[RFC PATCH 2/6] nvme: wire-up support for async-passthru on char-device.
Kanchan Joshi
joshi.k at samsung.com
Thu Aug 5 05:55:35 PDT 2021
Introduce handlers for fops->uring_cmd(), implementing async passthru
on char device (including the multipath one).
The handlers supports NVME_IOCTL_IO_CMD and NVME_IOCTL_IO64_CMD.
Signed-off-by: Kanchan Joshi <joshi.k at samsung.com>
Signed-off-by: Anuj Gupta <anuj20.g at samsung.com>
---
drivers/nvme/host/core.c | 1 +
drivers/nvme/host/ioctl.c | 177 +++++++++++++++++++++++++++++++---
drivers/nvme/host/multipath.c | 1 +
drivers/nvme/host/nvme.h | 5 +
4 files changed, 169 insertions(+), 15 deletions(-)
diff --git a/drivers/nvme/host/core.c b/drivers/nvme/host/core.c
index 11779be42186..58fe7f6c94ba 100644
--- a/drivers/nvme/host/core.c
+++ b/drivers/nvme/host/core.c
@@ -3559,6 +3559,7 @@ static const struct file_operations nvme_ns_chr_fops = {
.release = nvme_ns_chr_release,
.unlocked_ioctl = nvme_ns_chr_ioctl,
.compat_ioctl = compat_ptr_ioctl,
+ .uring_cmd = nvme_ns_chr_async_ioctl,
};
static int nvme_add_ns_cdev(struct nvme_ns *ns)
diff --git a/drivers/nvme/host/ioctl.c b/drivers/nvme/host/ioctl.c
index 305ddd415e45..2730c5dfdf78 100644
--- a/drivers/nvme/host/ioctl.c
+++ b/drivers/nvme/host/ioctl.c
@@ -18,6 +18,93 @@ static void __user *nvme_to_user_ptr(uintptr_t ptrval)
ptrval = (compat_uptr_t)ptrval;
return (void __user *)ptrval;
}
+/*
+ * This is carved within the io_uring_cmd, to avoid dynamic allocation.
+ * Care should be taken not to grow this beyond what is available.
+ * Expect build warning otherwise.
+ */
+struct uring_cmd_data {
+ union {
+ struct bio *bio;
+ u64 result; /* nvme cmd result */
+ };
+ void *meta; /* kernel-resident buffer */
+ int status; /* nvme cmd status */
+};
+
+inline u64 *nvme_ioucmd_data_addr(struct io_uring_cmd *ioucmd)
+{
+ return &(((struct block_uring_cmd *)&ioucmd->pdu)->unused2[1]);
+}
+
+static void nvme_pt_task_cb(struct io_uring_cmd *ioucmd)
+{
+ struct uring_cmd_data *ucd;
+ struct nvme_passthru_cmd64 __user *ptcmd64 = NULL;
+ struct block_uring_cmd *bcmd;
+
+ bcmd = (struct block_uring_cmd *) &ioucmd->pdu;
+ ptcmd64 = (void __user *) bcmd->unused2[0];
+ ucd = (struct uring_cmd_data *) nvme_ioucmd_data_addr(ioucmd);
+
+ if (ucd->meta) {
+ void __user *umeta = nvme_to_user_ptr(ptcmd64->metadata);
+
+ if (!ucd->status)
+ if (copy_to_user(umeta, ucd->meta, ptcmd64->metadata_len))
+ ucd->status = -EFAULT;
+ kfree(ucd->meta);
+ }
+ if (likely(bcmd->ioctl_cmd == NVME_IOCTL_IO64_CMD)) {
+ if (put_user(ucd->result, &ptcmd64->result))
+ ucd->status = -EFAULT;
+ } else {
+ struct nvme_passthru_cmd __user *ptcmd = (void *)bcmd->unused2[0];
+
+ if (put_user(ucd->result, &ptcmd->result))
+ ucd->status = -EFAULT;
+ }
+ io_uring_cmd_done(ioucmd, ucd->status);
+}
+
+static void nvme_end_async_pt(struct request *req, blk_status_t err)
+{
+ struct io_uring_cmd *ioucmd;
+ struct uring_cmd_data *ucd;
+ struct bio *bio;
+
