UbiquitousOS
/
OpenCloudOS-Kernel
mirror of https://gitee.com/OpenCloudOS/OpenCloudOS-Kernel.git
11
0
Fork 0
Code Issues Projects Releases Wiki Activity
OpenCloudOS-Kernel/drivers/infiniband/hw/mthca/mthca_eq.c

919 lines
25 KiB
C
Raw Normal View History

Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
/*
* Copyright (c) 2004, 2005 Topspin Communications. All rights reserved.
[PATCH] IB: Add copyright notices Make some lawyers happy and add copyright notices for people who forgot to include them when they actually touched the code. Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-08-11 14:03:10 +08:00
* Copyright (c) 2005 Mellanox Technologies. All rights reserved.
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
*
* 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.
*/
#include <linux/errno.h>
#include <linux/interrupt.h>
#include <linux/pci.h>
#include "mthca_dev.h"
#include "mthca_cmd.h"
#include "mthca_config_reg.h"
enum {
MTHCA_NUM_ASYNC_EQE = 0x80,
MTHCA_NUM_CMD_EQE = 0x80,
IB/mthca: prevent event queue overrun I am seeing EQ overruns in SDP stress tests: if the CQ completion handler arms a CQ, this could generate more EQEs, so that EQ will never get empty and consumer index will never get updated. This is similiar to what we have with command interface: /* * cmd_event() may add more commands. * The card will think the queue has overflowed if * we don't tell it we've been processing events. */ However, for completion events, we *don't* want to update the consumer index on each event. So, perform EQ doorbell coalescing: allocate EQs with some spare EQEs, and update once we run out of them. The value 0x80 was selected to avoid any performance impact. Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-01-10 06:04:40 +08:00
MTHCA_NUM_SPARE_EQE = 0x80,
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
MTHCA_EQ_ENTRY_SIZE = 0x20
};
/*
* Must be packed because start is 64 bits but only aligned to 32 bits.
*/
struct mthca_eq_context {
[PATCH] IB: sparse endianness cleanup Fix sparse warnings. Use __be* where appropriate. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-08-14 12:05:57 +08:00
__be32 flags;
__be64 start;
__be32 logsize_usrpage;
__be32 tavor_pd; /* reserved for Arbel */
u8 reserved1[3];
u8 intr;
__be32 arbel_pd; /* lost_count for Tavor */
__be32 lkey;
u32 reserved2[2];
__be32 consumer_index;
__be32 producer_index;
u32 reserved3[4];
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
} __attribute__((packed));
#define MTHCA_EQ_STATUS_OK ( 0 << 28)
#define MTHCA_EQ_STATUS_OVERFLOW ( 9 << 28)
#define MTHCA_EQ_STATUS_WRITE_FAIL (10 << 28)
#define MTHCA_EQ_OWNER_SW ( 0 << 24)
#define MTHCA_EQ_OWNER_HW ( 1 << 24)
#define MTHCA_EQ_FLAG_TR ( 1 << 18)
#define MTHCA_EQ_FLAG_OI ( 1 << 17)
#define MTHCA_EQ_STATE_ARMED ( 1 << 8)
#define MTHCA_EQ_STATE_FIRED ( 2 << 8)
#define MTHCA_EQ_STATE_ALWAYS_ARMED ( 3 << 8)
#define MTHCA_EQ_STATE_ARBEL ( 8 << 8)
enum {
MTHCA_EVENT_TYPE_COMP = 0x00,
MTHCA_EVENT_TYPE_PATH_MIG = 0x01,
MTHCA_EVENT_TYPE_COMM_EST = 0x02,
MTHCA_EVENT_TYPE_SQ_DRAINED = 0x03,
[IB] mthca: SRQ limit reached events Our hardware supports generating an event when the number of receives posted to a shared receive queue (SRQ) falls below a user-specified limit. Implement mthca_modify_srq() to arm the limit, and add code to handle dispatching SRQ events when they occur. Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-07 04:15:56 +08:00
MTHCA_EVENT_TYPE_SRQ_QP_LAST_WQE = 0x13,
MTHCA_EVENT_TYPE_SRQ_LIMIT = 0x14,
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
MTHCA_EVENT_TYPE_CQ_ERROR = 0x04,
MTHCA_EVENT_TYPE_WQ_CATAS_ERROR = 0x05,
MTHCA_EVENT_TYPE_EEC_CATAS_ERROR = 0x06,
MTHCA_EVENT_TYPE_PATH_MIG_FAILED = 0x07,
MTHCA_EVENT_TYPE_WQ_INVAL_REQ_ERROR = 0x10,
MTHCA_EVENT_TYPE_WQ_ACCESS_ERROR = 0x11,
MTHCA_EVENT_TYPE_SRQ_CATAS_ERROR = 0x12,
MTHCA_EVENT_TYPE_LOCAL_CATAS_ERROR = 0x08,
MTHCA_EVENT_TYPE_PORT_CHANGE = 0x09,
MTHCA_EVENT_TYPE_EQ_OVERFLOW = 0x0f,
MTHCA_EVENT_TYPE_ECC_DETECT = 0x0e,
MTHCA_EVENT_TYPE_CMD = 0x0a
};
#define MTHCA_ASYNC_EVENT_MASK ((1ULL << MTHCA_EVENT_TYPE_PATH_MIG) | \
(1ULL << MTHCA_EVENT_TYPE_COMM_EST) | \
(1ULL << MTHCA_EVENT_TYPE_SQ_DRAINED) | \
(1ULL << MTHCA_EVENT_TYPE_CQ_ERROR) | \
(1ULL << MTHCA_EVENT_TYPE_WQ_CATAS_ERROR) | \
(1ULL << MTHCA_EVENT_TYPE_EEC_CATAS_ERROR) | \
(1ULL << MTHCA_EVENT_TYPE_PATH_MIG_FAILED) | \
(1ULL << MTHCA_EVENT_TYPE_WQ_INVAL_REQ_ERROR) | \
(1ULL << MTHCA_EVENT_TYPE_WQ_ACCESS_ERROR) | \
(1ULL << MTHCA_EVENT_TYPE_LOCAL_CATAS_ERROR) | \
(1ULL << MTHCA_EVENT_TYPE_PORT_CHANGE) | \
(1ULL << MTHCA_EVENT_TYPE_ECC_DETECT))
[IB] mthca: SRQ limit reached events Our hardware supports generating an event when the number of receives posted to a shared receive queue (SRQ) falls below a user-specified limit. Implement mthca_modify_srq() to arm the limit, and add code to handle dispatching SRQ events when they occur. Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-07 04:15:56 +08:00
#define MTHCA_SRQ_EVENT_MASK ((1ULL << MTHCA_EVENT_TYPE_SRQ_CATAS_ERROR) | \
(1ULL << MTHCA_EVENT_TYPE_SRQ_QP_LAST_WQE) | \
(1ULL << MTHCA_EVENT_TYPE_SRQ_LIMIT))
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
#define MTHCA_CMD_EVENT_MASK (1ULL << MTHCA_EVENT_TYPE_CMD)
#define MTHCA_EQ_DB_INC_CI (1 << 24)
#define MTHCA_EQ_DB_REQ_NOT (2 << 24)
#define MTHCA_EQ_DB_DISARM_CQ (3 << 24)
#define MTHCA_EQ_DB_SET_CI (4 << 24)
#define MTHCA_EQ_DB_ALWAYS_ARM (5 << 24)
struct mthca_eqe {
u8 reserved1;
u8 type;
u8 reserved2;
u8 subtype;
union {
u32 raw[6];
struct {
[PATCH] IB: sparse endianness cleanup Fix sparse warnings. Use __be* where appropriate. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-08-14 12:05:57 +08:00
__be32 cqn;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
} __attribute__((packed)) comp;
struct {
[PATCH] IB: sparse endianness cleanup Fix sparse warnings. Use __be* where appropriate. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-08-14 12:05:57 +08:00
u16 reserved1;
__be16 token;
u32 reserved2;
u8 reserved3[3];
u8 status;
__be64 out_param;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
} __attribute__((packed)) cmd;
struct {
[PATCH] IB: sparse endianness cleanup Fix sparse warnings. Use __be* where appropriate. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-08-14 12:05:57 +08:00
__be32 qpn;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
} __attribute__((packed)) qp;
[IB] mthca: SRQ limit reached events Our hardware supports generating an event when the number of receives posted to a shared receive queue (SRQ) falls below a user-specified limit. Implement mthca_modify_srq() to arm the limit, and add code to handle dispatching SRQ events when they occur. Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-07 04:15:56 +08:00
struct {
__be32 srqn;
} __attribute__((packed)) srq;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
struct {
[PATCH] IB: sparse endianness cleanup Fix sparse warnings. Use __be* where appropriate. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-08-14 12:05:57 +08:00
__be32 cqn;
u32 reserved1;
u8 reserved2[3];
u8 syndrome;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
} __attribute__((packed)) cq_err;
struct {
[PATCH] IB: sparse endianness cleanup Fix sparse warnings. Use __be* where appropriate. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-08-14 12:05:57 +08:00
u32 reserved1[2];
__be32 port;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
} __attribute__((packed)) port_change;
} event;
u8 reserved3[3];
u8 owner;
} __attribute__((packed));
#define MTHCA_EQ_ENTRY_OWNER_SW (0 << 7)
#define MTHCA_EQ_ENTRY_OWNER_HW (1 << 7)
static inline u64 async_mask(struct mthca_dev *dev)
{
return dev->mthca_flags & MTHCA_FLAG_SRQ ?
