linux-rdma/rdma-core v30.6

rdma-core-30.6

on GitHub

• 2dcde91 libhns: The content of the header file should be protected with #define
• f928a76 libhns: Remove unused macros
• 08dac01 ibverbs: Don't memcpy padding in ibv_qp_init_attr
• 0a6f4c3 verbs: enable query IBV_EVENT_WQ_FATAL event string
• acd7387 Bump to version 30.4
• 42c9215 iwpmd: Zero-initialize the remote addr info
• 912a608 ibdiags: Make escape character tack effect
• 8b12ba3 ibdiags: Don't suggest non-existing ibqueryerrors command line argument
• c25b3da verbs: Add a man page note for IBV_EVENT_WQ_FATAL
• d8bf82f tests: Make sure that end_poll() is always called
• df2462b mlx5: DR, Fix error flow in rehash process
• d1bd600 mlx5: DR, Fix DM allocation when the size is not aligned
• 08f58f6 tests: Fix QP extended MW bind test
• de5fd77 Bump to version 30.3
• 46c76bf libhns: Bugfix for calculation of extended sge
• cd42c84 iwpmd: Check returned value of parse_iwpm_msg
• 4139d5b bnxt_re/lib: Check AH handler validity before use
• b5e8609 mlx5: Fix mlx5_read_clock returned errno value
• ecf862a suse: Fix path to dracut dir
• d74dc85 infiniband-diags: Reorder library build order
• 0c3de67 libhns: Avoid accessing NULL pointer when locking/unlocking CQ
• f22fa92 buildlib: Continue build on old GCC versions without SSE
• d728f85 mlx4: Fix mlx4_read_clock returned errno value
• 77f5084 verbs: Fix attr_optional() when 'IOCTL_MODE=write' is used
• 17ad6c4 libhns: Fix wrong range of a mask
• 7118700 verbs: Fix create CQ comp_mask check
• 605ba52 rdma-ndd: fix udev racy issue for system with multiple InfiniBand HCAs
• e569514 libhns: Remove the unnecessary mask on QPN of CQE
• 74cf23b libhns: Remove assert to check whether a pointer is NULL
• 1d0db84 libhns: Correct definition of DB_BYTE_4_TAG_M
• 01332cb libhns: Avoid double release of a pointer
• 220718f util: Add GENMASK helper
• 1aa3ee8 mlx5: DR, Force QP drain on table creation
• 8eba400 kernel-boot: Fix VF lookup
• 5d046bc suse: trigger udev rules after installing rdma-ndd
• 9f0eeb2 ibacm: Fix possible port loop overflow
• bb05128 verbs: Fix possible port loop overflow
• a605f3e efa: Fix DV extension clear check
• 081fe4a tests: Manage exceptions in rdmacm processes
• 151c43e mlx5: Fix uuars to have the 'uar_mmap_offset' data
• a314a4d Bump to version 30.2
• 6bde54a srp_daemon: Fix systemd dependency
• 3524d03 bnxt_re: Fix reported error code from create_cq
• b6946a0 verbs: Replace SQ with RQ in max_recv_sge's documents
• 58c5d4b verbs: Update the type of some variables in documents
• 22eba60 cxgb4: Fix reported error code from create_cq
• 82626f1 libqedr: Fix reported error code from create_cq
• 5207a4a librdmacm: Don't overwrite errno returned from libibverbs
• b168928 rdma_server: Add '-s' option in rdma_server's manual
• 0a59343 mlx5: Consider single threaded mode for shared UAR
• c6d4a06 udaddy: Fix create_reply_ah error flow
• 9ca169d redhat: no need to recursively remove srp_daemon.sh
• a9aedfb mlx5: DR, Create NC UAR as default but fall-back to WC if failed
