openkylin
/
linux
mirror of https://gitee.com/openkylin/linux.git
9
0
Fork 0
Code Issues Projects Releases Wiki Activity
linux/drivers/infiniband/core/device.c

1020 lines
27 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 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 Sun Microsystems, Inc. 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/module.h>
#include <linux/string.h>
#include <linux/errno.h>
IB/core: Use ARRAY_SIZE macro for mandatory_table Use ARRAY_SIZE() macro already defined in kernel.h instead of open coding equivalent code. Signed-off-by: Ahmed S. Darwish <darwish.07@gmail.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-02-07 00:07:25 +08:00
#include <linux/kernel.h>
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
#include <linux/slab.h>
#include <linux/init.h>
IB: convert from semaphores to mutexes semaphore to mutex conversion by Ingo and Arjan's script. Signed-off-by: Ingo Molnar <mingo@elte.hu> [ Sanity-checked on real IB hardware ] Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-01-14 06:51:39 +08:00
#include <linux/mutex.h>
IB/core: Find the network device matching connection parameters In the case of IPoIB, and maybe in other cases, the network device is managed by an upper-layer protocol (ULP). In order to expose this network device to other users of the IB device, let ULPs implement a callback that returns network device according to connection parameters. The IB device and port, together with the P_Key and the GID should be enough to uniquely identify the ULP net device. However, in current kernels there can be multiple IPoIB interfaces created with the same GID. Furthermore, such configuration may be desireable to support ipvlan-like configurations for RDMA CM with IPoIB. To resolve the device in these cases the code will also take the IP address as an additional input. Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Signed-off-by: Haggai Eran <haggaie@mellanox.com> Signed-off-by: Yotam Kenneth <yotamke@mellanox.com> Signed-off-by: Shachar Raindel <raindel@mellanox.com> Signed-off-by: Guy Shapiro <guysh@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-07-30 22:50:15 +08:00
#include <linux/netdevice.h>
RDMA: Add netlink infrastructure Add basic RDMA netlink infrastructure that allows for registration of RDMA clients for which data is to be exported and supplies message construction callbacks. Signed-off-by: Nir Muchtar <nirm@voltaire.com> [ Reorganize a few things, add CONFIG_NET dependency. - Roland ] Signed-off-by: Roland Dreier <roland@purestorage.com>
2011-05-21 02:46:11 +08:00
#include <rdma/rdma_netlink.h>
IB/core: Add RoCE GID table management RoCE GIDs are based on IP addresses configured on Ethernet net-devices which relate to the RDMA (RoCE) device port. Currently, each of the low-level drivers that support RoCE (ocrdma, mlx4) manages its own RoCE port GID table. As there's nothing which is essentially vendor specific, we generalize that, and enhance the RDMA core GID cache to do this job. In order to populate the GID table, we listen for events: (a) netdev up/down/change_addr events - if a netdev is built onto our RoCE device, we need to add/delete its IPs. This involves adding all GIDs related to this ndev, add default GIDs, etc. (b) inet events - add new GIDs (according to the IP addresses) to the table. For programming the port RoCE GID table, providers must implement the add_gid and del_gid callbacks. RoCE GID management requires us to state the associated net_device alongside the GID. This information is necessary in order to manage the GID table. For example, when a net_device is removed, its associated GIDs need to be removed as well. RoCE mandates generating a default GID for each port, based on the related net-device's IPv6 link local. In contrast to the GID based on the regular IPv6 link-local (as we generate GID per IP address), the default GID is also available when the net device is down (in order to support loopback). Locking is done as follows: The patch modify the GID table code both for new RoCE drivers implementing the add_gid/del_gid callbacks and for current RoCE and IB drivers that do not. The flows for updating the table are different, so the locking requirements are too. While updating RoCE GID table, protection against multiple writers is achieved via mutex_lock(&table->lock). Since writing to a table requires us to find an entry (possible a free entry) in the table and then modify it, this mutex protects both the find_gid and write_gid ensuring the atomicity of the action. Each entry in the GID cache is protected by rwlock. In RoCE, writing (usually results from netdev notifier) involves invoking the vendor's add_gid and del_gid callbacks, which could sleep. Therefore, an invalid flag is added for each entry. Updates for RoCE are done via a workqueue, thus sleeping is permitted. In IB, updates are done in write_lock_irq(&device->cache.lock), thus write_gid isn't allowed to sleep and add_gid/del_gid are not called. When passing net-device into/out-of the GID cache, the device is always passed held (dev_hold). The code uses a single work item for updating all RDMA devices, following a netdev or inet notifier. The patch moves the cache from being a client (which was incorrect, as the cache is part of the IB infrastructure) to being explicitly initialized/freed when a device is registered/removed. Signed-off-by: Matan Barak <matanb@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-07-30 23:33:26 +08:00
#include <rdma/ib_addr.h>
#include <rdma/ib_cache.h>
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
#include "core_priv.h"
MODULE_AUTHOR("Roland Dreier");
MODULE_DESCRIPTION("core kernel InfiniBand API");
MODULE_LICENSE("Dual BSD/GPL");
struct ib_client_data {
struct list_head list;
struct ib_client *client;
void * data;
IB/core: lock client data with lists_rwsem An ib_client callback that is called with the lists_rwsem locked only for read is protected from changes to the IB client lists, but not from ib_unregister_device() freeing its client data. This is because ib_unregister_device() will remove the device from the device list with lists_rwsem locked for write, but perform the rest of the cleanup, including the call to remove() without that lock. Mark client data that is undergoing de-registration with a new going_down flag in the client data context. Lock the client data list with lists_rwsem for write in addition to using the spinlock, so that functions calling the callback would be able to lock only lists_rwsem for read and let callbacks sleep. Since ib_unregister_client() now marks the client data context, no need for remove() to search the context again, so pass the client data directly to remove() callbacks. Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Signed-off-by: Haggai Eran <haggaie@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-07-30 22:50:14 +08:00
/* The device or client is going down. Do not call client or device
* callbacks other than remove(). */
bool going_down;
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: add a proper completion queue abstraction This adds an abstraction that allows ULPs to simply pass a completion object and completion callback with each submitted WR and let the RDMA core handle the nitty gritty details of how to handle completion interrupts and poll the CQ. In detail there is a new ib_cqe structure which just contains the completion callback, and which can be used to get at the containing object using container_of. It is pointed to by the WR and WC as an alternative to the wr_id field, similar to how many ULPs already use the field to store a pointer using casts. A driver using the new completion callbacks allocates it's CQs using the new ib_create_cq API, which in addition to the number of CQEs and the completion vectors also takes a mode on how we poll for CQEs. Three modes are available: direct for drivers that never take CQ interrupts and just poll for them, softirq to poll from softirq context using the to be renamed blk-iopoll infrastructure which takes care of rearming and budgeting, or a workqueue for consumer who want to be called from user context. Thanks a lot to Sagi Grimberg who helped reviewing the API, wrote the current version of the workqueue code because my two previous attempts sucked too much and converted the iSER initiator to the new API. Signed-off-by: Christoph Hellwig <hch@lst.de>
2015-12-12 03:53:03 +08:00
struct workqueue_struct *ib_comp_wq;
RDMA: Update workqueue usage * ib_wq is added, which is used as the common workqueue for infiniband instead of the system workqueue. All system workqueue usages including flush_scheduled_work() callers are converted to use and flush ib_wq. * cancel_delayed_work() + flush_scheduled_work() converted to cancel_delayed_work_sync(). * qib_wq is removed and ib_wq is used instead. This is to prepare for deprecation of flush_scheduled_work(). Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2010-10-19 23:24:36 +08:00
struct workqueue_struct *ib_wq;
EXPORT_SYMBOL_GPL(ib_wq);
IB/core: Add rwsem to allow reading device list or client list Currently the RDMA subsystem's device list and client list are protected by a single mutex. This prevents adding user-facing APIs that iterate these lists, since using them may cause a deadlock. The patch attempts to solve this problem by adding a read-write semaphore to protect the lists. Readers now don't need the mutex, and are safe just by read-locking the semaphore. The ib_register_device, ib_register_client, ib_unregister_device, and ib_unregister_client functions are modified to lock the semaphore for write during their respective list modification. Also, in order to make sure client callbacks are called only between add() and remove() calls, the code is changed to only add items to the lists after the add() calls and remove from the lists before the remove() calls. This patch attempts to solve a similar need [1] that was seen in the RoCE v2 patch series. [1] http://www.spinics.net/lists/linux-rdma/msg24733.html Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Cc: Matan Barak <matanb@mellanox.com> Signed-off-by: Haggai Eran <haggaie@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-07-30 22:50:13 +08:00
/* The device_list and client_list contain devices and clients after their
* registration has completed, and the devices and clients are removed
* during unregistration. */
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 LIST_HEAD(device_list);
static LIST_HEAD(client_list);
/*
IB/core: Add rwsem to allow reading device list or client list Currently the RDMA subsystem's device list and client list are protected by a single mutex. This prevents adding user-facing APIs that iterate these lists, since using them may cause a deadlock. The patch attempts to solve this problem by adding a read-write semaphore to protect the lists. Readers now don't need the mutex, and are safe just by read-locking the semaphore. The ib_register_device, ib_register_client, ib_unregister_device, and ib_unregister_client functions are modified to lock the semaphore for write during their respective list modification. Also, in order to make sure client callbacks are called only between add() and remove() calls, the code is changed to only add items to the lists after the add() calls and remove from the lists before the remove() calls. This patch attempts to solve a similar need [1] that was seen in the RoCE v2 patch series. [1] http://www.spinics.net/lists/linux-rdma/msg24733.html Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Cc: Matan Barak <matanb@mellanox.com> Signed-off-by: Haggai Eran <haggaie@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-07-30 22:50:13 +08:00
* device_mutex and lists_rwsem protect access to both device_list and
* client_list. device_mutex protects writer access by device and client
* registration / de-registration. lists_rwsem protects reader access to
* these lists. Iterators of these lists must lock it for read, while updates
* to the lists must be done with a write lock. A special case is when the
* device_mutex is locked. In this case locking the lists for read access is
* not necessary as the device_mutex implies it.
IB/core: lock client data with lists_rwsem An ib_client callback that is called with the lists_rwsem locked only for read is protected from changes to the IB client lists, but not from ib_unregister_device() freeing its client data. This is because ib_unregister_device() will remove the device from the device list with lists_rwsem locked for write, but perform the rest of the cleanup, including the call to remove() without that lock. Mark client data that is undergoing de-registration with a new going_down flag in the client data context. Lock the client data list with lists_rwsem for write in addition to using the spinlock, so that functions calling the callback would be able to lock only lists_rwsem for read and let callbacks sleep. Since ib_unregister_client() now marks the client data context, no need for remove() to search the context again, so pass the client data directly to remove() callbacks. Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Signed-off-by: Haggai Eran <haggaie@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-07-30 22:50:14 +08:00
*
* lists_rwsem also protects access to the client data list.
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: convert from semaphores to mutexes semaphore to mutex conversion by Ingo and Arjan's script. Signed-off-by: Ingo Molnar <mingo@elte.hu> [ Sanity-checked on real IB hardware ] Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-01-14 06:51:39 +08:00
static DEFINE_MUTEX(device_mutex);
IB/core: Add rwsem to allow reading device list or client list Currently the RDMA subsystem's device list and client list are protected by a single mutex. This prevents adding user-facing APIs that iterate these lists, since using them may cause a deadlock. The patch attempts to solve this problem by adding a read-write semaphore to protect the lists. Readers now don't need the mutex, and are safe just by read-locking the semaphore. The ib_register_device, ib_register_client, ib_unregister_device, and ib_unregister_client functions are modified to lock the semaphore for write during their respective list modification. Also, in order to make sure client callbacks are called only between add() and remove() calls, the code is changed to only add items to the lists after the add() calls and remove from the lists before the remove() calls. This patch attempts to solve a similar need [1] that was seen in the RoCE v2 patch series. [1] http://www.spinics.net/lists/linux-rdma/msg24733.html Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Cc: Matan Barak <matanb@mellanox.com> Signed-off-by: Haggai Eran <haggaie@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-07-30 22:50:13 +08:00
static DECLARE_RWSEM(lists_rwsem);
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 int ib_device_check_mandatory(struct ib_device *device)
{
#define IB_MANDATORY_FUNC(x) { offsetof(struct ib_device, x), #x }
static const struct {
size_t offset;
char *name;
} mandatory_table[] = {
IB_MANDATORY_FUNC(query_device),
IB_MANDATORY_FUNC(query_port),
IB_MANDATORY_FUNC(query_pkey),
IB_MANDATORY_FUNC(query_gid),
IB_MANDATORY_FUNC(alloc_pd),
IB_MANDATORY_FUNC(dealloc_pd),
IB_MANDATORY_FUNC(create_ah),
IB_MANDATORY_FUNC(destroy_ah),
IB_MANDATORY_FUNC(create_qp),
IB_MANDATORY_FUNC(modify_qp),
IB_MANDATORY_FUNC(destroy_qp),
IB_MANDATORY_FUNC(post_send),
IB_MANDATORY_FUNC(post_recv),
IB_MANDATORY_FUNC(create_cq),
IB_MANDATORY_FUNC(destroy_cq),
IB_MANDATORY_FUNC(poll_cq),
IB_MANDATORY_FUNC(req_notify_cq),
IB_MANDATORY_FUNC(get_dma_mr),
IB/core: Add per port immutable struct to ib_device As of commit 5eb620c81ce3 "IB/core: Add helpers for uncached GID and P_Key searches"; pkey_tbl_len and gid_tbl_len are immutable data which are stored in the ib_device. The per port core capability flags to be added later are also immutable data to be stored in the ib_device object. In preparation for this create a structure for per port immutable data and place the pkey and gid table lengths within this structure. "get_port_immutable" is added as a mandatory device function to allow the drivers to fill in this data. Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-05-14 08:02:58 +08:00
IB_MANDATORY_FUNC(dereg_mr),
IB_MANDATORY_FUNC(get_port_immutable)
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 i;
IB/core: Use ARRAY_SIZE macro for mandatory_table Use ARRAY_SIZE() macro already defined in kernel.h instead of open coding equivalent code. Signed-off-by: Ahmed S. Darwish <darwish.07@gmail.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-02-07 00:07:25 +08:00
for (i = 0; i < ARRAY_SIZE(mandatory_table); ++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
if (!*(void **) ((void *) device + mandatory_table[i].offset)) {
IB/core: trivial prink cleanup. 1. Replaced printk with appropriate pr_warn, pr_err, pr_info. 2. Removed unnecessary prints around memory allocation failure which are not required, as reported by the checkpatch script. Signed-off-by: Parav Pandit <pandit.parav@gmail.com> Reviewed-by: Haggai Eran <haggaie@mellanox.com> Reviewed-by: Sagi Grimberg <sagig@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-03-02 03:20:29 +08:00
pr_warn("Device %s is missing mandatory function %s\n",
device->name, mandatory_table[i].name);
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 -EINVAL;
}
}
return 0;
}
static struct ib_device *__ib_device_get_by_name(const char *name)
{
struct ib_device *device;
list_for_each_entry(device, &device_list, core_list)
if (!strncmp(name, device->name, IB_DEVICE_NAME_MAX))
return device;
return NULL;
}
static int alloc_name(char *name)
{
IB: find_first_zero_bit() takes unsigned pointer Fix sparse warning drivers/infiniband/core/device.c:142:6: warning: incorrect type in argument 1 (different signedness) drivers/infiniband/core/device.c:142:6: expected unsigned long const *addr drivers/infiniband/core/device.c:142:6: got long *[assigned] inuse by making the local variable inuse unsigned. Does not affect generated code at all. Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-10-10 10:59:04 +08:00
unsigned long *inuse;
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
char buf[IB_DEVICE_NAME_MAX];
struct ib_device *device;
int i;
IB: find_first_zero_bit() takes unsigned pointer Fix sparse warning drivers/infiniband/core/device.c:142:6: warning: incorrect type in argument 1 (different signedness) drivers/infiniband/core/device.c:142:6: expected unsigned long const *addr drivers/infiniband/core/device.c:142:6: got long *[assigned] inuse by making the local variable inuse unsigned. Does not affect generated code at all. Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-10-10 10:59:04 +08:00
inuse = (unsigned long *) get_zeroed_page(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 (!inuse)
return -ENOMEM;
list_for_each_entry(device, &device_list, core_list) {
if (!sscanf(device->name, name, &i))
continue;
if (i < 0 || i >= PAGE_SIZE * 8)
continue;
snprintf(buf, sizeof buf, name, i);
if (!strncmp(buf, device->name, IB_DEVICE_NAME_MAX))
set_bit(i, inuse);
}
i = find_first_zero_bit(inuse, PAGE_SIZE * 8);
free_page((unsigned long) inuse);
snprintf(buf, sizeof buf, name, i);
if (__ib_device_get_by_name(buf))
return -ENFILE;
strlcpy(name, buf, IB_DEVICE_NAME_MAX);
return 0;
}
IB/core: Make ib_alloc_device init the kobject This gets rid of the weird in-between state where struct ib_device was allocated but the kobject didn't work. Consequently ib_device_release is now guaranteed to be called in all situations and we needn't duplicate its kfrees on error paths. Signed-off-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-05 05:23:34 +08:00
static void ib_device_release(struct device *device)
{
struct ib_device *dev = container_of(device, struct ib_device, dev);
IB/core: Add RoCE GID table management RoCE GIDs are based on IP addresses configured on Ethernet net-devices which relate to the RDMA (RoCE) device port. Currently, each of the low-level drivers that support RoCE (ocrdma, mlx4) manages its own RoCE port GID table. As there's nothing which is essentially vendor specific, we generalize that, and enhance the RDMA core GID cache to do this job. In order to populate the GID table, we listen for events: (a) netdev up/down/change_addr events - if a netdev is built onto our RoCE device, we need to add/delete its IPs. This involves adding all GIDs related to this ndev, add default GIDs, etc. (b) inet events - add new GIDs (according to the IP addresses) to the table. For programming the port RoCE GID table, providers must implement the add_gid and del_gid callbacks. RoCE GID management requires us to state the associated net_device alongside the GID. This information is necessary in order to manage the GID table. For example, when a net_device is removed, its associated GIDs need to be removed as well. RoCE mandates generating a default GID for each port, based on the related net-device's IPv6 link local. In contrast to the GID based on the regular IPv6 link-local (as we generate GID per IP address), the default GID is also available when the net device is down (in order to support loopback). Locking is done as follows: The patch modify the GID table code both for new RoCE drivers implementing the add_gid/del_gid callbacks and for current RoCE and IB drivers that do not. The flows for updating the table are different, so the locking requirements are too. While updating RoCE GID table, protection against multiple writers is achieved via mutex_lock(&table->lock). Since writing to a table requires us to find an entry (possible a free entry) in the table and then modify it, this mutex protects both the find_gid and write_gid ensuring the atomicity of the action. Each entry in the GID cache is protected by rwlock. In RoCE, writing (usually results from netdev notifier) involves invoking the vendor's add_gid and del_gid callbacks, which could sleep. Therefore, an invalid flag is added for each entry. Updates for RoCE are done via a workqueue, thus sleeping is permitted. In IB, updates are done in write_lock_irq(&device->cache.lock), thus write_gid isn't allowed to sleep and add_gid/del_gid are not called. When passing net-device into/out-of the GID cache, the device is always passed held (dev_hold). The code uses a single work item for updating all RDMA devices, following a netdev or inet notifier. The patch moves the cache from being a client (which was incorrect, as the cache is part of the IB infrastructure) to being explicitly initialized/freed when a device is registered/removed. Signed-off-by: Matan Barak <matanb@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-07-30 23:33:26 +08:00
ib_cache_release_one(dev);
IB/core: Make ib_alloc_device init the kobject This gets rid of the weird in-between state where struct ib_device was allocated but the kobject didn't work. Consequently ib_device_release is now guaranteed to be called in all situations and we needn't duplicate its kfrees on error paths. Signed-off-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-05 05:23:34 +08:00
kfree(dev->port_immutable);
kfree(dev);
}
static int ib_device_uevent(struct device *device,
struct kobj_uevent_env *env)
{
struct ib_device *dev = container_of(device, struct ib_device, dev);
if (add_uevent_var(env, "NAME=%s", dev->name))
return -ENOMEM;
/*
* It would be nice to pass the node GUID with the event...
