Moni Shoua [Thu, 14 Jan 2016 15:50:33 +0000 (17:50 +0200)]
IB/mlx4: Add gid_type to GID properties
IB core driver adds a property of type to struct ib_gid_attr.
The mlx4 driver should take that in consideration when modifying or
querying the hardware gid table.
We now alwasy have a per-PD local_dma_lkey available. Make use of that
fact in svc_rdma and stop registering our own MR.
Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Sagi Grimberg <sagig@mellanox.com> Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Reviewed-by: Chuck Lever <chuck.lever@oracle.com> Reviewed-by: Steve Wise <swise@opengridcomputing.com> Acked-by: J. Bruce Fields <bfields@redhat.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
Chuck Lever [Thu, 7 Jan 2016 19:50:10 +0000 (14:50 -0500)]
svcrdma: Add class for RDMA backwards direction transport
To support the server-side of an NFSv4.1 backchannel on RDMA
connections, add a transport class that enables backward
direction messages on an existing forward channel connection.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com> Acked-by: Bruce Fields <bfields@fieldses.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
Chuck Lever [Thu, 7 Jan 2016 19:50:02 +0000 (14:50 -0500)]
svcrdma: Define maximum number of backchannel requests
Extra resources for handling backchannel requests have to be
pre-allocated when a transport instance is created. Set up
additional fields in svcxprt_rdma to track these resources.
The max_requests fields are elements of the RPC-over-RDMA
protocol, so they should be u32. To ensure that unsigned
arithmetic is used everywhere, some other fields in the
svcxprt_rdma struct are updated.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com> Acked-by: Bruce Fields <bfields@fieldses.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
Chuck Lever [Thu, 7 Jan 2016 19:49:37 +0000 (14:49 -0500)]
svcrdma: Add gfp flags to svc_rdma_post_recv()
svc_rdma_post_recv() allocates pages for receive buffers on-demand.
It uses GFP_KERNEL so the allocator tries hard, and may sleep. But
I'm about to add a call to svc_rdma_post_recv() from a function
that may not sleep.
Since all svc_rdma_post_recv() call sites can tolerate its failure,
allow it to fail if the page allocator returns nothing. Longer term,
receive buffers, being a finite resource per-connection, should be
pre-allocated and re-used.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com> Acked-by: Bruce Fields <bfields@fieldses.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
Chuck Lever [Thu, 7 Jan 2016 19:49:12 +0000 (14:49 -0500)]
svcrdma: Improve allocation of struct svc_rdma_op_ctxt
When the maximum payload size of NFS READ and WRITE was increased
by commit cc9a903d915c ("svcrdma: Change maximum server payload back
to RPCSVC_MAXPAYLOAD"), the size of struct svc_rdma_op_ctxt
increased to over 6KB (on x86_64). That makes allocating one of
these from a kmem_cache more likely to fail in situations when
system memory is exhausted.
Since I'm about to add a caller where this allocation must always
work _and_ it cannot sleep, pre-allocate ctxts for each connection.
Another motivation for this change is that NFSv4.x servers are
required by specification not to drop NFS requests. Pre-allocating
memory resources reduces the likelihood of a drop.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com> Acked-by: Bruce Fields <bfields@fieldses.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
Chuck Lever [Thu, 7 Jan 2016 19:48:55 +0000 (14:48 -0500)]
svcrdma: Clean up rdma_create_xprt()
kzalloc is used here, so setting the atomic fields to zero is
unnecessary. sc_ord is set again in handle_connect_req. The other
fields are re-initialized in svc_rdma_accept().
Signed-off-by: Chuck Lever <chuck.lever@oracle.com> Acked-by: Bruce Fields <bfields@fieldses.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
Matan Barak [Mon, 4 Jan 2016 08:49:54 +0000 (10:49 +0200)]
IB/core: Use hop-limit from IP stack for RoCE
Previously, IPV6_DEFAULT_HOPLIMIT was used as the hop limit value for
RoCE. Fixing that by taking ip4_dst_hoplimit and ip6_dst_hoplimit as
hop limit values.
