Hi Linus,
Here are the RDMA patches for 4.19.
This time there are many merge conflicts, all due to various tree-wide or RDMA
subystem wide changes done by various people. The resolution is tricky, as git
does not highlight two areas that need revision.
An overview of the conflicts:
- Mark Rutland renamed __atomic_add_unless() conflicting with major rework of
rdma_core.c
- Steve Wise split ib_device_attr.max_sge into max_send_sge and max_recv_sge,
and updated all ULPs. This clashed with other changes in the NVMe over
fabrics driver adding new users of max_sge. This is tricky as the correct
use of send/recv in the merge is essential. Christoph Hellwig has reviewded
this resolution in linux-next
- Bart did a tree wide change to use NULL for the last argument in
ibv_post_send(), clashing with various RDP ULP changes in netdev
- Parav revised the ugly ib_query_gid API with something saner and
did a treewide update. This clashed horribly with general work on the SMC
code in netdev. We've handled this by providing a compat ib_query_gid wrapper
and reverting the SMC changes in rdma.git, so no merge fixup neeed.
However, I will send a cleanup for this to remove the wrapper as soon as
possible.
- Various small conflicts between RDMA -rc and -next resolved by merging the
v4.18 tag.
Notably we have again avoided the historical merge conflicts in the Mellanox
drivers with the new 'shared branch' arrangement with DaveM, this seems to be
working well.
The above tree wide changes explain most of the hunks outside the tree, the
only other cases are a new function check_shl_overflow() in
include/linux/overflow.h ack'd by Kees and an edit to idr.h from Matthew Wilcox
that was needed for other patches he sent to the RDMA tree.
As before I've resolved the conflicts in the for-next tag (this makes the
diffstat below more accurate) and provided the unmerged tree in the
for-linus-unmerged tag. The merge resolution diff is a bit big so it is
included at the end of this message for your reference.
The following changes since commit 5c60a7389d795e001c8748b458eb76e3a5b6008c:
Merge tag 'for-linus-4.19-ofs1' of git://git.kernel.org/pub/scm/linux/kernel/git/hubcap/linux (2018-08-16 10:53:45 -0700)
are available in the Git repository at:
git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma.git tags/for-linus
for you to fetch changes up to 0a3173a5f09bc58a3638ecfd0a80bdbae55e123c:
Merge branch 'linus/master' into rdma.git for-next (2018-08-16 14:21:29 -0600)
----------------------------------------------------------------
This has been a large cycle for RDMA, with several major patch series
reworking parts of the core code.
- Rework the so-called 'gid cache' and internal APIs to use a kref'd
pointer to a struct instead of copying, push this upwards into the
callers and add more stuff to the struct. The new design avoids some
ugly races the old one suffered with. This is part of the namespace
enablement work as the new struct is learning to be namespace aware.
- Various uapi cleanups, moving more stuff to include/uapi and fixing some
long standing bugs that have recently been discovered.
- Driver updates for mlx5, mlx4 i40iw, rxe, cxgb4, hfi1, usnic, pvrdma,
and hns
- Provide max_send_sge and max_recv_sge attributes to better support HW
where these values are asymmetric.
- mlx5 user API 'devx' allows sending commands directly to the device FW,
instead of trying to cram every wild and niche feature into the common
API. Sort of like what GPU does.
- Major write() and ioctl() API rework to cleanly support PCI device hot
unplug and advance the ioctl conversion work
- Sparse and compile warning cleanups
- Add 'const' to the ib_poll_cq() signature, and permit a NULL 'bad_wr',
which is the common use case
- Various patches to avoid high order allocations across the stack
- SRQ support for cxgb4, hns and qedr
- Changes to IPoIB to better follow the netdev model for working with
struct net_device liftime
----------------------------------------------------------------
Arnd Bergmann (1):
infiniband: i40iw, nes: don't use wall time for TCP sequence numbers
Artemy Kovalyov (2):
RDMA: Validate grh_required when handling AVs
IB/uverbs: Pass IB_UVERBS_QPF_GRH_REQUIRED to user space
Bart Van Assche (51):
MAINTAINERS: Update SRP entries
IB/srpt: Support HCAs with more than two ports
include/rdma/opa_addr.h: Fix an endianness issue
IB/srp: Remove driver version and release data information
IB/mlx5: Remove set-but-not-used variables
ib_srpt: Fix a use-after-free in srpt_close_ch()
ib_srpt: Fix a use-after-free in __srpt_close_all_ch()
RDMA/core: Remove ib_find_cached_gid() and ib_find_cached_gid_by_port()
RDMA/core: Remove set-but-not-used variables
IB/nes: Fix a compiler warning
RDMA/ocrdma: Remove a set-but-not-used variable
IB/iser: Remove set-but-not-used variables
RDMA/rxe: Simplify the error handling code in rxe_create_ah()
RDMA/cxgb3: Make iwch_poll_cq_one() easier to analyze
RDMA/cxgb4: Make c4iw_poll_cq_one() easier to analyze
RDMA/rw: Fix rdma_rw_ctx_signature_init() kernel-doc header
RDMA/nes: Avoid complaints about unused variables
RDMA/ocrdma: Make ocrdma_destroy_qp() easier to analyze
IB/hfi1: Suppress a compiler warning
hns: Remove a set-but-not-used variable
IB/srpt: Fix srpt_cm_req_recv() error path (1/2)
IB/srpt: Fix srpt_cm_req_recv() error path (2/2)
RDMA/bnxt_re: Modify a fall-through annotation
MAINTAINERS: Remove Dave Goodell from the usnic RDMA driver maintainer list
IB/core: Allow ULPs to specify NULL as the third ib_post_(send|recv|srq_recv)() argument
RDMA/core: Simplify ib_post_(send|recv|srq_recv)() calls
IB/IPoIB: Simplify ib_post_(send|recv|srq_recv)() calls
IB/iser: Simplify ib_post_(send|recv|srq_recv)() calls
IB/isert: Simplify ib_post_(send|recv|srq_recv)() calls
IB/srp: Simplify ib_post_(send|recv|srq_recv)() calls
IB/srpt: Simplify ib_post_(send|recv|srq_recv)() calls
nvme-rdma: Simplify ib_post_(send|recv|srq_recv)() calls
nvmet-rdma: Simplify ib_post_(send|recv|srq_recv)() calls
fs/cifs: Simplify ib_post_(send|recv|srq_recv)() calls
net/9p: Simplify ib_post_(send|recv|srq_recv)() calls
net/rds: Remove two WARN_ON() statements
net/rds: Simplify ib_post_(send|recv|srq_recv)() calls
net/smc: Remove a WARN_ON() statement
net/smc: Simplify ib_post_(send|recv|srq_recv)() calls
net/xprtrdma: Simplify ib_post_(send|recv|srq_recv)() calls
RDMA/ocrdma: Suppress a compiler warning
RDMA/usnic: Suppress a compiler warning
