RDMA: Split kernel-only global device caps from uverbs device caps

Split out flags from ib_device::device_cap_flags that are only used
internally to the kernel into kernel_cap_flags that is not part of the
uapi. This limits the device_cap_flags to being the same bitmap that will
be copied to userspace.

This cleanly splits out the uverbs flags from the kernel flags to avoid
confusion in the flags bitmap.

Add some short comments describing which each of the kernel flags is
connected to. Remove unused kernel flags.

Link: https://lore.kernel.org/r/0-v2-22c19e565eef+139a-kern_caps_jgg@nvidia.com
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Max Gurtovoy <mgurtovoy@nvidia.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>

RDMA/ipoib: Distribute cq completion vector better

Currently ipoib choose cq completion vector based on port number, when HCA
only have one port, all the interface recv queue completion are bind to cq
completion vector 0.

To better distribute the load, use same method as __ib_alloc_cq_any to
choose completion vector, with the change, each interface now use
different completion vectors.

Link: https://lore.kernel.org/r/20201013074342.15867-1-jinpu.wang@cloud.ionos.com
Signed-off-by: Jack Wang <jinpu.wang@cloud.ionos.com>
Reviewed-by: Gioh Kim <gi-oh.kim@cloud.ionos.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>

IB/{rdmavt, hfi1}: Implement creation of accelerated UD QPs

Adds capability to create a qpn to be recognized as an accelerated
UD QP for ipoib.

This is accomplished by reserving 0x81 in byte[0] of the qpn as the
prefix for these qp types and reserving qpns between 0x810000 and
0x81ffff.

The hfi1 capability mask already contained a flag for the VNIC netdev.
This has been renamed and extended to include both VNIC and ipoib.

The rvt code to allocate qps now recognizes this flag and sets 0x81
into byte[0] of the qpn.

The code to allocate qpns is modified to reset the qpn numbering when it
is detected that a value is located in byte[0] for a UD QP and it is a
qpn being requested for net dev use. If it is a regular UD QP then it is
allowable to have bits set in byte[0] of the qpn and provide the
previously normal behavior.

The code to free the qpn now checks for the AIP prefix value of 0x81 and
removes it from the qpn before being freed so that the lower 16 bit
number can be reused.

This patch requires minor changes in the IB core and ipoib to facilitate
the creation of accelerated UP QPs.

Link: https://lore.kernel.org/r/20200511160607.173205.11757.stgit@awfm-01.aw.intel.com
Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Gary Leshner <Gary.S.Leshner@intel.com>
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>

RDMA/ipoib: Remove check of destroy CQ

There are nothing to do from user side with knowledge that destroy CQ
fails.

Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>

IB/core, ipoib: Do not overreact to SM LID change event

When IPoIB receives an SM LID change event, it reacts by flushing its
path record cache and rejoining multicast groups. This is the same
behavior it performs when it receives a reregistration event. This
behavior is unnecessary as an SM may have database backup or
synchronization mechanisms which permit the SM location or LID to change
without loss of multicast membership and without impact to path records.

Both opensm and the OPA FM issue reregistration events if a new SM is
started (or restarted with a new config) or an SM event occurs which
results in loss of multicast membership records by the SM (such as
opensm failover) or the SM encounters new nodes with Active ports (such
as after joining 2 fabrics by connecting switches via ISLs). Hence this
event can be depended on as the trigger for IPoIB cache and multicast
flushing.

It appears that some drivers, such as qib, and hfi1 issue the
IB_EVENT_SM_CHANGE but other drivers such as mlx4 and mlx5 do not.
Empirical testing on Mellanox EDR using ibv_asyncwatch has confirmed
that Mellanox EDR HCAs do not generate SM change events and that opensm
does generate reregistration.

An SM LID change event is generated by the mentioned drivers to reflect
that sm_lid and/or sm_sl in the local port info has changed. The intent
of this event is to permit applications and ULPs which have a local copy
of this information (or an address handle using it) to update their
information.

The intent is that the reregistration event (caused by the SM via a bit
in Set(PortInfo)) be used to inform nodes that they need to rejoin
multicast groups, resubscribe for notices and potentially update path
records.

When an SM migrates or fails over, a SM LID change event can occur. In
response IPoIB discards path records and multicast membership and loses
connectivity until these records are restored via SA requests. In very
large fabrics, it may take minutes for the SM to be ready and for the SA
responses to be supplied. This can result in undesirable and
unnecessary IPoIB connectivity impacts. It also can result in an
unnecessary storm of SA queries from all nodes in a cluster potentially
followed by yet another storm if the SM issues the reregistration
request.

