)]}' { "log": [ { "commit": "de593a02932a5ee7c729bd92df90e1fe8892b584", "tree": "abbc584c98bda6ace7cbaaf8dc39c23f0e4165a7", "parents": [ "0dee6b4ade37a38a8912a9190311c1ecab210641", "72e065b7e7477ff2caceb62d745b06d25ea1291d" ], "author": { "name": "Jason Gunthorpe", "email": "jgg@nvidia.com", "time": "Fri Apr 10 10:43:26 2026 -0300" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Fri Apr 10 10:43:26 2026 -0300" }, "message": "Merge pull request #1718 from amzn/rename-comp-mask\n\nefa: Rename alloc_ucontext comp_mask to supported_caps" }, { "commit": "0dee6b4ade37a38a8912a9190311c1ecab210641", "tree": "100d332d07722f94daf8de1b832618afe6ac16f6", "parents": [ "2b75cb458497d6756a15c1259bafeead6fe154dc", "4c1aee0a78cb540e0a4d7f29c31783d64fe41488" ], "author": { "name": "Jason Gunthorpe", "email": "jgg@nvidia.com", "time": "Fri Apr 10 10:41:54 2026 -0300" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Fri Apr 10 10:41:54 2026 -0300" }, "message": "Merge pull request #1709 from TaranovK/kotaranov/mw\n\nproviders/mana: allocate/destroy MWs" }, { "commit": "2b75cb458497d6756a15c1259bafeead6fe154dc", "tree": "0ee5f2aecbf243cd8f878d44a347df9fd4cc76a7", "parents": [ "3084eadfeacf3c049a1036ec178d3ee38b56dc41", "f4176cd713928fdbdce24515b6219f351746000a" ], "author": { "name": "Jason Gunthorpe", "email": "jgg@nvidia.com", "time": "Fri Apr 10 10:40:57 2026 -0300" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Fri Apr 10 10:40:57 2026 -0300" }, "message": "Merge pull request #1708 from amzn/wide_wqe\n\nefa: Add support for 128-byte WQE" }, { "commit": "3084eadfeacf3c049a1036ec178d3ee38b56dc41", "tree": "adceb0dfb68cb150501efe15dde93a76bbe9cce7", "parents": [ "cc0e95cbeba62b17db62f20f32454b514bb3cb7b", "4a05c56368f3a3600f680b05127d3e4af2e731f7" ], "author": { "name": "Jason Gunthorpe", "email": "jgg@nvidia.com", "time": "Fri Apr 10 10:23:11 2026 -0300" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Fri Apr 10 10:23:11 2026 -0300" }, "message": "Merge pull request #1676 from abhijitG-xlnx/wqe_idx\n\nionic: Reserve bytes in CQE format for future use" }, { "commit": "72e065b7e7477ff2caceb62d745b06d25ea1291d", "tree": "0746d54cf2dd5ced1fe0033508283863b2f6b9cb", "parents": [ "e38372106e9c78f8477efc8489358db44b3b289e" ], "author": { "name": "Michael Margolin", "email": "mrgolin@amazon.com", "time": "Tue Mar 17 16:43:25 2026 +0200" }, "committer": { "name": "Michael Margolin", "email": "mrgolin@amazon.com", "time": "Tue Mar 17 14:45:42 2026 +0000" }, "message": "providers/efa: Rename alloc_ucontext comp_mask to supported_caps\n\nRename the field and related constants to align with kernel changes.\nThis avoids confusion with the standard comp_mask semantics and makes\nit easier to add a proper comp_mask later if needed.\n\nReviewed-by: Tom Sela \u003ctomsela@amazon.com\u003e\nReviewed-by: Yonatan Nachum \u003cynachum@amazon.com\u003e\nSigned-off-by: Michael Margolin \u003cmrgolin@amazon.com\u003e\n" }, { "commit": "e38372106e9c78f8477efc8489358db44b3b289e", "tree": "b801f2eb9ed3c09c8570e25d64ea4a839c5d0bb3", "parents": [ "cc0e95cbeba62b17db62f20f32454b514bb3cb7b" ], "author": { "name": "Michael Margolin", "email": "mrgolin@amazon.com", "time": "Tue Mar 17 16:43:16 2026 +0200" }, "committer": { "name": "Michael Margolin", "email": "mrgolin@amazon.com", "time": "Tue Mar 17 14:45:42 2026 +0000" }, "message": "Update kernel headers\n\nTo commit: 5122be2a19aa (\"RDMA/efa: Rename alloc_ucontext comp_mask to\nsupported_caps\").\n\nReviewed-by: Tom Sela \u003ctomsela@amazon.com\u003e\nReviewed-by: Yonatan Nachum \u003cynachum@amazon.com\u003e\nSigned-off-by: Michael Margolin \u003cmrgolin@amazon.com\u003e\n" }, { "commit": "f4176cd713928fdbdce24515b6219f351746000a", "tree": "ee1fd35da93ee74cfedb90b190289aa5aa0af73e", "parents": [ "d2df166b90a1ad59f618673015eb0d298f310c4c" ], "author": { "name": "Yonatan Nachum", "email": "ynachum@amazon.com", "time": "Mon Mar 02 18:15:31 2026 +0000" }, "committer": { "name": "Yonatan Nachum", "email": "ynachum@amazon.com", "time": "Tue Mar 17 13:05:43 2026 +0000" }, "message": "efa: Validate RQ depth relatively to max WR SGEs\n\nEFA RQ max depth reported is the max descriptors a single RQ supports.\nFor RQ EFA support multiple SGEs which means a single recv WR consumes\nmultiple descriptors. This means that if the ULP requests more than 1\ndescriptor per recv WR, it can\u0027t use the max RQ depth reported. Validate\nthe requested RQ depth relatively to the supported max RQ depth with the\nmax SGEs requested.\n\nSigned-off-by: Yonatan Nachum \u003cynachum@amazon.com\u003e\n" }, { "commit": "d2df166b90a1ad59f618673015eb0d298f310c4c", "tree": "9824131fae45cb45bacd8fb51679c60ed81d7dd2", "parents": [ "1282261d905a9924363477dc6efb50b11b8dfb2d" ], "author": { "name": "Yonatan Nachum", "email": "ynachum@amazon.com", "time": "Sun Feb 08 13:31:31 2026 +0000" }, "committer": { "name": "Yonatan Nachum", "email": "ynachum@amazon.com", "time": "Tue Mar 17 13:05:37 2026 +0000" }, "message": "efa: Add EFA DV to get max RQ depth based on init params\n\nEFA RQ max depth reported is the max descriptors a single RQ supports.\nFor RQ EFA support multiple SGEs which means a single recv WR consumes\nmultiple descriptors. This means that if the ULP requests more than 1\ndescriptor per recv WR, it can\u0027t use the max RQ depth reported. To help\nULPs, add a new EFA DV that given the max recv SGEs requested, reports\nback the max RQ depth.\n\nSigned-off-by: Yonatan Nachum \u003cynachum@amazon.com\u003e\n" }, { "commit": "1282261d905a9924363477dc6efb50b11b8dfb2d", "tree": "67998957305b87efd7d4f596b383da2cdbd69bb0", "parents": [ "00e9a1579aaa7bd8ed63201d69675f9f1a4c06aa" ], "author": { "name": "Yonatan Nachum", "email": "ynachum@amazon.com", "time": "Sun Feb 08 12:15:34 2026 +0000" }, "committer": { "name": "Yonatan Nachum", "email": "ynachum@amazon.com", "time": "Tue Mar 17 12:47:14 2026 +0000" }, "message": "efa: Add EFA DV to get max SQ depth based on init params\n\nWith 128-byte WQE support, the max SQ depth is half the size compared to\nwhen using 64-byte WQE. Add an EFA DV that given SQ init parameters,\ncalculates the max SQ depth so ULPs can use it to create QPs without\nfailing on SQ depth.\n\nSigned-off-by: Yonatan Nachum \u003cynachum@amazon.com\u003e\n" }, { "commit": "00e9a1579aaa7bd8ed63201d69675f9f1a4c06aa", "tree": "3ac6746923b40f1fceae15f3f3cdd4df7ebd1578", "parents": [ "88917d9a9cdc45808e3025b9941fb4eff48a273d" ], "author": { "name": "Yonatan Nachum", "email": "ynachum@amazon.com", "time": "Sun Feb 08 10:32:07 2026 +0000" }, "committer": { "name": "Yonatan Nachum", "email": "ynachum@amazon.com", "time": "Tue Mar 17 12:47:14 2026 +0000" }, "message": "efa: Add support to query new extended inline size in query DV\n\nReport the new extended max inline size in EFA query device DV. The new\ninline size deprecates the legacy one and should be used from now.\n\nSigned-off-by: Yonatan Nachum \u003cynachum@amazon.com\u003e\n" }, { "commit": "88917d9a9cdc45808e3025b9941fb4eff48a273d", "tree": "9bdbbf17d5d3c21162164b46952d60dfe6a07541", "parents": [ "2166b90e284df10dfccea840765b863f905a28c2" ], "author": { "name": "Yonatan Nachum", "email": "ynachum@amazon.com", "time": "Mon Dec 08 16:53:55 2025 +0000" }, "committer": { "name": "Yonatan Nachum", "email": "ynachum@amazon.com", "time": "Tue Mar 17 12:47:14 2026 +0000" }, "message": "efa: Add inline support for RDMA write\n\nAdd inline support for RDMA write with 128-byte WQE. To support write\nwith inline ULP must provide a new flag in EFA create QP DV since EFA\nsupport write with inline only with 128-byte WQE. When the flag is\nprovided 128-byte WQE is used for any requested inline size.