Matan Barak [Mon, 4 Jan 2016 08:49:53 +0000 (10:49 +0200)]
IB/core: Rename rdma_addr_find_dmac_by_grh
rdma_addr_find_dmac_by_grh resolves dmac, vlan_id and if_index and
downsteram patch will also add hop_limit as an output parameter,
thus we rename it to rdma_addr_find_l2_eth_by_grh.
Bart Van Assche [Fri, 1 Jan 2016 12:17:46 +0000 (13:17 +0100)]
IB/cm: Fix a recently introduced deadlock
ib_send_cm_drep() calls cm_enter_timewait() while holding a spinlock
that can be locked from inside an interrupt handler. Hence do not
enable interrupts inside cm_enter_timewait() if called with interrupts
disabled.
This patch fixes e.g. the following deadlock: Acked-by: Erez Shitrit <erezsh@mellanox.com>
=================================
[ INFO: inconsistent lock state ]
4.4.0-rc7+ #1 Tainted: G E
---------------------------------
inconsistent {HARDIRQ-ON-W} -> {IN-HARDIRQ-W} usage.
swapper/8/0 [HC1[1]:SC0[0]:HE0:SE1] takes:
(&(&cm_id_priv->lock)->rlock){?.+...}, at: [<ffffffffa036eec4>] cm_establish+0x
74/0x1b0 [ib_cm]
{HARDIRQ-ON-W} state was registered at:
[<ffffffff810a3c11>] mark_held_locks+0x71/0x90
[<ffffffff810a3e87>] trace_hardirqs_on_caller+0xa7/0x1c0
[<ffffffff810a3fad>] trace_hardirqs_on+0xd/0x10
[<ffffffff8151c40b>] _raw_spin_unlock_irq+0x2b/0x40
[<ffffffffa036ea8e>] cm_enter_timewait+0xae/0x100 [ib_cm]
[<ffffffffa036ff76>] ib_send_cm_drep+0xb6/0x190 [ib_cm]
[<ffffffffa052ed08>] srp_cm_handler+0x128/0x1a0 [ib_srp]
[<ffffffffa0370340>] cm_process_work+0x20/0xf0 [ib_cm]
[<ffffffffa0371335>] cm_dreq_handler+0x135/0x2c0 [ib_cm]
[<ffffffffa03733c5>] cm_work_handler+0x75/0xd0 [ib_cm]
[<ffffffff8107184d>] process_one_work+0x1bd/0x460
[<ffffffff81073148>] worker_thread+0x118/0x420
[<ffffffff81078454>] kthread+0xe4/0x100
[<ffffffff8151cbbf>] ret_from_fork+0x3f/0x70
irq event stamp: 1672286
hardirqs last enabled at (1672283): [<ffffffff81408ec0>] poll_idle+0x10/0x80
hardirqs last disabled at (1672284): [<ffffffff8151d304>] common_interrupt+0x84/0x89
softirqs last enabled at (1672286): [<ffffffff8105b4dc>] _local_bh_enable+0x1c/0x50
softirqs last disabled at (1672285): [<ffffffff8105b697>] irq_enter+0x47/0x70
other info that might help us debug this:
Possible unsafe locking scenario:
Fixes: commit be4b499323bf ("IB/cm: Do not queue work to a device that's going away") Signed-off-by: Bart Van Assche <bart.vanassche@sandisk.com> Cc: Erez Shitrit <erezsh@mellanox.com> Cc: stable <stable@vger.kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
When write_gid function needs to do a sleep-able operation, it unlocks
table->rwlock and then relocks it. Sparse complains about context
imbalance.
This is safe as write_gid is always called with table->rwlock.
write_gid protects from simultaneous writes to this GID entry
by setting the GID_TABLE_ENTRY_INVALID flag.
Fixes: 9c584f049596 ('IB/core: Change per-entry lock in RoCE GID table to
one lock') Signed-off-by: Matan Barak <matanb@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
Port number is not part of ClassPortInfo attribute but is
still needed as a parameter when invoking process_mad.