*/
return 0;
}
static struct class ib_class = {
.name = "infiniband",
.dev_release = ib_device_release,
.dev_uevent = ib_device_uevent,
};
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_alloc_device - allocate an IB device struct
* @size:size of structure to allocate
*
* Low-level drivers should use ib_alloc_device() to allocate &struct
* ib_device. @size is the size of the structure to be allocated,
* including any private data used by the low-level driver.
* ib_dealloc_device() must be used to free structures allocated with
* ib_alloc_device().
*/
struct ib_device *ib_alloc_device(size_t size)
{
IB/core: Make ib_alloc_device init the kobject This gets rid of the weird in-between state where struct ib_device was allocated but the kobject didn't work. Consequently ib_device_release is now guaranteed to be called in all situations and we needn't duplicate its kfrees on error paths. Signed-off-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-05 05:23:34 +08:00
struct ib_device *device;
if (WARN_ON(size < sizeof(struct ib_device)))
return NULL;
device = kzalloc(size, GFP_KERNEL);
if (!device)
return NULL;
device->dev.class = &ib_class;
device_initialize(&device->dev);
dev_set_drvdata(&device->dev, device);
INIT_LIST_HEAD(&device->event_handler_list);
spin_lock_init(&device->event_handler_lock);
spin_lock_init(&device->client_data_lock);
INIT_LIST_HEAD(&device->client_data_list);
INIT_LIST_HEAD(&device->port_list);
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/core: Make ib_alloc_device init the kobject This gets rid of the weird in-between state where struct ib_device was allocated but the kobject didn't work. Consequently ib_device_release is now guaranteed to be called in all situations and we needn't duplicate its kfrees on error paths. Signed-off-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-05 05:23:34 +08:00
return device;
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
}
EXPORT_SYMBOL(ib_alloc_device);
/**
* ib_dealloc_device - free an IB device struct
* @device:structure to free
*
* Free a structure allocated with ib_alloc_device().
*/
void ib_dealloc_device(struct ib_device *device)
{
IB/core: Make ib_alloc_device init the kobject This gets rid of the weird in-between state where struct ib_device was allocated but the kobject didn't work. Consequently ib_device_release is now guaranteed to be called in all situations and we needn't duplicate its kfrees on error paths. Signed-off-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-05 05:23:34 +08:00
WARN_ON(device->reg_state != IB_DEV_UNREGISTERED &&
device->reg_state != IB_DEV_UNINITIALIZED);
IB: Remove sysfs files before unregistering device Move the ib_device_unregister_sysfs() call from ib_dealloc_device() to ib_unregister_device(). The old code allows device unregister to proceed even if some sysfs files are open, which leaves a window where userspace can open a file before a device is removed but then end up reading the file after the device is removed, which leads to various kernel crashes either because the device data structure is freed or because the low-level driver code is gone after module removal. By not returning from ib_unregister_device() until after all sysfs entries are removed, we make sure that data structures and/or module code is not freed until after all sysfs access is done. Reported-by: Jack Morgenstein <jackm@dev.mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2009-02-26 05:27:46 +08:00
kobject_put(&device->dev.kobj);
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
}
EXPORT_SYMBOL(ib_dealloc_device);
static int add_client_context(struct ib_device *device, struct ib_client *client)
{
struct ib_client_data *context;
unsigned long flags;
context = kmalloc(sizeof *context, GFP_KERNEL);
if (!context) {
IB/core: trivial prink cleanup. 1. Replaced printk with appropriate pr_warn, pr_err, pr_info. 2. Removed unnecessary prints around memory allocation failure which are not required, as reported by the checkpatch script. Signed-off-by: Parav Pandit <pandit.parav@gmail.com> Reviewed-by: Haggai Eran <haggaie@mellanox.com> Reviewed-by: Sagi Grimberg <sagig@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-03-02 03:20:29 +08:00
pr_warn("Couldn't allocate client context for %s/%s\n",
device->name, client->name);
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;
}
context->client = client;
context->data = NULL;
IB/core: lock client data with lists_rwsem An ib_client callback that is called with the lists_rwsem locked only for read is protected from changes to the IB client lists, but not from ib_unregister_device() freeing its client data. This is because ib_unregister_device() will remove the device from the device list with lists_rwsem locked for write, but perform the rest of the cleanup, including the call to remove() without that lock. Mark client data that is undergoing de-registration with a new going_down flag in the client data context. Lock the client data list with lists_rwsem for write in addition to using the spinlock, so that functions calling the callback would be able to lock only lists_rwsem for read and let callbacks sleep. Since ib_unregister_client() now marks the client data context, no need for remove() to search the context again, so pass the client data directly to remove() callbacks. Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Signed-off-by: Haggai Eran <haggaie@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-07-30 22:50:14 +08:00
context->going_down = false;
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/core: lock client data with lists_rwsem An ib_client callback that is called with the lists_rwsem locked only for read is protected from changes to the IB client lists, but not from ib_unregister_device() freeing its client data. This is because ib_unregister_device() will remove the device from the device list with lists_rwsem locked for write, but perform the rest of the cleanup, including the call to remove() without that lock. Mark client data that is undergoing de-registration with a new going_down flag in the client data context. Lock the client data list with lists_rwsem for write in addition to using the spinlock, so that functions calling the callback would be able to lock only lists_rwsem for read and let callbacks sleep. Since ib_unregister_client() now marks the client data context, no need for remove() to search the context again, so pass the client data directly to remove() callbacks. Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Signed-off-by: Haggai Eran <haggaie@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-07-30 22:50:14 +08:00
down_write(&lists_rwsem);
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
spin_lock_irqsave(&device->client_data_lock, flags);
list_add(&context->list, &device->client_data_list);
spin_unlock_irqrestore(&device->client_data_lock, flags);
IB/core: lock client data with lists_rwsem An ib_client callback that is called with the lists_rwsem locked only for read is protected from changes to the IB client lists, but not from ib_unregister_device() freeing its client data. This is because ib_unregister_device() will remove the device from the device list with lists_rwsem locked for write, but perform the rest of the cleanup, including the call to remove() without that lock. Mark client data that is undergoing de-registration with a new going_down flag in the client data context. Lock the client data list with lists_rwsem for write in addition to using the spinlock, so that functions calling the callback would be able to lock only lists_rwsem for read and let callbacks sleep. Since ib_unregister_client() now marks the client data context, no need for remove() to search the context again, so pass the client data directly to remove() callbacks. Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Signed-off-by: Haggai Eran <haggaie@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-07-30 22:50:14 +08:00
up_write(&lists_rwsem);
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 0;
}
IB/core: Add ability for drivers to report an alternate MAD size. Add max MAD size to the device immutable data set and have all drivers that support MADs report the current IB MAD size (IB_MGMT_MAD_SIZE) to the core. Verify MAD size data in both the MAD core and when reading the immutable data. OPA drivers will report alternate MAD sizes in subsequent patches. Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-06-07 02:38:29 +08:00
static int verify_immutable(const struct ib_device *dev, u8 port)
{
return WARN_ON(!rdma_cap_ib_mad(dev, port) &&
rdma_max_mad_size(dev, port) != 0);
}
IB/core: Add per port immutable struct to ib_device As of commit 5eb620c81ce3 "IB/core: Add helpers for uncached GID and P_Key searches"; pkey_tbl_len and gid_tbl_len are immutable data which are stored in the ib_device. The per port core capability flags to be added later are also immutable data to be stored in the ib_device object. In preparation for this create a structure for per port immutable data and place the pkey and gid table lengths within this structure. "get_port_immutable" is added as a mandatory device function to allow the drivers to fill in this data. Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-05-14 08:02:58 +08:00
static int read_port_immutable(struct ib_device *device)
IB/core: Add helpers for uncached GID and P_Key searches Add ib_find_gid() and ib_find_pkey() functions that use uncached device queries. The calls might block but the returns are always up-to-date. Cache P_Key and GID table lengths in core to avoid extra port info queries. Signed-off-by: Yosef Etigin <yosefe@voltaire.com> Acked-by: Michael S. Tsirkin <mst@dev.mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-05-14 12:26:51 +08:00
{
IB/core: Make ib_alloc_device init the kobject This gets rid of the weird in-between state where struct ib_device was allocated but the kobject didn't work. Consequently ib_device_release is now guaranteed to be called in all situations and we needn't duplicate its kfrees on error paths. Signed-off-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-05 05:23:34 +08:00
int ret;
IB/core: Add per port immutable struct to ib_device As of commit 5eb620c81ce3 "IB/core: Add helpers for uncached GID and P_Key searches"; pkey_tbl_len and gid_tbl_len are immutable data which are stored in the ib_device. The per port core capability flags to be added later are also immutable data to be stored in the ib_device object. In preparation for this create a structure for per port immutable data and place the pkey and gid table lengths within this structure. "get_port_immutable" is added as a mandatory device function to allow the drivers to fill in this data. Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-05-14 08:02:58 +08:00
u8 start_port = rdma_start_port(device);
u8 end_port = rdma_end_port(device);
u8 port;
/**
* device->port_immutable is indexed directly by the port number to make
* access to this data as efficient as possible.
*
* Therefore port_immutable is declared as a 1 based array with
* potential empty slots at the beginning.