The fact the Mellanox HCAs do not even generate this event, is further
evidence that on modern IB fabrics there will be no ill side effects
from the proposed changes below to reduce the reaction by 3 kernel
components to this event. So these changes should be benign for Mellanox
IB fabrics and will benefit OPA fabrics while also making ib_core and
ULP behavor "correct" as intended by the IBTA spec and kernel RDMA event
APIs.

Address these issues by removing IB_EVENT_SM_CHANGE handling from ipoib.
IPoIB does not locally store sm_lid nor sm_sl, so it does not need to do
anything on SM LID change. IPoIB makes use of other ib_core components
to issue SA requests for it and those components correctly track SM LID
and SM LID changes.

Also in ib_core multicast handling, remove the test for
IB_EVENT_SM_CHANGE. This code is moving all multicast groups to the
error state, which will trigger rejoins. This code is used by IPoIB as
well as the connection manager and other clients of multicast groups.
This kernel module centralizes group membership status and joins since a
node can only join a given group once but multiple ULPs or applications
may want to join the same group. It makes use of the sa_query.c
component in ib_core, which correctly trackes SM LID and SL. This
component does not track SM LID nor SL itself and hence need not react
to their changes.

Similarly in the ib_core cache code remove the handling for the
IB_EVENT_SM_CHANGE. In this function. The ib_cache_update function
which is ultimately called is updating local copies of the pkey table,
gid table and lmc. It does not update nor retain sm_lid nor sm_sl. As
such it does not need to be called on an SM LID change. It technically
also does not need to be called on a reregistration. The LID_CHANGE,
PKEY_CHANGE, GID_CHANGE and port state change events (PORT_ERR,
PORT_ACTICE) should be sufficient triggers.

It is worth noting that the alternative of simply having the hfi1 and
qib drivers not generate the SM LID change event was explored. While
this would duplicate what Mellanox drivers do now, it is not the correct
behavior and removes the ability for an SM to migrate without requiring
reregistration. Since both opensm and OPA SM have mechanisms to backup
or synchronize registration information, it is desirable to let them
perform SM migrations (with LID or SL changes) without requiring
reregistration when they deem it appropriate.

Suggested-by: Todd Rimmer <todd.rimmer@intel.com>
Tested-by: Michael Brooks <michael.brooks@intel.com>
Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Reviewed-by: Todd Rimmer <todd.rimmer@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>

RDMA/ulp: Use dev_name instead of ibdev->name

These return the same thing but dev_name is a more conventional use of the
kernel API.

Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Reviewed-by: Steve Wise <swise@opengridcomputing.com>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com>

RDMA/ipoib: Fix return code from ipoib_cm_dev_init

The proper return code is -EOPNOTSUPP and not -ENOSYS when the function
isn't supported, also make sure to return the right error code
from ipoib_transport_dev_init() when ipoib_cm_dev_init() is supported.

Signed-off-by: Kamal Heib <kamalheib1@gmail.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>

RDMA/ipoib: Fix use of sizeof()

Make sure to use sizeof(...) instead of sizeof ... which is more
preferred.

Signed-off-by: Kamal Heib <kamalheib1@gmail.com>
Reviewed-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>

IB/core: add max_send_sge and max_recv_sge attributes

This patch replaces the ib_device_attr.max_sge with max_send_sge and
max_recv_sge. It allows ulps to take advantage of devices that have very
different send and recv sge depths. For example cxgb4 has a max_recv_sge
of 4, yet a max_send_sge of 16. Splitting out these attributes allows
much more efficient use of the SQ for cxgb4 with ulps that use the RDMA_RW
API. Consider a large RDMA WRITE that has 16 scattergather entries.
With max_sge of 4, the ulp would send 4 WRITE WRs, but with max_sge of
16, it can be done with 1 WRITE WR.

Acked-by: Sagi Grimberg <sagi@grimberg.me>
Acked-by: Christoph Hellwig <hch@lst.de>
Acked-by: Selvin Xavier <selvin.xavier@broadcom.com>
Acked-by: Shiraz Saleem <shiraz.saleem@intel.com>
Acked-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>

IB/ipoib: Replace printk with pr_warn

pr_* is the preferred way to print messages, replace all
printk(KERN_WARN, ...) with pr_warn.

Signed-off-by: Yuval Shaia <yuval.shaia@oracle.com>
Reviewed-by: Zhu Yanjun <yanjun.zhu@oracle.com>
Reviewed-by: Leon Romanovsky <leonro@mellanox.com>
Reviewed-by: Santosh Shilimkar <santosh.shilimkar@oracle.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>

IB/ipoib: Use NAPI in UD/TX flows

Instead of explicit call to poll_cq of the tx ring, use the NAPI mechanism
to handle the completions of each packet that has been sent to the HW.

