1
0
Fork 0
alistair23-linux/drivers/infiniband/hw/ehca/ehca_classes.h

449 lines
11 KiB
C
Raw Normal View History

/*
* IBM eServer eHCA Infiniband device driver for Linux on POWER
*
* Struct definition for eHCA internal structures
*
* Authors: Heiko J Schick <schickhj@de.ibm.com>
* Christoph Raisch <raisch@de.ibm.com>
* Joachim Fenkes <fenkes@de.ibm.com>
*
* Copyright (c) 2005 IBM Corporation
*
* All rights reserved.
*
* This source code is distributed under a dual license of GPL v2.0 and OpenIB
* BSD.
*
* OpenIB BSD License
*
* 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.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
* AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
* LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
* SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
* BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER
* IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
* ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
* POSSIBILITY OF SUCH DAMAGE.
*/
#ifndef __EHCA_CLASSES_H__
#define __EHCA_CLASSES_H__
struct ehca_module;
struct ehca_qp;
struct ehca_cq;
struct ehca_eq;
struct ehca_mr;
struct ehca_mw;
struct ehca_pd;
struct ehca_av;
#include <linux/wait.h>
#include <linux/mutex.h>
#include <rdma/ib_verbs.h>
#include <rdma/ib_user_verbs.h>
#ifdef CONFIG_PPC64
#include "ehca_classes_pSeries.h"
#endif
#include "ipz_pt_fn.h"
#include "ehca_qes.h"
#include "ehca_irq.h"
#define EHCA_EQE_CACHE_SIZE 20
#define EHCA_MAX_NUM_QUEUES 0xffff
struct ehca_eqe_cache_entry {
struct ehca_eqe *eqe;
struct ehca_cq *cq;
};
struct ehca_eq {
u32 length;
struct ipz_queue ipz_queue;
struct ipz_eq_handle ipz_eq_handle;
struct work_struct work;
struct h_galpas galpas;
int is_initialized;
struct ehca_pfeq pf;
spinlock_t spinlock;
struct tasklet_struct interrupt_task;
u32 ist;
spinlock_t irq_spinlock;
struct ehca_eqe_cache_entry eqe_cache[EHCA_EQE_CACHE_SIZE];
};
struct ehca_sma_attr {
u16 lid, lmc, sm_sl, sm_lid;
u16 pkey_tbl_len, pkeys[16];
};
struct ehca_sport {
struct ib_cq *ibcq_aqp1;
struct ib_qp *ibqp_sqp[2];
IB/ehca: Add "port connection autodetect mode" This patch enhances ehca with a capability to "autodetect" the ports being connected physically. In order to utilize that function the module option nr_ports must be set to -1 (default is 2 - two ports). This feature is experimental and will made the default later. More detail: If the user connects only one port to the switch, current code requires 1) port one to be connected and 2) module option nr_ports=1 to be given. If autodetect is enabled, ehca will not wait at creation of the GSI QP for the respective port to become active. Since firmware does not accept modify_qp() while the port is down at initialization, we need to cache all calls to modify_qp() for the SMI/GSI QP and just return a good return code. When a port is activated and we get a PORT_ACTIVE event, we replay the cached modify-qp() parms and re-trigger any posted recv WRs. Only then do we forward the PORT_ACTIVE event to registered clients. The result of this autodetect patch is that all ports will be accessible by the users. Depending on their respective cabling only those ports that are connected properly will become operable. If a user tries to modify a regular QP of a non-connected port, modify_qp() will fail. Furthermore, ibv_devinfo should show the port state accordingly. Note that this patch primarily improves the loading behaviour of ehca. If the cable is removed while the driver is operating and plugged in again, firmware will handle that properly by sending an appropriate async event. Signed-off-by: Hoang-Nam Nguyen <hnguyen@de.ibm.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2008-01-17 07:05:45 -07:00
