From 8d36eb01da5d371feffa280e501377b5c450f5a5 Mon Sep 17 00:00:00 2001 From: Sean Hefty Date: Wed, 29 May 2013 10:09:07 -0700 Subject: RDMA/cma: Define native IB address Define AF_IB and sockaddr_ib to allow the rdma_cm to use native IB addressing. Signed-off-by: Sean Hefty Signed-off-by: Roland Dreier --- include/linux/socket.h | 2 ++ include/rdma/ib.h | 89 ++++++++++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 91 insertions(+) create mode 100644 include/rdma/ib.h (limited to 'include') diff --git a/include/linux/socket.h b/include/linux/socket.h index b10ce4b341ea..230c04bda3e2 100644 --- a/include/linux/socket.h +++ b/include/linux/socket.h @@ -167,6 +167,7 @@ struct ucred { #define AF_PPPOX 24 /* PPPoX sockets */ #define AF_WANPIPE 25 /* Wanpipe API Sockets */ #define AF_LLC 26 /* Linux LLC */ +#define AF_IB 27 /* Native InfiniBand address */ #define AF_CAN 29 /* Controller Area Network */ #define AF_TIPC 30 /* TIPC sockets */ #define AF_BLUETOOTH 31 /* Bluetooth sockets */ @@ -211,6 +212,7 @@ struct ucred { #define PF_PPPOX AF_PPPOX #define PF_WANPIPE AF_WANPIPE #define PF_LLC AF_LLC +#define PF_IB AF_IB #define PF_CAN AF_CAN #define PF_TIPC AF_TIPC #define PF_BLUETOOTH AF_BLUETOOTH diff --git a/include/rdma/ib.h b/include/rdma/ib.h new file mode 100644 index 000000000000..cf8f9e700e48 --- /dev/null +++ b/include/rdma/ib.h @@ -0,0 +1,89 @@ +/* + * Copyright (c) 2010 Intel Corporation. All rights reserved. + * + * This software is available to you under a choice of one of two + * licenses. You may choose to be licensed under the terms of the GNU + * General Public License (GPL) Version 2, available from the file + * COPYING in the main directory of this source tree, or the + * OpenIB.org BSD license below: + * + * Redistribution and use in source and binary forms, with or + * without modification, are permitted provided that the following + * conditions are met: + * + * - Redistributions of source code must retain the above + * copyright notice, this list of conditions and the following + * disclaimer. + * + * - Redistributions in binary form must reproduce the above + * copyright notice, this list of conditions and the following + * disclaimer in the documentation and/or other materials + * provided with the distribution. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND + * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS + * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN + * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN + * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#if !defined(_RDMA_IB_H) +#define _RDMA_IB_H + +#include + +struct ib_addr { + union { + __u8 uib_addr8[16]; + __be16 uib_addr16[8]; + __be32 uib_addr32[4]; + __be64 uib_addr64[2]; + } ib_u; +#define sib_addr8 ib_u.uib_addr8 +#define sib_addr16 ib_u.uib_addr16 +#define sib_addr32 ib_u.uib_addr32 +#define sib_addr64 ib_u.uib_addr64 +#define sib_raw ib_u.uib_addr8 +#define sib_subnet_prefix ib_u.uib_addr64[0] +#define sib_interface_id ib_u.uib_addr64[1] +}; + +static inline int ib_addr_any(const struct ib_addr *a) +{ + return ((a->sib_addr64[0] | a->sib_addr64[1]) == 0); +} + +static inline int ib_addr_loopback(const struct ib_addr *a) +{ + return ((a->sib_addr32[0] | a->sib_addr32[1] | + a->sib_addr32[2] | (a->sib_addr32[3] ^ htonl(1))) == 0); +} + +static inline void ib_addr_set(struct ib_addr *addr, + __be32 w1, __be32 w2, __be32 w3, __be32 w4) +{ + addr->sib_addr32[0] = w1; + addr->sib_addr32[1] = w2; + addr->sib_addr32[2] = w3; + addr->sib_addr32[3] = w4; +} + +static inline int ib_addr_cmp(const struct ib_addr *a1, const struct ib_addr *a2) +{ + return memcmp(a1, a2, sizeof(struct ib_addr)); +} + +struct sockaddr_ib { + unsigned short int sib_family; /* AF_IB */ + __be16 sib_pkey; + __be32 sib_flowinfo; + struct ib_addr sib_addr; + __be64 sib_sid; + __be64 sib_sid_mask; + __u64 sib_scope_id; +}; + +#endif /* _RDMA_IB_H */ -- cgit v1.2.3 From ef560861c01c301cde3da154eb9c1c2619924c3a Mon Sep 17 00:00:00 2001 From: Sean Hefty Date: Wed, 29 May 2013 10:09:10 -0700 Subject: IB/addr: Add AF_IB support to ip_addr_size Add support for AF_IB to ip_addr_size, and rename the function to account for the change. Give the compiler more control over whether the call should be inline or not by moving the definition into the .c file, removing the static inline, and exporting it. Signed-off-by: Sean Hefty Signed-off-by: Roland Dreier --- include/rdma/ib_addr.h | 6 +----- 1 file changed, 1 insertion(+), 5 deletions(-) (limited to 'include') diff --git a/include/rdma/ib_addr.h b/include/rdma/ib_addr.h index 99965395c5f3..f3ac0f2c4c66 100644 --- a/include/rdma/ib_addr.h +++ b/include/rdma/ib_addr.h @@ -102,11 +102,7 @@ void rdma_addr_cancel(struct rdma_dev_addr *addr); int rdma_copy_addr(struct rdma_dev_addr *dev_addr, struct net_device *dev, const unsigned char *dst_dev_addr); -static inline int ip_addr_size(struct sockaddr *addr) -{ - return addr->sa_family == AF_INET6 ? - sizeof(struct sockaddr_in6) : sizeof(struct sockaddr_in); -} +int rdma_addr_size(struct sockaddr *addr); static inline u16 ib_addr_get_pkey(struct rdma_dev_addr *dev_addr) { -- cgit v1.2.3 From 58afdcb7382234ebd780e43b17edde92a5853cca Mon Sep 17 00:00:00 2001 From: Sean Hefty Date: Wed, 29 May 2013 10:09:11 -0700 Subject: RDMA/cma: Update port reservation to support AF_IB The AF_IB uses a 64-bit service id (SID), which the user can control through the use of a mask. The rdma_cm will assign values to the unmasked portions of the SID based on the selected port space and port number. Because the IB spec divides the SID range into several regions, a SID/mask combination may fall into one of the existing port space ranges as defined by the RDMA CM IP Annex. Map the AF_IB SID to the correct RDMA port space. Signed-off-by: Sean Hefty Signed-off-by: Roland Dreier --- include/rdma/rdma_cm.h | 5 +++++ 1 file changed, 5 insertions(+) (limited to 'include') diff --git a/include/rdma/rdma_cm.h b/include/rdma/rdma_cm.h index ad3a3142383a..1e6c3c7af898 100644 --- a/include/rdma/rdma_cm.h +++ b/include/rdma/rdma_cm.h @@ -70,6 +70,11 @@ enum rdma_port_space { RDMA_PS_UDP = 0x0111, }; +#define RDMA_IB_IP_PS_MASK 0xFFFFFFFFFFFF0000ULL +#define RDMA_IB_IP_PS_TCP 0x0000000001060000ULL +#define RDMA_IB_IP_PS_UDP 0x0000000001110000ULL +#define RDMA_IB_IP_PS_IB 0x00000000013F0000ULL + struct rdma_addr { struct sockaddr_storage src_addr; struct sockaddr_storage dst_addr; -- cgit v1.2.3 From 5c438135adf90b33cb00e5351becf1e557bbdd9d Mon Sep 17 00:00:00 2001 From: Sean Hefty Date: Wed, 29 May 2013 10:09:23 -0700 Subject: RDMA/cma: Set qkey for AF_IB Allow the user to specify the qkey when using AF_IB. The qkey is added to struct rdma_ucm_conn_param in place of a reserved field, but for backwards compatability, is only accessed if the associated rdma_cm_id is using AF_IB. Signed-off-by: Sean Hefty Signed-off-by: Roland Dreier --- include/rdma/rdma_cm.h | 1 + include/uapi/rdma/rdma_user_cm.h | 2 +- 2 files changed, 2 insertions(+), 1 deletion(-) (limited to 'include') diff --git a/include/rdma/rdma_cm.h b/include/rdma/rdma_cm.h index 1e6c3c7af898..966f90ba8d8a 100644 --- a/include/rdma/rdma_cm.h +++ b/include/rdma/rdma_cm.h @@ -98,6 +98,7 @@ struct rdma_conn_param { /* Fields below ignored if a QP is created on the rdma_cm_id. */ u8 srq; u32 qp_num; + u32 qkey; }; struct rdma_ud_param { diff --git a/include/uapi/rdma/rdma_user_cm.h b/include/uapi/rdma/rdma_user_cm.h index 1ee9239ff8c2..29de08f603ac 100644 --- a/include/uapi/rdma/rdma_user_cm.h +++ b/include/uapi/rdma/rdma_user_cm.h @@ -131,7 +131,7 @@ struct rdma_ucm_query_route_resp { struct rdma_ucm_conn_param { __u32 qp_num; - __u32 reserved; + __u32 qkey; __u8 private_data[RDMA_MAX_PRIVATE_DATA]; __u8 private_data_len; __u8 srq; -- cgit v1.2.3 From ee7aed4528fb3c44a36abd79eb23fd5401a5b697 Mon Sep 17 00:00:00 2001 From: Sean Hefty Date: Wed, 29 May 2013 10:09:25 -0700 Subject: RDMA/ucma: Support querying for AF_IB addresses The sockaddr structure for AF_IB is larger than sockaddr_in6. The rdma cm user space ABI uses the latter to exchange address information between user space and the kernel. To support querying for larger addresses, define a new query command that exchanges data using sockaddr_storage, rather than sockaddr_in6. Unlike the existing query_route command, the new command only returns address information. Route (i.e. path record) data is separated. Signed-off-by: Sean Hefty Signed-off-by: Roland Dreier --- include/uapi/rdma/rdma_user_cm.h | 22 +++++++++++++++++++--- 1 file changed, 19 insertions(+), 3 deletions(-) (limited to 'include') diff --git a/include/uapi/rdma/rdma_user_cm.h b/include/uapi/rdma/rdma_user_cm.h index 29de08f603ac..3ea7e7a4d54b 100644 --- a/include/uapi/rdma/rdma_user_cm.h +++ b/include/uapi/rdma/rdma_user_cm.h @@ -61,7 +61,8 @@ enum { RDMA_USER_CM_CMD_NOTIFY, RDMA_USER_CM_CMD_JOIN_MCAST, RDMA_USER_CM_CMD_LEAVE_MCAST, - RDMA_USER_CM_CMD_MIGRATE_ID + RDMA_USER_CM_CMD_MIGRATE_ID, + RDMA_USER_CM_CMD_QUERY }; /* @@ -113,10 +114,14 @@ struct rdma_ucm_resolve_route { __u32 timeout_ms; }; -struct rdma_ucm_query_route { +enum { + RDMA_USER_CM_QUERY_ADDR +}; + +struct rdma_ucm_query { __u64 response; __u32 id; - __u32 reserved; + __u32 option; }; struct rdma_ucm_query_route_resp { @@ -129,6 +134,17 @@ struct rdma_ucm_query_route_resp { __u8 reserved[3]; }; +struct rdma_ucm_query_addr_resp { + __u64 node_guid; + __u8 port_num; + __u8 reserved; + __u16 pkey; + __u16 src_size; + __u16 dst_size; + struct sockaddr_storage src_addr; + struct sockaddr_storage dst_addr; +}; + struct rdma_ucm_conn_param { __u32 qp_num; __u32 qkey; -- cgit v1.2.3 From 2e08b5879e9244fa893fe09f5b887f72f4e6c29b Mon Sep 17 00:00:00 2001 From: Sean Hefty Date: Wed, 29 May 2013 10:09:26 -0700 Subject: IB/sa: Export function to pack a path record into wire format Allow converting from struct ib_sa_path_rec to the IB defined SA path record wire format. This will be used to report path data from the rdma cm into user space. Signed-off-by: Sean Hefty Signed-off-by: Roland Dreier --- include/rdma/ib_sa.h | 7 +++++++ 1 file changed, 7 insertions(+) (limited to 'include') diff --git a/include/rdma/ib_sa.h b/include/rdma/ib_sa.h index 8275e539bace..125f8714301d 100644 --- a/include/rdma/ib_sa.h +++ b/include/rdma/ib_sa.h @@ -401,6 +401,12 @@ int ib_init_ah_from_path(struct ib_device *device, u8 port_num, struct ib_sa_path_rec *rec, struct ib_ah_attr *ah_attr); +/** + * ib_sa_pack_path - Conert a path record from struct ib_sa_path_rec + * to IB MAD wire format. + */ +void ib_sa_pack_path(struct ib_sa_path_rec *rec, void *attribute); + /** * ib_sa_unpack_path - Convert a path record from MAD format to struct * ib_sa_path_rec. @@ -418,4 +424,5 @@ int ib_sa_guid_info_rec_query(struct ib_sa_client *client, void *context), void *context, struct ib_sa_query **sa_query); + #endif /* IB_SA_H */ -- cgit v1.2.3 From ac53b264b2f39e89781e3b855008123dfdb44aea Mon Sep 17 00:00:00 2001 From: Sean Hefty Date: Wed, 29 May 2013 10:09:27 -0700 Subject: RDMA/ucma: Support querying when IB paths are not reversible The current query_route call can return up to two path records. The assumption being that one is the primary path, with optional support for an alternate path. In both cases, the paths are assumed to be reversible and are used to send CM MADs. With the ability to manually set IB path data, the rdma cm can eventually be capable of using up to 6 paths per connection: forward primary, reverse primary, forward alternate, reverse alternate, reversible primary path for CM MADs reversible alternate path for CM MADs. (It is unclear at this time if IB routing will complicate this) In order to handle more flexible routing topologies, add a new command to report any number of paths. Signed-off-by: Sean Hefty Signed-off-by: Roland Dreier --- include/uapi/rdma/rdma_user_cm.h | 9 ++++++++- 1 file changed, 8 insertions(+), 1 deletion(-) (limited to 'include') diff --git a/include/uapi/rdma/rdma_user_cm.h b/include/uapi/rdma/rdma_user_cm.h index 3ea7e7a4d54b..07eb6cfa926c 100644 --- a/include/uapi/rdma/rdma_user_cm.h +++ b/include/uapi/rdma/rdma_user_cm.h @@ -115,7 +115,8 @@ struct rdma_ucm_resolve_route { }; enum { - RDMA_USER_CM_QUERY_ADDR + RDMA_USER_CM_QUERY_ADDR, + RDMA_USER_CM_QUERY_PATH }; struct rdma_ucm_query { @@ -145,6 +146,12 @@ struct rdma_ucm_query_addr_resp { struct sockaddr_storage dst_addr; }; +struct rdma_ucm_query_path_resp { + __u32 num_paths; + __u32 reserved; + struct ib_path_rec_data path_data[0]; +}; + struct rdma_ucm_conn_param { __u32 qp_num; __u32 qkey; -- cgit v1.2.3 From cf53936f229d81131fef475919f163ce566a205f Mon Sep 17 00:00:00 2001 From: Sean Hefty Date: Wed, 29 May 2013 10:09:28 -0700 Subject: RDMA/cma: Export cma_get_service_id() Allow the rdma_ucm to query the IB service ID formed or allocated by the rdma_cm by exporting the cma_get_service_id() functionality. Signed-off-by: Sean Hefty Signed-off-by: Roland Dreier --- include/rdma/rdma_cm.h | 7 +++++++ 1 file changed, 7 insertions(+) (limited to 'include') diff --git a/include/rdma/rdma_cm.h b/include/rdma/rdma_cm.h index 966f90ba8d8a..1ed2088dc9f5 100644 --- a/include/rdma/rdma_cm.h +++ b/include/rdma/rdma_cm.h @@ -373,4 +373,11 @@ int rdma_set_reuseaddr(struct rdma_cm_id *id, int reuse); */ int rdma_set_afonly(struct rdma_cm_id *id, int afonly); + /** + * rdma_get_service_id - Return the IB service ID for a specified address. + * @id: Communication identifier associated with the address. + * @addr: Address for the service ID. + */ +__be64 rdma_get_service_id(struct rdma_cm_id *id, struct sockaddr *addr); + #endif /* RDMA_CM_H */ -- cgit v1.2.3 From edaa7a5578988bcf12f68f14fb002bc0c87e2801 Mon Sep 17 00:00:00 2001 From: Sean Hefty Date: Wed, 29 May 2013 10:09:29 -0700 Subject: RDMA/ucma: Add ability to query GID addresses Part of address resolution is mapping IP addresses to IB GIDs. With the changes to support querying larger addresses and more path records, also provide a way to query IB GIDs after resolution completes. Signed-off-by: Sean Hefty Signed-off-by: Roland Dreier --- include/uapi/rdma/rdma_user_cm.h | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'include') diff --git a/include/uapi/rdma/rdma_user_cm.h b/include/uapi/rdma/rdma_user_cm.h index 07eb6cfa926c..ea79253ceb13 100644 --- a/include/uapi/rdma/rdma_user_cm.h +++ b/include/uapi/rdma/rdma_user_cm.h @@ -116,7 +116,8 @@ struct rdma_ucm_resolve_route { enum { RDMA_USER_CM_QUERY_ADDR, - RDMA_USER_CM_QUERY_PATH + RDMA_USER_CM_QUERY_PATH, + RDMA_USER_CM_QUERY_GID }; struct rdma_ucm_query { -- cgit v1.2.3 From 05ad94577ecd5a101889d04aa099b738ec5ee34f Mon Sep 17 00:00:00 2001 From: Sean Hefty Date: Wed, 29 May 2013 10:09:30 -0700 Subject: RDMA/ucma: Name changes to indicate only IP addresses supported Several commands into the RDMA CM from user space are restricted to supporting addresses which fit into a sockaddr_in6 structure: bind address, resolve address, and join multicast. With the addition of AF_IB, we need to support addresses which are larger than sockaddr_in6. This will be done by adding new commands that exchange address information using sockaddr_storage. However, to support existing applications, we maintain the current commands and structures, but rename them to indicate that they only support IPv4 and v6 addresses. Signed-off-by: Sean Hefty Signed-off-by: Roland Dreier --- include/uapi/rdma/rdma_user_cm.h | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) (limited to 'include') diff --git a/include/uapi/rdma/rdma_user_cm.h b/include/uapi/rdma/rdma_user_cm.h index ea79253ceb13..79f68f7c9fd2 100644 --- a/include/uapi/rdma/rdma_user_cm.h +++ b/include/uapi/rdma/rdma_user_cm.h @@ -45,8 +45,8 @@ enum { RDMA_USER_CM_CMD_CREATE_ID, RDMA_USER_CM_CMD_DESTROY_ID, - RDMA_USER_CM_CMD_BIND_ADDR, - RDMA_USER_CM_CMD_RESOLVE_ADDR, + RDMA_USER_CM_CMD_BIND_IP, + RDMA_USER_CM_CMD_RESOLVE_IP, RDMA_USER_CM_CMD_RESOLVE_ROUTE, RDMA_USER_CM_CMD_QUERY_ROUTE, RDMA_USER_CM_CMD_CONNECT, @@ -59,7 +59,7 @@ enum { RDMA_USER_CM_CMD_GET_OPTION, RDMA_USER_CM_CMD_SET_OPTION, RDMA_USER_CM_CMD_NOTIFY, - RDMA_USER_CM_CMD_JOIN_MCAST, + RDMA_USER_CM_CMD_JOIN_IP_MCAST, RDMA_USER_CM_CMD_LEAVE_MCAST, RDMA_USER_CM_CMD_MIGRATE_ID, RDMA_USER_CM_CMD_QUERY @@ -96,13 +96,13 @@ struct rdma_ucm_destroy_id_resp { __u32 events_reported; }; -struct rdma_ucm_bind_addr { +struct rdma_ucm_bind_ip { __u64 response; struct sockaddr_in6 addr; __u32 id; }; -struct rdma_ucm_resolve_addr { +struct rdma_ucm_resolve_ip { struct sockaddr_in6 src_addr; struct sockaddr_in6 dst_addr; __u32 id; @@ -216,7 +216,7 @@ struct rdma_ucm_notify { __u32 event; }; -struct rdma_ucm_join_mcast { +struct rdma_ucm_join_ip_mcast { __u64 response; /* rdma_ucm_create_id_resp */ __u64 uid; struct sockaddr_in6 addr; -- cgit v1.2.3 From eebe4c3a62aadb64ba30bde97b96d656e369d934 Mon Sep 17 00:00:00 2001 From: Sean Hefty Date: Wed, 29 May 2013 10:09:31 -0700 Subject: RDMA/ucma: Allow user space to bind to AF_IB Support user space binding to addresses using AF_IB. Since sockaddr_ib is larger than sockaddr_in6, we need to define a larger structure when binding using AF_IB. This time we use sockaddr_storage to cover future cases. Signed-off-by: Sean Hefty Signed-off-by: Roland Dreier --- include/uapi/rdma/rdma_user_cm.h | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-) (limited to 'include') diff --git a/include/uapi/rdma/rdma_user_cm.h b/include/uapi/rdma/rdma_user_cm.h index 79f68f7c9fd2..895a427bfc91 100644 --- a/include/uapi/rdma/rdma_user_cm.h +++ b/include/uapi/rdma/rdma_user_cm.h @@ -62,7 +62,8 @@ enum { RDMA_USER_CM_CMD_JOIN_IP_MCAST, RDMA_USER_CM_CMD_LEAVE_MCAST, RDMA_USER_CM_CMD_MIGRATE_ID, - RDMA_USER_CM_CMD_QUERY + RDMA_USER_CM_CMD_QUERY, + RDMA_USER_CM_CMD_BIND }; /* @@ -102,6 +103,13 @@ struct rdma_ucm_bind_ip { __u32 id; }; +struct rdma_ucm_bind { + __u32 id; + __u16 addr_size; + __u16 reserved; + struct sockaddr_storage addr; +}; + struct rdma_ucm_resolve_ip { struct sockaddr_in6 src_addr; struct sockaddr_in6 dst_addr; -- cgit v1.2.3 From 209cf2a751f9ff2a516102339e54fcac0176fa78 Mon Sep 17 00:00:00 2001 From: Sean Hefty Date: Wed, 29 May 2013 10:09:32 -0700 Subject: RDMA/ucma: Allow user space to pass AF_IB into resolve Allow user space applications to call resolve_addr using AF_IB. To support sockaddr_ib, we need to define a new structure capable of handling the larger address size. Signed-off-by: Sean Hefty Signed-off-by: Roland Dreier --- include/uapi/rdma/rdma_user_cm.h | 13 ++++++++++++- 1 file changed, 12 insertions(+), 1 deletion(-) (limited to 'include') diff --git a/include/uapi/rdma/rdma_user_cm.h b/include/uapi/rdma/rdma_user_cm.h index 895a427bfc91..6d03f9c16f1e 100644 --- a/include/uapi/rdma/rdma_user_cm.h +++ b/include/uapi/rdma/rdma_user_cm.h @@ -63,7 +63,8 @@ enum { RDMA_USER_CM_CMD_LEAVE_MCAST, RDMA_USER_CM_CMD_MIGRATE_ID, RDMA_USER_CM_CMD_QUERY, - RDMA_USER_CM_CMD_BIND + RDMA_USER_CM_CMD_BIND, + RDMA_USER_CM_CMD_RESOLVE_ADDR }; /* @@ -117,6 +118,16 @@ struct rdma_ucm_resolve_ip { __u32 timeout_ms; }; +struct rdma_ucm_resolve_addr { + __u32 id; + __u32 timeout_ms; + __u16 src_size; + __u16 dst_size; + __u32 reserved; + struct sockaddr_storage src_addr; + struct sockaddr_storage dst_addr; +}; + struct rdma_ucm_resolve_route { __u32 id; __u32 timeout_ms; -- cgit v1.2.3 From 5bc2b7b397b02026a0596a7807443a18422733fa Mon Sep 17 00:00:00 2001 From: Sean Hefty Date: Wed, 29 May 2013 10:09:33 -0700 Subject: RDMA/ucma: Allow user space to specify AF_IB when joining multicast Allow user space applications to join multicast groups using MGIDs directly. MGIDs may be passed using AF_IB addresses. Since the current multicast join command only supports addresses as large as sockaddr_in6, define a new structure for joining addresses specified using sockaddr_ib. Since AF_IB allows the user to specify the qkey when resolving a remote UD QP address, when joining the multicast group use the qkey value, if one has been assigned. Signed-off-by: Sean Hefty Signed-off-by: Roland Dreier --- include/uapi/rdma/rdma_user_cm.h | 12 +++++++++++- 1 file changed, 11 insertions(+), 1 deletion(-) (limited to 'include') diff --git a/include/uapi/rdma/rdma_user_cm.h b/include/uapi/rdma/rdma_user_cm.h index 6d03f9c16f1e..99b80abf360a 100644 --- a/include/uapi/rdma/rdma_user_cm.h +++ b/include/uapi/rdma/rdma_user_cm.h @@ -64,7 +64,8 @@ enum { RDMA_USER_CM_CMD_MIGRATE_ID, RDMA_USER_CM_CMD_QUERY, RDMA_USER_CM_CMD_BIND, - RDMA_USER_CM_CMD_RESOLVE_ADDR + RDMA_USER_CM_CMD_RESOLVE_ADDR, + RDMA_USER_CM_CMD_JOIN_MCAST }; /* @@ -242,6 +243,15 @@ struct rdma_ucm_join_ip_mcast { __u32 id; }; +struct rdma_ucm_join_mcast { + __u64 response; /* rdma_ucma_create_id_resp */ + __u64 uid; + __u32 id; + __u16 addr_size; + __u16 reserved; + struct sockaddr_storage addr; +}; + struct rdma_ucm_get_event { __u64 response; }; -- cgit v1.2.3 From 0134f16bc91cc15a38c867b81568b791c9b626aa Mon Sep 17 00:00:00 2001 From: Jack Morgenstein Date: Sun, 7 Jul 2013 17:25:52 +0300 Subject: IB/core: Add reserved values to enums for low-level driver use Continue the approach taken by commit d2b57063e4a ("IB/core: Reserve bits in enum ib_qp_create_flags for low-level driver use") and add reserved entries