The next major changes were taken:
* The driver init completion function in the creation of the send CQ,
that function triggers the napi scheduling.
* The driver uses CQ for RX for both modes UD and CM, and CQ for TX
for CM and UD.

Cc: Kamal Heib <kamalh@mellanox.com>
Signed-off-by: Erez Shitrit <erezsh@mellanox.com>
Reviewed-by: Alex Vesker <valex@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>

IB/IPoIB: Support acceleration options callbacks

IPoIB driver now uses the new set of callback functions.

If the hardware provider supports the new ipoib_options implementation,
the driver uses the callbacks in its data path flows, otherwise it uses the
driver default implementation for all data flows in its code.

The default implementation wasn't change and it is exactly as it was before
introduction of acceleration support.

Signed-off-by: Erez Shitrit <erezsh@mellanox.com>
Reviewed-by: Alex Vesker <valex@mellanox.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Doug Ledford <dledford@redhat.com>

IB/IPoIB: Use defined function for netdev_priv function

Make ipoib_priv point to netdev_priv where the code calls netdev_priv.

Signed-off-by: Erez Shitrit <erezsh@mellanox.com>
Reviewed-by: Alex Vesker <valex@mellanox.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Doug Ledford <dledford@redhat.com>

IB/IPoIB: Separate control and data related initializations

This patch prepares init and teardown flows so we can call them
through ipoib_options function pointers.

It arranges that area of code as the following:
* All operations which deal with the resource allocation/deletion
are performed in one place.
* All operations that are control oriented, meaning that they are not
connected to a specific hardware, are performed in a separate place.

The operations for allocation of hardware resources are now in the
function ipoib_dev_init_default, and the deletion of all the resources
are in ipoib_dev_uninit_default

The only exception is the creation of the PD object,
which is used both for resource allocation (create QP etc.)
and for control flows like creating AH.

It also does:
* Move creation of rx_ring and tx_ring to be in the resources
allocation area.
* Move the function ipoib_ib_dev_open that does the open device
to the control area instead of the dev_init which creates resources.

Signed-off-by: Erez Shitrit <erezsh@mellanox.com>
Reviewed-by: Alex Vesker <valex@mellanox.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Doug Ledford <dledford@redhat.com>

IB/ipoib_verbs: Remove deprecated create_singlethread_workqueue

alloc_ordered_workqueue() with WQ_MEM_RECLAIM set, replaces
deprecated create_singlethread_workqueue(). This is the identity
conversion.

The workqueue "wq" queues mulitple work items viz &priv->restart_task,
&priv->cm.rx_reap_task, &priv->cm.skb_task, &priv->neigh_reap_task,
&priv->ah_reap_task, &priv->mcast_task and &priv->carrier_on_task.
The work items require strict execution ordering.
Hence, an ordered dedicated workqueue has been used.

WQ_MEM_RECLAIM has been set to ensure forward progress under
memory pressure.

Signed-off-by: Bhaktipriya Shridhar <bhaktipriya96@gmail.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>

IB/core: add support to create a unsafe global rkey to ib_create_pd

Instead of exposing ib_get_dma_mr to ULPs and letting them use it more or
less unchecked, this moves the capability of creating a global rkey into
the RDMA core, where it can be easily audited. It also prints a warning
everytime this feature is used as well.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Reviewed-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>

IB/ipoib: Report SG feature regardless of HW UD CSUM capability

Decouple SG support from HW ability to do UD checksum.
This coupling is for historical reasons and removed with 'commit
ec5f06156423 ("net: Kill link between CSUM and SG features.")'

During driver load it is assumed that device does not supports SG. The
final decision is taken after creating UD QP based on device capability.

Signed-off-by: Yuval Shaia <yuval.shaia@oracle.com>
Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>

IB/IPoIB: Allow setting the device address

In IB networks, and specifically in IPoIB/rdmacm traffic, the device
address of an IPoIB interface is used as a means to exchange information
between nodes needed for communication.

Currently an IPoIB interface will always be created with a device
address based on its node GUID without a way to change that.

This change adds the ability to set the device address of an IPoIB
interface by value. We use the set mac address ndo to do that.

The flow should be broken down to two:
1) The GID value is already in the GID table,
in this case the interface will be able to set carrier up.

2) The GID value is not yet in the GID table,
in this case the interface won't try to join the multicast group
and will wait (listen on GID_CHANGE event) until the GID is inserted.