/* lock to serialze modify_qp() calls for sqp in normal
* and irq path (when event PORT_ACTIVE is received first time)
*/
spinlock_t mod_sqp_lock;
enum ib_port_state port_state;
struct ehca_sma_attr saved_attr;
u32 pma_qp_nr;
};
#define HCA_CAP_MR_PGSIZE_4K 0x80000000
#define HCA_CAP_MR_PGSIZE_64K 0x40000000
#define HCA_CAP_MR_PGSIZE_1M 0x20000000
#define HCA_CAP_MR_PGSIZE_16M 0x10000000
struct ehca_shca {
struct ib_device ib_device;
struct of_device *ofdev;
u8 num_ports;
int hw_level;
struct list_head shca_list;
struct ipz_adapter_handle ipz_hca_handle;
struct ehca_sport sport[2];
struct ehca_eq eq;
struct ehca_eq neq;
struct ehca_mr *maxmr;
struct ehca_pd *pd;
struct h_galpas galpas;
struct mutex modify_mutex;
u64 hca_cap;
/* MR pgsize: bit 0-3 means 4K, 64K, 1M, 16M respectively */
u32 hca_cap_mr_pgsize;
int max_mtu;
atomic_t num_cqs;
atomic_t num_qps;
};
struct ehca_pd {
struct ib_pd ib_pd;
struct ipz_pd fw_pd;
/* small queue mgmt */
struct mutex lock;
struct list_head free[2];
struct list_head full[2];
};
enum ehca_ext_qp_type {
EQPT_NORMAL = 0,
EQPT_LLQP = 1,
EQPT_SRQBASE = 2,
EQPT_SRQ = 3,
};
IB/ehca: Add "port connection autodetect mode" This patch enhances ehca with a capability to "autodetect" the ports being connected physically. In order to utilize that function the module option nr_ports must be set to -1 (default is 2 - two ports). This feature is experimental and will made the default later. More detail: If the user connects only one port to the switch, current code requires 1) port one to be connected and 2) module option nr_ports=1 to be given. If autodetect is enabled, ehca will not wait at creation of the GSI QP for the respective port to become active. Since firmware does not accept modify_qp() while the port is down at initialization, we need to cache all calls to modify_qp() for the SMI/GSI QP and just return a good return code. When a port is activated and we get a PORT_ACTIVE event, we replay the cached modify-qp() parms and re-trigger any posted recv WRs. Only then do we forward the PORT_ACTIVE event to registered clients. The result of this autodetect patch is that all ports will be accessible by the users. Depending on their respective cabling only those ports that are connected properly will become operable. If a user tries to modify a regular QP of a non-connected port, modify_qp() will fail. Furthermore, ibv_devinfo should show the port state accordingly. Note that this patch primarily improves the loading behaviour of ehca. If the cable is removed while the driver is operating and plugged in again, firmware will handle that properly by sending an appropriate async event. Signed-off-by: Hoang-Nam Nguyen <hnguyen@de.ibm.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2008-01-17 07:05:45 -07:00
/* struct to cache modify_qp()'s parms for GSI/SMI qp */
struct ehca_mod_qp_parm {
int mask;
struct ib_qp_attr attr;
};
#define EHCA_MOD_QP_PARM_MAX 4
struct ehca_qp {
union {
struct ib_qp ib_qp;
struct ib_srq ib_srq;
};
u32 qp_type;
enum ehca_ext_qp_type ext_type;
enum ib_qp_state state;
struct ipz_queue ipz_squeue;
struct ipz_queue ipz_rqueue;
struct h_galpas galpas;
u32 qkey;
u32 real_qp_num;
u32 token;
spinlock_t spinlock_s;
spinlock_t spinlock_r;
u32 sq_max_inline_data_size;
struct ipz_qp_handle ipz_qp_handle;
struct ehca_pfqp pf;
struct ib_qp_init_attr init_attr;
struct ehca_cq *send_cq;
struct ehca_cq *recv_cq;
unsigned int sqerr_purgeflag;
struct hlist_node list_entries;