*/
device->port_immutable = kzalloc(sizeof(*device->port_immutable)
* (end_port + 1),
GFP_KERNEL);
if (!device->port_immutable)
IB/core: Make ib_alloc_device init the kobject This gets rid of the weird in-between state where struct ib_device was allocated but the kobject didn't work. Consequently ib_device_release is now guaranteed to be called in all situations and we needn't duplicate its kfrees on error paths. Signed-off-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-05 05:23:34 +08:00
return -ENOMEM;
IB/core: Add helpers for uncached GID and P_Key searches Add ib_find_gid() and ib_find_pkey() functions that use uncached device queries. The calls might block but the returns are always up-to-date. Cache P_Key and GID table lengths in core to avoid extra port info queries. Signed-off-by: Yosef Etigin <yosefe@voltaire.com> Acked-by: Michael S. Tsirkin <mst@dev.mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-05-14 12:26:51 +08:00
IB/core: Add per port immutable struct to ib_device As of commit 5eb620c81ce3 "IB/core: Add helpers for uncached GID and P_Key searches"; pkey_tbl_len and gid_tbl_len are immutable data which are stored in the ib_device. The per port core capability flags to be added later are also immutable data to be stored in the ib_device object. In preparation for this create a structure for per port immutable data and place the pkey and gid table lengths within this structure. "get_port_immutable" is added as a mandatory device function to allow the drivers to fill in this data. Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-05-14 08:02:58 +08:00
for (port = start_port; port <= end_port; ++port) {
ret = device->get_port_immutable(device, port,
&device->port_immutable[port]);
IB/core: Add helpers for uncached GID and P_Key searches Add ib_find_gid() and ib_find_pkey() functions that use uncached device queries. The calls might block but the returns are always up-to-date. Cache P_Key and GID table lengths in core to avoid extra port info queries. Signed-off-by: Yosef Etigin <yosefe@voltaire.com> Acked-by: Michael S. Tsirkin <mst@dev.mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-05-14 12:26:51 +08:00
if (ret)
IB/core: Make ib_alloc_device init the kobject This gets rid of the weird in-between state where struct ib_device was allocated but the kobject didn't work. Consequently ib_device_release is now guaranteed to be called in all situations and we needn't duplicate its kfrees on error paths. Signed-off-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-05 05:23:34 +08:00
return ret;
IB/core: Add ability for drivers to report an alternate MAD size. Add max MAD size to the device immutable data set and have all drivers that support MADs report the current IB MAD size (IB_MGMT_MAD_SIZE) to the core. Verify MAD size data in both the MAD core and when reading the immutable data. OPA drivers will report alternate MAD sizes in subsequent patches. Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-06-07 02:38:29 +08:00
IB/core: Make ib_alloc_device init the kobject This gets rid of the weird in-between state where struct ib_device was allocated but the kobject didn't work. Consequently ib_device_release is now guaranteed to be called in all situations and we needn't duplicate its kfrees on error paths. Signed-off-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-05 05:23:34 +08:00
if (verify_immutable(device, port))
return -EINVAL;
IB/core: Add helpers for uncached GID and P_Key searches Add ib_find_gid() and ib_find_pkey() functions that use uncached device queries. The calls might block but the returns are always up-to-date. Cache P_Key and GID table lengths in core to avoid extra port info queries. Signed-off-by: Yosef Etigin <yosefe@voltaire.com> Acked-by: Michael S. Tsirkin <mst@dev.mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-05-14 12:26:51 +08:00
}
IB/core: Make ib_alloc_device init the kobject This gets rid of the weird in-between state where struct ib_device was allocated but the kobject didn't work. Consequently ib_device_release is now guaranteed to be called in all situations and we needn't duplicate its kfrees on error paths. Signed-off-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-05 05:23:34 +08:00
return 0;
IB/core: Add helpers for uncached GID and P_Key searches Add ib_find_gid() and ib_find_pkey() functions that use uncached device queries. The calls might block but the returns are always up-to-date. Cache P_Key and GID table lengths in core to avoid extra port info queries. Signed-off-by: Yosef Etigin <yosefe@voltaire.com> Acked-by: Michael S. Tsirkin <mst@dev.mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-05-14 12:26:51 +08:00
}
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_register_device - Register an IB device with IB core
* @device:Device to register
*
* Low-level drivers use ib_register_device() to register their
* devices with the IB core. All registered clients will receive a
* callback for each device that is added. @device must be allocated
* with ib_alloc_device().
*/
IB/core: Allow device-specific per-port sysfs files Add a new parameter to ib_register_device() so that low-level device drivers can pass in a pointer to a callback function that will be called for each port that is registered in sysfs. This allows low-level device drivers to create files in /sys/class/infiniband/<hca>/ports/<N>/ without having to poke through the internals of the RDMA sysfs handling. There is no need for an unregister function since the kobject reference will go to zero when ib_unregister_device() is called. Signed-off-by: Ralph Campbell <ralph.campbell@qlogic.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2010-05-07 08:03:25 +08:00
int ib_register_device(struct ib_device *device,
int (*port_callback)(struct ib_device *,
u8, struct kobject *))
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;
IB/core: Remove needless bracketization Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-15 22:16:14 +08:00
struct ib_client *client;
IB/core: Save the device attributes on the device structure This way both the IB core and upper level drivers can access these cached device attributes rather than querying or caching them on their own. Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-12-18 16:59:44 +08:00
struct ib_udata uhw = {.outlen = 0, .inlen = 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
IB: convert from semaphores to mutexes semaphore to mutex conversion by Ingo and Arjan's script. Signed-off-by: Ingo Molnar <mingo@elte.hu> [ Sanity-checked on real IB hardware ] Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-01-14 06:51:39 +08:00
mutex_lock(&device_mutex);
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 (strchr(device->name, '%')) {
ret = alloc_name(device->name);
if (ret)
goto out;
}
if (ib_device_check_mandatory(device)) {
ret = -EINVAL;
goto out;
}
IB/core: Add per port immutable struct to ib_device As of commit 5eb620c81ce3 "IB/core: Add helpers for uncached GID and P_Key searches"; pkey_tbl_len and gid_tbl_len are immutable data which are stored in the ib_device. The per port core capability flags to be added later are also immutable data to be stored in the ib_device object. In preparation for this create a structure for per port immutable data and place the pkey and gid table lengths within this structure. "get_port_immutable" is added as a mandatory device function to allow the drivers to fill in this data. Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-05-14 08:02:58 +08:00
ret = read_port_immutable(device);
IB/core: Add helpers for uncached GID and P_Key searches Add ib_find_gid() and ib_find_pkey() functions that use uncached device queries. The calls might block but the returns are always up-to-date. Cache P_Key and GID table lengths in core to avoid extra port info queries. Signed-off-by: Yosef Etigin <yosefe@voltaire.com> Acked-by: Michael S. Tsirkin <mst@dev.mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-05-14 12:26:51 +08:00
if (ret) {
IB/core: trivial prink cleanup. 1. Replaced printk with appropriate pr_warn, pr_err, pr_info. 2. Removed unnecessary prints around memory allocation failure which are not required, as reported by the checkpatch script. Signed-off-by: Parav Pandit <pandit.parav@gmail.com> Reviewed-by: Haggai Eran <haggaie@mellanox.com> Reviewed-by: Sagi Grimberg <sagig@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-03-02 03:20:29 +08:00
pr_warn("Couldn't create per port immutable data %s\n",
device->name);
IB/core: Add helpers for uncached GID and P_Key searches Add ib_find_gid() and ib_find_pkey() functions that use uncached device queries. The calls might block but the returns are always up-to-date. Cache P_Key and GID table lengths in core to avoid extra port info queries. Signed-off-by: Yosef Etigin <yosefe@voltaire.com> Acked-by: Michael S. Tsirkin <mst@dev.mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-05-14 12:26:51 +08:00
goto out;
}
IB/core: Add RoCE GID table management RoCE GIDs are based on IP addresses configured on Ethernet net-devices which relate to the RDMA (RoCE) device port. Currently, each of the low-level drivers that support RoCE (ocrdma, mlx4) manages its own RoCE port GID table. As there's nothing which is essentially vendor specific, we generalize that, and enhance the RDMA core GID cache to do this job. In order to populate the GID table, we listen for events: (a) netdev up/down/change_addr events - if a netdev is built onto our RoCE device, we need to add/delete its IPs. This involves adding all GIDs related to this ndev, add default GIDs, etc. (b) inet events - add new GIDs (according to the IP addresses) to the table. For programming the port RoCE GID table, providers must implement the add_gid and del_gid callbacks. RoCE GID management requires us to state the associated net_device alongside the GID. This information is necessary in order to manage the GID table. For example, when a net_device is removed, its associated GIDs need to be removed as well. RoCE mandates generating a default GID for each port, based on the related net-device's IPv6 link local. In contrast to the GID based on the regular IPv6 link-local (as we generate GID per IP address), the default GID is also available when the net device is down (in order to support loopback). Locking is done as follows: The patch modify the GID table code both for new RoCE drivers implementing the add_gid/del_gid callbacks and for current RoCE and IB drivers that do not. The flows for updating the table are different, so the locking requirements are too. While updating RoCE GID table, protection against multiple writers is achieved via mutex_lock(&table->lock). Since writing to a table requires us to find an entry (possible a free entry) in the table and then modify it, this mutex protects both the find_gid and write_gid ensuring the atomicity of the action. Each entry in the GID cache is protected by rwlock. In RoCE, writing (usually results from netdev notifier) involves invoking the vendor's add_gid and del_gid callbacks, which could sleep. Therefore, an invalid flag is added for each entry. Updates for RoCE are done via a workqueue, thus sleeping is permitted. In IB, updates are done in write_lock_irq(&device->cache.lock), thus write_gid isn't allowed to sleep and add_gid/del_gid are not called. When passing net-device into/out-of the GID cache, the device is always passed held (dev_hold). The code uses a single work item for updating all RDMA devices, following a netdev or inet notifier. The patch moves the cache from being a client (which was incorrect, as the cache is part of the IB infrastructure) to being explicitly initialized/freed when a device is registered/removed. Signed-off-by: Matan Barak <matanb@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-07-30 23:33:26 +08:00
ret = ib_cache_setup_one(device);
if (ret) {
IB/core: trivial prink cleanup. 1. Replaced printk with appropriate pr_warn, pr_err, pr_info. 2. Removed unnecessary prints around memory allocation failure which are not required, as reported by the checkpatch script. Signed-off-by: Parav Pandit <pandit.parav@gmail.com> Reviewed-by: Haggai Eran <haggaie@mellanox.com> Reviewed-by: Sagi Grimberg <sagig@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-03-02 03:20:29 +08:00
pr_warn("Couldn't set up InfiniBand P_Key/GID cache\n");
IB/core: Add RoCE GID table management RoCE GIDs are based on IP addresses configured on Ethernet net-devices which relate to the RDMA (RoCE) device port. Currently, each of the low-level drivers that support RoCE (ocrdma, mlx4) manages its own RoCE port GID table. As there's nothing which is essentially vendor specific, we generalize that, and enhance the RDMA core GID cache to do this job. In order to populate the GID table, we listen for events: (a) netdev up/down/change_addr events - if a netdev is built onto our RoCE device, we need to add/delete its IPs. This involves adding all GIDs related to this ndev, add default GIDs, etc. (b) inet events - add new GIDs (according to the IP addresses) to the table. For programming the port RoCE GID table, providers must implement the add_gid and del_gid callbacks. RoCE GID management requires us to state the associated net_device alongside the GID. This information is necessary in order to manage the GID table. For example, when a net_device is removed, its associated GIDs need to be removed as well. RoCE mandates generating a default GID for each port, based on the related net-device's IPv6 link local. In contrast to the GID based on the regular IPv6 link-local (as we generate GID per IP address), the default GID is also available when the net device is down (in order to support loopback). Locking is done as follows: The patch modify the GID table code both for new RoCE drivers implementing the add_gid/del_gid callbacks and for current RoCE and IB drivers that do not. The flows for updating the table are different, so the locking requirements are too. While updating RoCE GID table, protection against multiple writers is achieved via mutex_lock(&table->lock). Since writing to a table requires us to find an entry (possible a free entry) in the table and then modify it, this mutex protects both the find_gid and write_gid ensuring the atomicity of the action. Each entry in the GID cache is protected by rwlock. In RoCE, writing (usually results from netdev notifier) involves invoking the vendor's add_gid and del_gid callbacks, which could sleep. Therefore, an invalid flag is added for each entry. Updates for RoCE are done via a workqueue, thus sleeping is permitted. In IB, updates are done in write_lock_irq(&device->cache.lock), thus write_gid isn't allowed to sleep and add_gid/del_gid are not called. When passing net-device into/out-of the GID cache, the device is always passed held (dev_hold). The code uses a single work item for updating all RDMA devices, following a netdev or inet notifier. The patch moves the cache from being a client (which was incorrect, as the cache is part of the IB infrastructure) to being explicitly initialized/freed when a device is registered/removed. Signed-off-by: Matan Barak <matanb@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-07-30 23:33:26 +08:00
goto out;
}
IB/core: Save the device attributes on the device structure This way both the IB core and upper level drivers can access these cached device attributes rather than querying or caching them on their own. Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-12-18 16:59:44 +08:00
memset(&device->attrs, 0, sizeof(device->attrs));
ret = device->query_device(device, &device->attrs, &uhw);
if (ret) {
IB/core: trivial prink cleanup. 1. Replaced printk with appropriate pr_warn, pr_err, pr_info. 2. Removed unnecessary prints around memory allocation failure which are not required, as reported by the checkpatch script. Signed-off-by: Parav Pandit <pandit.parav@gmail.com> Reviewed-by: Haggai Eran <haggaie@mellanox.com> Reviewed-by: Sagi Grimberg <sagig@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-03-02 03:20:29 +08:00
pr_warn("Couldn't query the device attributes\n");
IB/core: Fix missed clean call in registration path In case of failure returned from query function in IB device registration, we need to clean IB cache which was missed. This change fixes it. Fixes: 3e153a93a1c1 ('IB/core: Save the device attributes on the device structure') Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Sagi Grimberg <sagig@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-02-22 00:12:26 +08:00
ib_cache_cleanup_one(device);
IB/core: Save the device attributes on the device structure This way both the IB core and upper level drivers can access these cached device attributes rather than querying or caching them on their own. Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-12-18 16:59:44 +08:00
goto out;
}
IB/core: Allow device-specific per-port sysfs files Add a new parameter to ib_register_device() so that low-level device drivers can pass in a pointer to a callback function that will be called for each port that is registered in sysfs. This allows low-level device drivers to create files in /sys/class/infiniband/<hca>/ports/<N>/ without having to poke through the internals of the RDMA sysfs handling. There is no need for an unregister function since the kobject reference will go to zero when ib_unregister_device() is called. Signed-off-by: Ralph Campbell <ralph.campbell@qlogic.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2010-05-07 08:03:25 +08:00
ret = ib_device_register_sysfs(device, port_callback);
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 (ret) {
IB/core: trivial prink cleanup. 1. Replaced printk with appropriate pr_warn, pr_err, pr_info. 2. Removed unnecessary prints around memory allocation failure which are not required, as reported by the checkpatch script. Signed-off-by: Parav Pandit <pandit.parav@gmail.com> Reviewed-by: Haggai Eran <haggaie@mellanox.com> Reviewed-by: Sagi Grimberg <sagig@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-03-02 03:20:29 +08:00
pr_warn("Couldn't register device %s with driver model\n",
device->name);
IB/core: Add RoCE GID table management RoCE GIDs are based on IP addresses configured on Ethernet net-devices which relate to the RDMA (RoCE) device port. Currently, each of the low-level drivers that support RoCE (ocrdma, mlx4) manages its own RoCE port GID table. As there's nothing which is essentially vendor specific, we generalize that, and enhance the RDMA core GID cache to do this job. In order to populate the GID table, we listen for events: (a) netdev up/down/change_addr events - if a netdev is built onto our RoCE device, we need to add/delete its IPs. This involves adding all GIDs related to this ndev, add default GIDs, etc. (b) inet events - add new GIDs (according to the IP addresses) to the table. For programming the port RoCE GID table, providers must implement the add_gid and del_gid callbacks. RoCE GID management requires us to state the associated net_device alongside the GID. This information is necessary in order to manage the GID table. For example, when a net_device is removed, its associated GIDs need to be removed as well. RoCE mandates generating a default GID for each port, based on the related net-device's IPv6 link local. In contrast to the GID based on the regular IPv6 link-local (as we generate GID per IP address), the default GID is also available when the net device is down (in order to support loopback). Locking is done as follows: The patch modify the GID table code both for new RoCE drivers implementing the add_gid/del_gid callbacks and for current RoCE and IB drivers that do not. The flows for updating the table are different, so the locking requirements are too. While updating RoCE GID table, protection against multiple writers is achieved via mutex_lock(&table->lock). Since writing to a table requires us to find an entry (possible a free entry) in the table and then modify it, this mutex protects both the find_gid and write_gid ensuring the atomicity of the action. Each entry in the GID cache is protected by rwlock. In RoCE, writing (usually results from netdev notifier) involves invoking the vendor's add_gid and del_gid callbacks, which could sleep. Therefore, an invalid flag is added for each entry. Updates for RoCE are done via a workqueue, thus sleeping is permitted. In IB, updates are done in write_lock_irq(&device->cache.lock), thus write_gid isn't allowed to sleep and add_gid/del_gid are not called. When passing net-device into/out-of the GID cache, the device is always passed held (dev_hold). The code uses a single work item for updating all RDMA devices, following a netdev or inet notifier. The patch moves the cache from being a client (which was incorrect, as the cache is part of the IB infrastructure) to being explicitly initialized/freed when a device is registered/removed. Signed-off-by: Matan Barak <matanb@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-07-30 23:33:26 +08:00
ib_cache_cleanup_one(device);
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 out;
}
device->reg_state = IB_DEV_REGISTERED;
IB/core: Remove needless bracketization Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-15 22:16:14 +08:00
list_for_each_entry(client, &client_list, list)
if (client->add && !add_client_context(device, client))
client->add(device);
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/core: Add rwsem to allow reading device list or client list Currently the RDMA subsystem's device list and client list are protected by a single mutex. This prevents adding user-facing APIs that iterate these lists, since using them may cause a deadlock. The patch attempts to solve this problem by adding a read-write semaphore to protect the lists. Readers now don't need the mutex, and are safe just by read-locking the semaphore. The ib_register_device, ib_register_client, ib_unregister_device, and ib_unregister_client functions are modified to lock the semaphore for write during their respective list modification. Also, in order to make sure client callbacks are called only between add() and remove() calls, the code is changed to only add items to the lists after the add() calls and remove from the lists before the remove() calls. This patch attempts to solve a similar need [1] that was seen in the RoCE v2 patch series. [1] http://www.spinics.net/lists/linux-rdma/msg24733.html Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Cc: Matan Barak <matanb@mellanox.com> Signed-off-by: Haggai Eran <haggaie@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-07-30 22:50:13 +08:00
down_write(&lists_rwsem);
list_add_tail(&device->core_list, &device_list);
up_write(&lists_rwsem);
out:
IB: convert from semaphores to mutexes semaphore to mutex conversion by Ingo and Arjan's script. Signed-off-by: Ingo Molnar <mingo@elte.hu> [ Sanity-checked on real IB hardware ] Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-01-14 06:51:39 +08:00
mutex_unlock(&device_mutex);
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 ret;
}
EXPORT_SYMBOL(ib_register_device);
/**
* ib_unregister_device - Unregister an IB device
* @device:Device to unregister
*
* Unregister an IB device. All clients will receive a remove callback.