IB/iser: Inline two work request conversion functions
RDMA: Constify the argument of the work request conversion functions
IB/mlx5, ib_post_send(), IB_WR_REG_SIG_MR: Do not modify the 'wr' argument
RDMA, core and ULPs: Declare ib_post_send() and ib_post_recv() arguments const
rdma/cxgb4: Remove a set-but-not-used variable
rdma/cxgb4: Fix SRQ endianness annotations
rdma/cxgb4: Simplify a structure initialization
RDMA/rxe: Set wqe->status correctly if an unexpected response is received
Update the e-mail address of Bart Van Assche
Bharat Potnuri (1):
iw_cxgb4: remove duplicate memcpy() in c4iw_create_listen()
Dan Carpenter (4):
IB/core: type promotion bug in rdma_rw_init_one_mr()
RDMA/bnxt_re: Fix a couple off by one bugs
RDMA/bnxt_re: Fix a bunch of off by one bugs in qplib_fp.c
rdma/cxgb4: fix some info leaks
Denis Drozdov (1):
IB/IPoIB: Set ah valid flag in multicast send flow
Erez Shitrit (1):
IB/ipoib: Use cancel_delayed_work_sync for neigh-clean task
Gal Pressman (1):
RDMA/hns: Fix usage of bitmap allocation functions return values
Håkon Bugge (1):
IB/cm: Remove unused and erroneous msg sequence encoding
Jack Morgenstein (1):
IB/mlx4: Use 4K pages for kernel QP's WQE buffer
Jan Dakinevich (2):
ib_srpt: use kvmalloc to allocate ring pointers
IPoIB: use kvzalloc to allocate an array of bucket pointers
Jann Horn (1):
IB/mlx5: fix uaccess beyond "count" in debugfs read/write handlers
Jason Gunthorpe (90):
IB/core: Provide rdma_ versions of the gid cache API
IB/core: Make rdma_find_gid_by_filter support all protocols
IB/rxe: Do not hide uABI stuff in memcpy
IB/core: Add a sgid_attr pointer to struct rdma_ah_attr
IB{cm, core}: Introduce and use ah_attr copy, move, replace APIs
RDMA: Hold the sgid_attr inside the struct ib_ah/qp
Merge branch 'mellanox/mlx5-next' into RDMA for-next
Merge branch 'icrc-counter' into rdma.git for-next
IB/uverbs: Delete type and id from uverbs_obj_attr
RDMA/core: Remove unused ib cache functions
IB/usnic: Update with bug fixes from core code
RDMA/uverbs: Check existence of create_flow callback
IB/mlx4: Create slave AH's directly
IB/core: Check for rdma_protocol_ib only after validating port_num
IB/cm: Remove now useless rcu_lock in dst_fetch_ha
Merge branch 'mlx5-dump-fill-mkey' into rdma.git for-next
RDMA/uverbs: Store the specs_root in the struct ib_uverbs_device
RDMA/uverbs: Split UVERBS_ATTR_FLOW_ACTION_ESP_HANDLE
RDMA/uverbs: Simplify UVERBS_ATTR family of macros
RDMA/uverbs: Simplify method definition macros
RDMA/uverbs: Simplify UVERBS_OBJECT and _TREE family of macros
RDMA/uverbs: Get rid of the & in method specifications
RDMA/uverbs: Remove UA_FLAGS
RDMA/uverbs: Use UVERBS_ATTR_MIN_SIZE correctly and uniformly
RDMA/uverbs: Combine MIN_SZ_OR_ZERO with UVERBS_ATTR_STRUCT
IB/uverbs: Add a uobj_perform_destroy helper
IB/uverbs: Move non driver related elements from ib_ucontext to ib_ufile
IB/uverbs: Replace ib_ucontext with ib_uverbs_file in core function calls
IB/uverbs: Replace file->ucontext with file in uverbs_cmd.c
IB/uverbs: Tidy up remaining references to ucontext
IB/uverbs: Remove ib_uobject_file
IB/uverbs: Replace ib_ucq_object uverbs_file with the one in ib_uobject
IB/uverbs: Do not use uverbs_cmd_mask in the ioctl path
RDMA: Fix storage of PortInfo CapabilityMask in the kernel
RDMA/hfi1: Move grh_required into update_sm_ah
RDMA/cxgb4: Restore the dropped uninitialized_var
IB/cm: Remove cma_multicast->igmp_joined
IB/uverbs: Move ib_access_flags and ib_read_counters_flags to uapi
Merge branch 'mellanox/mlx5-next' into rdma.git for-next
IB/uverbs: Get rid of null_obj_type
IB/uverbs: Handle IDR and FD types without truncation
IB/uverbs: Clarify the kref'ing ordering for alloc_commit
IB/uverbs: Revise the placement of get/puts on uobject
IB/uverbs: Clarify and revise uverbs_close_fd
IB/uverbs: Revise and clarify the rwsem and uobjects_lock
IB/uverbs: Rework the locking for cleaning up the ucontext
IB/uverbs: Always propagate errors from rdma_alloc_commit_uobject()
IB/uverbs: Move the FD uobj type struct file allocation to alloc_commit
IB/mlx5: Use the ucontext from the uobj, not the file
IB/uverbs: Fix locking around struct ib_uverbs_file ucontext
IB/usnic: usnic should not select INFINIBAND_USER_ACCESS
net/xprtrdma: Restore needed argument to ib_post_send
IB/cache: Restore compatibility for ib_query_gid
IB/uverbs: Add UVERBS_ATTR_FLAGS_IN to the specs language
IB/uverbs: Remove rdma_explicit_destroy() from the ioctl methods
IB/uverbs: Make the write path destroy methods use the same flow as ioctl
IB/uverbs: Consolidate uobject destruction
IB/uverbs: Convert 'bool exclusive' into an enum
IB/uverbs: Allow RDMA_REMOVE_DESTROY to work concurrently with disassociate
IB/uverbs: Allow uobject allocation to work concurrently with disassociate
IB/uverbs: Lower the test for ongoing disassociation
IB/uverbs: Do not pass struct ib_device to the write based methods
IB/uverbs: Do not pass struct ib_device to the ioctl methods
IB/uverbs: Do not block disassociate during write()
IB/uverbs: Allow all DESTROY commands to succeed after disassociate
IB/ipoib: Get rid of IPOIB_FLAG_GOING_DOWN
IB/ipoib: Move all uninit code into ndo_uninit
IB/ipoib: Move init code to ndo_init
RDMA/netdev: Use priv_destructor for netdev cleanup
IB/ipoib: Get rid of the sysfs_mutex
IB/ipoib: Do not remove child devices from within the ndo_uninit
IB/ipoib: Maintain the child_intfs list from ndo_init/uninit
IB/ipoib: Consolidate checking of the proposed child interface
overflow.h: Add arithmetic shift helper
IB/uverbs: Fix reading of 32 bit flags
IB/uverbs: Have the core code create the uverbs_root_spec
IB/uverbs: Build the specs into a radix tree at runtime
IB/uverbs: Use uverbs_api to manage the object type inside the uobject
IB/uverbs: Provide implementation private memory for the uverbs_attr_bundle
IB/uverbs: Remove the ib_uverbs_attr pointer from each attr
IB/uverbs: Add a simple allocator to uverbs_attr_bundle
IB/uverbs: Use uverbs_alloc for allocations
IB/uverbs: Use uverbs_api to unmarshal ioctl commands
IB/uverbs: Remove struct uverbs_root_spec and all supporting code
IB/uverbs: Do not check for device disassociation during ioctl
IB/ucm: Fix compiling ucm.c
IB/mlx5: Fix leaking stack memory to userspace
Merge tag 'v4.18' into rdma.git for-next
Revert "net/smc: Replace ib_query_gid with rdma_get_gid_attr"
Merge branch 'linus/master' into rdma.git for-next
Kamal Heib (11):
RDMA/ipoib: Use min_t() macro instead of min()
