From: Jack Morgenstein <ja...@dev.mellanox.co.il>
Add support for core userspace XRC operations (alloc/dealloc XRC
domain, create XRC SRQ), including adding an ABI for marshalling
requests and responses.
Signed-off-by: Jack Morgenstein <ja...@dev.mellanox.co.il>
Signed-off-by: Roland Dreier <rola...@cisco.com>
---
drivers/infiniband/core/uverbs.h | 4 +
drivers/infiniband/core/uverbs_cmd.c | 274 ++++++++++++++++++++++++++++++++-
drivers/infiniband/core/uverbs_main.c | 35 +++--
include/rdma/ib_user_verbs.h | 35 ++++-
4 files changed, 334 insertions(+), 14 deletions(-)
diff --git a/drivers/infiniband/core/uverbs.h b/drivers/infiniband/core/uverbs.h
index b3ea958..f9c051e 100644
--- a/drivers/infiniband/core/uverbs.h
+++ b/drivers/infiniband/core/uverbs.h
@@ -141,6 +141,7 @@ extern struct idr ib_uverbs_ah_idr;
extern struct idr ib_uverbs_cq_idr;
extern struct idr ib_uverbs_qp_idr;
extern struct idr ib_uverbs_srq_idr;
+extern struct idr ib_uverbs_xrcd_idr;
void idr_remove_uobj(struct idr *idp, struct ib_uobject *uobj);
@@ -194,5 +195,8 @@ IB_UVERBS_DECLARE_CMD(create_srq);
IB_UVERBS_DECLARE_CMD(modify_srq);
IB_UVERBS_DECLARE_CMD(query_srq);
IB_UVERBS_DECLARE_CMD(destroy_srq);
+IB_UVERBS_DECLARE_CMD(create_xrc_srq);
+IB_UVERBS_DECLARE_CMD(open_xrcd);
+IB_UVERBS_DECLARE_CMD(close_xrcd);
#endif /* UVERBS_H */
diff --git a/drivers/infiniband/core/uverbs_cmd.c
b/drivers/infiniband/core/uverbs_cmd.c
index 112d397..92f9f11 100644
--- a/drivers/infiniband/core/uverbs_cmd.c
+++ b/drivers/infiniband/core/uverbs_cmd.c
@@ -46,6 +46,7 @@ static struct lock_class_key cq_lock_key;
static struct lock_class_key qp_lock_key;
static struct lock_class_key ah_lock_key;
static struct lock_class_key srq_lock_key;
+static struct lock_class_key xrcd_lock_key;
#define INIT_UDATA(udata, ibuf, obuf, ilen, olen) \
do { \
@@ -254,6 +255,17 @@ static void put_srq_read(struct ib_srq *srq)
put_uobj_read(srq->uobject);
}
+static struct ib_xrcd *idr_read_xrcd(int xrcd_handle,
+ struct ib_ucontext *context)
+{
+ return idr_read_obj(&ib_uverbs_xrcd_idr, xrcd_handle, context, 0);
+}
+
+static void put_xrcd_read(struct ib_xrcd *xrcd)
+{
+ put_uobj_read(xrcd->uobject);
+}
+
ssize_t ib_uverbs_get_context(struct ib_uverbs_file *file,
const char __user *buf,
int in_len, int out_len)
@@ -297,6 +309,7 @@ ssize_t ib_uverbs_get_context(struct ib_uverbs_file *file,
INIT_LIST_HEAD(&ucontext->qp_list);
INIT_LIST_HEAD(&ucontext->srq_list);
INIT_LIST_HEAD(&ucontext->ah_list);
+ INIT_LIST_HEAD(&ucontext->xrcd_list);
ucontext->closing = 0;
resp.num_comp_vectors = file->device->num_comp_vectors;
@@ -1026,6 +1039,7 @@ ssize_t ib_uverbs_create_qp(struct ib_uverbs_file *file,
struct ib_srq *srq;
struct ib_qp *qp;
struct ib_qp_init_attr attr;
+ struct ib_xrcd *xrcd;
int ret;
if (out_len < sizeof resp)
@@ -1045,13 +1059,17 @@ ssize_t ib_uverbs_create_qp(struct ib_uverbs_file *file,
init_uobj(&obj->uevent.uobject, cmd.user_handle, file->ucontext,
&qp_lock_key);
down_write(&obj->uevent.uobject.mutex);
- srq = cmd.is_srq ? idr_read_srq(cmd.srq_handle, file->ucontext) : NULL;
+ srq = (cmd.is_srq && cmd.qp_type != IB_QPT_XRC) ?