In order to track those changes, we add a new flag:
* IPOIB_FLAG_DEV_ADDR_SET.

When set, it means the dev_addr is a based on a value in the gid
table. this bit will be cleared upon a dev_addr change triggered
by the user and set after validation.

Per IB spec the port GUID can't change if the module is loaded.
port GUID is the basis for GID at index 0 which is the basis for
the default device address of a ipoib interface.

The issue is that there are devices that don't follow the spec,
they change the port GUID while HCA is powered on, so in order
not to break userspace applications. We need to check if the
user wanted to control the device address and we assume that
if he sets the device address back to be based on GID index 0,
he no longer wishs to control it.

In order to track this, we add an additional flag:
* IPOIB_FLAG_DEV_ADDR_CTRL

When setting the device address, there is no validation of the upper
twelve bytes of the device address (flags, qpn, subnet prefix) as those
bytes are not under the control of the user.

Signed-off-by: Mark Bloch <markb@mellanox.com>
Reviewed-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Doug Ledford <dledford@redhat.com>

IB/ipoib: Add handling for sending of skb with many frags

IPoIB converts skb-fragments to sge adding 1 extra sge when SG is enabled.
Current codepath assumes that the max number of sge a device support
is at least MAX_SKB_FRAGS+1, there is no interaction with upper layers
to limit number of fragments in an skb if a device suports fewer
sges. The assumptions also lead to requesting a fixed number of sge
when IPoIB creates queue-pairs with SG enabled.

A fallback/slowpath is implemented using skb_linearize to
handle cases where the conversion would result in more sges than supported.

Signed-off-by: Hans Westgaard Ry <hans.westgaard.ry@oracle.com>
Reviewed-by: Håkon Bugge <haakon.bugge@oracle.com>
Reviewed-by: Wei Lin Guay <wei.lin.guay@oracle.com>
Reviewed-by: Yuval Shaia <yuval.shaia@oracle.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>

IB: split struct ib_send_wr

This patch split up struct ib_send_wr so that all non-trivial verbs
use their own structure which embedds struct ib_send_wr. This dramaticly
shrinks the size of a WR for most common operations:

sizeof(struct ib_send_wr) (old): 96

sizeof(struct ib_send_wr): 48
sizeof(struct ib_rdma_wr): 64
sizeof(struct ib_atomic_wr): 96
sizeof(struct ib_ud_wr): 88
sizeof(struct ib_fast_reg_wr): 88
sizeof(struct ib_bind_mw_wr): 96
sizeof(struct ib_sig_handover_wr): 80

And with Sagi's pending MR rework the fast registration WR will also be
down to a reasonable size:

sizeof(struct ib_fastreg_wr): 64

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com> [srp, srpt]
Reviewed-by: Chuck Lever <chuck.lever@oracle.com> [sunrpc]
Tested-by: Haggai Eran <haggaie@mellanox.com>
Tested-by: Sagi Grimberg <sagig@mellanox.com>
Tested-by: Steve Wise <swise@opengridcomputing.com>

IB/core: Make ib_dealloc_pd return void

The majority of callers never check the return value, and even if they
did, they can't do anything about a failure.

All possible failure cases represent a bug in the caller, so just
WARN_ON inside the function instead.

This fixes a few random errors:
net/rd/iw.c infinite loops while it fails. (racing with EBUSY?)

This also lays the ground work to get rid of error return from the
drivers. Most drivers do not error, the few that do are broken since
it cannot be handled.

Since uverbs can legitimately make use of EBUSY, open code the
check.

Signed-off-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Reviewed-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>

IB/ipoib: Remove ib_get_dma_mr calls

The pd now has a local_dma_lkey member which completely replaces
ib_get_dma_mr, use it instead.

Signed-off-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>

IB/ipoib: Fix CONFIG_INFINIBAND_IPOIB_CM

If the above is turned off then ipoib_cm_dev_init unconditionally
returns ENOSYS, and the newly added error handling in
0b3957 prevents ipoib from coming up at all:

kernel: mlx4_0: ipoib_transport_dev_init failed
kernel: mlx4_0: failed to initialize port 1 (ret = -12)

Fixes: 0b39578bcde4 (IB/ipoib: Use dedicated workqueues per interface)
Signed-off-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>

IB/core: Change ib_create_cq to use struct ib_cq_init_attr

Currently, ib_create_cq uses cqe and comp_vecotr instead
of the extendible ib_cq_init_attr struct.

Earlier patches already changed the vendors to work with
ib_cq_init_attr. This patch changes the consumers too.

