WSL2-Linux-Kernel/drivers/infiniband/core/multicast.c

907 строки
24 KiB
C
Исходник Обычный вид История

/*
* Copyright (c) 2006 Intel Corporation. All rights reserved.
*
* This software is available to you under a choice of one of two
* licenses. You may choose to be licensed under the terms of the GNU
* General Public License (GPL) Version 2, available from the file
* COPYING in the main directory of this source tree, or the
* OpenIB.org BSD license below:
*
* Redistribution and use in source and binary forms, with or
* without modification, are permitted provided that the following
* conditions are met:
*
* - Redistributions of source code must retain the above
* copyright notice, this list of conditions and the following
* disclaimer.
*
* - Redistributions in binary form must reproduce the above
* copyright notice, this list of conditions and the following
* disclaimer in the documentation and/or other materials
* provided with the distribution.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
* NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
* BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
* ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
* CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
* SOFTWARE.
*/
#include <linux/completion.h>
#include <linux/dma-mapping.h>
#include <linux/err.h>
#include <linux/interrupt.h>
#include <linux/export.h>
include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h percpu.h is included by sched.h and module.h and thus ends up being included when building most .c files. percpu.h includes slab.h which in turn includes gfp.h making everything defined by the two files universally available and complicating inclusion dependencies. percpu.h -> slab.h dependency is about to be removed. Prepare for this change by updating users of gfp and slab facilities include those headers directly instead of assuming availability. As this conversion needs to touch large number of source files, the following script is used as the basis of conversion. http://userweb.kernel.org/~tj/misc/slabh-sweep.py The script does the followings. * Scan files for gfp and slab usages and update includes such that only the necessary includes are there. ie. if only gfp is used, gfp.h, if slab is used, slab.h. * When the script inserts a new include, it looks at the include blocks and try to put the new include such that its order conforms to its surrounding. It's put in the include block which contains core kernel includes, in the same order that the rest are ordered - alphabetical, Christmas tree, rev-Xmas-tree or at the end if there doesn't seem to be any matching order. * If the script can't find a place to put a new include (mostly because the file doesn't have fitting include block), it prints out an error message indicating which .h file needs to be added to the file. The conversion was done in the following steps. 1. The initial automatic conversion of all .c files updated slightly over 4000 files, deleting around 700 includes and adding ~480 gfp.h and ~3000 slab.h inclusions. The script emitted errors for ~400 files. 2. Each error was manually checked. Some didn't need the inclusion, some needed manual addition while adding it to implementation .h or embedding .c file was more appropriate for others. This step added inclusions to around 150 files. 3. The script was run again and the output was compared to the edits from #2 to make sure no file was left behind. 4. Several build tests were done and a couple of problems were fixed. e.g. lib/decompress_*.c used malloc/free() wrappers around slab APIs requiring slab.h to be added manually. 5. The script was run on all .h files but without automatically editing them as sprinkling gfp.h and slab.h inclusions around .h files could easily lead to inclusion dependency hell. Most gfp.h inclusion directives were ignored as stuff from gfp.h was usually wildly available and often used in preprocessor macros. Each slab.h inclusion directive was examined and added manually as necessary. 6. percpu.h was updated not to include slab.h. 7. Build test were done on the following configurations and failures were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my distributed build env didn't work with gcov compiles) and a few more options had to be turned off depending on archs to make things build (like ipr on powerpc/64 which failed due to missing writeq). * x86 and x86_64 UP and SMP allmodconfig and a custom test config. * powerpc and powerpc64 SMP allmodconfig * sparc and sparc64 SMP allmodconfig * ia64 SMP allmodconfig * s390 SMP allmodconfig * alpha SMP allmodconfig * um on x86_64 SMP allmodconfig 8. percpu.h modifications were reverted so that it could be applied as a separate patch and serve as bisection point. Given the fact that I had only a couple of failures from tests on step 6, I'm fairly confident about the coverage of this conversion patch. If there is a breakage, it's likely to be something in one of the arch headers which should be easily discoverable easily on most builds of the specific arch. Signed-off-by: Tejun Heo <tj@kernel.org> Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org> Cc: Ingo Molnar <mingo@redhat.com> Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
2010-03-24 11:04:11 +03:00
#include <linux/slab.h>
#include <linux/bitops.h>
#include <linux/random.h>
#include <rdma/ib_cache.h>
#include "sa.h"
static int mcast_add_one(struct ib_device *device);
static void mcast_remove_one(struct ib_device *device, void *client_data);
static struct ib_client mcast_client = {