\n\nAssign the inline buffer address in the SQ field to enable inline data\nsetters for 128-byte WQE and set the remote length the same as local\nlength.\n\nSigned-off-by: Yonatan Nachum \u003cynachum@amazon.com\u003e\n" }, { "commit": "2166b90e284df10dfccea840765b863f905a28c2", "tree": "bac54f989e6d017edcc82bf1d1b7a8a0a1358672", "parents": [ "2255d7c7e59c2b947dac04cea4590e74b0de7832" ], "author": { "name": "Yonatan Nachum", "email": "ynachum@amazon.com", "time": "Mon Dec 08 16:51:34 2025 +0000" }, "committer": { "name": "Yonatan Nachum", "email": "ynachum@amazon.com", "time": "Tue Mar 17 12:47:14 2026 +0000" }, "message": "efa: Add support for 128-byte WQE\n\nAdd the new 128-byte WQE structs and add the 128-byte WQE size to the SQ\nWQE calculation utility.\nFor our datapath the logic is different between old and new APIs:\n1. Old API - we use a buffer large enough for both WQEs and cast the\n buffer to the right pointers based on the WQE size.\n2. New API - To have generic setters in the QP function pointers, add a\n new TX WQE context struct that holds the building blocks of the WQE\n and set them to the right fields for the WQE being constructed. This\n way the setters in the flow can access the fields in a generic way\n without being aware of the WQE type ore its size.\n\nSigned-off-by: Yonatan Nachum \u003cynachum@amazon.com\u003e\n" }, { "commit": "2255d7c7e59c2b947dac04cea4590e74b0de7832", "tree": "ee00cae86a81408979be92a41caea5d7472be8ec", "parents": [ "0f5d6ec6f277f47641719253a677e26bc2a6f7f3" ], "author": { "name": "Yonatan Nachum", "email": "ynachum@amazon.com", "time": "Mon Dec 08 16:33:00 2025 +0000" }, "committer": { "name": "Yonatan Nachum", "email": "ynachum@amazon.com", "time": "Tue Mar 17 12:47:14 2026 +0000" }, "message": "efa: Store SQ WQE size for dynamic usage\n\nAs preparation for 128-byte WQE support, add a new SQ WQE size field to\nthe SQ struct to store the dynamically configured WQE size.\nChange static usage of sizeof of the 64-byte WQE size with the new\nfield.\n\nSigned-off-by: Yonatan Nachum \u003cynachum@amazon.com\u003e\n" }, { "commit": "0f5d6ec6f277f47641719253a677e26bc2a6f7f3", "tree": "57c72e067846c35ef80026215d10d0bec29ffe96", "parents": [ "d9ae1f9c023a08089f1b6e9449cae05f57f7dadd" ], "author": { "name": "Yonatan Nachum", "email": "ynachum@amazon.com", "time": "Mon Dec 08 16:27:15 2025 +0000" }, "committer": { "name": "Yonatan Nachum", "email": "ynachum@amazon.com", "time": "Tue Mar 17 12:47:14 2026 +0000" }, "message": "efa: Use IO structs in WR setters instead of TX WQE\n\nAs preparation for supporting 128-byte WQE, remove direct usage of\nstruct efa_io_tx_wqe wherever possible in WQE setters. This gives the\nability to reuse those structs when needed for different WQE formats.\n\nSigned-off-by: Yonatan Nachum \u003cynachum@amazon.com\u003e\n" }, { "commit": "d9ae1f9c023a08089f1b6e9449cae05f57f7dadd", "tree": "6922163bb052da4ac782f69be863ae26c17926f8", "parents": [ "7ae33f982582059fce117b5b618d4f1dffa9cba2" ], "author": { "name": "Yonatan Nachum", "email": "ynachum@amazon.com", "time": "Mon Dec 08 11:52:45 2025 +0000" }, "committer": { "name": "Yonatan Nachum", "email": "ynachum@amazon.com", "time": "Tue Mar 17 12:47:14 2026 +0000" }, "message": "efa: Add max inline validation on QP creation\n\nAs preparation for 128-byte WQE, add a validation for the requested\ninline size compared to the device\u0027s max supported inline size. We must\ndo this validation prior to calculating the WQE size.\n\nUse the newly extended inline size which fallback to legacy value if\ndriver didn\u0027t support query it.\n\nSigned-off-by: Yonatan Nachum \u003cynachum@amazon.com\u003e\n" }, { "commit": "7ae33f982582059fce117b5b618d4f1dffa9cba2", "tree": "a057e102dff11634e4ee4e48ca3da57a8a5ff3dd", "parents": [ "cc0e95cbeba62b17db62f20f32454b514bb3cb7b" ], "author": { "name": "Yonatan Nachum", "email": "ynachum@amazon.com", "time": "Sun Feb 08 10:20:18 2026 +0000" }, "committer": { "name": "Yonatan Nachum", "email": "ynachum@amazon.com", "time": "Tue Mar 17 12:47:14 2026 +0000" }, "message": "efa: Store the new extended inline buf size in EFA ctx\n\nStore the newly added max inline buffer size in EFA context. For\ncompatibility, if the driver doesn\u0027t support the new extended inline\nsize, set it to the legacy one.\n\nSigned-off-by: Yonatan Nachum \u003cynachum@amazon.com\u003e\n" }, { "commit": "cc0e95cbeba62b17db62f20f32454b514bb3cb7b", "tree": "ef92a510d4b0bef841ebf4e3fe1342531783030b", "parents": [ "109428d4533f6909cf84e7cabd0fc9a02ba61beb", "ae50fd76852cd17a7d39806e74f35c2a1e7e14a3" ], "author": { "name": "Leon Romanovsky", "email": "leon@kernel.org", "time": "Mon Mar 16 16:11:09 2026 +0200" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Mon Mar 16 16:11:09 2026 +0200" }, "message": "Merge pull request #1712 from guyharris/fix-signedness-inconsistency\n\nlibibverbs: fix a signedness inconsistency in a return value." }, { "commit": "109428d4533f6909cf84e7cabd0fc9a02ba61beb", "tree": "13306a72d9154215206ba13104b3966c53791791", "parents": [ "e5608da566e4456d87fc86615a59136efaf9803c", "6b3416ec842d38fb3a4309919a7b2f34816c02b9" ], "author": { "name": "Leon Romanovsky", "email": "leon@kernel.org", "time": "Mon Mar 16 16:05:50 2026 +0200" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Mon Mar 16 16:05:50 2026 +0200" }, "message": "Merge pull request #1705 from bluca/upstream_sysusers\n\nInstall sysusers.d config file for rdma system group" }, { "commit": "e5608da566e4456d87fc86615a59136efaf9803c", "tree": "af6e7da31e8fdc89ac13d938a7fe23418bcd1215", "parents": [ "cfedf9ea1a3bb1a81c391c4a3ba32d7c0573d13b", "f0c4973f3dc87a225dabc07f0d3d155f6135ef79" ], "author": { "name": "Leon Romanovsky", "email": "leon@kernel.org", "time": "Mon Mar 16 16:04:56 2026 +0200" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Mon Mar 16 16:04:56 2026 +0200" }, "message": "Merge pull request #1704 from selvintxavier/inline_perf\n\nbnxt_re/lib: Optimize the inline data copy" }, { "commit": "cfedf9ea1a3bb1a81c391c4a3ba32d7c0573d13b", "tree": "59bdd66fdc06149db378f3d85ecddc8494b448b8", "parents": [ "ff72a96b757515d5990e512db186446f6fe219f3", "a2d37c5f6cbb4d563802e416963edd9b922e8467" ], "author": { "name": "Leon Romanovsky", "email": "leon@kernel.org", "time": "Mon Mar 16 16:02:38 2026 +0200" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Mon Mar 16 16:02:38 2026 +0200" }, "message": "Merge pull request #1707 from yehuday/master\n\nCMakeLists: Fix LTTNG trace build" }, { "commit": "ff72a96b757515d5990e512db186446f6fe219f3", "tree": "58edbdd412a1b9937e577805ba3fe3fa1a5b3fe9", "parents": [ "e991ee4ec284cd0e25a5fe1d68562f6ed256d128", "fdd5dbfb71a1f5c80b90f4e1d6340ce21338bf6c" ], "author": { "name": "Yishai Hadas", "email": "yishaih@mellanox.com", "time": "Mon Mar 16 11:30:17 2026 +0200" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Mon Mar 16 11:30:17 2026 +0200" }, "message": "Merge pull request #1711 from yishaih/mlx5_dv\n\nmlx5: Fix mlx5dv_create_flow to keep legacy ABI" }, { "commit": "ae50fd76852cd17a7d39806e74f35c2a1e7e14a3", "tree": "b66cc91a9be6b29074f0a8a6ffcbd56dbc1b5f9c", "parents": [ "e991ee4ec284cd0e25a5fe1d68562f6ed256d128" ], "author": { "name": "Guy Harris", "email": "gharris@sonic.net", "time": "Sat Mar 14 12:02:54 2026 -0700" }, "committer": { "name": "Guy Harris", "email": "gharris@sonic.net", "time": "Sat Mar 14 12:05:02 2026 -0700" }, "message": "libibverbs: fix a signedness inconsistency in a return value.\n\nfind_sysfs_devs_nl() returns 0 on success, -EOPNOTSUPP if\nrdmanl_socket_alloc() fails, and EINVAL on other errors.\n\nIt appears the intent is to return 0 on success and an errno value on\nfailure; given that, it should return EOPNOTSUPP, not -EOPNOTSUPP.