RDMA/ipoib: Prefer unsigned int to bare use of unsigned
RDMA/ipoib: Fix use of sizeof()
RDMA/ipoib: Fix return code from ipoib_cm_dev_init
RDMA/mlx5: Remove set but not used variables
RDMA/providers: Fix return value from create_srq callbacks
RDMA/ipoib: Fix check for return code from ib_create_srq
RDMA/core: Remove {create,destroy}_ah from mandatory verbs
RDMA/core: Check for verbs callbacks before using them
RDMA/providers: Remove pointless functions
RDMA: Fix return code check in rdma_set_cq_moderation
Kees Cook (1):
test_overflow: Add shift overflow tests
Leon Romanovsky (16):
RDMA/nldev: Return port capability flag for IB only
RDMA/uverbs: Refactor flow_resources_alloc() function
RDMA/mlx5: Refactor transport domain checks
RDMA/verbs: Drop kernel variant of create_flow
RDMA/verbs: Drop kernel variant of destroy_flow
RDMA/uverbs: Don't overwrite NULL pointer with ZERO_SIZE_PTR
RDMA/umem: Don't check for a negative return value of dma_map_sg_attrs()
RDMA/uverbs: Remove redundant check
RDMA/mlx5: Don't leak UARs in case of free fails
RDMA/i40w: Hold read semaphore while looking after VMA
RDMA/mlx5: Remove unused port number parameter
RDMA/mlx5: Melt consecutive calls to alloc_bfreg() in one call
RDMA/mlx5: Check that supplied blue flame index doesn't overflow
RDMA/umem: Don't hold mmap_sem for too long
RDMA/umem: Refactor exit paths in ib_umem_get
RDMA/mlx5: Fix shift overflow in mlx5_ib_create_wq
Lijun Ou (12):
RDMA/hns: Do not overwrite the error code during error unwind in hns_roce_init
RDMA/hns: Add 50GE type of hnae3 device match
RDMA/hns: Return correct error code from hns_roce_v1_rsv_lp_qp()
RDMA/hns: Add illegal hop_num judgement
RDMA/hns: Use delay instead of usleep
RDMA/hns: Update the data type of immediate data
RDMA/hns: Enable modify_cq for uverbs.
RDMA/hns: Only assgin the fields of the av if IB_QP_AV bit is set
RDMA/hns: Assign the value for vlan field of qp context
RDMA/hns: Modify qp will return errno when qp type is illegal
RDMA/hns: Use macro instead of magic number
RDMA/hns: Program the tclass and flow label into the hardware
Maor Gottlieb (1):
IB/mlx5: Fix GRE flow specification
Matan Barak (5):
IB/uverbs: Export uverbs idr and fd types
IB/uverbs: Refactor uverbs_finalize_objects
IB/uverbs: Add PTR_IN attributes that are allocated/copied automatically
IB/uverbs: Add a macro to define a type with no kernel known size
IB/uverbs: Allow an empty namespace in ioctl() framework
Matthew Wilcox (1):
IB/mad: Agent registration is process context only
Michael J. Ruhl (3):
IB/hfi1: Remove INTx support and simplify MSIx usage
IB/hfi1: Set in_use_ctxts bits for user ctxts only
IB/hfi1: Remove incorrect call to do_interrupt callback
Mike Marciniszyn (10):
IB/hfi1: Move normal functions from hfi1_devdata to const array
IB/hfi1: Move rhf_offset from devdata to ctxtdata
IB/hfi1: Remove rcvhdrsize
IB/rdmavt, IB/hfi1: Create device dependent s_flags
IB/hfi1: Remove rcvhdrq_size
IB/hfi1: Remove rcvctrl from ctxtdata
IB/hfi1: Rightsize ctxt_eager_bufs fields
IB/hfi1: Remove unused/writeonly devdata fields
IB/hfi1: Remove caches of chip CSRs
IB/hfi1: Reorg ctxtdata and rightsize fields
Neil Horman (1):
vmw_pvrdma: Release netdev when vmxnet3 module is removed
Or Gerlitz (1):
MAINTAINERS: Moving out...
Parav Pandit (45):
IB/core: Do not set the gid type when reserving default entries
IB/core: Store default GID property per-table instead of per-entry
IB/core: Introduce GID entry reference counts
RDMA: Use GID from the ib_gid_attr during the add_gid() callback
IB/core: Introduce GID attribute get, put and hold APIs
IB/core: Replace ib_query_gid with rdma_get_gid_attr
net/smc: Replace ib_query_gid with rdma_get_gid_attr
IB: Replace ib_query_gid/ib_get_cached_gid with rdma_query_gid
IB/rxe: Use rdma GID API
IB: Ensure that all rdma_ah_attr's are zero initialized
IB/core: Tidy ib_resolve_eth_dmac
RDMA: Convert drivers to use sgid_attr instead of sgid_index
IB/mlx4: Use GID attribute from ah attribute
RDMA: Convert drivers to use the AH's sgid_attr in post_wr paths
RDMA/cma: Consider net namespace while leaving multicast group
IB/core: Free GID table entry during GID deletion
IB: Make ib_init_ah_attr_from_wc set sgid_attr
IB: Make ib_init_ah_from_mcmember set sgid_attr
IB: Make init_ah_attr_grh_fields set sgid_attr
IB/cm: Keep track of the sgid_attr that created the cm id
IB/cm: Pass the sgid_attr through various events
IB/cm: Replace members of sa_path_rec with 'struct sgid_attr *'
IB/cm: Use sgid_attr from the AV
IB/mlx5: Honor cnt_set_id_valid flag instead of set_id
IB/core: Simplify check for RoCE route resolve
IB/core: Introduce and use sgid_attr in CM requests
RDMA/cma: Consider netdevice for RoCE ports
RDMA/cma: Do not ignore net namespace for unbound cm_id
RDMA/cma: Initialize resource type in __rdma_create_id()
RDMA/cma: Simplify rdma_resolve_addr() error flow
RDMA/core: Constify dst_addr argument
RDMA/cma: Constify path record, ib_cm_event, listen_id pointers
RDMA/cma: Get rid of 1 bit boolean
RDMA/core: Return bool instead of int
RDMA/core: Avoid holding lock while initializing fields on stack
RDMA/core: Simplify gid type check in cma_acquire_dev()
RDMA/core: Prefix _ib to IB/RoCE specific functions
IB/ucm: Initialize sgid request GID attribute pointer
IB/core: Add comment for change upper netevent handling
IB/core: Avoid confusing del_netdev_default_ips
IB/core: Delete lower netdevice default GID entries in bonding scenario
IB/core: Consider adding default GIDs of bond device
IB/core: Add default GIDs of the bond master netdev
IB/core: Update GID entries for netdevice whose mac address changes
IB/core: Change filter function return type from int to bool
Potnuri Bharat Teja (3):
iw_cxgb4: RDMA write with immediate support
iw_cxgb4: Support FW write completion WR
iw_cxgb4: pass window scale in flowc work request
Qing Huang (1):
IB/mlx5: avoid excessive warning msgs when creating VFs on 2nd port
Raju Rangoju (3):
rdma/cxgb4: Add support for 64Byte cqes
rdma/cxgb4: Add support for srq functions & structs
rdma/cxgb4: Add support for kernel mode SRQ's
Sagi Grimberg (1):
IB/iser: set can_queue earlier to allow setting higher queue depth
Shiraz Saleem (1):
i40iw: Reorganize acquire/release of locks in i40iw_manage_apbvt
Steve Wise (1):
IB/core: add max_send_sge and max_recv_sge attributes
Talat Batheesh (1):
IB/mlx5: Support RoCE ICRC encapsulated error counter
Tarick Bedeir (1):
IB/mlx4: Test port number before querying type.
