RDMA/efa: Allow passing of optional access flags for MR registration
As part of adding a range of optional access flags that drivers need to be
able to accept, mask this range inside efa driver. This will prevent the
driver from failing when an access flag from that range is passed.
Jason Gunthorpe [Wed, 8 Jan 2020 18:05:33 +0000 (20:05 +0200)]
RDMA/core: Remove ucontext_lock from the uverbs_destry_ufile_hw() path
This lock only serializes ucontext creation. Instead of checking the
ucontext_lock during destruction hold the existing hw_destroy_rwsem during
creation, which is the standard pattern for object creation.
The simplification of locking is needed for the next patch.
Jason Gunthorpe [Wed, 8 Jan 2020 18:05:32 +0000 (20:05 +0200)]
RDMA/core: Add UVERBS_METHOD_ASYNC_EVENT_ALLOC
Allow the async FD to be allocated separately from the context.
This is necessary to introduce the ioctl to create a context, as an ioctl
should only ever create a single uobject at a time.
If multiple async FDs are created then the first one is used to deliver
affiliated events from any ib_uevent_object, with all subsequent ones will
receive only unaffiliated events.
Sergey Gorenko [Wed, 15 Jan 2020 13:30:55 +0000 (13:30 +0000)]
IB/srp: Never use immediate data if it is disabled by a user
Some SRP targets that do not support specification SRP-2, put the garbage
to the reserved bits of the SRP login response. The problem was not
detected for a long time because the SRP initiator ignored those bits. But
now one of them is used as SRP_LOGIN_RSP_IMMED_SUPP. And it causes a
critical error on the target when the initiator sends immediate data.
The ib_srp module has a use_imm_date parameter to enable or disable
immediate data manually. But it does not help in the above case, because
use_imm_date is ignored at handling the SRP login response. The problem is
definitely caused by a bug on the target side, but the initiator's
behavior also does not look correct. The initiator should not use
immediate data if use_imm_date is disabled by a user.
This commit adds an additional checking of use_imm_date at the handling of
SRP login response to avoid unexpected use of immediate data.
Fixes: 882981f4a411 ("RDMA/srp: Add support for immediate data") Link: https://lore.kernel.org/r/20200115133055.30232-1-sergeygo@mellanox.com Signed-off-by: Sergey Gorenko <sergeygo@mellanox.com> Reviewed-by: Bart Van Assche <bvanassche@acm.org> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Rao Shoaib [Tue, 14 Jan 2020 00:41:20 +0000 (16:41 -0800)]
RDMA/rxe: Compute the maximum sges and inline size based on the WQE size
The SGE buffer size and max_inline data should be derived from the size of
the WQE. Each value individually sets the WQE size, so compute the actual
sizes based on the actual WQE size and configure the QP with the maximums.
Also fix the missing return of the actual maximum capability to the caller.
Jiaran Zhang [Wed, 15 Jan 2020 01:42:26 +0000 (09:42 +0800)]
RDMA/hns: Add support for extended atomic in userspace
To support extended atomic operations including cmp & swap and fetch & add
of 8 bytes, 16 bytes, 32 bytes, 64 bytes in userspace, some field in qpc
should be configured.
Lijun Ou [Sat, 11 Jan 2020 10:32:40 +0000 (18:32 +0800)]
RDMA/hns: Add interfaces to get pf capabilities from firmware
pf capabilities are set by default for hip08 previously which should
depends on different types of hardware. So add new interfaces to get them
from firmware.
Jason Gunthorpe [Wed, 8 Jan 2020 17:22:06 +0000 (19:22 +0200)]
RDMA/core: Use READ_ONCE for ib_ufile.async_file
The writer for async_file holds the ucontext_lock, while the readers are
left unlocked. Most readers rely on an implicit locking, either by having
a uobject (which cannot be created before a context) or by holding the
ib_ufile kref.
However ib_uverbs_free_hw_resources() has no implicit lock and has a
possible race. Make this all clear and sane by using READ_ONCE
consistently.
Jason Gunthorpe [Wed, 8 Jan 2020 17:22:04 +0000 (19:22 +0200)]
RDMA/core: Remove the ufile arg from rdma_alloc_begin_uobject
Now that all callers provide a non-NULL attrs the ufile is redundant.
Adjust things so that the context handling is done inside alloc_uobj,
and the ib_uverbs_get_ucontext_file() is avoided if we already have the
context.
Jason Gunthorpe [Wed, 8 Jan 2020 17:22:02 +0000 (19:22 +0200)]
RDMA/core: Simplify type usage for ib_uverbs_async_handler()
This function works on an ib_uverbs_async_file. Accept that as a parameter
instead of the struct ib_uverbs_file.