*/
void ib_unregister_device(struct ib_device *device)
{
struct ib_client_data *context, *tmp;
unsigned long flags;
IB: convert from semaphores to mutexes semaphore to mutex conversion by Ingo and Arjan's script. Signed-off-by: Ingo Molnar <mingo@elte.hu> [ Sanity-checked on real IB hardware ] Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-01-14 06:51:39 +08:00
mutex_lock(&device_mutex);
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/core: Add rwsem to allow reading device list or client list Currently the RDMA subsystem's device list and client list are protected by a single mutex. This prevents adding user-facing APIs that iterate these lists, since using them may cause a deadlock. The patch attempts to solve this problem by adding a read-write semaphore to protect the lists. Readers now don't need the mutex, and are safe just by read-locking the semaphore. The ib_register_device, ib_register_client, ib_unregister_device, and ib_unregister_client functions are modified to lock the semaphore for write during their respective list modification. Also, in order to make sure client callbacks are called only between add() and remove() calls, the code is changed to only add items to the lists after the add() calls and remove from the lists before the remove() calls. This patch attempts to solve a similar need [1] that was seen in the RoCE v2 patch series. [1] http://www.spinics.net/lists/linux-rdma/msg24733.html Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Cc: Matan Barak <matanb@mellanox.com> Signed-off-by: Haggai Eran <haggaie@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-07-30 22:50:13 +08:00
down_write(&lists_rwsem);
list_del(&device->core_list);
IB/core: lock client data with lists_rwsem An ib_client callback that is called with the lists_rwsem locked only for read is protected from changes to the IB client lists, but not from ib_unregister_device() freeing its client data. This is because ib_unregister_device() will remove the device from the device list with lists_rwsem locked for write, but perform the rest of the cleanup, including the call to remove() without that lock. Mark client data that is undergoing de-registration with a new going_down flag in the client data context. Lock the client data list with lists_rwsem for write in addition to using the spinlock, so that functions calling the callback would be able to lock only lists_rwsem for read and let callbacks sleep. Since ib_unregister_client() now marks the client data context, no need for remove() to search the context again, so pass the client data directly to remove() callbacks. Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Signed-off-by: Haggai Eran <haggaie@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-07-30 22:50:14 +08:00
spin_lock_irqsave(&device->client_data_lock, flags);
list_for_each_entry_safe(context, tmp, &device->client_data_list, list)
context->going_down = true;
spin_unlock_irqrestore(&device->client_data_lock, flags);
downgrade_write(&lists_rwsem);
IB/core: Add rwsem to allow reading device list or client list Currently the RDMA subsystem's device list and client list are protected by a single mutex. This prevents adding user-facing APIs that iterate these lists, since using them may cause a deadlock. The patch attempts to solve this problem by adding a read-write semaphore to protect the lists. Readers now don't need the mutex, and are safe just by read-locking the semaphore. The ib_register_device, ib_register_client, ib_unregister_device, and ib_unregister_client functions are modified to lock the semaphore for write during their respective list modification. Also, in order to make sure client callbacks are called only between add() and remove() calls, the code is changed to only add items to the lists after the add() calls and remove from the lists before the remove() calls. This patch attempts to solve a similar need [1] that was seen in the RoCE v2 patch series. [1] http://www.spinics.net/lists/linux-rdma/msg24733.html Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Cc: Matan Barak <matanb@mellanox.com> Signed-off-by: Haggai Eran <haggaie@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-07-30 22:50:13 +08:00
IB/core: lock client data with lists_rwsem An ib_client callback that is called with the lists_rwsem locked only for read is protected from changes to the IB client lists, but not from ib_unregister_device() freeing its client data. This is because ib_unregister_device() will remove the device from the device list with lists_rwsem locked for write, but perform the rest of the cleanup, including the call to remove() without that lock. Mark client data that is undergoing de-registration with a new going_down flag in the client data context. Lock the client data list with lists_rwsem for write in addition to using the spinlock, so that functions calling the callback would be able to lock only lists_rwsem for read and let callbacks sleep. Since ib_unregister_client() now marks the client data context, no need for remove() to search the context again, so pass the client data directly to remove() callbacks. Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Signed-off-by: Haggai Eran <haggaie@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-07-30 22:50:14 +08:00
list_for_each_entry_safe(context, tmp, &device->client_data_list,
list) {
if (context->client->remove)
context->client->remove(device, context->data);
}
up_read(&lists_rwsem);
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: convert from semaphores to mutexes semaphore to mutex conversion by Ingo and Arjan's script. Signed-off-by: Ingo Molnar <mingo@elte.hu> [ Sanity-checked on real IB hardware ] Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-01-14 06:51:39 +08:00
mutex_unlock(&device_mutex);
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: Remove sysfs files before unregistering device Move the ib_device_unregister_sysfs() call from ib_dealloc_device() to ib_unregister_device(). The old code allows device unregister to proceed even if some sysfs files are open, which leaves a window where userspace can open a file before a device is removed but then end up reading the file after the device is removed, which leads to various kernel crashes either because the device data structure is freed or because the low-level driver code is gone after module removal. By not returning from ib_unregister_device() until after all sysfs entries are removed, we make sure that data structures and/or module code is not freed until after all sysfs access is done. Reported-by: Jack Morgenstein <jackm@dev.mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2009-02-26 05:27:46 +08:00
ib_device_unregister_sysfs(device);
IB/core: Add RoCE GID table management RoCE GIDs are based on IP addresses configured on Ethernet net-devices which relate to the RDMA (RoCE) device port. Currently, each of the low-level drivers that support RoCE (ocrdma, mlx4) manages its own RoCE port GID table. As there's nothing which is essentially vendor specific, we generalize that, and enhance the RDMA core GID cache to do this job. In order to populate the GID table, we listen for events: (a) netdev up/down/change_addr events - if a netdev is built onto our RoCE device, we need to add/delete its IPs. This involves adding all GIDs related to this ndev, add default GIDs, etc. (b) inet events - add new GIDs (according to the IP addresses) to the table. For programming the port RoCE GID table, providers must implement the add_gid and del_gid callbacks. RoCE GID management requires us to state the associated net_device alongside the GID. This information is necessary in order to manage the GID table. For example, when a net_device is removed, its associated GIDs need to be removed as well. RoCE mandates generating a default GID for each port, based on the related net-device's IPv6 link local. In contrast to the GID based on the regular IPv6 link-local (as we generate GID per IP address), the default GID is also available when the net device is down (in order to support loopback). Locking is done as follows: The patch modify the GID table code both for new RoCE drivers implementing the add_gid/del_gid callbacks and for current RoCE and IB drivers that do not. The flows for updating the table are different, so the locking requirements are too. While updating RoCE GID table, protection against multiple writers is achieved via mutex_lock(&table->lock). Since writing to a table requires us to find an entry (possible a free entry) in the table and then modify it, this mutex protects both the find_gid and write_gid ensuring the atomicity of the action. Each entry in the GID cache is protected by rwlock. In RoCE, writing (usually results from netdev notifier) involves invoking the vendor's add_gid and del_gid callbacks, which could sleep. Therefore, an invalid flag is added for each entry. Updates for RoCE are done via a workqueue, thus sleeping is permitted. In IB, updates are done in write_lock_irq(&device->cache.lock), thus write_gid isn't allowed to sleep and add_gid/del_gid are not called. When passing net-device into/out-of the GID cache, the device is always passed held (dev_hold). The code uses a single work item for updating all RDMA devices, following a netdev or inet notifier. The patch moves the cache from being a client (which was incorrect, as the cache is part of the IB infrastructure) to being explicitly initialized/freed when a device is registered/removed. Signed-off-by: Matan Barak <matanb@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-07-30 23:33:26 +08:00
ib_cache_cleanup_one(device);
IB: Remove sysfs files before unregistering device Move the ib_device_unregister_sysfs() call from ib_dealloc_device() to ib_unregister_device(). The old code allows device unregister to proceed even if some sysfs files are open, which leaves a window where userspace can open a file before a device is removed but then end up reading the file after the device is removed, which leads to various kernel crashes either because the device data structure is freed or because the low-level driver code is gone after module removal. By not returning from ib_unregister_device() until after all sysfs entries are removed, we make sure that data structures and/or module code is not freed until after all sysfs access is done. Reported-by: Jack Morgenstein <jackm@dev.mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2009-02-26 05:27:46 +08:00
IB/core: lock client data with lists_rwsem An ib_client callback that is called with the lists_rwsem locked only for read is protected from changes to the IB client lists, but not from ib_unregister_device() freeing its client data. This is because ib_unregister_device() will remove the device from the device list with lists_rwsem locked for write, but perform the rest of the cleanup, including the call to remove() without that lock. Mark client data that is undergoing de-registration with a new going_down flag in the client data context. Lock the client data list with lists_rwsem for write in addition to using the spinlock, so that functions calling the callback would be able to lock only lists_rwsem for read and let callbacks sleep. Since ib_unregister_client() now marks the client data context, no need for remove() to search the context again, so pass the client data directly to remove() callbacks. Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Signed-off-by: Haggai Eran <haggaie@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-07-30 22:50:14 +08:00
down_write(&lists_rwsem);
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
spin_lock_irqsave(&device->client_data_lock, flags);
list_for_each_entry_safe(context, tmp, &device->client_data_list, list)
kfree(context);
spin_unlock_irqrestore(&device->client_data_lock, flags);
IB/core: lock client data with lists_rwsem An ib_client callback that is called with the lists_rwsem locked only for read is protected from changes to the IB client lists, but not from ib_unregister_device() freeing its client data. This is because ib_unregister_device() will remove the device from the device list with lists_rwsem locked for write, but perform the rest of the cleanup, including the call to remove() without that lock. Mark client data that is undergoing de-registration with a new going_down flag in the client data context. Lock the client data list with lists_rwsem for write in addition to using the spinlock, so that functions calling the callback would be able to lock only lists_rwsem for read and let callbacks sleep. Since ib_unregister_client() now marks the client data context, no need for remove() to search the context again, so pass the client data directly to remove() callbacks. Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Signed-off-by: Haggai Eran <haggaie@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-07-30 22:50:14 +08:00
up_write(&lists_rwsem);
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
device->reg_state = IB_DEV_UNREGISTERED;
}
EXPORT_SYMBOL(ib_unregister_device);
/**
* ib_register_client - Register an IB client
* @client:Client to register
*
* Upper level users of the IB drivers can use ib_register_client() to
* register callbacks for IB device addition and removal. When an IB
* device is added, each registered client's add method will be called
* (in the order the clients were registered), and when a device is
* removed, each client's remove method will be called (in the reverse
* order that clients were registered). In addition, when
* ib_register_client() is called, the client will receive an add
* callback for all devices already registered.
*/
int ib_register_client(struct ib_client *client)
{
struct ib_device *device;
IB: convert from semaphores to mutexes semaphore to mutex conversion by Ingo and Arjan's script. Signed-off-by: Ingo Molnar <mingo@elte.hu> [ Sanity-checked on real IB hardware ] Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-01-14 06:51:39 +08:00
mutex_lock(&device_mutex);
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
list_for_each_entry(device, &device_list, core_list)
if (client->add && !add_client_context(device, client))
client->add(device);
IB/core: Add rwsem to allow reading device list or client list Currently the RDMA subsystem's device list and client list are protected by a single mutex. This prevents adding user-facing APIs that iterate these lists, since using them may cause a deadlock. The patch attempts to solve this problem by adding a read-write semaphore to protect the lists. Readers now don't need the mutex, and are safe just by read-locking the semaphore. The ib_register_device, ib_register_client, ib_unregister_device, and ib_unregister_client functions are modified to lock the semaphore for write during their respective list modification. Also, in order to make sure client callbacks are called only between add() and remove() calls, the code is changed to only add items to the lists after the add() calls and remove from the lists before the remove() calls. This patch attempts to solve a similar need [1] that was seen in the RoCE v2 patch series. [1] http://www.spinics.net/lists/linux-rdma/msg24733.html Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Cc: Matan Barak <matanb@mellanox.com> Signed-off-by: Haggai Eran <haggaie@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-07-30 22:50:13 +08:00
down_write(&lists_rwsem);
list_add_tail(&client->list, &client_list);
up_write(&lists_rwsem);
IB: convert from semaphores to mutexes semaphore to mutex conversion by Ingo and Arjan's script. Signed-off-by: Ingo Molnar <mingo@elte.hu> [ Sanity-checked on real IB hardware ] Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-01-14 06:51:39 +08:00
mutex_unlock(&device_mutex);
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 0;
}
EXPORT_SYMBOL(ib_register_client);
/**
* ib_unregister_client - Unregister an IB client
* @client:Client to unregister
*
* Upper level users use ib_unregister_client() to remove their client
* registration. When ib_unregister_client() is called, the client
* will receive a remove callback for each IB device still registered.