Varsha Rao (1):
IB/core: Remove extra parentheses
Vijay Immanuel (3):
IB/rxe: increase max MR limit
IB/rxe: support for 802.1q VLAN on the listener
IB/rxe: don't clear the tx queue on every transfer
Wei Yongjun (1):
IB/ipoib: Fix error return code in ipoib_dev_init()
Yishai Hadas (22):
IB/core: Introduce DECLARE_UVERBS_GLOBAL_METHODS
IB/core: Expose ib_ucontext from a given ib_uverbs_file
IB/mlx5: Introduce DEVX
IB/mlx5: Add support for DEVX general command
IB/mlx5: Add obj create and destroy functionality
IB/mlx5: Add DEVX support for modify and query commands
IB/mlx5: Add support for DEVX query UAR
IB/mlx5: Add DEVX support for memory registration
IB/mlx5: Add DEVX query EQN support
IB/mlx5: Expose DEVX tree
IB/mlx5: Add support for drain SQ & RQ
IB/mlx4: Add support for drain SQ & RQ
IB: Improve uverbs_cleanup_ucontext algorithm
IB: Enable uverbs_destroy_def_handler to be used by drivers
net/mlx5: Add forward compatible support for the FTE match data
net/mlx5: Add support for flow table destination number
IB/mlx5: Introduce flow steering matcher uapi object
IB: Support ib_flow creation in drivers
IB/mlx5: Introduce driver create and destroy flow methods
IB/mlx5: Support adding flow steering rule by raw description
IB/mlx5: Add support for a flow table destination for driver flow steering
IB/mlx5: Enable driver uapi commands for flow steering
Yixian Liu (1):
RDMA/hns: Support flush cqe for hip08 in kernel space
Yonatan Cohen (1):
IB/mlx5: Expose dump and fill memory key
Yuval Bason (3):
qedr: Add wrapping generic structure for qpidr and adjust idr routines.
qedr: Add support for kernel mode SRQ's
qedr: Add user space support for SRQ
Yuval Shaia (1):
RDMA/vmw_pvrdma: Delete unused function
Zhu Yanjun (2):
IB/rxe: avoid unnecessary NULL check
IB/rxe: Drop QP0 silently
oulijun (5):
RDMA/hns: Fix endian conversions and annotations
RDMA/hns: Add TSQ link table support
RDMA/hns: Add TPQ link table support
RDMA/hns: Update the implementation of set_gid
RDMA/hns: Update the implementation of set_mac
willy@xxxxxxxxxxxxx (2):
IDR: Expose the XArray lock
IB/mad: Use IDR for agent IDs
.mailmap | 2 +
MAINTAINERS | 16 +-
drivers/infiniband/Kconfig | 2 +-
drivers/infiniband/core/Makefile | 5 +-
drivers/infiniband/core/addr.c | 16 +-
drivers/infiniband/core/cache.c | 724 ++++++++-----
drivers/infiniband/core/cm.c | 147 +--
drivers/infiniband/core/cm_msgs.h | 7 -
drivers/infiniband/core/cma.c | 362 ++++---
drivers/infiniband/core/core_priv.h | 4 +-
drivers/infiniband/core/device.c | 23 +-
drivers/infiniband/core/mad.c | 113 +-
drivers/infiniband/core/mad_priv.h | 7 +-
drivers/infiniband/core/multicast.c | 40 +-
drivers/infiniband/core/nldev.c | 16 +-
drivers/infiniband/core/rdma_core.c | 1018 +++++++++++-------
drivers/infiniband/core/rdma_core.h | 96 +-
drivers/infiniband/core/roce_gid_mgmt.c | 306 ++++--
drivers/infiniband/core/rw.c | 8 +-
drivers/infiniband/core/sa_query.c | 138 ++-
drivers/infiniband/core/sysfs.c | 66 +-
drivers/infiniband/core/ucm.c | 15 +-
drivers/infiniband/core/umem.c | 62 +-
drivers/infiniband/core/user_mad.c | 1 +
drivers/infiniband/core/uverbs.h | 34 +-
drivers/infiniband/core/uverbs_cmd.c | 599 +++++------
drivers/infiniband/core/uverbs_ioctl.c | 709 ++++++++-----
drivers/infiniband/core/uverbs_ioctl_merge.c | 664 ------------
drivers/infiniband/core/uverbs_main.c | 232 ++--
drivers/infiniband/core/uverbs_marshall.c | 2 -
drivers/infiniband/core/uverbs_std_types.c | 200 ++--
.../infiniband/core/uverbs_std_types_counters.c | 108 +-
drivers/infiniband/core/uverbs_std_types_cq.c | 154 ++-
drivers/infiniband/core/uverbs_std_types_dm.c | 61 +-
.../infiniband/core/uverbs_std_types_flow_action.c | 170 +--
drivers/infiniband/core/uverbs_std_types_mr.c | 88 +-
drivers/infiniband/core/uverbs_uapi.c | 346 ++++++
drivers/infiniband/core/verbs.c | 523 ++++++---
drivers/infiniband/hw/bnxt_re/ib_verbs.c | 144 +--
drivers/infiniband/hw/bnxt_re/ib_verbs.h | 15 +-
drivers/infiniband/hw/bnxt_re/qplib_fp.c | 12 +-
drivers/infiniband/hw/bnxt_re/qplib_sp.c | 4 +-
drivers/infiniband/hw/cxgb3/iwch_cq.c | 64 +-
drivers/infiniband/hw/cxgb3/iwch_provider.c | 44 +-
drivers/infiniband/hw/cxgb3/iwch_provider.h | 8 +-
drivers/infiniband/hw/cxgb3/iwch_qp.c | 32 +-
drivers/infiniband/hw/cxgb4/cm.c | 90 +-
drivers/infiniband/hw/cxgb4/cq.c | 269 +++--
