mirrors/qemu
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
c04cfb4596
qemu/hw/rdma/rdma_rm.h

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

98 lines
4.7 KiB
C
Raw Normal View History

hw/rdma: Implementation of generic rdma device layers This layer is composed of two sub-modules, backend and resource manager. Backend sub-module is responsible for all the interaction with IB layers such as ibverbs and umad (external libraries). Resource manager is a collection of functions and structures to manage RDMA resources such as QPs, CQs and MRs. Reviewed-by: Dotan Barak <dotanb@mellanox.com> Reviewed-by: Zhu Yanjun <yanjun.zhu@oracle.com> Signed-off-by: Yuval Shaia <yuval.shaia@oracle.com> Signed-off-by: Marcel Apfelbaum <marcel@redhat.com>
2018-02-09 16:23:18 +03:00
/*
* RDMA device: Definitions of Resource Manager functions
*
* Copyright (C) 2018 Oracle
* Copyright (C) 2018 Red Hat Inc
*
* Authors:
* Yuval Shaia <yuval.shaia@oracle.com>
* Marcel Apfelbaum <marcel@redhat.com>
*
* This work is licensed under the terms of the GNU GPL, version 2 or later.
* See the COPYING file in the top-level directory.
*
*/
#ifndef RDMA_RM_H
#define RDMA_RM_H
rdma: fix up include directives Our rule right now is to use <> for external headers only. RDMA code violates that, fix it up. Signed-off-by: Michael S. Tsirkin <mst@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Reviewed-by: Marcel Apfelbaum <marcel@redhat.com> Signed-off-by: Marcel Apfelbaum <marcel@redhat.com>
2018-03-21 18:22:07 +03:00
#include "qapi/error.h"
hw/rdma: Implementation of generic rdma device layers This layer is composed of two sub-modules, backend and resource manager. Backend sub-module is responsible for all the interaction with IB layers such as ibverbs and umad (external libraries). Resource manager is a collection of functions and structures to manage RDMA resources such as QPs, CQs and MRs. Reviewed-by: Dotan Barak <dotanb@mellanox.com> Reviewed-by: Zhu Yanjun <yanjun.zhu@oracle.com> Signed-off-by: Yuval Shaia <yuval.shaia@oracle.com> Signed-off-by: Marcel Apfelbaum <marcel@redhat.com>
2018-02-09 16:23:18 +03:00
#include "rdma_backend_defs.h"
#include "rdma_rm_defs.h"
hw/rdma: Switch to generic error reporting way Utilize error_report for all pr_err calls and some pr_dbg that are considered as errors. For the remaining pr_dbg calls, the important ones were replaced by trace points while other deleted. Some of the functions got renamed to include prefix "rdma/pvrdma" in the function name. Signed-off-by: Yuval Shaia <yuval.shaia@oracle.com> Reviewed-by: Marcel Apfelbaum <marcel.apfelbaum@gmail.com> Message-Id: <1552300155-25216-2-git-send-email-yuval.shaia@oracle.com> Reviewed-by: Kamal Heib <kamalheib1@gmail.com> Signed-off-by: Marcel Apfelbaum <marcel.apfelbaum@gmail.com>
2019-03-11 13:29:05 +03:00
int rdma_rm_init(RdmaDeviceResources *dev_res,
struct ibv_device_attr *dev_attr);
hw/pvrdma: Add support to allow guest to configure GID table The control over the RDMA device's GID table is done by updating the device's Ethernet function addresses. Usually the first GID entry is determined by the MAC address, the second by the first IPv6 address and the third by the IPv4 address. Other entries can be added by adding more IP addresses. The opposite is the same, i.e. whenever an address is removed, the corresponding GID entry is removed. The process is done by the network and RDMA stacks. Whenever an address is added the ib_core driver is notified and calls the device driver add_gid function which in turn update the device. To support this in pvrdma device we need to hook into the create_bind and destroy_bind HW commands triggered by pvrdma driver in guest. Whenever a change is made to the pvrdma port's GID table a special QMP message is sent to be processed by libvirt to update the address of the backend Ethernet device. Signed-off-by: Yuval Shaia <yuval.shaia@oracle.com> Reviewed-by: Marcel Apfelbaum<marcel.apfelbaum@gmail.com> Signed-off-by: Marcel Apfelbaum <marcel.apfelbaum@gmail.com>