MTHCA_ASYNC_EVENT_MASK | MTHCA_SRQ_EVENT_MASK :
MTHCA_ASYNC_EVENT_MASK;
}
static inline void tavor_set_eq_ci(struct mthca_dev *dev, struct mthca_eq *eq, u32 ci)
{
/*
* This barrier makes sure that all updates to ownership bits
* done by set_eqe_hw() hit memory before the consumer index
* is updated. set_eq_ci() allows the HCA to possibly write
* more EQ entries, and we want to avoid the exceedingly
* unlikely possibility of the HCA writing an entry and then
* having set_eqe_hw() overwrite the owner field.
*/
wmb();
IB/mthca: Avoid alignment traps when writing doorbells Architectures such as ia64 see alignment traps when doing a 64-bit read from __be32 doorbell[2] arrays to do doorbell writes in mthca_write64(). Fix this by just passing the two halves of the doorbell value into mthca_write64(). This actually improves the generated code by allowing the compiler to see what's going on better. Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-10-15 11:40:27 +08:00
mthca_write64(MTHCA_EQ_DB_SET_CI | eq->eqn, ci & (eq->nent - 1),
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
dev->kar + MTHCA_EQ_DOORBELL,
MTHCA_GET_DOORBELL_LOCK(&dev->doorbell_lock));
}
static inline void arbel_set_eq_ci(struct mthca_dev *dev, struct mthca_eq *eq, u32 ci)
{
/* See comment in tavor_set_eq_ci() above. */
wmb();
[PATCH] IB: sparse endianness cleanup Fix sparse warnings. Use __be* where appropriate. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-08-14 12:05:57 +08:00
__raw_writel((__force u32) cpu_to_be32(ci),
dev->eq_regs.arbel.eq_set_ci_base + eq->eqn * 8);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
/* We still want ordering, just not swabbing, so add a barrier */
mb();
}
static inline void set_eq_ci(struct mthca_dev *dev, struct mthca_eq *eq, u32 ci)
{
[PATCH] IB/mthca: encapsulate mem-free check into mthca_is_memfree() Clean up mem-free mode support by introducing mthca_is_memfree() function, which encapsulates the logic of deciding if a device is mem-free. Signed-off-by: Roland Dreier <roland@topspin.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-04-17 06:26:32 +08:00
if (mthca_is_memfree(dev))
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
arbel_set_eq_ci(dev, eq, ci);
else
tavor_set_eq_ci(dev, eq, ci);
}
static inline void tavor_eq_req_not(struct mthca_dev *dev, int eqn)
{
IB/mthca: Avoid alignment traps when writing doorbells Architectures such as ia64 see alignment traps when doing a 64-bit read from __be32 doorbell[2] arrays to do doorbell writes in mthca_write64(). Fix this by just passing the two halves of the doorbell value into mthca_write64(). This actually improves the generated code by allowing the compiler to see what's going on better. Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-10-15 11:40:27 +08:00
mthca_write64(MTHCA_EQ_DB_REQ_NOT | eqn, 0,
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
dev->kar + MTHCA_EQ_DOORBELL,
MTHCA_GET_DOORBELL_LOCK(&dev->doorbell_lock));
}
static inline void arbel_eq_req_not(struct mthca_dev *dev, u32 eqn_mask)
{
writel(eqn_mask, dev->eq_regs.arbel.eq_arm);
}
static inline void disarm_cq(struct mthca_dev *dev, int eqn, int cqn)
{
[PATCH] IB/mthca: encapsulate mem-free check into mthca_is_memfree() Clean up mem-free mode support by introducing mthca_is_memfree() function, which encapsulates the logic of deciding if a device is mem-free. Signed-off-by: Roland Dreier <roland@topspin.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-04-17 06:26:32 +08:00
if (!mthca_is_memfree(dev)) {
IB/mthca: Avoid alignment traps when writing doorbells Architectures such as ia64 see alignment traps when doing a 64-bit read from __be32 doorbell[2] arrays to do doorbell writes in mthca_write64(). Fix this by just passing the two halves of the doorbell value into mthca_write64(). This actually improves the generated code by allowing the compiler to see what's going on better. Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-10-15 11:40:27 +08:00
mthca_write64(MTHCA_EQ_DB_DISARM_CQ | eqn, cqn,
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
dev->kar + MTHCA_EQ_DOORBELL,
MTHCA_GET_DOORBELL_LOCK(&dev->doorbell_lock));
}
}
static inline struct mthca_eqe *get_eqe(struct mthca_eq *eq, u32 entry)
{
unsigned long off = (entry & (eq->nent - 1)) * MTHCA_EQ_ENTRY_SIZE;
return eq->page_list[off / PAGE_SIZE].buf + off % PAGE_SIZE;
}
IB/mthca: Formatting cleanups Fix a few whitespace and other coding style problems. Signed-off-by: Roland Dreier <rolandd@cisco.com>
2008-04-17 12:01:03 +08:00
static inline struct mthca_eqe *next_eqe_sw(struct mthca_eq *eq)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
{
IB/mthca: Formatting cleanups Fix a few whitespace and other coding style problems. Signed-off-by: Roland Dreier <rolandd@cisco.com>
2008-04-17 12:01:03 +08:00
struct mthca_eqe *eqe;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
eqe = get_eqe(eq, eq->cons_index);
return (MTHCA_EQ_ENTRY_OWNER_HW & eqe->owner) ? NULL : eqe;
}
static inline void set_eqe_hw(struct mthca_eqe *eqe)
{
eqe->owner = MTHCA_EQ_ENTRY_OWNER_HW;
}
static void port_change(struct mthca_dev *dev, int port, int active)
{
struct ib_event record;
mthca_dbg(dev, "Port change to %s for port %d\n",
active ? "active" : "down", port);
record.device = &dev->ib_dev;
record.event = active ? IB_EVENT_PORT_ACTIVE : IB_EVENT_PORT_ERR;
record.element.port_num = port;
ib_dispatch_event(&record);
}
static int mthca_eq_int(struct mthca_dev *dev, struct mthca_eq *eq)
{
struct mthca_eqe *eqe;
int disarm_cqn;
IB/mthca: prevent event queue overrun I am seeing EQ overruns in SDP stress tests: if the CQ completion handler arms a CQ, this could generate more EQEs, so that EQ will never get empty and consumer index will never get updated. This is similiar to what we have with command interface: /* * cmd_event() may add more commands. * The card will think the queue has overflowed if * we don't tell it we've been processing events. */ However, for completion events, we *don't* want to update the consumer index on each event. So, perform EQ doorbell coalescing: allocate EQs with some spare EQEs, and update once we run out of them. The value 0x80 was selected to avoid any performance impact. Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-01-10 06:04:40 +08:00
int eqes_found = 0;
int set_ci = 0;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
while ((eqe = next_eqe_sw(eq))) {
/*
* Make sure we read EQ entry contents after we've
* checked the ownership bit.
*/
rmb();
switch (eqe->type) {
case MTHCA_EVENT_TYPE_COMP:
disarm_cqn = be32_to_cpu(eqe->event.comp.cqn) & 0xffffff;
disarm_cq(dev, eq->eqn, disarm_cqn);
[IB] mthca: report asynchronous CQ events Implement reporting asynchronous CQ events in Mellanox HCA driver. Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-29 22:39:42 +08:00
mthca_cq_completion(dev, disarm_cqn);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
break;
case MTHCA_EVENT_TYPE_PATH_MIG:
mthca_qp_event(dev, be32_to_cpu(eqe->event.qp.qpn) & 0xffffff,
IB_EVENT_PATH_MIG);
break;
case MTHCA_EVENT_TYPE_COMM_EST:
mthca_qp_event(dev, be32_to_cpu(eqe->event.qp.qpn) & 0xffffff,
IB_EVENT_COMM_EST);
break;
case MTHCA_EVENT_TYPE_SQ_DRAINED:
mthca_qp_event(dev, be32_to_cpu(eqe->event.qp.qpn) & 0xffffff,
IB_EVENT_SQ_DRAINED);
break;
[IB] mthca: SRQ limit reached events Our hardware supports generating an event when the number of receives posted to a shared receive queue (SRQ) falls below a user-specified limit. Implement mthca_modify_srq() to arm the limit, and add code to handle dispatching SRQ events when they occur. Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-07 04:15:56 +08:00
case MTHCA_EVENT_TYPE_SRQ_QP_LAST_WQE:
mthca_qp_event(dev, be32_to_cpu(eqe->event.qp.qpn) & 0xffffff,
IB_EVENT_QP_LAST_WQE_REACHED);
break;
case MTHCA_EVENT_TYPE_SRQ_LIMIT:
mthca_srq_event(dev, be32_to_cpu(eqe->event.srq.srqn) & 0xffffff,
IB_EVENT_SRQ_LIMIT_REACHED);
break;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
case MTHCA_EVENT_TYPE_WQ_CATAS_ERROR:
mthca_qp_event(dev, be32_to_cpu(eqe->event.qp.qpn) & 0xffffff,
IB_EVENT_QP_FATAL);
break;
case MTHCA_EVENT_TYPE_PATH_MIG_FAILED:
mthca_qp_event(dev, be32_to_cpu(eqe->event.qp.qpn) & 0xffffff,
IB_EVENT_PATH_MIG_ERR);
break;
case MTHCA_EVENT_TYPE_WQ_INVAL_REQ_ERROR:
mthca_qp_event(dev, be32_to_cpu(eqe->event.qp.qpn) & 0xffffff,
IB_EVENT_QP_REQ_ERR);
break;
case MTHCA_EVENT_TYPE_WQ_ACCESS_ERROR:
mthca_qp_event(dev, be32_to_cpu(eqe->event.qp.qpn) & 0xffffff,
IB_EVENT_QP_ACCESS_ERR);
break;
case MTHCA_EVENT_TYPE_CMD:
mthca_cmd_event(dev,
be16_to_cpu(eqe->event.cmd.token),
eqe->event.cmd.status,
be64_to_cpu(eqe->event.cmd.out_param));
break;
case MTHCA_EVENT_TYPE_PORT_CHANGE:
port_change(dev,
(be32_to_cpu(eqe->event.port_change.port) >> 28) & 3,
eqe->subtype == 0x4);
break;
case MTHCA_EVENT_TYPE_CQ_ERROR:
[PATCH] IB/mthca: fix format of CQ number for CQ events CQ numbers are only 24 bits, so only print 6 hex digits and mask off reserved part when reporting a CQ event. Signed-off-by: Roland Dreier <roland@topspin.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-04-17 06:26:22 +08:00
mthca_warn(dev, "CQ %s on CQN %06x\n",
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
eqe->event.cq_err.syndrome == 1 ?