+ idr_read_srq(cmd.srq_handle, file->ucontext) : NULL;
+ xrcd = cmd.qp_type == IB_QPT_XRC ?
+ idr_read_xrcd(cmd.srq_handle, file->ucontext) : NULL;
pd = idr_read_pd(cmd.pd_handle, file->ucontext);
scq = idr_read_cq(cmd.send_cq_handle, file->ucontext, 0);
rcq = cmd.recv_cq_handle == cmd.send_cq_handle ?
scq : idr_read_cq(cmd.recv_cq_handle, file->ucontext, 1);
- if (!pd || !scq || !rcq || (cmd.is_srq && !srq)) {
+ if (!pd || !scq || !rcq || (cmd.is_srq && !srq) ||
+ (cmd.qp_type == IB_QPT_XRC && !xrcd)) {
ret = -EINVAL;
goto err_put;
}
@@ -1063,6 +1081,7 @@ ssize_t ib_uverbs_create_qp(struct ib_uverbs_file *file,
attr.srq = srq;
attr.sq_sig_type = cmd.sq_sig_all ? IB_SIGNAL_ALL_WR :
IB_SIGNAL_REQ_WR;
attr.qp_type = cmd.qp_type;
+ attr.xrcd = xrcd;
attr.create_flags = 0;
attr.cap.max_send_wr = cmd.max_send_wr;
@@ -1090,11 +1109,14 @@ ssize_t ib_uverbs_create_qp(struct ib_uverbs_file *file,
qp->event_handler = attr.event_handler;
qp->qp_context = attr.qp_context;
qp->qp_type = attr.qp_type;
+ qp->xrcd = attr.xrcd;
atomic_inc(&pd->usecnt);
atomic_inc(&attr.send_cq->usecnt);
atomic_inc(&attr.recv_cq->usecnt);
if (attr.srq)
atomic_inc(&attr.srq->usecnt);
+ else if (attr.xrcd)
+ atomic_inc(&attr.xrcd->usecnt);
obj->uevent.uobject.object = qp;
ret = idr_add_uobj(&ib_uverbs_qp_idr, &obj->uevent.uobject);
@@ -1122,6 +1144,8 @@ ssize_t ib_uverbs_create_qp(struct ib_uverbs_file *file,
put_cq_read(rcq);
if (srq)
put_srq_read(srq);
+ if (xrcd)
+ put_xrcd_read(xrcd);
mutex_lock(&file->mutex);
list_add_tail(&obj->uevent.uobject.list, &file->ucontext->qp_list);
@@ -1148,6 +1172,8 @@ err_put:
put_cq_read(rcq);
if (srq)
put_srq_read(srq);
+ if (xrcd)
+ put_xrcd_read(xrcd);
put_uobj_write(&obj->uevent.uobject);
return ret;
@@ -2000,6 +2026,8 @@ ssize_t ib_uverbs_create_srq(struct ib_uverbs_file *file,
srq->uobject = &obj->uobject;
srq->event_handler = attr.event_handler;
srq->srq_context = attr.srq_context;
+ srq->xrc_cq = NULL;
+ srq->xrcd = NULL;
atomic_inc(&pd->usecnt);
atomic_set(&srq->usecnt, 0);
@@ -2045,6 +2073,134 @@ err:
return ret;
}
+ssize_t ib_uverbs_create_xrc_srq(struct ib_uverbs_file *file,
+ const char __user *buf, int in_len,
+ int out_len)
+{
+ struct ib_uverbs_create_xrc_srq cmd;
+ struct ib_uverbs_create_srq_resp resp;
+ struct ib_udata udata;
+ struct ib_uevent_object *obj;
+ struct ib_pd *pd;
+ struct ib_srq *srq;
+ struct ib_cq *xrc_cq;
+ struct ib_xrcd *xrcd;
+ struct ib_srq_init_attr attr;
+ int ret;
+
+ if (out_len < sizeof resp)
+ return -ENOSPC;
+
+ if (copy_from_user(&cmd, buf, sizeof cmd))
+ return -EFAULT;
+
+ INIT_UDATA(&udata, buf + sizeof cmd,
+ (unsigned long) cmd.response + sizeof resp,
+ in_len - sizeof cmd, out_len - sizeof resp);
+
+ obj = kmalloc(sizeof *obj, GFP_KERNEL);
+ if (!obj)
+ return -ENOMEM;
+