*/
void ib_unregister_client(struct ib_client *client)
{
struct ib_client_data *context, *tmp;
struct ib_device *device;
unsigned long flags;
IB: convert from semaphores to mutexes semaphore to mutex conversion by Ingo and Arjan's script. Signed-off-by: Ingo Molnar <mingo@elte.hu> [ Sanity-checked on real IB hardware ] Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-01-14 06:51:39 +08:00
mutex_lock(&device_mutex);
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/core: Add rwsem to allow reading device list or client list Currently the RDMA subsystem's device list and client list are protected by a single mutex. This prevents adding user-facing APIs that iterate these lists, since using them may cause a deadlock. The patch attempts to solve this problem by adding a read-write semaphore to protect the lists. Readers now don't need the mutex, and are safe just by read-locking the semaphore. The ib_register_device, ib_register_client, ib_unregister_device, and ib_unregister_client functions are modified to lock the semaphore for write during their respective list modification. Also, in order to make sure client callbacks are called only between add() and remove() calls, the code is changed to only add items to the lists after the add() calls and remove from the lists before the remove() calls. This patch attempts to solve a similar need [1] that was seen in the RoCE v2 patch series. [1] http://www.spinics.net/lists/linux-rdma/msg24733.html Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Cc: Matan Barak <matanb@mellanox.com> Signed-off-by: Haggai Eran <haggaie@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-07-30 22:50:13 +08:00
down_write(&lists_rwsem);
list_del(&client->list);
up_write(&lists_rwsem);
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
list_for_each_entry(device, &device_list, core_list) {
IB/core: lock client data with lists_rwsem An ib_client callback that is called with the lists_rwsem locked only for read is protected from changes to the IB client lists, but not from ib_unregister_device() freeing its client data. This is because ib_unregister_device() will remove the device from the device list with lists_rwsem locked for write, but perform the rest of the cleanup, including the call to remove() without that lock. Mark client data that is undergoing de-registration with a new going_down flag in the client data context. Lock the client data list with lists_rwsem for write in addition to using the spinlock, so that functions calling the callback would be able to lock only lists_rwsem for read and let callbacks sleep. Since ib_unregister_client() now marks the client data context, no need for remove() to search the context again, so pass the client data directly to remove() callbacks. Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Signed-off-by: Haggai Eran <haggaie@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-07-30 22:50:14 +08:00
struct ib_client_data *found_context = NULL;
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/core: lock client data with lists_rwsem An ib_client callback that is called with the lists_rwsem locked only for read is protected from changes to the IB client lists, but not from ib_unregister_device() freeing its client data. This is because ib_unregister_device() will remove the device from the device list with lists_rwsem locked for write, but perform the rest of the cleanup, including the call to remove() without that lock. Mark client data that is undergoing de-registration with a new going_down flag in the client data context. Lock the client data list with lists_rwsem for write in addition to using the spinlock, so that functions calling the callback would be able to lock only lists_rwsem for read and let callbacks sleep. Since ib_unregister_client() now marks the client data context, no need for remove() to search the context again, so pass the client data directly to remove() callbacks. Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Signed-off-by: Haggai Eran <haggaie@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-07-30 22:50:14 +08:00
down_write(&lists_rwsem);
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
spin_lock_irqsave(&device->client_data_lock, flags);
list_for_each_entry_safe(context, tmp, &device->client_data_list, list)
if (context->client == client) {
IB/core: lock client data with lists_rwsem An ib_client callback that is called with the lists_rwsem locked only for read is protected from changes to the IB client lists, but not from ib_unregister_device() freeing its client data. This is because ib_unregister_device() will remove the device from the device list with lists_rwsem locked for write, but perform the rest of the cleanup, including the call to remove() without that lock. Mark client data that is undergoing de-registration with a new going_down flag in the client data context. Lock the client data list with lists_rwsem for write in addition to using the spinlock, so that functions calling the callback would be able to lock only lists_rwsem for read and let callbacks sleep. Since ib_unregister_client() now marks the client data context, no need for remove() to search the context again, so pass the client data directly to remove() callbacks. Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Signed-off-by: Haggai Eran <haggaie@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-07-30 22:50:14 +08:00
context->going_down = true;
found_context = context;
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
}
spin_unlock_irqrestore(&device->client_data_lock, flags);
IB/core: lock client data with lists_rwsem An ib_client callback that is called with the lists_rwsem locked only for read is protected from changes to the IB client lists, but not from ib_unregister_device() freeing its client data. This is because ib_unregister_device() will remove the device from the device list with lists_rwsem locked for write, but perform the rest of the cleanup, including the call to remove() without that lock. Mark client data that is undergoing de-registration with a new going_down flag in the client data context. Lock the client data list with lists_rwsem for write in addition to using the spinlock, so that functions calling the callback would be able to lock only lists_rwsem for read and let callbacks sleep. Since ib_unregister_client() now marks the client data context, no need for remove() to search the context again, so pass the client data directly to remove() callbacks. Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Signed-off-by: Haggai Eran <haggaie@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-07-30 22:50:14 +08:00
up_write(&lists_rwsem);
if (client->remove)
client->remove(device, found_context ?
found_context->data : NULL);
if (!found_context) {
pr_warn("No client context found for %s/%s\n",
device->name, client->name);
continue;
}
down_write(&lists_rwsem);
spin_lock_irqsave(&device->client_data_lock, flags);
list_del(&found_context->list);
kfree(found_context);
spin_unlock_irqrestore(&device->client_data_lock, flags);
up_write(&lists_rwsem);
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: convert from semaphores to mutexes semaphore to mutex conversion by Ingo and Arjan's script. Signed-off-by: Ingo Molnar <mingo@elte.hu> [ Sanity-checked on real IB hardware ] Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-01-14 06:51:39 +08:00
mutex_unlock(&device_mutex);
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
}
EXPORT_SYMBOL(ib_unregister_client);
/**
* ib_get_client_data - Get IB client context
* @device:Device to get context for
* @client:Client to get context for
*
* ib_get_client_data() returns client context set with
* ib_set_client_data().
*/
void *ib_get_client_data(struct ib_device *device, struct ib_client *client)
{
struct ib_client_data *context;
void *ret = NULL;
unsigned long flags;
spin_lock_irqsave(&device->client_data_lock, flags);
list_for_each_entry(context, &device->client_data_list, list)
if (context->client == client) {
ret = context->data;
break;
}
spin_unlock_irqrestore(&device->client_data_lock, flags);
return ret;
}
EXPORT_SYMBOL(ib_get_client_data);
/**
IB: Fix typo in kerneldoc for ib_set_client_data() Signed-off-by: Krishna Kumar <krkumar2@in.ibm.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-09-23 06:22:58 +08:00
* ib_set_client_data - Set IB client context
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
* @device:Device to set context for
* @client:Client to set context for
* @data:Context to set
*
* ib_set_client_data() sets client context that can be retrieved with
* ib_get_client_data().
*/
void ib_set_client_data(struct ib_device *device, struct ib_client *client,
void *data)
{
struct ib_client_data *context;
unsigned long flags;
spin_lock_irqsave(&device->client_data_lock, flags);
list_for_each_entry(context, &device->client_data_list, list)
if (context->client == client) {
context->data = data;
goto out;
}
IB/core: trivial prink cleanup. 1. Replaced printk with appropriate pr_warn, pr_err, pr_info. 2. Removed unnecessary prints around memory allocation failure which are not required, as reported by the checkpatch script. Signed-off-by: Parav Pandit <pandit.parav@gmail.com> Reviewed-by: Haggai Eran <haggaie@mellanox.com> Reviewed-by: Sagi Grimberg <sagig@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-03-02 03:20:29 +08:00
pr_warn("No client context found for %s/%s\n",
device->name, client->name);
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
out:
spin_unlock_irqrestore(&device->client_data_lock, flags);
}
EXPORT_SYMBOL(ib_set_client_data);
/**
* ib_register_event_handler - Register an IB event handler
* @event_handler:Handler to register
*
* ib_register_event_handler() registers an event handler that will be
* called back when asynchronous IB events occur (as defined in
* chapter 11 of the InfiniBand Architecture Specification). This
* callback may occur in interrupt context.
*/
int ib_register_event_handler (struct ib_event_handler *event_handler)
{
unsigned long flags;
spin_lock_irqsave(&event_handler->device->event_handler_lock, flags);
list_add_tail(&event_handler->list,
&event_handler->device->event_handler_list);
spin_unlock_irqrestore(&event_handler->device->event_handler_lock, flags);
return 0;
}
EXPORT_SYMBOL(ib_register_event_handler);
/**
* ib_unregister_event_handler - Unregister an event handler
* @event_handler:Handler to unregister
*
* Unregister an event handler registered with
* ib_register_event_handler().
*/
int ib_unregister_event_handler(struct ib_event_handler *event_handler)
{
unsigned long flags;
spin_lock_irqsave(&event_handler->device->event_handler_lock, flags);
list_del(&event_handler->list);
spin_unlock_irqrestore(&event_handler->device->event_handler_lock, flags);
return 0;
}
EXPORT_SYMBOL(ib_unregister_event_handler);
/**
* ib_dispatch_event - Dispatch an asynchronous event
* @event:Event to dispatch
*
* Low-level drivers must call ib_dispatch_event() to dispatch the
* event to all registered event handlers when an asynchronous event
* occurs.
*/
void ib_dispatch_event(struct ib_event *event)
{
unsigned long flags;
struct ib_event_handler *handler;
spin_lock_irqsave(&event->device->event_handler_lock, flags);
list_for_each_entry(handler, &event->device->event_handler_list, list)
handler->handler(handler, event);
spin_unlock_irqrestore(&event->device->event_handler_lock, flags);
}
EXPORT_SYMBOL(ib_dispatch_event);
/**
* ib_query_port - Query IB port attributes
* @device:Device to query
* @port_num:Port number to query
* @port_attr:Port attributes
*
* ib_query_port() returns the attributes of a port through the
* @port_attr pointer.
*/
int ib_query_port(struct ib_device *device,
u8 port_num,
struct ib_port_attr *port_attr)
{
IB/core: Add subnet prefix to port info The subnet prefix is a part of the port_info MAD returned and should be available at the ib_port_attr struct. We define it here and provide a default implementation in case the hardware driver does not provide one. The subnet prefix is required when creating the address vector to access the SA in networks where GRH must be used. Signed-off-by: Eli Cohen <eli@mellanox.com> Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-03-12 04:58:36 +08:00
union ib_gid gid;
int err;
IB/core: Create common start/end port functions Previously start_port and end_port were defined in 2 places, cache.c and device.c and this prevented their use in other modules. Make these common functions, change the name to reflect the rdma name space, and update existing users. Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-05-14 08:02:55 +08:00
if (port_num < rdma_start_port(device) || port_num > rdma_end_port(device))
[IB] Check port number in ib_query_port()/ib_modify_port() Check port number before passing query_port or modify_port operations on to device driver. Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-04 00:32:33 +08:00
return -EINVAL;
IB/core: Add subnet prefix to port info The subnet prefix is a part of the port_info MAD returned and should be available at the ib_port_attr struct. We define it here and provide a default implementation in case the hardware driver does not provide one. The subnet prefix is required when creating the address vector to access the SA in networks where GRH must be used. Signed-off-by: Eli Cohen <eli@mellanox.com> Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-03-12 04:58:36 +08:00
memset(port_attr, 0, sizeof(*port_attr));
err = device->query_port(device, port_num, port_attr);
if (err || port_attr->subnet_prefix)
return err;
err = ib_query_gid(device, port_num, 0, &gid, NULL);
if (err)
return err;
port_attr->subnet_prefix = be64_to_cpu(gid.global.subnet_prefix);
return 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
}
EXPORT_SYMBOL(ib_query_port);
/**
* ib_query_gid - Get GID table entry
* @device:Device to query
* @port_num:Port number to query
* @index:GID table index to query
* @gid:Returned GID
IB/core: Add netdev and gid attributes paramteres to cache Adding an ability to query the IB cache by a netdev and get the attributes of a GID. These parameters are necessary in order to successfully resolve the required GID (when the netdevice is known) and get the Ethernet L2 attributes from a GID. Signed-off-by: Matan Barak <matanb@mellanox.com> Reviewed-By: Devesh Sharma <devesh.sharma@avagotech.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-10-15 23:38:45 +08:00
* @attr: Returned GID attributes related to this GID index (only in RoCE).
* NULL means ignore.
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_query_gid() fetches the specified GID table entry.
*/
int ib_query_gid(struct ib_device *device,
IB/core: Add netdev and gid attributes paramteres to cache Adding an ability to query the IB cache by a netdev and get the attributes of a GID. These parameters are necessary in order to successfully resolve the required GID (when the netdevice is known) and get the Ethernet L2 attributes from a GID. Signed-off-by: Matan Barak <matanb@mellanox.com> Reviewed-By: Devesh Sharma <devesh.sharma@avagotech.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-10-15 23:38:45 +08:00
u8 port_num, int index, union ib_gid *gid,
struct ib_gid_attr *attr)
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/core: Add RoCE GID table management RoCE GIDs are based on IP addresses configured on Ethernet net-devices which relate to the RDMA (RoCE) device port. Currently, each of the low-level drivers that support RoCE (ocrdma, mlx4) manages its own RoCE port GID table. As there's nothing which is essentially vendor specific, we generalize that, and enhance the RDMA core GID cache to do this job. In order to populate the GID table, we listen for events: (a) netdev up/down/change_addr events - if a netdev is built onto our RoCE device, we need to add/delete its IPs. This involves adding all GIDs related to this ndev, add default GIDs, etc. (b) inet events - add new GIDs (according to the IP addresses) to the table. For programming the port RoCE GID table, providers must implement the add_gid and del_gid callbacks. RoCE GID management requires us to state the associated net_device alongside the GID. This information is necessary in order to manage the GID table. For example, when a net_device is removed, its associated GIDs need to be removed as well. RoCE mandates generating a default GID for each port, based on the related net-device's IPv6 link local. In contrast to the GID based on the regular IPv6 link-local (as we generate GID per IP address), the default GID is also available when the net device is down (in order to support loopback). Locking is done as follows: The patch modify the GID table code both for new RoCE drivers implementing the add_gid/del_gid callbacks and for current RoCE and IB drivers that do not. The flows for updating the table are different, so the locking requirements are too. While updating RoCE GID table, protection against multiple writers is achieved via mutex_lock(&table->lock). Since writing to a table requires us to find an entry (possible a free entry) in the table and then modify it, this mutex protects both the find_gid and write_gid ensuring the atomicity of the action. Each entry in the GID cache is protected by rwlock. In RoCE, writing (usually results from netdev notifier) involves invoking the vendor's add_gid and del_gid callbacks, which could sleep. Therefore, an invalid flag is added for each entry. Updates for RoCE are done via a workqueue, thus sleeping is permitted. In IB, updates are done in write_lock_irq(&device->cache.lock), thus write_gid isn't allowed to sleep and add_gid/del_gid are not called. When passing net-device into/out-of the GID cache, the device is always passed held (dev_hold). The code uses a single work item for updating all RDMA devices, following a netdev or inet notifier. The patch moves the cache from being a client (which was incorrect, as the cache is part of the IB infrastructure) to being explicitly initialized/freed when a device is registered/removed. Signed-off-by: Matan Barak <matanb@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-07-30 23:33:26 +08:00
if (rdma_cap_roce_gid_table(device, port_num))
IB/core: Add netdev and gid attributes paramteres to cache Adding an ability to query the IB cache by a netdev and get the attributes of a GID. These parameters are necessary in order to successfully resolve the required GID (when the netdevice is known) and get the Ethernet L2 attributes from a GID. Signed-off-by: Matan Barak <matanb@mellanox.com> Reviewed-By: Devesh Sharma <devesh.sharma@avagotech.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-10-15 23:38:45 +08:00
return ib_get_cached_gid(device, port_num, index, gid, attr);
if (attr)
return -EINVAL;
IB/core: Add RoCE GID table management RoCE GIDs are based on IP addresses configured on Ethernet net-devices which relate to the RDMA (RoCE) device port. Currently, each of the low-level drivers that support RoCE (ocrdma, mlx4) manages its own RoCE port GID table. As there's nothing which is essentially vendor specific, we generalize that, and enhance the RDMA core GID cache to do this job. In order to populate the GID table, we listen for events: (a) netdev up/down/change_addr events - if a netdev is built onto our RoCE device, we need to add/delete its IPs. This involves adding all GIDs related to this ndev, add default GIDs, etc. (b) inet events - add new GIDs (according to the IP addresses) to the table. For programming the port RoCE GID table, providers must implement the add_gid and del_gid callbacks. RoCE GID management requires us to state the associated net_device alongside the GID. This information is necessary in order to manage the GID table. For example, when a net_device is removed, its associated GIDs need to be removed as well. RoCE mandates generating a default GID for each port, based on the related net-device's IPv6 link local. In contrast to the GID based on the regular IPv6 link-local (as we generate GID per IP address), the default GID is also available when the net device is down (in order to support loopback). Locking is done as follows: The patch modify the GID table code both for new RoCE drivers implementing the add_gid/del_gid callbacks and for current RoCE and IB drivers that do not. The flows for updating the table are different, so the locking requirements are too. While updating RoCE GID table, protection against multiple writers is achieved via mutex_lock(&table->lock). Since writing to a table requires us to find an entry (possible a free entry) in the table and then modify it, this mutex protects both the find_gid and write_gid ensuring the atomicity of the action. Each entry in the GID cache is protected by rwlock. In RoCE, writing (usually results from netdev notifier) involves invoking the vendor's add_gid and del_gid callbacks, which could sleep. Therefore, an invalid flag is added for each entry. Updates for RoCE are done via a workqueue, thus sleeping is permitted. In IB, updates are done in write_lock_irq(&device->cache.lock), thus write_gid isn't allowed to sleep and add_gid/del_gid are not called. When passing net-device into/out-of the GID cache, the device is always passed held (dev_hold). The code uses a single work item for updating all RDMA devices, following a netdev or inet notifier. The patch moves the cache from being a client (which was incorrect, as the cache is part of the IB infrastructure) to being explicitly initialized/freed when a device is registered/removed. Signed-off-by: Matan Barak <matanb@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-07-30 23:33:26 +08:00
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 device->query_gid(device, port_num, index, gid);
}
EXPORT_SYMBOL(ib_query_gid);
IB/core: Add RoCE GID table management RoCE GIDs are based on IP addresses configured on Ethernet net-devices which relate to the RDMA (RoCE) device port. Currently, each of the low-level drivers that support RoCE (ocrdma, mlx4) manages its own RoCE port GID table. As there's nothing which is essentially vendor specific, we generalize that, and enhance the RDMA core GID cache to do this job. In order to populate the GID table, we listen for events: (a) netdev up/down/change_addr events - if a netdev is built onto our RoCE device, we need to add/delete its IPs. This involves adding all GIDs related to this ndev, add default GIDs, etc. (b) inet events - add new GIDs (according to the IP addresses) to the table. For programming the port RoCE GID table, providers must implement the add_gid and del_gid callbacks. RoCE GID management requires us to state the associated net_device alongside the GID. This information is necessary in order to manage the GID table. For example, when a net_device is removed, its associated GIDs need to be removed as well. RoCE mandates generating a default GID for each port, based on the related net-device's IPv6 link local. In contrast to the GID based on the regular IPv6 link-local (as we generate GID per IP address), the default GID is also available when the net device is down (in order to support loopback). Locking is done as follows: The patch modify the GID table code both for new RoCE drivers implementing the add_gid/del_gid callbacks and for current RoCE and IB drivers that do not. The flows for updating the table are different, so the locking requirements are too. While updating RoCE GID table, protection against multiple writers is achieved via mutex_lock(&table->lock). Since writing to a table requires us to find an entry (possible a free entry) in the table and then modify it, this mutex protects both the find_gid and write_gid ensuring the atomicity of the action. Each entry in the GID cache is protected by rwlock. In RoCE, writing (usually results from netdev notifier) involves invoking the vendor's add_gid and del_gid callbacks, which could sleep. Therefore, an invalid flag is added for each entry. Updates for RoCE are done via a workqueue, thus sleeping is permitted. In IB, updates are done in write_lock_irq(&device->cache.lock), thus write_gid isn't allowed to sleep and add_gid/del_gid are not called. When passing net-device into/out-of the GID cache, the device is always passed held (dev_hold). The code uses a single work item for updating all RDMA devices, following a netdev or inet notifier. The patch moves the cache from being a client (which was incorrect, as the cache is part of the IB infrastructure) to being explicitly initialized/freed when a device is registered/removed. Signed-off-by: Matan Barak <matanb@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-07-30 23:33:26 +08:00
/**
* ib_enum_roce_netdev - enumerate all RoCE ports
* @ib_dev : IB device we want to query
* @filter: Should we call the callback?
