From patchwork Fri Sep 28 22:53:50 2012 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: "Hefty, Sean" X-Patchwork-Id: 1527341 Return-Path: X-Original-To: patchwork-linux-rdma@patchwork.kernel.org Delivered-To: patchwork-process-083081@patchwork2.kernel.org Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by patchwork2.kernel.org (Postfix) with ESMTP id 7EB83DF283 for ; Fri, 28 Sep 2012 22:53:55 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1756000Ab2I1Wxy (ORCPT ); Fri, 28 Sep 2012 18:53:54 -0400 Received: from mga02.intel.com ([134.134.136.20]:39151 "EHLO mga02.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754427Ab2I1Wxw convert rfc822-to-8bit (ORCPT ); Fri, 28 Sep 2012 18:53:52 -0400 Received: from orsmga002.jf.intel.com ([10.7.209.21]) by orsmga101.jf.intel.com with ESMTP; 28 Sep 2012 15:53:51 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="4.80,504,1344236400"; d="scan'208";a="215764100" Received: from fmsmsx103.amr.corp.intel.com ([10.19.9.34]) by orsmga002.jf.intel.com with ESMTP; 28 Sep 2012 15:53:51 -0700 Received: from fmsmsx151.amr.corp.intel.com ([169.254.6.213]) by FMSMSX103.amr.corp.intel.com ([169.254.3.13]) with mapi id 14.01.0355.002; Fri, 28 Sep 2012 15:53:51 -0700 From: "Hefty, Sean" To: "linux-rdma (linux-rdma@vger.kernel.org)" , "miked@mellanox.com" , "Tzahi Oved (tzahio@mellanox.com)" , Jason Gunthorpe , "yishaih@mellanox.com" Subject: [PATCH v3 2/7] libibverbs: libibverbs: Introduce XRC domains Thread-Topic: [PATCH v3 2/7] libibverbs: libibverbs: Introduce XRC domains Thread-Index: Ac2dyl6LDrx4e+oSSN++zoIv4rl+sg== Date: Fri, 28 Sep 2012 22:53:50 +0000 Message-ID: <1828884A29C6694DAF28B7E6B8A8237346A981A6@FMSMSX151.amr.corp.intel.com> Accept-Language: en-US Content-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: x-originating-ip: [10.22.254.139] MIME-Version: 1.0 Sender: linux-rdma-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-rdma@vger.kernel.org XRC introduces several new concepts and structures, one of which is the XRC domain. XRC domains: xrcd's are a type of protection domain used to associate shared receive queues with xrc queue pairs. Since xrcd are meant to be shared among multiple processes, we introduce new APIs to open/close xrcd's. The user to kernel ABI is extended to account for opening/ closing the xrcd. Signed-off-by: Sean Hefty --- change from v2: Used structure to specify initial xrcd attributes in place of direct function parameters. include/infiniband/driver.h | 16 ++++++++++++ include/infiniband/kern-abi.h | 27 +++++++++++++++++++- include/infiniband/verbs.h | 55 +++++++++++++++++++++++++++++++++++++++-- src/cmd.c | 41 +++++++++++++++++++++++++++++++ src/libibverbs.map | 4 +++ 5 files changed, 139 insertions(+), 4 deletions(-) -- To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html diff --git a/include/infiniband/driver.h b/include/infiniband/driver.h index f22f287..ce88442 100644 --- a/include/infiniband/driver.h +++ b/include/infiniband/driver.h @@ -53,6 +53,17 @@ */ #define IBV_DEVICE_LIBRARY_EXTENSION rdmav2 +enum verbs_xrcd_mask { + VERBS_XRCD_HANDLE = 1 << 0, + VERBS_XRCD_RESERVED = 1 << 1 +}; + +struct verbs_xrcd { + struct ibv_xrcd xrcd; + uint64_t comp_mask; + uint32_t handle; +}; + typedef struct ibv_device *(*ibv_driver_init_func)(const char *uverbs_sys_path, int