To properly handle this attribute, port_num is added as a
parameter to get_counter_table and get_perf_mad was changed
not to store port_num in the attribute itself when it's
querying the ClassPortInfo attribute.
This handles issue pointed out by Matan Barak <matanb@dev.mellanox.co.il>
Fixes: 145d9c541032 ('IB/core: Display extended counter set if available') Signed-off-by: Hal Rosenstock <hal@mellanox.com> Acked-by: Matan Barak <matanb@mellanox.com> Acked-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Christoph Lameter <cl@linux.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
Leon Romanovsky [Tue, 19 Jan 2016 09:11:24 +0000 (11:11 +0200)]
IB/mlx5: Delete locally redefined variable
Fix the following sparse warning:
drivers/infiniband/hw/mlx5/main.c:1061:29: warning: symbol 'pfn' shadows
an earlier one
drivers/infiniband/hw/mlx5/main.c:1030:21: originally declared here
Fixes: d69e3bcf7976 ('IB/mlx5: Mmap the HCA's core clock register to user-space') Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
Moni Shoua [Thu, 14 Jan 2016 15:47:38 +0000 (17:47 +0200)]
IB/mlx4: Take source mac from AH instead from the port
In commit dbf727de7440 ("IB/core: Use GID table in AH creation and dmac
resolution") we copy source mac to mlx4_ah from the attributes of
gid at ib_ah_attr.grh.sgid_index. Now we can use it.
Signed-off-by: Moni Shoua <monis@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
Matan Barak [Thu, 14 Jan 2016 15:47:02 +0000 (17:47 +0200)]
IB/mlx4: Initialize hop_limit when creating address handle
Hop limit value wasn't copied from attributes when ah was created.
This may influence packets for unconnected services to get dropped in
routers when endpoints are not in the same subnet.
Fixes: fa417f7b520e ("IB/mlx4: Add support for IBoE") Signed-off-by: Matan Barak <matanb@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
Hariprasad S [Wed, 13 Jan 2016 04:33:14 +0000 (10:03 +0530)]
iw_cxgb4: Take clip reference before starting IPv6 listen
The h/w is designed in such a way that, if you do anything IPv6
related, a valid clip entry must be there. So take clip reference
before creating IPv6 listening servers, and then if we fail to
create server, release the clip entry.
Hariprasad S [Tue, 12 Jan 2016 11:03:21 +0000 (16:33 +0530)]
iw_cxgb4: Fixes static checker warning in c4iw_rdev_open()
Commit c5dfb000b904 ("iw_cxgb4: Pass qid range to user space driver")
from Dec 11, 2015, leads to the following static checker warning:
drivers/infiniband/hw/cxgb4/device.c:857 c4iw_rdev_open()
warn: variable dereferenced before check 'rdev->status_page'
Also we weren't deallocating ocqp pool in error path when failed to
allocate status page. Fixing it too.
Fixes: c5dfb000b904 ("iw_cxgb4: Pass qid range to user space driver") Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Hariprasad Shenai <hariprasad@chelsio.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
Dan Carpenter [Tue, 12 Jan 2016 09:29:21 +0000 (12:29 +0300)]
IB/cma: allocating too much memory in make_cma_ports()
The issue here is that there is a cut and paste bug. When we allocate
cma_dev_group->default_ports_group we use "sizeof(*cma_dev_group->ports)"
instead of "sizeof(*cma_dev_group->default_ports_group)".
We're bumping up against the 80 character limit so I introduced a new
local pointer "ports_group" to get around that.
Fixes: 045959db65c6 ('IB/cma: Add configfs for rdma_cm') Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com> Acked-by: Matan Barak <matanb@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
Dan Carpenter [Tue, 12 Jan 2016 09:27:43 +0000 (12:27 +0300)]
RDMA/nes: checking for NULL instead of IS_ERR
nes_reg_phys_mr() returns ERR_PTRs on error. It doesn't return NULL.
This bug has been there for a while, but we recently changed from
calling a function pointer to calling nes_reg_phys_mr() directly so now
Smatch is able to detect the bug.