+ ioucmd = req->end_io_data;
+ ucd = (struct uring_cmd_data *) nvme_ioucmd_data_addr(ioucmd);
+ /* extract bio before reusing the same field for status */
+ bio = ucd->bio;
+
+ if (nvme_req(req)->flags & NVME_REQ_CANCELLED)
+ ucd->status = -EINTR;
+ else
+ ucd->status = nvme_req(req)->status;
+ ucd->result = le64_to_cpu(nvme_req(req)->result.u64);
+
+ /* this takes care of setting up task-work */
+ io_uring_cmd_complete_in_task(ioucmd, nvme_pt_task_cb);
+
+ /* we can unmap pages, free bio and request */
+ blk_rq_unmap_user(bio);
+ blk_mq_free_request(req);
+}
+
+static void nvme_setup_uring_cmd_data(struct request *rq,
+ struct io_uring_cmd *ioucmd, void *meta, bool write)
+{
+ struct uring_cmd_data *ucd;
+
+ ucd = (struct uring_cmd_data *) nvme_ioucmd_data_addr(ioucmd);
+ /* to free bio on completion, as req->bio will be null at that time */
+ ucd->bio = rq->bio;
+ /* meta update is required only for read requests */
+ if (meta && !write)
+ ucd->meta = meta;
+ rq->end_io_data = ioucmd;
+}
static void *nvme_add_user_metadata(struct bio *bio, void __user *ubuf,
unsigned len, u32 seed, bool write)
@@ -56,7 +143,8 @@ static void *nvme_add_user_metadata(struct bio *bio, void __user *ubuf,
static int nvme_submit_user_cmd(struct request_queue *q,
struct nvme_command *cmd, void __user *ubuffer,
unsigned bufflen, void __user *meta_buffer, unsigned meta_len,
- u32 meta_seed, u64 *result, unsigned timeout)
+ u32 meta_seed, u64 *result, unsigned timeout,
+ struct io_uring_cmd *ioucmd)
{
bool write = nvme_is_write(cmd);
struct nvme_ns *ns = q->queuedata;
@@ -92,6 +180,12 @@ static int nvme_submit_user_cmd(struct request_queue *q,
req->cmd_flags |= REQ_INTEGRITY;
}
}
+ if (ioucmd) { /* async dispatch */
+ nvme_setup_uring_cmd_data(req, ioucmd, meta, write);
+ blk_execute_rq_nowait(ns ? ns->disk : NULL, req, 0,
+ nvme_end_async_pt);
+ return 0;
+ }
ret = nvme_execute_passthru_rq(req);
if (result)
@@ -170,7 +264,8 @@ static int nvme_submit_io(struct nvme_ns *ns, struct nvme_user_io __user *uio)
return nvme_submit_user_cmd(ns->queue, &c,
nvme_to_user_ptr(io.addr), length,
- metadata, meta_len, lower_32_bits(io.slba), NULL, 0);
+ metadata, meta_len, lower_32_bits(io.slba), NULL, 0,
+ NULL);
}
static bool nvme_validate_passthru_nsid(struct nvme_ctrl *ctrl,
@@ -188,7 +283,8 @@ static bool nvme_validate_passthru_nsid(struct nvme_ctrl *ctrl,
}
static int nvme_user_cmd(struct nvme_ctrl *ctrl, struct nvme_ns *ns,
- struct nvme_passthru_cmd __user *ucmd)
+ struct nvme_passthru_cmd __user *ucmd,
+ struct io_uring_cmd *ioucmd)
{
struct nvme_passthru_cmd cmd;
struct nvme_command c;
@@ -224,9 +320,9 @@ static int nvme_user_cmd(struct nvme_ctrl *ctrl, struct nvme_ns *ns,
status = nvme_submit_user_cmd(ns ? ns->queue : ctrl->admin_q, &c,
nvme_to_user_ptr(cmd.addr), cmd.data_len,
nvme_to_user_ptr(cmd.metadata), cmd.metadata_len,
- 0, &result, timeout);
+ 0, &result, timeout, ioucmd);
- if (status >= 0) {