Consoldiate all the callers working from an ib_uevent_object to a single
function and locate the async_file directly from the struct ib_uobject
instead of using context_ptr.
Jason Gunthorpe [Wed, 8 Jan 2020 17:21:57 +0000 (19:21 +0200)]
RDMA/core: Make ib_ucq_object use ib_uevent_object
Any uobject that sends events into the async_event_file should be using
ib_uevent_object so it can use the standard uevent based helper
functions. CQ pushes events into both the async_event and the comp_channel
in an open coded way. Move the async events related stuff to
ib_uevent_object.
Jason Gunthorpe [Wed, 8 Jan 2020 17:21:56 +0000 (19:21 +0200)]
RDMA/core: Do not allow alloc_commit to fail
This is a left over from an earlier version that creates a lot of
complexity for error unwind, particularly for FD uobjects.
The only reason this was done is so that anon_inode_get_file() could be
called with the final fops and a fully setup uobject. Both need to be
setup since unwinding anon_inode_get_file() via fput will call the
driver's release().
Now that the driver does not provide release, we no longer need to worry
about this complicated sequence, simply create the struct file at the
start and allow the core code's release function to deal with the abort
case.
This allows all the confusing error paths around commit to be removed.
Jason Gunthorpe [Wed, 8 Jan 2020 17:21:55 +0000 (19:21 +0200)]
RDMA/mlx5: Simplify devx async commands
With the new FD structure the async commands do not need to hold any
references while running. The existing mlx5_cmd_exec_cb() and
mlx5_cmd_cleanup_async_ctx() provide enough synchronization to ensure
that all outstanding commands are completed before the uobject can be
destructed.
Remove the now confusing get_file() and the type erasure of the
devx_async_cmd_event_file.
Jason Gunthorpe [Wed, 8 Jan 2020 17:21:54 +0000 (19:21 +0200)]
RDMA/core: Simplify destruction of FD uobjects
FD uobjects have a weird split between the struct file and uobject
world. Simplify this to make them pure uobjects and use a generic release
method for all struct file operations.
This fixes the control flow so that mlx5_cmd_cleanup_async_ctx() is always
called before erasing the linked list contents to make the concurrancy
simpler to understand.
For this to work the uobject destruction must fence anything that it is
cleaning up - the design must not rely on struct file lifetime.
Only deliver_event() relies on the struct file to when adding new events
to the queue, add a is_destroyed check under lock to block it.
Jason Gunthorpe [Wed, 8 Jan 2020 17:21:53 +0000 (19:21 +0200)]
RDMA/mlx5: Use RCU and direct refcounts to keep memory alive
dispatch_event_fd() runs from a notifier with minimal locking, and relies
on RCU and a file refcount to keep the uobject and eventfd alive.
As the next patch wants to remove the file_operations release function
from the drivers, re-organize things so that the devx_event_notifier()
path uses the existing RCU to manage the lifetime of the uobject and
eventfd.
Move the refcount puts to a call_rcu so that the objects are guaranteed to
exist and remove the indirect file refcount.
Jason Gunthorpe [Mon, 13 Jan 2020 14:33:10 +0000 (14:33 +0000)]
RDMA/uverbs: Remove needs_kfree_rcu from uverbs_obj_type_class
After device disassociation the uapi_objects are destroyed and freed,
however it is still possible that core code can be holding a kref on the
uobject. When it finally goes to uverbs_uobject_free() via the kref_put()
it can trigger a use-after-free on the uapi_object.
Since needs_kfree_rcu is a micro optimization that only benefits file
uobjects, just get rid of it. There is no harm in using kfree_rcu even if
it isn't required, and the number of involved objects is small.
Expose vDPA emulation device capabilities from the core layer.
It includes reading the capabilities from the firmware and exposing
helper functions to access the data.
Xi Wang [Thu, 9 Jan 2020 12:20:12 +0000 (20:20 +0800)]
RDMA/hns: Add support for reporting wc as software mode
When hardware is in resetting stage, we may can't poll back all the
expected work completions as the hardware won't generate cqe anymore.
This patch allows the driver to compose the expected wc instead of the
hardware during resetting stage. Once the hardware finished resetting, we
can poll cq from hardware again.
Lijun Ou [Thu, 9 Jan 2020 12:10:52 +0000 (20:10 +0800)]
RDMA/hns: Bugfix for posting a wqe with sge
Driver should first check whether the sge is valid, then fill the valid
sge and the caculated total into hardware, otherwise invalid sges will
cause an error.