abi_version); typedef struct verbs_device *(*verbs_driver_init_func)(const char *uverbs_sys_path, @@ -78,6 +89,11 @@ int ibv_cmd_alloc_pd(struct ibv_context *context, struct ibv_pd *pd, struct ibv_alloc_pd *cmd, size_t cmd_size, struct ibv_alloc_pd_resp *resp, size_t resp_size); int ibv_cmd_dealloc_pd(struct ibv_pd *pd); +int ibv_cmd_open_xrcd(struct ibv_context *context, struct verbs_xrcd *xrcd, + struct ibv_xrcd_init_attr *attr, + struct ibv_open_xrcd *cmd, size_t cmd_size, + struct ibv_open_xrcd_resp *resp, size_t resp_size); +int ibv_cmd_close_xrcd(struct verbs_xrcd *xrcd); #define IBV_CMD_REG_MR_HAS_RESP_PARAMS int ibv_cmd_reg_mr(struct ibv_pd *pd, void *addr, size_t length, uint64_t hca_va, int access, diff --git a/include/infiniband/kern-abi.h b/include/infiniband/kern-abi.h index 619ea7e..d7c673f 100644 --- a/include/infiniband/kern-abi.h +++ b/include/infiniband/kern-abi.h @@ -85,7 +85,9 @@ enum { IB_USER_VERBS_CMD_MODIFY_SRQ, IB_USER_VERBS_CMD_QUERY_SRQ, IB_USER_VERBS_CMD_DESTROY_SRQ, - IB_USER_VERBS_CMD_POST_SRQ_RECV + IB_USER_VERBS_CMD_POST_SRQ_RECV, + IB_USER_VERBS_CMD_OPEN_XRCD, + IB_USER_VERBS_CMD_CLOSE_XRCD, }; /* @@ -246,6 +248,27 @@ struct ibv_dealloc_pd { __u32 pd_handle; }; +struct ibv_open_xrcd { + __u32 command; + __u16 in_words; + __u16 out_words; + __u64 response; + __u32 fd; + __u32 oflags; + __u64 driver_data[0]; +}; + +struct ibv_open_xrcd_resp { + __u32 xrcd_handle; +}; + +struct ibv_close_xrcd { + __u32 command; + __u16 in_words; + __u16 out_words; + __u32 xrcd_handle; +}; + struct ibv_reg_mr { __u32 command; __u16 in_words; @@ -804,6 +827,8 @@ enum { * trick opcodes in IBV_INIT_CMD() doesn't break. */ IB_USER_VERBS_CMD_CREATE_COMP_CHANNEL_V2 = -1, + IB_USER_VERBS_CMD_OPEN_XRCD_V2 = -1, + IB_USER_VERBS_CMD_CLOSE_XRCD_V2 = -1, }; struct ibv_destroy_cq_v1 { diff --git a/include/infiniband/verbs.h b/include/infiniband/verbs.h index b196b83..cf8299a 100644 --- a/include/infiniband/verbs.h +++ b/include/infiniband/verbs.h @@ -1,6 +1,6 @@ /* * Copyright (c) 2004, 2005 Topspin Communications. All rights reserved. - * Copyright (c) 2004 Intel Corporation. All rights reserved. + * Copyright (c) 2004, 2011-2012 Intel Corporation. All rights reserved. * Copyright (c) 2005, 2006, 2007 Cisco Systems, Inc. All rights reserved. * Copyright (c) 2005 PathScale, Inc. All rights reserved. * @@ -39,6 +39,7 @@ #include #include #include +#include #ifdef __cplusplus # define BEGIN_C_DECLS extern "C" { @@ -312,6 +313,22 @@ struct ibv_pd { uint32_t handle; }; +enum ibv_xrcd_init_attr_mask { + IBV_XRCD_INIT_ATTR_FD = 1 << 0, + IBV_XRCD_INIT_ATTR_OFLAGS = 1 << 1, + IBV_XRCD_INIT_ATTR_RESERVED = 1 << 2 +}; + +struct ibv_xrcd_init_attr { + uint32_t comp_mask; + int fd; + int oflags; +}; + +struct ibv_xrcd { + struct ibv_context *context; +}; + enum ibv_rereg_mr_flags { IBV_REREG_MR_CHANGE_TRANSLATION = (1 << 0), IBV_REREG_MR_CHANGE_PD = (1 << 1), @@ -728,11 +745,15 @@ struct ibv_context { }; enum verbs_context_mask { - VERBS_CONTEXT_RESERVED = 1 << 0 + VERBS_CONTEXT_XRCD = 1 << 0, + VERBS_CONTEXT_RESERVED = 1 << 1 }; struct verbs_context { /* "grows up" - new fields go here */ + struct ibv_xrcd * (*open_xrcd)(struct ibv_context *context, + struct ibv_xrcd_init_attr *xrcd_init_attr); + int (*close_xrcd)(struct ibv_xrcd *xrcd); uint64_t has_comp_mask; size_t sz; /* Must