Signed-off-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>

IB/ipoib: Use one linear skb in RX flow

The current code in the RX flow uses two sg entries for each incoming
packet, the first one was for the IB headers and the second for the rest
of the data, that causes two dma map/unmap and two allocations, and few
more actions that were done at the data path.

Use only one linear skb on each incoming packet, for the data (IB
headers and payload), that reduces the packet processing in the
data-path (only one skb, no frags, the first frag was not used anyway,
less memory allocations) and the dma handling (only one dma map/unmap
over each incoming packet instead of two map/unmap per each incoming packet).

After commit 73d3fe6d1c6d ("gro: fix aggregation for skb using frag_list") from
Eric Dumazet, we will get full aggregation for large packets.

When running bandwidth tests before and after the (over the card's numa node),
using "netperf -H 1.1.1.3 -T -t TCP_STREAM", the results before are ~12Gbs before
and after ~16Gbs on my setup (Mellanox's ConnectX3).

Signed-off-by: Erez Shitrit <erezsh@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>

IB/ipoib: Use dedicated workqueues per interface

During my recent work on the rtnl lock deadlock in the IPoIB driver, I
saw that even once I fixed the apparent races for a single device, as
soon as that device had any children, new races popped up. It turns
out that this is because no matter how well we protect against races
on a single device, the fact that all devices use the same workqueue,
and flush_workqueue() flushes *everything* from that workqueue means
that we would also have to prevent all races between different devices
(for instance, ipoib_mcast_restart_task on interface ib0 can race with
ipoib_mcast_flush_dev on interface ib0.8002, resulting in a deadlock on
the rtnl_lock).

There are several possible solutions to this problem:

Make carrier_on_task and mcast_restart_task try to take the rtnl for
some set period of time and if they fail, then bail. This runs the
real risk of dropping work on the floor, which can end up being its
own separate kind of deadlock.

Set some global flag in the driver that says some device is in the
middle of going down, letting all tasks know to bail. Again, this can
drop work on the floor.

Or the method this patch attempts to use, which is when we bring an
interface up, create a workqueue specifically for that interface, so
that when we take it back down, we are flushing only those tasks
associated with our interface. In addition, keep the global
workqueue, but now limit it to only flush tasks. In this way, the
flush tasks can always flush the device specific work queues without
having deadlock issues.

Signed-off-by: Doug Ledford <dledford@redhat.com>

Revert "IPoIB: Use dedicated workqueues per interface"

This reverts commit 5141861cd5e17eac9676ff49c5abfafbea2b0e98.

The series of IPoIB bug fixes that went into 3.19-rc1 introduce
regressions, and after trying to sort things out, we decided to revert
to 3.18's IPoIB driver and get things right for 3.20.

Signed-off-by: Roland Dreier <roland@purestorage.com>

IPoIB: Use dedicated workqueues per interface

During my recent work on the rtnl lock deadlock in the IPoIB driver, I
saw that even once I fixed the apparent races for a single device, as
soon as that device had any children, new races popped up. It turns
out that this is because no matter how well we protect against races
on a single device, the fact that all devices use the same workqueue,
and flush_workqueue() flushes *everything* from that workqueue, we can
have one device in the middle of a down and holding the rtnl lock and
another totally unrelated device needing to run mcast_restart_task,
which wants the rtnl lock and will loop trying to take it unless is
sees its own FLAG_ADMIN_UP flag go away. Because the unrelated
interface will never see its own ADMIN_UP flag drop, the interface
going down will deadlock trying to flush the queue. There are several
possible solutions to this problem:

Make carrier_on_task and mcast_restart_task try to take the rtnl for
some set period of time and if they fail, then bail. This runs the
real risk of dropping work on the floor, which can end up being its
own separate kind of deadlock.

Set some global flag in the driver that says some device is in the
middle of going down, letting all tasks know to bail. Again, this can
drop work on the floor. I suppose if our own ADMIN_UP flag doesn't go
away, then maybe after a few tries on the rtnl lock we can queue our
own task back up as a delayed work and return and avoid dropping work
on the floor that way. But I'm not 100% convinced that we won't cause
other problems.

Or the method this patch attempts to use, which is when we bring an
interface up, create a workqueue specifically for that interface, so
that when we take it back down, we are flushing only those tasks
associated with our interface. In addition, keep the global
workqueue, but now limit it to only flush tasks. In this way, the
flush tasks can always flush the device specific work queues without
having deadlock issues.