* @filter_cookie: Cookie passed to filter
* @cb: Callback to call for each found RoCE ports
* @cookie: Cookie passed back to the callback
*
* Enumerates all of the physical RoCE ports of ib_dev
* which are related to netdevice and calls callback() on each
* device for which filter() function returns non zero.
*/
void ib_enum_roce_netdev(struct ib_device *ib_dev,
roce_netdev_filter filter,
void *filter_cookie,
roce_netdev_callback cb,
void *cookie)
{
u8 port;
for (port = rdma_start_port(ib_dev); port <= rdma_end_port(ib_dev);
port++)
if (rdma_protocol_roce(ib_dev, port)) {
struct net_device *idev = NULL;
if (ib_dev->get_netdev)
idev = ib_dev->get_netdev(ib_dev, port);
if (idev &&
idev->reg_state >= NETREG_UNREGISTERED) {
dev_put(idev);
idev = NULL;
}
if (filter(ib_dev, port, idev, filter_cookie))
cb(ib_dev, port, idev, cookie);
if (idev)
dev_put(idev);
}
}
/**
* ib_enum_all_roce_netdevs - enumerate all RoCE devices
* @filter: Should we call the callback?
* @filter_cookie: Cookie passed to filter
* @cb: Callback to call for each found RoCE ports
* @cookie: Cookie passed back to the callback
*
* Enumerates all RoCE devices' physical ports which are related
* to netdevices and calls callback() on each device for which
* filter() function returns non zero.
*/
void ib_enum_all_roce_netdevs(roce_netdev_filter filter,
void *filter_cookie,
roce_netdev_callback cb,
void *cookie)
{
struct ib_device *dev;
down_read(&lists_rwsem);
list_for_each_entry(dev, &device_list, core_list)
ib_enum_roce_netdev(dev, filter, filter_cookie, cb, cookie);
up_read(&lists_rwsem);
}
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_query_pkey - Get P_Key table entry
* @device:Device to query
* @port_num:Port number to query
* @index:P_Key table index to query
* @pkey:Returned P_Key
*
* ib_query_pkey() fetches the specified P_Key table entry.
*/
int ib_query_pkey(struct ib_device *device,
u8 port_num, u16 index, u16 *pkey)
{
return device->query_pkey(device, port_num, index, pkey);
}
EXPORT_SYMBOL(ib_query_pkey);
/**
* ib_modify_device - Change IB device attributes
* @device:Device to modify
* @device_modify_mask:Mask of attributes to change
* @device_modify:New attribute values
*
* ib_modify_device() changes a device's attributes as specified by
* the @device_modify_mask and @device_modify structure.
*/
int ib_modify_device(struct ib_device *device,
int device_modify_mask,
struct ib_device_modify *device_modify)
{
RDMA: Allow for NULL .modify_device() and .modify_port() methods These methods don't make sense for iWARP devices, so rather than forcing them to implement stubs, just return -ENOSYS in the core if the hardware driver doesn't set .modify_device and/or .modify_port. Signed-off-by: Roland Dreier <roland@purestorage.com>
2011-06-19 00:35:42 +08:00
if (!device->modify_device)
return -ENOSYS;
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 device->modify_device(device, device_modify_mask,
device_modify);
}
EXPORT_SYMBOL(ib_modify_device);
/**
* ib_modify_port - Modifies the attributes for the specified port.
* @device: The device to modify.
* @port_num: The number of the port to modify.
* @port_modify_mask: Mask used to specify which attributes of the port
* to change.
* @port_modify: New attribute values for the port.
*
* ib_modify_port() changes a port's attributes as specified by the
* @port_modify_mask and @port_modify structure.
*/
int ib_modify_port(struct ib_device *device,
u8 port_num, int port_modify_mask,
struct ib_port_modify *port_modify)
{
RDMA: Allow for NULL .modify_device() and .modify_port() methods These methods don't make sense for iWARP devices, so rather than forcing them to implement stubs, just return -ENOSYS in the core if the hardware driver doesn't set .modify_device and/or .modify_port. Signed-off-by: Roland Dreier <roland@purestorage.com>
2011-06-19 00:35:42 +08:00
if (!device->modify_port)
return -ENOSYS;
IB/core: Create common start/end port functions Previously start_port and end_port were defined in 2 places, cache.c and device.c and this prevented their use in other modules. Make these common functions, change the name to reflect the rdma name space, and update existing users. Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-05-14 08:02:55 +08:00
if (port_num < rdma_start_port(device) || port_num > rdma_end_port(device))
[IB] Check port number in ib_query_port()/ib_modify_port() Check port number before passing query_port or modify_port operations on to device driver. Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-04 00:32:33 +08:00
return -EINVAL;
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 device->modify_port(device, port_num, port_modify_mask,
port_modify);
}
EXPORT_SYMBOL(ib_modify_port);
IB/core: Add helpers for uncached GID and P_Key searches Add ib_find_gid() and ib_find_pkey() functions that use uncached device queries. The calls might block but the returns are always up-to-date. Cache P_Key and GID table lengths in core to avoid extra port info queries. Signed-off-by: Yosef Etigin <yosefe@voltaire.com> Acked-by: Michael S. Tsirkin <mst@dev.mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-05-14 12:26:51 +08:00
/**
* ib_find_gid - Returns the port number and GID table index where
* a specified GID value occurs.
* @device: The device to query.
* @gid: The GID value to search for.
IB/core: Add gid_type to gid attribute In order to support multiple GID types, we need to store the gid_type with each GID. This is also aligned with the RoCE v2 annex "RoCEv2 PORT GID table entries shall have a "GID type" attribute that denotes the L3 Address type". The currently supported GID is IB_GID_TYPE_IB which is also RoCE v1 GID type. This implies that gid_type should be added to roce_gid_table meta-data. Signed-off-by: Matan Barak <matanb@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-12-23 20:56:47 +08:00
* @gid_type: Type of GID.
IB/core: Add netdev and gid attributes paramteres to cache Adding an ability to query the IB cache by a netdev and get the attributes of a GID. These parameters are necessary in order to successfully resolve the required GID (when the netdevice is known) and get the Ethernet L2 attributes from a GID. Signed-off-by: Matan Barak <matanb@mellanox.com> Reviewed-By: Devesh Sharma <devesh.sharma@avagotech.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-10-15 23:38:45 +08:00
* @ndev: The ndev related to the GID to search for.
IB/core: Add helpers for uncached GID and P_Key searches Add ib_find_gid() and ib_find_pkey() functions that use uncached device queries. The calls might block but the returns are always up-to-date. Cache P_Key and GID table lengths in core to avoid extra port info queries. Signed-off-by: Yosef Etigin <yosefe@voltaire.com> Acked-by: Michael S. Tsirkin <mst@dev.mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-05-14 12:26:51 +08:00
* @port_num: The port number of the device where the GID value was found.
* @index: The index into the GID table where the GID was found. This
* parameter may be NULL.
*/
int ib_find_gid(struct ib_device *device, union ib_gid *gid,
IB/core: Add gid_type to gid attribute In order to support multiple GID types, we need to store the gid_type with each GID. This is also aligned with the RoCE v2 annex "RoCEv2 PORT GID table entries shall have a "GID type" attribute that denotes the L3 Address type". The currently supported GID is IB_GID_TYPE_IB which is also RoCE v1 GID type. This implies that gid_type should be added to roce_gid_table meta-data. Signed-off-by: Matan Barak <matanb@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-12-23 20:56:47 +08:00
enum ib_gid_type gid_type, struct net_device *ndev,
u8 *port_num, u16 *index)
IB/core: Add helpers for uncached GID and P_Key searches Add ib_find_gid() and ib_find_pkey() functions that use uncached device queries. The calls might block but the returns are always up-to-date. Cache P_Key and GID table lengths in core to avoid extra port info queries. Signed-off-by: Yosef Etigin <yosefe@voltaire.com> Acked-by: Michael S. Tsirkin <mst@dev.mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-05-14 12:26:51 +08:00
{
union ib_gid tmp_gid;
int ret, port, i;
IB/core: Create common start/end port functions Previously start_port and end_port were defined in 2 places, cache.c and device.c and this prevented their use in other modules. Make these common functions, change the name to reflect the rdma name space, and update existing users. Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-05-14 08:02:55 +08:00
for (port = rdma_start_port(device); port <= rdma_end_port(device); ++port) {
IB/core: Add RoCE GID table management RoCE GIDs are based on IP addresses configured on Ethernet net-devices which relate to the RDMA (RoCE) device port. Currently, each of the low-level drivers that support RoCE (ocrdma, mlx4) manages its own RoCE port GID table. As there's nothing which is essentially vendor specific, we generalize that, and enhance the RDMA core GID cache to do this job. In order to populate the GID table, we listen for events: (a) netdev up/down/change_addr events - if a netdev is built onto our RoCE device, we need to add/delete its IPs. This involves adding all GIDs related to this ndev, add default GIDs, etc. (b) inet events - add new GIDs (according to the IP addresses) to the table. For programming the port RoCE GID table, providers must implement the add_gid and del_gid callbacks. RoCE GID management requires us to state the associated net_device alongside the GID. This information is necessary in order to manage the GID table. For example, when a net_device is removed, its associated GIDs need to be removed as well. RoCE mandates generating a default GID for each port, based on the related net-device's IPv6 link local. In contrast to the GID based on the regular IPv6 link-local (as we generate GID per IP address), the default GID is also available when the net device is down (in order to support loopback). Locking is done as follows: The patch modify the GID table code both for new RoCE drivers implementing the add_gid/del_gid callbacks and for current RoCE and IB drivers that do not. The flows for updating the table are different, so the locking requirements are too. While updating RoCE GID table, protection against multiple writers is achieved via mutex_lock(&table->lock). Since writing to a table requires us to find an entry (possible a free entry) in the table and then modify it, this mutex protects both the find_gid and write_gid ensuring the atomicity of the action. Each entry in the GID cache is protected by rwlock. In RoCE, writing (usually results from netdev notifier) involves invoking the vendor's add_gid and del_gid callbacks, which could sleep. Therefore, an invalid flag is added for each entry. Updates for RoCE are done via a workqueue, thus sleeping is permitted. In IB, updates are done in write_lock_irq(&device->cache.lock), thus write_gid isn't allowed to sleep and add_gid/del_gid are not called. When passing net-device into/out-of the GID cache, the device is always passed held (dev_hold). The code uses a single work item for updating all RDMA devices, following a netdev or inet notifier. The patch moves the cache from being a client (which was incorrect, as the cache is part of the IB infrastructure) to being explicitly initialized/freed when a device is registered/removed. Signed-off-by: Matan Barak <matanb@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-07-30 23:33:26 +08:00
if (rdma_cap_roce_gid_table(device, port)) {
IB/core: Add gid_type to gid attribute In order to support multiple GID types, we need to store the gid_type with each GID. This is also aligned with the RoCE v2 annex "RoCEv2 PORT GID table entries shall have a "GID type" attribute that denotes the L3 Address type". The currently supported GID is IB_GID_TYPE_IB which is also RoCE v1 GID type. This implies that gid_type should be added to roce_gid_table meta-data. Signed-off-by: Matan Barak <matanb@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-12-23 20:56:47 +08:00
if (!ib_find_cached_gid_by_port(device, gid, gid_type, port,
IB/core: Expose and rename ib_find_cached_gid_by_port cache API Sometime consumers might want to search for a GID in a specific port. For example, when a WC arrives and we want to search the GID that matches that port - it's better to search only the relevant port. Exposing and renaming ib_cache_gid_find_by_port in order to match the naming convention of the module. Signed-off-by: Matan Barak <matanb@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-10-15 23:38:46 +08:00
ndev, index)) {
IB/core: Add RoCE GID table management RoCE GIDs are based on IP addresses configured on Ethernet net-devices which relate to the RDMA (RoCE) device port. Currently, each of the low-level drivers that support RoCE (ocrdma, mlx4) manages its own RoCE port GID table. As there's nothing which is essentially vendor specific, we generalize that, and enhance the RDMA core GID cache to do this job. In order to populate the GID table, we listen for events: (a) netdev up/down/change_addr events - if a netdev is built onto our RoCE device, we need to add/delete its IPs. This involves adding all GIDs related to this ndev, add default GIDs, etc. (b) inet events - add new GIDs (according to the IP addresses) to the table. For programming the port RoCE GID table, providers must implement the add_gid and del_gid callbacks. RoCE GID management requires us to state the associated net_device alongside the GID. This information is necessary in order to manage the GID table. For example, when a net_device is removed, its associated GIDs need to be removed as well. RoCE mandates generating a default GID for each port, based on the related net-device's IPv6 link local. In contrast to the GID based on the regular IPv6 link-local (as we generate GID per IP address), the default GID is also available when the net device is down (in order to support loopback). Locking is done as follows: The patch modify the GID table code both for new RoCE drivers implementing the add_gid/del_gid callbacks and for current RoCE and IB drivers that do not. The flows for updating the table are different, so the locking requirements are too. While updating RoCE GID table, protection against multiple writers is achieved via mutex_lock(&table->lock). Since writing to a table requires us to find an entry (possible a free entry) in the table and then modify it, this mutex protects both the find_gid and write_gid ensuring the atomicity of the action. Each entry in the GID cache is protected by rwlock. In RoCE, writing (usually results from netdev notifier) involves invoking the vendor's add_gid and del_gid callbacks, which could sleep. Therefore, an invalid flag is added for each entry. Updates for RoCE are done via a workqueue, thus sleeping is permitted. In IB, updates are done in write_lock_irq(&device->cache.lock), thus write_gid isn't allowed to sleep and add_gid/del_gid are not called. When passing net-device into/out-of the GID cache, the device is always passed held (dev_hold). The code uses a single work item for updating all RDMA devices, following a netdev or inet notifier. The patch moves the cache from being a client (which was incorrect, as the cache is part of the IB infrastructure) to being explicitly initialized/freed when a device is registered/removed. Signed-off-by: Matan Barak <matanb@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-07-30 23:33:26 +08:00