"overrun" : "access violation",
[PATCH] IB/mthca: fix format of CQ number for CQ events CQ numbers are only 24 bits, so only print 6 hex digits and mask off reserved part when reporting a CQ event. Signed-off-by: Roland Dreier <roland@topspin.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-04-17 06:26:22 +08:00
be32_to_cpu(eqe->event.cq_err.cqn) & 0xffffff);
[IB] mthca: report asynchronous CQ events Implement reporting asynchronous CQ events in Mellanox HCA driver. Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-29 22:39:42 +08:00
mthca_cq_event(dev, be32_to_cpu(eqe->event.cq_err.cqn),
IB_EVENT_CQ_ERR);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
break;
case MTHCA_EVENT_TYPE_EQ_OVERFLOW:
mthca_warn(dev, "EQ overrun on EQN %d\n", eq->eqn);
break;
case MTHCA_EVENT_TYPE_EEC_CATAS_ERROR:
case MTHCA_EVENT_TYPE_SRQ_CATAS_ERROR:
case MTHCA_EVENT_TYPE_LOCAL_CATAS_ERROR:
case MTHCA_EVENT_TYPE_ECC_DETECT:
default:
mthca_warn(dev, "Unhandled event %02x(%02x) on EQ %d\n",
eqe->type, eqe->subtype, eq->eqn);
break;
};
set_eqe_hw(eqe);
++eq->cons_index;
eqes_found = 1;
IB/mthca: prevent event queue overrun I am seeing EQ overruns in SDP stress tests: if the CQ completion handler arms a CQ, this could generate more EQEs, so that EQ will never get empty and consumer index will never get updated. This is similiar to what we have with command interface: /* * cmd_event() may add more commands. * The card will think the queue has overflowed if * we don't tell it we've been processing events. */ However, for completion events, we *don't* want to update the consumer index on each event. So, perform EQ doorbell coalescing: allocate EQs with some spare EQEs, and update once we run out of them. The value 0x80 was selected to avoid any performance impact. Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-01-10 06:04:40 +08:00
++set_ci;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
IB/mthca: prevent event queue overrun I am seeing EQ overruns in SDP stress tests: if the CQ completion handler arms a CQ, this could generate more EQEs, so that EQ will never get empty and consumer index will never get updated. This is similiar to what we have with command interface: /* * cmd_event() may add more commands. * The card will think the queue has overflowed if * we don't tell it we've been processing events. */ However, for completion events, we *don't* want to update the consumer index on each event. So, perform EQ doorbell coalescing: allocate EQs with some spare EQEs, and update once we run out of them. The value 0x80 was selected to avoid any performance impact. Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-01-10 06:04:40 +08:00
/*
* The HCA will think the queue has overflowed if we
* don't tell it we've been processing events. We
* create our EQs with MTHCA_NUM_SPARE_EQE extra
* entries, so we must update our consumer index at
* least that often.
*/
if (unlikely(set_ci >= MTHCA_NUM_SPARE_EQE)) {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
/*
* Conditional on hca_type is OK here because
* this is a rare case, not the fast path.
*/
set_eq_ci(dev, eq, eq->cons_index);
set_ci = 0;
}
}
/*
* Rely on caller to set consumer index so that we don't have
* to test hca_type in our interrupt handling fast path.
*/
return eqes_found;
}
IRQ: Maintain regs pointer globally rather than passing to IRQ handlers Maintain a per-CPU global "struct pt_regs *" variable which can be used instead of passing regs around manually through all ~1800 interrupt handlers in the Linux kernel. The regs pointer is used in few places, but it potentially costs both stack space and code to pass it around. On the FRV arch, removing the regs parameter from all the genirq function results in a 20% speed up of the IRQ exit path (ie: from leaving timer_interrupt() to leaving do_IRQ()). Where appropriate, an arch may override the generic storage facility and do something different with the variable. On FRV, for instance, the address is maintained in GR28 at all times inside the kernel as part of general exception handling. Having looked over the code, it appears that the parameter may be handed down through up to twenty or so layers of functions. Consider a USB character device attached to a USB hub, attached to a USB controller that posts its interrupts through a cascaded auxiliary interrupt controller. A character device driver may want to pass regs to the sysrq handler through the input layer which adds another few layers of parameter passing. I've build this code with allyesconfig for x86_64 and i386. I've runtested the main part of the code on FRV and i386, though I can't test most of the drivers. I've also done partial conversion for powerpc and MIPS - these at least compile with minimal configurations. This will affect all archs. Mostly the changes should be relatively easy. Take do_IRQ(), store the regs pointer at the beginning, saving the old one: struct pt_regs *old_regs = set_irq_regs(regs); And put the old one back at the end: set_irq_regs(old_regs); Don't pass regs through to generic_handle_irq() or __do_IRQ(). In timer_interrupt(), this sort of change will be necessary: - update_process_times(user_mode(regs)); - profile_tick(CPU_PROFILING, regs); + update_process_times(user_mode(get_irq_regs())); + profile_tick(CPU_PROFILING); I'd like to move update_process_times()'s use of get_irq_regs() into itself, except that i386, alone of the archs, uses something other than user_mode(). Some notes on the interrupt handling in the drivers: (*) input_dev() is now gone entirely. The regs pointer is no longer stored in the input_dev struct. (*) finish_unlinks() in drivers/usb/host/ohci-q.c needs checking. It does something different depending on whether it's been supplied with a regs pointer or not. (*) Various IRQ handler function pointers have been moved to type irq_handler_t. Signed-Off-By: David Howells <dhowells@redhat.com> (cherry picked from 1b16e7ac850969f38b375e511e3fa2f474a33867 commit)
2006-10-05 21:55:46 +08:00
static irqreturn_t mthca_tavor_interrupt(int irq, void *dev_ptr)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
{
struct mthca_dev *dev = dev_ptr;
u32 ecr;
int i;
if (dev->eq_table.clr_mask)
writel(dev->eq_table.clr_mask, dev->eq_table.clr_int);
ecr = readl(dev->eq_regs.tavor.ecr_base + 4);
[IB] mthca: Always re-arm EQs in mthca_tavor_interrupt() We should always re-arm an event queue's interrupt in mthca_tavor_interrupt() if the corresponding bit is set in the event cause register (ECR), even if we didn't find any entries in the EQ. If we don't, then there's a window where we miss an EQ entry and then get stuck because we don't get another EQ event. Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-23 00:43:29 +08:00
if (!ecr)
return IRQ_NONE;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
[IB] mthca: Always re-arm EQs in mthca_tavor_interrupt() We should always re-arm an event queue's interrupt in mthca_tavor_interrupt() if the corresponding bit is set in the event cause register (ECR), even if we didn't find any entries in the EQ. If we don't, then there's a window where we miss an EQ entry and then get stuck because we don't get another EQ event. Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-23 00:43:29 +08:00
writel(ecr, dev->eq_regs.tavor.ecr_base +
MTHCA_ECR_CLR_BASE - MTHCA_ECR_BASE + 4);
for (i = 0; i < MTHCA_NUM_EQ; ++i)
if (ecr & dev->eq_table.eq[i].eqn_mask) {
if (mthca_eq_int(dev, &dev->eq_table.eq[i]))
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
tavor_set_eq_ci(dev, &dev->eq_table.eq[i],
dev->eq_table.eq[i].cons_index);
[IB] mthca: Always re-arm EQs in mthca_tavor_interrupt() We should always re-arm an event queue's interrupt in mthca_tavor_interrupt() if the corresponding bit is set in the event cause register (ECR), even if we didn't find any entries in the EQ. If we don't, then there's a window where we miss an EQ entry and then get stuck because we don't get another EQ event. Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-23 00:43:29 +08:00
tavor_eq_req_not(dev, dev->eq_table.eq[i].eqn);
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
[IB] mthca: Always re-arm EQs in mthca_tavor_interrupt() We should always re-arm an event queue's interrupt in mthca_tavor_interrupt() if the corresponding bit is set in the event cause register (ECR), even if we didn't find any entries in the EQ. If we don't, then there's a window where we miss an EQ entry and then get stuck because we don't get another EQ event. Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-23 00:43:29 +08:00
return IRQ_HANDLED;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