+ if (!ioucmd && status >= 0) {
if (put_user(result, &ucmd->result))
return -EFAULT;
}
@@ -235,7 +331,8 @@ static int nvme_user_cmd(struct nvme_ctrl *ctrl, struct nvme_ns *ns,
}
static int nvme_user_cmd64(struct nvme_ctrl *ctrl, struct nvme_ns *ns,
- struct nvme_passthru_cmd64 __user *ucmd)
+ struct nvme_passthru_cmd64 __user *ucmd,
+ struct io_uring_cmd *ioucmd)
{
struct nvme_passthru_cmd64 cmd;
struct nvme_command c;
@@ -270,9 +367,9 @@ static int nvme_user_cmd64(struct nvme_ctrl *ctrl, struct nvme_ns *ns,
status = nvme_submit_user_cmd(ns ? ns->queue : ctrl->admin_q, &c,
nvme_to_user_ptr(cmd.addr), cmd.data_len,
nvme_to_user_ptr(cmd.metadata), cmd.metadata_len,
- 0, &cmd.result, timeout);
+ 0, &cmd.result, timeout, ioucmd);
- if (status >= 0) {
+ if (!ioucmd && status >= 0) {
if (put_user(cmd.result, &ucmd->result))
return -EFAULT;
}
@@ -294,9 +391,9 @@ static int nvme_ctrl_ioctl(struct nvme_ctrl *ctrl, unsigned int cmd,
{
switch (cmd) {
case NVME_IOCTL_ADMIN_CMD:
- return nvme_user_cmd(ctrl, NULL, argp);
+ return nvme_user_cmd(ctrl, NULL, argp, NULL);
case NVME_IOCTL_ADMIN64_CMD:
- return nvme_user_cmd64(ctrl, NULL, argp);
+ return nvme_user_cmd64(ctrl, NULL, argp, NULL);
default:
return sed_ioctl(ctrl->opal_dev, cmd, argp);
}
@@ -328,7 +425,7 @@ static int nvme_ns_ioctl(struct nvme_ns *ns, unsigned int cmd,
force_successful_syscall_return();
return ns->head->ns_id;
case NVME_IOCTL_IO_CMD:
- return nvme_user_cmd(ns->ctrl, ns, argp);
+ return nvme_user_cmd(ns->ctrl, ns, argp, NULL);
/*
* struct nvme_user_io can have different padding on some 32-bit ABIs.
* Just accept the compat version as all fields that are used are the
@@ -340,7 +437,7 @@ static int nvme_ns_ioctl(struct nvme_ns *ns, unsigned int cmd,
case NVME_IOCTL_SUBMIT_IO:
return nvme_submit_io(ns, argp);
case NVME_IOCTL_IO64_CMD:
- return nvme_user_cmd64(ns->ctrl, ns, argp);
+ return nvme_user_cmd64(ns->ctrl, ns, argp, NULL);
default:
if (!ns->ndev)
return -ENOTTY;
@@ -371,6 +468,41 @@ long nvme_ns_chr_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
return __nvme_ioctl(ns, cmd, (void __user *)arg);
}
+static int nvme_ns_async_ioctl(struct nvme_ns *ns, struct io_uring_cmd *ioucmd)
+{
+ struct block_uring_cmd *bcmd = (struct block_uring_cmd *)&ioucmd->pdu;
+ void __user *argp = (void __user *) bcmd->unused2[0];
+ int ret;
+
+ BUILD_BUG_ON(sizeof(struct uring_cmd_data) >
+ sizeof(struct block_uring_cmd) -
+ offsetof(struct block_uring_cmd, unused2[1]));
+
+ switch (bcmd->ioctl_cmd) {
+ case NVME_IOCTL_IO_CMD:
+ ret = nvme_user_cmd(ns->ctrl, ns, argp, ioucmd);
+ break;
+ case NVME_IOCTL_IO64_CMD:
+ ret = nvme_user_cmd64(ns->ctrl, ns, argp, ioucmd);
+ break;
+ default:
+ ret = -ENOTTY;
+ }
+
+ if (ret >= 0)
+ ret = -EIOCBQUEUED;
+ return ret;
+}
+
+int nvme_ns_chr_async_ioctl(struct io_uring_cmd *ioucmd,
+ enum io_uring_cmd_flags flags)
+{
+ struct nvme_ns *ns = container_of(file_inode(ioucmd->file)->i_cdev,
+ struct nvme_ns, cdev);