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Doug Ledford <dledford@redhat.com>
Vinit Agnihotri [Mon, 11 Jan 2016 17:57:25 +0000 (12:57 -0500)]
IB/qib: Support creating qps with GFP_NOIO flag
The current code is problematic when the QP creation and ipoib is used to
support NFS and NFS desires to do IO for paging purposes. In that case, the
GFP_KERNEL allocation in qib_qp.c causes a deadlock in tight memory
situations.
This fix adds support to create queue pair with GFP_NOIO flag for connected
mode only to cleanly fail the create queue pair in those situations.
Ira Weiny [Mon, 4 Jan 2016 03:44:25 +0000 (22:44 -0500)]
IB/sysfs: Fix sparse warning on attr_id
Attributed ID was declared as an int while the value should really be big
endian 16.
Fixes: 35c4cbb17811 ("IB/core: Create get_perf_mad function in sysfs.c") Reported-by: Bart Van Assche <bart.vanassche@sandisk.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com> Reviewed-by: Christoph Lameter <cl@linux.com> Reviewed-by: Hal Rosenstock <hal@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
Devesh Sharma [Thu, 24 Dec 2015 18:14:08 +0000 (13:14 -0500)]
RDMA/be2net: Remove open and close entry points
Recently Dough Ledford reported a deadlock happening
between ocrdma-load sequence and NetworkManager service
issueing "open" on be2net interface.
The deadlock happens when any be2net hook (e.g. open/close) is called
in parallel to insmod ocrdma.ko.
A. be2net is sending administrative open/close event to ocrdma holding
device_list_mutex. It does this from ndo_open/ndo_stop hooks of be2net.
So sequence of locks is rtnl_lock---> device_list lock
B. When new ocrdma roce device gets registered, infiniband stack now
takes rtnl_lock in ib_register_device() in GID initialization routines.
So sequence of locks in this path is device_list lock ---> rtnl_lock.
This improper locking sequence causes deadlock.
In order to resolve the above deadlock condition, ocrdma intorduced a
patch to stop listening to administrative open/close events generated from
be2net driver. It now depends on link-state-change async-event generated from
CNA. This change leaves behind dead code which used to generate administrative
open/close events. This patch cleans-up all that dead code from be2net.
Devesh Sharma [Thu, 24 Dec 2015 18:14:07 +0000 (13:14 -0500)]
RDMA/ocrdma: Depend on async link events from CNA
Recently Dough Ledford reported a deadlock happening
between ocrdma-load sequence and NetworkManager service
issuing "open" on be2net interface.
The deadlock happens when any be2net hook (e.g. open/close) is called
in parallel to insmod ocrdma.ko.
A. be2net is sending administrative open/close event to ocrdma holding
device_list_mutex. It does this from ndo_open/ndo_stop hooks of be2net.
So sequence of locks is rtnl_lock---> device_list lock
B. When new ocrdma roce device gets registered, infiniband stack now
takes rtnl_lock in ib_register_device() in GID initialization routines.
So sequence of locks in this path is device_list lock ---> rtnl_lock.
This improper locking sequence causes deadlock.
With this patch we stop using administrative open and close events
injected by be2net driver. These events were used to dispatch PORT_ACTIVE
and PORT_ERROR events to the IB-stack. This patch implements a logic
to receive async-link-events generated from CNA whenever link-state-change
is detected. Now on, these async-events will be used to dispatch
PORT_ACTIVE and PORT_ERROR events to IB-stack.
Depending on async-events from CNA removes the need to hold device-list-mutex
and thus breaks the busy-wait scenario.
Devesh Sharma [Thu, 24 Dec 2015 18:14:06 +0000 (13:14 -0500)]
RDMA/ocrdma: Dispatch only port event when port state changes
Dispatch only port event to IB stack when port state changes.
Don't explicitly modify qps to error. Let application listen to
port events on async event queue or let QP fail with retry-exceeded
completion error.
Devesh Sharma [Thu, 24 Dec 2015 18:14:05 +0000 (13:14 -0500)]
RDMA/ocrdma: Fix vlan-id assignment in qp parameters
vlan-id is wrongly getting as 0 when PFC is enabled.
