Skip to content

Releases: linux-rdma/rdma-core

rdma-core-25.3:

25 Nov 15:10
v25.3
Compare
Choose a tag to compare

Changes:

  • ba0fbcf Bump to version 25.3
  • d83578b build/cbuild: Update cbuild to work with python3
  • bdae1b9 Bump to version 25.2
  • be3fe89 buildlib: Build devel stable branches on Azure
  • 3ac84b9 man: Fix return value for ibv_reg_dm_mr
  • 7e8caf1 build: Update ABI files
  • fc1ab96 build: Run CI builds on the stable branches with azp support
  • 2aa095c buildlib: Remove travis CI
  • 742f07e build/azp: Have Azure Pipelines create releases when tags are made
  • ca18bd6 cxgb4: always query device before initializing chip version
See More
  • a38d9d4 cxgb4: free appropriate pointer in error case
  • a44fc5c mlx5: Allow insertion of duplicate rules using DR API
  • 1b765b7 verbs: Set missing errno in ibv_cmd_reg_mr
  • d919a75 pyverbs: Fix CQ and PD assignment in QPAttr
  • 2217846 pyverbs: Fix WC creation process
  • 7041ce3 man: Fix wrong field in ibv_wr_post's man page
  • 4827f10 build: Do not enable -Wredundant-decls twice
  • cd6b1d7 srp_daemon: fix a double free segment fault for ibsrpdm
  • 93fd8af pyverbs: Fix PD assignment in QPInitAttrEx
  • 21e1b1d mlx5: Fix incorrect size of QPN variable as part of direct rules
  • 2d8e50c mlx5: Set miss address on copied STE during rehash
  • 18e0ef1 mlx5: Fix incorrect postsend of new rehashed/formatted table

This list of changes was auto generated.

rdma-core-24.2:

25 Nov 14:54
v24.2
Compare
Choose a tag to compare

Updates from version 24.1

  • Backport fixes:
    • man: Fix return value for ibv_reg_dm_mr
    • cxgb4: free appropriate pointer in error case
    • mlx5: Allow insertion of duplicate rules using DR API
    • verbs: Set missing errno in ibv_cmd_reg_mr
    • pyverbs: Fix WC creation process
    • man: Fix wrong field in ibv_wr_post's man page
    • build: Do not enable -Wredundant-decls twice
    • srp_daemon: fix a double free segment fault for ibsrpdm
    • pyverbs: Fix PD assignment in QPInitAttrEx
    • mlx5: Fix incorrect size of QPN variable as part of direct rules
    • mlx5: Set miss address on copied STE during rehash
    • mlx5: Fix incorrect postsend of new rehashed/formatted table

rdma-core-26.0:

02 Oct 11:00
4159490
Compare
Choose a tag to compare

Updates from version 25.0

  • mlx4: Update errno where necessary
  • pyverbs: Fix PD assignment in QPInitAttrEx
  • verbs: Fix UD pingpong default message size to match default MTU
  • mlx5: Add support for ibv_open_qp
  • ibdiag: Remove wrongly added ibtypes.py file
  • redhat: BuildRequires python3
  • srp_daemon: fix a double free segment fault for ibsrpdm
  • cbuild: Run gpg with a home directy in the tmpdir
  • mlx5: Don't assume that input is rounded to power two
  • mlx5: Fix incorrect size of QPN variable as part of direct rules
  • mlx5: Set miss address on copied STE during rehash
  • mlx5: Fix incorrect postsend of new rehashed/formatted table
  • Fix static library regex check
  • suse: provide libibmad-devel
  • suse: make all infiniband-diags config file noreplace
  • suse: add perl dependency for infiniband-diags
  • suse: fix dependency to rst2man
  • suse: Fix name for libefa RPM
  • mlx5: Fix shift of bit in conversion between mlx5_ilog2 and ilog32
  • ccam: Properly enable ilog32() calculations
  • Revert "Revert "mlx5: Use ilog32 instead of mlx5_ilog2""
  • Revert "mlx5: Use ilog32 instead of mlx5_ilog2"
  • libqedr: Add support for send with invalidate
  • ibdiags: Support arbitrary number of IB devices in ibstat
  • libibumad: Redesign resolve_ca_name to support arbitrary number of IB devices
  • libibumad: Support arbitrary number of IB devices
  • efa: Support send using extended QP API
  • efa: Add create extended QP direct verb
  • mlx5: Allow creating a RDMA_RX flow table
  • Update kernel headers
  • efa: Introduce create extended QP support
  • efa: Store verbs_qp in EFA's QP
  • efa: Split send validation function
  • efa: Split efa_post_send to auxiliary functions
  • efa: Split post send SGL API to auxiliary functions
  • efa: Set errno upon verbs failure
  • mlx5: Report ODP capabilities for DC transport
  • Document how to setup rxe/siw.
  • debian: Depend on python3-docutils instead of python-docutils
  • mlx5: Use ilog32 instead of mlx5_ilog2
  • libhns: Modify pi vlaue when cq overflows
  • libhns: Remove unused headerfiles
  • libhns: Adjust resource release order
  • libhns: Refactor for creating qp
  • libhns: Change type of wqe_shift to unsigned
  • libhns: Remove unnecessary memset calls
  • siw: Change user mmapped CQ notifications flags to 32bit.
  • verbs: Add unspecified node/transport types
  • efa: Fill send operation type in TX descriptor
  • efa: Protect WQ access with a lock on poll CQ flow
  • efa: Remove various unused fields from structs
  • cxgb4: remove unused c4iw_match_device
  • cxgb4: fix chipversion initialization
  • efa: Remove redundant zero of port_attr
  • efa: Use {} for zero initialization of structs
  • efa: Add missing initialization of command structs
  • efa: Use proper error labels in alloc context flow
  • efa: Handle memory leaks in free context flow
  • efa: Don't use opportunistic TX inline
  • efa: Fix direct verbs wrong version number in debian symbols
  • efa: Remove reimplementation of field_avail macro
  • srp_daemon: check that port LID is valid before calling create_ah
  • build/azp: Fix centos6 spec file to version 26
  • Update library version to be 26.0
  • kernel-boot: Set default prefix for RDMA devices with unknown protocol
  • kernel-boot: Separate PCI fill function
  • efa: Add query AH direct verb
  • efa: Add query device direct verb
  • build/azp: Have Azure Pipelines create releases when tags are made
  • build: Use the CMake variable -DENABLE_WERROR to turn on WERROR mode
  • build/azp: Run lintian over the bionic .debs
  • build/azp: Add centos6 to the test distributions
  • build/azp: Reduce the package list
  • build/cbuild: Update cbuild to work with python3
  • build/azp: Update check-build to work with python3
  • build/azp: Add Fedora 30 to the distro testing
  • build/azp: Run a test compile on ppc64el as well
  • build/azp: Use clang 8.0 for building
  • build/azp: Use gcc 9.3 for building
  • build/cbuild: Add push-azp-images
  • build/azp: Use a version number for the docker images
  • build/travis: Do not cross compile for ARM64
  • build/travis: Do not run checkpatch
  • build/travis: Do not build packages in travis anymore
  • build/cbuild: Remove docker-gc
  • util: Enable uninitialized_var on powerpc
  • rdmacm: Fix missing libraries on centos6 build
  • kernel-boot: Instrument rename utility to allow sane bug reports