2018-12-21 17:40:25 +03:00
void rdma_rm_fini(RdmaDeviceResources *dev_res, RdmaBackendDev *backend_dev,
const char *ifname);
hw/rdma: Implementation of generic rdma device layers This layer is composed of two sub-modules, backend and resource manager. Backend sub-module is responsible for all the interaction with IB layers such as ibverbs and umad (external libraries). Resource manager is a collection of functions and structures to manage RDMA resources such as QPs, CQs and MRs. Reviewed-by: Dotan Barak <dotanb@mellanox.com> Reviewed-by: Zhu Yanjun <yanjun.zhu@oracle.com> Signed-off-by: Yuval Shaia <yuval.shaia@oracle.com> Signed-off-by: Marcel Apfelbaum <marcel@redhat.com>
2018-02-09 16:23:18 +03:00
int rdma_rm_alloc_pd(RdmaDeviceResources *dev_res, RdmaBackendDev *backend_dev,
uint32_t *pd_handle, uint32_t ctx_handle);
RdmaRmPD *rdma_rm_get_pd(RdmaDeviceResources *dev_res, uint32_t pd_handle);
void rdma_rm_dealloc_pd(RdmaDeviceResources *dev_res, uint32_t pd_handle);
int rdma_rm_alloc_mr(RdmaDeviceResources *dev_res, uint32_t pd_handle,
hw/rdma: Switch to generic error reporting way Utilize error_report for all pr_err calls and some pr_dbg that are considered as errors. For the remaining pr_dbg calls, the important ones were replaced by trace points while other deleted. Some of the functions got renamed to include prefix "rdma/pvrdma" in the function name. Signed-off-by: Yuval Shaia <yuval.shaia@oracle.com> Reviewed-by: Marcel Apfelbaum <marcel.apfelbaum@gmail.com> Message-Id: <1552300155-25216-2-git-send-email-yuval.shaia@oracle.com> Reviewed-by: Kamal Heib <kamalheib1@gmail.com> Signed-off-by: Marcel Apfelbaum <marcel.apfelbaum@gmail.com>
2019-03-11 13:29:05 +03:00
uint64_t guest_start, uint64_t guest_length,
void *host_virt, int access_flags, uint32_t *mr_handle,
uint32_t *lkey, uint32_t *rkey);
hw/rdma: Implementation of generic rdma device layers This layer is composed of two sub-modules, backend and resource manager. Backend sub-module is responsible for all the interaction with IB layers such as ibverbs and umad (external libraries). Resource manager is a collection of functions and structures to manage RDMA resources such as QPs, CQs and MRs. Reviewed-by: Dotan Barak <dotanb@mellanox.com> Reviewed-by: Zhu Yanjun <yanjun.zhu@oracle.com> Signed-off-by: Yuval Shaia <yuval.shaia@oracle.com> Signed-off-by: Marcel Apfelbaum <marcel@redhat.com>
2018-02-09 16:23:18 +03:00
RdmaRmMR *rdma_rm_get_mr(RdmaDeviceResources *dev_res, uint32_t mr_handle);
void rdma_rm_dealloc_mr(RdmaDeviceResources *dev_res, uint32_t mr_handle);
int rdma_rm_alloc_uc(RdmaDeviceResources *dev_res, uint32_t pfn,
uint32_t *uc_handle);
RdmaRmUC *rdma_rm_get_uc(RdmaDeviceResources *dev_res, uint32_t uc_handle);
void rdma_rm_dealloc_uc(RdmaDeviceResources *dev_res, uint32_t uc_handle);
int rdma_rm_alloc_cq(RdmaDeviceResources *dev_res, RdmaBackendDev *backend_dev,
uint32_t cqe, uint32_t *cq_handle, void *opaque);
RdmaRmCQ *rdma_rm_get_cq(RdmaDeviceResources *dev_res, uint32_t cq_handle);
void rdma_rm_req_notify_cq(RdmaDeviceResources *dev_res, uint32_t cq_handle,
bool notify);
void rdma_rm_dealloc_cq(RdmaDeviceResources *dev_res, uint32_t cq_handle);
int rdma_rm_alloc_qp(RdmaDeviceResources *dev_res, uint32_t pd_handle,
uint8_t qp_type, uint32_t max_send_wr,
uint32_t max_send_sge, uint32_t send_cq_handle,
uint32_t max_recv_wr, uint32_t max_recv_sge,