.name = "ib_multicast",
.add = mcast_add_one,
.remove = mcast_remove_one
};
static struct ib_sa_client sa_client;
static struct workqueue_struct *mcast_wq;
static union ib_gid mgid0;
struct mcast_device;
struct mcast_port {
struct mcast_device *dev;
spinlock_t lock;
struct rb_root table;
refcount_t refcount;
struct completion comp;
RDMA: Support more than 255 rdma ports Current code uses many different types when dealing with a port of a RDMA device: u8, unsigned int and u32. Switch to u32 to clean up the logic. This allows us to make (at least) the core view consistent and use the same type. Unfortunately not all places can be converted. Many uverbs functions expect port to be u8 so keep those places in order not to break UAPIs. HW/Spec defined values must also not be changed. With the switch to u32 we now can support devices with more than 255 ports. U32_MAX is reserved to make control logic a bit easier to deal with. As a device with U32_MAX ports probably isn't going to happen any time soon this seems like a non issue. When a device with more than 255 ports is created uverbs will report the RDMA device as having 255 ports as this is the max currently supported. The verbs interface is not changed yet because the IBTA spec limits the port size in too many places to be u8 and all applications that relies in verbs won't be able to cope with this change. At this stage, we are extending the interfaces that are using vendor channel solely Once the limitation is lifted mlx5 in switchdev mode will be able to have thousands of SFs created by the device. As the only instance of an RDMA device that reports more than 255 ports will be a representor device and it exposes itself as a RAW Ethernet only device CM/MAD/IPoIB and other ULPs aren't effected by this change and their sysfs/interfaces that are exposes to userspace can remain unchanged. While here cleanup some alignment issues and remove unneeded sanity checks (mainly in rdmavt), Link: https://lore.kernel.org/r/20210301070420.439400-1-leon@kernel.org Signed-off-by: Mark Bloch <mbloch@nvidia.com> Signed-off-by: Leon Romanovsky <leonro@nvidia.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2021-03-01 10:04:20 +03:00
u32 port_num;
};
struct mcast_device {
struct ib_device *device;
struct ib_event_handler event_handler;
int start_port;
int end_port;
struct mcast_port port[];
};
enum mcast_state {
MCAST_JOINING,
MCAST_MEMBER,
MCAST_ERROR,
};
enum mcast_group_state {
MCAST_IDLE,
MCAST_BUSY,
MCAST_GROUP_ERROR,
MCAST_PKEY_EVENT
};
enum {
MCAST_INVALID_PKEY_INDEX = 0xFFFF
};
struct mcast_member;
struct mcast_group {
struct ib_sa_mcmember_rec rec;
struct rb_node node;
struct mcast_port *port;
spinlock_t lock;
struct work_struct work;
struct list_head pending_list;
struct list_head active_list;
struct mcast_member *last_join;
int members[NUM_JOIN_MEMBERSHIP_TYPES];
atomic_t refcount;
enum mcast_group_state state;
struct ib_sa_query *query;
u16 pkey_index;
u8 leave_state;
int retries;
};
struct mcast_member {
struct ib_sa_multicast multicast;
struct ib_sa_client *client;
struct mcast_group *group;
struct list_head list;
enum mcast_state state;
refcount_t refcount;
struct completion comp;
};
static void join_handler(int status, struct ib_sa_mcmember_rec *rec,
void *context);
static void leave_handler(int status, struct ib_sa_mcmember_rec *rec,
void *context);
static struct mcast_group *mcast_find(struct mcast_port *port,
union ib_gid *mgid)
{
struct rb_node *node = port->table.rb_node;
struct mcast_group *group;
int ret;
while (node) {
group = rb_entry(node, struct mcast_group, node);
ret = memcmp(mgid->raw, group->rec.mgid.raw, sizeof *mgid);
if (!ret)
return group;
if (ret < 0)
node = node->rb_left;
else
node = node->rb_right;
}
return NULL;
}
static struct mcast_group *mcast_insert(struct mcast_port *port,
struct mcast_group *group,
int allow_duplicates)
{
struct rb_node **link = &port->table.rb_node;
struct rb_node *parent = NULL;
struct mcast_group *cur_group;
int ret;
while (*link) {
parent = *link;
cur_group = rb_entry(parent, struct mcast_group, node);
ret = memcmp(group->rec.mgid.raw, cur_group->rec.mgid.raw,
sizeof group->rec.mgid);
if (ret < 0)
link = &(*link)->rb_left;
else if (ret > 0)
link = &(*link)->rb_right;
else if (allow_duplicates)
link = &(*link)->rb_left;
else
return cur_group;
}
rb_link_node(&group->node, parent, link);
rb_insert_color(&group->node, &port->table);
return NULL;
}
static void deref_port(struct mcast_port *port)
{
if (refcount_dec_and_test(&port->refcount))
complete(&port->comp);
}
static void release_group(struct mcast_group *group)
{
struct mcast_port *port = group->port;
unsigned long flags;
spin_lock_irqsave(&port->lock, flags);
if (atomic_dec_and_test(&group->refcount)) {
rb_erase(&group->node, &port->table);
spin_unlock_irqrestore(&port->lock, flags);
kfree(group);
deref_port(port);
} else
spin_unlock_irqrestore(&port->lock, flags);
}
static void deref_member(struct mcast_member *member)
{
if (refcount_dec_and_test(&member->refcount))
complete(&member->comp);
}
static void queue_join(struct mcast_member *member)
{
struct mcast_group *group = member->group;
unsigned long flags;
spin_lock_irqsave(&group->lock, flags);
list_add_tail(&member->list, &group->pending_list);
if (group->state == MCAST_IDLE) {
group->state = MCAST_BUSY;
atomic_inc(&group->refcount);
queue_work(mcast_wq, &group->work);
}
spin_unlock_irqrestore(&group->lock, flags);
}
/*
* A multicast group has four types of members: full member, non member,
* sendonly non member and sendonly full member.