Set vlan-id configured by user in QP parameters.
In case vlan interface is not used, flash a warning to
user to configure vlan and assign vlan-id as 0 in qp params.
Fixes: dbf727de7440 ('IB/core: Use GID table in AH creation and dmac resolution') Cc: Matan Barak <matanb@mellanox.com> Signed-off-by: Devesh Sharma <devesh.sharma@avagotech.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
Matan Barak [Thu, 7 Jan 2016 09:19:29 +0000 (11:19 +0200)]
IB/cma: Fix RDMA port validation for iWarp
cma_validate_port wrongly assumed that Ethernet devices are RoCE
devices and thus their ndev should be matched in the GID table.
This broke the iWarp support. Fixing that matching the ndev only if
we work on a RoCE port.
Cc: <stable@vger.kernel.org> # 4.4.x- Fixes: abae1b71dd37 ('IB/cma: cma_validate_port should verify the port
and netdevice') Reported-by: Hariprasad Shenai <hariprasad@chelsio.com> Tested-by: Hariprasad Shenai <hariprasad@chelsio.com> Signed-off-by: Matan Barak <matanb@mellanox.com> Reviewed-by: Steve Wise <swise@opengridcomputing.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
Erez Shitrit [Thu, 7 Jan 2016 07:28:08 +0000 (09:28 +0200)]
IB/IPoIB: Fix kernel panic on multicast flow
ipoib_mcast_restart_task calls ipoib_mcast_remove_list with the
parameter mcast->dev. That mcast is a temporary (used as an iterator)
variable that may be uninitialized.
There is no need to send the variable dev to the function, as each mcast
has its dev as a member in the mcast struct.
Jenny Derzhavetz [Thu, 24 Dec 2015 10:20:48 +0000 (12:20 +0200)]
IB/iser: Support the remote invalidation exception
Declare that we support remote invalidation in case we are:
1. using fastreg method
2. always registering memory
Detect the invalidated rkey from the work completion info so we
won't invalidate it locally. The spec mandates that we must not rely
on the target remote invalidate our rkey so we must check it upon
a receive (scsi response) completion.
Sagi Grimberg [Wed, 9 Dec 2015 12:12:07 +0000 (14:12 +0200)]
IB/iser: Change the increment rkey flow logic
When we enable remote invalidate support we won't want to perform
local invalidates at the same time we do today, but we still need
to get new rkeys. So, decouple the rkey update from the local
invalidate and tie it to memory reg instead.
IB/isert: Support the remote invalidation exception
We'll use remote invalidate, according to negotiation result
during connection establishment. If the initiator declared that
it supports the remote invalidate exception and the local HCA
supports IB_DEVICE_MEM_MGT_EXTENSIONS then the target will
use IB_WR_SEND_WITH_INV with the correct rkey for the response.
Sagi Grimberg [Wed, 9 Dec 2015 12:12:03 +0000 (14:12 +0200)]
IB/iser,isert: Create and use new shared header
The iser RDMA_CM negotiation protocol is shared by
the initiator and the target, so have a shared header
for the defines and structure. Move relevant items from
the initiator and target headers.
This parameter is described as "is mr valid indicator".
In other words, it indicates whether memory registration
is valid or not. So intuitive values would be:
mr_valid=True, when memory registration is valid and
mr_valid=False otherwise.
Sagi Grimberg [Wed, 9 Dec 2015 12:12:00 +0000 (14:12 +0200)]
IB/iser: Reuse ib_sg_to_pages
We have in iser iser_sg_to_page_vec which has exactly
the same role as ib_sg_to_pages. Customize the page_vec
to hold a fake MR so we can reuse ib_sg_to_pages.
Leon Romanovsky [Thu, 17 Dec 2015 07:31:52 +0000 (09:31 +0200)]
IB/mlx4: Suppress non-fatal memory allocations
Failure in kmalloc memory allocations will throw a warning about it.
Such warnings are not needed anymore, since in commit 0ef2f05c7e02
("IB/mlx4: Use vmalloc for WR buffers when needed"), fallback mechanism
from kmalloc() to __vmalloc() was added.
Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com> Reviewed-by: Sagi Grimberg <sagig@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
Eran Ben Elisha [Mon, 14 Dec 2015 14:34:10 +0000 (16:34 +0200)]
IB/mlx5: Advertise atomic capabilities in query device
In order to ensure IB spec atomic correctness in atomic operations, if
HW is configured to host endianness, advertise IB_ATOMIC_HCA. if not,
advertise IB_ATOMIC_NONE.
Signed-off-by: Eran Ben Elisha <eranbe@mellanox.com> Reviewed-by: Yishai Hadas <yishaih@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
Eran Ben Elisha [Mon, 14 Dec 2015 14:34:09 +0000 (16:34 +0200)]
net/mlx5_core: Add setting ATOMIC endian mode
HW is capable of 2 requestor endianness modes for standard 8 Bytes
atomic: BE (0x0) and host endianness (0x1). Read the supported modes
from hca atomic capabilities and configure HW to host endianness mode if
supported.
Signed-off-by: Eran Ben Elisha <eranbe@mellanox.com> Reviewed-by: Yishai Hadas <yishaih@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
Hariprasad S [Fri, 11 Dec 2015 07:32:01 +0000 (13:02 +0530)]
iw_cxgb4: Pass qid range to user space driver
Enhances the t4_dev_status_page to pass the qid start and size
attributes from iw_cxgb4 to libcxgb4.
Bump the ABI Version to 3 -> To allow libcxgb4 to detect old drivers and
revert to the old way of computing the qid ranges.
Dean Luick [Thu, 10 Dec 2015 21:52:30 +0000 (16:52 -0500)]
IB/mad: Ensure fairness in ib_mad_completion_handler
It was found that when a process was rapidly sending MADs other processes could
be hung in their unregister calls.
This would happen when process A was injecting packets fast enough that the
single threaded workqueue was never exiting ib_mad_completion_handler.
Therefore when process B called flush_workqueue via the unregister call it
would hang until process A stopped sending MADs.
The fix is to periodically reschedule ib_mad_completion_handler after
processing a large number of completions. The number of completions chosen was
decided based on the defaults for the recv queue size. However, it was kept
fixed such that increasing those queue sizes would not adversely affect
fairness in the future.
Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Dean Luick <dean.luick@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
Leon Romanovsky [Sun, 20 Dec 2015 10:16:11 +0000 (12:16 +0200)]
IB/mlx5: Add driver cross-channel support
Add support of cross-channel functionality to mlx5
driver. This includes ability to ignore overrun for CQ
which intended for cross-channel, export device capability and
configure the QP to be sync master/slave queues.
The cross-channel enabled QP supports combination of
three possible properties:
* WQE processing on the receive queue of this QP
* WQE processing on the send queue of this QP
* WQE are supported on the send queue
Leon Romanovsky [Sun, 20 Dec 2015 10:16:10 +0000 (12:16 +0200)]
IB/core: Add cross-channel support
The cross-channel feature allows to execute WQEs that involve
synchronization of I/O operations’ on different QPs.
This capability enables to program complex flows with a single
function call, hereby significantly reducing overhead associated
with I/O processing.
Cross-channel operations support is indicated by HCA capability
information.
The queue pairs can be configured to work as a “sync master queue”
or “sync slave queues”.
The added flags are:
1. Device capability flag IB_DEVICE_CROSS_CHANNEL for the
devices that can perform cross-channel operations.
2. CQ property flag IB_CQ_FLAGS_IGNORE_OVERRUN to disable CQ overrun
check. This check is useless in cross-channel scenario.
3. QP property flags to indicate if queues are slave or master:
* IB_QP_CREATE_MANAGED_SEND indicates that posted send work requests
will not be executed immediately and requires enabling.
* IB_QP_CREATE_MANAGED_RECV indicates that posted receive work
requests will not be executed immediately and requires enabling.