+ init_uobj(&obj->uobject, cmd.user_handle, file->ucontext,
&srq_lock_key);
+ down_write(&obj->uobject.mutex);
+
+ pd = idr_read_pd(cmd.pd_handle, file->ucontext);
+ if (!pd) {
+ ret = -EINVAL;
+ goto err;
+ }
+
+ xrc_cq = idr_read_cq(cmd.xrc_cq, file->ucontext, 0);
+ if (!xrc_cq) {
+ ret = -EINVAL;
+ goto err_put_pd;
+ }
+
+ xrcd = idr_read_xrcd(cmd.xrcd_handle, file->ucontext);
+ if (!xrcd) {
+ ret = -EINVAL;
+ goto err_put_cq;
+ }
+
+ attr.event_handler = ib_uverbs_srq_event_handler;
+ attr.srq_context = file;
+ attr.attr.max_wr = cmd.max_wr;
+ attr.attr.max_sge = cmd.max_sge;
+ attr.attr.srq_limit = cmd.srq_limit;
+
+ obj->events_reported = 0;
+ INIT_LIST_HEAD(&obj->event_list);
+
+ srq = pd->device->create_xrc_srq(pd, xrc_cq, xrcd, &attr, &udata);
+ if (IS_ERR(srq)) {
+ ret = PTR_ERR(srq);
+ goto err_put;
+ }
+
+ srq->device = pd->device;
+ srq->pd = pd;
+ srq->uobject = &obj->uobject;
+ srq->event_handler = attr.event_handler;
+ srq->srq_context = attr.srq_context;
+ srq->xrc_cq = xrc_cq;
+ srq->xrcd = xrcd;
+ atomic_inc(&pd->usecnt);
+ atomic_inc(&xrc_cq->usecnt);
+ atomic_inc(&xrcd->usecnt);
+
+ atomic_set(&srq->usecnt, 0);
+
+ obj->uobject.object = srq;
+ ret = idr_add_uobj(&ib_uverbs_srq_idr, &obj->uobject);
+ if (ret)
+ goto err_destroy;
+
+ memset(&resp, 0, sizeof resp);
+ resp.srq_handle = obj->uobject.id;
+ resp.max_wr = attr.attr.max_wr;
+ resp.max_sge = attr.attr.max_sge;
+
+ if (copy_to_user((void __user *) (unsigned long) cmd.response,
+ &resp, sizeof resp)) {
+ ret = -EFAULT;
+ goto err_copy;
+ }
+
+ put_xrcd_read(xrcd);
+ put_cq_read(xrc_cq);
+ put_pd_read(pd);
+
+ mutex_lock(&file->mutex);
+ list_add_tail(&obj->uobject.list, &file->ucontext->srq_list);
+ mutex_unlock(&file->mutex);
+
+ obj->uobject.live = 1;
+
+ up_write(&obj->uobject.mutex);
+
+ return in_len;
+
+err_copy:
+ idr_remove_uobj(&ib_uverbs_srq_idr, &obj->uobject);
+
+err_destroy:
+ ib_destroy_srq(srq);
+
+err_put:
+ put_xrcd_read(xrcd);
+
+err_put_cq:
+ put_cq_read(xrc_cq);
+
+err_put_pd:
+ put_pd_read(pd);
+
+err:
+ put_uobj_write(&obj->uobject);
+ return ret;
+}
+
ssize_t ib_uverbs_modify_srq(struct ib_uverbs_file *file,
const char __user *buf, int in_len,
int out_len)
@@ -2163,3 +2319,117 @@ ssize_t ib_uverbs_destroy_srq(struct ib_uverbs_file
*file,
return ret ? ret : in_len;
}
+
+ssize_t ib_uverbs_open_xrcd(struct ib_uverbs_file *file,
+ const char __user *buf, int in_len,
+ int out_len)
+{
+ struct ib_uverbs_open_xrcd cmd;
+ struct ib_uverbs_open_xrcd_resp resp;
+ struct ib_udata udata;
+ struct ib_uobject *uobj;
+ struct ib_xrcd *xrcd;
+ int ret;
+
+ if (out_len < sizeof resp)
+ return -ENOSPC;
+
+ if (copy_from_user(&cmd, buf, sizeof cmd))
+ return -EFAULT;
+
+ /* file descriptors/inodes not yet implemented */
+ if (cmd.fd != -1)
+ return -ENOSYS;
+