IRQ: Maintain regs pointer globally rather than passing to IRQ handlers Maintain a per-CPU global "struct pt_regs *" variable which can be used instead of passing regs around manually through all ~1800 interrupt handlers in the Linux kernel. The regs pointer is used in few places, but it potentially costs both stack space and code to pass it around. On the FRV arch, removing the regs parameter from all the genirq function results in a 20% speed up of the IRQ exit path (ie: from leaving timer_interrupt() to leaving do_IRQ()). Where appropriate, an arch may override the generic storage facility and do something different with the variable. On FRV, for instance, the address is maintained in GR28 at all times inside the kernel as part of general exception handling. Having looked over the code, it appears that the parameter may be handed down through up to twenty or so layers of functions. Consider a USB character device attached to a USB hub, attached to a USB controller that posts its interrupts through a cascaded auxiliary interrupt controller. A character device driver may want to pass regs to the sysrq handler through the input layer which adds another few layers of parameter passing. I've build this code with allyesconfig for x86_64 and i386. I've runtested the main part of the code on FRV and i386, though I can't test most of the drivers. I've also done partial conversion for powerpc and MIPS - these at least compile with minimal configurations. This will affect all archs. Mostly the changes should be relatively easy. Take do_IRQ(), store the regs pointer at the beginning, saving the old one: struct pt_regs *old_regs = set_irq_regs(regs); And put the old one back at the end: set_irq_regs(old_regs); Don't pass regs through to generic_handle_irq() or __do_IRQ(). In timer_interrupt(), this sort of change will be necessary: - update_process_times(user_mode(regs)); - profile_tick(CPU_PROFILING, regs); + update_process_times(user_mode(get_irq_regs())); + profile_tick(CPU_PROFILING); I'd like to move update_process_times()'s use of get_irq_regs() into itself, except that i386, alone of the archs, uses something other than user_mode(). Some notes on the interrupt handling in the drivers: (*) input_dev() is now gone entirely. The regs pointer is no longer stored in the input_dev struct. (*) finish_unlinks() in drivers/usb/host/ohci-q.c needs checking. It does something different depending on whether it's been supplied with a regs pointer or not. (*) Various IRQ handler function pointers have been moved to type irq_handler_t. Signed-Off-By: David Howells <dhowells@redhat.com> (cherry picked from 1b16e7ac850969f38b375e511e3fa2f474a33867 commit)
2006-10-05 21:55:46 +08:00
static irqreturn_t mthca_tavor_msi_x_interrupt(int irq, void *eq_ptr)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
{
struct mthca_eq *eq = eq_ptr;
struct mthca_dev *dev = eq->dev;
mthca_eq_int(dev, eq);
tavor_set_eq_ci(dev, eq, eq->cons_index);
tavor_eq_req_not(dev, eq->eqn);
/* MSI-X vectors always belong to us */
return IRQ_HANDLED;
}
IRQ: Maintain regs pointer globally rather than passing to IRQ handlers Maintain a per-CPU global "struct pt_regs *" variable which can be used instead of passing regs around manually through all ~1800 interrupt handlers in the Linux kernel. The regs pointer is used in few places, but it potentially costs both stack space and code to pass it around. On the FRV arch, removing the regs parameter from all the genirq function results in a 20% speed up of the IRQ exit path (ie: from leaving timer_interrupt() to leaving do_IRQ()). Where appropriate, an arch may override the generic storage facility and do something different with the variable. On FRV, for instance, the address is maintained in GR28 at all times inside the kernel as part of general exception handling. Having looked over the code, it appears that the parameter may be handed down through up to twenty or so layers of functions. Consider a USB character device attached to a USB hub, attached to a USB controller that posts its interrupts through a cascaded auxiliary interrupt controller. A character device driver may want to pass regs to the sysrq handler through the input layer which adds another few layers of parameter passing. I've build this code with allyesconfig for x86_64 and i386. I've runtested the main part of the code on FRV and i386, though I can't test most of the drivers. I've also done partial conversion for powerpc and MIPS - these at least compile with minimal configurations. This will affect all archs. Mostly the changes should be relatively easy. Take do_IRQ(), store the regs pointer at the beginning, saving the old one: struct pt_regs *old_regs = set_irq_regs(regs); And put the old one back at the end: set_irq_regs(old_regs); Don't pass regs through to generic_handle_irq() or __do_IRQ(). In timer_interrupt(), this sort of change will be necessary: - update_process_times(user_mode(regs)); - profile_tick(CPU_PROFILING, regs); + update_process_times(user_mode(get_irq_regs())); + profile_tick(CPU_PROFILING); I'd like to move update_process_times()'s use of get_irq_regs() into itself, except that i386, alone of the archs, uses something other than user_mode(). Some notes on the interrupt handling in the drivers: (*) input_dev() is now gone entirely. The regs pointer is no longer stored in the input_dev struct. (*) finish_unlinks() in drivers/usb/host/ohci-q.c needs checking. It does something different depending on whether it's been supplied with a regs pointer or not. (*) Various IRQ handler function pointers have been moved to type irq_handler_t. Signed-Off-By: David Howells <dhowells@redhat.com> (cherry picked from 1b16e7ac850969f38b375e511e3fa2f474a33867 commit)
2006-10-05 21:55:46 +08:00
static irqreturn_t mthca_arbel_interrupt(int irq, void *dev_ptr)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
{
struct mthca_dev *dev = dev_ptr;
int work = 0;
int i;
if (dev->eq_table.clr_mask)
writel(dev->eq_table.clr_mask, dev->eq_table.clr_int);
for (i = 0; i < MTHCA_NUM_EQ; ++i)
if (mthca_eq_int(dev, &dev->eq_table.eq[i])) {
work = 1;
arbel_set_eq_ci(dev, &dev->eq_table.eq[i],
dev->eq_table.eq[i].cons_index);
}
arbel_eq_req_not(dev, dev->eq_table.arm_mask);
return IRQ_RETVAL(work);
}
IRQ: Maintain regs pointer globally rather than passing to IRQ handlers Maintain a per-CPU global "struct pt_regs *" variable which can be used instead of passing regs around manually through all ~1800 interrupt handlers in the Linux kernel. The regs pointer is used in few places, but it potentially costs both stack space and code to pass it around. On the FRV arch, removing the regs parameter from all the genirq function results in a 20% speed up of the IRQ exit path (ie: from leaving timer_interrupt() to leaving do_IRQ()). Where appropriate, an arch may override the generic storage facility and do something different with the variable. On FRV, for instance, the address is maintained in GR28 at all times inside the kernel as part of general exception handling. Having looked over the code, it appears that the parameter may be handed down through up to twenty or so layers of functions. Consider a USB character device attached to a USB hub, attached to a USB controller that posts its interrupts through a cascaded auxiliary interrupt controller. A character device driver may want to pass regs to the sysrq handler through the input layer which adds another few layers of parameter passing. I've build this code with allyesconfig for x86_64 and i386. I've runtested the main part of the code on FRV and i386, though I can't test most of the drivers. I've also done partial conversion for powerpc and MIPS - these at least compile with minimal configurations. This will affect all archs. Mostly the changes should be relatively easy. Take do_IRQ(), store the regs pointer at the beginning, saving the old one: struct pt_regs *old_regs = set_irq_regs(regs); And put the old one back at the end: set_irq_regs(old_regs); Don't pass regs through to generic_handle_irq() or __do_IRQ(). In timer_interrupt(), this sort of change will be necessary: - update_process_times(user_mode(regs)); - profile_tick(CPU_PROFILING, regs); + update_process_times(user_mode(get_irq_regs())); + profile_tick(CPU_PROFILING); I'd like to move update_process_times()'s use of get_irq_regs() into itself, except that i386, alone of the archs, uses something other than user_mode(). Some notes on the interrupt handling in the drivers: (*) input_dev() is now gone entirely. The regs pointer is no longer stored in the input_dev struct. (*) finish_unlinks() in drivers/usb/host/ohci-q.c needs checking. It does something different depending on whether it's been supplied with a regs pointer or not. (*) Various IRQ handler function pointers have been moved to type irq_handler_t. Signed-Off-By: David Howells <dhowells@redhat.com> (cherry picked from 1b16e7ac850969f38b375e511e3fa2f474a33867 commit)
2006-10-05 21:55:46 +08:00
static irqreturn_t mthca_arbel_msi_x_interrupt(int irq, void *eq_ptr)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
{
struct mthca_eq *eq = eq_ptr;
struct mthca_dev *dev = eq->dev;
mthca_eq_int(dev, eq);
arbel_set_eq_ci(dev, eq, eq->cons_index);
arbel_eq_req_not(dev, eq->eqn_mask);
/* MSI-X vectors always belong to us */
return IRQ_HANDLED;
}
IB/mthca: Fix section mismatches Commit b3b30f5e ("IB/mthca: Recover from catastrophic errors") introduced some section mismatch breakage, because the error recovery code tears down and reinitializes the device, which calls into lots of code originally marked __devinit and __devexit from regular .text. Fix this by getting rid of these now-incorrect section markers. Reported by Randy Dunlap <randy.dunlap@oracle.com>. Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-11-30 07:33:06 +08:00