* We need to keep track of the number of members of each
* type based on their join state. Adjust the number of members the belong to
* the specified join states.
*/
static void adjust_membership(struct mcast_group *group, u8 join_state, int inc)
{
int i;
for (i = 0; i < NUM_JOIN_MEMBERSHIP_TYPES; i++, join_state >>= 1)
if (join_state & 0x1)
group->members[i] += inc;
}
/*
* If a multicast group has zero members left for a particular join state, but
* the group is still a member with the SA, we need to leave that join state.
* Determine which join states we still belong to, but that do not have any
* active members.
*/
static u8 get_leave_state(struct mcast_group *group)
{
u8 leave_state = 0;
int i;
for (i = 0; i < NUM_JOIN_MEMBERSHIP_TYPES; i++)
if (!group->members[i])
leave_state |= (0x1 << i);
return leave_state & group->rec.join_state;
}
static int check_selector(ib_sa_comp_mask comp_mask,
ib_sa_comp_mask selector_mask,
ib_sa_comp_mask value_mask,
u8 selector, u8 src_value, u8 dst_value)
{
int err;
if (!(comp_mask & selector_mask) || !(comp_mask & value_mask))
return 0;
switch (selector) {
case IB_SA_GT:
err = (src_value <= dst_value);
break;
case IB_SA_LT:
err = (src_value >= dst_value);
break;
case IB_SA_EQ:
err = (src_value != dst_value);
break;
default:
err = 0;
break;
}
return err;
}
static int cmp_rec(struct ib_sa_mcmember_rec *src,
struct ib_sa_mcmember_rec *dst, ib_sa_comp_mask comp_mask)
{
/* MGID must already match */
if (comp_mask & IB_SA_MCMEMBER_REC_PORT_GID &&
memcmp(&src->port_gid, &dst->port_gid, sizeof src->port_gid))
return -EINVAL;
if (comp_mask & IB_SA_MCMEMBER_REC_QKEY && src->qkey != dst->qkey)
return -EINVAL;
if (comp_mask & IB_SA_MCMEMBER_REC_MLID && src->mlid != dst->mlid)
return -EINVAL;
if (check_selector(comp_mask, IB_SA_MCMEMBER_REC_MTU_SELECTOR,
IB_SA_MCMEMBER_REC_MTU, dst->mtu_selector,
src->mtu, dst->mtu))
return -EINVAL;
if (comp_mask & IB_SA_MCMEMBER_REC_TRAFFIC_CLASS &&
src->traffic_class != dst->traffic_class)
return -EINVAL;
if (comp_mask & IB_SA_MCMEMBER_REC_PKEY && src->pkey != dst->pkey)
return -EINVAL;
if (check_selector(comp_mask, IB_SA_MCMEMBER_REC_RATE_SELECTOR,
IB_SA_MCMEMBER_REC_RATE, dst->rate_selector,
src->rate, dst->rate))
return -EINVAL;
if (check_selector(comp_mask,
IB_SA_MCMEMBER_REC_PACKET_LIFE_TIME_SELECTOR,
IB_SA_MCMEMBER_REC_PACKET_LIFE_TIME,
dst->packet_life_time_selector,
src->packet_life_time, dst->packet_life_time))
return -EINVAL;
if (comp_mask & IB_SA_MCMEMBER_REC_SL && src->sl != dst->sl)
return -EINVAL;
if (comp_mask & IB_SA_MCMEMBER_REC_FLOW_LABEL &&
src->flow_label != dst->flow_label)
return -EINVAL;
if (comp_mask & IB_SA_MCMEMBER_REC_HOP_LIMIT &&
src->hop_limit != dst->hop_limit)
return -EINVAL;
if (comp_mask & IB_SA_MCMEMBER_REC_SCOPE && src->scope != dst->scope)
return -EINVAL;
/* join_state checked separately, proxy_join ignored */
return 0;
}
static int send_join(struct mcast_group *group, struct mcast_member *member)
{
struct mcast_port *port = group->port;
int ret;
group->last_join = member;
ret = ib_sa_mcmember_rec_query(&sa_client, port->dev->device,
port->port_num, IB_MGMT_METHOD_SET,
&member->multicast.rec,
member->multicast.comp_mask,
3000, GFP_KERNEL, join_handler, group,
&group->query);
return (ret > 0) ? 0 : ret;
}
static int send_leave(struct mcast_group *group, u8 leave_state)
{
struct mcast_port *port = group->port;
struct ib_sa_mcmember_rec rec;
int ret;
rec = group->rec;