IB/ehca: Add "port connection autodetect mode" This patch enhances ehca with a capability to "autodetect" the ports being connected physically. In order to utilize that function the module option nr_ports must be set to -1 (default is 2 - two ports). This feature is experimental and will made the default later. More detail: If the user connects only one port to the switch, current code requires 1) port one to be connected and 2) module option nr_ports=1 to be given. If autodetect is enabled, ehca will not wait at creation of the GSI QP for the respective port to become active. Since firmware does not accept modify_qp() while the port is down at initialization, we need to cache all calls to modify_qp() for the SMI/GSI QP and just return a good return code. When a port is activated and we get a PORT_ACTIVE event, we replay the cached modify-qp() parms and re-trigger any posted recv WRs. Only then do we forward the PORT_ACTIVE event to registered clients. The result of this autodetect patch is that all ports will be accessible by the users. Depending on their respective cabling only those ports that are connected properly will become operable. If a user tries to modify a regular QP of a non-connected port, modify_qp() will fail. Furthermore, ibv_devinfo should show the port state accordingly. Note that this patch primarily improves the loading behaviour of ehca. If the cable is removed while the driver is operating and plugged in again, firmware will handle that properly by sending an appropriate async event. Signed-off-by: Hoang-Nam Nguyen <hnguyen@de.ibm.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2008-01-17 07:05:45 -07:00
/* array to cache modify_qp()'s parms for GSI/SMI qp */
struct ehca_mod_qp_parm *mod_qp_parm;
int mod_qp_parm_idx;
/* mmap counter for resources mapped into user space */
u32 mm_count_squeue;
u32 mm_count_rqueue;
u32 mm_count_galpa;
/* unsolicited ack circumvention */
int unsol_ack_circ;
int mtu_shift;
u32 message_count;
u32 packet_count;
atomic_t nr_events; /* events seen */
wait_queue_head_t wait_completion;
};
#define IS_SRQ(qp) (qp->ext_type == EQPT_SRQ)
#define HAS_SQ(qp) (qp->ext_type != EQPT_SRQ)
#define HAS_RQ(qp) (qp->ext_type != EQPT_SRQBASE)
/* must be power of 2 */
#define QP_HASHTAB_LEN 8
struct ehca_cq {
struct ib_cq ib_cq;
struct ipz_queue ipz_queue;
struct h_galpas galpas;
spinlock_t spinlock;
u32 cq_number;
u32 token;
u32 nr_of_entries;
struct ipz_cq_handle ipz_cq_handle;
struct ehca_pfcq pf;
spinlock_t cb_lock;
struct hlist_head qp_hashtab[QP_HASHTAB_LEN];
struct list_head entry;
u32 nr_callbacks; /* #events assigned to cpu by scaling code */
atomic_t nr_events; /* #events seen */
wait_queue_head_t wait_completion;
spinlock_t task_lock;
/* mmap counter for resources mapped into user space */
u32 mm_count_queue;
u32 mm_count_galpa;
};
enum ehca_mr_flag {
EHCA_MR_FLAG_FMR = 0x80000000, /* FMR, created with ehca_alloc_fmr */
EHCA_MR_FLAG_MAXMR = 0x40000000, /* max-MR */
};
struct ehca_mr {
union {
struct ib_mr ib_mr; /* must always be first in ehca_mr */
struct ib_fmr ib_fmr; /* must always be first in ehca_mr */
} ib;
IB/uverbs: Export ib_umem_get()/ib_umem_release() to modules Export ib_umem_get()/ib_umem_release() and put low-level drivers in control of when to call ib_umem_get() to pin and DMA map userspace, rather than always calling it in ib_uverbs_reg_mr() before calling the low-level driver's reg_user_mr method. Also move these functions to be in the ib_core module instead of ib_uverbs, so that driver modules using them do not depend on ib_uverbs. This has a number of advantages: - It is better design from the standpoint of making generic code a library that can be used or overridden by device-specific code as the details of specific devices dictate. - Drivers that do not need to pin userspace memory regions do not need to take the performance hit of calling ib_mem_get(). For example, although I have not tried to implement it in