drivers/infiniband/hw/cxgb4/device.c | 20 +-
drivers/infiniband/hw/cxgb4/ev.c | 5 +-
drivers/infiniband/hw/cxgb4/iw_cxgb4.h | 57 +-
drivers/infiniband/hw/cxgb4/provider.c | 56 +-
drivers/infiniband/hw/cxgb4/qp.c | 1051 +++++++++++++++---
drivers/infiniband/hw/cxgb4/resource.c | 51 +-
drivers/infiniband/hw/cxgb4/t4.h | 164 ++-
drivers/infiniband/hw/cxgb4/t4fw_ri_api.h | 68 +-
drivers/infiniband/hw/hfi1/chip.c | 205 ++--
drivers/infiniband/hw/hfi1/chip.h | 30 +
drivers/infiniband/hw/hfi1/driver.c | 63 +-
drivers/infiniband/hw/hfi1/file_ops.c | 10 +-
drivers/infiniband/hw/hfi1/hfi.h | 243 ++---
drivers/infiniband/hw/hfi1/init.c | 44 +-
drivers/infiniband/hw/hfi1/pcie.c | 19 +-
drivers/infiniband/hw/hfi1/pio.c | 14 +-
drivers/infiniband/hw/hfi1/qp.c | 6 +-
drivers/infiniband/hw/hfi1/qp.h | 24 +-
drivers/infiniband/hw/hfi1/rc.c | 6 +-
drivers/infiniband/hw/hfi1/ruc.c | 14 +-
drivers/infiniband/hw/hfi1/sdma.c | 10 +-
drivers/infiniband/hw/hfi1/verbs.c | 18 +-
drivers/infiniband/hw/hfi1/vnic_main.c | 12 +-
drivers/infiniband/hw/hns/hns_roce_ah.c | 21 +-
drivers/infiniband/hw/hns/hns_roce_common.h | 9 -
drivers/infiniband/hw/hns/hns_roce_db.c | 2 +
drivers/infiniband/hw/hns/hns_roce_device.h | 45 +-
drivers/infiniband/hw/hns/hns_roce_hem.c | 7 +-
drivers/infiniband/hw/hns/hns_roce_hw_v1.c | 430 +++++---
drivers/infiniband/hw/hns/hns_roce_hw_v1.h | 2 +-
drivers/infiniband/hw/hns/hns_roce_hw_v2.c | 698 ++++++++++--
drivers/infiniband/hw/hns/hns_roce_hw_v2.h | 136 ++-
drivers/infiniband/hw/hns/hns_roce_main.c | 15 +-
drivers/infiniband/hw/hns/hns_roce_pd.c | 2 +-
drivers/infiniband/hw/hns/hns_roce_qp.c | 55 +-
drivers/infiniband/hw/i40iw/Kconfig | 1 +
drivers/infiniband/hw/i40iw/i40iw_cm.c | 26 +-
drivers/infiniband/hw/i40iw/i40iw_hw.c | 83 +-
drivers/infiniband/hw/i40iw/i40iw_verbs.c | 54 +-
drivers/infiniband/hw/mlx4/ah.c | 70 +-
drivers/infiniband/hw/mlx4/mad.c | 29 +-
drivers/infiniband/hw/mlx4/main.c | 41 +-
drivers/infiniband/hw/mlx4/mlx4_ib.h | 21 +-
drivers/infiniband/hw/mlx4/qp.c | 421 ++++----
drivers/infiniband/hw/mlx4/srq.c | 4 +-
drivers/infiniband/hw/mlx5/Makefile | 2 +
drivers/infiniband/hw/mlx5/ah.c | 11 +-
drivers/infiniband/hw/mlx5/cmd.c | 12 +
drivers/infiniband/hw/mlx5/cmd.h | 1 +
drivers/infiniband/hw/mlx5/cong.c | 9 +-
drivers/infiniband/hw/mlx5/cq.c | 2 +-
drivers/infiniband/hw/mlx5/devx.c | 1119 ++++++++++++++++++++
drivers/infiniband/hw/mlx5/flow.c | 252 +++++
drivers/infiniband/hw/mlx5/gsi.c | 8 +-
drivers/infiniband/hw/mlx5/main.c | 570 +++++++---
drivers/infiniband/hw/mlx5/mlx5_ib.h | 85 +-
drivers/infiniband/hw/mlx5/mr.c | 34 +-
drivers/infiniband/hw/mlx5/qp.c | 294 +++--
drivers/infiniband/hw/mlx5/srq.c | 4 +-
drivers/infiniband/hw/mthca/mthca_av.c | 5 +-
drivers/infiniband/hw/mthca/mthca_dev.h | 24 +-
drivers/infiniband/hw/mthca/mthca_provider.c | 7 +-
drivers/infiniband/hw/mthca/mthca_qp.c | 24 +-
drivers/infiniband/hw/mthca/mthca_srq.c | 8 +-
drivers/infiniband/hw/nes/nes.h | 2 +-
drivers/infiniband/hw/nes/nes_cm.c | 8 +-
drivers/infiniband/hw/nes/nes_hw.c | 6 +-
drivers/infiniband/hw/nes/nes_verbs.c | 74 +-
drivers/infiniband/hw/ocrdma/ocrdma_ah.c | 32 +-
drivers/infiniband/hw/ocrdma/ocrdma_ah.h | 1 -
drivers/infiniband/hw/ocrdma/ocrdma_hw.c | 26 +-
drivers/infiniband/hw/ocrdma/ocrdma_main.c | 1 -
drivers/infiniband/hw/ocrdma/ocrdma_verbs.c | 50 +-
drivers/infiniband/hw/ocrdma/ocrdma_verbs.h | 12 +-
drivers/infiniband/hw/qedr/main.c | 107 +-
drivers/infiniband/hw/qedr/qedr.h | 43 +-
drivers/infiniband/hw/qedr/qedr_hsi_rdma.h | 11 +
drivers/infiniband/hw/qedr/qedr_iw_cm.c | 12 +-
drivers/infiniband/hw/qedr/qedr_roce_cm.c | 37 +-
drivers/infiniband/hw/qedr/qedr_roce_cm.h | 8 +-
drivers/infiniband/hw/qedr/verbs.c | 625 +++++++++--
drivers/infiniband/hw/qedr/verbs.h | 17 +-
drivers/infiniband/hw/qib/qib_verbs.c | 3 +-
drivers/infiniband/hw/qib/qib_verbs.h | 5 +-
drivers/infiniband/hw/usnic/Kconfig | 2 +-
drivers/infiniband/hw/usnic/usnic_fwd.c | 4 +-
drivers/infiniband/hw/usnic/usnic_fwd.h | 2 +-
drivers/infiniband/hw/usnic/usnic_ib_verbs.c | 10 +-
drivers/infiniband/hw/usnic/usnic_ib_verbs.h | 8 +-
drivers/infiniband/hw/usnic/usnic_uiom.c | 40 +-
drivers/infiniband/hw/usnic/usnic_uiom.h | 5 +-
drivers/infiniband/hw/vmw_pvrdma/pvrdma.h | 5 -
drivers/infiniband/hw/vmw_pvrdma/pvrdma_cq.c | 26 -
drivers/infiniband/hw/vmw_pvrdma/pvrdma_main.c | 52 +-