• d4d7ed5 mlx5: Fix wqe size parameter in wqe signature calculation
• f75cd37 Bump to version 30.1
• dd1fc97 mlx5: DR, Fix error flows while adding new STE
• 5a52da0 mlx5: DR, Support match value of size zero
• f04d8bc srp_daemon: Avoid extra permissions for the lock file
• 4b45dff mlx5: Add fork support for DEVX umem
• 47eeb2d verbs: Prevent dontfork on ODP MR
• 6fbbe39 srp_daemon: Fix the spelling of the name of two symbolic constants
• e80189e verbs: Set attributes to zero if query_device_ex() is not supported
• aa9f3c3 verbs: Make sure VM_DONTCOPY removed
• 99cf316 verbs: Try access device before adding it to device list
• 8d2302f tests: Allow zero vendor_part_id
• bf3b859 verbs: Use WQ creation flags properly
• 44841ea mlx5: Fix an issue on P9 with legacy UAR
• d1b3e35 pyverbs: Memset the memory after posix_memalign
• 8c203a8 verbs: Close async_fd only when it was previously created
• 947c7f3 buildlib: Use the right container for azure pipelines release mode
• 1d86c4e redhat: Fix the condition for pyverbs enablement on Fedora 32 and up
• 9d0c84f mlx5: Fix potential arithmetic overflow
• eb7b4ab ABI Files
• 0f4dd71 Merge pull request #774 from amzn/efa-summer-cleanups-pr
• 635cc82 efa: Check the correct field on SQ/RQ terminate
• 0a0c0b2 efa: Use variable to store SQ on init/terminate functions
• 67da43f efa: Use variable to store RQ on init/terminate functions
• 5fd2c1d efa: Rename descriptor index to producer counter
• d9117e3 efa: Add helper functions for ringing the doorbell
• d0ee510 efa: Move sub CQ index parameter to base class
• f33aaa0 efa: Move doorbell parameter to base class
• 7142fa6 Merge pull request #773 from amzn/pyverbs-fixes-pr
• 8e18345 tests: Verify QP type support on CQ events test
• 51a5bdd tests: Verify QP type support on relaxed ordering tests
• 0f046a8 tests: Verify QP type support on extended QP tests
• 4069762 tests: Skip RC and UC QP type tests on unsupported devices
• d1a11a9 tests: Update verification of unsupported direct memory devices
• a882a50 tests: Update verification of RoCE support
• baa1d6d Merge pull request #772 from Honggang-LI/verbs
• 88789b7 libibverbs: Fix ABI_placeholder1 and ABI_placeholder2 assignment
• 6f214ec Merge pull request #771 from Li-Weihang/cleanup_0601
• 1fe4b2e libhns: Optimize the offset information of mmap areas of device
• 6942d69 libhns: Remove redundant parameters of two SRQ related functions
• a95f62b libhns: Check number of extended sge when using extended atomic
• ebda2f3 Merge pull request #766 from yishaih/mlx5_dr
• 4d09cd4 Merge pull request #764 from yishaih/mlx5_misc
• c79e6c4 tests: Add LAG TX affinity test
• 17a4af2 pyverbs/mlx5: Add support to modify/query affinity
• 1ead972 mlx5: Enable modifying QP's lag tx_affinity port
• 7926212 mlx5: Enable querying QP's configured and active lag port
• abd51d1 mlx5: Add num_lag_ports support in mlx5dv_query_device()
• 7047745 Merge branch 'EdwardSro-pr-rdmacm'
• fdaf5d9 Merge branch 'master' into pr-rdmacm
• 76a5f9a Merge pull request #763 from EdwardSro/pr-bug-fixes
• 40792b6 Merge pull request #757 from EdwardSro/pr-prefetch