this patch, the ipath driver should be able to avoid pinning memory and just use copy_{to,from}_user() to access userspace memory regions. - Buffers that need special mapping treatment can be identified by the low-level driver. For example, it may be possible to solve some Altix-specific memory ordering issues with mthca CQs in userspace by mapping CQ buffers with extra flags. - Drivers that need to pin and DMA map userspace memory for things other than memory regions can use ib_umem_get() directly, instead of hacks using extra parameters to their reg_phys_mr method. For example, the mlx4 driver that is pending being merged needs to pin and DMA map QP and CQ buffers, but it does not need to create a memory key for these buffers. So the cleanest solution is for mlx4 to call ib_umem_get() in the create_qp and create_cq methods. Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-03-04 17:15:11 -07:00
struct ib_umem *umem;
spinlock_t mrlock;
enum ehca_mr_flag flags;
u32 num_kpages; /* number of kernel pages */
u32 num_hwpages; /* number of hw pages to form MR */
u64 hwpage_size; /* hw page size used for this MR */
int acl; /* ACL (stored here for usage in reregister) */
u64 *start; /* virtual start address (stored here for */
/* usage in reregister) */
u64 size; /* size (stored here for usage in reregister) */
u32 fmr_page_size; /* page size for FMR */
u32 fmr_max_pages; /* max pages for FMR */
u32 fmr_max_maps; /* max outstanding maps for FMR */
u32 fmr_map_cnt; /* map counter for FMR */
/* fw specific data */
struct ipz_mrmw_handle ipz_mr_handle; /* MR handle for h-calls */
struct h_galpas galpas;
};
struct ehca_mw {
struct ib_mw ib_mw; /* gen2 mw, must always be first in ehca_mw */
spinlock_t mwlock;
u8 never_bound; /* indication MW was never bound */
struct ipz_mrmw_handle ipz_mw_handle; /* MW handle for h-calls */
struct h_galpas galpas;
};
enum ehca_mr_pgi_type {
EHCA_MR_PGI_PHYS = 1, /* type of ehca_reg_phys_mr,
* ehca_rereg_phys_mr,
* ehca_reg_internal_maxmr */
EHCA_MR_PGI_USER = 2, /* type of ehca_reg_user_mr */
EHCA_MR_PGI_FMR = 3 /* type of ehca_map_phys_fmr */
};
struct ehca_mr_pginfo {
enum ehca_mr_pgi_type type;
u64 num_kpages;
u64 kpage_cnt;
u64 hwpage_size; /* hw page size used for this MR */
u64 num_hwpages; /* number of hw pages */
u64 hwpage_cnt; /* counter for hw pages */
u64 next_hwpage; /* next hw page in buffer/chunk/listelem */
union {
struct { /* type EHCA_MR_PGI_PHYS section */
int num_phys_buf;
struct ib_phys_buf *phys_buf_array;
u64 next_buf;
} phy;
struct { /* type EHCA_MR_PGI_USER section */
struct ib_umem *region;
struct ib_umem_chunk *next_chunk;
u64 next_nmap;
} usr;
struct { /* type EHCA_MR_PGI_FMR section */
u64 fmr_pgsize;
u64 *page_list;
u64 next_listelem;
} fmr;
} u;
};
/* output parameters for MR/FMR hipz calls */
struct ehca_mr_hipzout_parms {
struct ipz_mrmw_handle handle;
u32 lkey;
u32 rkey;
u64 len;
u64 vaddr;
u32 acl;
};
/* output parameters for MW hipz calls */
struct ehca_mw_hipzout_parms {
struct ipz_mrmw_handle handle;
u32 rkey;
};
struct ehca_av {
struct ib_ah ib_ah;
struct ehca_ud_av av;
};
struct ehca_ucontext {
struct ib_ucontext ib_ucontext;
};
int ehca_init_pd_cache(void);
void ehca_cleanup_pd_cache(void);
int ehca_init_cq_cache(void);
void ehca_cleanup_cq_cache(void);
int ehca_init_qp_cache(void);
void ehca_cleanup_qp_cache(void);
int ehca_init_av_cache(void);
void ehca_cleanup_av_cache(void);
int ehca_init_mrmw_cache(void);
void ehca_cleanup_mrmw_cache(void);
int ehca_init_small_qp_cache(void);
void ehca_cleanup_small_qp_cache(void);
extern rwlock_t ehca_qp_idr_lock;
extern rwlock_t ehca_cq_idr_lock;
extern struct idr ehca_qp_idr;
extern struct idr ehca_cq_idr;
extern int ehca_static_rate;