hw/rdma: Modify create/destroy QP to support SRQ Modify create/destroy QP to support shared receive queue and rearrange the destroy_qp() code to avoid touching the QP after calling rdma_rm_dealloc_qp(). Signed-off-by: Kamal Heib <kamalheib1@gmail.com> Message-Id: <20190403113343.26384-4-kamalheib1@gmail.com> Reviewed-by: Yuval Shaia <yuval.shaia@oracle.com> Signed-off-by: Marcel Apfelbaum <marcel.apfelbaum@gmail.com>
2019-04-03 14:33:42 +03:00
uint32_t recv_cq_handle, void *opaque, uint32_t *qpn,
uint8_t is_srq, uint32_t srq_handle);
hw/rdma: Implementation of generic rdma device layers This layer is composed of two sub-modules, backend and resource manager. Backend sub-module is responsible for all the interaction with IB layers such as ibverbs and umad (external libraries). Resource manager is a collection of functions and structures to manage RDMA resources such as QPs, CQs and MRs. Reviewed-by: Dotan Barak <dotanb@mellanox.com> Reviewed-by: Zhu Yanjun <yanjun.zhu@oracle.com> Signed-off-by: Yuval Shaia <yuval.shaia@oracle.com> Signed-off-by: Marcel Apfelbaum <marcel@redhat.com>
2018-02-09 16:23:18 +03:00
RdmaRmQP *rdma_rm_get_qp(RdmaDeviceResources *dev_res, uint32_t qpn);
int rdma_rm_modify_qp(RdmaDeviceResources *dev_res, RdmaBackendDev *backend_dev,
hw/pvrdma: Add support to allow guest to configure GID table The control over the RDMA device's GID table is done by updating the device's Ethernet function addresses. Usually the first GID entry is determined by the MAC address, the second by the first IPv6 address and the third by the IPv4 address. Other entries can be added by adding more IP addresses. The opposite is the same, i.e. whenever an address is removed, the corresponding GID entry is removed. The process is done by the network and RDMA stacks. Whenever an address is added the ib_core driver is notified and calls the device driver add_gid function which in turn update the device. To support this in pvrdma device we need to hook into the create_bind and destroy_bind HW commands triggered by pvrdma driver in guest. Whenever a change is made to the pvrdma port's GID table a special QMP message is sent to be processed by libvirt to update the address of the backend Ethernet device. Signed-off-by: Yuval Shaia <yuval.shaia@oracle.com> Reviewed-by: Marcel Apfelbaum<marcel.apfelbaum@gmail.com> Signed-off-by: Marcel Apfelbaum <marcel.apfelbaum@gmail.com>
2018-12-21 17:40:25 +03:00
uint32_t qp_handle, uint32_t attr_mask, uint8_t sgid_idx,
hw/rdma: Implementation of generic rdma device layers This layer is composed of two sub-modules, backend and resource manager. Backend sub-module is responsible for all the interaction with IB layers such as ibverbs and umad (external libraries). Resource manager is a collection of functions and structures to manage RDMA resources such as QPs, CQs and MRs. Reviewed-by: Dotan Barak <dotanb@mellanox.com> Reviewed-by: Zhu Yanjun <yanjun.zhu@oracle.com> Signed-off-by: Yuval Shaia <yuval.shaia@oracle.com> Signed-off-by: Marcel Apfelbaum <marcel@redhat.com>
2018-02-09 16:23:18 +03:00
union ibv_gid *dgid, uint32_t dqpn,
enum ibv_qp_state qp_state, uint32_t qkey,
uint32_t rq_psn, uint32_t sq_psn);
hw/rdma: Add Query QP operation This operation is needed by rdma devices - implement it. Signed-off-by: Yuval Shaia <yuval.shaia@oracle.com> Reviewed-by: Marcel Apfelbaum <marcel@redhat.com> Signed-off-by: Marcel Apfelbaum <marcel@redhat.com>
2018-03-20 22:00:22 +03:00
int rdma_rm_query_qp(RdmaDeviceResources *dev_res, RdmaBackendDev *backend_dev,
uint32_t qp_handle, struct ibv_qp_attr *attr,
int attr_mask, struct ibv_qp_init_attr *init_attr);
hw/rdma: Implementation of generic rdma device layers This layer is composed of two sub-modules, backend and resource manager. Backend sub-module is responsible for all the interaction with IB layers such as ibverbs and umad (external libraries). Resource manager is a collection of functions and structures to manage RDMA resources such as QPs, CQs and MRs. Reviewed-by: Dotan Barak <dotanb@mellanox.com> Reviewed-by: Zhu Yanjun <yanjun.zhu@oracle.com> Signed-off-by: Yuval Shaia <yuval.shaia@oracle.com> Signed-off-by: Marcel Apfelbaum <marcel@redhat.com>