* IB_QP_CREATE_CROSS_CHANNEL declares the QP to work in cross-channel
mode. If IB_QP_CREATE_MANAGED_SEND and IB_QP_CREATE_MANAGED_RECV are
not provided, this QP will be sync master queue, else it will be sync
slave.
Matan Barak [Tue, 15 Dec 2015 18:30:12 +0000 (20:30 +0200)]
IB/mlx5: Add hca_core_clock_offset to udata in init_ucontext
Pass hca_core_clock_offset to user-space is mandatory in order to
let the user-space read the free-running clock register from the
right offset in the memory mapped page.
Passing this value is done by changing the vendor's command
and response of init_ucontext to be in extensible form.
Matan Barak [Tue, 15 Dec 2015 18:30:11 +0000 (20:30 +0200)]
IB/mlx5: Add support for hca_core_clock and timestamp_mask
Reporting the hca_core_clock (in kHZ) and the timestamp_mask in
query_device extended verb. timestamp_mask is used by users in order
to know what is the valid range of the raw timestamps, while
hca_core_clock reports the clock frequency that is used for
timestamps.
Matan Barak [Tue, 15 Dec 2015 18:30:10 +0000 (20:30 +0200)]
IB/core: Add ib_is_udata_cleared
Extending core and vendor verb commands require us to check that the
unknown part of the user's given command is all zeros.
Adding ib_is_udata_cleared in order to do so.
Remove the unused ib_allow_mw and ib_bind_mw functions, remove the
unused IB_WR_BIND_MW and IB_WC_BIND_MW opcodes and move ib_dealloc_mw
into the uverbs module.
Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Sagi Grimberg <sagig@mellanox.com> Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> [core] Reviewed-by: Steve Wise <swise@opengridcomputing.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
IB/IPoIB: Move multicast specific code out of ipoib_main.c
Code cleanup to move multicast specific code that checks for
a sendonly join to ipoib_multicast.c. This allows the removal
of the export of __ipoib_mcast_find().
Signed-off-by: Christoph Lameter <cl@linux.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
IB/IPoIB: factor out common multicast list removal code
Code cleanup to remove multicast specific code from ipoib_main.c
The removal of a list of multicast groups occurs in three places.
Create a new function ipoib_mcast_remove_list(). Use this new
function in ipoib_main.c too.
That in turn allows the dropping of two functions that were
exported from ipoib_multicast.c for expiration of mc groups.
Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Christoph Lameter <cl@linux.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
Achiad Shochat [Wed, 23 Dec 2015 16:47:22 +0000 (18:47 +0200)]
IB/mlx5: Set network_hdr_type upon RoCE responder completion
When handling a responder completion, if the link layer is Ethernet,
set the work completion network_hdr_type field according to CQE's
info and the IB_WC_WITH_NETWORK_HDR_TYPE flag.
Achiad Shochat [Wed, 23 Dec 2015 16:47:19 +0000 (18:47 +0200)]
net/mlx5_core: Introduce access functions to enable/disable RoCE
A mlx5 Ethernet port must be explicitly enabled for RoCE.
When RoCE is not enabled on the port, the NIC will refuse to create
QPs attached to it and incoming RoCE packets will be considered by the
NIC as plain Ethernet packets.
Achiad Shochat [Wed, 23 Dec 2015 16:47:18 +0000 (18:47 +0200)]
net/mlx5_core: Break down the vport mac address query function
Introduce a new function called mlx5_query_nic_vport_context().
This function gets all the NIC vport attributes from the device.
The MAC address is just one of the NIC vport attributes, so
mlx5_query_nic_vport_mac_address() is now just a wrapper function
above mlx5_query_nic_vport_context().
More NIC vport attributes will be used in following commits.
Achiad Shochat [Wed, 23 Dec 2015 16:47:17 +0000 (18:47 +0200)]
IB/mlx5: Support IB device's callback for getting its netdev
For Eth ports only:
Maintain a net device pointer in mlx5_ib_device and update it
upon NETDEV_REGISTER and NETDEV_UNREGISTER events if the
net-device and IB device have the same PCI parent device.
Implement the get_netdev callback to return this net device.
projects / karo-tx-linux.git / log