rec.join_state = leave_state;
group->leave_state = leave_state;
ret = ib_sa_mcmember_rec_query(&sa_client, port->dev->device,
port->port_num, IB_SA_METHOD_DELETE, &rec,
IB_SA_MCMEMBER_REC_MGID |
IB_SA_MCMEMBER_REC_PORT_GID |
IB_SA_MCMEMBER_REC_JOIN_STATE,
3000, GFP_KERNEL, leave_handler,
group, &group->query);
return (ret > 0) ? 0 : ret;
}
static void join_group(struct mcast_group *group, struct mcast_member *member,
u8 join_state)
{
member->state = MCAST_MEMBER;
adjust_membership(group, join_state, 1);
group->rec.join_state |= join_state;
member->multicast.rec = group->rec;
member->multicast.rec.join_state = join_state;
list_move(&member->list, &group->active_list);
}
static int fail_join(struct mcast_group *group, struct mcast_member *member,
int status)
{
spin_lock_irq(&group->lock);
list_del_init(&member->list);
spin_unlock_irq(&group->lock);
return member->multicast.callback(status, &member->multicast);
}
static void process_group_error(struct mcast_group *group)
{
struct mcast_member *member;
int ret = 0;
u16 pkey_index;
if (group->state == MCAST_PKEY_EVENT)
ret = ib_find_pkey(group->port->dev->device,
group->port->port_num,
be16_to_cpu(group->rec.pkey), &pkey_index);
spin_lock_irq(&group->lock);
if (group->state == MCAST_PKEY_EVENT && !ret &&
group->pkey_index == pkey_index)
goto out;
while (!list_empty(&group->active_list)) {
member = list_entry(group->active_list.next,
struct mcast_member, list);
refcount_inc(&member->refcount);
list_del_init(&member->list);
adjust_membership(group, member->multicast.rec.join_state, -1);
member->state = MCAST_ERROR;
spin_unlock_irq(&group->lock);
ret = member->multicast.callback(-ENETRESET,
&member->multicast);
deref_member(member);
if (ret)
ib_sa_free_multicast(&member->multicast);
spin_lock_irq(&group->lock);
}
group->rec.join_state = 0;
out:
group->state = MCAST_BUSY;
spin_unlock_irq(&group->lock);
}
static void mcast_work_handler(struct work_struct *work)
{
struct mcast_group *group;
struct mcast_member *member;
struct ib_sa_multicast *multicast;
int status, ret;
u8 join_state;
group = container_of(work, typeof(*group), work);
retest:
spin_lock_irq(&group->lock);
while (!list_empty(&group->pending_list) ||
(group->state != MCAST_BUSY)) {
if (group->state != MCAST_BUSY) {
spin_unlock_irq(&group->lock);
process_group_error(group);
goto retest;
}
member = list_entry(group->pending_list.next,
struct mcast_member, list);
multicast = &member->multicast;
join_state = multicast->rec.join_state;
refcount_inc(&member->refcount);
if (join_state == (group->rec.join_state & join_state)) {
status = cmp_rec(&group->rec, &multicast->rec,
multicast->comp_mask);
if (!status)
join_group(group, member, join_state);
else
list_del_init(&member->list);
spin_unlock_irq(&group->lock);
ret = multicast->callback(status, multicast);
} else {
spin_unlock_irq(&group->lock);
status = send_join(group, member);
if (!status) {
deref_member(member);
return;
}
ret = fail_join(group, member, status);
}
deref_member(member);
if (ret)
ib_sa_free_multicast(&member->multicast);
spin_lock_irq(&group->lock);
}
join_state = get_leave_state(group);
if (join_state) {
group->rec.join_state &= ~join_state;
spin_unlock_irq(&group->lock);
if (send_leave(group, join_state))
goto retest;
} else {
group->state = MCAST_IDLE;
spin_unlock_irq(&group->lock);
release_group(group);
}
}
/*
* Fail a join request if it is still active - at the head of the pending queue.