Fixes: 52e3b42a2f58 ("RDMA/hns: Filter for zero length of sge in hip08 kernel mode") Fixes: 7bdee4158b37 ("RDMA/hns: Fill sq wqe context of ud type in hip08") Link: https://lore.kernel.org/r/1578571852-13704-1-git-send-email-liweihang@huawei.com Signed-off-by: Lijun Ou <oulijun@huawei.com> Signed-off-by: Weihang Li <liweihang@huawei.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
All other code paths increment some form of drop counter.
This was missed in the original implementation.
Fixes: 82c2611daaf0 ("staging/rdma/hfi1: Handle packets with invalid RHF on context 0") Link: https://lore.kernel.org/r/20200106134228.119356.96828.stgit@awfm-01.aw.intel.com Reviewed-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com> Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
IB/hfi1: Return void in packet receiving functions
Packet receiving functions returns int value, and yet the return values
are not used at all.
This patch converts the functions to return void.
Link: https://lore.kernel.org/r/20200106134222.119356.84098.stgit@awfm-01.aw.intel.com Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com> Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Grzegorz Andrejczuk <grzegorz.andrejczuk@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>
IRQ name was connected to IRQ type, this is not sufficient and it would be
better to use name as argument to msix_request_irq instead of assigning it
to variables when function is called.
Index argument was required to generate name and now it can be removed.
To generate name correctly helpers function were added and updated.
Link: https://lore.kernel.org/r/20200106134216.119356.44478.stgit@awfm-01.aw.intel.com Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com> Reviewed-by: Michael J. Ruhl <michael.j.ruhl@intel.com> Signed-off-by: Grzegorz Andrejczuk <grzegorz.andrejczuk@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>
Add an auto activate routine for use by the interrupt handler.
Link: https://lore.kernel.org/r/20200106134210.119356.43079.stgit@awfm-01.aw.intel.com Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Mike Marciniszyn <mike.marciniszyn@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>
IB/hfi1: IB/hfi1: Add an API to handle special case drop
This patch pushes special case drop logic into an API to be shared by all
interrupt handlers.
Additionally, convert do_drop to a bool.
Link: https://lore.kernel.org/r/20200106134203.119356.36962.stgit@awfm-01.aw.intel.com Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Mike Marciniszyn <mike.marciniszyn@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>
Tracing interrupts, incrementing interrupt counter and ASPM are part that
will be reused by HFI1 receive IRQ handlers.
Create common function to have shared code in one place.
Link: https://lore.kernel.org/r/20200106134157.119356.32656.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: Grzegorz Andrejczuk <grzegorz.andrejczuk@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>
IB/hfi1: Add fast and slow handlers for receive context
This patch eliminate special cases by adding a fast_handler member to the
receive context and changes to the fast handler as specified in the new
variable. Initialize the variable as soon as the setting for dma tail is
known when the context is created.
Setting fast path is called every time when any context has entered slow
path. Add function to check if contexts is using fast path and do not set
fast path when it is already done to improve RCD fastpath setting.
Link: https://lore.kernel.org/r/20200106134150.119356.87558.stgit@awfm-01.aw.intel.com Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Reviewed-by: Michael J. Ruhl <michael.j.ruhl@intel.com> Signed-off-by: Grzegorz Andrejczuk <grzegorz.andrejczuk@intel.com> Signed-off-by: Sadanand Warrier <sadanand.warrier@intel.com> Signed-off-by: Mike Marciniszyn <mike.marciniszyn@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>
Move routines and defines associated with hdrq size validation to a chip
specific routine since the limits are specific to the device.
Fix incorrect value for min size 2 -> 32
CSR writes should also be in chip.c.
Create a chip routine to write the hdrq specific CSRs and call as
appropriate.
Link: https://lore.kernel.org/r/20200106134144.119356.74312.stgit@awfm-01.aw.intel.com Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Reviewed-by: Michael J. Ruhl <michael.j.ruhl@intel.com> Signed-off-by: Mike Marciniszyn <mike.marciniszyn@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>
Jason Gunthorpe [Wed, 8 Jan 2020 17:22:03 +0000 (19:22 +0200)]
RDMA/core: Fix locking in ib_uverbs_event_read
This should not be using ib_dev to test for disassociation, during
disassociation is_closed is set under lock and the waitq is triggered.
Instead check is_closed and be sure to re-obtain the lock to test the
value after the wait_event returns.