• 1674a82 Merge pull request #768 from Honggang-LI/master
• a51ad65 libibverbs/man/ibv_create_wq.3 replace verbs_exp.h with verbs.h
• d26cc66 Merge pull request #765 from amzn/create-qp-ioctl-fix-pr
• cd88ca8 efa: Fix create QP command over ioctl interface
• 7ebd3fc mlx5: Skip source port matching on DR FDB RX domain
• 075ccbb mlx5: Fix matching on vport gvmi
• 6047912 mlx5: Update matching order, TTL before ipv4 5-tuple
• a02ba04 Merge pull request #760 from yishaih/mlx5_dr
• fdfc34d Merge pull request #762 from yishaih/mlx5_misc
• 8c4721f Merge pull request #753 from yishaih/verbs_fd
• 4244c79 verbs: Move QP create and destroy commands to ioctl
• 5839bdb verbs: Move WQ create and destroy to ioctl
• 130dc94 verbs: Fix ibv_create_wq() to set wq_context
• 9a09dd8 verbs: Move SRQ create and destroy to ioctl
• 71c1e3f verbs: Fix ibv_get_srq_num() man page
• 62c4b32 mlx4: Delete comp_mask from verbs_srq
• 90a4d0c verbs: Extend CQ KABI to get an async FD
• 6d048da Update kernel headers
• 087c941 pyverbs: Add MWs weakrefs to QPEx
• e50a081 mlx5: Fix reg_mr bad flow
• e9aed71 mlx5: Add a note to mlx5dv_create_qp man page
• 5bcc7a2 tests: Add prefetch ODP MR testing
• e8c9b7f tests: Rename get_send_element method in utils module
• 8cbff59 pyverbs: Add support for ODP MR prefetch feature
• c725238 mlx5: Remove extra comma from TAG action dump print
• 571249f mlx5: Update mlx5dv_dr_action dest_ft dump tool info
• 9f7b242 mlx5: Add support for dumping mlx5dv_dr meter action
• a26d474 pyverbs: Remove add_ref from Mlx5Context
• 00ccb43 tests: Add RDMACM Multicast test
• 714eb94 pyverbs: Add RDMACM multicast
• c557b83 tests: Add RDMACM UDP test
• f81c073 pyverbs: Support UDP RDMACM control path
• 01a3673 tests: Fix rdmacm client/server sync mechanism
• 21afa61 pyverbs: Fix AH creation from WC
• bcaebb2 pyverbs: AddrInfo class extension
• 7ba7f54 Merge pull request #758 from Kamalheib/tso_caps
• a6abe9b libibverbs: Fix display tso_caps section
• 7af01c7 Merge pull request #748 from yishaih/mlx5_misc
• ac695dc mlx5: Add DR drop action support over root tables
• 5867bda mlx5: Add support for drop action in DV steering
• 005c591 mlx5: Add DR support for steering default miss
• 510e478 Update kernel headers
• 6d0ee2a Merge pull request #744 from yishaih/mlx5_dv
• 77c031b mlx5: Add support for steering default miss
• 7e2f5a9 Update kernel headers
• d45a831 Merge pull request #754 from EdwardSro/pr-rdmacm
• d9cccc2 Merge pull request #750 from rleon/rdma-cm-hotplug-full
• 8b976be Merge pull request #749 from rleon/rdma-cm-hotplug
• 4deaad0 librdmacm: Add and delete CMA devices dynamically
• e5d371c librdmacm: Globally store and sort IB device list
• 191c934 librdmacm: Reference count access to verbs context
• 0e3e5c5 librdmacm: Make CMA device list to be dynamic
• 28da4a8 librdmacm: Rely on IB device index if available
• 307f922 libibverbs: Get stable IB device index
• 4ca5caf libibverbs: Fix description of ibv_get_device_guid man page
• d92ca93 Update kernel headers
• 19184c8 tests: Prioritize device config with inet IP.