*/
static void process_join_error(struct mcast_group *group, int status)
{
struct mcast_member *member;
int ret;
spin_lock_irq(&group->lock);
member = list_entry(group->pending_list.next,
struct mcast_member, list);
if (group->last_join == member) {
refcount_inc(&member->refcount);
list_del_init(&member->list);
spin_unlock_irq(&group->lock);
ret = member->multicast.callback(status, &member->multicast);
deref_member(member);
if (ret)
ib_sa_free_multicast(&member->multicast);
} else
spin_unlock_irq(&group->lock);
}
static void join_handler(int status, struct ib_sa_mcmember_rec *rec,
void *context)
{
struct mcast_group *group = context;
u16 pkey_index = MCAST_INVALID_PKEY_INDEX;
if (status)
process_join_error(group, status);
else {
int mgids_changed, is_mgid0;
if (ib_find_pkey(group->port->dev->device,
group->port->port_num, be16_to_cpu(rec->pkey),
&pkey_index))
pkey_index = MCAST_INVALID_PKEY_INDEX;
spin_lock_irq(&group->port->lock);
if (group->state == MCAST_BUSY &&
group->pkey_index == MCAST_INVALID_PKEY_INDEX)
group->pkey_index = pkey_index;
mgids_changed = memcmp(&rec->mgid, &group->rec.mgid,
sizeof(group->rec.mgid));
group->rec = *rec;
if (mgids_changed) {
rb_erase(&group->node, &group->port->table);
is_mgid0 = !memcmp(&mgid0, &group->rec.mgid,
sizeof(mgid0));
mcast_insert(group->port, group, is_mgid0);
}
spin_unlock_irq(&group->port->lock);
}
mcast_work_handler(&group->work);
}
static void leave_handler(int status, struct ib_sa_mcmember_rec *rec,
void *context)
{
struct mcast_group *group = context;
if (status && group->retries > 0 &&
!send_leave(group, group->leave_state))
group->retries--;
else
mcast_work_handler(&group->work);
}
static struct mcast_group *acquire_group(struct mcast_port *port,
union ib_gid *mgid, gfp_t gfp_mask)
{
struct mcast_group *group, *cur_group;
unsigned long flags;
int is_mgid0;
is_mgid0 = !memcmp(&mgid0, mgid, sizeof mgid0);
if (!is_mgid0) {
spin_lock_irqsave(&port->lock, flags);
group = mcast_find(port, mgid);
if (group)
goto found;
spin_unlock_irqrestore(&port->lock, flags);
}
group = kzalloc(sizeof *group, gfp_mask);
if (!group)
return NULL;
group->retries = 3;
group->port = port;
group->rec.mgid = *mgid;
group->pkey_index = MCAST_INVALID_PKEY_INDEX;
INIT_LIST_HEAD(&group->pending_list);
INIT_LIST_HEAD(&group->active_list);
INIT_WORK(&group->work, mcast_work_handler);
spin_lock_init(&group->lock);
spin_lock_irqsave(&port->lock, flags);
cur_group = mcast_insert(port, group, is_mgid0);
if (cur_group) {
kfree(group);
group = cur_group;
} else
refcount_inc(&port->refcount);
found:
atomic_inc(&group->refcount);
spin_unlock_irqrestore(&port->lock, flags);
return group;
}
/*
* We serialize all join requests to a single group to make our lives much
* easier. Otherwise, two users could try to join the same group
* simultaneously, with different configurations, one could leave while the
* join is in progress, etc., which makes locking around error recovery
* difficult.
*/
struct ib_sa_multicast *
ib_sa_join_multicast(struct ib_sa_client *client,
RDMA: Support more than 255 rdma ports Current code uses many different types when dealing with a port of a RDMA device: u8, unsigned int and u32. Switch to u32 to clean up the logic. This allows us to make (at least) the core view consistent and use the same type. Unfortunately not all places can be converted. Many uverbs functions expect port to be u8 so keep those places in order not to break UAPIs. HW/Spec defined values must also not be changed. With the switch to u32 we now can support devices with more than 255 ports. U32_MAX is reserved to make control logic a bit easier to deal with. As a device with U32_MAX ports probably isn't going to happen any time soon this seems like a non issue. When a device with more than 255 ports is created uverbs will report the RDMA device as having 255 ports as this is the max currently supported. The verbs interface is not changed yet because the IBTA spec limits the port size in too many places to be u8 and all applications that relies in verbs won't be able to cope with this change. At this stage, we are extending the interfaces that are using vendor channel solely Once the limitation is lifted mlx5 in switchdev mode will be able to have thousands of SFs created by the device. As the only instance of an RDMA device that reports more than 255 ports will be a representor device and it exposes itself as a RAW Ethernet only device CM/MAD/IPoIB and other ULPs aren't effected by this change and their sysfs/interfaces that are exposes to userspace can remain unchanged. While here cleanup some alignment issues and remove unneeded sanity checks (mainly in rdmavt), Link: https://lore.kernel.org/r/20210301070420.439400-1-leon@kernel.org Signed-off-by: Mark Bloch <mbloch@nvidia.com> Signed-off-by: Leon Romanovsky <leonro@nvidia.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2021-03-01 10:04:20 +03:00
struct ib_device *device, u32 port_num,
struct ib_sa_mcmember_rec *rec,
ib_sa_comp_mask comp_mask, gfp_t gfp_mask,
int (*callback)(int status,
struct ib_sa_multicast *multicast),
void *context)
{
struct mcast_device *dev;
struct mcast_member *member;
struct ib_sa_multicast *multicast;
int ret;
dev = ib_get_client_data(device, &mcast_client);
if (!dev)
return ERR_PTR(-ENODEV);
member = kmalloc(sizeof *member, gfp_mask);
if (!member)
return ERR_PTR(-ENOMEM);
ib_sa_client_get(client);
member->client = client;
member->multicast.rec = *rec;
member->multicast.comp_mask = comp_mask;
member->multicast.callback = callback;
member->multicast.context = context;
init_completion(&member->comp);
refcount_set(&member->refcount, 1);
member->state = MCAST_JOINING;
member->group = acquire_group(&dev->port[port_num - dev->start_port],
&rec->mgid, gfp_mask);
if (!member->group) {
ret = -ENOMEM;
goto err;
}
/*
* The user will get the multicast structure in their callback. They
* could then free the multicast structure before we can return from
* this routine. So we save the pointer to return before queuing
* any callback.