\n\nCurrently, this makes no difference, as its return value is only checked\nfor success vs. failure, but if the return value is ever tested for\nspecific errors, or assigned to errno, it\u0027s safer to always return an\nerrno rether than the negative of an errno.\n\nSigned-off-by: Guy Harris \u003cgharris@sonic.net\u003e\n" }, { "commit": "a2d37c5f6cbb4d563802e416963edd9b922e8467", "tree": "eb20a679bf35818f1adf72ec1005c5dd3a8c2c08", "parents": [ "c6d03b4113323e29567a95975206f129cf71994e" ], "author": { "name": "Yehuda Yitschak", "email": "yehuday@amazon.com", "time": "Mon Mar 02 06:14:11 2026 +0000" }, "committer": { "name": "Yehuda Yitschak", "email": "yehuday@amazon.com", "time": "Thu Mar 12 12:32:28 2026 +0000" }, "message": "CMakeLists: Fix LTTNG trace build\n\ncommit \"d4dd187600 providers: add support for USDT tracing using\nlibbpf/usdt\" changed the make flags required for building with trace\nsupport but left out setting ENABLE_LTTNG in the main CMakeLists.txt.\n\nThe provider CMakeLists.txt files check for both ENABLE_LTTNG and\nLTTNGUST_FOUND, but ENABLE_LTTNG was never set. Fix by setting it\nwhen TRACING\u003dLTTNG is specified.\n\nFixes: d4dd1876008d (\"providers: add support for USDT tracing using\n 4 libbpf/usdt\")\nSigned-off-by: Yehuda Yitschak \u003cyehuday@amazon.com\u003e\n" }, { "commit": "fdd5dbfb71a1f5c80b90f4e1d6340ce21338bf6c", "tree": "58edbdd412a1b9937e577805ba3fe3fa1a5b3fe9", "parents": [ "e991ee4ec284cd0e25a5fe1d68562f6ed256d128" ], "author": { "name": "Yishai Hadas", "email": "yishaih@nvidia.com", "time": "Thu Mar 05 09:51:24 2026 +0200" }, "committer": { "name": "Yishai Hadas", "email": "yishaih@nvidia.com", "time": "Wed Mar 11 17:06:35 2026 +0200" }, "message": "mlx5: Fix mlx5dv_create_flow to keep legacy ABI\n\nThe commit ID mentioned by the Fixes line increased the layout of struct\nmlx5dv_flow_action_attr from 16 to 24 bytes, breaking ABI for users of\nthe mlx5dv_create_flow() API and caused a SF where more than one action\nis supplied.\n\nFix this ABI break, by keeping the layout for the original symbol\nversion of mlx5dv_create_flow() while introducing a new symbol to\nsupport a dynamic stride based on the compiled mlx5dv_flow_action_attr\nsize.\n\nFixes: 1dc3e8d9866e (\"mlx5: Add support for bulk flow counters in mlx5dv_create_flow\")\nSigned-off-by: Yishai Hadas \u003cyishaih@nvidia.com\u003e\n" }, { "commit": "e991ee4ec284cd0e25a5fe1d68562f6ed256d128", "tree": "f0be8d616bbe9be3f17d7f88f3b39d1f56dcedbe", "parents": [ "c6d03b4113323e29567a95975206f129cf71994e", "86f6aee0b347536a437050d2c65edfe18b39feda" ], "author": { "name": "Yishai Hadas", "email": "yishaih@mellanox.com", "time": "Sun Mar 08 14:56:23 2026 +0200" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sun Mar 08 14:56:23 2026 +0200" }, "message": "Merge pull request #1706 from yishaih/mlx5_dv\n\nmlx5: Add support for TLP emulation" }, { "commit": "86f6aee0b347536a437050d2c65edfe18b39feda", "tree": "f0be8d616bbe9be3f17d7f88f3b39d1f56dcedbe", "parents": [ "c7d1531f82551679fec998e027267cc23da20dc1" ], "author": { "name": "Maher Sanalla", "email": "msanalla@nvidia.com", "time": "Thu Feb 12 18:49:40 2026 +0200" }, "committer": { "name": "Yishai Hadas", "email": "yishaih@nvidia.com", "time": "Sun Mar 08 11:33:14 2026 +0200" }, "message": "mlx5: Add mlx5dv_devx_umem_export/import\n\nIntroduce mlx5dv_devx_umem_export(), mlx5dv_devx_umem_import() and\nmlx5dv_devx_umem_unimport() APIs to enable cross-process sharing of DEVX\nUMEM objects.\n\nmlx5dv_devx_umem_export() exports the DEVX UMEM into an opaque data\nbuffer whose size is obtained via mlx5dv_get_export_sizes().\n\nmlx5dv_devx_umem_import() creates a DEVX UMEM object from the exported\ndata on a context that originally registered the UMEM, or shares the\nsame kernel resources (e.g., via ibv_import_device()).\n\nmlx5dv_devx_umem_unimport() releases the imported DEVX UMEM without\nderegistering the underlying kernel object.\n\nSigned-off-by: Maher Sanalla \u003cmsanalla@nvidia.com\u003e\nSigned-off-by: Yishai Hadas \u003cyishaih@nvidia.com\u003e\n" }, { "commit": "c7d1531f82551679fec998e027267cc23da20dc1", "tree": "c790de2258f447c7535bbb9e40c0912e33f82400", "parents": [ "d02057685f11bc7201b294b19314cb6ee1a6ac36" ], "author": { "name": "Maher Sanalla", "email": "msanalla@nvidia.com", "time": "Thu Feb 12 18:44:31 2026 +0200" }, "committer": { "name": "Yishai Hadas", "email": "yishaih@nvidia.com", "time": "Sun Mar 08 11:33:14 2026 +0200" }, "message": "mlx5: Add mlx5dv_devx_obj_export/import\n\nIntroduce mlx5dv_devx_obj_export(), mlx5dv_devx_obj_import() and\nmlx5dv_devx_obj_unimport() APIs to enable cross-process sharing of DEVX\nobjects.\n\nmlx5dv_devx_obj_export() exports the DEVX object into an opaque data\nbuffer whose size is obtained via mlx5dv_get_export_sizes().\n\nmlx5dv_devx_obj_import() creates a DEVX object from the exported data on\na context that originally created the object, or shares the same kernel\nresources (e.g., via ibv_import_device()).\n\nmlx5dv_devx_obj_unimport() releases the imported DEVX object without\ndestroying the underlying kernel object.\n\nSigned-off-by: Maher Sanalla \u003cmsanalla@nvidia.com\u003e\nSigned-off-by: Yishai Hadas \u003cyishaih@nvidia.com\u003e\n" }, { "commit": "d02057685f11bc7201b294b19314cb6ee1a6ac36", "tree": "c20e6ae48deaf6c22b88eba2f1a48ca8086de890", "parents": [ "5068f26d1aa8b4ef4ffb7d10cd5d7084b41afb93" ], "author": { "name": "Maher Sanalla", "email": "msanalla@nvidia.com", "time": "Tue Feb 17 17:19:37 2026 +0200" }, "committer": { "name": "Yishai Hadas", "email": "yishaih@nvidia.com", "time": "Sun Mar 08 11:33:14 2026 +0200" }, "message": "mlx5: Add mlx5dv_var_export/import\n\nIntroduce mlx5dv_var_export(), mlx5dv_var_import() and\nmlx5dv_var_unimport() APIs to enable cross-process sharing of VAR\nobjects.\n\nmlx5dv_var_export() exports the VAR into an opaque data buffer whose\nsize is obtained via mlx5dv_get_export_sizes().\n\nmlx5dv_var_import() creates a VAR object from the exported data on a\ncontext that originally allocated the VAR, or shares the same kernel\nresources (e.g., via ibv_import_device()).\n\nmlx5dv_var_unimport() releases the imported VAR without destroying the\nunderlying kernel object.\n\nSigned-off-by: Maher Sanalla \u003cmsanalla@nvidia.com\u003e\nSigned-off-by: Yishai Hadas \u003cyishaih@nvidia.com\u003e\n" }, { "commit": "5068f26d1aa8b4ef4ffb7d10cd5d7084b41afb93", "tree": "c6530abe7e54caf77389ca0426788d39e4b1566a", "parents": [ "1c8a486daaa54700f498cd0583c1e29e951795d8" ], "author": { "name": "Maher Sanalla", "email": "msanalla@nvidia.com", "time": "Thu Feb 12 18:41:46 2026 +0200" }, "committer": { "name": "Yishai Hadas", "email": "yishaih@nvidia.com", "time": "Sun Mar 08 11:33:13 2026 +0200" }, "message": "mlx5: Add mlx5dv_get_export_sizes() for cross-process object sharing\n\nIntroduce mlx5dv_get_export_sizes() API that returns the buffer sizes\nrequired by the export/import APIs, that will be introduced in upcoming\npatches, to allow cross-process sharing of mlx5 objects (VAR, UMEM, DEVX\nobject).\n\nApplications call this API to obtain the required buffer sizes, allocate\naccordingly, then pass the buffers to the per-object export/import\nfunctions.\n\nSigned-off-by: Maher Sanalla \u003cmsanalla@nvidia.com\u003e\nSigned-off-by: Yishai Hadas \u003cyishaih@nvidia.com\u003e\n" }, { "commit": "1c8a486daaa54700f498cd0583c1e29e951795d8", "tree": "67aa280db7f604a39c8ac67afb0e5386f925b394", "parents": [ "8a0194ef86ff5f6d5e340b2de778450091455366" ], "author": { "name": "Maher Sanalla", "email": "msanalla@nvidia.com", "time": "Sat Dec 20 22:26:30 2025 +0200" }, "committer": { "name": "Yishai Hadas", "email": "yishaih@nvidia.com", "time": "Sun Mar 08 11:33:13 2026 +0200" }, "message": "mlx5: Add TLP VAR allocation support\n\nExtend mlx5dv_alloc_var() to support allocating TLP VAR regions used for\nTransaction Layer Packet emulation scenarios.