*port_num = port;
return 0;
IB/core: missing curly braces in ib_find_gid() Smatch says that, based on the indenting, we should probably add curly braces here. Fixes: 03db3a2d81e6 ('IB/core: Add RoCE GID table management') Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-18 17:22:10 +08:00
}
IB/core: Add RoCE GID table management RoCE GIDs are based on IP addresses configured on Ethernet net-devices which relate to the RDMA (RoCE) device port. Currently, each of the low-level drivers that support RoCE (ocrdma, mlx4) manages its own RoCE port GID table. As there's nothing which is essentially vendor specific, we generalize that, and enhance the RDMA core GID cache to do this job. In order to populate the GID table, we listen for events: (a) netdev up/down/change_addr events - if a netdev is built onto our RoCE device, we need to add/delete its IPs. This involves adding all GIDs related to this ndev, add default GIDs, etc. (b) inet events - add new GIDs (according to the IP addresses) to the table. For programming the port RoCE GID table, providers must implement the add_gid and del_gid callbacks. RoCE GID management requires us to state the associated net_device alongside the GID. This information is necessary in order to manage the GID table. For example, when a net_device is removed, its associated GIDs need to be removed as well. RoCE mandates generating a default GID for each port, based on the related net-device's IPv6 link local. In contrast to the GID based on the regular IPv6 link-local (as we generate GID per IP address), the default GID is also available when the net device is down (in order to support loopback). Locking is done as follows: The patch modify the GID table code both for new RoCE drivers implementing the add_gid/del_gid callbacks and for current RoCE and IB drivers that do not. The flows for updating the table are different, so the locking requirements are too. While updating RoCE GID table, protection against multiple writers is achieved via mutex_lock(&table->lock). Since writing to a table requires us to find an entry (possible a free entry) in the table and then modify it, this mutex protects both the find_gid and write_gid ensuring the atomicity of the action. Each entry in the GID cache is protected by rwlock. In RoCE, writing (usually results from netdev notifier) involves invoking the vendor's add_gid and del_gid callbacks, which could sleep. Therefore, an invalid flag is added for each entry. Updates for RoCE are done via a workqueue, thus sleeping is permitted. In IB, updates are done in write_lock_irq(&device->cache.lock), thus write_gid isn't allowed to sleep and add_gid/del_gid are not called. When passing net-device into/out-of the GID cache, the device is always passed held (dev_hold). The code uses a single work item for updating all RDMA devices, following a netdev or inet notifier. The patch moves the cache from being a client (which was incorrect, as the cache is part of the IB infrastructure) to being explicitly initialized/freed when a device is registered/removed. Signed-off-by: Matan Barak <matanb@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-07-30 23:33:26 +08:00
}
IB/core: Add gid_type to gid attribute In order to support multiple GID types, we need to store the gid_type with each GID. This is also aligned with the RoCE v2 annex "RoCEv2 PORT GID table entries shall have a "GID type" attribute that denotes the L3 Address type". The currently supported GID is IB_GID_TYPE_IB which is also RoCE v1 GID type. This implies that gid_type should be added to roce_gid_table meta-data. Signed-off-by: Matan Barak <matanb@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-12-23 20:56:47 +08:00
if (gid_type != IB_GID_TYPE_IB)
continue;
IB/core: Add per port immutable struct to ib_device As of commit 5eb620c81ce3 "IB/core: Add helpers for uncached GID and P_Key searches"; pkey_tbl_len and gid_tbl_len are immutable data which are stored in the ib_device. The per port core capability flags to be added later are also immutable data to be stored in the ib_device object. In preparation for this create a structure for per port immutable data and place the pkey and gid table lengths within this structure. "get_port_immutable" is added as a mandatory device function to allow the drivers to fill in this data. Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-05-14 08:02:58 +08:00
for (i = 0; i < device->port_immutable[port].gid_tbl_len; ++i) {
IB/core: Add netdev and gid attributes paramteres to cache Adding an ability to query the IB cache by a netdev and get the attributes of a GID. These parameters are necessary in order to successfully resolve the required GID (when the netdevice is known) and get the Ethernet L2 attributes from a GID. Signed-off-by: Matan Barak <matanb@mellanox.com> Reviewed-By: Devesh Sharma <devesh.sharma@avagotech.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-10-15 23:38:45 +08:00
ret = ib_query_gid(device, port, i, &tmp_gid, NULL);
IB/core: Add helpers for uncached GID and P_Key searches Add ib_find_gid() and ib_find_pkey() functions that use uncached device queries. The calls might block but the returns are always up-to-date. Cache P_Key and GID table lengths in core to avoid extra port info queries. Signed-off-by: Yosef Etigin <yosefe@voltaire.com> Acked-by: Michael S. Tsirkin <mst@dev.mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-05-14 12:26:51 +08:00
if (ret)
return ret;
if (!memcmp(&tmp_gid, gid, sizeof *gid)) {
*port_num = port;
if (index)
*index = i;
return 0;
}
}
}
return -ENOENT;
}
EXPORT_SYMBOL(ib_find_gid);
/**
* ib_find_pkey - Returns the PKey table index where a specified
* PKey value occurs.
* @device: The device to query.
* @port_num: The port number of the device to search for the PKey.
* @pkey: The PKey value to search for.
* @index: The index into the PKey table where the PKey was found.
*/
int ib_find_pkey(struct ib_device *device,
u8 port_num, u16 pkey, u16 *index)
{
int ret, i;
u16 tmp_pkey;
IB/core: Handle table with full and partial membership for the same P_Key Extend the cached and non-cached P_Key table lookups to handle limited and full membership of the same P_Key to co-exist in the P_Key table. This is necessary for SR-IOV, to allow for some guests would to have the full membership P_Key in their virtual P_Key table, while other guests on the same physical HCA would have the limited one. To support this, we need both the limited and full membership P_Keys to be present in the master's (hypervisor physical port) P_Key table. The algorithm for handling P_Key tables which contain both the limited and the full membership versions of the same P_Key works as follows: When scanning the P_Key table for a 15-bit P_Key: A. If there is a full member version of that P_Key anywhere in the table, return its index (even if a limited-member version of the P_Key exists earlier in the table). B. If the full member version is not in the table, but the limited-member version is in the table, return the index of the limited P_Key. Signed-off-by: Liran Liss <liranl@mellanox.com> Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il> Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com> Signed-off-by: Roland Dreier <roland@purestorage.com>
2012-08-03 16:40:38 +08:00
int partial_ix = -1;
IB/core: Add helpers for uncached GID and P_Key searches Add ib_find_gid() and ib_find_pkey() functions that use uncached device queries. The calls might block but the returns are always up-to-date. Cache P_Key and GID table lengths in core to avoid extra port info queries. Signed-off-by: Yosef Etigin <yosefe@voltaire.com> Acked-by: Michael S. Tsirkin <mst@dev.mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-05-14 12:26:51 +08:00
IB/core: Add per port immutable struct to ib_device As of commit 5eb620c81ce3 "IB/core: Add helpers for uncached GID and P_Key searches"; pkey_tbl_len and gid_tbl_len are immutable data which are stored in the ib_device. The per port core capability flags to be added later are also immutable data to be stored in the ib_device object. In preparation for this create a structure for per port immutable data and place the pkey and gid table lengths within this structure. "get_port_immutable" is added as a mandatory device function to allow the drivers to fill in this data. Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-05-14 08:02:58 +08:00
for (i = 0; i < device->port_immutable[port_num].pkey_tbl_len; ++i) {
IB/core: Add helpers for uncached GID and P_Key searches Add ib_find_gid() and ib_find_pkey() functions that use uncached device queries. The calls might block but the returns are always up-to-date. Cache P_Key and GID table lengths in core to avoid extra port info queries. Signed-off-by: Yosef Etigin <yosefe@voltaire.com> Acked-by: Michael S. Tsirkin <mst@dev.mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-05-14 12:26:51 +08:00
ret = ib_query_pkey(device, port_num, i, &tmp_pkey);
if (ret)
return ret;
IB/core: Ignore membership bit in ib_find_pkey() ib_find_pkey() is used as a replacement for ib_find_cached_pkey(), and the original function ignored the membership bit when searching for a P_Key, so ib_find_pkey() should ignore the bit too. In particular, IPoIB turns on the P_Key membership bit of limited membership P_Keys when creating a child interface and looks for the full membership P_key. This broke if a port was a partial member of a partition when IPoIB switched from ib_find_cached_pkey() to ib_find_pkey(), and this change fixes things again. Signed-off-by: Moni Shoua <monis@voltaire.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-07-23 15:07:42 +08:00
if ((pkey & 0x7fff) == (tmp_pkey & 0x7fff)) {
IB/core: Handle table with full and partial membership for the same P_Key Extend the cached and non-cached P_Key table lookups to handle limited and full membership of the same P_Key to co-exist in the P_Key table. This is necessary for SR-IOV, to allow for some guests would to have the full membership P_Key in their virtual P_Key table, while other guests on the same physical HCA would have the limited one. To support this, we need both the limited and full membership P_Keys to be present in the master's (hypervisor physical port) P_Key table. The algorithm for handling P_Key tables which contain both the limited and the full membership versions of the same P_Key works as follows: When scanning the P_Key table for a 15-bit P_Key: A. If there is a full member version of that P_Key anywhere in the table, return its index (even if a limited-member version of the P_Key exists earlier in the table). B. If the full member version is not in the table, but the limited-member version is in the table, return the index of the limited P_Key. Signed-off-by: Liran Liss <liranl@mellanox.com> Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il> Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com> Signed-off-by: Roland Dreier <roland@purestorage.com>
2012-08-03 16:40:38 +08:00
/* if there is full-member pkey take it.*/
if (tmp_pkey & 0x8000) {
*index = i;
return 0;
}
if (partial_ix < 0)
partial_ix = i;
IB/core: Add helpers for uncached GID and P_Key searches Add ib_find_gid() and ib_find_pkey() functions that use uncached device queries. The calls might block but the returns are always up-to-date. Cache P_Key and GID table lengths in core to avoid extra port info queries. Signed-off-by: Yosef Etigin <yosefe@voltaire.com> Acked-by: Michael S. Tsirkin <mst@dev.mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-05-14 12:26:51 +08:00
}
}
IB/core: Handle table with full and partial membership for the same P_Key Extend the cached and non-cached P_Key table lookups to handle limited and full membership of the same P_Key to co-exist in the P_Key table. This is necessary for SR-IOV, to allow for some guests would to have the full membership P_Key in their virtual P_Key table, while other guests on the same physical HCA would have the limited one. To support this, we need both the limited and full membership P_Keys to be present in the master's (hypervisor physical port) P_Key table. The algorithm for handling P_Key tables which contain both the limited and the full membership versions of the same P_Key works as follows: When scanning the P_Key table for a 15-bit P_Key: A. If there is a full member version of that P_Key anywhere in the table, return its index (even if a limited-member version of the P_Key exists earlier in the table). B. If the full member version is not in the table, but the limited-member version is in the table, return the index of the limited P_Key. Signed-off-by: Liran Liss <liranl@mellanox.com> Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il> Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com> Signed-off-by: Roland Dreier <roland@purestorage.com>
2012-08-03 16:40:38 +08:00
/*no full-member, if exists take the limited*/
if (partial_ix >= 0) {
*index = partial_ix;
return 0;
}
IB/core: Add helpers for uncached GID and P_Key searches Add ib_find_gid() and ib_find_pkey() functions that use uncached device queries. The calls might block but the returns are always up-to-date. Cache P_Key and GID table lengths in core to avoid extra port info queries. Signed-off-by: Yosef Etigin <yosefe@voltaire.com> Acked-by: Michael S. Tsirkin <mst@dev.mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-05-14 12:26:51 +08:00
return -ENOENT;
}
EXPORT_SYMBOL(ib_find_pkey);
IB/core: Find the network device matching connection parameters In the case of IPoIB, and maybe in other cases, the network device is managed by an upper-layer protocol (ULP). In order to expose this network device to other users of the IB device, let ULPs implement a callback that returns network device according to connection parameters. The IB device and port, together with the P_Key and the GID should be enough to uniquely identify the ULP net device. However, in current kernels there can be multiple IPoIB interfaces created with the same GID. Furthermore, such configuration may be desireable to support ipvlan-like configurations for RDMA CM with IPoIB. To resolve the device in these cases the code will also take the IP address as an additional input. Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Signed-off-by: Haggai Eran <haggaie@mellanox.com> Signed-off-by: Yotam Kenneth <yotamke@mellanox.com> Signed-off-by: Shachar Raindel <raindel@mellanox.com> Signed-off-by: Guy Shapiro <guysh@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-07-30 22:50:15 +08:00
/**
* ib_get_net_dev_by_params() - Return the appropriate net_dev
* for a received CM request
* @dev: An RDMA device on which the request has been received.
* @port: Port number on the RDMA device.
* @pkey: The Pkey the request came on.
* @gid: A GID that the net_dev uses to communicate.
* @addr: Contains the IP address that the request specified as its
* destination.