Signed-off-by: Doug Ledford <dledford@redhat.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>

IB/core: Add flow steering support for IPoIB UD traffic

When creating an IPoIB UD QP, provide a hint to the low level driver
that the QP should support flow-steering. This means that privileged
user space applications can steer TCP/IP IPoIB traffic from the
network stack, in a similar manner done with Ethernet RAW_PACKET QPs.

The hint is provided through new QP creation flag called NETIF_QP.

Signed-off-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>

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>

IPoIB: Get rid of ipoib_mcast_detach() wrapper

ipoib_mcast_detach() does nothing except call ib_detach_mcast(), so just
use the core API in the one place that does a multicast group detach.

add/remove: 0/1 grow/shrink: 0/1 up/down: 0/-105 (-105)
function old new delta
ipoib_mcast_leave 357 319 -38
ipoib_mcast_detach 67 - -67

Signed-off-by: Roland Dreier <rolandd@cisco.com>

IPoIB: Only set Q_Key once: after joining broadcast group

The current code will set the Q_Key for any join of a non-sendonly
multicast group. The operation involves a modify QP operation, which
is fairly heavyweight, and is only really required after the join of
the broadcast group. Fix this by adding a parameter to ipoib_mcast_attach()
to control when the Q_Key is set.

Signed-off-by: Eli Cohen <eli@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>

IPoIB: Remove priv->mcast_mutex

No need for a mutex around calls to ib_attach_mcast/ib_detach_mcast
since these operations are synchronized at the HW driver layer.

Signed-off-by: Eli Cohen <eli@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>

IPoIB: Refresh paths instead of flushing them on SM change events

The patch tries to solve the problem of device going down and paths being
flushed on an SM change event. The method is to mark the paths as candidates for
refresh (by setting the new valid flag to 0), and wait for an ARP
probe a new path record query.

The solution requires a different and less intrusive handling of SM
change event. For that, the second argument of the flush function
changes its meaning from a boolean flag to a level. In most cases, SM
failover doesn't cause LID change so traffic won't stop. In the rare
cases of LID change, the remote host (the one that hadn't changed its
LID) will lose connectivity until paths are refreshed. This is no
worse than the current state. In fact, preventing the device from
going down saves packets that otherwise would be lost.

Signed-off-by: Moni Levy <monil@voltaire.com>
Signed-off-by: Moni Shoua <monis@voltaire.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>

IPoIB: Use multicast loopback blocking if available

Set IB_QP_CREATE_BLOCK_MULTICAST_LOOPBACK for IPoIB's UD QPs if
supported by the underlying device. This creates an improvement of up
to 39% in bandwidth when sending multicast packets with IPoIB, and an
improvment of 12% in cpu usage.

Signed-off-by: Ron Livne <ronli@voltaire.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>

RDMA: Remove subversion $Id tags

They don't get updated by git and so they're worse than useless.

Signed-off-by: Roland Dreier <rolandd@cisco.com>

IB/ipoib: Fix transmit queue stalling forever

Commit f56bcd80 ("IPoIB: Use separate CQ for UD send completions")
introduced a bug where the transmit queue could get stopped and never
woken up. The problem is that send completions are only polled at the
end of the xmit function, so if the send queue fills up and the xmit
path stops the queue, then there is no way for send completions to
ever get polled, and so the transmit queue stays stopped forever.

Fix this by arming the send CQ just before posting the last send
request that fills the send queue. Then, when the completion event
handler is called, drain the send CQ. Since it is possible that not
enough send completions are in the CQ, verify that the the net queue
has been woken up after draining the send CQ, and if not arm a timer
and drain again at the timer function.

Signed-off-by: Roland Dreier <rolandd@cisco.com>

IPoIB: Use separate CQ for UD send completions

Use a dedicated CQ for UD send completions. Also, do not arm the UD
send CQ, which reduces the number of interrupts generated. This patch
farther reduces overhead by not calling poll CQ for every posted send
WR -- it does polls only when there 16 or more outstanding work requests.

Signed-off-by: Eli Cohen <eli@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>

IPoIB: Handle 4K IB MTU for UD (datagram) mode

This patch enables IPoIB to use 4K UD messages (when the underlying
device and fabrics support a 4K MTU) by using two scatter buffers when
PAGE_SIZE is less than or equal to thhe HCA IB MTU size. The first
buffer is for IPoIB header + GRH header, and the second buffer is the
IPoIB payload, which is 4K-4.

Signed-off-by: Shirley Ma <xma@us.ibm.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>

IPoIB: Add LSO support

For HCAs that support TCP segmentation offload (IB_DEVICE_UD_TSO), set
NETIF_F_TSO and use HW LSO to offload TCP segmentation.