to the ib_qp_type and ib_wr_opcode enums. Low-level drivers can then define macros to use these reserved values, giving proper names to the macros for readability. Also add a range of reserved flags to enum ib_send_flags. The mlx5 IB driver uses the new additions. Signed-off-by: Jack Morgenstein Signed-off-by: Roland Dreier --- include/rdma/ib_verbs.h | 35 +++++++++++++++++++++++++++++++++-- 1 file changed, 33 insertions(+), 2 deletions(-) (limited to 'include') diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h index 98cc4b29fc5b..645c3cedce9c 100644 --- a/include/rdma/ib_verbs.h +++ b/include/rdma/ib_verbs.h @@ -610,7 +610,21 @@ enum ib_qp_type { IB_QPT_RAW_PACKET = 8, IB_QPT_XRC_INI = 9, IB_QPT_XRC_TGT, - IB_QPT_MAX + IB_QPT_MAX, + /* Reserve a range for qp types internal to the low level driver. + * These qp types will not be visible at the IB core layer, so the + * IB_QPT_MAX usages should not be affected in the core layer + */ + IB_QPT_RESERVED1 = 0x1000, + IB_QPT_RESERVED2, + IB_QPT_RESERVED3, + IB_QPT_RESERVED4, + IB_QPT_RESERVED5, + IB_QPT_RESERVED6, + IB_QPT_RESERVED7, + IB_QPT_RESERVED8, + IB_QPT_RESERVED9, + IB_QPT_RESERVED10, }; enum ib_qp_create_flags { @@ -766,6 +780,19 @@ enum ib_wr_opcode { IB_WR_MASKED_ATOMIC_CMP_AND_SWP, IB_WR_MASKED_ATOMIC_FETCH_AND_ADD, IB_WR_BIND_MW, + /* reserve values for low level drivers' internal use. + * These values will not be used at all in the ib core layer. + */ + IB_WR_RESERVED1 = 0xf0, + IB_WR_RESERVED2, + IB_WR_RESERVED3, + IB_WR_RESERVED4, + IB_WR_RESERVED5, + IB_WR_RESERVED6, + IB_WR_RESERVED7, + IB_WR_RESERVED8, + IB_WR_RESERVED9, + IB_WR_RESERVED10, }; enum ib_send_flags { @@ -773,7 +800,11 @@ enum ib_send_flags { IB_SEND_SIGNALED = (1<<1), IB_SEND_SOLICITED = (1<<2), IB_SEND_INLINE = (1<<3), - IB_SEND_IP_CSUM = (1<<4) + IB_SEND_IP_CSUM = (1<<4), + + /* reserve bits 26-31 for low level drivers' internal use */ + IB_SEND_RESERVED_START = (1 << 26), + IB_SEND_RESERVED_END = (1 << 31), }; struct ib_sge { -- cgit v1.2.3 From e126ba97dba9edeb6fafa3665b5f8497fc9cdf8c Mon Sep 17 00:00:00 2001 From: Eli Cohen Date: Sun, 7 Jul 2013 17:25:49 +0300 Subject: mlx5: Add driver for Mellanox Connect-IB adapters The driver is comprised of two kernel modules: mlx5_ib and mlx5_core. This partitioning resembles what we have for mlx4, except that mlx5_ib is the pci device driver and not mlx5_core. mlx5_core is essentially a library that provides general functionality that is intended to be used by other Mellanox devices that will be introduced in the future. mlx5_ib has a similar role as any hardware device under drivers/infiniband/hw. Signed-off-by: Eli Cohen Signed-off-by: Jack Morgenstein Signed-off-by: Or Gerlitz [ Merge in coccinelle fixes from Fengguang Wu . - Roland ] Signed-off-by: Roland Dreier --- include/linux/mlx5/cmd.h | 51 +++ include/linux/mlx5/cq.h | 165 ++++++++ include/linux/mlx5/device.h | 893 ++++++++++++++++++++++++++++++++++++++++++ include/linux/mlx5/doorbell.h | 79 ++++ include/linux/mlx5/driver.h | 769 ++++++++++++++++++++++++++++++++++++ include/linux/mlx5/qp.h | 467 ++++++++++++++++++++++ include/linux/mlx5/srq.h | 41 ++ 7 files changed, 2465 insertions(+) create mode 100644 include/linux/mlx5/cmd.h create mode 100644 include/linux/mlx5/cq.h create mode 100644 include/linux/mlx5/device.h create mode 100644 include/linux/mlx5/doorbell.h create mode 100644 include/linux/mlx5/driver.h create mode 100644 include/linux/mlx5/qp.h create mode 100644 include/linux/mlx5/srq.h (limited to 'include') diff --git a/include/linux/mlx5/cmd.h b/include/linux/mlx5/cmd.h new file mode 100644 index 000000000000..2826a4b6071e --- /dev/null +++ b/include/linux/mlx5/cmd.h @@ -0,0 +1,51 @@ +/* + * Copyright (c) 2013, Mellanox Technologies inc. All rights reserved. + * + * This software is available to you under a choice of one of two + * licenses. You may choose to be licensed under the terms of the GNU + * General Public License (GPL) Version 2, available from the file + * COPYING in the main directory of this source tree, or the + * OpenIB.org BSD license below: + * + * Redistribution and use in source and binary forms, with or + * without modification, are permitted provided that the following + * conditions are met: + * + * - Redistributions of source code must retain the above + * copyright notice, this list of conditions and the following + * disclaimer. + * + * - Redistributions in binary form must reproduce the above + * copyright notice, this list of conditions and the following + * disclaimer in the documentation and/or other materials + * provided with the distribution. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND + * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS + * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN + * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN + * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#ifndef MLX5_CMD_H +#define MLX5_CMD_H + +#include + +struct manage_pages_layout { + u64 ptr; + u32 reserved; + u16 num_entries; + u16 func_id; +}; + + +struct mlx5_cmd_alloc_uar_imm_out { + u32 rsvd[3]; + u32 uarn; +}; + +#endif /* MLX5_CMD_H */ diff --git a/include/linux/mlx5/cq.h b/include/linux/mlx5/cq.h new file mode 100644 index 000000000000..3db67f73d96d --- /dev/null +++ b/include/linux/mlx5/cq.h @@ -0,0 +1,165 @@ +/* + * Copyright (c) 2013, Mellanox Technologies inc. All rights reserved. + * + * This software is available to you under a choice of one of two + * licenses. You may choose to be licensed under the terms of the GNU + * General Public License (GPL) Version 2, available from the file + * COPYING in the main directory of this source tree, or the + * OpenIB.org BSD license below: + * + * Redistribution and use in source and binary forms, with or + * without modification, are permitted provided that the following + * conditions are met: + * + * - Redistributions of source code must retain the above + * copyright notice, this list of conditions and the following + * disclaimer. + * + * - Redistributions in binary form must reproduce the above + * copyright notice, this list of conditions and the following + * disclaimer in the documentation and/or other materials + * provided with the distribution. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND + * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS + * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN + * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN + * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#ifndef MLX5_CORE_CQ_H +#define MLX5_CORE_CQ_H + +#include +#include + + +struct mlx5_core_cq { + u32 cqn; + int cqe_sz; + __be32 *set_ci_db; + __be32 *arm_db; + atomic_t refcount; + struct completion free; + unsigned vector; + int irqn; + void (*comp) (struct mlx5_core_cq *); + void (*event) (struct mlx5_core_cq *, enum mlx5_event); + struct mlx5_uar *uar; + u32 cons_index; + unsigned arm_sn; + struct mlx5_rsc_debug *dbg; + int pid; +}; + + +enum { + MLX5_CQE_SYNDROME_LOCAL_LENGTH_ERR = 0x01, + MLX5_CQE_SYNDROME_LOCAL_QP_OP_ERR = 0x02, + MLX5_CQE_SYNDROME_LOCAL_PROT_ERR = 0x04, + MLX5_CQE_SYNDROME_WR_FLUSH_ERR = 0x05, + MLX5_CQE_SYNDROME_MW_BIND_ERR = 0x06, + MLX5_CQE_SYNDROME_BAD_RESP_ERR = 0x10, + MLX5_CQE_SYNDROME_LOCAL_ACCESS_ERR = 0x11, + MLX5_CQE_SYNDROME_REMOTE_INVAL_REQ_ERR = 0x12, + MLX5_CQE_SYNDROME_REMOTE_ACCESS_ERR = 0x13, + MLX5_CQE_SYNDROME_REMOTE_OP_ERR = 0x14, + MLX5_CQE_SYNDROME_TRANSPORT_RETRY_EXC_ERR = 0x15, + MLX5_CQE_SYNDROME_RNR_RETRY_EXC_ERR = 0x16, + MLX5_CQE_SYNDROME_REMOTE_ABORTED_ERR = 0x22, +}; + +enum { + MLX5_CQE_OWNER_MASK = 1, + MLX5_CQE_REQ = 0, + MLX5_CQE_RESP_WR_IMM = 1, + MLX5_CQE_RESP_SEND = 2, + MLX5_CQE_RESP_SEND_IMM = 3, + MLX5_CQE_RESP_SEND_INV = 4, + MLX5_CQE_RESIZE_CQ = 0xff, /* TBD */ + MLX5_CQE_REQ_ERR = 13, + MLX5_CQE_RESP_ERR = 14, +}; + +enum { + MLX5_CQ_MODIFY_RESEIZE = 0, + MLX5_CQ_MODIFY_MODER = 1, + MLX5_CQ_MODIFY_MAPPING = 2, +}; + +struct mlx5_cq_modify_params { + int type; + union { + struct { + u32 page_offset; + u8 log_cq_size; + } resize; + + struct { + } moder; + + struct { + } mapping; + } params; +}; + +enum { + CQE_SIZE_64 = 0, + CQE_SIZE_128 = 1, +}; + +static inline int cqe_sz_to_mlx_sz(u8 size) +{ + return size == 64 ? CQE_SIZE_64 : CQE_SIZE_128; +} + +static inline void mlx5_cq_set_ci(struct mlx5_core_cq *cq) +{ + *cq->set_ci_db = cpu_to_be32(cq->cons_index & 0xffffff); +} + +enum { + MLX5_CQ_DB_REQ_NOT_SOL = 1 << 24, + MLX5_CQ_DB_REQ_NOT = 0 << 24 +}; + +static inline void mlx5_cq_arm(struct mlx5_core_cq *cq, u32 cmd, + void __iomem *uar_page, + spinlock_t *doorbell_lock) +{ + __be32 doorbell[2]; + u32 sn; + u32 ci; + + sn = cq->arm_sn & 3; + ci = cq->cons_index & 0xffffff; + + *cq->arm_db = cpu_to_be32(sn << 28 | cmd | ci); + + /* Make sure that the doorbell record in host memory is + * written before ringing the doorbell via PCI MMIO. + */ + wmb(); + + doorbell[0] = cpu_to_be32(sn << 28 | cmd | ci); + doorbell[1] = cpu_to_be32(cq->cqn); + + mlx5_write64(doorbell, uar_page + MLX5_CQ_DOORBELL, doorbell_lock); +} + +int mlx5_init_cq_table(struct mlx5_core_dev *dev); +void mlx5_cleanup_cq_table(struct mlx5_core_dev *dev); +int mlx5_core_create_cq(struct mlx5_core_dev *dev, struct mlx5_core_cq *cq, + struct mlx5_create_cq_mbox_in *in, int inlen); +int mlx5_core_destroy_cq(struct mlx5_core_dev *dev, struct mlx5_core_cq *cq); +int mlx5_core_query_cq(struct mlx5_core_dev *dev, struct mlx5_core_cq *cq, + struct mlx5_query_cq_mbox_out *out); +int mlx5_core_modify_cq(struct mlx5_core_dev *dev, struct mlx5_core_cq *cq, + int type, struct mlx5_cq_modify_params *params); +int mlx5_debug_cq_add(struct mlx5_core_dev *dev, struct mlx5_core_cq *cq); +void mlx5_debug_cq_remove(struct mlx5_core_dev *dev, struct mlx5_core_cq *cq); + +#endif /* MLX5_CORE_CQ_H */ diff --git a/include/linux/mlx5/device.h b/include/linux/mlx5/device.h new file mode 100644 index 000000000000..51390915e538 --- /dev/null +++ b/include/linux/mlx5/device.h @@ -0,0 +1,893 @@ +/* + * Copyright (c) 2013, Mellanox Technologies inc. All rights reserved. + * + * This software is available to you under a choice of one of two + * licenses. You may choose to be licensed under the terms of the GNU + * General Public License (GPL) Version 2, available from the file + * COPYING in the main directory of this source tree, or the + * OpenIB.org BSD license below: + * + * Redistribution and use in source and binary forms, with or + * without modification, are permitted provided that the following + * conditions are met: + * + * - Redistributions of source code must retain the above + * copyright notice, this list of conditions and the following + * disclaimer. + * + * - Redistributions in binary form must reproduce the above + * copyright notice, this list of conditions and the following + * disclaimer in the documentation and/or other materials + * provided with the distribution. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND + * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS + * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN + * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN + * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#ifndef MLX5_DEVICE_H +#define MLX5_DEVICE_H + +#include +#include + +#if defined(__LITTLE_ENDIAN) +#define MLX5_SET_HOST_ENDIANNESS 0 +#elif defined(__BIG_ENDIAN) +#define MLX5_SET_HOST_ENDIANNESS 0x80 +#else +#error Host endianness not defined +#endif + +enum { + MLX5_MAX_COMMANDS = 32, + MLX5_CMD_DATA_BLOCK_SIZE = 512, + MLX5_PCI_CMD_XPORT = 7, +}; + +enum { + MLX5_EXTENDED_UD_AV = 0x80000000, +}; + +enum { + MLX5_CQ_STATE_ARMED = 9, + MLX5_CQ_STATE_ALWAYS_ARMED = 0xb, + MLX5_CQ_STATE_FIRED = 0xa, +}; + +enum { + MLX5_STAT_RATE_OFFSET = 5, +}; + +enum { + MLX5_INLINE_SEG = 0x80000000, +}; + +enum { + MLX5_PERM_LOCAL_READ = 1 << 2, + MLX5_PERM_LOCAL_WRITE = 1 << 3, + MLX5_PERM_REMOTE_READ = 1 << 4, + MLX5_PERM_REMOTE_WRITE = 1 << 5, + MLX5_PERM_ATOMIC = 1 << 6, + MLX5_PERM_UMR_EN = 1 << 7, +}; + +enum { + MLX5_PCIE_CTRL_SMALL_FENCE = 1 << 0, + MLX5_PCIE_CTRL_RELAXED_ORDERING = 1 << 2, + MLX5_PCIE_CTRL_NO_SNOOP = 1 << 3, + MLX5_PCIE_CTRL_TLP_PROCE_EN = 1 << 6, + MLX5_PCIE_CTRL_TPH_MASK = 3 << 4, +}; + +enum { + MLX5_ACCESS_MODE_PA = 0, + MLX5_ACCESS_MODE_MTT = 1, + MLX5_ACCESS_MODE_KLM = 2 +}; + +enum { + MLX5_MKEY_REMOTE_INVAL = 1 << 24, + MLX5_MKEY_FLAG_SYNC_UMR = 1 << 29, + MLX5_MKEY_BSF_EN = 1 << 30, + MLX5_MKEY_LEN64 = 1 << 31, +}; + +enum { + MLX5_EN_RD = (u64)1, + MLX5_EN_WR = (u64)2 +}; + +enum { + MLX5_BF_REGS_PER_PAGE = 4, + MLX5_MAX_UAR_PAGES = 1 << 8, + MLX5_MAX_UUARS = MLX5_MAX_UAR_PAGES * MLX5_BF_REGS_PER_PAGE, +}; + +enum { + MLX5_MKEY_MASK_LEN = 1ull << 0, + MLX5_MKEY_MASK_PAGE_SIZE = 1ull << 1, + MLX5_MKEY_MASK_START_ADDR = 1ull << 6, + MLX5_MKEY_MASK_PD = 1ull << 7, + MLX5_MKEY_MASK_EN_RINVAL = 1ull << 8, + MLX5_MKEY_MASK_BSF_EN = 1ull << 12, + MLX5_MKEY_MASK_KEY = 1ull << 13, + MLX5_MKEY_MASK_QPN = 1ull << 14, + MLX5_MKEY_MASK_LR = 1ull << 17, + MLX5_MKEY_MASK_LW = 1ull << 18, + MLX5_MKEY_MASK_RR = 1ull << 19, + MLX5_MKEY_MASK_RW = 1ull << 20, + MLX5_MKEY_MASK_A = 1ull << 21, + MLX5_MKEY_MASK_SMALL_FENCE = 1ull << 23, + MLX5_MKEY_MASK_FREE = 1ull << 29, +}; + +enum mlx5_event { + MLX5_EVENT_TYPE_COMP = 0x0, + + MLX5_EVENT_TYPE_PATH_MIG = 0x01, + MLX5_EVENT_TYPE_COMM_EST = 0x02, + MLX5_EVENT_TYPE_SQ_DRAINED = 0x03, + MLX5_EVENT_TYPE_SRQ_LAST_WQE = 0x13, + MLX5_EVENT_TYPE_SRQ_RQ_LIMIT = 0x14, + + MLX5_EVENT_TYPE_CQ_ERROR = 0x04, + MLX5_EVENT_TYPE_WQ_CATAS_ERROR = 0x05, + MLX5_EVENT_TYPE_PATH_MIG_FAILED = 0x07, + MLX5_EVENT_TYPE_WQ_INVAL_REQ_ERROR = 0x10, + MLX5_EVENT_TYPE_WQ_ACCESS_ERROR = 0x11, + MLX5_EVENT_TYPE_SRQ_CATAS_ERROR = 0x12, + + MLX5_EVENT_TYPE_INTERNAL_ERROR = 0x08, + MLX5_EVENT_TYPE_PORT_CHANGE = 0x09, + MLX5_EVENT_TYPE_GPIO_EVENT = 0x15, + MLX5_EVENT_TYPE_REMOTE_CONFIG = 0x19, + + MLX5_EVENT_TYPE_DB_BF_CONGESTION = 0x1a, + MLX5_EVENT_TYPE_STALL_EVENT = 0x1b, + + MLX5_EVENT_TYPE_CMD = 0x0a, + MLX5_EVENT_TYPE_PAGE_REQUEST = 0xb, +}; + +enum { + MLX5_PORT_CHANGE_SUBTYPE_DOWN = 1, + MLX5_PORT_CHANGE_SUBTYPE_ACTIVE = 4, + MLX5_PORT_CHANGE_SUBTYPE_INITIALIZED = 5, + MLX5_PORT_CHANGE_SUBTYPE_LID = 6, + MLX5_PORT_CHANGE_SUBTYPE_PKEY = 7, + MLX5_PORT_CHANGE_SUBTYPE_GUID = 8, + MLX5_PORT_CHANGE_SUBTYPE_CLIENT_REREG = 9, +}; + +enum { + MLX5_DEV_CAP_FLAG_RC = 1LL << 0, + MLX5_DEV_CAP_FLAG_UC = 1LL << 1, + MLX5_DEV_CAP_FLAG_UD = 1LL << 2, + MLX5_DEV_CAP_FLAG_XRC = 1LL << 3, + MLX5_DEV_CAP_FLAG_SRQ = 1LL << 6, + MLX5_DEV_CAP_FLAG_BAD_PKEY_CNTR = 1LL << 8, + MLX5_DEV_CAP_FLAG_BAD_QKEY_CNTR = 1LL << 9, + MLX5_DEV_CAP_FLAG_APM = 1LL << 17, + MLX5_DEV_CAP_FLAG_ATOMIC = 1LL << 18, + MLX5_DEV_CAP_FLAG_ON_DMND_PG = 1LL << 24, + MLX5_DEV_CAP_FLAG_RESIZE_SRQ = 1LL << 32, + MLX5_DEV_CAP_FLAG_REMOTE_FENCE = 1LL << 38, + MLX5_DEV_CAP_FLAG_TLP_HINTS = 1LL << 39, + MLX5_DEV_CAP_FLAG_SIG_HAND_OVER = 1LL << 40, + MLX5_DEV_CAP_FLAG_DCT = 1LL << 41, + MLX5_DEV_CAP_FLAG_CMDIF_CSUM = 1LL << 46, +}; + +enum { + MLX5_OPCODE_NOP = 0x00, + MLX5_OPCODE_SEND_INVAL = 0x01, + MLX5_OPCODE_RDMA_WRITE = 0x08, + MLX5_OPCODE_RDMA_WRITE_IMM = 0x09, + MLX5_OPCODE_SEND = 0x0a, + MLX5_OPCODE_SEND_IMM = 0x0b, + MLX5_OPCODE_RDMA_READ = 0x10, + MLX5_OPCODE_ATOMIC_CS = 0x11, + MLX5_OPCODE_ATOMIC_FA = 0x12, + MLX5_OPCODE_ATOMIC_MASKED_CS = 0x14, + MLX5_OPCODE_ATOMIC_MASKED_FA = 0x15, + MLX5_OPCODE_BIND_MW = 0x18, + MLX5_OPCODE_CONFIG_CMD = 0x1f, + + MLX5_RECV_OPCODE_RDMA_WRITE_IMM = 0x00, + MLX5_RECV_OPCODE_SEND = 0x01, + MLX5_RECV_OPCODE_SEND_IMM = 0x02, + MLX5_RECV_OPCODE_SEND_INVAL = 0x03, + + MLX5_CQE_OPCODE_ERROR = 0x1e, + MLX5_CQE_OPCODE_RESIZE = 0x16, + + MLX5_OPCODE_SET_PSV = 0x20, + MLX5_OPCODE_GET_PSV = 0x21, + MLX5_OPCODE_CHECK_PSV = 0x22, + MLX5_OPCODE_RGET_PSV = 0x26, + MLX5_OPCODE_RCHECK_PSV = 0x27, + + MLX5_OPCODE_UMR = 0x25, + +}; + +enum { + MLX5_SET_PORT_RESET_QKEY = 0, + MLX5_SET_PORT_GUID0 = 16, + MLX5_SET_PORT_NODE_GUID = 17, + MLX5_SET_PORT_SYS_GUID = 18, + MLX5_SET_PORT_GID_TABLE = 19, + MLX5_SET_PORT_PKEY_TABLE = 20, +}; + +enum { + MLX5_MAX_PAGE_SHIFT = 31 +}; + +struct mlx5_inbox_hdr { + __be16 opcode; + u8 rsvd[4]; + __be16 opmod; +}; + +struct mlx5_outbox_hdr { + u8 status; + u8 rsvd[3]; + __be32 syndrome; +}; + +struct mlx5_cmd_query_adapter_mbox_in { + struct mlx5_inbox_hdr hdr; + u8 rsvd[8]; +}; + +struct mlx5_cmd_query_adapter_mbox_out { + struct mlx5_outbox_hdr hdr; + u8 rsvd0[24]; + u8 intapin; + u8 rsvd1[13]; + __be16 vsd_vendor_id; + u8 vsd[208]; + u8 vsd_psid[16]; +}; + +struct mlx5_hca_cap { + u8 rsvd1[16]; + u8 log_max_srq_sz; + u8 log_max_qp_sz; + u8 rsvd2; + u8 log_max_qp; + u8 log_max_strq_sz; + u8 log_max_srqs; + u8 rsvd4[2]; + u8 rsvd5; + u8 log_max_cq_sz; + u8 rsvd6; + u8 log_max_cq; + u8 log_max_eq_sz; + u8 log_max_mkey; + u8 rsvd7; + u8 log_max_eq; + u8 max_indirection; + u8 log_max_mrw_sz; + u8 log_max_bsf_list_sz; + u8 log_max_klm_list_sz; + u8 rsvd_8_0; + u8 log_max_ra_req_dc; + u8 rsvd_8_1; + u8 log_max_ra_res_dc; + u8 rsvd9; + u8 log_max_ra_req_qp; + u8 rsvd10; + u8 log_max_ra_res_qp; + u8 rsvd11[4]; + __be16 max_qp_count; + __be16 rsvd12; + u8 rsvd13; + u8 local_ca_ack_delay; + u8 rsvd14; + u8 num_ports; + u8 log_max_msg; + u8 rsvd15[3]; + __be16 stat_rate_support; + u8 rsvd16[2]; + __be64 flags; + u8 rsvd17; + u8 uar_sz; + u8 rsvd18; + u8 log_pg_sz; + __be16 bf_log_bf_reg_size; + u8 rsvd19[4]; + __be16 max_desc_sz_sq; + u8 rsvd20[2]; + __be16 max_desc_sz_rq; + u8 rsvd21[2]; + __be16 max_desc_sz_sq_dc; + u8 rsvd22[4]; + __be16 max_qp_mcg; + u8 rsvd23; + u8 log_max_mcg; + u8 rsvd24; + u8 log_max_pd; + u8 rsvd25; + u8 log_max_xrcd; + u8 rsvd26[40]; + __be32 uar_page_sz; + u8 rsvd27[28]; + u8 log_msx_atomic_size_qp; + u8 rsvd28[2]; + u8 log_msx_atomic_size_dc; + u8 rsvd29[76]; +}; + + +struct mlx5_cmd_query_hca_cap_mbox_in { + struct mlx5_inbox_hdr hdr; + u8 rsvd[8]; +}; + + +struct mlx5_cmd_query_hca_cap_mbox_out { + struct mlx5_outbox_hdr hdr; + u8 rsvd0[8]; + struct mlx5_hca_cap hca_cap; +}; + + +struct mlx5_cmd_set_hca_cap_mbox_in { + struct mlx5_inbox_hdr hdr; + u8 rsvd[8]; + struct mlx5_hca_cap hca_cap; +}; + + +struct mlx5_cmd_set_hca_cap_mbox_out { + struct mlx5_outbox_hdr hdr; + u8 rsvd0[8]; +}; + + +struct mlx5_cmd_init_hca_mbox_in { + struct mlx5_inbox_hdr hdr; + u8 rsvd0[2]; + __be16 profile; + u8 rsvd1[4]; +}; + +struct mlx5_cmd_init_hca_mbox_out { + struct mlx5_outbox_hdr hdr; + u8 rsvd[8]; +}; + +struct mlx5_cmd_teardown_hca_mbox_in { + struct mlx5_inbox_hdr hdr; + u8 rsvd0[2]; + __be16 profile; + u8 rsvd1[4]; +}; + +struct mlx5_cmd_teardown_hca_mbox_out { + struct mlx5_outbox_hdr hdr; + u8 rsvd[8]; +}; + +struct mlx5_cmd_layout { + u8 type; + u8 rsvd0[3]; + __be32 inlen; + __be64 in_ptr; + __be32 in[4]; + __be32 out[4]; + __be64 out_ptr; + __be32 outlen; + u8 token; + u8 sig; + u8 rsvd1; + u8 status_own; +}; + + +struct health_buffer { + __be32 assert_var[5]; + __be32 rsvd0[3]; + __be32 assert_exit_ptr; + __be32 assert_callra; + __be32 rsvd1[2]; + __be32 fw_ver; + __be32 hw_id; + __be32 rsvd2; + u8 irisc_index; + u8 synd; + __be16 ext_sync; +}; + +struct mlx5_init_seg { + __be32 fw_rev; + __be32 cmdif_rev_fw_sub; + __be32 rsvd0[2]; + __be32 cmdq_addr_h; + __be32 cmdq_addr_l_sz; + __be32 cmd_dbell; + __be32 rsvd1[121]; + struct health_buffer health; + __be32 rsvd2[884]; + __be32 health_counter; + __be32 rsvd3[1023]; + __be64 ieee1588_clk; + __be32 ieee1588_clk_type; + __be32 clr_intx; +}; + +struct mlx5_eqe_comp { + __be32 reserved[6]; + __be32 cqn; +}; + +struct mlx5_eqe_qp_srq { + __be32 reserved[6]; + __be32 qp_srq_n; +}; + +struct mlx5_eqe_cq_err { + __be32 cqn; + u8 reserved1[7]; + u8 syndrome; +}; + +struct mlx5_eqe_dropped_packet { +}; + +struct mlx5_eqe_port_state { + u8 reserved0[8]; + u8 port; +}; + +struct mlx5_eqe_gpio { + __be32 reserved0[2]; + __be64 gpio_event; +}; + +struct mlx5_eqe_congestion { + u8 type; + u8 rsvd0; + u8 congestion_level; +}; + +struct mlx5_eqe_stall_vl { + u8 rsvd0[3]; + u8 port_vl; +}; + +struct mlx5_eqe_cmd { + __be32 vector; + __be32 rsvd[6]; +}; + +struct mlx5_eqe_page_req { + u8 rsvd0[2]; + __be16 func_id; + u8 rsvd1[2]; + __be16 num_pages; + __be32 rsvd2[5]; +}; + +union ev_data { + __be32 raw[7]; + struct mlx5_eqe_cmd cmd; + struct mlx5_eqe_comp comp; + struct mlx5_eqe_qp_srq qp_srq; + struct mlx5_eqe_cq_err cq_err; + struct mlx5_eqe_dropped_packet dp; + struct mlx5_eqe_port_state port; + struct mlx5_eqe_gpio gpio; + struct mlx5_eqe_congestion cong; + struct mlx5_eqe_stall_vl stall_vl; + struct mlx5_eqe_page_req req_pages; +} __packed; + +struct mlx5_eqe { + u8 rsvd0; + u8 type; + u8 rsvd1; + u8 sub_type; + __be32 rsvd2[7]; + union ev_data data; + __be16 rsvd3; + u8 signature; + u8 owner; +} __packed; + +struct mlx5_cmd_prot_block { + u8 data[MLX5_CMD_DATA_BLOCK_SIZE]; + u8 rsvd0[48]; + __be64 next; + __be32 block_num; + u8 rsvd1; + u8 token; + u8 ctrl_sig; + u8 sig; +}; + +struct mlx5_err_cqe { + u8 rsvd0[32]; + __be32 srqn; + u8 rsvd1[18]; + u8 vendor_err_synd; + u8 syndrome; + __be32 s_wqe_opcode_qpn; + __be16 wqe_counter; + u8 signature; + u8 op_own; +}; + +struct mlx5_cqe64 { + u8 rsvd0[17]; + u8 ml_path; + u8 rsvd20[4]; + __be16 slid; + __be32 flags_rqpn; + u8 rsvd28[4]; + __be32 srqn; + __be32 imm_inval_pkey; + u8 rsvd40[4]; + __be32 byte_cnt; + __be64 