\n\nAdd MLX5DV_VAR_ALLOC_FLAG_TLP flag that applications pass to\nmlx5dv_alloc_var() to request a TLP VAR instead of a standard VAR. When\nthis flag is set, the driver passes it to the kernel via\nMLX5_IB_ATTR_VAR_OBJ_ALLOC_FLAGS so the kernel allocates from the TLP\nVAR region.\n\nSigned-off-by: Maher Sanalla \u003cmsanalla@nvidia.com\u003e\nSigned-off-by: Yishai Hadas \u003cyishaih@nvidia.com\u003e\n" }, { "commit": "8a0194ef86ff5f6d5e340b2de778450091455366", "tree": "5adb1c9d43a33f96c569d5e673bf6662d3cacb99", "parents": [ "c6d03b4113323e29567a95975206f129cf71994e" ], "author": { "name": "Yishai Hadas", "email": "yishaih@nvidia.com", "time": "Sun Mar 08 11:30:55 2026 +0200" }, "committer": { "name": "Yishai Hadas", "email": "yishaih@nvidia.com", "time": "Sun Mar 08 11:30:55 2026 +0200" }, "message": "Update kernel headers\n\nTo commit: 75b864f08773 (\"RDMA/mlx5: Add support for TLP VAR allocation\").\n\nSigned-off-by: Yishai Hadas \u003cyishaih@nvidia.com\u003e\n" }, { "commit": "4c1aee0a78cb540e0a4d7f29c31783d64fe41488", "tree": "6666d27fa9e68db250693b72adede35e9a003283", "parents": [ "c6d03b4113323e29567a95975206f129cf71994e" ], "author": { "name": "Konstantin Taranov", "email": "kotaranov@microsoft.com", "time": "Wed Jan 14 17:01:46 2026 +0100" }, "committer": { "name": "Konstantin Taranov", "email": "kotaranov@microsoft.com", "time": "Fri Mar 06 12:02:01 2026 +0100" }, "message": "providers/mana: allocate/destroy MWs\n\nImplement .alloc_mw() and .dealloc_mw() for mana.\n\nSigned-off-by: Konstantin Taranov \u003ckotaranov@microsoft.com\u003e\n" }, { "commit": "6b3416ec842d38fb3a4309919a7b2f34816c02b9", "tree": "4bfd5e571f08ac2181759086f465a368a63df086", "parents": [ "c6d03b4113323e29567a95975206f129cf71994e" ], "author": { "name": "Luca Boccassi", "email": "luca.boccassi@gmail.com", "time": "Tue Feb 24 22:47:11 2026 +0000" }, "committer": { "name": "Luca Boccassi", "email": "luca.boccassi@gmail.com", "time": "Wed Feb 25 13:31:55 2026 +0000" }, "message": "Install sysusers.d config file for rdma system group\n\nsysusers.d config files allow a package to use declarative\nconfiguration instead of manually written maintainer scripts.\nThis also allows image-based systems to be created with /usr/\nonly, and also allows for factory resetting a system and\nrecreating /etc/ on boot.\n\nhttps://www.freedesktop.org/software/systemd/man/latest/sysusers.d.html\nSigned-off-by: Luca Boccassi \u003cluca.boccassi@gmail.com\u003e\n" }, { "commit": "c6d03b4113323e29567a95975206f129cf71994e", "tree": "86ea59ca77eff580a7552ea21173e7de2ebf0f11", "parents": [ "31af04ec84378724cb6256814d4ffde359a7123b" ], "author": { "name": "Leon Romanovsky", "email": "leonro@nvidia.com", "time": "Tue Feb 24 08:29:58 2026 -0500" }, "committer": { "name": "Leon Romanovsky", "email": "leonro@nvidia.com", "time": "Tue Feb 24 08:29:58 2026 -0500" }, "message": "Update library version to be 63.0\n\nSigned-off-by: Leon Romanovsky \u003cleonro@nvidia.com\u003e\n" }, { "commit": "31af04ec84378724cb6256814d4ffde359a7123b", "tree": "4dc9ab0a1bb0cad5346714d57e685c6e469d1485", "parents": [ "97d14a845cc4807906d47b859c2f643cba2601ec", "f167dea17e40ec49acdb47282075c0dd51bfdde0" ], "author": { "name": "Leon Romanovsky", "email": "leon@kernel.org", "time": "Tue Feb 24 15:09:52 2026 +0200" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Tue Feb 24 15:09:52 2026 +0200" }, "message": "Merge pull request #1703 from ShacharKagan/mem_fixes\n\npyverbs: Fix memory leaks and update Python/Cython compatibility" }, { "commit": "97d14a845cc4807906d47b859c2f643cba2601ec", "tree": "7f85aa684a2bd8bcd9304eee1a1e7248f98fb109", "parents": [ "125c731ea0087aabad1041073244de519bbad732", "d1c38b0cc3f6a1007b774fcd71992b368472ca21" ], "author": { "name": "Leon Romanovsky", "email": "leon@kernel.org", "time": "Tue Feb 24 15:07:40 2026 +0200" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Tue Feb 24 15:07:40 2026 +0200" }, "message": "Merge pull request #1697 from amzn/error-feature-mismatch\n\nEFA: Add new completion status for feature mismatch" }, { "commit": "125c731ea0087aabad1041073244de519bbad732", "tree": "7e282fa397a7f3ee3d4157da22d6bc7b3afb8d14", "parents": [ "8220d19ef03943839fb83259e88b1357bdf1989b", "b7f6164e5612742a9c6b60cd848c6e832725b76f" ], "author": { "name": "Leon Romanovsky", "email": "leon@kernel.org", "time": "Tue Feb 24 15:04:44 2026 +0200" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Tue Feb 24 15:04:44 2026 +0200" }, "message": "Merge pull request #1695 from TaranovK/kotaranov/cq_poll_opt\n\nproviders/mana: Optimize and generalize poll completion and post request paths" }, { "commit": "8220d19ef03943839fb83259e88b1357bdf1989b", "tree": "aa16b09bb5ee57ed33386af12e0436a47e5078e7", "parents": [ "8b9cdb7c6bd2b6e4e64e08888c10124b0d1873f2", "3f091b0f65121f72471c7dfe20639cdc637716cf" ], "author": { "name": "Leon Romanovsky", "email": "leon@kernel.org", "time": "Tue Feb 24 14:59:25 2026 +0200" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Tue Feb 24 14:59:25 2026 +0200" }, "message": "Merge pull request #1692 from kaleshap86/qp_rate_limit\n\nRDMA/bnxt_re: Add QP rate limit support" }, { "commit": "f0c4973f3dc87a225dabc07f0d3d155f6135ef79", "tree": "3a3df2997a7f10ae30fad576c2318b871d6046bc", "parents": [ "8b9cdb7c6bd2b6e4e64e08888c10124b0d1873f2" ], "author": { "name": "Selvin Xavier", "email": "selvin.xavier@broadcom.com", "time": "Tue Feb 17 21:27:50 2026 -0800" }, "committer": { "name": "Selvin Xavier", "email": "selvin.xavier@broadcom.com", "time": "Thu Feb 19 09:18:40 2026 -0800" }, "message": "bnxt_re/lib: Optimize the inline data copy\n\nInline data is currently copied in 16-byte chunks\nto avoid ring-buffer wraparound. Wraparound is rare\nfor typical post_send usage, so when it does not\noccur we can copy the whole inline payload in one go.\n\nImplements a mechanism to handle wraparound and\nnon-wraparound cases separately.\n\nSigned-off-by: Selvin Xavier \u003cselvin.xavier@broadcom.com\u003e\n" }, { "commit": "f167dea17e40ec49acdb47282075c0dd51bfdde0", "tree": "b9e786c3d83857612a3c97bf1602e751a161005e", "parents": [ "0fe9c86990c70ade5b0d3b44bf14d6a2b2d99923" ], "author": { "name": "Shachar Kagan", "email": "skagan@nvidia.com", "time": "Mon Dec 08 17:26:49 2025 +0200" }, "committer": { "name": "Shachar Kagan", "email": "skagan@nvidia.com", "time": "Thu Feb 19 11:17:22 2026 +0200" }, "message": "pyverbs: Replace WeakSet.pop() with list cleanup\n\nReplace the pop()-based cleanup loop with a safer approach that:\n- Converts WeakSet to a list (atomic snapshot)\n- Clears the WeakSet to prevent concurrent access\n- Iterates over the list to close items safely\n\nThis is a workaround for a bug in Python 3.12\u0027s weakref.WeakSet.pop()\nmethod that can cause KeyError exceptions to escape try/except blocks\nduring concurrent modifications. The issue occurs when weakref\ncallbacks raise KeyError from C extension code, bypassing normal Python\nexception handling.\n\nThe fix maintains the same cleanup behavior while avoiding the\nproblematic WeakSet.pop() method entirely. WeakSet fixes are available\nfrom Python 3.14.\n\nSigned-off-by: Shachar Kagan \u003cskagan@nvidia.com\u003e\n" }, { "commit": "0fe9c86990c70ade5b0d3b44bf14d6a2b2d99923", "tree": "5ed291db460080d43a2a9632889047cbbf300cf8", "parents": [ "b788f2136ab40b3ca4afcbe01c0af3d7db5541b1" ], "author": { "name": "Daria Velikovsky", "email": "daria@nvidia.com", "time": "Fri Oct 03 16:22:44 2025 +0300" }, "committer": { "name": "Shachar Kagan", "email": "skagan@nvidia.com", "time": "Thu Feb 19 11:17:22 2026 +0200" }, "message": "pyverbs: Change devx object return type\n\nChange devx object return type to be compatible with Cython 3.x.