extern int ehca_port_act_time;
extern int ehca_use_hp_mr;
extern int ehca_scaling_code;
extern int ehca_lock_hcalls;
IB/ehca: Add "port connection autodetect mode" This patch enhances ehca with a capability to "autodetect" the ports being connected physically. In order to utilize that function the module option nr_ports must be set to -1 (default is 2 - two ports). This feature is experimental and will made the default later. More detail: If the user connects only one port to the switch, current code requires 1) port one to be connected and 2) module option nr_ports=1 to be given. If autodetect is enabled, ehca will not wait at creation of the GSI QP for the respective port to become active. Since firmware does not accept modify_qp() while the port is down at initialization, we need to cache all calls to modify_qp() for the SMI/GSI QP and just return a good return code. When a port is activated and we get a PORT_ACTIVE event, we replay the cached modify-qp() parms and re-trigger any posted recv WRs. Only then do we forward the PORT_ACTIVE event to registered clients. The result of this autodetect patch is that all ports will be accessible by the users. Depending on their respective cabling only those ports that are connected properly will become operable. If a user tries to modify a regular QP of a non-connected port, modify_qp() will fail. Furthermore, ibv_devinfo should show the port state accordingly. Note that this patch primarily improves the loading behaviour of ehca. If the cable is removed while the driver is operating and plugged in again, firmware will handle that properly by sending an appropriate async event. Signed-off-by: Hoang-Nam Nguyen <hnguyen@de.ibm.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2008-01-17 07:05:45 -07:00
extern int ehca_nr_ports;
extern int ehca_max_cq;
extern int ehca_max_qp;
struct ipzu_queue_resp {
u32 qe_size; /* queue entry size */
u32 act_nr_of_sg;
u32 queue_length; /* queue length allocated in bytes */
u32 pagesize;
u32 toggle_state;
u32 offset; /* save offset within a page for small_qp */
};
struct ehca_create_cq_resp {
u32 cq_number;
u32 token;
struct ipzu_queue_resp ipz_queue;
u32 fw_handle_ofs;
u32 dummy;
};
struct ehca_create_qp_resp {
u32 qp_num;
u32 token;
u32 qp_type;
u32 ext_type;
u32 qkey;
/* qp_num assigned by ehca: sqp0/1 may have got different numbers */
u32 real_qp_num;
u32 fw_handle_ofs;
u32 dummy;
struct ipzu_queue_resp ipz_squeue;
struct ipzu_queue_resp ipz_rqueue;
};
struct ehca_alloc_cq_parms {
u32 nr_cqe;
u32 act_nr_of_entries;
u32 act_pages;
struct ipz_eq_handle eq_handle;
};
enum ehca_service_type {
ST_RC = 0,
ST_UC = 1,
ST_RD = 2,
ST_UD = 3,
};
enum ehca_ll_comp_flags {
LLQP_SEND_COMP = 0x20,
LLQP_RECV_COMP = 0x40,
LLQP_COMP_MASK = 0x60,
};
struct ehca_alloc_queue_parms {
/* input parameters */
int max_wr;
int max_sge;
int page_size;
int is_small;
/* output parameters */
u16 act_nr_wqes;
u8 act_nr_sges;
u32 queue_size; /* bytes for small queues, pages otherwise */
};
struct ehca_alloc_qp_parms {
struct ehca_alloc_queue_parms squeue;
struct ehca_alloc_queue_parms rqueue;
/* input parameters */
enum ehca_service_type servicetype;
int qp_storage;
int sigtype;
enum ehca_ext_qp_type ext_type;
enum ehca_ll_comp_flags ll_comp_flags;
int ud_av_l_key_ctl;
u32 token;
struct ipz_eq_handle eq_handle;
struct ipz_pd pd;
struct ipz_cq_handle send_cq_handle, recv_cq_handle;
u32 srq_qpn, srq_token, srq_limit;
/* output parameters */
u32 real_qp_num;
struct ipz_qp_handle qp_handle;
struct h_galpas galpas;
};
int ehca_cq_assign_qp(struct ehca_cq *cq, struct ehca_qp *qp);
int ehca_cq_unassign_qp(struct ehca_cq *cq, unsigned int qp_num);
struct ehca_qp *ehca_cq_get_qp(struct ehca_cq *cq, int qp_num);
#endif