Fixes: 036b10635739 ("IB/uverbs: Enable device removal when there are active user space applications") Link: https://lore.kernel.org/r/1578504126-9400-12-git-send-email-yishaih@mellanox.com Signed-off-by: Yishai Hadas <yishaih@mellanox.com> Reviewed-by: Håkon Bugge <haakon.bugge@oracle.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Arnd Bergmann [Thu, 9 Jan 2020 08:47:30 +0000 (09:47 +0100)]
IB/core: Fix build failure without hugepages
HPAGE_SHIFT is only defined on architectures that support hugepages:
drivers/infiniband/core/umem_odp.c: In function 'ib_umem_odp_get':
drivers/infiniband/core/umem_odp.c:245:26: error: 'HPAGE_SHIFT' undeclared (first use in this function); did you mean 'PAGE_SHIFT'?
Enclose this in an #ifdef.
Fixes: 9ff1b6466a29 ("IB/core: Fix ODP with IB_ACCESS_HUGETLB handling") Link: https://lore.kernel.org/r/20200109084740.2872079-1-arnd@arndb.de Signed-off-by: Arnd Bergmann <arnd@arndb.de> Reviewed-by: Jason Gunthorpe <jgg@mellanox.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Parav Pandit [Thu, 12 Dec 2019 11:30:24 +0000 (13:30 +0200)]
IB/core: Rename event_handler_lock to qp_open_list_lock
This lock is used to protect the qp->open_list linked list. As a side
effect it seems to also globally serialize the qp event_handler, but it
isn't clear if that is a deliberate design.
Parav Pandit [Thu, 12 Dec 2019 11:30:22 +0000 (13:30 +0200)]
IB/core: Let IB core distribute cache update events
Currently when the low level driver notifies Pkey, GID, and port change
events they are notified to the registered handlers in the order they are
registered.
IB core and other ULPs such as IPoIB are interested in GID, LID, Pkey
change events.
Since all GID queries done by ULPs are serviced by IB core, and the IB
core deferes cache updates to a work queue, it is possible for other
clients to see stale cache data when they handle their own events.
For example, the below call tree shows how ipoib will call
rdma_query_gid() concurrently with the update to the cache sitting in the
WQ.
Lijun Ou [Mon, 6 Jan 2020 12:21:12 +0000 (20:21 +0800)]
RDMA/hns: Update the value of qp type
The values used to represent service type of RC and UD should be
interchanged according to design of hardware. And it's better to define
these types in enumeration than macros.
Some features of the output:
- The lifetime and owner PD of each MR is clearly visible.
- The type of MR is captured, as is the SGE array size.
- Failing MR allocation can be recorded.
Chuck Lever [Wed, 18 Dec 2019 20:18:15 +0000 (15:18 -0500)]
RDMA/core: Trace points for diagnosing completion queue issues
Sample trace events:
kworker/u29:0-300 [007] 120.042217: cq_alloc: cq.id=4 nr_cqe=161 comp_vector=2 poll_ctx=WORKQUEUE
<idle>-0 [002] 120.056292: cq_schedule: cq.id=4
kworker/2:1H-482 [002] 120.056402: cq_process: cq.id=4 wake-up took 109 [us] from interrupt
kworker/2:1H-482 [002] 120.056407: cq_poll: cq.id=4 requested 16, returned 1
<idle>-0 [002] 120.067503: cq_schedule: cq.id=4
kworker/2:1H-482 [002] 120.067537: cq_process: cq.id=4 wake-up took 34 [us] from interrupt
kworker/2:1H-482 [002] 120.067541: cq_poll: cq.id=4 requested 16, returned 1
<idle>-0 [002] 120.067657: cq_schedule: cq.id=4
kworker/2:1H-482 [002] 120.067672: cq_process: cq.id=4 wake-up took 15 [us] from interrupt
kworker/2:1H-482 [002] 120.067674: cq_poll: cq.id=4 requested 16, returned 1
...
systemd-1 [002] 122.392653: cq_schedule: cq.id=4
kworker/2:1H-482 [002] 122.392688: cq_process: cq.id=4 wake-up took 35 [us] from interrupt
kworker/2:1H-482 [002] 122.392693: cq_poll: cq.id=4 requested 16, returned 16
kworker/2:1H-482 [002] 122.392836: cq_poll: cq.id=4 requested 16, returned 16
kworker/2:1H-482 [002] 122.392970: cq_poll: cq.id=4 requested 16, returned 16
kworker/2:1H-482 [002] 122.393083: cq_poll: cq.id=4 requested 16, returned 16
kworker/2:1H-482 [002] 122.393195: cq_poll: cq.id=4 requested 16, returned 3
Several features to note in this output:
- The WCE count and context type are reported at allocation time
- The CPU and kworker for each CQ is evident
- The CQ's restracker ID is tagged on each trace event
- CQ poll scheduling latency is measured
- Details about how often single completions occur versus multiple
completions are evident
- The cost of the ULP's completion handler is recorded
Chuck Lever [Wed, 18 Dec 2019 20:18:10 +0000 (15:18 -0500)]
RDMA/cma: Add trace points in RDMA Connection Manager
Record state transitions as each connection is established. The IP address
of both peers and the Type of Service is reported. These trace points are
not in performance hot paths.