drivers/infiniband/hw/vmw_pvrdma/pvrdma_qp.c | 11 +-
drivers/infiniband/hw/vmw_pvrdma/pvrdma_srq.c | 7 -
drivers/infiniband/hw/vmw_pvrdma/pvrdma_verbs.c | 6 +-
drivers/infiniband/hw/vmw_pvrdma/pvrdma_verbs.h | 15 +-
drivers/infiniband/sw/rdmavt/ah.c | 4 +-
drivers/infiniband/sw/rdmavt/qp.c | 27 +-
drivers/infiniband/sw/rdmavt/qp.h | 12 +-
drivers/infiniband/sw/rdmavt/srq.c | 2 +-
drivers/infiniband/sw/rxe/rxe.c | 3 +-
drivers/infiniband/sw/rxe/rxe_av.c | 30 +-
drivers/infiniband/sw/rxe/rxe_comp.c | 1 +
drivers/infiniband/sw/rxe/rxe_loc.h | 5 +-
drivers/infiniband/sw/rxe/rxe_net.c | 67 +-
drivers/infiniband/sw/rxe/rxe_param.h | 2 +-
drivers/infiniband/sw/rxe/rxe_qp.c | 31 +-
drivers/infiniband/sw/rxe/rxe_recv.c | 24 +-
drivers/infiniband/sw/rxe/rxe_resp.c | 5 +
drivers/infiniband/sw/rxe/rxe_verbs.c | 70 +-
drivers/infiniband/ulp/ipoib/ipoib.h | 32 +-
drivers/infiniband/ulp/ipoib/ipoib_cm.c | 81 +-
drivers/infiniband/ulp/ipoib/ipoib_ethtool.c | 2 +-
drivers/infiniband/ulp/ipoib/ipoib_fs.c | 6 +-
drivers/infiniband/ulp/ipoib/ipoib_ib.c | 15 +-
drivers/infiniband/ulp/ipoib/ipoib_main.c | 444 ++++----
drivers/infiniband/ulp/ipoib/ipoib_multicast.c | 7 +-
drivers/infiniband/ulp/ipoib/ipoib_netlink.c | 23 -
drivers/infiniband/ulp/ipoib/ipoib_verbs.c | 8 +-
drivers/infiniband/ulp/ipoib/ipoib_vlan.c | 261 +++--
drivers/infiniband/ulp/iser/iscsi_iser.c | 16 +-
drivers/infiniband/ulp/iser/iser_memory.c | 5 +-
drivers/infiniband/ulp/iser/iser_verbs.c | 14 +-
drivers/infiniband/ulp/isert/ib_isert.c | 26 +-
drivers/infiniband/ulp/srp/ib_srp.c | 27 +-
drivers/infiniband/ulp/srpt/ib_srpt.c | 71 +-
drivers/infiniband/ulp/srpt/ib_srpt.h | 4 +-
drivers/net/ethernet/chelsio/cxgb4/t4_msg.h | 4 +-
drivers/net/ethernet/mellanox/mlx5/core/cmd.c | 5 +
.../mellanox/mlx5/core/diag/fs_tracepoint.c | 3 +
drivers/net/ethernet/mellanox/mlx5/core/fs_cmd.c | 24 +-
drivers/net/ethernet/mellanox/mlx5/core/fs_core.c | 81 +-
.../net/ethernet/mellanox/mlx5/core/ipoib/ipoib.c | 37 +-
drivers/nvme/host/rdma.c | 13 +-
drivers/nvme/target/rdma.c | 13 +-
fs/cifs/smbdirect.c | 32 +-
include/linux/idr.h | 11 +
include/linux/mlx5/driver.h | 3 -
include/linux/mlx5/fs.h | 1 +
include/linux/mlx5/mlx5_ifc.h | 5 +
include/linux/overflow.h | 31 +
include/rdma/ib.h | 4 +-
include/rdma/ib_addr.h | 6 +-
include/rdma/ib_cache.h | 83 +-
include/rdma/ib_cm.h | 18 +-
include/rdma/ib_mad.h | 33 +
include/rdma/ib_sa.h | 49 +-
include/rdma/ib_verbs.h | 257 +++--
include/rdma/opa_addr.h | 2 +-
include/rdma/rdma_cm.h | 2 +-
include/rdma/rdmavt_qp.h | 30 +-
include/rdma/uverbs_ioctl.h | 627 ++++++-----
include/rdma/uverbs_named_ioctl.h | 109 +-
include/rdma/uverbs_std_types.h | 96 +-
include/rdma/uverbs_types.h | 133 +--
include/uapi/rdma/cxgb4-abi.h | 32 +-
include/uapi/rdma/hns-abi.h | 1 +
include/uapi/rdma/ib_user_ioctl_cmds.h | 7 +-
include/uapi/rdma/ib_user_ioctl_verbs.h | 58 +
include/uapi/rdma/ib_user_verbs.h | 5 +-
include/uapi/rdma/mlx5-abi.h | 6 +-
include/uapi/rdma/mlx5_user_ioctl_cmds.h | 121 +++
include/uapi/rdma/qedr-abi.h | 17 +
include/uapi/rdma/rdma_user_ioctl_cmds.h | 7 +-
lib/test_overflow.c | 198 +++-
net/9p/trans_rdma.c | 8 +-
net/core/secure_seq.c | 1 +
net/rds/ib.c | 2 +-
net/rds/ib_frmr.c | 11 +-
net/rds/ib_recv.c | 6 +-
net/rds/ib_send.c | 6 +-
net/smc/smc_core.c | 1 +
net/smc/smc_ib.c | 1 +
net/smc/smc_tx.c | 3 +-
net/smc/smc_wr.c | 9 +-
net/smc/smc_wr.h | 3 +-
net/sunrpc/xprtrdma/fmr_ops.c | 4 +-
net/sunrpc/xprtrdma/frwr_ops.c | 7 +-
net/sunrpc/xprtrdma/svc_rdma_recvfrom.c | 3 +-
net/sunrpc/xprtrdma/svc_rdma_rw.c | 3 +-
net/sunrpc/xprtrdma/svc_rdma_sendto.c | 3 +-
net/sunrpc/xprtrdma/svc_rdma_transport.c | 2 +-
net/sunrpc/xprtrdma/verbs.c | 5 +-
233 files changed, 12380 insertions(+), 7059 deletions(-)
delete mode 100644 drivers/infiniband/core/uverbs_ioctl_merge.c
create mode 100644 drivers/infiniband/core/uverbs_uapi.c
create mode 100644 drivers/infiniband/hw/mlx5/devx.c
create mode 100644 drivers/infiniband/hw/mlx5/flow.c
Merge resolution diff:
diff --cc drivers/infiniband/core/rdma_core.c
index 12e7c6c102c136,475910ffbcb680..6eb64c6f08028b
--- a/drivers/infiniband/core/rdma_core.c
+++ b/drivers/infiniband/core/rdma_core.c
@@@ -77,201 -120,19 +77,201 @@@ static int uverbs_try_lock_object(struc
* concurrently, setting the counter to zero is enough for releasing
* this lock.