+ INIT_UDATA(&udata, buf + sizeof cmd,
+ (unsigned long) cmd.response + sizeof resp,
+ in_len - sizeof cmd, out_len - sizeof resp);
+
+ uobj = kmalloc(sizeof *uobj, GFP_KERNEL);
+ if (!uobj)
+ return -ENOMEM;
+
+ init_uobj(uobj, 0, file->ucontext, &xrcd_lock_key);
+ down_write(&uobj->mutex);
+
+ xrcd = file->device->ib_dev->alloc_xrcd(file->device->ib_dev,
+ file->ucontext, &udata);
+ if (IS_ERR(xrcd)) {
+ ret = PTR_ERR(xrcd);
+ goto err;
+ }
+
+ xrcd->uobject = uobj;
+ xrcd->device = file->device->ib_dev;
+ atomic_set(&xrcd->usecnt, 0);
+
+ uobj->object = xrcd;
+ ret = idr_add_uobj(&ib_uverbs_xrcd_idr, uobj);
+ if (ret)
+ goto err_idr;
+
+ memset(&resp, 0, sizeof resp);
+ resp.xrcd_handle = uobj->id;
+
+ if (copy_to_user((void __user *) (unsigned long) cmd.response,
+ &resp, sizeof resp)) {
+ ret = -EFAULT;
+ goto err_copy;
+ }
+
+ mutex_lock(&file->mutex);
+ list_add_tail(&uobj->list, &file->ucontext->xrcd_list);
+ mutex_unlock(&file->mutex);
+
+ uobj->live = 1;
+
+ up_write(&uobj->mutex);
+
+ return in_len;
+
+err_copy:
+ idr_remove_uobj(&ib_uverbs_xrcd_idr, uobj);
+
+err_idr:
+ ib_dealloc_xrcd(xrcd);
+
+err:
+ put_uobj_write(uobj);
+ return ret;
+}
+
+ssize_t ib_uverbs_close_xrcd(struct ib_uverbs_file *file,
+ const char __user *buf, int in_len,
+ int out_len)
+{
+ struct ib_uverbs_close_xrcd cmd;
+ struct ib_uobject *uobj;
+ int ret;
+
+ if (copy_from_user(&cmd, buf, sizeof cmd))
+ return -EFAULT;
+
+ uobj = idr_write_uobj(&ib_uverbs_xrcd_idr, cmd.xrcd_handle,
+ file->ucontext);
+ if (!uobj)
+ return -EINVAL;
+
+ ret = ib_dealloc_xrcd(uobj->object);
+ if (!ret)
+ uobj->live = 0;
+
+ put_uobj_write(uobj);
+
+ if (ret)
+ return ret;
+
+ idr_remove_uobj(&ib_uverbs_xrcd_idr, uobj);
+
+ mutex_lock(&file->mutex);
+ list_del(&uobj->list);
+ mutex_unlock(&file->mutex);
+
+ put_uobj(uobj);
+
+ return in_len;
+}
diff --git a/drivers/infiniband/core/uverbs_main.c
b/drivers/infiniband/core/uverbs_main.c
index 5f284ff..09b5b58 100644
--- a/drivers/infiniband/core/uverbs_main.c
+++ b/drivers/infiniband/core/uverbs_main.c
@@ -73,6 +73,7 @@ DEFINE_IDR(ib_uverbs_ah_idr);
DEFINE_IDR(ib_uverbs_cq_idr);
DEFINE_IDR(ib_uverbs_qp_idr);
DEFINE_IDR(ib_uverbs_srq_idr);
+DEFINE_IDR(ib_uverbs_xrcd_idr);
static DEFINE_SPINLOCK(map_lock);
static struct ib_uverbs_device *dev_table[IB_UVERBS_MAX_DEVICES];
@@ -109,6 +110,9 @@ static ssize_t (*uverbs_cmd_table[])(struct ib_uverbs_file
*file,
[IB_USER_VERBS_CMD_MODIFY_SRQ] = ib_uverbs_modify_srq,
[IB_USER_VERBS_CMD_QUERY_SRQ] = ib_uverbs_query_srq,
[IB_USER_VERBS_CMD_DESTROY_SRQ] = ib_uverbs_destroy_srq,
+ [IB_USER_VERBS_CMD_CREATE_XRC_SRQ] = ib_uverbs_create_xrc_srq,
+ [IB_USER_VERBS_CMD_OPEN_XRCD] = ib_uverbs_open_xrcd,
+ [IB_USER_VERBS_CMD_CLOSE_XRCD] = ib_uverbs_close_xrcd,
};
static struct vfsmount *uverbs_event_mnt;