static int mthca_create_eq(struct mthca_dev *dev,
int nent,
u8 intr,
struct mthca_eq *eq)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
{
IB/mthca: create_eq with size not a power of 2 Fix mthca_create_eq for when the EQ size is not a power of 2. Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-01-06 08:17:38 +08:00
int npages;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
u64 *dma_list = NULL;
dma_addr_t t;
[PATCH] IB/mthca: Align FW command mailboxes to 4K Future versions of Mellanox HCA firmware will require command mailboxes to be aligned to 4K. Support this by using a pci_pool to allocate all mailboxes. This has the added benefit of shrinking the source and text of mthca. Signed-off-by: Roland Dreier <roland@topspin.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-28 05:36:45 +08:00
struct mthca_mailbox *mailbox;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
struct mthca_eq_context *eq_context;
int err = -ENOMEM;
int i;
u8 status;
[PATCH] IB/mthca: Initialize eq->nent before we use it In mthca_create_eq(), we call get_eqe() before setting eq->nent. This is wrong, because get_eqe() uses eq->nent. Fix this, and clean up the code a little while we're at it. (We got lucky with the current code, because eq->nent was cleared to 0, which get_eqe() made happen to do the right thing) Pointed out by Michael S. Tsirkin <mst@mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-09-19 04:52:06 +08:00
eq->dev = dev;
eq->nent = roundup_pow_of_two(max(nent, 2));
IB/mthca: Whitespace cleanups Remove trailing whitespace and fix indentation that with spaces instead of tabs. Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-02-02 05:38:24 +08:00
npages = ALIGN(eq->nent * MTHCA_EQ_ENTRY_SIZE, PAGE_SIZE) / PAGE_SIZE;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
eq->page_list = kmalloc(npages * sizeof *eq->page_list,
GFP_KERNEL);
if (!eq->page_list)
goto err_out;
for (i = 0; i < npages; ++i)
eq->page_list[i].buf = NULL;
dma_list = kmalloc(npages * sizeof *dma_list, GFP_KERNEL);
if (!dma_list)
goto err_out_free;
[PATCH] IB/mthca: Align FW command mailboxes to 4K Future versions of Mellanox HCA firmware will require command mailboxes to be aligned to 4K. Support this by using a pci_pool to allocate all mailboxes. This has the added benefit of shrinking the source and text of mthca. Signed-off-by: Roland Dreier <roland@topspin.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-28 05:36:45 +08:00
mailbox = mthca_alloc_mailbox(dev, GFP_KERNEL);
if (IS_ERR(mailbox))
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
goto err_out_free;
[PATCH] IB/mthca: Align FW command mailboxes to 4K Future versions of Mellanox HCA firmware will require command mailboxes to be aligned to 4K. Support this by using a pci_pool to allocate all mailboxes. This has the added benefit of shrinking the source and text of mthca. Signed-off-by: Roland Dreier <roland@topspin.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-28 05:36:45 +08:00
eq_context = mailbox->buf;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
for (i = 0; i < npages; ++i) {
[PATCH] IB/mthca: Use dma_alloc_coherent instead of pci_alloc_consistent Switch all allocations of coherent memory from pci_alloc_consistent() to dma_alloc_coherent(), so that we can pass GFP_KERNEL. This should help when the system is low on memory. Signed-off-by: Roland Dreier <roland@topspin.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-28 05:36:40 +08:00
eq->page_list[i].buf = dma_alloc_coherent(&dev->pdev->dev,
PAGE_SIZE, &t, GFP_KERNEL);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
if (!eq->page_list[i].buf)
[PATCH] IB/mthca: Align FW command mailboxes to 4K Future versions of Mellanox HCA firmware will require command mailboxes to be aligned to 4K. Support this by using a pci_pool to allocate all mailboxes. This has the added benefit of shrinking the source and text of mthca. Signed-off-by: Roland Dreier <roland@topspin.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-28 05:36:45 +08:00
goto err_out_free_pages;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
dma_list[i] = t;
pci_unmap_addr_set(&eq->page_list[i], mapping, t);
IB/mthca: Replace memset(<addr>, 0, PAGE_SIZE) with clear_page(<addr>) Signed-off-by: Shani Moideen <shani.moideen@wipro.com> Signed-off-by: Roland Dreier <rolandd@cisco.com> ----
2007-06-18 11:16:41 +08:00
clear_page(eq->page_list[i].buf);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
[PATCH] IB/mthca: Initialize eq->nent before we use it In mthca_create_eq(), we call get_eqe() before setting eq->nent. This is wrong, because get_eqe() uses eq->nent. Fix this, and clean up the code a little while we're at it. (We got lucky with the current code, because eq->nent was cleared to 0, which get_eqe() made happen to do the right thing) Pointed out by Michael S. Tsirkin <mst@mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-09-19 04:52:06 +08:00
for (i = 0; i < eq->nent; ++i)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
set_eqe_hw(get_eqe(eq, i));
eq->eqn = mthca_alloc(&dev->eq_table.alloc);
if (eq->eqn == -1)
[PATCH] IB/mthca: Align FW command mailboxes to 4K Future versions of Mellanox HCA firmware will require command mailboxes to be aligned to 4K. Support this by using a pci_pool to allocate all mailboxes. This has the added benefit of shrinking the source and text of mthca. Signed-off-by: Roland Dreier <roland@topspin.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-28 05:36:45 +08:00
goto err_out_free_pages;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
err = mthca_mr_alloc_phys(dev, dev->driver_pd.pd_num,
dma_list, PAGE_SHIFT, npages,
0, npages * PAGE_SIZE,
MTHCA_MPT_FLAG_LOCAL_WRITE |
MTHCA_MPT_FLAG_LOCAL_READ,
&eq->mr);
if (err)
goto err_out_free_eq;
memset(eq_context, 0, sizeof *eq_context);
eq_context->flags = cpu_to_be32(MTHCA_EQ_STATUS_OK |
MTHCA_EQ_OWNER_HW |
MTHCA_EQ_STATE_ARMED |
MTHCA_EQ_FLAG_TR);
[PATCH] IB/mthca: encapsulate mem-free check into mthca_is_memfree() Clean up mem-free mode support by introducing mthca_is_memfree() function, which encapsulates the logic of deciding if a device is mem-free. Signed-off-by: Roland Dreier <roland@topspin.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-04-17 06:26:32 +08:00
if (mthca_is_memfree(dev))
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
eq_context->flags |= cpu_to_be32(MTHCA_EQ_STATE_ARBEL);
[PATCH] IB/mthca: Initialize eq->nent before we use it In mthca_create_eq(), we call get_eqe() before setting eq->nent. This is wrong, because get_eqe() uses eq->nent. Fix this, and clean up the code a little while we're at it. (We got lucky with the current code, because eq->nent was cleared to 0, which get_eqe() made happen to do the right thing) Pointed out by Michael S. Tsirkin <mst@mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-09-19 04:52:06 +08:00
eq_context->logsize_usrpage = cpu_to_be32((ffs(eq->nent) - 1) << 24);
[PATCH] IB/mthca: encapsulate mem-free check into mthca_is_memfree() Clean up mem-free mode support by introducing mthca_is_memfree() function, which encapsulates the logic of deciding if a device is mem-free. Signed-off-by: Roland Dreier <roland@topspin.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-04-17 06:26:32 +08:00
if (mthca_is_memfree(dev)) {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
eq_context->arbel_pd = cpu_to_be32(dev->driver_pd.pd_num);
} else {
eq_context->logsize_usrpage |= cpu_to_be32(dev->driver_uar.index);
eq_context->tavor_pd = cpu_to_be32(dev->driver_pd.pd_num);
}
eq_context->intr = intr;
eq_context->lkey = cpu_to_be32(eq->mr.ibmr.lkey);
[PATCH] IB/mthca: Align FW command mailboxes to 4K Future versions of Mellanox HCA firmware will require command mailboxes to be aligned to 4K. Support this by using a pci_pool to allocate all mailboxes. This has the added benefit of shrinking the source and text of mthca. Signed-off-by: Roland Dreier <roland@topspin.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-28 05:36:45 +08:00
err = mthca_SW2HW_EQ(dev, mailbox, eq->eqn, &status);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
if (err) {
mthca_warn(dev, "SW2HW_EQ failed (%d)\n", err);
goto err_out_free_mr;
}
if (status) {
mthca_warn(dev, "SW2HW_EQ returned status 0x%02x\n",
status);
err = -EINVAL;
goto err_out_free_mr;
}
kfree(dma_list);
[PATCH] IB/mthca: Align FW command mailboxes to 4K Future versions of Mellanox HCA firmware will require command mailboxes to be aligned to 4K. Support this by using a pci_pool to allocate all mailboxes. This has the added benefit of shrinking the source and text of mthca. Signed-off-by: Roland Dreier <roland@topspin.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-28 05:36:45 +08:00
mthca_free_mailbox(dev, mailbox);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
eq->eqn_mask = swab32(1 << eq->eqn);
eq->cons_index = 0;
dev->eq_table.arm_mask |= eq->eqn_mask;
