpasim = container_of(work, struct vdpasim, work); + bool reschedule = false; int i; - spin_lock(&vdpasim->lock); + mutex_lock(&vdpasim->mutex); if (!(vdpasim->status & VIRTIO_CONFIG_S_DRIVER_OK)) goto out; + if (!vdpasim->running) + goto out; + for (i = 0; i < VDPASIM_BLK_VQ_NUM; i++) { struct vdpasim_virtqueue *vq = &vdpasim->vqs[i]; + int reqs = 0; if (!vq->ready) continue; @@ -218,10 +349,18 @@ static void vdpasim_blk_work(struct work_struct *work) if (vringh_need_notify_iotlb(&vq->vring) > 0) vringh_notify(&vq->vring); local_bh_enable(); + + if (++reqs > 4) { + reschedule = true; + break; + } } } out: - spin_unlock(&vdpasim->lock); + mutex_unlock(&vdpasim->mutex); + + if (reschedule) + vdpasim_schedule_work(vdpasim); } static void vdpasim_blk_get_config(struct vdpasim *vdpasim, void *config) @@ -237,6 +376,25 @@ static void vdpasim_blk_get_config(struct vdpasim *vdpasim, void *config) blk_config->min_io_size = cpu_to_vdpasim16(vdpasim, 1); blk_config->opt_io_size = cpu_to_vdpasim32(vdpasim, 1); blk_config->blk_size = cpu_to_vdpasim32(vdpasim, SECTOR_SIZE); + /* VIRTIO_BLK_F_DISCARD */ + blk_config->discard_sector_alignment = + cpu_to_vdpasim32(vdpasim, SECTOR_SIZE); + blk_config->max_discard_sectors = + cpu_to_vdpasim32(vdpasim, VDPASIM_BLK_DWZ_MAX_SECTORS); + blk_config->max_discard_seg = cpu_to_vdpasim32(vdpasim, 1); + /* VIRTIO_BLK_F_WRITE_ZEROES */ + blk_config->max_write_zeroes_sectors = + cpu_to_vdpasim32(vdpasim, VDPASIM_BLK_DWZ_MAX_SECTORS); + blk_config->max_write_zeroes_seg = cpu_to_vdpasim32(vdpasim, 1); + +} + +static void vdpasim_blk_free(struct vdpasim *vdpasim) +{ + struct vdpasim_blk *blk = sim_to_blk(vdpasim); + + if (!blk->shared_backend) + kvfree(blk->buffer); } static void vdpasim_blk_mgmtdev_release(struct device *dev) @@ -252,6 +410,7 @@ static int vdpasim_blk_dev_add(struct vdpa_mgmt_dev *mdev, const char *name, const struct vdpa_dev_set_config *config) { struct vdpasim_dev_attr dev_attr = {}; + struct vdpasim_blk *blk; struct vdpasim *simdev; int ret; @@ -260,15 +419,32 @@ static int vdpasim_blk_dev_add(struct vdpa_mgmt_dev *mdev, const char *name, dev_attr.id = VIRTIO_ID_BLOCK; dev_attr.supported_features = VDPASIM_BLK_FEATURES; dev_attr.nvqs = VDPASIM_BLK_VQ_NUM; + dev_attr.ngroups = VDPASIM_BLK_GROUP_NUM; + dev_attr.nas = VDPASIM_BLK_AS_NUM; + dev_attr.alloc_size = sizeof(struct vdpasim_blk); dev_attr.config_size = sizeof(struct virtio_blk_config); dev_attr.get_config = vdpasim_blk_get_config; dev_attr.work_fn = vdpasim_blk_work; - dev_attr.buffer_size = VDPASIM_BLK_CAPACITY << SECTOR_SHIFT; + dev_attr.free = vdpasim_blk_free; - simdev = vdpasim_create(&dev_attr); + simdev = vdpasim_create(&dev_attr, config); if (IS_ERR(simdev)) return PTR_ERR(simdev); + blk = sim_to_blk(simdev); + blk->shared_backend = shared_backend; + + if (blk->shared_backend) { + blk->buffer = shared_buffer; + } else { + blk->buffer = kvzalloc(VDPASIM_BLK_CAPACITY << SECTOR_SHIFT, + GFP_KERNEL); + if (!blk->buffer) { + ret = -ENOMEM; + goto put_dev; + } + } + ret = _vdpa_register_device(&simdev->vdpa, VDPASIM_BLK_VQ_NUM); if (ret) goto put_dev; @@ -309,15 +485,27 @@ static int __init vdpasim_blk_init(void) int ret; ret = device_register(&vdpasim_blk_mgmtdev); - if (ret) + if (ret) { + put_device(&vdpasim_blk_mgmtdev); return ret; + } ret = vdpa_mgmtdev_register(&mgmt_dev); if (ret) goto parent_err; - return 0; + if (shared_backend) { + shared_buffer = kvzalloc(VDPASIM_BLK_CAPACITY << SECTOR_SHIFT, + GFP_KERNEL); + if (!shared_buffer) { + ret = -ENOMEM; + goto mgmt_dev_err; + } + } + return 0; +mgmt_dev_err: + vdpa_mgmtdev_unregister(&mgmt_dev); parent_err: device_unregister(&vdpasim_blk_mgmtdev); return ret; @@ -325,6 +513,7 @@ parent_err: static void __exit vdpasim_blk_exit(void) { + kvfree(shared_buffer); vdpa_mgmtdev_unregister(&mgmt_dev); device_unregister(&vdpasim_blk_mgmtdev); } |