Also, record each cm_event_handler call to ULPs. This eliminates the need
for each ULP to add its own similar trace point in its CM event handler
function.
These new trace points appear in a new trace subsystem called "rdma_cma".
Some features to note:
- restracker ID of the rdma_cm_id is tagged on each trace event
- The source and destination IP addresses and TOS are reported
- CM event upcalls are shown with decoded event and status
- CM state transitions are reported
- rdma_cm_id lifetime events are captured
- The latency of ULP CM event handlers is reported
- Lifetime events of associated QPs are reported
- Device removal and insertion is reported
Jiewei Ke [Fri, 27 Dec 2019 11:36:13 +0000 (19:36 +0800)]
RDMA/rxe: Fix error type of mmap_offset
The type of mmap_offset should be u64 instead of int to match the type of
mminfo.offset. If otherwise, after we create several thousands of CQs, it
will run into overflow issues.
zhengbin [Tue, 24 Dec 2019 08:40:12 +0000 (16:40 +0800)]
RDMA/mlx5: use true,false for bool variable
Fixes coccicheck warning:
drivers/infiniband/hw/mlx5/mr.c:150:2-26: WARNING: Assignment of 0/1 to bool variable
drivers/infiniband/hw/mlx5/mr.c:1455:2-26: WARNING: Assignment of 0/1 to bool variable
drivers/infiniband/hw/mlx5/qp.c:1874:6-20: WARNING: Assignment of 0/1 to bool variable
zhengbin [Tue, 24 Dec 2019 08:40:11 +0000 (16:40 +0800)]
RDMA/mlx4: use true,false for bool variable
Fixes coccicheck warning:
drivers/infiniband/hw/mlx4/qp.c:852:2-14: WARNING: Assignment of 0/1 to bool variable
drivers/infiniband/hw/mlx4/qp.c:3087:3-10: WARNING: Assignment of 0/1 to bool variable
zhengbin [Tue, 24 Dec 2019 08:40:10 +0000 (16:40 +0800)]
IB/iser: use true,false for bool variable
Fixes coccicheck warning:
drivers/infiniband/ulp/iser/iser_memory.c:530:2-21: WARNING: Assignment of 0/1 to bool variable
drivers/infiniband/ulp/iser/iser_verbs.c:1096:2-21: WARNING: Assignment of 0/1 to bool variable
Yishai Hadas [Sun, 22 Dec 2019 12:46:49 +0000 (14:46 +0200)]
IB/core: Fix ODP with IB_ACCESS_HUGETLB handling
As VMAs for a given range might not be available as part of the
registration phase in ODP.
ib_init_umem_odp() considered the expected page shift value that was
previously set and initializes its internals accordingly.
If memory isn't backed by physical contiguous pages aligned to a hugepage
boundary an error will be set as part of the page fault flow and come back
to the user as some failed RDMA operation.
Fixes: 0008b84ea9af ("IB/umem: Add support to huge ODP") Link: https://lore.kernel.org/r/20191222124649.52300-4-leon@kernel.org Signed-off-by: Yishai Hadas <yishaih@mellanox.com> Reviewed-by: Artemy Kovalyov <artemyko@mellanox.com> Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Reviewed-by: Jason Gunthorpe <jgg@mellanox.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Yishai Hadas [Sun, 22 Dec 2019 12:46:48 +0000 (14:46 +0200)]
IB/core: Fix ODP get user pages flow
The nr_pages argument of get_user_pages_remote() should always be in terms
of the system page size, not the MR page size. Use PAGE_SIZE instead of
umem_odp->page_shift.
Fixes: 403cd12e2cf7 ("IB/umem: Add contiguous ODP support") Link: https://lore.kernel.org/r/20191222124649.52300-3-leon@kernel.org Signed-off-by: Yishai Hadas <yishaih@mellanox.com> Reviewed-by: Artemy Kovalyov <artemyko@mellanox.com> Reviewed-by: Jason Gunthorpe <jgg@mellanox.com> Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Artemy Kovalyov [Sun, 22 Dec 2019 12:46:47 +0000 (14:46 +0200)]
IB/mlx5: Unify ODP MR code paths to allow extra flexibility
Building MR translation table in the ODP case requires additional
flexibility, namely random access to DMA addresses. Make both direct and
indirect ODP MR use same code path, separated from the non-ODP MR code
path.