+
+ return nvme_ns_async_ioctl(ns, ioucmd);
+}
+
#ifdef CONFIG_NVME_MULTIPATH
static int nvme_ns_head_ctrl_ioctl(struct nvme_ns *ns, unsigned int cmd,
void __user *argp, struct nvme_ns_head *head, int srcu_idx)
@@ -387,6 +519,21 @@ static int nvme_ns_head_ctrl_ioctl(struct nvme_ns *ns, unsigned int cmd,
return ret;
}
+int nvme_ns_head_chr_async_ioctl(struct io_uring_cmd *ioucmd,
+ enum io_uring_cmd_flags flags)
+{
+ struct cdev *cdev = file_inode(ioucmd->file)->i_cdev;
+ struct nvme_ns_head *head = container_of(cdev, struct nvme_ns_head, cdev);
+ int srcu_idx = srcu_read_lock(&head->srcu);
+ struct nvme_ns *ns = nvme_find_path(head);
+ int ret = -EWOULDBLOCK;
+
+ if (ns)
+ ret = nvme_ns_async_ioctl(ns, ioucmd);
+ srcu_read_unlock(&head->srcu, srcu_idx);
+ return ret;
+}
+
int nvme_ns_head_ioctl(struct block_device *bdev, fmode_t mode,
unsigned int cmd, unsigned long arg)
{
@@ -463,7 +610,7 @@ static int nvme_dev_user_cmd(struct nvme_ctrl *ctrl, void __user *argp)
kref_get(&ns->kref);
up_read(&ctrl->namespaces_rwsem);
- ret = nvme_user_cmd(ctrl, ns, argp);
+ ret = nvme_user_cmd(ctrl, ns, argp, NULL);
nvme_put_ns(ns);
return ret;
@@ -480,9 +627,9 @@ long nvme_dev_ioctl(struct file *file, unsigned int cmd,
switch (cmd) {
case NVME_IOCTL_ADMIN_CMD:
- return nvme_user_cmd(ctrl, NULL, argp);
+ return nvme_user_cmd(ctrl, NULL, argp, NULL);
case NVME_IOCTL_ADMIN64_CMD:
- return nvme_user_cmd64(ctrl, NULL, argp);
+ return nvme_user_cmd64(ctrl, NULL, argp, NULL);
case NVME_IOCTL_IO_CMD:
return nvme_dev_user_cmd(ctrl, argp);
case NVME_IOCTL_RESET:
diff --git a/drivers/nvme/host/multipath.c b/drivers/nvme/host/multipath.c
index 0ea5298469c3..e89a6334249a 100644
--- a/drivers/nvme/host/multipath.c
+++ b/drivers/nvme/host/multipath.c
@@ -403,6 +403,7 @@ static const struct file_operations nvme_ns_head_chr_fops = {
.release = nvme_ns_head_chr_release,
.unlocked_ioctl = nvme_ns_head_chr_ioctl,
.compat_ioctl = compat_ptr_ioctl,
+ .uring_cmd = nvme_ns_head_chr_async_ioctl,
};
static int nvme_add_ns_head_cdev(struct nvme_ns_head *head)
diff --git a/drivers/nvme/host/nvme.h b/drivers/nvme/host/nvme.h
index 18ef8dd03a90..e43e73eda4c6 100644
--- a/drivers/nvme/host/nvme.h
+++ b/drivers/nvme/host/nvme.h
@@ -17,6 +17,7 @@
#include <linux/rcupdate.h>
#include <linux/wait.h>
#include <linux/t10-pi.h>
+#include <linux/io_uring.h>
#include <trace/events/block.h>
@@ -688,6 +689,10 @@ long nvme_ns_head_chr_ioctl(struct file *file, unsigned int cmd,
unsigned long arg);
long nvme_dev_ioctl(struct file *file, unsigned int cmd,
unsigned long arg);
+int nvme_ns_chr_async_ioctl(struct io_uring_cmd *ucmd,
+ enum io_uring_cmd_flags flags);
+int nvme_ns_head_chr_async_ioctl(struct io_uring_cmd *ucmd,
+ enum io_uring_cmd_flags flags);
int nvme_getgeo(struct block_device *bdev, struct hd_geometry *geo);
extern const struct attribute_group *nvme_ns_id_attr_groups[];
--
2.25.1
More information about the Linux-nvme
mailing list