Signed-off-by: Eli Cohen <eli@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>

IPoIB: Add send gather support

This patch acts as a preparation for using checksum offload for IB
devices capable of inserting/verifying checksum in IP packets. The
patch does not actaully turn on NETIF_F_SG - we defer that to the
patches adding checksum offload capabilities.

We only add support for send gathers for datagram mode, since existing
HW does not support checksum offload on connected QPs.

Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il>
Signed-off-by: Eli Cohen <eli@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>

IPoIB/cm: Add connected mode support for devices without SRQs

Some IB adapters (notably IBM's eHCA) do not implement SRQs (shared
receive queues). The current IPoIB connected mode support only works
on devices that support SRQs.

Fix this by adding support for using the receive queue of each
connected mode receive QP. The disadvantage of this compared to using
an SRQ is that it means a full queue of receives must be posted for
each remote connected mode peer, which means that total memory usage
is potentially much higher than when using SRQs. To manage this, add
a new module parameter "max_nonsrq_conn_qp" that limits the number of
connections allowed per interface.

The rest of the changes are fairly straightforward: we use a table of
struct ipoib_cm_rx to hold all the active connections, and put the
table index of the connection in the high bits of receive WR IDs.
This is needed because we cannot rely on the struct ib_wc.qp field for
non-SRQ receive completions. Most of the rest of the changes just
test whether or not an SRQ is available, and post receives or find
received packets in the right place depending on the answer.

Cleaning up dead connections actually becomes simpler, because we do
not have to do the "last WQE reached" dance that is required to
destroy QPs attached to an SRQ. We just move the QP to the error
state and wait for all pending receives to be flushed.

Signed-off-by: Pradeep Satyanarayana <pradeeps@linux.vnet.ibm.com>

[ Completely rewritten and split up, based on Pradeep's work. Several
bugs fixed and no doubt several bugs introduced. - Roland ]

Signed-off-by: Roland Dreier <rolandd@cisco.com>

IPoIB: Trivial formatting cleanups

Fix whitespace blunders, convert "foo* bar" to "foo *bar", etc.

Signed-off-by: Roland Dreier <rolandd@cisco.com>

IPoIB: Fix typo to end statement with ';' instead of ','

Signed-off-by: Eli Cohen <eli@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>

IPoIB: Fix leak in ipoib_transport_dev_init() error path

ipoib_transport_dev_init() calls ipoib_cm_dev_init(), so it needs to
call ipoib_cm_dev_cleanup() to unwind that on the error path.

Found by Dotan Barak of Mellanox.

Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>

IPoIB/cm: Fix SRQ WR leak

SRQ WR leakage has been observed with IPoIB/CM: e.g. flipping ports on
and off will, with time, leak out all WRs and then all connections
will start getting RNR NAKs. Fix this in the way suggested by spec:
move the QP being destroyed to the error state, wait for "Last WQE
Reached" event and then post WR on a "drain QP" connected to the same
CQ. Once we observe a completion on the drain QP, it's safe to call
ib_destroy_qp.

Signed-off-by: Michael S. Tsirkin <mst@dev.mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>

IPoIB: Handle P_Key table reordering

SM reconfiguration or failover possibly causes a shuffling of the values
in the P_Key table. Right now, IPoIB only queries for the P_Key index
once when it creates the device QP, and hence there are problems if the
index of a P_Key value changes. Fix this by using the PKEY_CHANGE event
to trigger a recheck of the P_Key index.

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>

IB: Add CQ comp_vector support

Add a num_comp_vectors member to struct ib_device and extend
ib_create_cq() to pass in a comp_vector parameter -- this parallels
the userspace libibverbs API. Update all hardware drivers to set
num_comp_vectors to 1 and have all ULPs pass 0 for the comp_vector
value. Pass the value of num_comp_vectors to userspace rather than
hard-coding a value of 1.

We want multiple CQ event vector support (via MSI-X or similar for
adapters that can generate multiple interrupts), but it's not clear
how many vectors we want, or how we want to deal with policy issues
such as how to decide which vector to use or how to set up interrupt
affinity. This patch is useful for experimenting, since no core
changes will be necessary when updating a driver to support multiple
vectors, and we know that we want to make at least these changes
anyway.

Signed-off-by: Michael S. Tsirkin <mst@dev.mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>

IPoIB: Only handle async events for one port

An asynchronous event carries the port number that the event occurred
on, so there's no reason for an IPoIB interface to process an event
associated with a different local HCA port.