\n\nSigned-off-by: Daria Velikovsky \u003cdaria@nvidia.com\u003e\nSigned-off-by: Shachar Kagan \u003cskagan@nvidia.com\u003e\n" }, { "commit": "b788f2136ab40b3ca4afcbe01c0af3d7db5541b1", "tree": "5b4c3f60c045431f3f32e8426eb546a79096320f", "parents": [ "8b9cdb7c6bd2b6e4e64e08888c10124b0d1873f2" ], "author": { "name": "Elyashiv Cohen", "email": "elyashivc@nvidia.com", "time": "Wed Aug 20 17:54:05 2025 +0300" }, "committer": { "name": "Shachar Kagan", "email": "skagan@nvidia.com", "time": "Thu Feb 19 11:16:58 2026 +0200" }, "message": "tests: Fix memory leak in test_mlx5_crypto\n\nWhen skipping crypto test the context object need to be close to\nprevent a memory leak.\n\nSigned-off-by: Elyashiv Cohen \u003celyashivc@nvidia.com\u003e\nSigned-off-by: Shachar Kagan \u003cskagan@nvidia.com\u003e\n" }, { "commit": "3f091b0f65121f72471c7dfe20639cdc637716cf", "tree": "c9dd2555045196aa754433421102e083aa6029ae", "parents": [ "8706cef967e0d22b683565d06eb0ebc88220c625" ], "author": { "name": "Kalesh AP", "email": "kalesh-anakkur.purayil@broadcom.com", "time": "Wed Jan 07 11:14:31 2026 +0530" }, "committer": { "name": "Kalesh AP", "email": "kalesh-anakkur.purayil@broadcom.com", "time": "Fri Feb 13 11:08:24 2026 +0530" }, "message": "bnxt_re: Support ibv_modify_qp_rate_limit\n\nImplement the verb ibv_modify_qp_rate_limit to support configuration of\nQP rate limiting. Broadcom device support rate limit only for RC QPs.\nQP must be in RTS or RTS state.\n\nIn case the device doesn\u0027t support QP rate limit or rate_limit is 0\nan error will be returned.\n\nSigned-off-by: Kalesh AP \u003ckalesh-anakkur.purayil@broadcom.com\u003e\n" }, { "commit": "8706cef967e0d22b683565d06eb0ebc88220c625", "tree": "7552209e0ed66fac8729134b3aa09847082cf45f", "parents": [ "eb71afff6c4f8593ff95aaba297a21a6c7506f8e" ], "author": { "name": "Kalesh AP", "email": "kalesh-anakkur.purayil@broadcom.com", "time": "Wed Jan 07 11:10:34 2026 +0530" }, "committer": { "name": "Kalesh AP", "email": "kalesh-anakkur.purayil@broadcom.com", "time": "Fri Feb 13 11:08:20 2026 +0530" }, "message": "bnxt_re: Report QP rate limit capabilities\n\nAdded changes to report QP rate limit capabilities through device query.\nThe new capabilities provide information on minimum and maximum rate\nlimit values and the supported QP types.\n\nSigned-off-by: Kalesh AP \u003ckalesh-anakkur.purayil@broadcom.com\u003e\n" }, { "commit": "eb71afff6c4f8593ff95aaba297a21a6c7506f8e", "tree": "711f72e5ba42199f3bbedfb526fbaafc8db6fda9", "parents": [ "f52b26723f2507466bd5ce368546e18b1ca1668f" ], "author": { "name": "Kalesh AP", "email": "kalesh-anakkur.purayil@broadcom.com", "time": "Tue Jan 06 11:20:01 2026 +0530" }, "committer": { "name": "Kalesh AP", "email": "kalesh-anakkur.purayil@broadcom.com", "time": "Fri Feb 13 11:08:17 2026 +0530" }, "message": "libibverbs: Enhance display of print_packet_pacing_caps\n\nBroadcom P7 devices support rate limiting on RC QPs.\nModified print_packet_pacing_caps() to parse RC QPs.\n\nThe output would look something like below:\n$ ibv_devinfo -v -d bnxt_re0\npacket_pacing_caps:\n\tqp_rate_limit_min:\t8000kbps\n\tqp_rate_limit_max:\t100000000kbps\n\tsupported_qp:\n\t\t\t\tSUPPORT_RC\n\nSigned-off-by: Kalesh AP \u003ckalesh-anakkur.purayil@broadcom.com\u003e\n" }, { "commit": "f52b26723f2507466bd5ce368546e18b1ca1668f", "tree": "333f9f551087e6149200ecb738f2aabef98041b6", "parents": [ "9f95e3b84bc6c87e70ea865e003bc09b5fed821c" ], "author": { "name": "Kalesh AP", "email": "kalesh-anakkur.purayil@broadcom.com", "time": "Tue Jan 06 11:06:56 2026 +0530" }, "committer": { "name": "Kalesh AP", "email": "kalesh-anakkur.purayil@broadcom.com", "time": "Fri Feb 13 11:08:14 2026 +0530" }, "message": "bnxt_re: Add support for rate limit in modify_qp\n\nUpdated to use the extended command to pass rate limit\nto the kernel. (i.e. ibv_cmd_modify_qp_ex).\n\nSigned-off-by: Kalesh AP \u003ckalesh-anakkur.purayil@broadcom.com\u003e\n" }, { "commit": "9f95e3b84bc6c87e70ea865e003bc09b5fed821c", "tree": "81a7d5651b2c9900768391e64505dc7d5ddd0ce8", "parents": [ "6e9643ea97d1f312223801b07c19c0a16615eeb4" ], "author": { "name": "Kalesh AP", "email": "kalesh-anakkur.purayil@broadcom.com", "time": "Fri Feb 13 11:07:58 2026 +0530" }, "committer": { "name": "Kalesh AP", "email": "kalesh-anakkur.purayil@broadcom.com", "time": "Fri Feb 13 11:07:58 2026 +0530" }, "message": "Update kernel headers\n\nTo commit: 13edc7d4e0aa (\"RDMA/bnxt_re: Report packet pacing capabilities\nwhen querying device\").\n\nSigned-off-by: Kalesh AP \u003ckalesh-anakkur.purayil@broadcom.com\u003e\n" }, { "commit": "d1c38b0cc3f6a1007b774fcd71992b368472ca21", "tree": "3535c1c80b09f3c749e41906628ed8f2ed931734", "parents": [ "8b9cdb7c6bd2b6e4e64e08888c10124b0d1873f2" ], "author": { "name": "Chen Brasch", "email": "cbrasch@amazon.com", "time": "Mon Feb 09 14:28:40 2026 +0200" }, "committer": { "name": "Chen Brasch", "email": "cbrasch@amazon.com", "time": "Mon Feb 09 12:52:33 2026 +0000" }, "message": "providers/efa: Add new completion status for feature mismatch\n\nAdd new completion status for remote feature mismatch.\n\nReviewed-by: Dan Cherniavsky \u003cdchern@amazon.com\u003e\nReviewed-by: Michael Margolin \u003cmrgolin@amazon.com\u003e\nSigned-off-by: Chen Brasch \u003ccbrasch@amazon.com\u003e\n" }, { "commit": "8b9cdb7c6bd2b6e4e64e08888c10124b0d1873f2", "tree": "6f004d0d86bb42988e15df2a103374f6afdac48e", "parents": [ "4969c01fe6e5a39ecbcea0f31fdedff870281594", "b685f21fce5247b8f5cecc9d4c44a128dfeee9ef" ], "author": { "name": "Yishai Hadas", "email": "yishaih@mellanox.com", "time": "Sun Feb 08 11:11:58 2026 +0200" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sun Feb 08 11:11:58 2026 +0200" }, "message": "Merge pull request #1678 from yishaih/verbs_misc\n\nverbs: Add support for exporting dma-buf file descriptors" }, { "commit": "b685f21fce5247b8f5cecc9d4c44a128dfeee9ef", "tree": "6f004d0d86bb42988e15df2a103374f6afdac48e", "parents": [ "1b8cbbe9f6c1dedc3c20dca1123f12b3e59faa44" ], "author": { "name": "Yishai Hadas", "email": "yishaih@nvidia.com", "time": "Sun Dec 07 15:05:50 2025 +0200" }, "committer": { "name": "Yishai Hadas", "email": "yishaih@nvidia.com", "time": "Sun Feb 08 10:57:07 2026 +0200" }, "message": "mlx5: Introduce mlx5dv_devx_uar_export_dmabuf_fd()\n\nThis API exports a dmabuf fd that is associated with a given\nmlx5dv_devx_uar.\n\nUpon success the returned fd can be used for DMA and RDMA operations\nassociated with it.\n\nOnce the usage has been ended, close() should be called while supplying\nthe fd. This call will release the resources that were earlier allocated\nusing the mlx5dv_devx_uar_export_dmabuf_fd() API.\n\nA detailed man page was added as well.\n\nSigned-off-by: Yishai Hadas \u003cyishaih@nvidia.com\u003e\n" }, { "commit": "1b8cbbe9f6c1dedc3c20dca1123f12b3e59faa44", "tree": "520880f78f3c759a1874230de7bdb0713086afbc", "parents": [ "f824da4bcd757a5f597c437c78de853bc6a0b6d9" ], "author": { "name": "Yishai Hadas", "email": "yishaih@nvidia.com", "time": "Sun Dec 07 14:49:25 2025 +0200" }, "committer": { "name": "Yishai Hadas", "email": "yishaih@nvidia.com", "time": "Sun Feb 08 10:57:06 2026 +0200" }, "message": "mlx5: Add support for DM export DMABUF fd\n\nAdd support for DM export DMABUF fd.\n\nUpon the allocation flow of the DM the driver may save the pg_off on its\ninternal mlx5_dm structure.\n\nThis data will be used upon exporting a DMABUF fd for the DM by calling\nibv_cmd_export_dmabuf_fd().