*/
- if (!exclusive)
+ switch (mode) {
+ case UVERBS_LOOKUP_READ:
- return __atomic_add_unless(&uobj->usecnt, 1, -1) == -1 ?
+ return atomic_fetch_add_unless(&uobj->usecnt, 1, -1) == -1 ?
-EBUSY : 0;
+ case UVERBS_LOOKUP_WRITE:
+ /* lock is exclusive */
+ return atomic_cmpxchg(&uobj->usecnt, 0, -1) == 0 ? 0 : -EBUSY;
+ case UVERBS_LOOKUP_DESTROY:
+ return 0;
+ }
+ return 0;
+}
+
+static void assert_uverbs_usecnt(struct ib_uobject *uobj,
+ enum rdma_lookup_mode mode)
+{
+#ifdef CONFIG_LOCKDEP
+ switch (mode) {
+ case UVERBS_LOOKUP_READ:
+ WARN_ON(atomic_read(&uobj->usecnt) <= 0);
+ break;
+ case UVERBS_LOOKUP_WRITE:
+ WARN_ON(atomic_read(&uobj->usecnt) != -1);
+ break;
+ case UVERBS_LOOKUP_DESTROY:
+ break;
+ }
+#endif
+}
+
+/*
+ * This must be called with the hw_destroy_rwsem locked for read or write,
+ * also the uobject itself must be locked for write.
+ *
+ * Upon return the HW object is guaranteed to be destroyed.
+ *
+ * For RDMA_REMOVE_ABORT, the hw_destroy_rwsem is not required to be held,
+ * however the type's allocat_commit function cannot have been called and the
+ * uobject cannot be on the uobjects_lists
+ *
+ * For RDMA_REMOVE_DESTROY the caller shold be holding a kref (eg via
+ * rdma_lookup_get_uobject) and the object is left in a state where the caller
+ * needs to call rdma_lookup_put_uobject.
+ *
+ * For all other destroy modes this function internally unlocks the uobject
+ * and consumes the kref on the uobj.
+ */
+static int uverbs_destroy_uobject(struct ib_uobject *uobj,
+ enum rdma_remove_reason reason)
+{
+ struct ib_uverbs_file *ufile = uobj->ufile;
+ unsigned long flags;
+ int ret;
+
+ lockdep_assert_held(&ufile->hw_destroy_rwsem);
+ assert_uverbs_usecnt(uobj, UVERBS_LOOKUP_WRITE);
+
+ if (uobj->object) {
+ ret = uobj->uapi_object->type_class->destroy_hw(uobj, reason);
+ if (ret) {
+ if (ib_is_destroy_retryable(ret, reason, uobj))
+ return ret;
+
+ /* Nothing to be done, dangle the memory and move on */
+ WARN(true,
+ "ib_uverbs: failed to remove uobject id %d, driver err=%d",
+ uobj->id, ret);
+ }
+
+ uobj->object = NULL;
+ }
- /* lock is either WRITE or DESTROY - should be exclusive */
- return atomic_cmpxchg(&uobj->usecnt, 0, -1) == 0 ? 0 : -EBUSY;
+ if (reason == RDMA_REMOVE_ABORT) {
+ WARN_ON(!list_empty(&uobj->list));
+ WARN_ON(!uobj->context);
+ uobj->uapi_object->type_class->alloc_abort(uobj);
+ }
+
+ uobj->context = NULL;
+
+ /*
+ * For DESTROY the usecnt is held write locked, the caller is expected
+ * to put it unlock and put the object when done with it. Only DESTROY
+ * can remove the IDR handle.
+ */
+ if (reason != RDMA_REMOVE_DESTROY)
+ atomic_set(&uobj->usecnt, 0);
+ else
+ uobj->uapi_object->type_class->remove_handle(uobj);
+
+ if (!list_empty(&uobj->list)) {
+ spin_lock_irqsave(&ufile->uobjects_lock, flags);
+ list_del_init(&uobj->list);
+ spin_unlock_irqrestore(&ufile->uobjects_lock, flags);
+
+ /*
+ * Pairs with the get in rdma_alloc_commit_uobject(), could
+ * destroy uobj.
+ */
+ uverbs_uobject_put(uobj);
+ }
+
+ /*
+ * When aborting the stack kref remains owned by the core code, and is
+ * not transferred into the type. Pairs with the get in alloc_uobj
+ */
+ if (reason == RDMA_REMOVE_ABORT)
+ uverbs_uobject_put(uobj);
+
+ return 0;
}
-static struct ib_uobject *alloc_uobj(struct ib_ucontext *context,
- const struct uverbs_obj_type *type)
+/*
+ * This calls uverbs_destroy_uobject() using the RDMA_REMOVE_DESTROY
+ * sequence. It should only be used from command callbacks. On success the
+ * caller must pair this with rdma_lookup_put_uobject(LOOKUP_WRITE). This
+ * version requires the caller to have already obtained an
+ * LOOKUP_DESTROY uobject kref.
+ */
+int uobj_destroy(struct ib_uobject *uobj)
{
- struct ib_uobject *uobj = kzalloc(type->obj_size, GFP_KERNEL);
+ struct ib_uverbs_file *ufile = uobj->ufile;
+ int ret;
+
+ down_read(&ufile->hw_destroy_rwsem);
+
+ ret = uverbs_try_lock_object(uobj, UVERBS_LOOKUP_WRITE);
+ if (ret)
+ goto out_unlock;
+
+ ret = uverbs_destroy_uobject(uobj, RDMA_REMOVE_DESTROY);
+ if (ret) {
+ atomic_set(&uobj->usecnt, 0);
+ goto out_unlock;
+ }
+out_unlock:
+ up_read(&ufile->hw_destroy_rwsem);
+ return ret;
+}
+
+/*
+ * uobj_get_destroy destroys the HW object and returns a handle to the uobj
+ * with a NULL object pointer. The caller must pair this with
+ * uverbs_put_destroy.
+ */
+struct ib_uobject *__uobj_get_destroy(const struct uverbs_api_object *obj,
+ u32 id, struct ib_uverbs_file *ufile)
+{
+ struct ib_uobject *uobj;
+ int ret;
+
+ uobj = rdma_lookup_get_uobject(obj, ufile, id, UVERBS_LOOKUP_DESTROY);
+ if (IS_ERR(uobj))
+ return uobj;
+
+ ret = uobj_destroy(uobj);
+ if (ret) {
+ rdma_lookup_put_uobject(uobj, UVERBS_LOOKUP_DESTROY);
+ return ERR_PTR(ret);
+ }
+
+ return uobj;
+}
+
+/*
+ * Does both uobj_get_destroy() and uobj_put_destroy(). Returns success_res
+ * on success (negative errno on failure). For use by callers that do not need
+ * the uobj.