• 5e1cb32 tests: RDMACM tests infrastructure refactoring
• 8d05999 Merge pull request #755 from yishaih/mlx5_dr
• f89f113 mlx5: Memory pools sync performance improvement
• 56019aa mlx5: Support DR ability to enable device memory reclaim
• 378fda5 Merge pull request #752 from Li-Weihang/tiny_cleanups
• 55b46a9 Merge pull request #747 from Kamalheib/code_dup
• 0e8f8bc Merge pull request #751 from jgunthorpe/azp_fix
• d17653d tests: Delete code duplication from CMTestCase
• 37faa01 libhns: Let software PI/CI grow naturally
• c257f1d libhns: Move memory barrier to the same position
• be31cd3 libhns: Check for valid sge number before processing sq inline
• 0060c3b Merge pull request #746 from haimbo/function_sort_ca_device_list
• bb1e49f Merge pull request #739 from Li-Weihang/refactor
• 7a75c0a build/azp: Use focal as the base OS for azp
• ef392f8 build/azp: Update to clang-10
• 0136516 build/azp: Update to gcc-10
• e78b5df build/azp: Add Ubuntu bionic to the deb tests
• 7ba12af buildlib: Fix a warning from newer pythons
• 162212e pyverbs: Fix compiling with clang on python 3.8
• 92f6a0a build/sparse: Support glibc 31
• 069ee04 ibacm: Work around using a flex array in a union
• 523de14 ibacm: Use [] for flex arrays
• 224b03c ibacm: Ensure strings are NULL terminated
• 457b345 cbuild: Fix build breakage from APT
• 3f71fdc cbuild: Adjust to the new clang CDN
• b51b4ce ibdiags: Displays IB CA device list sorted in alphabetical order
• fda070a libibumad: Alphabetically sorting CA device list
• 8dd9269 libhns: Optimize SRQ buffer size calculating process
• b75ddaa libhns: Optimize WQE buffer parameters calculating process
• 06a15db Merge pull request #741 from EdwardSro/pr-dyn-uar
• 01f12e3 Merge pull request #737 from amzn/efa-pyverbs-pr
• e9afdfe tests: Add a test for EFA query device direct verb
• 7453ba5 pyverbs: Add a name property to the context class
• a1be0c0 pyverbs/efa: Support printing device caps in human readable format
• bc13248 pyverbs/efa: Initial support for EFA direct verbs
• 4580ca1 Merge pull request #743 from Honggang-LI/redhat
• e8e24c4 Merge pull request #736 from Kamalheib/cmd_line_arg
• 76bf492 redhat: unversioned documentation directory
• 17c2d4e Merge pull request #742 from bdrung/master
• 2956adb Merge pull request #734 from yishaih/mlx5_dr
• 57b2f8c tests: Change DeviceTest to inherit from PyverbsAPITestCase
• 0859dc5 tests: Start use dev command line argument
• bf65fb5 tests: Add support for parsing command arguments
• fafa025 ibacm: Fix move of man page from section 1 to section 8
• 37d4919 debian: Bump Standards-Version to 4.5.0
• 7d4c75d tests: Add Mlx5 UAR test
• a45a0a4 Documentation: Add code snippet for Mlx5UAR creation
• fd6f963 pyverbs/mlx5: Support mlx5 UAR allocation
• 75438a6 Update library version to be 30.0
• c7c6985 Merge pull request #740 from galpress/fix-testing-doc-pr
• 88595b7 Documentation: Fix formatting of shell commands
• 0571579 mlx5: Handle ICM memory via buddy allocation instead of bucket management
• 6a06235 mlx5: Add buddy mechanism utilities
• 96bd547 mlx5: Allocate accurate aligned DM memory size
• d6dc276 Merge pull request #735 from yishaih/mlx5_misc
• e089511 mlx5: Add a man page note regarding DV indirect MKEY
• 5510034 Merge pull request #732 from haimbo/set_the_right_subnet_prefix
• a3152b6 Merge pull request #730 from markhaywood/startswith
• f7317ec Merge pull request #728 from Li-Weihang/compatibility
• 613643c ibdiag: Set the right subnet prefix when using the -G option
• c40a0f9 Merge pull request #729 from yishaih/mlx5_dv
• 4c512ff mlx5: Add RDMA_TX flow table type
• b09151a Update kernel headers
• 116253b Merge pull request #731 from Kamalheib/fix_roce_v2_ignore
• a2bd0a6 tests: Fix check for unsupported RoCEv2 devices
• 3f7bad7 Merge pull request #727 from yishaih/mlx5_misc
• 7fdcd25 mlx5: Move to fully dynamic UAR mode including QPs
• 0c7212c mlx5: Support dyn UAR for CQ
• 7886e39 mlx5: Adapt UAR legacy dynamic to the new mode
• 84ac527 mlx5: Enhance mlx5dv_devx_alloc_uar() functionality
• cb40082 Update kernel headers
• 89a4012 Merge pull request #724 from yishaih/verbs_misc
• 873945b Merge pull request #713 from mkalderon/master
• ad5d934 ibacm: check provider file ends with .so extension
• 0b7d09e libhns: Forbid to create 0-depth qp
• f4d1f9e Merge pull request #726 from markhaywood/string_id
• 58ccb63 libibverbs: Fix query_device_ex dummy function not to return EOPNOTSUPP
• cd909d8 Merge pull request #723 from yishaih/mlx5_misc

This list of changes was auto generated.