mthca_dbg(dev, "Allocated EQ %d with %d entries\n",
[PATCH] IB/mthca: Initialize eq->nent before we use it In mthca_create_eq(), we call get_eqe() before setting eq->nent. This is wrong, because get_eqe() uses eq->nent. Fix this, and clean up the code a little while we're at it. (We got lucky with the current code, because eq->nent was cleared to 0, which get_eqe() made happen to do the right thing) Pointed out by Michael S. Tsirkin <mst@mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-09-19 04:52:06 +08:00
eq->eqn, eq->nent);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
return err;
err_out_free_mr:
mthca_free_mr(dev, &eq->mr);
err_out_free_eq:
mthca_free(&dev->eq_table.alloc, eq->eqn);
[PATCH] IB/mthca: Align FW command mailboxes to 4K Future versions of Mellanox HCA firmware will require command mailboxes to be aligned to 4K. Support this by using a pci_pool to allocate all mailboxes. This has the added benefit of shrinking the source and text of mthca. Signed-off-by: Roland Dreier <roland@topspin.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-28 05:36:45 +08:00
err_out_free_pages:
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
for (i = 0; i < npages; ++i)
if (eq->page_list[i].buf)
[PATCH] IB/mthca: Use dma_alloc_coherent instead of pci_alloc_consistent Switch all allocations of coherent memory from pci_alloc_consistent() to dma_alloc_coherent(), so that we can pass GFP_KERNEL. This should help when the system is low on memory. Signed-off-by: Roland Dreier <roland@topspin.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-28 05:36:40 +08:00
dma_free_coherent(&dev->pdev->dev, PAGE_SIZE,
eq->page_list[i].buf,
pci_unmap_addr(&eq->page_list[i],
mapping));
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
[PATCH] IB/mthca: Align FW command mailboxes to 4K Future versions of Mellanox HCA firmware will require command mailboxes to be aligned to 4K. Support this by using a pci_pool to allocate all mailboxes. This has the added benefit of shrinking the source and text of mthca. Signed-off-by: Roland Dreier <roland@topspin.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-28 05:36:45 +08:00
mthca_free_mailbox(dev, mailbox);
err_out_free:
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
kfree(eq->page_list);
kfree(dma_list);
err_out:
return err;
}
static void mthca_free_eq(struct mthca_dev *dev,
struct mthca_eq *eq)
{
[PATCH] IB/mthca: Align FW command mailboxes to 4K Future versions of Mellanox HCA firmware will require command mailboxes to be aligned to 4K. Support this by using a pci_pool to allocate all mailboxes. This has the added benefit of shrinking the source and text of mthca. Signed-off-by: Roland Dreier <roland@topspin.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-28 05:36:45 +08:00
struct mthca_mailbox *mailbox;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
int err;
u8 status;
int npages = (eq->nent * MTHCA_EQ_ENTRY_SIZE + PAGE_SIZE - 1) /
PAGE_SIZE;
int i;
[PATCH] IB/mthca: Align FW command mailboxes to 4K Future versions of Mellanox HCA firmware will require command mailboxes to be aligned to 4K. Support this by using a pci_pool to allocate all mailboxes. This has the added benefit of shrinking the source and text of mthca. Signed-off-by: Roland Dreier <roland@topspin.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-28 05:36:45 +08:00
mailbox = mthca_alloc_mailbox(dev, GFP_KERNEL);
if (IS_ERR(mailbox))
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
return;
[PATCH] IB/mthca: Align FW command mailboxes to 4K Future versions of Mellanox HCA firmware will require command mailboxes to be aligned to 4K. Support this by using a pci_pool to allocate all mailboxes. This has the added benefit of shrinking the source and text of mthca. Signed-off-by: Roland Dreier <roland@topspin.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-28 05:36:45 +08:00
err = mthca_HW2SW_EQ(dev, mailbox, eq->eqn, &status);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
if (err)
mthca_warn(dev, "HW2SW_EQ failed (%d)\n", err);
if (status)
[PATCH] IB/mthca: Clean up error messages - Fix incorrect cut-n-paste in error messages. - Add missing newlines in error messages. - Use DRV_NAME instead of "ib_mthca" in a couple of places. Signed-off-by: Roland Dreier <roland@topspin.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-28 05:36:39 +08:00
mthca_warn(dev, "HW2SW_EQ returned status 0x%02x\n", status);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
dev->eq_table.arm_mask &= ~eq->eqn_mask;
if (0) {
mthca_dbg(dev, "Dumping EQ context %02x:\n", eq->eqn);
for (i = 0; i < sizeof (struct mthca_eq_context) / 4; ++i) {
if (i % 4 == 0)
printk("[%02x] ", i * 4);
[PATCH] IB/mthca: Align FW command mailboxes to 4K Future versions of Mellanox HCA firmware will require command mailboxes to be aligned to 4K. Support this by using a pci_pool to allocate all mailboxes. This has the added benefit of shrinking the source and text of mthca. Signed-off-by: Roland Dreier <roland@topspin.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-28 05:36:45 +08:00
printk(" %08x", be32_to_cpup(mailbox->buf + i * 4));
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
if ((i + 1) % 4 == 0)
printk("\n");
}
}
mthca_free_mr(dev, &eq->mr);
for (i = 0; i < npages; ++i)
pci_free_consistent(dev->pdev, PAGE_SIZE,
eq->page_list[i].buf,
pci_unmap_addr(&eq->page_list[i], mapping));
kfree(eq->page_list);
[PATCH] IB/mthca: Align FW command mailboxes to 4K Future versions of Mellanox HCA firmware will require command mailboxes to be aligned to 4K. Support this by using a pci_pool to allocate all mailboxes. This has the added benefit of shrinking the source and text of mthca. Signed-off-by: Roland Dreier <roland@topspin.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-28 05:36:45 +08:00
mthca_free_mailbox(dev, mailbox);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
static void mthca_free_irqs(struct mthca_dev *dev)
{
int i;
if (dev->eq_table.have_irq)
free_irq(dev->pdev->irq, dev);
for (i = 0; i < MTHCA_NUM_EQ; ++i)
if (dev->eq_table.eq[i].have_irq)
free_irq(dev->eq_table.eq[i].msi_x_vector,
dev->eq_table.eq + i);
}
IB/mthca: Fix section mismatches Commit b3b30f5e ("IB/mthca: Recover from catastrophic errors") introduced some section mismatch breakage, because the error recovery code tears down and reinitializes the device, which calls into lots of code originally marked __devinit and __devexit from regular .text. Fix this by getting rid of these now-incorrect section markers. Reported by Randy Dunlap <randy.dunlap@oracle.com>. Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-11-30 07:33:06 +08:00
static int mthca_map_reg(struct mthca_dev *dev,
unsigned long offset, unsigned long size,
void __iomem **map)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
{
unsigned long base = pci_resource_start(dev->pdev, 0);
*map = ioremap(base + offset, size);
IB/mthca: Use pci_request_regions() Back in prehistoric (pre-git!) days, the kernel's MSI-X support did request_mem_region() on a device's MSI-X tables, which meant that a driver that enabled MSI-X couldn't use pci_request_regions() (since that would clash with the PCI layer's MSI-X request). However, that was removed (by me!) years ago, so mthca can just use pci_request_regions() and pci_release_regions() instead of its own much more complicated code that avoids requesting the MSI-X tables. Signed-off-by: Roland Dreier <rolandd@cisco.com>
2008-09-30 12:37:33 +08:00
if (!*map)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
return -ENOMEM;
return 0;
}
IB/mthca: Fix section mismatches Commit b3b30f5e ("IB/mthca: Recover from catastrophic errors") introduced some section mismatch breakage, because the error recovery code tears down and reinitializes the device, which calls into lots of code originally marked __devinit and __devexit from regular .text. Fix this by getting rid of these now-incorrect section markers. Reported by Randy Dunlap <randy.dunlap@oracle.com>. Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-11-30 07:33:06 +08:00
static int mthca_map_eq_regs(struct mthca_dev *dev)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
{
[PATCH] IB/mthca: encapsulate mem-free check into mthca_is_memfree() Clean up mem-free mode support by introducing mthca_is_memfree() function, which encapsulates the logic of deciding if a device is mem-free. Signed-off-by: Roland Dreier <roland@topspin.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-04-17 06:26:32 +08:00
if (mthca_is_memfree(dev)) {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
/*
* We assume that the EQ arm and EQ set CI registers
* fall within the first BAR. We can't trust the
* values firmware gives us, since those addresses are
* valid on the HCA's side of the PCI bus but not
* necessarily the host side.
*/
if (mthca_map_reg(dev, (pci_resource_len(dev->pdev, 0) - 1) &
dev->fw.arbel.clr_int_base, MTHCA_CLR_INT_SIZE,
&dev->clr_base)) {
mthca_err(dev, "Couldn't map interrupt clear register, "
"aborting.\n");
return -ENOMEM;
}
/*
* Add 4 because we limit ourselves to EQs 0 ... 31,
* so we only need the low word of the register.