*/
multicast = &member->multicast;
queue_join(member);
return multicast;
err:
ib_sa_client_put(client);
kfree(member);
return ERR_PTR(ret);
}
EXPORT_SYMBOL(ib_sa_join_multicast);
void ib_sa_free_multicast(struct ib_sa_multicast *multicast)
{
struct mcast_member *member;
struct mcast_group *group;
member = container_of(multicast, struct mcast_member, multicast);
group = member->group;
spin_lock_irq(&group->lock);
if (member->state == MCAST_MEMBER)
adjust_membership(group, multicast->rec.join_state, -1);
list_del_init(&member->list);
if (group->state == MCAST_IDLE) {
group->state = MCAST_BUSY;
spin_unlock_irq(&group->lock);
/* Continue to hold reference on group until callback */
queue_work(mcast_wq, &group->work);
} else {
spin_unlock_irq(&group->lock);
release_group(group);
}
deref_member(member);
wait_for_completion(&member->comp);
ib_sa_client_put(member->client);
kfree(member);
}
EXPORT_SYMBOL(ib_sa_free_multicast);
RDMA: Support more than 255 rdma ports Current code uses many different types when dealing with a port of a RDMA device: u8, unsigned int and u32. Switch to u32 to clean up the logic. This allows us to make (at least) the core view consistent and use the same type. Unfortunately not all places can be converted. Many uverbs functions expect port to be u8 so keep those places in order not to break UAPIs. HW/Spec defined values must also not be changed. With the switch to u32 we now can support devices with more than 255 ports. U32_MAX is reserved to make control logic a bit easier to deal with. As a device with U32_MAX ports probably isn't going to happen any time soon this seems like a non issue. When a device with more than 255 ports is created uverbs will report the RDMA device as having 255 ports as this is the max currently supported. The verbs interface is not changed yet because the IBTA spec limits the port size in too many places to be u8 and all applications that relies in verbs won't be able to cope with this change. At this stage, we are extending the interfaces that are using vendor channel solely Once the limitation is lifted mlx5 in switchdev mode will be able to have thousands of SFs created by the device. As the only instance of an RDMA device that reports more than 255 ports will be a representor device and it exposes itself as a RAW Ethernet only device CM/MAD/IPoIB and other ULPs aren't effected by this change and their sysfs/interfaces that are exposes to userspace can remain unchanged. While here cleanup some alignment issues and remove unneeded sanity checks (mainly in rdmavt), Link: https://lore.kernel.org/r/20210301070420.439400-1-leon@kernel.org Signed-off-by: Mark Bloch <mbloch@nvidia.com> Signed-off-by: Leon Romanovsky <leonro@nvidia.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2021-03-01 10:04:20 +03:00
int ib_sa_get_mcmember_rec(struct ib_device *device, u32 port_num,
union ib_gid *mgid, struct ib_sa_mcmember_rec *rec)
{
struct mcast_device *dev;
struct mcast_port *port;
struct mcast_group *group;
unsigned long flags;
int ret = 0;
dev = ib_get_client_data(device, &mcast_client);
if (!dev)
return -ENODEV;
port = &dev->port[port_num - dev->start_port];
spin_lock_irqsave(&port->lock, flags);
group = mcast_find(port, mgid);
if (group)
*rec = group->rec;
else
ret = -EADDRNOTAVAIL;
spin_unlock_irqrestore(&port->lock, flags);
return ret;
}
EXPORT_SYMBOL(ib_sa_get_mcmember_rec);
/**
* ib_init_ah_from_mcmember - Initialize AH attribute from multicast
* member record and gid of the device.