\n\nSigned-off-by: Yishai Hadas \u003cyishaih@nvidia.com\u003e\n" }, { "commit": "f824da4bcd757a5f597c437c78de853bc6a0b6d9", "tree": "15bf5462a192da7dd8852eda80ae9de11f97b526", "parents": [ "60917319bfd1ced83b1f76f48dc5082e90a5ba69" ], "author": { "name": "Yishai Hadas", "email": "yishaih@nvidia.com", "time": "Mon Dec 08 10:30:58 2025 +0200" }, "committer": { "name": "Yishai Hadas", "email": "yishaih@nvidia.com", "time": "Sun Feb 08 10:57:06 2026 +0200" }, "message": "verbs: Expose ibv_cmd_export_dmabuf_fd() to be used by drivers\n\nExpose ibv_cmd_export_dmabuf_fd() to be used by drivers to call the\nkernel for executing the command.\n\nSigned-off-by: Yishai Hadas \u003cyishaih@nvidia.com\u003e\n" }, { "commit": "60917319bfd1ced83b1f76f48dc5082e90a5ba69", "tree": "bc82709644648d454e2bf889172f6ae048334109", "parents": [ "c7899191531f6a6d7547aeb6f2d3563dd2524b53" ], "author": { "name": "Yishai Hadas", "email": "yishaih@nvidia.com", "time": "Sun Dec 07 13:49:48 2025 +0200" }, "committer": { "name": "Yishai Hadas", "email": "yishaih@nvidia.com", "time": "Sun Feb 08 10:57:06 2026 +0200" }, "message": "verbs: Introduce ibv_dm_export_dmabuf_fd() verb\n\nThis API exports a dmabuf fd that is associated with a given ibv_dm.\n\nUpon success the returned fd can be used for DMA and RDMA operations\nassociated with it.\n\nOnce the usage has been ended, close() should be called while supplying\nthe fd. This call will release the resources that were earlier allocated\nusing the ibv_dm_export_dmabuf_fd() API.\n\nA detailed man page was added as well.\n\nSigned-off-by: Yishai Hadas \u003cyishaih@nvidia.com\u003e\n" }, { "commit": "c7899191531f6a6d7547aeb6f2d3563dd2524b53", "tree": "cf6aa8e9d23dea66965dcb8cc69311cb8039cba6", "parents": [ "4969c01fe6e5a39ecbcea0f31fdedff870281594" ], "author": { "name": "Yishai Hadas", "email": "yishaih@nvidia.com", "time": "Sun Feb 08 10:25:49 2026 +0200" }, "committer": { "name": "Yishai Hadas", "email": "yishaih@nvidia.com", "time": "Sun Feb 08 10:56:43 2026 +0200" }, "message": "Update kernel headers\n\nTo commit: 92a14bb2150a (\"RDMA/mlx5: Implement DMABUF export ops\").\n\nSigned-off-by: Yishai Hadas \u003cyishaih@nvidia.com\u003e\n" }, { "commit": "b7f6164e5612742a9c6b60cd848c6e832725b76f", "tree": "d1054e3cfe4600dfb28fee39fc094691b9bb3f28", "parents": [ "f9a143fb4fb5c8a10008a87bdd4a57e261a104cb" ], "author": { "name": "Konstantin Taranov", "email": "kotaranov@microsoft.com", "time": "Wed Dec 17 15:59:39 2025 +0100" }, "committer": { "name": "Konstantin Taranov", "email": "kotaranov@microsoft.com", "time": "Wed Feb 04 11:11:32 2026 +0100" }, "message": "providers/mana: unify rc_qp to be rnic_qp\n\nAbstract helpers and data structures to handle a general RNIC QP.\nThis change helps to introduce new QP types and reuse the same\nhelpers and the same general fields (e.g., psn, msn).\n\nSigned-off-by: Konstantin Taranov \u003ckotaranov@microsoft.com\u003e\n" }, { "commit": "f9a143fb4fb5c8a10008a87bdd4a57e261a104cb", "tree": "dc942f5bc16d79caf0b914a5f77149f09beb183b", "parents": [ "e277e90b580d69255bab1cac3c3c2f11e47785b9" ], "author": { "name": "Konstantin Taranov", "email": "kotaranov@microsoft.com", "time": "Wed Dec 17 15:57:00 2025 +0100" }, "committer": { "name": "Konstantin Taranov", "email": "kotaranov@microsoft.com", "time": "Wed Feb 04 11:06:13 2026 +0100" }, "message": "providers/mana: Retrieve queue type from queue ID\n\nUse the two least significant bits of the queue ID to identify\nthe queue type. This simplifies QP lookup during CQE processing.\nUse helper functions to retrieve the required queue for QPs.\n\nSigned-off-by: Konstantin Taranov \u003ckotaranov@microsoft.com\u003e\n" }, { "commit": "e277e90b580d69255bab1cac3c3c2f11e47785b9", "tree": "0e17f4574a9c1e1dec9c4fd0f675821365977efc", "parents": [ "4969c01fe6e5a39ecbcea0f31fdedff870281594" ], "author": { "name": "Shiraz Saleem", "email": "shirazsaleem@microsoft.com", "time": "Wed Jun 18 19:03:38 2025 -0500" }, "committer": { "name": "Konstantin Taranov", "email": "kotaranov@microsoft.com", "time": "Wed Feb 04 11:00:48 2026 +0100" }, "message": "providers/mana: Optimize poll completion path\n\nPerformance profiling on ibv_poll_cq shows 2 issues:\n\n1. 0-CQE shows latency that is O(QP_count)\n2. We have hotpaths in mana_poll_cq that point to our list walks\n--48.29%--mana_poll_cq\n |--20.68%--mana_process_completions (inlined)\n |--15.89%--mana_flush_completions (inlined)\n\nRe-design mana poll completion to greedily return WC as and when they\nare available by indexing into the shadow-queue of the QP for this CQE.\nThis avoids expensive list walk of all healthy QPs to find the shadow\nqueue in mana_process_completions.\n\nAdditionally maintain an error qp list to walk for CQEs in error to\nflush.\n\nImprovements in performance measured as avg-latency/CQE for 64 QPs, and\npoll batch size \u003d 32 via ib_write_bw\nBefore,\nCQE (ne\u003e\u003d1): 512.4 ns\nCQE (ne\u003d\u003d0): 292.3 ns\n\nAfter,\nCQE (ne\u003e\u003d1): 344 ns\nCQE (ne\u003d\u003d0): 32.7 ns\n\nSigned-off-by: Konstantin Taranov \u003ckotaranov@microsoft.com\u003e\nSigned-off-by: Shiraz Saleem \u003cshirazsaleem@microsoft.com\u003e\n" }, { "commit": "4969c01fe6e5a39ecbcea0f31fdedff870281594", "tree": "b5ad914d529487598f0cd60396814fd639e09677", "parents": [ "d646a57072b26f82ccafad0cc23ca000ed90f224", "d4dd1876008dfd39fa0548077398e44ef064f298" ], "author": { "name": "Leon Romanovsky", "email": "leon@kernel.org", "time": "Tue Feb 03 14:38:51 2026 +0200" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Tue Feb 03 14:38:51 2026 +0200" }, "message": "Merge pull request #1663 from natoscott/usdt-tracing\n\nproviders: add support for USDT tracing using libbpf/usdt" }, { "commit": "d4dd1876008dfd39fa0548077398e44ef064f298", "tree": "eeebd233f46447c4de7560632539f35a46fb2413", "parents": [ "6e9643ea97d1f312223801b07c19c0a16615eeb4" ], "author": { "name": "Nathan Scott", "email": "nathans@redhat.com", "time": "Thu Nov 13 15:41:43 2025 +1100" }, "committer": { "name": "Nathan Scott", "email": "nathans@redhat.com", "time": "Tue Feb 03 10:31:59 2026 +1100" }, "message": "providers: add support for USDT tracing using libbpf/usdt\n\nProvide lightweight USDT tracing as an alternative to LTTng.\nThis piggybacks on the existing tracing code added for LTTng\nfor a minimal set of changes.\n\nThe cmake build configuration is updated to allow either one\nof -DTRACING\u003dLTTNG or -DTRACING\u003dUSDT for this functionality\n(by default tracing is disabled, as before).\n\n\u003e $ sudo bpftrace -l usdt:build/lib/lib*.so:*\n\u003e usdt:build/lib/libefa-rdmav59.so:rdma_core_efa:post_recv\n\u003e usdt:build/lib/libefa-rdmav59.so:rdma_core_efa:post_send\n\u003e usdt:build/lib/libefa-rdmav59.so:rdma_core_efa:process_completion\n\u003e usdt:build/lib/libefa.so:rdma_core_efa:post_recv\n\u003e usdt:build/lib/libefa.so:rdma_core_efa:post_send\n\u003e usdt:build/lib/libefa.so:rdma_core_efa:process_completion\n\u003e usdt:build/lib/libhns-rdmav59.so:rdma_core_hns:poll_cq\n\u003e usdt:build/lib/libhns-rdmav59.so:rdma_core_hns:post_recv\n\u003e usdt:build/lib/libhns-rdmav59.so:rdma_core_hns:post_send\n\u003e usdt:build/lib/libhns.so:rdma_core_hns:poll_cq\n\u003e usdt:build/lib/libhns.so:rdma_core_hns:post_recv\n\u003e usdt:build/lib/libhns.so:rdma_core_hns:post_send\n\u003e usdt:build/lib/libmlx5-rdmav59.so:rdma_core_mlx5:post_send\n\u003e usdt:build/lib/libmlx5.so:rdma_core_mlx5:post_send\n\u003e usdt:build/lib/librxe-rdmav59.so:rdma_core_rxe:post_send\n\nThe USDT header used here is from the libbpf/usdt project at\nhttps://github.com/libbpf/usdt.git\n\nFurther background discussion for this commit is included