2018-02-09 16:23:18 +03:00
void rdma_rm_dealloc_qp(RdmaDeviceResources *dev_res, uint32_t qp_handle);
hw/rdma: Add support for managing SRQ resource Adding the required functions and definitions for support managing the shared receive queues (SRQs). Signed-off-by: Kamal Heib <kamalheib1@gmail.com> Message-Id: <20190403113343.26384-3-kamalheib1@gmail.com> Reviewed-by: Yuval Shaia <yuval.shaia@oracle.com> Signed-off-by: Marcel Apfelbaum <marcel.apfelbaum@gmail.com>
2019-04-03 14:33:41 +03:00
RdmaRmSRQ *rdma_rm_get_srq(RdmaDeviceResources *dev_res, uint32_t srq_handle);
int rdma_rm_alloc_srq(RdmaDeviceResources *dev_res, uint32_t pd_handle,
uint32_t max_wr, uint32_t max_sge, uint32_t srq_limit,
uint32_t *srq_handle, void *opaque);
int rdma_rm_query_srq(RdmaDeviceResources *dev_res, uint32_t srq_handle,
struct ibv_srq_attr *srq_attr);
int rdma_rm_modify_srq(RdmaDeviceResources *dev_res, uint32_t srq_handle,
struct ibv_srq_attr *srq_attr, int srq_attr_mask);
void rdma_rm_dealloc_srq(RdmaDeviceResources *dev_res, uint32_t srq_handle);
hw/rdma: Implementation of generic rdma device layers This layer is composed of two sub-modules, backend and resource manager. Backend sub-module is responsible for all the interaction with IB layers such as ibverbs and umad (external libraries). Resource manager is a collection of functions and structures to manage RDMA resources such as QPs, CQs and MRs. Reviewed-by: Dotan Barak <dotanb@mellanox.com> Reviewed-by: Zhu Yanjun <yanjun.zhu@oracle.com> Signed-off-by: Yuval Shaia <yuval.shaia@oracle.com> Signed-off-by: Marcel Apfelbaum <marcel@redhat.com>
2018-02-09 16:23:18 +03:00
int rdma_rm_alloc_cqe_ctx(RdmaDeviceResources *dev_res, uint32_t *cqe_ctx_id,
void *ctx);
void *rdma_rm_get_cqe_ctx(RdmaDeviceResources *dev_res, uint32_t cqe_ctx_id);
void rdma_rm_dealloc_cqe_ctx(RdmaDeviceResources *dev_res, uint32_t cqe_ctx_id);
hw/pvrdma: Add support to allow guest to configure GID table The control over the RDMA device's GID table is done by updating the device's Ethernet function addresses. Usually the first GID entry is determined by the MAC address, the second by the first IPv6 address and the third by the IPv4 address. Other entries can be added by adding more IP addresses. The opposite is the same, i.e. whenever an address is removed, the corresponding GID entry is removed. The process is done by the network and RDMA stacks. Whenever an address is added the ib_core driver is notified and calls the device driver add_gid function which in turn update the device. To support this in pvrdma device we need to hook into the create_bind and destroy_bind HW commands triggered by pvrdma driver in guest. Whenever a change is made to the pvrdma port's GID table a special QMP message is sent to be processed by libvirt to update the address of the backend Ethernet device. Signed-off-by: Yuval Shaia <yuval.shaia@oracle.com> Reviewed-by: Marcel Apfelbaum<marcel.apfelbaum@gmail.com> Signed-off-by: Marcel Apfelbaum <marcel.apfelbaum@gmail.com>
2018-12-21 17:40:25 +03:00
int rdma_rm_add_gid(RdmaDeviceResources *dev_res, RdmaBackendDev *backend_dev,
const char *ifname, union ibv_gid *gid, int gid_idx);
int rdma_rm_del_gid(RdmaDeviceResources *dev_res, RdmaBackendDev *backend_dev,
const char *ifname, int gid_idx);
int rdma_rm_get_backend_gid_index(RdmaDeviceResources *dev_res,
RdmaBackendDev *backend_dev, int sgid_idx);
static inline union ibv_gid *rdma_rm_get_gid(RdmaDeviceResources *dev_res,
int sgid_idx)
{