*/
if (mthca_map_reg(dev, ((pci_resource_len(dev->pdev, 0) - 1) &
dev->fw.arbel.eq_arm_base) + 4, 4,
&dev->eq_regs.arbel.eq_arm)) {
[PATCH] IB/mthca: Clean up error messages - Fix incorrect cut-n-paste in error messages. - Add missing newlines in error messages. - Use DRV_NAME instead of "ib_mthca" in a couple of places. Signed-off-by: Roland Dreier <roland@topspin.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-28 05:36:39 +08:00
mthca_err(dev, "Couldn't map EQ arm register, aborting.\n");
IB/mthca: Use pci_request_regions() Back in prehistoric (pre-git!) days, the kernel's MSI-X support did request_mem_region() on a device's MSI-X tables, which meant that a driver that enabled MSI-X couldn't use pci_request_regions() (since that would clash with the PCI layer's MSI-X request). However, that was removed (by me!) years ago, so mthca can just use pci_request_regions() and pci_release_regions() instead of its own much more complicated code that avoids requesting the MSI-X tables. Signed-off-by: Roland Dreier <rolandd@cisco.com>
2008-09-30 12:37:33 +08:00
iounmap(dev->clr_base);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
return -ENOMEM;
}
if (mthca_map_reg(dev, (pci_resource_len(dev->pdev, 0) - 1) &
dev->fw.arbel.eq_set_ci_base,
MTHCA_EQ_SET_CI_SIZE,
&dev->eq_regs.arbel.eq_set_ci_base)) {
[PATCH] IB/mthca: Clean up error messages - Fix incorrect cut-n-paste in error messages. - Add missing newlines in error messages. - Use DRV_NAME instead of "ib_mthca" in a couple of places. Signed-off-by: Roland Dreier <roland@topspin.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-28 05:36:39 +08:00
mthca_err(dev, "Couldn't map EQ CI register, aborting.\n");
IB/mthca: Use pci_request_regions() Back in prehistoric (pre-git!) days, the kernel's MSI-X support did request_mem_region() on a device's MSI-X tables, which meant that a driver that enabled MSI-X couldn't use pci_request_regions() (since that would clash with the PCI layer's MSI-X request). However, that was removed (by me!) years ago, so mthca can just use pci_request_regions() and pci_release_regions() instead of its own much more complicated code that avoids requesting the MSI-X tables. Signed-off-by: Roland Dreier <rolandd@cisco.com>
2008-09-30 12:37:33 +08:00
iounmap(dev->eq_regs.arbel.eq_arm);
iounmap(dev->clr_base);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
return -ENOMEM;
}
} else {
if (mthca_map_reg(dev, MTHCA_CLR_INT_BASE, MTHCA_CLR_INT_SIZE,
&dev->clr_base)) {
mthca_err(dev, "Couldn't map interrupt clear register, "
"aborting.\n");
return -ENOMEM;
}
if (mthca_map_reg(dev, MTHCA_ECR_BASE,
MTHCA_ECR_SIZE + MTHCA_ECR_CLR_SIZE,
&dev->eq_regs.tavor.ecr_base)) {
mthca_err(dev, "Couldn't map ecr register, "
"aborting.\n");
IB/mthca: Use pci_request_regions() Back in prehistoric (pre-git!) days, the kernel's MSI-X support did request_mem_region() on a device's MSI-X tables, which meant that a driver that enabled MSI-X couldn't use pci_request_regions() (since that would clash with the PCI layer's MSI-X request). However, that was removed (by me!) years ago, so mthca can just use pci_request_regions() and pci_release_regions() instead of its own much more complicated code that avoids requesting the MSI-X tables. Signed-off-by: Roland Dreier <rolandd@cisco.com>
2008-09-30 12:37:33 +08:00
iounmap(dev->clr_base);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
return -ENOMEM;
}
}
return 0;
}
IB/mthca: Fix section mismatch problems Quite a few cleanup functions in mthca were marked as __devexit. However, they could also be called from error paths during initialization, so they cannot be marked that way. Just delete all of the incorrect annotations. Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-03-30 01:36:46 +08:00
static void mthca_unmap_eq_regs(struct mthca_dev *dev)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
{
[PATCH] IB/mthca: encapsulate mem-free check into mthca_is_memfree() Clean up mem-free mode support by introducing mthca_is_memfree() function, which encapsulates the logic of deciding if a device is mem-free. Signed-off-by: Roland Dreier <roland@topspin.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-04-17 06:26:32 +08:00
if (mthca_is_memfree(dev)) {
IB/mthca: Use pci_request_regions() Back in prehistoric (pre-git!) days, the kernel's MSI-X support did request_mem_region() on a device's MSI-X tables, which meant that a driver that enabled MSI-X couldn't use pci_request_regions() (since that would clash with the PCI layer's MSI-X request). However, that was removed (by me!) years ago, so mthca can just use pci_request_regions() and pci_release_regions() instead of its own much more complicated code that avoids requesting the MSI-X tables. Signed-off-by: Roland Dreier <rolandd@cisco.com>
2008-09-30 12:37:33 +08:00
iounmap(dev->eq_regs.arbel.eq_set_ci_base);
iounmap(dev->eq_regs.arbel.eq_arm);
iounmap(dev->clr_base);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
} else {
IB/mthca: Use pci_request_regions() Back in prehistoric (pre-git!) days, the kernel's MSI-X support did request_mem_region() on a device's MSI-X tables, which meant that a driver that enabled MSI-X couldn't use pci_request_regions() (since that would clash with the PCI layer's MSI-X request). However, that was removed (by me!) years ago, so mthca can just use pci_request_regions() and pci_release_regions() instead of its own much more complicated code that avoids requesting the MSI-X tables. Signed-off-by: Roland Dreier <rolandd@cisco.com>
2008-09-30 12:37:33 +08:00
iounmap(dev->eq_regs.tavor.ecr_base);
iounmap(dev->clr_base);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
}
IB/mthca: Fix section mismatches Commit b3b30f5e ("IB/mthca: Recover from catastrophic errors") introduced some section mismatch breakage, because the error recovery code tears down and reinitializes the device, which calls into lots of code originally marked __devinit and __devexit from regular .text. Fix this by getting rid of these now-incorrect section markers. Reported by Randy Dunlap <randy.dunlap@oracle.com>. Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-11-30 07:33:06 +08:00
int mthca_map_eq_icm(struct mthca_dev *dev, u64 icm_virt)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
{
int ret;
u8 status;
/*
* We assume that mapping one page is enough for the whole EQ
* context table. This is fine with all current HCAs, because
* we only use 32 EQs and each EQ uses 32 bytes of context
* memory, or 1 KB total.
*/
dev->eq_table.icm_virt = icm_virt;
dev->eq_table.icm_page = alloc_page(GFP_HIGHUSER);
if (!dev->eq_table.icm_page)
return -ENOMEM;
dev->eq_table.icm_dma = pci_map_page(dev->pdev, dev->eq_table.icm_page, 0,
PAGE_SIZE, PCI_DMA_BIDIRECTIONAL);
dma-mapping: add the device argument to dma_mapping_error() Add per-device dma_mapping_ops support for CONFIG_X86_64 as POWER architecture does: This enables us to cleanly fix the Calgary IOMMU issue that some devices are not behind the IOMMU (http://lkml.org/lkml/2008/5/8/423). I think that per-device dma_mapping_ops support would be also helpful for KVM people to support PCI passthrough but Andi thinks that this makes it difficult to support the PCI passthrough (see the above thread). So I CC'ed this to KVM camp. Comments are appreciated. A pointer to dma_mapping_ops to struct dev_archdata is added. If the pointer is non NULL, DMA operations in asm/dma-mapping.h use it. If it's NULL, the system-wide dma_ops pointer is used as before. If it's useful for KVM people, I plan to implement a mechanism to register a hook called when a new pci (or dma capable) device is created (it works with hot plugging). It enables IOMMUs to set up an appropriate dma_mapping_ops per device. The major obstacle is that dma_mapping_error doesn't take a pointer to the device unlike other DMA operations. So x86 can't have dma_mapping_ops per device. Note all the POWER IOMMUs use the same dma_mapping_error function so this is not a problem for POWER but x86 IOMMUs use different dma_mapping_error functions. The first patch adds the device argument to dma_mapping_error. The patch is trivial but large since it touches lots of drivers and dma-mapping.h in all the architecture. This patch: dma_mapping_error() doesn't take a pointer to the device unlike other DMA operations. So we can't have dma_mapping_ops per device. Note that POWER already has dma_mapping_ops per device but all the POWER IOMMUs use the same dma_mapping_error function. x86 IOMMUs use device argument. [akpm@linux-foundation.org: fix sge] [akpm@linux-foundation.org: fix svc_rdma] [akpm@linux-foundation.org: build fix] [akpm@linux-foundation.org: fix bnx2x] [akpm@linux-foundation.org: fix s2io] [akpm@linux-foundation.org: fix pasemi_mac] [akpm@linux-foundation.org: fix sdhci] [akpm@linux-foundation.org: build fix] [akpm@linux-foundation.org: fix sparc] [akpm@linux-foundation.org: fix ibmvscsi] Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp> Cc: Muli Ben-Yehuda <muli@il.ibm.com> Cc: Andi Kleen <andi@firstfloor.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Ingo Molnar <mingo@elte.hu> Cc: Avi Kivity <avi@qumranet.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-26 10:44:49 +08:00
if (pci_dma_mapping_error(dev->pdev, dev->eq_table.icm_dma)) {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
__free_page(dev->eq_table.icm_page);
return -ENOMEM;
}
ret = mthca_MAP_ICM_page(dev, dev->eq_table.icm_dma, icm_virt, &status);
if (!ret && status)
ret = -EINVAL;
if (ret) {
pci_unmap_page(dev->pdev, dev->eq_table.icm_dma, PAGE_SIZE,
PCI_DMA_BIDIRECTIONAL);
__free_page(dev->eq_table.icm_page);
}
return ret;
}
IB/mthca: Fix section mismatch problems Quite a few cleanup functions in mthca were marked as __devexit. However, they could also be called from error paths during initialization, so they cannot be marked that way. Just delete all of the incorrect annotations. Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-03-30 01:36:46 +08:00
void mthca_unmap_eq_icm(struct mthca_dev *dev)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
{
u8 status;