in\nhttps://github.com/linux-rdma/rdma-core/pull/1621\n\nSigned-off-by: Nathan Scott \u003cnathans@redhat.com\u003e\n" }, { "commit": "d646a57072b26f82ccafad0cc23ca000ed90f224", "tree": "f4170fc2e786b7b52333cafedbbcca3a7a2c39cf", "parents": [ "75bc96b885c1b077d6fb88bfd84a2971dd5415f6", "b48a28da3f587ccd515f074fe6018d15d0fba7ae" ], "author": { "name": "Leon Romanovsky", "email": "leon@kernel.org", "time": "Mon Feb 02 11:06:32 2026 +0200" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Mon Feb 02 11:06:32 2026 +0200" }, "message": "Merge pull request #1687 from bdrung/wrap-and-sort\n\ndebian: run wrap-and-sort" }, { "commit": "75bc96b885c1b077d6fb88bfd84a2971dd5415f6", "tree": "5636a45abc2e83860d77ae1aadbfade3aac63a09", "parents": [ "26a1cff510764d104079da33a47c18413a32d8dd", "3b94bdec5cab39375faf23735e5739b6469df990" ], "author": { "name": "Leon Romanovsky", "email": "leon@kernel.org", "time": "Mon Feb 02 11:06:07 2026 +0200" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Mon Feb 02 11:06:07 2026 +0200" }, "message": "Merge pull request #1691 from TaranovK/kotaranov/dm_mr\n\nproviders/mana: Device memory support" }, { "commit": "26a1cff510764d104079da33a47c18413a32d8dd", "tree": "c9c0ae07475091018629156e1621565d95c7ef89", "parents": [ "41fbbae76b97ef12a5fe8bca31e0c801a9b889e9" ], "author": { "name": "Yijing Zeng", "email": "zengyijing19900106@gmail.com", "time": "Sun Dec 07 20:40:02 2025 -0800" }, "committer": { "name": "Leon Romanovsky", "email": "leonro@nvidia.com", "time": "Mon Feb 02 04:04:39 2026 -0500" }, "message": "libibverbs: Avoid memcpy from NULL in fill_attr_in()\n\nfill_attr_in() unconditionally calls memcpy() when len \u003c\u003d sizeof(u64),\nregardless of whether the data pointer is NULL. In commit d9af4972d06e\n(\"verbs: Add ibv_cmd_alloc/free commands for DMA handle\"), the call\n\n\tfill_attr_in_enum(cmdb, UVERBS_ATTR_ALLOC_DMAH_TPH_MEM_TYPE,\n\t\t\t attr-\u003etph_mem_type, NULL, 0);\n\nstarted passing a NULL data pointer together with len \u003d\u003d 0, which leads\nto memcpy() being invoked with a NULL source address. While nothing is\nactually copied, some compilers and sanitizers treat this as undefined\nbehavior and emit errors.\n\nFix this by only inlining small attributes when len \u003c\u003d sizeof(u64) and\nthe data pointer is non-NULL. In all other cases, including zero-length\nattributes with a NULL data pointer, the pointer is stored via\nioctl_ptr_to_u64() instead. This preserves the existing behavior for\nvalid callers while avoiding memcpy() from NULL.\n\nfill_attr_in() was originally introduced in commit c34463569326 (\"verbs:\nAdd basic infrastructure support for the kabi ioctl\").\n\nFixes: d9af4972d06e (\"verbs: Add ibv_cmd_alloc/free commands for DMA handle\")\nSigned-off-by: Yijing Zeng \u003czengyijing19900106@gmail.com\u003e\nSigned-off-by: Leon Romanovsky \u003cleonro@nvidia.com\u003e\n" }, { "commit": "41fbbae76b97ef12a5fe8bca31e0c801a9b889e9", "tree": "6c0937dec302ed286abfca10aa0ffcd38636640c", "parents": [ "6e9643ea97d1f312223801b07c19c0a16615eeb4", "197d1a40a5e606a64dddff46daa6353e19c2dea6" ], "author": { "name": "Leon Romanovsky", "email": "leon@kernel.org", "time": "Mon Feb 02 10:53:27 2026 +0200" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Mon Feb 02 10:53:27 2026 +0200" }, "message": "Merge pull request #1690 from zhijianli88/resize_cq\n\ntests/test_cq: Prevent CQ overflows by introducing a delay before resize" }, { "commit": "3b94bdec5cab39375faf23735e5739b6469df990", "tree": "4541eb1921fb18dbc0f39c24adb456ef3af5ba61", "parents": [ "6e9643ea97d1f312223801b07c19c0a16615eeb4" ], "author": { "name": "Konstantin Taranov", "email": "kotaranov@microsoft.com", "time": "Fri Nov 14 15:10:54 2025 +0100" }, "committer": { "name": "Konstantin Taranov", "email": "kotaranov@microsoft.com", "time": "Fri Jan 16 09:47:41 2026 +0100" }, "message": "providers/mana: Device memory\n\nBasic implementation of DM allowing to create and register\nDM memory and use its memory keys for networking.\n\nSigned-off-by: Konstantin Taranov \u003ckotaranov@microsoft.com\u003e\n" }, { "commit": "197d1a40a5e606a64dddff46daa6353e19c2dea6", "tree": "6c0937dec302ed286abfca10aa0ffcd38636640c", "parents": [ "6e9643ea97d1f312223801b07c19c0a16615eeb4" ], "author": { "name": "Li Zhijian", "email": "lizhijian@fujitsu.com", "time": "Fri Jan 16 16:19:07 2026 +0800" }, "committer": { "name": "Li Zhijian", "email": "lizhijian@fujitsu.com", "time": "Fri Jan 16 16:35:19 2026 +0800" }, "message": "tests/test_cq: Prevent CQ overflows by introducing a delay before resize\n\nIn the RXE driver environment, intermittent test failures occur due to the\nresize_cq() operation succeeding before the completion of the posting WQE.\nAs a result, when WQE completion happens, the CQ can become full, and some\nCQEs may be discarded by the kernel, causing poll_cq() to fail.\n\nThe failure is evidenced by the following error message:\n\u003e rxe_enp2s0: cq#342 rxe_cq_post: queue full\n\nTo address this issue simply, introduce a short sleep after post_cq() to\nensure at least one WQE completes before performing the resize_cq()\noperation. This mitigates the race condition that leads to CQ overflow.\n\nSigned-off-by: Li Zhijian \u003clizhijian@fujitsu.com\u003e\n" }, { "commit": "b48a28da3f587ccd515f074fe6018d15d0fba7ae", "tree": "69c533780ffcb1a9f9023443c1a02b5df485b8f7", "parents": [ "6e9643ea97d1f312223801b07c19c0a16615eeb4" ], "author": { "name": "Benjamin Drung", "email": "bdrung@ubuntu.com", "time": "Thu Jan 15 10:39:59 2026 +0100" }, "committer": { "name": "Benjamin Drung", "email": "bdrung@ubuntu.com", "time": "Thu Jan 15 10:39:59 2026 +0100" }, "message": "debian: run wrap-and-sort\n\nKeep list sorted by running `wrap-and-sort`.\n\nSigned-off-by: Benjamin Drung \u003cbdrung@ubuntu.com\u003e\n" }, { "commit": "6e9643ea97d1f312223801b07c19c0a16615eeb4", "tree": "c8e5927633f16a4aad9c330e9d4c233ec6984193", "parents": [ "38eaf306583a51cf731acc5d4ca5e83bb1b0e9ff", "bb176a63b84ecad06859baa8120095bb900f1e15" ], "author": { "name": "Leon Romanovsky", "email": "leon@kernel.org", "time": "Thu Jan 15 10:45:41 2026 +0200" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Thu Jan 15 10:45:41 2026 +0200" }, "message": "Merge pull request #1689 from bdrung/exclude-ionic\n\ndebian: exclude ionic on architectures without coherent DMA" }, { "commit": "38eaf306583a51cf731acc5d4ca5e83bb1b0e9ff", "tree": "da3d8f39b442162f2bd40c38b02cc1f1e5658a3c", "parents": [ "8c0497624a2422caacb0f7068079de2065b3b165", "d7bfb2487b8af747c4bd6356f218717e23639776" ], "author": { "name": "Leon Romanovsky", "email": "leon@kernel.org", "time": "Thu Jan 15 10:45:11 2026 +0200" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Thu Jan 15 10:45:11 2026 +0200" }, "message": "Merge pull request #1688 from bdrung/fix-spelling\n\nFix spelling mistake \"allows to do\"" }, { "commit": "8c0497624a2422caacb0f7068079de2065b3b165", "tree": "4df6fd10873308fe8974d0ebbc8a60f689141b7c", "parents": [ "c289d3b4468e9b2b62f21098a17a60da42365720", "d774d001faeb418954f7c66611063b6400c9f1c4" ], "author": { "name": "Leon Romanovsky", "email": "leon@kernel.org", "time": "Thu Jan 15 10:43:57 2026 +0200" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Thu Jan 15 10:43:57 2026 +0200" }, "message": "Merge pull request #1686 from bdrung/update-year\n\ndebian: update year in copyright" }, { "commit": "c289d3b4468e9b2b62f21098a17a60da42365720", "tree": "b088b1b9fb79c26e25cb72e53d678551c856995e", "parents": [ "5692197186d8d7c7057dfad8ff836bbc7d9683b6", "39b955de18c616c3808718b5545067ba92ddea8c" ], "author": { "name": "Leon Romanovsky", "email": "leon@kernel.org", "time": "Thu Jan 15 10:43:23 2026 +0200" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Thu Jan 15 10:43:23 2026 +0200" }, "message": "Merge pull request #1685 from bdrung/debian-cross-build\n\ndebian: Fix cross compiling" }, { "commit": "bb176a63b84ecad06859baa8120095bb900f1e15", "tree": "1227d24e2f0f8af600cc589fe8bdede2d6b666fb", "parents": [ "5692197186d8d7c7057dfad8ff836bbc7d9683b6" ], "author": { "name": "Benjamin Drung", "email": "bdrung@ubuntu.com", "time": "Wed Jan 14 15:33:17 2026 +0100" }, "committer": { "name": "Benjamin Drung", "email": "bdrung@ubuntu.com", "time": "Wed Jan 14 15:33:17 2026 +0100" }, "message": "debian: exclude ionic on architectures without coherent DMA\n\nThe `ionic` provider is only built on architecture that have coherent\nDMA.