+ */
+int __uobj_perform_destroy(const struct uverbs_api_object *obj, u32 id,
+ struct ib_uverbs_file *ufile, int success_res)
+{
+ struct ib_uobject *uobj;
+
+ uobj = __uobj_get_destroy(obj, id, ufile);
+ if (IS_ERR(uobj))
+ return PTR_ERR(uobj);
+
+ rdma_lookup_put_uobject(uobj, UVERBS_LOOKUP_WRITE);
+ return success_res;
+}
+
+/* alloc_uobj must be undone by uverbs_destroy_uobject() */
+static struct ib_uobject *alloc_uobj(struct ib_uverbs_file *ufile,
+ const struct uverbs_api_object *obj)
+{
+ struct ib_uobject *uobj;
+ struct ib_ucontext *ucontext;
+
+ ucontext = ib_uverbs_get_ucontext(ufile);
+ if (IS_ERR(ucontext))
+ return ERR_CAST(ucontext);
+
+ uobj = kzalloc(obj->type_attrs->obj_size, GFP_KERNEL);
if (!uobj)
return ERR_PTR(-ENOMEM);
/*
diff --cc drivers/nvme/host/rdma.c
index 72e8e8e7d2d7cd,0805fa6215eefb..dc042017c293ad
--- a/drivers/nvme/host/rdma.c
+++ b/drivers/nvme/host/rdma.c
@@@ -374,6 -377,8 +377,8 @@@ nvme_rdma_find_get_device(struct rdma_c
goto out_free_pd;
}
+ ndev->num_inline_segments = min(NVME_RDMA_MAX_INLINE_SEGMENTS,
- ndev->dev->attrs.max_sge - 1);
++ ndev->dev->attrs.max_send_sge - 1);
list_add(&ndev->entry, &device_list);
out_unlock:
mutex_unlock(&device_list_mutex);
diff --cc drivers/nvme/target/rdma.c
index 1a642e214a4ce2,e7f43d1e17797f..3533e918ea376b
--- a/drivers/nvme/target/rdma.c
+++ b/drivers/nvme/target/rdma.c
@@@ -382,13 -435,22 +435,21 @@@ static void nvmet_rdma_free_rsps(struc
static int nvmet_rdma_post_recv(struct nvmet_rdma_device *ndev,
struct nvmet_rdma_cmd *cmd)
{
- struct ib_recv_wr *bad_wr;
+ int ret;
+
ib_dma_sync_single_for_device(ndev->device,
cmd->sge[0].addr, cmd->sge[0].length,
DMA_FROM_DEVICE);
if (ndev->srq)
- return ib_post_srq_recv(ndev->srq, &cmd->wr, NULL);
- return ib_post_recv(cmd->queue->cm_id->qp, &cmd->wr, NULL);
- ret = ib_post_srq_recv(ndev->srq, &cmd->wr, &bad_wr);
++ ret = ib_post_srq_recv(ndev->srq, &cmd->wr, NULL);
+ else
- ret = ib_post_recv(cmd->queue->cm_id->qp, &cmd->wr, &bad_wr);
++ ret = ib_post_recv(cmd->queue->cm_id->qp, &cmd->wr, NULL);
+
+ if (unlikely(ret))
+ pr_err("post_recv cmd failed\n");
+
+ return ret;
}
static void nvmet_rdma_process_wr_wait_list(struct nvmet_rdma_queue *queue)
@@@ -491,7 -553,7 +552,7 @@@ static void nvmet_rdma_queue_response(s
rsp->send_sge.addr, rsp->send_sge.length,
DMA_TO_DEVICE);
- if (ib_post_send(cm_id->qp, first_wr, NULL)) {
- if (unlikely(ib_post_send(cm_id->qp, first_wr, &bad_wr))) {
++ if (unlikely(ib_post_send(cm_id->qp, first_wr, NULL))) {
pr_err("sending cmd response failed\n");
nvmet_rdma_release_rsp(rsp);
}
@@@ -805,6 -890,18 +889,18 @@@ nvmet_rdma_find_get_device(struct rdma_
if (!ndev)
goto out_err;
+ inline_page_count = num_pages(port->inline_data_size);
+ inline_sge_count = max(cm_id->device->attrs.max_sge_rd,
- cm_id->device->attrs.max_sge) - 1;
++ cm_id->device->attrs.max_recv_sge) - 1;
+ if (inline_page_count > inline_sge_count) {
+ pr_warn("inline_data_size %d cannot be supported by device %s. Reducing to %lu.\n",
+ port->inline_data_size, cm_id->device->name,
+ inline_sge_count * PAGE_SIZE);
+ port->inline_data_size = inline_sge_count * PAGE_SIZE;
+ inline_page_count = inline_sge_count;
+ }
+ ndev->inline_data_size = port->inline_data_size;
+ ndev->inline_page_count = inline_page_count;
ndev->device = cm_id->device;
kref_init(&ndev->ref);
diff --cc net/rds/ib_recv.c
index 4c5a937304b274,d300186b8dc020..2f16146e4ec94e
--- a/net/rds/ib_recv.c
+++ b/net/rds/ib_recv.c
@@@ -416,10 -415,10 +414,10 @@@ void rds_ib_recv_refill(struct rds_conn
&recv->r_frag->f_sg));
/* XXX when can this fail? */
- ret = ib_post_recv(ic->i_cm_id->qp, &recv->r_wr, &failed_wr);
+ ret = ib_post_recv(ic->i_cm_id->qp, &recv->r_wr, NULL);
if (ret) {
rds_ib_conn_error(conn, "recv post on "
- "%pI4 returned %d, disconnecting and "
+ "%pI6c returned %d, disconnecting and "
"reconnecting\n", &conn->c_faddr,
ret);
break;
diff --cc net/rds/ib_send.c
index 8ac80c1b051e1a,c8dd3125d39877..2dcb555e6350d1
--- a/net/rds/ib_send.c
+++ b/net/rds/ib_send.c
@@@ -758,8 -758,7 +758,7 @@@ int rds_ib_xmit_atomic(struct rds_conne
{
struct rds_ib_connection *ic = conn->c_transport_data;
struct rds_ib_send_work *send = NULL;
- struct ib_send_wr *failed_wr;
+ const struct ib_send_wr *failed_wr;
- struct rds_ib_device *rds_ibdev;
u32 pos;
u32 work_alloc;
int ret;
diff --cc net/smc/smc_wr.c
index de1a438cf9773f,f856b8402b3fe5..3c458d27985574
--- a/net/smc/smc_wr.c
+++ b/net/smc/smc_wr.c
@@@ -246,14 -239,11 +238,10 @@@ int smc_wr_tx_send(struct smc_link *lin
ib_req_notify_cq(link->smcibdev->roce_cq_send,
IB_CQ_NEXT_COMP | IB_CQ_REPORT_MISSED_EVENTS);
pend = container_of(priv, struct smc_wr_tx_pend, priv);
- rc = ib_post_send(link->roce_qp, &link->wr_tx_ibs[pend->idx],
- &failed_wr);
+ rc = ib_post_send(link->roce_qp, &link->wr_tx_ibs[pend->idx], NULL);
if (rc) {
- struct smc_link_group *lgr =
- container_of(link, struct smc_link_group,
- lnk[SMC_SINGLE_LINK]);
-
smc_wr_tx_put_slot(link, priv);
- smc_lgr_terminate(lgr);
+ smc_lgr_terminate(smc_get_lgr(link));
}
return rc;
}
Attachment:
signature.asc
Description: PGP signature