rdma-core-25.1:

26 Sep 13:20
v25.1
Compare
Choose a tag to compare

Updates from version 25.0

  • Backport fixes:
    • suse: provide libibmad-devel
    • suse: add perl dependency for infiniband-diags
    • suse: fix dependency to rst2man
    • suse: Fix name for libefa RPM
    • ccam: Properly enable ilog32() calculations
    • siw: Change user mmapped CQ notifications flags to 32bit.
    • travis: fix aarch64 extraction
    • cxgb4: fix chipversion initialization
    • kernel-boot: Set default prefix for RDMA devices with unknown protocol
    • rdmacm: Fix missing libraries on centos6 build
    • ABI Files

rdma-core-24.1:

26 Sep 13:20
v24.1
Compare
Choose a tag to compare

Updates from version 24.0

  • Backport fixes:
    • suse: Fix name for libefa RPM
    • ccam: Properly enable ilog32() calculations
    • travis: fix aarch64 extraction
    • kernel-boot: Set default prefix for RDMA devices with unknown protocol
    • kernel-boot: Fix garbage name due to wrong usage of netlink API
    • mlx5: Fix mlx5_ifc metadata fields spelling
    • mlx5: Fix bucket allocation check
    • mlx5: Allow matching of source QP regardless the source port
    • mlx5: Set the proper flags upon dr_fill_data_segs
    • RDMA/hns: Bugfix for identify the last srq sge
    • kernel-boot: Reset buffer before refill
    • suse: fix dracut support
    • ABI Files

rdma-core-23.2:

26 Sep 13:20
v23.2
Compare
Choose a tag to compare

Updates from version 23.1

  • Backport fixes:
    • travis: fix aarch64 extraction
    • ccam: Properly enable ilog32() calculations
    • RDMA/hns: Bugfix for identify the last srq sge
    • suse: fix dracut support
    • mlx5: Fix man page of mlx5dv_create_flow_action_modify_header()
    • libhns: Bugfix for flush cqe in case multi-process

rdma-core-22.4:

26 Sep 13:19
v22.4
Compare
Choose a tag to compare

Updates from version 22.3

  • Backport fixes:
    • travis: fix aarch64 extraction
    • ccam: Properly enable ilog32() calculations

rdma-core-21.3:

26 Sep 13:19
v21.3
Compare
Choose a tag to compare

Updates from version 21.2

  • Backport fixes:
    • travis: fix aarch64 extraction
    • ccam: Properly enable ilog32() calculations
    • suse: fix dracut support
    • mlx5: Fix man page of mlx5dv_create_flow_action_modify_header()
    • libhns: Bugfix for flush cqe in case multi-process

rdma-core-20.4:

26 Sep 13:19
v20.4
Compare
Choose a tag to compare

Updates from version 20.3

  • Backport fixes:
    • cbuild: Fix packaging of SuSE leap
    • cbuild: extend CLI to pass --with[out] options to rpmbuild
    • cbuild: fix python path for leap
    • buildlib: update SUSE validation targets
    • travis: fix aarch64 extraction
    • ccam: Properly enable ilog32() calculations
    • suse: fix dracut support
    • libhns: Bugfix for flush cqe in case multi-process

rdma-core-19.4:

26 Sep 13:19
v19.4
Compare
Choose a tag to compare

Updates from version 19.3

  • Backport fixes:
    • cbuild: Fix packaging of SuSE leap
    • cbuild: extend CLI to pass --with[out] options to rpmbuild
    • cbuild: fix python path for leap
    • buildlib: update SUSE validation targets
    • travis: fix aarch64 extraction
    • ccam: Properly enable ilog32() calculations
    • suse: fix dracut support