With the restructuring the correct page_shift is now used around
__mlx5_ib_populate_pas().
Fixes: d2183c6f1958 ("RDMA/umem: Move page_shift from ib_umem to ib_odp_umem") Link: https://lore.kernel.org/r/20191222124649.52300-2-leon@kernel.org Signed-off-by: Artemy Kovalyov <artemyko@mellanox.com> Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Mike Marciniszyn [Thu, 19 Dec 2019 21:19:34 +0000 (16:19 -0500)]
IB/rdmavt: Correct comments in rdmavt_qp.h header
Comments need to be with the definition of rvt_restart_sge().
Other comments were duplicated in sw/rdmavt/rc.c and were removed.
Fixes: 385156c5f2a6 ("IB/hfi: Move RC functions into a header file") Link: https://lore.kernel.org/r/20191219211934.58387.88014.stgit@awfm-01.aw.intel.com Reviewed-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com> Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Michael J. Ruhl [Thu, 19 Dec 2019 21:19:28 +0000 (16:19 -0500)]
IB/hfi1: List all receive contexts from debugfs
The current debugfs output for receive contexts (rcds), stops after the
kernel receive contexts have been displayed. This is not enough
information to fully diagnose packet drops.
Display all of the receive contexts.
Augment the output with some more context information.
Limit the ring buffer header output to 5 entries to avoid overextending
the sequential file output.
Fixes: bf808b5039c ("IB/hfi1: Add kernel receive context info to debugfs") Link: https://lore.kernel.org/r/20191219211928.58387.20737.stgit@awfm-01.aw.intel.com Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com> Signed-off-by: Michael J. Ruhl <michael.j.ruhl@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>
Mike Marciniszyn [Thu, 19 Dec 2019 21:19:22 +0000 (16:19 -0500)]
IB/hfi1: Add accessor API routines to access context members
This patch adds a set of accessor routines to access context members.
Link: https://lore.kernel.org/r/20191219211922.58387.26548.stgit@awfm-01.aw.intel.com Reviewed-by: Michael J. Ruhl <michael.j.ruhl@intel.com> Signed-off-by: Mike Marciniszyn <mike.marciniszyn@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>
Eugene Crosser [Thu, 19 Dec 2019 13:48:47 +0000 (15:48 +0200)]
RDMA/mlx4: Redo TX checksum offload in line with docs
Ingress checksum offload was not working for IPv6 frames because the
conditional expression that checks validation status passed from the
hardware was not matching the algorithm described in the documentation.
This patch defines L4_CSUM flag (which falls inside the badfcs_enc field
in the existing definition of the CQE layout) and replaces the conditional
expression with the one defined in the "ConnectX(r) Family Programmer's
Manual" document.
Link: https://lore.kernel.org/r/20191219134847.413582-1-leon@kernel.org Signed-off-by: Eugene Crosser <evgenii.cherkashin@profitbricks.com> Reviewed-by: Jack Wang <jinpu.wang@profitbricks.com> Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Danit Goldberg [Thu, 19 Dec 2019 13:47:50 +0000 (15:47 +0200)]
RDMA/cm: Use RCU synchronization mechanism to protect cm_id_private xa_load()
The RCU mechanism is optimized for read-mostly scenarios and therefore
more suitable to protect the cm_id_private to decrease "cm.lock"
congestion.
This patch replaces the existing spinlock locking mechanism and kfree with
RCU mechanism in places where spinlock(cm.lock) protected xa_load
returning the cm_id_priv
In addition, delete the cm_get_id() function as there is no longer a
distinction if the caller already holds the cm_lock.
Remove an open coded version of cm_get_id().
Link: https://lore.kernel.org/r/20191219134750.413429-1-leon@kernel.org Signed-off-by: Danit Goldberg <danitg@mellanox.com> Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Reviewed-by: Jason Gunthorpe <jgg@mellanox.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Aditya Pakki [Tue, 17 Dec 2019 19:44:37 +0000 (13:44 -0600)]
RDMA/srpt: Remove unnecessary assertion in srpt_queue_response
Since ch has already been de-referenced by the time we get to the BUG_ON,
it is useless. The back trace alone is enough to tell what is going on,
delete the redundant BUG_ON.