* @device: RDMA device
* @port_num: Port of the rdma device to consider
* @rec: Multicast member record to use
* @ndev: Optional netdevice, applicable only for RoCE
* @gid_type: GID type to consider
* @ah_attr: AH attribute to fillup on successful completion
*
* ib_init_ah_from_mcmember() initializes AH attribute based on multicast
* member record and other device properties. On success the caller is
* responsible to call rdma_destroy_ah_attr on the ah_attr. Returns 0 on
* success or appropriate error code.
*
*/
RDMA: Support more than 255 rdma ports Current code uses many different types when dealing with a port of a RDMA device: u8, unsigned int and u32. Switch to u32 to clean up the logic. This allows us to make (at least) the core view consistent and use the same type. Unfortunately not all places can be converted. Many uverbs functions expect port to be u8 so keep those places in order not to break UAPIs. HW/Spec defined values must also not be changed. With the switch to u32 we now can support devices with more than 255 ports. U32_MAX is reserved to make control logic a bit easier to deal with. As a device with U32_MAX ports probably isn't going to happen any time soon this seems like a non issue. When a device with more than 255 ports is created uverbs will report the RDMA device as having 255 ports as this is the max currently supported. The verbs interface is not changed yet because the IBTA spec limits the port size in too many places to be u8 and all applications that relies in verbs won't be able to cope with this change. At this stage, we are extending the interfaces that are using vendor channel solely Once the limitation is lifted mlx5 in switchdev mode will be able to have thousands of SFs created by the device. As the only instance of an RDMA device that reports more than 255 ports will be a representor device and it exposes itself as a RAW Ethernet only device CM/MAD/IPoIB and other ULPs aren't effected by this change and their sysfs/interfaces that are exposes to userspace can remain unchanged. While here cleanup some alignment issues and remove unneeded sanity checks (mainly in rdmavt), Link: https://lore.kernel.org/r/20210301070420.439400-1-leon@kernel.org Signed-off-by: Mark Bloch <mbloch@nvidia.com> Signed-off-by: Leon Romanovsky <leonro@nvidia.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2021-03-01 10:04:20 +03:00
int ib_init_ah_from_mcmember(struct ib_device *device, u32 port_num,
struct ib_sa_mcmember_rec *rec,
struct net_device *ndev,
enum ib_gid_type gid_type,
struct rdma_ah_attr *ah_attr)
{
const struct ib_gid_attr *sgid_attr;
/* GID table is not based on the netdevice for IB link layer,
* so ignore ndev during search.
*/
if (rdma_protocol_ib(device, port_num))
ndev = NULL;
else if (!rdma_protocol_roce(device, port_num))
return -EINVAL;
sgid_attr = rdma_find_gid_by_port(device, &rec->port_gid,
gid_type, port_num, ndev);
if (IS_ERR(sgid_attr))
return PTR_ERR(sgid_attr);
memset(ah_attr, 0, sizeof(*ah_attr));
ah_attr->type = rdma_ah_find_type(device, port_num);
rdma_ah_set_dlid(ah_attr, be16_to_cpu(rec->mlid));
rdma_ah_set_sl(ah_attr, rec->sl);
rdma_ah_set_port_num(ah_attr, port_num);
rdma_ah_set_static_rate(ah_attr, rec->rate);
rdma_move_grh_sgid_attr(ah_attr, &rec->mgid,
be32_to_cpu(rec->flow_label),
rec->hop_limit, rec->traffic_class,
sgid_attr);
return 0;
}
EXPORT_SYMBOL(ib_init_ah_from_mcmember);
static void mcast_groups_event(struct mcast_port *port,
enum mcast_group_state state)
{
struct mcast_group *group;
struct rb_node *node;
unsigned long flags;
spin_lock_irqsave(&port->lock, flags);
for (node = rb_first(&port->table); node; node = rb_next(node)) {
group = rb_entry(node, struct mcast_group, node);
spin_lock(&group->lock);
if (group->state == MCAST_IDLE) {
atomic_inc(&group->refcount);
queue_work(mcast_wq, &group->work);
}
if (group->state != MCAST_GROUP_ERROR)
group->state = state;
spin_unlock(&group->lock);
}
spin_unlock_irqrestore(&port->lock, flags);
}
static void mcast_event_handler(struct ib_event_handler *handler,
struct ib_event *event)
{