*/
struct net_device *ib_get_net_dev_by_params(struct ib_device *dev,
u8 port,
u16 pkey,
const union ib_gid *gid,
const struct sockaddr *addr)
{
struct net_device *net_dev = NULL;
struct ib_client_data *context;
if (!rdma_protocol_ib(dev, port))
return NULL;
down_read(&lists_rwsem);
list_for_each_entry(context, &dev->client_data_list, list) {
struct ib_client *client = context->client;
if (context->going_down)
continue;
if (client->get_net_dev_by_params) {
net_dev = client->get_net_dev_by_params(dev, port, pkey,
gid, addr,
context->data);
if (net_dev)
break;
}
}
up_read(&lists_rwsem);
return net_dev;
}
EXPORT_SYMBOL(ib_get_net_dev_by_params);
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 int __init ib_core_init(void)
{
int ret;
RDMA: Update workqueue usage * ib_wq is added, which is used as the common workqueue for infiniband instead of the system workqueue. All system workqueue usages including flush_scheduled_work() callers are converted to use and flush ib_wq. * cancel_delayed_work() + flush_scheduled_work() converted to cancel_delayed_work_sync(). * qib_wq is removed and ib_wq is used instead. This is to prepare for deprecation of flush_scheduled_work(). Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2010-10-19 23:24:36 +08:00
ib_wq = alloc_workqueue("infiniband", 0, 0);
if (!ib_wq)
return -ENOMEM;
IB: add a proper completion queue abstraction This adds an abstraction that allows ULPs to simply pass a completion object and completion callback with each submitted WR and let the RDMA core handle the nitty gritty details of how to handle completion interrupts and poll the CQ. In detail there is a new ib_cqe structure which just contains the completion callback, and which can be used to get at the containing object using container_of. It is pointed to by the WR and WC as an alternative to the wr_id field, similar to how many ULPs already use the field to store a pointer using casts. A driver using the new completion callbacks allocates it's CQs using the new ib_create_cq API, which in addition to the number of CQEs and the completion vectors also takes a mode on how we poll for CQEs. Three modes are available: direct for drivers that never take CQ interrupts and just poll for them, softirq to poll from softirq context using the to be renamed blk-iopoll infrastructure which takes care of rearming and budgeting, or a workqueue for consumer who want to be called from user context. Thanks a lot to Sagi Grimberg who helped reviewing the API, wrote the current version of the workqueue code because my two previous attempts sucked too much and converted the iSER initiator to the new API. Signed-off-by: Christoph Hellwig <hch@lst.de>
2015-12-12 03:53:03 +08:00
ib_comp_wq = alloc_workqueue("ib-comp-wq",
WQ_UNBOUND | WQ_HIGHPRI | WQ_MEM_RECLAIM,
WQ_UNBOUND_MAX_ACTIVE);
if (!ib_comp_wq) {
ret = -ENOMEM;
goto err;
}
IB/core: Make ib_alloc_device init the kobject This gets rid of the weird in-between state where struct ib_device was allocated but the kobject didn't work. Consequently ib_device_release is now guaranteed to be called in all situations and we needn't duplicate its kfrees on error paths. Signed-off-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-05 05:23:34 +08:00
ret = class_register(&ib_class);
RDMA: Add error handling to ib_core_init() Fail RDMA midlayer initialization if sysfs setup fails. Signed-off-by: Nir Muchtar <nirm@voltaire.com> Signed-off-by: Roland Dreier <roland@purestorage.com>
2011-05-21 02:46:10 +08:00
if (ret) {
IB/core: trivial prink cleanup. 1. Replaced printk with appropriate pr_warn, pr_err, pr_info. 2. Removed unnecessary prints around memory allocation failure which are not required, as reported by the checkpatch script. Signed-off-by: Parav Pandit <pandit.parav@gmail.com> Reviewed-by: Haggai Eran <haggaie@mellanox.com> Reviewed-by: Sagi Grimberg <sagig@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-03-02 03:20:29 +08:00
pr_warn("Couldn't create InfiniBand device class\n");
IB: add a proper completion queue abstraction This adds an abstraction that allows ULPs to simply pass a completion object and completion callback with each submitted WR and let the RDMA core handle the nitty gritty details of how to handle completion interrupts and poll the CQ. In detail there is a new ib_cqe structure which just contains the completion callback, and which can be used to get at the containing object using container_of. It is pointed to by the WR and WC as an alternative to the wr_id field, similar to how many ULPs already use the field to store a pointer using casts. A driver using the new completion callbacks allocates it's CQs using the new ib_create_cq API, which in addition to the number of CQEs and the completion vectors also takes a mode on how we poll for CQEs. Three modes are available: direct for drivers that never take CQ interrupts and just poll for them, softirq to poll from softirq context using the to be renamed blk-iopoll infrastructure which takes care of rearming and budgeting, or a workqueue for consumer who want to be called from user context. Thanks a lot to Sagi Grimberg who helped reviewing the API, wrote the current version of the workqueue code because my two previous attempts sucked too much and converted the iSER initiator to the new API. Signed-off-by: Christoph Hellwig <hch@lst.de>
2015-12-12 03:53:03 +08:00
goto err_comp;
RDMA: Add error handling to ib_core_init() Fail RDMA midlayer initialization if sysfs setup fails. Signed-off-by: Nir Muchtar <nirm@voltaire.com> Signed-off-by: Roland Dreier <roland@purestorage.com>
2011-05-21 02:46:10 +08:00
}
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
RDMA: Add netlink infrastructure Add basic RDMA netlink infrastructure that allows for registration of RDMA clients for which data is to be exported and supplies message construction callbacks. Signed-off-by: Nir Muchtar <nirm@voltaire.com> [ Reorganize a few things, add CONFIG_NET dependency. - Roland ] Signed-off-by: Roland Dreier <roland@purestorage.com>
2011-05-21 02:46:11 +08:00
ret = ibnl_init();
if (ret) {
IB/core: trivial prink cleanup. 1. Replaced printk with appropriate pr_warn, pr_err, pr_info. 2. Removed unnecessary prints around memory allocation failure which are not required, as reported by the checkpatch script. Signed-off-by: Parav Pandit <pandit.parav@gmail.com> Reviewed-by: Haggai Eran <haggaie@mellanox.com> Reviewed-by: Sagi Grimberg <sagig@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-03-02 03:20:29 +08:00
pr_warn("Couldn't init IB netlink interface\n");
RDMA: Add netlink infrastructure Add basic RDMA netlink infrastructure that allows for registration of RDMA clients for which data is to be exported and supplies message construction callbacks. Signed-off-by: Nir Muchtar <nirm@voltaire.com> [ Reorganize a few things, add CONFIG_NET dependency. - Roland ] Signed-off-by: Roland Dreier <roland@purestorage.com>
2011-05-21 02:46:11 +08:00
goto err_sysfs;
}
IB/core: Integrate IB address resolution module into core IB address resolution is declared as a module (ib_addr.ko) which loads itself before IB core module (ib_core.ko). It causes to the scenario where IB netlink which is initialized by IB core can't be used by ib_addr.ko. In order to solve it, we are converting ib_addr.ko to be part of IB core module. Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Mark Bloch <markb@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-05-19 22:12:31 +08:00
ret = addr_init();
if (ret) {
pr_warn("Could't init IB address resolution\n");
goto err_ibnl;
}
IB/core: Add RoCE GID table management RoCE GIDs are based on IP addresses configured on Ethernet net-devices which relate to the RDMA (RoCE) device port. Currently, each of the low-level drivers that support RoCE (ocrdma, mlx4) manages its own RoCE port GID table. As there's nothing which is essentially vendor specific, we generalize that, and enhance the RDMA core GID cache to do this job. In order to populate the GID table, we listen for events: (a) netdev up/down/change_addr events - if a netdev is built onto our RoCE device, we need to add/delete its IPs. This involves adding all GIDs related to this ndev, add default GIDs, etc. (b) inet events - add new GIDs (according to the IP addresses) to the table. For programming the port RoCE GID table, providers must implement the add_gid and del_gid callbacks. RoCE GID management requires us to state the associated net_device alongside the GID. This information is necessary in order to manage the GID table. For example, when a net_device is removed, its associated GIDs need to be removed as well. RoCE mandates generating a default GID for each port, based on the related net-device's IPv6 link local. In contrast to the GID based on the regular IPv6 link-local (as we generate GID per IP address), the default GID is also available when the net device is down (in order to support loopback). Locking is done as follows: The patch modify the GID table code both for new RoCE drivers implementing the add_gid/del_gid callbacks and for current RoCE and IB drivers that do not. The flows for updating the table are different, so the locking requirements are too. While updating RoCE GID table, protection against multiple writers is achieved via mutex_lock(&table->lock). Since writing to a table requires us to find an entry (possible a free entry) in the table and then modify it, this mutex protects both the find_gid and write_gid ensuring the atomicity of the action. Each entry in the GID cache is protected by rwlock. In RoCE, writing (usually results from netdev notifier) involves invoking the vendor's add_gid and del_gid callbacks, which could sleep. Therefore, an invalid flag is added for each entry. Updates for RoCE are done via a workqueue, thus sleeping is permitted. In IB, updates are done in write_lock_irq(&device->cache.lock), thus write_gid isn't allowed to sleep and add_gid/del_gid are not called. When passing net-device into/out-of the GID cache, the device is always passed held (dev_hold). The code uses a single work item for updating all RDMA devices, following a netdev or inet notifier. The patch moves the cache from being a client (which was incorrect, as the cache is part of the IB infrastructure) to being explicitly initialized/freed when a device is registered/removed. Signed-off-by: Matan Barak <matanb@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-07-30 23:33:26 +08:00
ib_cache_setup();
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
RDMA: Add error handling to ib_core_init() Fail RDMA midlayer initialization if sysfs setup fails. Signed-off-by: Nir Muchtar <nirm@voltaire.com> Signed-off-by: Roland Dreier <roland@purestorage.com>
2011-05-21 02:46:10 +08:00
return 0;
IB/core: Integrate IB address resolution module into core IB address resolution is declared as a module (ib_addr.ko) which loads itself before IB core module (ib_core.ko). It causes to the scenario where IB netlink which is initialized by IB core can't be used by ib_addr.ko. In order to solve it, we are converting ib_addr.ko to be part of IB core module. Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Mark Bloch <markb@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-05-19 22:12:31 +08:00
err_ibnl:
ibnl_cleanup();
RDMA: Add error handling to ib_core_init() Fail RDMA midlayer initialization if sysfs setup fails. Signed-off-by: Nir Muchtar <nirm@voltaire.com> Signed-off-by: Roland Dreier <roland@purestorage.com>
2011-05-21 02:46:10 +08:00
err_sysfs:
IB/core: Make ib_alloc_device init the kobject This gets rid of the weird in-between state where struct ib_device was allocated but the kobject didn't work. Consequently ib_device_release is now guaranteed to be called in all situations and we needn't duplicate its kfrees on error paths. Signed-off-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-05 05:23:34 +08:00
class_unregister(&ib_class);
IB: add a proper completion queue abstraction This adds an abstraction that allows ULPs to simply pass a completion object and completion callback with each submitted WR and let the RDMA core handle the nitty gritty details of how to handle completion interrupts and poll the CQ. In detail there is a new ib_cqe structure which just contains the completion callback, and which can be used to get at the containing object using container_of. It is pointed to by the WR and WC as an alternative to the wr_id field, similar to how many ULPs already use the field to store a pointer using casts. A driver using the new completion callbacks allocates it's CQs using the new ib_create_cq API, which in addition to the number of CQEs and the completion vectors also takes a mode on how we poll for CQEs. Three modes are available: direct for drivers that never take CQ interrupts and just poll for them, softirq to poll from softirq context using the to be renamed blk-iopoll infrastructure which takes care of rearming and budgeting, or a workqueue for consumer who want to be called from user context. Thanks a lot to Sagi Grimberg who helped reviewing the API, wrote the current version of the workqueue code because my two previous attempts sucked too much and converted the iSER initiator to the new API. Signed-off-by: Christoph Hellwig <hch@lst.de>
2015-12-12 03:53:03 +08:00
err_comp:
destroy_workqueue(ib_comp_wq);
RDMA: Add error handling to ib_core_init() Fail RDMA midlayer initialization if sysfs setup fails. Signed-off-by: Nir Muchtar <nirm@voltaire.com> Signed-off-by: Roland Dreier <roland@purestorage.com>
2011-05-21 02:46:10 +08:00
err:
destroy_workqueue(ib_wq);
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 ret;
}
static void __exit ib_core_cleanup(void)
{
ib_cache_cleanup();
IB/core: Integrate IB address resolution module into core IB address resolution is declared as a module (ib_addr.ko) which loads itself before IB core module (ib_core.ko). It causes to the scenario where IB netlink which is initialized by IB core can't be used by ib_addr.ko. In order to solve it, we are converting ib_addr.ko to be part of IB core module. Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Mark Bloch <markb@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-05-19 22:12:31 +08:00
addr_cleanup();
RDMA: Add netlink infrastructure Add basic RDMA netlink infrastructure that allows for registration of RDMA clients for which data is to be exported and supplies message construction callbacks. Signed-off-by: Nir Muchtar <nirm@voltaire.com> [ Reorganize a few things, add CONFIG_NET dependency. - Roland ] Signed-off-by: Roland Dreier <roland@purestorage.com>
2011-05-21 02:46:11 +08:00
ibnl_cleanup();
IB/core: Make ib_alloc_device init the kobject This gets rid of the weird in-between state where struct ib_device was allocated but the kobject didn't work. Consequently ib_device_release is now guaranteed to be called in all situations and we needn't duplicate its kfrees on error paths. Signed-off-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-05 05:23:34 +08:00
class_unregister(&ib_class);
IB: add a proper completion queue abstraction This adds an abstraction that allows ULPs to simply pass a completion object and completion callback with each submitted WR and let the RDMA core handle the nitty gritty details of how to handle completion interrupts and poll the CQ. In detail there is a new ib_cqe structure which just contains the completion callback, and which can be used to get at the containing object using container_of. It is pointed to by the WR and WC as an alternative to the wr_id field, similar to how many ULPs already use the field to store a pointer using casts. A driver using the new completion callbacks allocates it's CQs using the new ib_create_cq API, which in addition to the number of CQEs and the completion vectors also takes a mode on how we poll for CQEs. Three modes are available: direct for drivers that never take CQ interrupts and just poll for them, softirq to poll from softirq context using the to be renamed blk-iopoll infrastructure which takes care of rearming and budgeting, or a workqueue for consumer who want to be called from user context. Thanks a lot to Sagi Grimberg who helped reviewing the API, wrote the current version of the workqueue code because my two previous attempts sucked too much and converted the iSER initiator to the new API. Signed-off-by: Christoph Hellwig <hch@lst.de>
2015-12-12 03:53:03 +08:00
destroy_workqueue(ib_comp_wq);
IB/uverbs: Export ib_umem_get()/ib_umem_release() to modules Export ib_umem_get()/ib_umem_release() and put low-level drivers in control of when to call ib_umem_get() to pin and DMA map userspace, rather than always calling it in ib_uverbs_reg_mr() before calling the low-level driver's reg_user_mr method. Also move these functions to be in the ib_core module instead of ib_uverbs, so that driver modules using them do not depend on ib_uverbs. This has a number of advantages: - It is better design from the standpoint of making generic code a library that can be used or overridden by device-specific code as the details of specific devices dictate. - Drivers that do not need to pin userspace memory regions do not need to take the performance hit of calling ib_mem_get(). For example, although I have not tried to implement it in this patch, the ipath driver should be able to avoid pinning memory and just use copy_{to,from}_user() to access userspace memory regions. - Buffers that need special mapping treatment can be identified by the low-level driver. For example, it may be possible to solve some Altix-specific memory ordering issues with mthca CQs in userspace by mapping CQ buffers with extra flags. - Drivers that need to pin and DMA map userspace memory for things other than memory regions can use ib_umem_get() directly, instead of hacks using extra parameters to their reg_phys_mr method. For example, the mlx4 driver that is pending being merged needs to pin and DMA map QP and CQ buffers, but it does not need to create a memory key for these buffers. So the cleanest solution is for mlx4 to call ib_umem_get() in the create_qp and create_cq methods. Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-03-05 08:15:11 +08:00
/* Make sure that any pending umem accounting work is done. */
RDMA: Update workqueue usage * ib_wq is added, which is used as the common workqueue for infiniband instead of the system workqueue. All system workqueue usages including flush_scheduled_work() callers are converted to use and flush ib_wq. * cancel_delayed_work() + flush_scheduled_work() converted to cancel_delayed_work_sync(). * qib_wq is removed and ib_wq is used instead. This is to prepare for deprecation of flush_scheduled_work(). Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2010-10-19 23:24:36 +08:00
destroy_workqueue(ib_wq);
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
}
module_init(ib_core_init);
module_exit(ib_core_cleanup);