Link: https://lore.kernel.org/r/20191217194437.25568-1-pakki001@umn.edu Signed-off-by: Aditya Pakki <pakki001@umn.edu> Reviewed-by: Bart Van Assche <bvanassche@acm.org> Reviewed-by: Jason Gunthorpe <jgg@mellanox.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Håkon Bugge [Mon, 16 Dec 2019 12:04:36 +0000 (13:04 +0100)]
RDMA/netlink: Do not always generate an ACK for some netlink operations
In rdma_nl_rcv_skb(), the local variable err is assigned the return value
of the supplied callback function, which could be one of
ib_nl_handle_resolve_resp(), ib_nl_handle_set_timeout(), or
ib_nl_handle_ip_res_resp(). These three functions all return skb->len on
success.
rdma_nl_rcv_skb() is merely a copy of netlink_rcv_skb(). The callback
functions used by the latter have the convention: "Returns 0 on success or
a negative error code".
In particular, the statement (equal for both functions):
if (nlh->nlmsg_flags & NLM_F_ACK || err)
implies that rdma_nl_rcv_skb() always will ack a message, independent of
the NLM_F_ACK being set in nlmsg_flags or not.
The fix could be to change the above statement, but it is better to keep
the two *_rcv_skb() functions equal in this respect and instead change the
three callback functions in the rdma subsystem to the correct convention.
Fixes: 2ca546b92a02 ("IB/sa: Route SA pathrecord query through netlink") Fixes: ae43f8286730 ("IB/core: Add IP to GID netlink offload") Link: https://lore.kernel.org/r/20191216120436.3204814-1-haakon.bugge@oracle.com Suggested-by: Mark Haywood <mark.haywood@oracle.com> Signed-off-by: Håkon Bugge <haakon.bugge@oracle.com> Tested-by: Mark Haywood <mark.haywood@oracle.com> Reviewed-by: Leon Romanovsky <leonro@mellanox.com> Reviewed-by: Jason Gunthorpe <jgg@mellanox.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Prabhath Sajeepa [Fri, 13 Dec 2019 00:11:29 +0000 (17:11 -0700)]
IB/mlx5: Fix outstanding_pi index for GSI qps
Commit b0ffeb537f3a ("IB/mlx5: Fix iteration overrun in GSI qps") changed
the way outstanding WRs are tracked for the GSI QP. But the fix did not
cover the case when a call to ib_post_send() fails and updates index to
track outstanding.
Since the prior commmit outstanding_pi should not be bounded otherwise the
loop generate_completions() will fail.
Bernard Metzler [Tue, 10 Dec 2019 16:17:29 +0000 (17:17 +0100)]
RDMA/siw: Simplify QP representation
Change siw_qp to contain ib_qp. Use rdma_is_kernel_res() on contained
ib_qp to distinguish kernel level from user level applications
resources. Apply same mechanism for kernel/user level application
detection to completion queues.
Yixian Liu [Tue, 10 Dec 2019 12:45:02 +0000 (20:45 +0800)]
RDMA/hns: Simplify the calculation and usage of wqe idx for post verbs
Currently, the wqe idx is calculated repeatly everywhere it is used. This
patch defines wqe_idx and calculated it only once, then just use it as
needed.
Selvin Xavier [Mon, 25 Nov 2019 08:39:34 +0000 (00:39 -0800)]
RDMA/bnxt_re: Report more number of completion vectors
Report the the data path MSIx vectors allocated by driver as number of
completion vectors. One interrupt vector is used for Control path. So
reporting one less than the total number of MSIx vectors allocated by the
driver.
Michal Kalderon [Thu, 21 Nov 2019 11:29:57 +0000 (13:29 +0200)]
RDMA/qedr: Add kernel capability flags for dpm enabled mode
HW/FW support two types of latency enhancement features. Until now
user-space implemented only edpm (enhanced dpm). We add kernel capability
flags to differentiate between current FW in kernel that supports both
ldpm and edpm. Since edpm is not yet supported for iWARP we add different
flags for iWARP + RoCE. We also fix bad practice of defining sizes in
rdma-core and pass initialization to kernel, for forward compatibility.
The capability flags are added for backward-forward compatibility between
kernel and rdma-core for qedr.
Before this change there was a field called dpm_enabled which could hold
either 0 or 1 value, this indicated whether RoCE edpm was enabled or
not. We modified this field to be dpm_flags, and bit 1 still holds the
same meaning of RoCE edpm being enabled or not.