timestamp; + __be32 sop_drop_qpn; + __be16 wqe_counter; + u8 signature; + u8 op_own; +}; + +struct mlx5_wqe_srq_next_seg { + u8 rsvd0[2]; + __be16 next_wqe_index; + u8 signature; + u8 rsvd1[11]; +}; + +union mlx5_ext_cqe { + struct ib_grh grh; + u8 inl[64]; +}; + +struct mlx5_cqe128 { + union mlx5_ext_cqe inl_grh; + struct mlx5_cqe64 cqe64; +}; + +struct mlx5_srq_ctx { + u8 state_log_sz; + u8 rsvd0[3]; + __be32 flags_xrcd; + __be32 pgoff_cqn; + u8 rsvd1[4]; + u8 log_pg_sz; + u8 rsvd2[7]; + __be32 pd; + __be16 lwm; + __be16 wqe_cnt; + u8 rsvd3[8]; + __be64 db_record; +}; + +struct mlx5_create_srq_mbox_in { + struct mlx5_inbox_hdr hdr; + __be32 input_srqn; + u8 rsvd0[4]; + struct mlx5_srq_ctx ctx; + u8 rsvd1[208]; + __be64 pas[0]; +}; + +struct mlx5_create_srq_mbox_out { + struct mlx5_outbox_hdr hdr; + __be32 srqn; + u8 rsvd[4]; +}; + +struct mlx5_destroy_srq_mbox_in { + struct mlx5_inbox_hdr hdr; + __be32 srqn; + u8 rsvd[4]; +}; + +struct mlx5_destroy_srq_mbox_out { + struct mlx5_outbox_hdr hdr; + u8 rsvd[8]; +}; + +struct mlx5_query_srq_mbox_in { + struct mlx5_inbox_hdr hdr; + __be32 srqn; + u8 rsvd0[4]; +}; + +struct mlx5_query_srq_mbox_out { + struct mlx5_outbox_hdr hdr; + u8 rsvd0[8]; + struct mlx5_srq_ctx ctx; + u8 rsvd1[32]; + __be64 pas[0]; +}; + +struct mlx5_arm_srq_mbox_in { + struct mlx5_inbox_hdr hdr; + __be32 srqn; + __be16 rsvd; + __be16 lwm; +}; + +struct mlx5_arm_srq_mbox_out { + struct mlx5_outbox_hdr hdr; + u8 rsvd[8]; +}; + +struct mlx5_cq_context { + u8 status; + u8 cqe_sz_flags; + u8 st; + u8 rsvd3; + u8 rsvd4[6]; + __be16 page_offset; + __be32 log_sz_usr_page; + __be16 cq_period; + __be16 cq_max_count; + __be16 rsvd20; + __be16 c_eqn; + u8 log_pg_sz; + u8 rsvd25[7]; + __be32 last_notified_index; + __be32 solicit_producer_index; + __be32 consumer_counter; + __be32 producer_counter; + u8 rsvd48[8]; + __be64 db_record_addr; +}; + +struct mlx5_create_cq_mbox_in { + struct mlx5_inbox_hdr hdr; + __be32 input_cqn; + u8 rsvdx[4]; + struct mlx5_cq_context ctx; + u8 rsvd6[192]; + __be64 pas[0]; +}; + +struct mlx5_create_cq_mbox_out { + struct mlx5_outbox_hdr hdr; + __be32 cqn; + u8 rsvd0[4]; +}; + +struct mlx5_destroy_cq_mbox_in { + struct mlx5_inbox_hdr hdr; + __be32 cqn; + u8 rsvd0[4]; +}; + +struct mlx5_destroy_cq_mbox_out { + struct mlx5_outbox_hdr hdr; + u8 rsvd0[8]; +}; + +struct mlx5_query_cq_mbox_in { + struct mlx5_inbox_hdr hdr; + __be32 cqn; + u8 rsvd0[4]; +}; + +struct mlx5_query_cq_mbox_out { + struct mlx5_outbox_hdr hdr; + u8 rsvd0[8]; + struct mlx5_cq_context ctx; + u8 rsvd6[16]; + __be64 pas[0]; +}; + +struct mlx5_eq_context { + u8 status; + u8 ec_oi; + u8 st; + u8 rsvd2[7]; + __be16 page_pffset; + __be32 log_sz_usr_page; + u8 rsvd3[7]; + u8 intr; + u8 log_page_size; + u8 rsvd4[15]; + __be32 consumer_counter; + __be32 produser_counter; + u8 rsvd5[16]; +}; + +struct mlx5_create_eq_mbox_in { + struct mlx5_inbox_hdr hdr; + u8 rsvd0[3]; + u8 input_eqn; + u8 rsvd1[4]; + struct mlx5_eq_context ctx; + u8 rsvd2[8]; + __be64 events_mask; + u8 rsvd3[176]; + __be64 pas[0]; +}; + +struct mlx5_create_eq_mbox_out { + struct mlx5_outbox_hdr hdr; + u8 rsvd0[3]; + u8 eq_number; + u8 rsvd1[4]; +}; + +struct mlx5_destroy_eq_mbox_in { + struct mlx5_inbox_hdr hdr; + u8 rsvd0[3]; + u8 eqn; + u8 rsvd1[4]; +}; + +struct mlx5_destroy_eq_mbox_out { + struct mlx5_outbox_hdr hdr; + u8 rsvd[8]; +}; + +struct mlx5_map_eq_mbox_in { + struct mlx5_inbox_hdr hdr; + __be64 mask; + u8 mu; + u8 rsvd0[2]; + u8 eqn; + u8 rsvd1[24]; +}; + +struct mlx5_map_eq_mbox_out { + struct mlx5_outbox_hdr hdr; + u8 rsvd[8]; +}; + +struct mlx5_query_eq_mbox_in { + struct mlx5_inbox_hdr hdr; + u8 rsvd0[3]; + u8 eqn; + u8 rsvd1[4]; +}; + +struct mlx5_query_eq_mbox_out { + struct mlx5_outbox_hdr hdr; + u8 rsvd[8]; + struct mlx5_eq_context ctx; +}; + +struct mlx5_mkey_seg { + /* This is a two bit field occupying bits 31-30. + * bit 31 is always 0, + * bit 30 is zero for regular MRs and 1 (e.g free) for UMRs that do not have tanslation + */ + u8 status; + u8 pcie_control; + u8 flags; + u8 version; + __be32 qpn_mkey7_0; + u8 rsvd1[4]; + __be32 flags_pd; + __be64 start_addr; + __be64 len; + __be32 bsfs_octo_size; + u8 rsvd2[16]; + __be32 xlt_oct_size; + u8 rsvd3[3]; + u8 log2_page_size; + u8 rsvd4[4]; +}; + +struct mlx5_query_special_ctxs_mbox_in { + struct mlx5_inbox_hdr hdr; + u8 rsvd[8]; +}; + +struct mlx5_query_special_ctxs_mbox_out { + struct mlx5_outbox_hdr hdr; + __be32 dump_fill_mkey; + __be32 reserved_lkey; +}; + +struct mlx5_create_mkey_mbox_in { + struct mlx5_inbox_hdr hdr; + __be32 input_mkey_index; + u8 rsvd0[4]; + struct mlx5_mkey_seg seg; + u8 rsvd1[16]; + __be32 xlat_oct_act_size; + __be32 bsf_coto_act_size; + u8 rsvd2[168]; + __be64 pas[0]; +}; + +struct mlx5_create_mkey_mbox_out { + struct mlx5_outbox_hdr hdr; + __be32 mkey; + u8 rsvd[4]; +}; + +struct mlx5_destroy_mkey_mbox_in { + struct mlx5_inbox_hdr hdr; + __be32 mkey; + u8 rsvd[4]; +}; + +struct mlx5_destroy_mkey_mbox_out { + struct mlx5_outbox_hdr hdr; + u8 rsvd[8]; +}; + +struct mlx5_query_mkey_mbox_in { + struct mlx5_inbox_hdr hdr; + __be32 mkey; +}; + +struct mlx5_query_mkey_mbox_out { + struct mlx5_outbox_hdr hdr; + __be64 pas[0]; +}; + +struct mlx5_modify_mkey_mbox_in { + struct mlx5_inbox_hdr hdr; + __be32 mkey; + __be64 pas[0]; +}; + +struct mlx5_modify_mkey_mbox_out { + struct mlx5_outbox_hdr hdr; +}; + +struct mlx5_dump_mkey_mbox_in { + struct mlx5_inbox_hdr hdr; +}; + +struct mlx5_dump_mkey_mbox_out { + struct mlx5_outbox_hdr hdr; + __be32 mkey; +}; + +struct mlx5_mad_ifc_mbox_in { + struct mlx5_inbox_hdr hdr; + __be16 remote_lid; + u8 rsvd0; + u8 port; + u8 rsvd1[4]; + u8 data[256]; +}; + +struct mlx5_mad_ifc_mbox_out { + struct mlx5_outbox_hdr hdr; + u8 rsvd[8]; + u8 data[256]; +}; + +struct mlx5_access_reg_mbox_in { + struct mlx5_inbox_hdr hdr; + u8 rsvd0[2]; + __be16 register_id; + __be32 arg; + __be32 data[0]; +}; + +struct mlx5_access_reg_mbox_out { + struct mlx5_outbox_hdr hdr; + u8 rsvd[8]; + __be32 data[0]; +}; + +#define MLX5_ATTR_EXTENDED_PORT_INFO cpu_to_be16(0xff90) + +enum { + MLX_EXT_PORT_CAP_FLAG_EXTENDED_PORT_INFO = 1 << 0 +}; + +#endif /* MLX5_DEVICE_H */ diff --git a/include/linux/mlx5/doorbell.h b/include/linux/mlx5/doorbell.h new file mode 100644 index 000000000000..163a818411e7 --- /dev/null +++ b/include/linux/mlx5/doorbell.h @@ -0,0 +1,79 @@ +/* + * Copyright (c) 2013, Mellanox Technologies inc. All rights reserved. + * + * This software is available to you under a choice of one of two + * licenses. You may choose to be licensed under the terms of the GNU + * General Public License (GPL) Version 2, available from the file + * COPYING in the main directory of this source tree, or the + * OpenIB.org BSD license below: + * + * Redistribution and use in source and binary forms, with or + * without modification, are permitted provided that the following + * conditions are met: + * + * - Redistributions of source code must retain the above + * copyright notice, this list of conditions and the following + * disclaimer. + * + * - Redistributions in binary form must reproduce the above + * copyright notice, this list of conditions and the following + * disclaimer in the documentation and/or other materials + * provided with the distribution. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND + * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS + * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN + * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN + * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#ifndef MLX5_DOORBELL_H +#define MLX5_DOORBELL_H + +#define MLX5_BF_OFFSET 0x800 +#define MLX5_CQ_DOORBELL 0x20 + +#if BITS_PER_LONG == 64 +/* Assume that we can just write a 64-bit doorbell atomically. s390 + * actually doesn't have writeq() but S/390 systems don't even have + * PCI so we won't worry about it. + */ + +#define MLX5_DECLARE_DOORBELL_LOCK(name) +#define MLX5_INIT_DOORBELL_LOCK(ptr) do { } while (0) +#define MLX5_GET_DOORBELL_LOCK(ptr) (NULL) + +static inline void mlx5_write64(__be32 val[2], void __iomem *dest, + spinlock_t *doorbell_lock) +{ + __raw_writeq(*(u64 *)val, dest); +} + +#else + +/* Just fall back to a spinlock to protect the doorbell if + * BITS_PER_LONG is 32 -- there's no portable way to do atomic 64-bit + * MMIO writes. + */ + +#define MLX5_DECLARE_DOORBELL_LOCK(name) spinlock_t name; +#define MLX5_INIT_DOORBELL_LOCK(ptr) spin_lock_init(ptr) +#define MLX5_GET_DOORBELL_LOCK(ptr) (ptr) + +static inline void mlx5_write64(__be32 val[2], void __iomem *dest, + spinlock_t *doorbell_lock) +{ + unsigned long flags; + + spin_lock_irqsave(doorbell_lock, flags); + __raw_writel((__force u32) val[0], dest); + __raw_writel((__force u32) val[1], dest + 4); + spin_unlock_irqrestore(doorbell_lock, flags); +} + +#endif + +#endif /* MLX5_DOORBELL_H */ diff --git a/include/linux/mlx5/driver.h b/include/linux/mlx5/driver.h new file mode 100644 index 000000000000..e47f1e4c9b03 --- /dev/null +++ b/include/linux/mlx5/driver.h @@ -0,0 +1,769 @@ +/* + * Copyright (c) 2013, Mellanox Technologies inc. All rights reserved. + * + * This software is available to you under a choice of one of two + * licenses. You may choose to be licensed under the terms of the GNU + * General Public License (GPL) Version 2, available from the file + * COPYING in the main directory of this source tree, or the + * OpenIB.org BSD license below: + * + * Redistribution and use in source and binary forms, with or + * without modification, are permitted provided that the following + * conditions are met: + * + * - Redistributions of source code must retain the above + * copyright notice, this list of conditions and the following + * disclaimer. + * + * - Redistributions in binary form must reproduce the above + * copyright notice, this list of conditions and the following + * disclaimer in the documentation and/or other materials + * provided with the distribution. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND + * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS + * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN + * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN + * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#ifndef MLX5_DRIVER_H +#define MLX5_DRIVER_H + +#include +#include +#include +#include +#include +#include +#include +#include +#include + +enum { + MLX5_BOARD_ID_LEN = 64, + MLX5_MAX_NAME_LEN = 16, +}; + +enum { + /* one minute for the sake of bringup. Generally, commands must always + * complete and we may need to increase this timeout value + */ + MLX5_CMD_TIMEOUT_MSEC = 7200 * 1000, + MLX5_CMD_WQ_MAX_NAME = 32, +}; + +enum { + CMD_OWNER_SW = 0x0, + CMD_OWNER_HW = 0x1, + CMD_STATUS_SUCCESS = 0, +}; + +enum mlx5_sqp_t { + MLX5_SQP_SMI = 0, + MLX5_SQP_GSI = 1, + MLX5_SQP_IEEE_1588 = 2, + MLX5_SQP_SNIFFER = 3, + MLX5_SQP_SYNC_UMR = 4, +}; + +enum { + MLX5_MAX_PORTS = 2, +}; + +enum { + MLX5_EQ_VEC_PAGES = 0, + MLX5_EQ_VEC_CMD = 1, + MLX5_EQ_VEC_ASYNC = 2, + MLX5_EQ_VEC_COMP_BASE, +}; + +enum { + MLX5_MAX_EQ_NAME = 20 +}; + +enum { + MLX5_ATOMIC_MODE_IB_COMP = 1 << 16, + MLX5_ATOMIC_MODE_CX = 2 << 16, + MLX5_ATOMIC_MODE_8B = 3 << 16, + MLX5_ATOMIC_MODE_16B = 4 << 16, + MLX5_ATOMIC_MODE_32B = 5 << 16, + MLX5_ATOMIC_MODE_64B = 6 << 16, + MLX5_ATOMIC_MODE_128B = 7 << 16, + MLX5_ATOMIC_MODE_256B = 8 << 16, +}; + +enum { + MLX5_CMD_OP_QUERY_HCA_CAP = 0x100, + MLX5_CMD_OP_QUERY_ADAPTER = 0x101, + MLX5_CMD_OP_INIT_HCA = 0x102, + MLX5_CMD_OP_TEARDOWN_HCA = 0x103, + MLX5_CMD_OP_QUERY_PAGES = 0x107, + MLX5_CMD_OP_MANAGE_PAGES = 0x108, + MLX5_CMD_OP_SET_HCA_CAP = 0x109, + + MLX5_CMD_OP_CREATE_MKEY = 0x200, + MLX5_CMD_OP_QUERY_MKEY = 0x201, + MLX5_CMD_OP_DESTROY_MKEY = 0x202, + MLX5_CMD_OP_QUERY_SPECIAL_CONTEXTS = 0x203, + + MLX5_CMD_OP_CREATE_EQ = 0x301, + MLX5_CMD_OP_DESTROY_EQ = 0x302, + MLX5_CMD_OP_QUERY_EQ = 0x303, + + MLX5_CMD_OP_CREATE_CQ = 0x400, + MLX5_CMD_OP_DESTROY_CQ = 0x401, + MLX5_CMD_OP_QUERY_CQ = 0x402, + MLX5_CMD_OP_MODIFY_CQ = 0x403, + + MLX5_CMD_OP_CREATE_QP = 0x500, + MLX5_CMD_OP_DESTROY_QP = 0x501, + MLX5_CMD_OP_RST2INIT_QP = 0x502, + MLX5_CMD_OP_INIT2RTR_QP = 0x503, + MLX5_CMD_OP_RTR2RTS_QP = 0x504, + MLX5_CMD_OP_RTS2RTS_QP = 0x505, + MLX5_CMD_OP_SQERR2RTS_QP = 0x506, + MLX5_CMD_OP_2ERR_QP = 0x507, + MLX5_CMD_OP_RTS2SQD_QP = 0x508, + MLX5_CMD_OP_SQD2RTS_QP = 0x509, + MLX5_CMD_OP_2RST_QP = 0x50a, + MLX5_CMD_OP_QUERY_QP = 0x50b, + MLX5_CMD_OP_CONF_SQP = 0x50c, + MLX5_CMD_OP_MAD_IFC = 0x50d, + MLX5_CMD_OP_INIT2INIT_QP = 0x50e, + MLX5_CMD_OP_SUSPEND_QP = 0x50f, + MLX5_CMD_OP_UNSUSPEND_QP = 0x510, + MLX5_CMD_OP_SQD2SQD_QP = 0x511, + MLX5_CMD_OP_ALLOC_QP_COUNTER_SET = 0x512, + MLX5_CMD_OP_DEALLOC_QP_COUNTER_SET = 0x513, + MLX5_CMD_OP_QUERY_QP_COUNTER_SET = 0x514, + + MLX5_CMD_OP_CREATE_PSV = 0x600, + MLX5_CMD_OP_DESTROY_PSV = 0x601, + MLX5_CMD_OP_QUERY_PSV = 0x602, + MLX5_CMD_OP_QUERY_SIG_RULE_TABLE = 0x603, + MLX5_CMD_OP_QUERY_BLOCK_SIZE_TABLE = 0x604, + + MLX5_CMD_OP_CREATE_SRQ = 0x700, + MLX5_CMD_OP_DESTROY_SRQ = 0x701, + MLX5_CMD_OP_QUERY_SRQ = 0x702, + MLX5_CMD_OP_ARM_RQ = 0x703, + MLX5_CMD_OP_RESIZE_SRQ = 0x704, + + MLX5_CMD_OP_ALLOC_PD = 0x800, + MLX5_CMD_OP_DEALLOC_PD = 0x801, + MLX5_CMD_OP_ALLOC_UAR = 0x802, + MLX5_CMD_OP_DEALLOC_UAR = 0x803, + + MLX5_CMD_OP_ATTACH_TO_MCG = 0x806, + MLX5_CMD_OP_DETACH_FROM_MCG = 0x807, + + + MLX5_CMD_OP_ALLOC_XRCD = 0x80e, + MLX5_CMD_OP_DEALLOC_XRCD = 0x80f, + + MLX5_CMD_OP_ACCESS_REG = 0x805, + MLX5_CMD_OP_MAX = 0x810, +}; + +enum { + MLX5_REG_PCAP = 0x5001, + MLX5_REG_PMTU = 0x5003, + MLX5_REG_PTYS = 0x5004, + MLX5_REG_PAOS = 0x5006, + MLX5_REG_PMAOS = 0x5012, + MLX5_REG_PUDE = 0x5009, + MLX5_REG_PMPE = 0x5010, + MLX5_REG_PELC = 0x500e, + MLX5_REG_PMLP = 0, /* TBD */ + MLX5_REG_NODE_DESC = 0x6001, + MLX5_REG_HOST_ENDIANNESS = 0x7004, +}; + +enum dbg_rsc_type { + MLX5_DBG_RSC_QP, + MLX5_DBG_RSC_EQ, + MLX5_DBG_RSC_CQ, +}; + +struct mlx5_field_desc { + struct dentry *dent; + int i; +}; + +struct mlx5_rsc_debug { + struct mlx5_core_dev *dev; + void *object; + enum dbg_rsc_type type; + struct dentry *root; + struct mlx5_field_desc fields[0]; +}; + +enum mlx5_dev_event { + MLX5_DEV_EVENT_SYS_ERROR, + MLX5_DEV_EVENT_PORT_UP, + MLX5_DEV_EVENT_PORT_DOWN, + MLX5_DEV_EVENT_PORT_INITIALIZED, + MLX5_DEV_EVENT_LID_CHANGE, + MLX5_DEV_EVENT_PKEY_CHANGE, + MLX5_DEV_EVENT_GUID_CHANGE, + MLX5_DEV_EVENT_CLIENT_REREG, +}; + +struct mlx5_uuar_info { + struct mlx5_uar *uars; + int num_uars; + int num_low_latency_uuars; + unsigned long *bitmap; + unsigned int *count; + struct mlx5_bf *bfs; + + /* + * protect uuar allocation data structs + */ + struct mutex lock; +}; + +struct mlx5_bf { + void __iomem *reg; + void __iomem *regreg; + int buf_size; + struct mlx5_uar *uar; + unsigned long offset; + int need_lock; + /* protect blue flame buffer selection when needed + */ + spinlock_t lock; + + /* serialize 64 bit writes when done as two 32 bit accesses + */ + spinlock_t lock32; + int uuarn; +}; + +struct mlx5_cmd_first { + __be32 data[4]; +}; + +struct mlx5_cmd_msg { + struct list_head list; + struct cache_ent *cache; + u32 len; + struct mlx5_cmd_first first; + struct mlx5_cmd_mailbox *next; +}; + +struct mlx5_cmd_debug { + struct dentry *dbg_root; + struct dentry *dbg_in; + struct dentry *dbg_out; + struct dentry *dbg_outlen; + struct dentry *dbg_status; + struct dentry *dbg_run; + void *in_msg; + void *out_msg; + u8 status; + u16 inlen; + u16 outlen; +}; + +struct cache_ent { + /* protect block chain allocations + */ + spinlock_t lock; + struct list_head head; +}; + +struct cmd_msg_cache { + struct cache_ent large; + struct cache_ent med; + +}; + +struct mlx5_cmd_stats { + u64 sum; + u64 n; + struct dentry *root; + struct dentry *avg; + struct dentry *count; + /* protect command average calculations */ + spinlock_t lock; +}; + +struct mlx5_cmd { + void *cmd_buf; + dma_addr_t dma; + u16 cmdif_rev; + u8 log_sz; + u8 log_stride; + int max_reg_cmds; + int events; + u32 __iomem *vector; + + /* protect command queue allocations + */ + spinlock_t alloc_lock; + + /* protect token allocations + */ + spinlock_t token_lock; + u8 token; + unsigned long bitmask; + char wq_name[MLX5_CMD_WQ_MAX_NAME]; + struct workqueue_struct *wq; + struct semaphore sem; + struct semaphore pages_sem; + int mode; + struct mlx5_cmd_work_ent *ent_arr[MLX5_MAX_COMMANDS]; + struct pci_pool *pool; + struct mlx5_cmd_debug dbg; + struct cmd_msg_cache cache; + int checksum_disabled; + struct mlx5_cmd_stats stats[MLX5_CMD_OP_MAX]; +}; + +struct mlx5_port_caps { + int gid_table_len; + int pkey_table_len; +}; + +struct mlx5_caps { + u8 log_max_eq; + u8 log_max_cq; + u8 log_max_qp; + u8 log_max_mkey; + u8 log_max_pd; + u8 log_max_srq; + u32 max_cqes; + int max_wqes; + int max_sq_desc_sz; + int max_rq_desc_sz; + u64 flags; + u16 stat_rate_support; + int log_max_msg; + int num_ports; + int max_ra_res_qp; + int max_ra_req_qp; + int max_srq_wqes; + int bf_reg_size; + int bf_regs_per_page; + struct mlx5_port_caps port[MLX5_MAX_PORTS]; + u8 ext_port_cap[MLX5_MAX_PORTS]; + int max_vf; + u32 reserved_lkey; + u8 local_ca_ack_delay; + u8 log_max_mcg; + u16 max_qp_mcg; + int min_page_sz; +}; + +struct mlx5_cmd_mailbox { + void *buf; + dma_addr_t dma; + struct mlx5_cmd_mailbox *next; +}; + +struct mlx5_buf_list { + void *buf; + dma_addr_t map; +}; + +struct mlx5_buf { + struct mlx5_buf_list direct; + struct mlx5_buf_list *page_list; + int nbufs; + int npages; + int page_shift; + int size; +}; + +struct mlx5_eq { + struct mlx5_core_dev *dev; + __be32 __iomem *doorbell; + u32 cons_index; + struct mlx5_buf buf; + int size; + u8 irqn; + u8 eqn; + int nent; + u64 mask; + char name[MLX5_MAX_EQ_NAME]; + struct list_head list; + int index; + struct mlx5_rsc_debug *dbg; +}; + + +struct mlx5_core_mr { + u64 iova; + u64 size; + u32 key; + u32 pd; + u32 access; +}; + +struct mlx5_core_srq { + u32 srqn; + int max; + int max_gs; + int max_avail_gather; + int wqe_shift; + void (*event) (struct mlx5_core_srq *, enum mlx5_event); + + atomic_t refcount; + struct completion free; +}; + +struct mlx5_eq_table { + void __iomem *update_ci; + void __iomem *update_arm_ci; + struct list_head *comp_eq_head; + struct mlx5_eq pages_eq; + struct mlx5_eq async_eq; + struct mlx5_eq cmd_eq; + struct msix_entry *msix_arr; + int num_comp_vectors; + /* protect EQs list + */ + spinlock_t lock; +}; + +struct mlx5_uar { + u32 index; + struct list_head bf_list; + unsigned free_bf_bmap; + void __iomem *wc_map; + void __iomem *map; +}; + + +struct mlx5_core_health { + struct health_buffer __iomem *health; + __be32 __iomem *health_counter; + struct timer_list timer; + struct list_head list; + u32 prev; + int miss_counter; +}; + +struct mlx5_cq_table { + /* protect radix tree + */ + spinlock_t lock; + struct radix_tree_root tree; +}; + +struct mlx5_qp_table { + /* protect radix tree + */ + spinlock_t lock; + struct radix_tree_root tree; +}; + +struct mlx5_srq_table { + /* protect radix tree + */ + spinlock_t lock; + struct radix_tree_root tree; +}; + +struct mlx5_priv { + char name[MLX5_MAX_NAME_LEN]; + struct mlx5_eq_table eq_table; + struct mlx5_uuar_info uuari; + MLX5_DECLARE_DOORBELL_LOCK(cq_uar_lock); + + /* pages stuff */ + struct workqueue_struct *pg_wq; + struct rb_root page_root; + int fw_pages; + int reg_pages; + + struct mlx5_core_health health; + + struct mlx5_srq_table srq_table; + + /* start: qp staff */ + struct mlx5_qp_table qp_table; + struct dentry *qp_debugfs; + struct dentry *eq_debugfs; + struct dentry *cq_debugfs; + struct dentry *cmdif_debugfs; + /* end: qp staff */ + + /* start: cq staff */ + struct mlx5_cq_table cq_table; + /* end: cq staff */ + + /* start: alloc staff */ + struct mutex pgdir_mutex; + struct list_head pgdir_list; + /* end: alloc staff */ + struct dentry *dbg_root; + + /* protect mkey key part */ + spinlock_t mkey_lock; + u8 mkey_key; +}; + +struct mlx5_core_dev { + struct pci_dev *pdev; + u8 rev_id; + char board_id[MLX5_BOARD_ID_LEN]; + struct mlx5_cmd cmd; + struct mlx5_caps caps; + phys_addr_t iseg_base; + struct mlx5_init_seg __iomem *iseg; + void (*event) (struct mlx5_core_dev *dev, + enum mlx5_dev_event event, + void *data); + struct mlx5_priv priv; + struct mlx5_profile *profile; + atomic_t num_qps; +}; + +struct mlx5_db { + __be32 *db; + union { + struct mlx5_db_pgdir *pgdir; + struct mlx5_ib_user_db_page *user_page; + } u; + dma_addr_t dma; + int index; +}; + +enum { + MLX5_DB_PER_PAGE = PAGE_SIZE / L1_CACHE_BYTES, +}; + +enum { + MLX5_COMP_EQ_SIZE = 1024, +}; + +struct mlx5_db_pgdir { + struct list_head list; + DECLARE_BITMAP(bitmap, MLX5_DB_PER_PAGE); + __be32 *db_page; + dma_addr_t db_dma; +}; + +typedef void (*mlx5_cmd_cbk_t)(int status, void *context); + +struct mlx5_cmd_work_ent { + struct mlx5_cmd_msg *in; + struct mlx5_cmd_msg *out; + mlx5_cmd_cbk_t callback; + void *context; + int idx; + struct completion done; + struct mlx5_cmd *cmd; + struct work_struct work; + struct mlx5_cmd_layout *lay; + int ret; + int page_queue; + u8 status; + u8 token; + struct timespec ts1; + struct timespec ts2; +}; + +struct mlx5_pas { + u64 pa; + u8 log_sz; +}; + +static inline void *mlx5_buf_offset(struct mlx5_buf *buf, int offset) +{ + if (likely(BITS_PER_LONG == 64 || buf->nbufs == 1)) + return buf->direct.buf + offset; + else + return buf->page_list[offset >> PAGE_SHIFT].buf + + (offset & (PAGE_SIZE - 1)); +} + +extern struct workqueue_struct *mlx5_core_wq; + +#define STRUCT_FIELD(header, field) \ + .struct_offset_bytes = offsetof(struct ib_unpacked_ ## header, field), \ + .struct_size_bytes = sizeof((struct ib_unpacked_ ## header *)0)->field + +struct ib_field { + size_t struct_offset_bytes; + size_t struct_size_bytes; + int offset_bits; + int size_bits; +}; + +static inline struct mlx5_core_dev *pci2mlx5_core_dev(struct pci_dev *pdev) +{ + return pci_get_drvdata(pdev); +} + +extern struct dentry *mlx5_debugfs_root; + +static inline u16 fw_rev_maj(struct mlx5_core_dev *dev) +{ + return ioread32be(&dev->iseg->fw_rev) & 0xffff; +} + +static inline u16 fw_rev_min(struct mlx5_core_dev *dev) +{ + return ioread32be(&dev->iseg->fw_rev) >> 16; +} + +static inline u16 fw_rev_sub(struct mlx5_core_dev *dev) +{ + return ioread32be(&dev->iseg->cmdif_rev_fw_sub) & 0xffff; +} + +static inline u16 cmdif_rev(struct mlx5_core_dev *dev) +{ + return ioread32be(&dev->iseg->cmdif_rev_fw_sub) >> 16; +} + +static inline void *mlx5_vzalloc(unsigned long size) +{ + void *rtn; + + rtn = kzalloc(size, GFP_KERNEL | __GFP_NOWARN); + if (!rtn) + rtn = vzalloc(size); + return rtn; +} + +static inline void mlx5_vfree(const void *addr) +{ + if (addr && is_vmalloc_addr(addr)) + vfree(addr); + else + kfree(addr); +} + +int mlx5_dev_init(struct mlx5_core_dev *dev, struct pci_dev *pdev); +void mlx5_dev_cleanup(struct mlx5_core_dev *dev); +int mlx5_cmd_init(struct mlx5_core_dev *dev); +void mlx5_cmd_cleanup(struct mlx5_core_dev *dev); +void mlx5_cmd_use_events(struct mlx5_core_dev *dev); +void mlx5_cmd_use_polling(struct mlx5_core_dev *dev); +int mlx5_cmd_status_to_err(struct mlx5_outbox_hdr *hdr); +int mlx5_cmd_exec(struct mlx5_core_dev *dev, void *in, int in_size, void *out, + int out_size); +int mlx5_cmd_alloc_uar(struct mlx5_core_dev *dev, u32 *uarn); +int mlx5_cmd_free_uar(struct mlx5_core_dev *dev, u32 uarn); +int mlx5_alloc_uuars(struct mlx5_core_dev *dev, struct mlx5_uuar_info *uuari); +int mlx5_free_uuars(struct mlx5_core_dev *dev, struct mlx5_uuar_info *uuari); +void mlx5_health_cleanup(void); +void __init mlx5_health_init(void); +void mlx5_start_health_poll(struct mlx5_core_dev *dev); +void mlx5_stop_health_poll(struct mlx5_core_dev *dev); +int mlx5_buf_alloc(struct mlx5_core_dev *dev, int size, int max_direct, + struct mlx5_buf *buf); +void mlx5_buf_free(struct mlx5_core_dev *dev, struct mlx5_buf *buf); +struct mlx5_cmd_mailbox *mlx5_alloc_cmd_mailbox_chain(struct mlx5_core_dev *dev, + gfp_t flags, int npages); +void mlx5_free_cmd_mailbox_chain(struct mlx5_core_dev *dev, + struct mlx5_cmd_mailbox *head); +int mlx5_core_create_srq(struct mlx5_core_dev *dev, struct mlx5_core_srq *srq, + struct mlx5_create_srq_mbox_in *in, int inlen); +int mlx5_core_destroy_srq(struct mlx5_core_dev *dev, struct mlx5_core_srq *srq); +int mlx5_core_query_srq(struct mlx5_core_dev *dev, struct mlx5_core_srq *srq, + struct mlx5_query_srq_mbox_out *out); +int mlx5_core_arm_srq(struct mlx5_core_dev *dev, struct mlx5_core_srq *srq, + u16 lwm, int is_srq); +int mlx5_core_create_mkey(struct mlx5_core_dev *dev, struct mlx5_core_mr *mr, + struct mlx5_create_mkey_mbox_in *in, int inlen); +int mlx5_core_destroy_mkey(struct mlx5_core_dev *dev, struct mlx5_core_mr *mr); +int mlx5_core_query_mkey(struct mlx5_core_dev *dev, struct mlx5_core_mr *mr, + struct mlx5_query_mkey_mbox_out *out, int outlen); +int mlx5_core_dump_fill_mkey(struct mlx5_core_dev *dev, struct mlx5_core_mr *mr, + u32 *mkey); +int mlx5_core_alloc_pd(struct mlx5_core_dev *dev, u32 *pdn); +int mlx5_core_dealloc_pd(struct mlx5_core_dev *dev, u32 pdn); +int mlx5_core_mad_ifc(struct mlx5_core_dev *dev, void *inb, void *outb, + u16 opmod, int port); +void mlx5_pagealloc_init(struct mlx5_core_dev *dev); +void mlx5_pagealloc_cleanup(struct mlx5_core_dev *dev); +int mlx5_pagealloc_start(struct mlx5_core_dev *dev); +void mlx5_pagealloc_stop(struct mlx5_core_dev *dev); +void mlx5_core_req_pages_handler(struct mlx5_core_dev *dev, u16 func_id, + s16 npages); +int mlx5_satisfy_startup_pages(struct mlx5_core_dev *dev); +int mlx5_reclaim_startup_pages(struct mlx5_core_dev *dev); +void mlx5_register_debugfs(void); +void mlx5_unregister_debugfs(void); +int mlx5_eq_init(struct mlx5_core_dev *dev); +void mlx5_eq_cleanup(struct mlx5_core_dev *dev); +void mlx5_fill_page_array(struct mlx5_buf *buf, __be64 *pas); +void mlx5_cq_completion(struct mlx5_core_dev *dev, u32 cqn); +void mlx5_qp_event(struct mlx5_core_dev *dev, u32 qpn, int event_type); +void mlx5_srq_event(struct mlx5_core_dev *dev, u32 srqn, int event_type); +struct mlx5_core_srq *mlx5_core_get_srq(struct mlx5_core_dev *dev, u32 srqn); +void mlx5_cmd_comp_handler(struct mlx5_core_dev *dev, unsigned long vector); +void mlx5_cq_event(struct mlx5_core_dev *dev, u32 cqn, int event_type); +int mlx5_create_map_eq(struct mlx5_core_dev *dev, struct mlx5_eq *eq, u8 vecidx, + int nent, u64 mask, const char *name, struct mlx5_uar *uar); +int mlx5_destroy_unmap_eq(struct mlx5_core_dev *dev, struct mlx5_eq *eq); +int mlx5_start_eqs(struct mlx5_core_dev *dev); +int mlx5_stop_eqs(struct mlx5_core_dev *dev); +int mlx5_core_attach_mcg(struct mlx5_core_dev *dev, union ib_gid *mgid, u32 qpn); +int mlx5_core_detach_mcg(struct mlx5_core_dev *dev, union ib_gid *mgid, u32 qpn); + +int mlx5_qp_debugfs_init(struct mlx5_core_dev *dev); +void mlx5_qp_debugfs_cleanup(struct mlx5_core_dev *dev); +int mlx5_core_access_reg(struct mlx5_core_dev *dev, void *data_in, + int size_in, void *data_out, int size_out, + u16 reg_num, int arg, int write); +int mlx5_set_port_caps(struct mlx5_core_dev *dev, int port_num, u32 caps); + +int mlx5_debug_eq_add(struct mlx5_core_dev *dev, struct mlx5_eq *eq); +void mlx5_debug_eq_remove(struct mlx5_core_dev *dev, struct mlx5_eq *eq); +int mlx5_core_eq_query(struct mlx5_core_dev *dev, struct mlx5_eq *eq, + struct mlx5_query_eq_mbox_out *out, int outlen); +int mlx5_eq_debugfs_init(struct mlx5_core_dev *dev); +void mlx5_eq_debugfs_cleanup(struct mlx5_core_dev *dev); +int mlx5_cq_debugfs_init(struct mlx5_core_dev *dev); +void mlx5_cq_debugfs_cleanup(struct mlx5_core_dev *dev); +int mlx5_db_alloc(struct mlx5_core_dev *dev, struct mlx5_db *db); +void mlx5_db_free(struct mlx5_core_dev *dev, struct mlx5_db *db); + +typedef void (*health_handler_t)(struct pci_dev *pdev, void *buf, int size); +int mlx5_register_health_report_handler(health_handler_t handler); +void mlx5_unregister_health_report_handler(void); +const char *mlx5_command_str(int command); +int mlx5_cmdif_debugfs_init(struct mlx5_core_dev *dev); +void mlx5_cmdif_debugfs_cleanup(struct mlx5_core_dev *dev); + +static inline u32 mlx5_mkey_to_idx(u32 mkey) +{ + return mkey >> 8; +} + +static inline u32 mlx5_idx_to_mkey(u32 mkey_idx) +{ + return mkey_idx << 8; +} + +enum { + MLX5_PROF_MASK_QP_SIZE = (u64)1 << 0, + MLX5_PROF_MASK_CMDIF_CSUM = (u64)1 << 1, + MLX5_PROF_MASK_MR_CACHE = (u64)1 << 2, +}; + +enum { + MAX_MR_CACHE_ENTRIES = 16, +}; + +struct mlx5_profile { + u64 mask; + u32 log_max_qp; + int cmdif_csum; + struct { + int size; + int limit; + } mr_cache[MAX_MR_CACHE_ENTRIES]; +}; + +#endif /* MLX5_DRIVER_H */ diff --git a/include/linux/mlx5/qp.h b/include/linux/mlx5/qp.h new file mode 100644 index 000000000000..d9e3eacb3a7f --- /dev/null +++ b/include/linux/mlx5/qp.h @@ -0,0 +1,467 @@ +/* + * Copyright (c) 2013, Mellanox Technologies inc. All rights reserved. + * + * This software is available to you under a choice of one of two + * licenses. You may choose to be licensed under the terms of the GNU + * General Public License (GPL) Version 2, available from the file + * COPYING in the main directory of this source tree, or the + * OpenIB.org BSD license below: + * + * Redistribution and use in source and binary forms, with or + * without modification, are permitted provided that the following + * conditions are met: + * + * - Redistributions of source code must retain the above + * copyright notice, this list of conditions and the following + * disclaimer. + * + * - Redistributions in binary form must reproduce the above + * copyright notice, this list of conditions and the following + * disclaimer in the documentation and/or other materials + * provided with the distribution. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND + * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS + * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN + * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN + * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#ifndef MLX5_QP_H +#define MLX5_QP_H + +#include +#include + +#define MLX5_INVALID_LKEY 0x100 + +enum mlx5_qp_optpar { + MLX5_QP_OPTPAR_ALT_ADDR_PATH = 1 << 0, + MLX5_QP_OPTPAR_RRE = 1 << 1, + MLX5_QP_OPTPAR_RAE = 1 << 2, + MLX5_QP_OPTPAR_RWE = 1 << 3, + MLX5_QP_OPTPAR_PKEY_INDEX = 1 << 4, + MLX5_QP_OPTPAR_Q_KEY = 1 << 5, + MLX5_QP_OPTPAR_RNR_TIMEOUT = 1 << 6, + MLX5_QP_OPTPAR_PRIMARY_ADDR_PATH = 1 << 7, + MLX5_QP_OPTPAR_SRA_MAX = 1 << 8, + MLX5_QP_OPTPAR_RRA_MAX = 1 << 9, + MLX5_QP_OPTPAR_PM_STATE = 1 << 10, + MLX5_QP_OPTPAR_RETRY_COUNT = 1 << 12, + MLX5_QP_OPTPAR_RNR_RETRY = 1 << 13, + MLX5_QP_OPTPAR_ACK_TIMEOUT = 1 << 14, + MLX5_QP_OPTPAR_PRI_PORT = 1 << 16, + MLX5_QP_OPTPAR_SRQN = 1 << 18, + MLX5_QP_OPTPAR_CQN_RCV = 1 << 19, + MLX5_QP_OPTPAR_DC_HS = 1 << 20, + MLX5_QP_OPTPAR_DC_KEY = 1 << 21, +}; + +enum mlx5_qp_state { + MLX5_QP_STATE_RST = 0, + MLX5_QP_STATE_INIT = 1, + MLX5_QP_STATE_RTR = 2, + MLX5_QP_STATE_RTS = 3, + MLX5_QP_STATE_SQER = 4, + MLX5_QP_STATE_SQD = 5, + MLX5_QP_STATE_ERR = 6, + MLX5_QP_STATE_SQ_DRAINING = 7, + MLX5_QP_STATE_SUSPENDED = 9, + MLX5_QP_NUM_STATE +}; + +enum { + MLX5_QP_ST_RC = 0x0, + MLX5_QP_ST_UC = 0x1, + MLX5_QP_ST_UD = 0x2, + MLX5_QP_ST_XRC = 0x3, + MLX5_QP_ST_MLX = 0x4, + MLX5_QP_ST_DCI = 0x5, + MLX5_QP_ST_DCT = 0x6, + MLX5_QP_ST_QP0 = 0x7, + MLX5_QP_ST_QP1 = 0x8, + MLX5_QP_ST_RAW_ETHERTYPE = 0x9, + MLX5_QP_ST_RAW_IPV6 = 0xa, + MLX5_QP_ST_SNIFFER = 0xb, + MLX5_QP_ST_SYNC_UMR = 0xe, + MLX5_QP_ST_PTP_1588 = 0xd, + MLX5_QP_ST_REG_UMR = 0xc, + MLX5_QP_ST_MAX +}; + +enum { + MLX5_QP_PM_MIGRATED = 0x3, + MLX5_QP_PM_ARMED = 0x0, + MLX5_QP_PM_REARM = 0x1 +}; + +enum { + MLX5_NON_ZERO_RQ = 0 << 24, + MLX5_SRQ_RQ = 1 << 24, + MLX5_CRQ_RQ = 2 << 24, + MLX5_ZERO_LEN_RQ = 3 << 24 +}; + +enum { + /* params1 */ + MLX5_QP_BIT_SRE = 1 << 15, + MLX5_QP_BIT_SWE = 1 << 14, + MLX5_QP_BIT_SAE = 1 << 13, + /* params2 */ + MLX5_QP_BIT_RRE = 1 << 15, + MLX5_QP_BIT_RWE = 1 << 14, + MLX5_QP_BIT_RAE = 1 << 13, + MLX5_QP_BIT_RIC = 1 << 4, +}; + +enum { + MLX5_WQE_CTRL_CQ_UPDATE = 2 << 2, + MLX5_WQE_CTRL_SOLICITED = 1 << 1, +}; + +enum { + MLX5_SEND_WQE_BB = 64, +}; + +enum { + MLX5_WQE_FMR_PERM_LOCAL_READ = 1 << 27, + MLX5_WQE_FMR_PERM_LOCAL_WRITE = 1 << 28, + MLX5_WQE_FMR_PERM_REMOTE_READ = 1 << 29, + MLX5_WQE_FMR_PERM_REMOTE_WRITE = 1 << 30, + MLX5_WQE_FMR_PERM_ATOMIC = 1 << 31 +}; + +enum { + MLX5_FENCE_MODE_NONE = 0 << 5, + MLX5_FENCE_MODE_INITIATOR_SMALL = 1 << 5, + MLX5_FENCE_MODE_STRONG_ORDERING = 3 << 5, + MLX5_FENCE_MODE_SMALL_AND_FENCE = 4 << 5, +}; + +enum { + MLX5_QP_LAT_SENSITIVE = 1 << 28, + MLX5_QP_ENABLE_SIG = 1 << 31, +}; + +enum { + MLX5_RCV_DBR = 0, + MLX5_SND_DBR = 1, +}; + +struct mlx5_wqe_fmr_seg { + __be32 flags; + __be32 mem_key; + __be64 buf_list; + __be64 start_addr; + __be64 reg_len; + __be32 offset; + __be32 page_size; + u32 reserved[2]; +}; + +struct mlx5_wqe_ctrl_seg { + __be32 opmod_idx_opcode; + __be32 qpn_ds; + u8 signature; + u8 rsvd[2]; + u8 fm_ce_se; + __be32 imm; +}; + +struct mlx5_wqe_xrc_seg { + __be32 xrc_srqn; + u8 rsvd[12]; +}; + +struct mlx5_wqe_masked_atomic_seg { + __be64 swap_add; + __be64 compare; + __be64 swap_add_mask; + __be64 compare_mask; +}; + +struct mlx5_av { + union { + struct { + __be32 qkey; + __be32 reserved; + } qkey; + __be64 dc_key; + } key; + __be32 dqp_dct; + u8 stat_rate_sl; + u8 fl_mlid; + __be16 rlid; + u8 reserved0[10]; + u8 tclass; + u8 hop_limit; + __be32 grh_gid_fl; + u8 rgid[16]; +}; + +struct mlx5_wqe_datagram_seg { + struct mlx5_av av; +}; + +struct mlx5_wqe_raddr_seg { + __be64 raddr; + __be32 rkey; + u32 reserved; +}; + +struct mlx5_wqe_atomic_seg { + __be64 swap_add; + __be64 compare; +}; + +struct mlx5_wqe_data_seg { + __be32 byte_count; + __be32 lkey; + __be64 addr; +}; + +struct mlx5_wqe_umr_ctrl_seg { + u8 flags; + u8 rsvd0[3]; + __be16 klm_octowords; + __be16 bsf_octowords; + __be64 mkey_mask; + u8 rsvd1[32]; +}; + +struct mlx5_seg_set_psv { + __be32 psv_num; + __be16 syndrome; + __be16 status; + __be32 transient_sig; + __be32 ref_tag; +}; + +struct mlx5_seg_get_psv { + u8 rsvd[19]; + u8 num_psv; + __be32 l_key; + __be64 va; + __be32 psv_index[4]; +}; + +struct mlx5_seg_check_psv { + u8 rsvd0[2]; + __be16 err_coalescing_op; + u8 rsvd1[2]; + __be16 xport_err_op; + u8 rsvd2[2]; + __be16 xport_err_mask; + u8 rsvd3[7]; + u8 num_psv; + __be32 l_key; + __be64 va; + __be32 psv_index[4]; +}; + +struct mlx5_rwqe_sig { + u8 rsvd0[4]; + u8 signature; + u8 rsvd1[11]; +}; + +struct mlx5_wqe_signature_seg { + u8 rsvd0[4]; + u8 signature; + u8 rsvd1[11]; +}; + +struct mlx5_wqe_inline_seg { + __be32 byte_count; +}; + +struct mlx5_core_qp { + void (*event) (struct mlx5_core_qp *, int); + int qpn; + atomic_t refcount; + struct completion free; + struct mlx5_rsc_debug *dbg; + int pid; +}; + +struct mlx5_qp_path { + u8 fl; + u8 rsvd3; + u8 free_ar; + u8 pkey_index; + u8 rsvd0; + u8 grh_mlid; + __be16 rlid; + u8 ackto_lt; + u8 mgid_index; + u8 static_rate; + u8 hop_limit; + __be32 tclass_flowlabel; + u8 rgid[16]; + u8 rsvd1[4]; + u8 sl; + u8 port; + u8 rsvd2[6]; +}; + +struct mlx5_qp_context { + __be32 flags; + __be32 flags_pd; + u8 mtu_msgmax; + u8 rq_size_stride; + __be16 sq_crq_size; + __be32 qp_counter_set_usr_page; + __be32 wire_qpn; + __be32 log_pg_sz_remote_qpn; + struct mlx5_qp_path pri_path; + struct mlx5_qp_path alt_path; + __be32 params1; + u8 reserved2[4]; + __be32 next_send_psn; + __be32 cqn_send; + u8 reserved3[8]; + __be32 last_acked_psn; + __be32 ssn; + __be32 params2; + __be32 rnr_nextrecvpsn; + __be32 xrcd; + __be32 cqn_recv; + __be64 db_rec_addr; + __be32 qkey; + __be32 rq_type_srqn; + __be32 rmsn; + __be16 hw_sq_wqe_counter; + __be16 sw_sq_wqe_counter; + __be16 hw_rcyclic_byte_counter; + __be16 hw_rq_counter; + __be16 sw_rcyclic_byte_counter; + __be16 sw_rq_counter; + u8 rsvd0[5]; + u8 cgs; + u8 cs_req; + u8 cs_res; + __be64 dc_access_key; + u8 rsvd1[24]; +}; + +struct mlx5_create_qp_mbox_in { + struct mlx5_inbox_hdr hdr; + __be32 input_qpn; + u8 rsvd0[4]; + __be32 opt_param_mask; + u8 rsvd1[4]; + struct mlx5_qp_context ctx; + u8 rsvd3[16]; + __be64 pas[0]; +}; + +struct mlx5_create_qp_mbox_out { + struct mlx5_outbox_hdr hdr; + __be32 qpn; + u8 rsvd0[4]; +}; + +struct mlx5_destroy_qp_mbox_in { + struct mlx5_inbox_hdr hdr; + __be32 qpn; + u8 rsvd0[4]; +}; + +struct mlx5_destroy_qp_mbox_out { + struct mlx5_outbox_hdr hdr; + u8 rsvd0[8]; +}; + +struct mlx5_modify_qp_mbox_in { + struct mlx5_inbox_hdr hdr; + __be32 qpn; + u8 rsvd1[4]; + __be32 optparam; + u8 rsvd0[4]; + struct mlx5_qp_context ctx; +}; + +struct mlx5_modify_qp_mbox_out { + struct mlx5_outbox_hdr hdr; + u8 rsvd0[8]; +}; + +struct mlx5_query_qp_mbox_in { + struct mlx5_inbox_hdr hdr; + __be32 qpn; + u8 rsvd[4]; +}; + +struct mlx5_query_qp_mbox_out { + struct mlx5_outbox_hdr hdr; + u8 rsvd1[8]; + __be32 optparam; + u8 rsvd0[4]; + struct mlx5_qp_context ctx; + u8 rsvd2[16]; + __be64 pas[0]; +}; + +struct mlx5_conf_sqp_mbox_in { + struct mlx5_inbox_hdr hdr; + __be32 qpn; + u8 rsvd[3]; + u8 type; +}; + +struct mlx5_conf_sqp_mbox_out { + struct mlx5_outbox_hdr hdr; + u8 rsvd[8]; +}; + +struct mlx5_alloc_xrcd_mbox_in { + struct mlx5_inbox_hdr hdr; + u8 rsvd[8]; +}; + +struct mlx5_alloc_xrcd_mbox_out { + struct mlx5_outbox_hdr hdr; + __be32 xrcdn; + u8 rsvd[4]; +}; + +struct mlx5_dealloc_xrcd_mbox_in { + struct mlx5_inbox_hdr hdr; + __be32 xrcdn; + u8 rsvd[4]; +}; + +struct mlx5_dealloc_xrcd_mbox_out { + struct mlx5_outbox_hdr hdr; + u8 rsvd[8]; +}; + +static inline struct mlx5_core_qp *__mlx5_qp_lookup(struct mlx5_core_dev *dev, u32 qpn) +{ + return radix_tree_lookup(&dev->priv.qp_table.tree, qpn); +} + +int mlx5_core_create_qp(struct mlx5_core_dev *dev, + struct mlx5_core_qp *qp, + struct mlx5_create_qp_mbox_in *in, + int inlen); +int mlx5_core_qp_modify(struct mlx5_core_dev *dev, enum mlx5_qp_state cur_state, + enum mlx5_qp_state new_state, + struct mlx5_modify_qp_mbox_in *in, int sqd_event, + struct mlx5_core_qp *qp); +int mlx5_core_destroy_qp(struct mlx5_core_dev *dev, + struct mlx5_core_qp *qp); +int mlx5_core_qp_query(struct mlx5_core_dev *dev, struct mlx5_core_qp *qp, + struct mlx5_query_qp_mbox_out *out, int outlen); + +int mlx5_core_xrcd_alloc(struct mlx5_core_dev *dev, u32 *xrcdn); +int mlx5_core_xrcd_dealloc(struct mlx5_core_dev *dev, u32 xrcdn); +void mlx5_init_qp_table(struct mlx5_core_dev *dev); +void mlx5_cleanup_qp_table(struct mlx5_core_dev *dev); +int mlx5_debug_qp_add(struct mlx5_core_dev *dev, struct mlx5_core_qp *qp); +void mlx5_debug_qp_remove(struct mlx5_core_dev *dev, struct mlx5_core_qp *qp); + +#endif /* MLX5_QP_H */ diff --git a/include/linux/mlx5/srq.h b/include/linux/mlx5/srq.h new file mode 100644 index 000000000000..e1a363a33663 --- /dev/null +++ b/include/linux/mlx5/srq.h @@ -0,0 +1,41 @@ +/* + * Copyright (c) 2013, Mellanox Technologies inc. All rights reserved. + * + * This software is available to you under a choice of one of two + * licenses. You may choose to be licensed under the terms of the GNU + * General Public License (GPL) Version 2, available from the file + * COPYING in the main directory of this source tree, or the + * OpenIB.org BSD license below: + * + * Redistribution and use in source and binary forms, with or + * without modification, are permitted provided that the following + * conditions are met: + * + * - Redistributions of source code must retain the above + * copyright notice, this list of conditions and the following + * disclaimer. + * + * - Redistributions in binary form must reproduce the above + * copyright notice, this list of conditions and the following + * disclaimer in the documentation and/or other materials + * provided with the distribution. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND + * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS + * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN + * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN + * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#ifndef MLX5_SRQ_H +#define MLX5_SRQ_H + +#include + +void mlx5_init_srq_table(struct mlx5_core_dev *dev); +void mlx5_cleanup_srq_table(struct mlx5_core_dev *dev); + +#endif /* MLX5_SRQ_H */ -- cgit v1.2.3 From 63884c90ffa3f73a81b81f169c51c34d2b9cf75e Mon Sep 17 00:00:00 2001 From: Roland Dreier Date: Mon, 1 Jul 2013 14:15:17 -0700 Subject: mlx5: Fix parameter type of health_handler_t This deals with the sparse warning: drivers/net/ethernet/mellanox/mlx5/core/health.c:94:54: warning: incorrect type in argument 2 (different address spaces) drivers/net/ethernet/mellanox/mlx5/core/health.c:94:54: expected void *buf drivers/net/ethernet/mellanox/mlx5/core/health.c:94:54: got struct health_buffer [noderef] *health Signed-off-by: Roland Dreier --- include/linux/mlx5/driver.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'include') diff --git a/include/linux/mlx5/driver.h b/include/linux/mlx5/driver.h index e47f1e4c9b03..f22e4419839b 100644 --- a/include/linux/mlx5/driver.h +++ b/include/linux/mlx5/driver.h @@ -729,7 +729,7 @@ void mlx5_cq_debugfs_cleanup(struct mlx5_core_dev *dev); int mlx5_db_alloc(struct mlx5_core_dev *dev, struct mlx5_db *db); void mlx5_db_free(struct mlx5_core_dev *dev, struct mlx5_db *db); -typedef void (*health_handler_t)(struct pci_dev *pdev, void *buf, int size); +typedef void (*health_handler_t)(struct pci_dev *pdev, struct health_buffer __iomem *buf, int size); int mlx5_register_health_report_handler(health_handler_t handler); void mlx5_unregister_health_report_handler(void); const char *mlx5_command_str(int command); -- cgit v1.2.3 From 288dde9f23b6726c1e8147bf635721372bf77b16 Mon Sep 17 00:00:00 2001 From: Moshe Lazer Date: Wed, 10 Jul 2013 14:31:03 +0300 Subject: mlx5_core: Adjust hca_cap.uar_page_sz to conform to Connect-IB spec Sparse reported an endianness bug in the assignment to hca_cap.uar_page_sz. Fix the declaration of this field to be __be16 (which is what is in the firmware spec), renaming the field to log_uar_pg_size to conform to the spec, which fixes the endianness bug reported by sparse. Reported-by: Fengguang Wu Signed-off-by: Moshe Lazer Signed-off-by: Or Gerlitz Signed-off-by: Roland Dreier --- include/linux/mlx5/device.h | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'include') diff --git a/include/linux/mlx5/device.h b/include/linux/mlx5/device.h index 51390915e538..8de8d8f22384 100644 --- a/include/linux/mlx5/device.h +++ b/include/linux/mlx5/device.h @@ -317,8 +317,8 @@ struct mlx5_hca_cap { u8 log_max_pd; u8 rsvd25; u8 log_max_xrcd; - u8 rsvd26[40]; - __be32 uar_page_sz; + u8 rsvd26[42]; + __be16 log_uar_page_sz; u8 rsvd27[28]; u8 log_msx_atomic_size_qp; u8 rsvd28[2]; -- cgit v1.2.3