@@ -212,17 +216,6 @@ static int ib_uverbs_cleanup_ucontext(struct
ib_uverbs_file *file,
kfree(uqp);
}
- list_for_each_entry_safe(uobj, tmp, &context->cq_list, list) {
- struct ib_cq *cq = uobj->object;
- struct ib_uverbs_event_file *ev_file = cq->cq_context;
- struct ib_ucq_object *ucq =
- container_of(uobj, struct ib_ucq_object, uobject);
-
- idr_remove_uobj(&ib_uverbs_cq_idr, uobj);
- ib_destroy_cq(cq);
- ib_uverbs_release_ucq(file, ev_file, ucq);
- kfree(ucq);
- }
list_for_each_entry_safe(uobj, tmp, &context->srq_list, list) {
struct ib_srq *srq = uobj->object;
@@ -235,6 +228,18 @@ static int ib_uverbs_cleanup_ucontext(struct
ib_uverbs_file *file,
kfree(uevent);
}
+ list_for_each_entry_safe(uobj, tmp, &context->cq_list, list) {
+ struct ib_cq *cq = uobj->object;
+ struct ib_uverbs_event_file *ev_file = cq->cq_context;
+ struct ib_ucq_object *ucq =
+ container_of(uobj, struct ib_ucq_object, uobject);
+
+ idr_remove_uobj(&ib_uverbs_cq_idr, uobj);
+ ib_destroy_cq(cq);
+ ib_uverbs_release_ucq(file, ev_file, ucq);
+ kfree(ucq);
+ }
+
/* XXX Free MWs */
list_for_each_entry_safe(uobj, tmp, &context->mr_list, list) {
@@ -245,6 +250,14 @@ static int ib_uverbs_cleanup_ucontext(struct
ib_uverbs_file *file,
kfree(uobj);
}
+ list_for_each_entry_safe(uobj, tmp, &context->xrcd_list, list) {
+ struct ib_xrcd *xrcd = uobj->object;
+
+ idr_remove_uobj(&ib_uverbs_xrcd_idr, uobj);
+ ib_dealloc_xrcd(xrcd);
+ kfree(uobj);
+ }
+
list_for_each_entry_safe(uobj, tmp, &context->pd_list, list) {
struct ib_pd *pd = uobj->object;
diff --git a/include/rdma/ib_user_verbs.h b/include/rdma/ib_user_verbs.h
index a17f771..c9e540c 100644
--- a/include/rdma/ib_user_verbs.h
+++ b/include/rdma/ib_user_verbs.h
@@ -81,7 +81,10 @@ enum {
IB_USER_VERBS_CMD_MODIFY_SRQ,
IB_USER_VERBS_CMD_QUERY_SRQ,
IB_USER_VERBS_CMD_DESTROY_SRQ,
- IB_USER_VERBS_CMD_POST_SRQ_RECV
+ IB_USER_VERBS_CMD_POST_SRQ_RECV,
+ IB_USER_VERBS_CMD_CREATE_XRC_SRQ,
+ IB_USER_VERBS_CMD_OPEN_XRCD,
+ IB_USER_VERBS_CMD_CLOSE_XRCD
};
/*
@@ -647,6 +650,18 @@ struct ib_uverbs_create_srq {
__u64 driver_data[0];
};
+struct ib_uverbs_create_xrc_srq {
+ __u64 response;
+ __u64 user_handle;
+ __u32 pd_handle;
+ __u32 max_wr;
+ __u32 max_sge;
+ __u32 srq_limit;
+ __u32 xrcd_handle;
+ __u32 xrc_cq;
+ __u64 driver_data[0];
+};
+
struct ib_uverbs_create_srq_resp {
__u32 srq_handle;
__u32 max_wr;
@@ -686,4 +701,22 @@ struct ib_uverbs_destroy_srq_resp {
__u32 events_reported;
};
+struct ib_uverbs_open_xrcd {
+ __u64 response;
+ __s32 fd;
+ __u32 oflags;
+ __u64 driver_data[0];
+};
+
+struct ib_uverbs_open_xrcd_resp {
+ __u32 xrcd_handle;
+};
+
+struct ib_uverbs_close_xrcd {
+ __u64 response;
+ __u32 xrcd_handle;
+ __u32 reserved;
+ __u64 driver_data[0];
+};
+
#endif /* IB_USER_VERBS_H */
--
1.6.6.1
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