hw/rdma: Remove unneeded code that handles more that one port Device supports only one port, let's remove a dead code that handles more than one port. Signed-off-by: Yuval Shaia <yuval.shaia@oracle.com> Reviewed-by: Marcel Apfelbaum <marcel.apfelbaum@gmail.com> Signed-off-by: Marcel Apfelbaum <marcel.apfelbaum@gmail.com>
2018-12-21 17:40:32 +03:00
return &dev_res->port.gid_tbl[sgid_idx].gid;
hw/pvrdma: Add support to allow guest to configure GID table The control over the RDMA device's GID table is done by updating the device's Ethernet function addresses. Usually the first GID entry is determined by the MAC address, the second by the first IPv6 address and the third by the IPv4 address. Other entries can be added by adding more IP addresses. The opposite is the same, i.e. whenever an address is removed, the corresponding GID entry is removed. The process is done by the network and RDMA stacks. Whenever an address is added the ib_core driver is notified and calls the device driver add_gid function which in turn update the device. To support this in pvrdma device we need to hook into the create_bind and destroy_bind HW commands triggered by pvrdma driver in guest. Whenever a change is made to the pvrdma port's GID table a special QMP message is sent to be processed by libvirt to update the address of the backend Ethernet device. Signed-off-by: Yuval Shaia <yuval.shaia@oracle.com> Reviewed-by: Marcel Apfelbaum<marcel.apfelbaum@gmail.com> Signed-off-by: Marcel Apfelbaum <marcel.apfelbaum@gmail.com>
2018-12-21 17:40:25 +03:00
}
qapi: introduce x-query-rdma QMP command This is a counterpart to the HMP "info rdma" command. It is being added with an "x-" prefix because this QMP command is intended as an adhoc debugging tool and will thus not be modelled in QAPI as fully structured data, nor will it have long term guaranteed stability. The existing HMP command is rewritten to call the QMP command. Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com> Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
2021-09-08 12:35:43 +03:00
void rdma_format_device_counters(RdmaDeviceResources *dev_res, GString *buf);
hw/pvrdma: Add support to allow guest to configure GID table The control over the RDMA device's GID table is done by updating the device's Ethernet function addresses. Usually the first GID entry is determined by the MAC address, the second by the first IPv6 address and the third by the IPv4 address. Other entries can be added by adding more IP addresses. The opposite is the same, i.e. whenever an address is removed, the corresponding GID entry is removed. The process is done by the network and RDMA stacks. Whenever an address is added the ib_core driver is notified and calls the device driver add_gid function which in turn update the device. To support this in pvrdma device we need to hook into the create_bind and destroy_bind HW commands triggered by pvrdma driver in guest. Whenever a change is made to the pvrdma port's GID table a special QMP message is sent to be processed by libvirt to update the address of the backend Ethernet device. Signed-off-by: Yuval Shaia <yuval.shaia@oracle.com> Reviewed-by: Marcel Apfelbaum<marcel.apfelbaum@gmail.com> Signed-off-by: Marcel Apfelbaum <marcel.apfelbaum@gmail.com>
2018-12-21 17:40:25 +03:00
hw/rdma: Implementation of generic rdma device layers This layer is composed of two sub-modules, backend and resource manager. Backend sub-module is responsible for all the interaction with IB layers such as ibverbs and umad (external libraries). Resource manager is a collection of functions and structures to manage RDMA resources such as QPs, CQs and MRs. Reviewed-by: Dotan Barak <dotanb@mellanox.com> Reviewed-by: Zhu Yanjun <yanjun.zhu@oracle.com> Signed-off-by: Yuval Shaia <yuval.shaia@oracle.com> Signed-off-by: Marcel Apfelbaum <marcel@redhat.com>
2018-02-09 16:23:18 +03:00
#endif
Reference in New Issue Copy Permalink