struct mcast_device *dev;
int index;
dev = container_of(handler, struct mcast_device, event_handler);
if (!rdma_cap_ib_mcast(dev->device, event->element.port_num))
return;
index = event->element.port_num - dev->start_port;
switch (event->event) {
case IB_EVENT_PORT_ERR:
case IB_EVENT_LID_CHANGE:
case IB_EVENT_CLIENT_REREGISTER:
mcast_groups_event(&dev->port[index], MCAST_GROUP_ERROR);
break;
case IB_EVENT_PKEY_CHANGE:
mcast_groups_event(&dev->port[index], MCAST_PKEY_EVENT);
break;
default:
break;
}
}
static int mcast_add_one(struct ib_device *device)
{
struct mcast_device *dev;
struct mcast_port *port;
int i;
int count = 0;
treewide: Use struct_size() for kmalloc()-family One of the more common cases of allocation size calculations is finding the size of a structure that has a zero-sized array at the end, along with memory for some number of elements for that array. For example: struct foo { int stuff; void *entry[]; }; instance = kmalloc(sizeof(struct foo) + sizeof(void *) * count, GFP_KERNEL); Instead of leaving these open-coded and prone to type mistakes, we can now use the new struct_size() helper: instance = kmalloc(struct_size(instance, entry, count), GFP_KERNEL); This patch makes the changes for kmalloc()-family (and kvmalloc()-family) uses. It was done via automatic conversion with manual review for the "CHECKME" non-standard cases noted below, using the following Coccinelle script: // pkey_cache = kmalloc(sizeof *pkey_cache + tprops->pkey_tbl_len * // sizeof *pkey_cache->table, GFP_KERNEL); @@ identifier alloc =~ "kmalloc|kzalloc|kvmalloc|kvzalloc"; expression GFP; identifier VAR, ELEMENT; expression COUNT; @@ - alloc(sizeof(*VAR) + COUNT * sizeof(*VAR->ELEMENT), GFP) + alloc(struct_size(VAR, ELEMENT, COUNT), GFP) // mr = kzalloc(sizeof(*mr) + m * sizeof(mr->map[0]), GFP_KERNEL); @@ identifier alloc =~ "kmalloc|kzalloc|kvmalloc|kvzalloc"; expression GFP; identifier VAR, ELEMENT; expression COUNT; @@ - alloc(sizeof(*VAR) + COUNT * sizeof(VAR->ELEMENT[0]), GFP) + alloc(struct_size(VAR, ELEMENT, COUNT), GFP) // Same pattern, but can't trivially locate the trailing element name, // or variable name. @@ identifier alloc =~ "kmalloc|kzalloc|kvmalloc|kvzalloc"; expression GFP; expression SOMETHING, COUNT, ELEMENT; @@ - alloc(sizeof(SOMETHING) + COUNT * sizeof(ELEMENT), GFP) + alloc(CHECKME_struct_size(&SOMETHING, ELEMENT, COUNT), GFP) Signed-off-by: Kees Cook <keescook@chromium.org>
2018-05-08 23:45:50 +03:00
dev = kmalloc(struct_size(dev, port, device->phys_port_cnt),
GFP_KERNEL);
if (!dev)
return -ENOMEM;
dev->start_port = rdma_start_port(device);
dev->end_port = rdma_end_port(device);
for (i = 0; i <= dev->end_port - dev->start_port; i++) {
if (!rdma_cap_ib_mcast(device, dev->start_port + i))
continue;
port = &dev->port[i];
port->dev = dev;
port->port_num = dev->start_port + i;
spin_lock_init(&port->lock);
port->table = RB_ROOT;
init_completion(&port->comp);
refcount_set(&port->refcount, 1);
++count;
}
if (!count) {
kfree(dev);
return -EOPNOTSUPP;
}
dev->device = device;
ib_set_client_data(device, &mcast_client, dev);
INIT_IB_EVENT_HANDLER(&dev->event_handler, device, mcast_event_handler);
ib_register_event_handler(&dev->event_handler);
return 0;
}
static void mcast_remove_one(struct ib_device *device, void *client_data)
{
struct mcast_device *dev = client_data;
struct mcast_port *port;
int i;
ib_unregister_event_handler(&dev->event_handler);
flush_workqueue(mcast_wq);
for (i = 0; i <= dev->end_port - dev->start_port; i++) {
if (rdma_cap_ib_mcast(device, dev->start_port + i)) {
port = &dev->port[i];
deref_port(port);
wait_for_completion(&port->comp);
}
}
kfree(dev);
}
int mcast_init(void)
{
int ret;
mcast_wq = alloc_ordered_workqueue("ib_mcast", WQ_MEM_RECLAIM);
if (!mcast_wq)
return -ENOMEM;
ib_sa_register_client(&sa_client);
ret = ib_register_client(&mcast_client);
if (ret)
goto err;
return 0;
err:
ib_sa_unregister_client(&sa_client);
destroy_workqueue(mcast_wq);
return ret;
}
void mcast_cleanup(void)
{
ib_unregister_client(&mcast_client);
ib_sa_unregister_client(&sa_client);
destroy_workqueue(mcast_wq);
}