IB/mthca: Use an enum for HCA page size Use a named enum for the HCA's internal page size, rather than having magic values of 4096 and shifts by 12 all over the code. Also, fix one minor bug in EQ handling: only one HCA page is mapped to the HCA during initialization, but a full kernel page is unmapped during cleanup. This might cause problems when PAGE_SIZE != 4096. Signed-off-by: Ishai Rabinovitz <ishai@mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-03-02 14:33:11 +08:00
mthca_UNMAP_ICM(dev, dev->eq_table.icm_virt, 1, &status);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
pci_unmap_page(dev->pdev, dev->eq_table.icm_dma, PAGE_SIZE,
PCI_DMA_BIDIRECTIONAL);
__free_page(dev->eq_table.icm_page);
}
IB/mthca: Fix section mismatches Commit b3b30f5e ("IB/mthca: Recover from catastrophic errors") introduced some section mismatch breakage, because the error recovery code tears down and reinitializes the device, which calls into lots of code originally marked __devinit and __devexit from regular .text. Fix this by getting rid of these now-incorrect section markers. Reported by Randy Dunlap <randy.dunlap@oracle.com>. Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-11-30 07:33:06 +08:00
int mthca_init_eq_table(struct mthca_dev *dev)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
{
int err;
u8 status;
u8 intr;
int i;
err = mthca_alloc_init(&dev->eq_table.alloc,
dev->limits.num_eqs,
dev->limits.num_eqs - 1,
dev->limits.reserved_eqs);
if (err)
return err;
err = mthca_map_eq_regs(dev);
if (err)
goto err_out_free;
IB/mthca: Remove MSI support as scheduled Remove MSI support from the mthca driver, as scheduled. There is no reason to use MSI instead of MSI-X, since MSI-X performs better. No one has spoken up since MSI support was deprecated in commit f6be6fbe ("IB/mthca: Schedule MSI support for removal"), so apparently the MSI support is unused. Signed-off-by: Adrian Bunk <bunk@kernel.org> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2008-01-01 21:47:10 +08:00
if (dev->mthca_flags & MTHCA_FLAG_MSI_X) {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
dev->eq_table.clr_mask = 0;
} else {
dev->eq_table.clr_mask =
swab32(1 << (dev->eq_table.inta_pin & 31));
dev->eq_table.clr_int = dev->clr_base +
[IB] mthca: fix off by one in clr_int calculation We should use the first word of the clear interrupt register if the bit we're after is < 32, not < 31. Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-09-27 00:29:33 +08:00
(dev->eq_table.inta_pin < 32 ? 4 : 0);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
dev->eq_table.arm_mask = 0;
IB/mthca: Remove MSI support as scheduled Remove MSI support from the mthca driver, as scheduled. There is no reason to use MSI instead of MSI-X, since MSI-X performs better. No one has spoken up since MSI support was deprecated in commit f6be6fbe ("IB/mthca: Schedule MSI support for removal"), so apparently the MSI support is unused. Signed-off-by: Adrian Bunk <bunk@kernel.org> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2008-01-01 21:47:10 +08:00
intr = dev->eq_table.inta_pin;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
IB/mthca: prevent event queue overrun I am seeing EQ overruns in SDP stress tests: if the CQ completion handler arms a CQ, this could generate more EQEs, so that EQ will never get empty and consumer index will never get updated. This is similiar to what we have with command interface: /* * cmd_event() may add more commands. * The card will think the queue has overflowed if * we don't tell it we've been processing events. */ However, for completion events, we *don't* want to update the consumer index on each event. So, perform EQ doorbell coalescing: allocate EQs with some spare EQEs, and update once we run out of them. The value 0x80 was selected to avoid any performance impact. Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-01-10 06:04:40 +08:00
err = mthca_create_eq(dev, dev->limits.num_cqs + MTHCA_NUM_SPARE_EQE,
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
(dev->mthca_flags & MTHCA_FLAG_MSI_X) ? 128 : intr,
&dev->eq_table.eq[MTHCA_EQ_COMP]);
if (err)
goto err_out_unmap;
IB/mthca: prevent event queue overrun I am seeing EQ overruns in SDP stress tests: if the CQ completion handler arms a CQ, this could generate more EQEs, so that EQ will never get empty and consumer index will never get updated. This is similiar to what we have with command interface: /* * cmd_event() may add more commands. * The card will think the queue has overflowed if * we don't tell it we've been processing events. */ However, for completion events, we *don't* want to update the consumer index on each event. So, perform EQ doorbell coalescing: allocate EQs with some spare EQEs, and update once we run out of them. The value 0x80 was selected to avoid any performance impact. Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-01-10 06:04:40 +08:00
err = mthca_create_eq(dev, MTHCA_NUM_ASYNC_EQE + MTHCA_NUM_SPARE_EQE,
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
(dev->mthca_flags & MTHCA_FLAG_MSI_X) ? 129 : intr,
&dev->eq_table.eq[MTHCA_EQ_ASYNC]);
if (err)
goto err_out_comp;
IB/mthca: prevent event queue overrun I am seeing EQ overruns in SDP stress tests: if the CQ completion handler arms a CQ, this could generate more EQEs, so that EQ will never get empty and consumer index will never get updated. This is similiar to what we have with command interface: /* * cmd_event() may add more commands. * The card will think the queue has overflowed if * we don't tell it we've been processing events. */ However, for completion events, we *don't* want to update the consumer index on each event. So, perform EQ doorbell coalescing: allocate EQs with some spare EQEs, and update once we run out of them. The value 0x80 was selected to avoid any performance impact. Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-01-10 06:04:40 +08:00
err = mthca_create_eq(dev, MTHCA_NUM_CMD_EQE + MTHCA_NUM_SPARE_EQE,
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
(dev->mthca_flags & MTHCA_FLAG_MSI_X) ? 130 : intr,
&dev->eq_table.eq[MTHCA_EQ_CMD]);
if (err)
goto err_out_async;
if (dev->mthca_flags & MTHCA_FLAG_MSI_X) {
static const char *eq_name[] = {
[MTHCA_EQ_COMP] = DRV_NAME " (comp)",
[MTHCA_EQ_ASYNC] = DRV_NAME " (async)",
[MTHCA_EQ_CMD] = DRV_NAME " (cmd)"
};
for (i = 0; i < MTHCA_NUM_EQ; ++i) {
err = request_irq(dev->eq_table.eq[i].msi_x_vector,
[PATCH] IB/mthca: encapsulate mem-free check into mthca_is_memfree() Clean up mem-free mode support by introducing mthca_is_memfree() function, which encapsulates the logic of deciding if a device is mem-free. Signed-off-by: Roland Dreier <roland@topspin.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-04-17 06:26:32 +08:00
mthca_is_memfree(dev) ?
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
mthca_arbel_msi_x_interrupt :
mthca_tavor_msi_x_interrupt,
0, eq_name[i], dev->eq_table.eq + i);
if (err)
goto err_out_cmd;
dev->eq_table.eq[i].have_irq = 1;
}
} else {
err = request_irq(dev->pdev->irq,
[PATCH] IB/mthca: encapsulate mem-free check into mthca_is_memfree() Clean up mem-free mode support by introducing mthca_is_memfree() function, which encapsulates the logic of deciding if a device is mem-free. Signed-off-by: Roland Dreier <roland@topspin.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-04-17 06:26:32 +08:00
mthca_is_memfree(dev) ?
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
mthca_arbel_interrupt :
mthca_tavor_interrupt,
[PATCH] irq-flags: misc drivers: Use the new IRQF_ constants Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Cc: Ingo Molnar <mingo@elte.hu> Cc: "David S. Miller" <davem@davemloft.net> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-07-02 10:29:38 +08:00
IRQF_SHARED, DRV_NAME, dev);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
if (err)
goto err_out_cmd;
dev->eq_table.have_irq = 1;
}
err = mthca_MAP_EQ(dev, async_mask(dev),
0, dev->eq_table.eq[MTHCA_EQ_ASYNC].eqn, &status);
if (err)
mthca_warn(dev, "MAP_EQ for async EQ %d failed (%d)\n",
dev->eq_table.eq[MTHCA_EQ_ASYNC].eqn, err);
if (status)
mthca_warn(dev, "MAP_EQ for async EQ %d returned status 0x%02x\n",
dev->eq_table.eq[MTHCA_EQ_ASYNC].eqn, status);
err = mthca_MAP_EQ(dev, MTHCA_CMD_EVENT_MASK,
0, dev->eq_table.eq[MTHCA_EQ_CMD].eqn, &status);
if (err)
mthca_warn(dev, "MAP_EQ for cmd EQ %d failed (%d)\n",
dev->eq_table.eq[MTHCA_EQ_CMD].eqn, err);
if (status)
mthca_warn(dev, "MAP_EQ for cmd EQ %d returned status 0x%02x\n",
dev->eq_table.eq[MTHCA_EQ_CMD].eqn, status);
IB/mthca: Coverity fix to mthca_init_eq_table() Fix bug found by coverity: the loop body never executed, because it was doing for (i = 0; i < MTHCA_EQ_CMD; ++i), but MTHCA_EQ_CMD is 0. The correct loop bound is MTHCA_NUM_EQ, to loop over all EQs. Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-03-21 02:08:25 +08:00
for (i = 0; i < MTHCA_NUM_EQ; ++i)
[PATCH] IB/mthca: encapsulate mem-free check into mthca_is_memfree() Clean up mem-free mode support by introducing mthca_is_memfree() function, which encapsulates the logic of deciding if a device is mem-free. Signed-off-by: Roland Dreier <roland@topspin.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-04-17 06:26:32 +08:00
if (mthca_is_memfree(dev))
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
arbel_eq_req_not(dev, dev->eq_table.eq[i].eqn_mask);
else
tavor_eq_req_not(dev, dev->eq_table.eq[i].eqn);
return 0;
err_out_cmd:
mthca_free_irqs(dev);
mthca_free_eq(dev, &dev->eq_table.eq[MTHCA_EQ_CMD]);
err_out_async:
mthca_free_eq(dev, &dev->eq_table.eq[MTHCA_EQ_ASYNC]);
err_out_comp:
mthca_free_eq(dev, &dev->eq_table.eq[MTHCA_EQ_COMP]);
err_out_unmap:
mthca_unmap_eq_regs(dev);
err_out_free:
mthca_alloc_cleanup(&dev->eq_table.alloc);
return err;
}
IB/mthca: Fix section mismatch problems Quite a few cleanup functions in mthca were marked as __devexit. However, they could also be called from error paths during initialization, so they cannot be marked that way. Just delete all of the incorrect annotations. Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-03-30 01:36:46 +08:00
void mthca_cleanup_eq_table(struct mthca_dev *dev)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
{
u8 status;
int i;
mthca_free_irqs(dev);
mthca_MAP_EQ(dev, async_mask(dev),
1, dev->eq_table.eq[MTHCA_EQ_ASYNC].eqn, &status);
mthca_MAP_EQ(dev, MTHCA_CMD_EVENT_MASK,
1, dev->eq_table.eq[MTHCA_EQ_CMD].eqn, &status);
for (i = 0; i < MTHCA_NUM_EQ; ++i)
mthca_free_eq(dev, &dev->eq_table.eq[i]);
mthca_unmap_eq_regs(dev);
mthca_alloc_cleanup(&dev->eq_table.alloc);
}