Signed-off-by: Roland Dreier <rolandd@cisco.com>

IPoIB: Connected mode experimental support

The following patch adds experimental support for IPoIB connected
mode, as defined by the draft from the IETF ipoib working group. The
idea is to increase performance by increasing the MTU from the maximum
of 2K (theoretically 4K) supported by IPoIB on top of UD. With this
code, I'm able to get 800MByte/sec or more with netperf without
options on a Mellanox 4x back-to-back DDR system.

Some notes on code:
1. SRQ is used for scalability to large cluster sizes
2. Only RC connections are used (UC does not support SRQ now)
3. Retry count is set to 0 since spec draft warns against retries
4. Each connection is used for data transfers in only 1 direction, so
each connection is either active(TX) or passive (RX). 2 sides that
want to communicate create 2 connections.
5. Each active (TX) connection has a separate CQ for send completions -
this keeps the code simple without CQ resize and other tricks
6. To detect stale passive side connections (where the remote side is
down), we keep an LRU list of passive connections (updated once per
second per connection) and destroy a connection after it has been
unused for several seconds. The LRU rule makes it possible to avoid
scanning connections that have recently been active.

Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>

IPoIB: Handle client reregister events

Handle client reregister events by treating them just like LID or
SM changes -- flush all cached paths and rejoin multicast groups.

Signed-off-by: Leonid Arsh <leonida@voltaire.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>

IPoIB: Make send and receive queue sizes tunable

Make IPoIB's send and receive queue sizes tunable via module
parameters ("send_queue_size" and "recv_queue_size"). This allows the
queue sizes to be enlarged to fix disastrously bad performance on some
platforms and workloads, without bloating memory usage when large
queues aren't needed.

Signed-off-by: Shirley Ma <xma@us.ibm.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>

IPoIB: P_Key change event handling

This patch causes the network interface to respond to P_Key change
events correctly. As a result, you'll see a child interface in the
"RUNNING" state (netif_carrier_on()) only when the corresponding P_Key
is configured by the SM. When SM removes a P_Key, the "RUNNING" state
will be disabled for the corresponding network interface. To
implement this, I added IB_EVENT_PKEY_CHANGE event handling. To
prevent flushing the device before the device is open by the "delay
open" mechanism, I added an additional device flag called
IPOIB_FLAG_INITIALIZED.

This also prevents the child network interface from trying to join to
multicast groups until the PKEY is configured. We used to get error
messages like:

ib0.f2f2: couldn't attach QP to multicast group ff12:401b:f2f2:0:0:0:ffff:ffff

in this case. To fix this, I just check IPOIB_FLAG_OPER_UP flag in
ipoib_set_mcast_list().

Signed-off-by: Leonid Arsh <leonida@voltaire.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>

IPoIB: Fix network interface "RUNNING" status

With the current IPoIB driver, the status of network interfaces stays
"RUNNING" even if the link goes down (for example because a cable is
unplugged). Fix this by flushing the IPoIB interface when the link
goes down.

Signed-off-by: Leonid Arsh <leonida@voltaire.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>

IPoIB: Move ipoib_ib_dev_flush() to ipoib workqueue

Move ipoib_ib_dev_flush() to ipoib's workqueue. This keeps it ordered
with respect to other work scheduled by the ipoib driver. This fixes
problems with races, for example:
- ipoib_ib_dev_flush() has started running because of an IB event
- user does ifconfig ib0 down
- ipoib_mcast_stop_thread() gets called twice and waits for the same
completion twice

Signed-off-by: Jack Morgenstein <jackm@mellanox.co.il>
Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>

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>

[IPoIB] cleanups: fix comment, remove useless variables

Minor cleanups: fix a misleading comment, and get rid of attr_mask
variables that are only used to hold constants (just use the constants
directly).

Signed-off-by: Roland Dreier <rolandd@cisco.com>

[IPoIB] Rename ipoib_create_qp() -> ipoib_init_qp() and fix error cleanup

ipoib_create_qp() no longer creates IPoIB's QP, so it shouldn't
destroy the QP on failure -- that unwinding happens elsewhere, so the
current code can cause a double free. While we're at it, the
function's name should match what it actually does, so rename it to
ipoib_init_qp().

Signed-off-by: Roland Dreier <rolandd@cisco.com>

[PATCH] IB: move include files to include/rdma

Move the InfiniBand headers from drivers/infiniband/include to include/rdma.
This allows InfiniBand-using code to live elsewhere, and lets us remove the
ugly EXTRA_CFLAGS include path from the InfiniBand Makefiles.

Signed-off-by: Roland Dreier <rolandd@cisco.com>

[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>

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!