Linus Torvalds [Sun, 29 Dec 2019 18:01:27 +0000 (10:01 -0800)]
Merge tag 'riscv/for-v5.5-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/riscv/linux
Pull RISC-V fixes from Paul Walmsley:
"One important fix for RISC-V:
- Redirect any incoming syscall with an ID less than -1 to
sys_ni_syscall, rather than allowing them to fall through into the
syscall handler.
and two minor build fixes:
- Export __asm_copy_{from,to}_user() from where they are defined.
This fixes a build error triggered by some randconfigs.
- Export flush_icache_all(). I'd resisted this before, since
historically we didn't want modules to be able to flush the I$
directly; but apparently everyone else is doing it now"
* tag 'riscv/for-v5.5-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/riscv/linux:
riscv: export flush_icache_all to modules
riscv: reject invalid syscalls below -1
riscv: fix compile failure with EXPORT_SYMBOL() & !MMU
Linus Torvalds [Sun, 29 Dec 2019 17:50:57 +0000 (09:50 -0800)]
Merge tag 'locks-v5.5-1' of git://git.kernel.org/pub/scm/linux/kernel/git/jlayton/linux
Pull /proc/locks formatting fix from Jeff Layton:
"This is a trivial fix for a _very_ long standing bug in /proc/locks
formatting. Ordinarily, I'd wait for the merge window for something
like this, but it is making it difficult to validate some overlayfs
fixes.
I've also gone ahead and marked this for stable"
* tag 'locks-v5.5-1' of git://git.kernel.org/pub/scm/linux/kernel/git/jlayton/linux:
locks: print unsigned ino in /proc/locks
Linus Torvalds [Sun, 29 Dec 2019 17:48:47 +0000 (09:48 -0800)]
Merge tag '5.5-rc3-smb3-fixes' of git://git.samba.org/sfrench/cifs-2.6
Pull cifs fixes from Steve French:
"One performance fix for large directory searches, and one minor style
cleanup noticed by Clang"
* tag '5.5-rc3-smb3-fixes' of git://git.samba.org/sfrench/cifs-2.6:
cifs: Optimize readdir on reparse points
cifs: Adjust indentation in smb2_open_file
Olof Johansson [Tue, 17 Dec 2019 04:07:04 +0000 (20:07 -0800)]
riscv: export flush_icache_all to modules
This is needed by LKDTM (crash dump test module), it calls
flush_icache_range(), which on RISC-V turns into flush_icache_all(). On
other architectures, the actual implementation is exported, so follow
that precedence and export it here too.
Fixes build of CONFIG_LKDTM that fails with:
ERROR: "flush_icache_all" [drivers/misc/lkdtm/lkdtm.ko] undefined!
Signed-off-by: Olof Johansson <olof@lixom.net> Signed-off-by: Paul Walmsley <paul.walmsley@sifive.com>
Running "stress-ng --enosys 4 -t 20 -v" showed a large number of kernel oops
with "Unable to handle kernel paging request at virtual address" message. This
happens when enosys stressor starts testing random non-valid syscalls.
I forgot to redirect any syscall below -1 to sys_ni_syscall.
With the patch kernel oops messages are gone while running stress-ng enosys
stressor.
Signed-off-by: David Abdurachmanov <david.abdurachmanov@sifive.com> Fixes: 5340627e3fe0 ("riscv: add support for SECCOMP and SECCOMP_FILTER") Signed-off-by: Paul Walmsley <paul.walmsley@sifive.com>
riscv: fix compile failure with EXPORT_SYMBOL() & !MMU
When support for !MMU was added, the declaration of
__asm_copy_to_user() & __asm_copy_from_user() were #ifdefed
out hence their EXPORT_SYMBOL() give an error message like:
.../riscv_ksyms.c:13:15: error: '__asm_copy_to_user' undeclared here
.../riscv_ksyms.c:14:15: error: '__asm_copy_from_user' undeclared here
Since these symbols are not defined with !MMU it's wrong to export them.
Same for __clear_user() (even though this one is also declared in
include/asm-generic/uaccess.h and thus doesn't give an error message).
Fix this by doing the EXPORT_SYMBOL() directly where these symbols
are defined: inside lib/uaccess.S itself.
Fixes: 6bd33e1ece52 ("riscv: fix compile failure with EXPORT_SYMBOL() & !MMU") Reported-by: kbuild test robot <lkp@intel.com> Cc: Christoph Hellwig <hch@lst.de> Cc: Palmer Dabbelt <palmer@dabbelt.com> Cc: Paul Walmsley <paul.walmsley@sifive.com> Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com> Signed-off-by: Paul Walmsley <paul.walmsley@sifive.com>