be immediately before struct ibv_context */ struct ibv_context context;/* Must be last field in the struct */ @@ -745,6 +766,11 @@ static inline struct verbs_context *verbs_get_ctx( NULL : container_of(ctx, struct verbs_context, context); } +#define verbs_get_ctx_op(ctx, op) ({ \ + struct verbs_context *vctx = verbs_get_ctx(ctx); \ + (!vctx || (vctx->sz < sizeof(*vctx) - offsetof(struct verbs_context, op)) || \ + !vctx->op) ? NULL : vctx; }) + static inline struct verbs_device *verbs_get_device( const struct ibv_device *dev) { @@ -830,7 +856,7 @@ static inline int ___ibv_query_port(struct ibv_context *context, uint8_t port_num, struct ibv_port_attr *port_attr) { - /* For compatability when running with old libibverbs */ + /* For compatibility when running with old libibverbs */ port_attr->link_layer = IBV_LINK_LAYER_UNSPECIFIED; port_attr->reserved = 0; @@ -863,6 +889,29 @@ struct ibv_pd *ibv_alloc_pd(struct ibv_context *context); int ibv_dealloc_pd(struct ibv_pd *pd); /** + * ibv_open_xrcd - Open an extended connection domain + */ +static inline struct ibv_xrcd * +ibv_open_xrcd(struct ibv_context *context, struct ibv_xrcd_init_attr *xrcd_init_attr) +{ + struct verbs_context *vctx = verbs_get_ctx_op(context, open_xrcd); + if (!vctx) { + errno = ENOSYS; + return NULL; + } + return vctx->open_xrcd(context, xrcd_init_attr); +} + +/** + * ibv_close_xrcd - Close an extended connection domain + */ +static inline int ibv_close_xrcd(struct ibv_xrcd *xrcd) +{ + struct verbs_context *vctx = verbs_get_ctx(xrcd->context); + return vctx->close_xrcd(xrcd); +} + +/** * ibv_reg_mr - Register a memory region */ struct ibv_mr *ibv_reg_mr(struct ibv_pd *pd, void *addr, diff --git a/src/cmd.c b/src/cmd.c index dab8930..f0fb1dc 100644 --- a/src/cmd.c +++ b/src/cmd.c @@ -194,6 +194,47 @@ int ibv_cmd_dealloc_pd(struct ibv_pd *pd) return 0; } +int ibv_cmd_open_xrcd(struct ibv_context *context, struct verbs_xrcd *xrcd, + struct ibv_xrcd_init_attr *attr, + struct ibv_open_xrcd *cmd, size_t cmd_size, + struct ibv_open_xrcd_resp *resp, size_t resp_size) +{ + IBV_INIT_CMD_RESP(cmd, cmd_size, OPEN_XRCD, resp, resp_size); + + if (attr->comp_mask >= IBV_XRCD_INIT_ATTR_RESERVED) + return ENOSYS; + + if (!(attr->comp_mask & IBV_XRCD_INIT_ATTR_FD) || + !(attr->comp_mask & IBV_XRCD_INIT_ATTR_OFLAGS)) + return EINVAL; + + cmd->fd = attr->fd; + cmd->oflags = attr->oflags; + if (write(context->cmd_fd, cmd, cmd_size) != cmd_size) + return errno; + + VALGRIND_MAKE_MEM_DEFINED(resp, resp_size); + + xrcd->xrcd.context = context; + xrcd->comp_mask = VERBS_XRCD_HANDLE; + xrcd->handle = resp->xrcd_handle; + + return 0; +} + +int ibv_cmd_close_xrcd(struct verbs_xrcd *xrcd) +{ + struct ibv_close_xrcd cmd; + + IBV_INIT_CMD(&cmd, sizeof cmd, CLOSE_XRCD); + cmd.xrcd_handle = xrcd->handle; + + if (write(xrcd->xrcd.context->cmd_fd, &cmd, sizeof cmd) != sizeof cmd) + return errno; + + return 0; +} + int ibv_cmd_reg_mr(struct ibv_pd *pd, void *addr, size_t length, uint64_t hca_va, int access, struct ibv_mr *mr, struct ibv_reg_mr *cmd, diff --git a/src/libibverbs.map b/src/libibverbs.map index ee9adea..9a15f3f 100644 --- a/src/libibverbs.map +++ b/src/libibverbs.map @@ -97,4 +97,8 @@ IBVERBS_1.1 { ibv_port_state_str; ibv_event_type_str; ibv_wc_status_str; + + ibv_cmd_open_xrcd; + ibv_cmd_close_xrcd; + } IBVERBS_1.0;