\n\nSigned-off-by: Benjamin Drung \u003cbdrung@ubuntu.com\u003e\n" }, { "commit": "d7bfb2487b8af747c4bd6356f218717e23639776", "tree": "e101a0ce18988b15f554c1f29a00e8f1c75b2c72", "parents": [ "5692197186d8d7c7057dfad8ff836bbc7d9683b6" ], "author": { "name": "Benjamin Drung", "email": "bdrung@ubuntu.com", "time": "Wed Jan 14 13:26:07 2026 +0100" }, "committer": { "name": "Benjamin Drung", "email": "bdrung@ubuntu.com", "time": "Wed Jan 14 13:26:07 2026 +0100" }, "message": "Fix spelling mistake \"allows to do\"\n\nlintian complains: Using \"allows to do\" is not correct. Either use\n\"allows one to do\" or \"allows doing\".\n\nSigned-off-by: Benjamin Drung \u003cbdrung@ubuntu.com\u003e\n" }, { "commit": "d774d001faeb418954f7c66611063b6400c9f1c4", "tree": "d18f553c452db5c9017e4cd275b6e47b68a2c082", "parents": [ "5692197186d8d7c7057dfad8ff836bbc7d9683b6" ], "author": { "name": "Benjamin Drung", "email": "bdrung@ubuntu.com", "time": "Wed Jan 14 13:13:17 2026 +0100" }, "committer": { "name": "Benjamin Drung", "email": "bdrung@ubuntu.com", "time": "Wed Jan 14 13:13:17 2026 +0100" }, "message": "debian: update year in copyright\n\nSigned-off-by: Benjamin Drung \u003cbdrung@ubuntu.com\u003e\n" }, { "commit": "5692197186d8d7c7057dfad8ff836bbc7d9683b6", "tree": "52e6eee68e0d99b6a122b27e9a8bffbc9ecc99a3", "parents": [ "e420131a48cef717abc4ee6e72c59599a1c284d6", "51c365bfccead0738a8c9763eeb6f029527c2898" ], "author": { "name": "Leon Romanovsky", "email": "leon@kernel.org", "time": "Wed Jan 14 13:46:02 2026 +0200" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Wed Jan 14 13:46:02 2026 +0200" }, "message": "Merge pull request #1684 from bdrung/rules-requires-root\n\ndebian: drop Rules-Requires-Root" }, { "commit": "39b955de18c616c3808718b5545067ba92ddea8c", "tree": "ea8838a556558d5137232131694f59a1a63f9b93", "parents": [ "e420131a48cef717abc4ee6e72c59599a1c284d6" ], "author": { "name": "Helmut Grohne", "email": "helmut@subdivi.de", "time": "Wed Dec 17 15:23:03 2025 +0100" }, "committer": { "name": "Benjamin Drung", "email": "bdrung@ubuntu.com", "time": "Wed Jan 14 12:13:14 2026 +0100" }, "message": "debian: Fix cross compiling\n\nrdma-core fails to cross build from source and all of the causes relate\nto the Python extension build.\n\nA build dependency on python3-dev requests the host architecture\ninterpreter. It can neither be installed nor run during build. For\nbuilding extensions, the typical dependency should be libpython3-dev +\npython3-dev:native. Beyond this, cython3 is requested for the host\narchitecture. It likewise cannot be run that way. While cython3 can have\narchitecture-dependent behavior and therefore is not marked M-A:foreign,\nits use in rdma-core looks ok to annotate the dependency :native.\n\nWith these changes, rdma-core successfully builds broken packages. The\nPython extensions use suffixes derived from the build architecture. To\nfix this, the host\u0027s sysconfigdata should be used by exporting the\nrelevant environment variable during build.\n\nBug-Debian: https://bugs.debian.org/1123693\nSigned-off-by: Benjamin Drung \u003cbdrung@ubuntu.com\u003e\n" }, { "commit": "51c365bfccead0738a8c9763eeb6f029527c2898", "tree": "52e6eee68e0d99b6a122b27e9a8bffbc9ecc99a3", "parents": [ "e420131a48cef717abc4ee6e72c59599a1c284d6" ], "author": { "name": "Benjamin Drung", "email": "bdrung@ubuntu.com", "time": "Wed Jan 14 11:46:52 2026 +0100" }, "committer": { "name": "Benjamin Drung", "email": "bdrung@ubuntu.com", "time": "Wed Jan 14 11:46:52 2026 +0100" }, "message": "debian: drop Rules-Requires-Root\n\nSpecifying `Rules-Requires-Root: no` is not needed any more since it is\nthe default.\n\nSigned-off-by: Benjamin Drung \u003cbdrung@ubuntu.com\u003e\n" }, { "commit": "e420131a48cef717abc4ee6e72c59599a1c284d6", "tree": "8bfd2d07576392e560f58f37d95c2f023c73b3c0", "parents": [ "35bad8c933a1b77d4576dd86674f88817bb93288", "5570deee0546236e98674a89ec08c38484894a36" ], "author": { "name": "Leon Romanovsky", "email": "leon@kernel.org", "time": "Wed Jan 14 10:13:27 2026 +0200" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Wed Jan 14 10:13:27 2026 +0200" }, "message": "Merge pull request #1683 from bdrung/debian-copyright\n\ndebian: replace old FSF postal address by URL" }, { "commit": "35bad8c933a1b77d4576dd86674f88817bb93288", "tree": "fb77da9137460e3bc5232c7fd8bf240ef68fca6a", "parents": [ "19c4baa8a3be0f54c4c2c67c233a174018265688", "0b5f67fbe78f61945151a4a3e620d88b7b94876a" ], "author": { "name": "Leon Romanovsky", "email": "leon@kernel.org", "time": "Wed Jan 14 10:12:40 2026 +0200" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Wed Jan 14 10:12:40 2026 +0200" }, "message": "Merge pull request #1682 from bdrung/debian-python-dependency\n\ndebian: add ${python3:Depends} to rdma-core for rdma_topo" }, { "commit": "19c4baa8a3be0f54c4c2c67c233a174018265688", "tree": "5e535a813b7dc431913c87511951d3ec547eec6e", "parents": [ "db312dd339f0a6a655d05d793ef9f002083e31a7", "18d8571321d28b50da42953c61e5dded42d06a37" ], "author": { "name": "Leon Romanovsky", "email": "leon@kernel.org", "time": "Wed Jan 14 10:06:39 2026 +0200" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Wed Jan 14 10:06:39 2026 +0200" }, "message": "Merge pull request #1680 from ShacharKagan/pyverbs_resolve_addr\n\ntests: Add resolve and query addrinfo test" }, { "commit": "db312dd339f0a6a655d05d793ef9f002083e31a7", "tree": "7135cf01f434b371f88fe31d8c33320b27a5d41a", "parents": [ "ce001cc64f499774fcd6bfb7fb8f8225dedac153", "a09db86ec3552408785648b30fafe3e82e51bba2" ], "author": { "name": "Leon Romanovsky", "email": "leon@kernel.org", "time": "Wed Jan 14 10:05:03 2026 +0200" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Wed Jan 14 10:05:03 2026 +0200" }, "message": "Merge pull request #1679 from TaranovK/kotaranov/fix_cq_check\n\nproviders/mana: do not check cqid on creation" }, { "commit": "ce001cc64f499774fcd6bfb7fb8f8225dedac153", "tree": "4df9fd48bb6b4686c7bf3a4052efee676d0efa6c", "parents": [ "c7cc0c8ef7f05cb951f47e8c003c905529913446", "fdb42b169ed7a774c3c73636d326450ca0960b15" ], "author": { "name": "Leon Romanovsky", "email": "leon@kernel.org", "time": "Wed Jan 14 09:56:49 2026 +0200" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Wed Jan 14 09:56:49 2026 +0200" }, "message": "Merge pull request #1681 from bdrung/debian-add-libionic1\n\ndebian: add libionic1 to ibverbs-providers lintian-overrides" }, { "commit": "5570deee0546236e98674a89ec08c38484894a36", "tree": "212e9b51dadd542ea42e71046b368c7823547097", "parents": [ "c7cc0c8ef7f05cb951f47e8c003c905529913446" ], "author": { "name": "Benjamin Drung", "email": "bdrung@ubuntu.com", "time": "Wed Jan 14 01:55:29 2026 +0100" }, "committer": { "name": "Benjamin Drung", "email": "bdrung@ubuntu.com", "time": "Wed Jan 14 01:55:29 2026 +0100" }, "message": "debian: replace old FSF postal address by URL\n\nLintian complains: The debian copyright file refers to the old postal\naddress of the Free Software Foundation (FSF). You should refer to\nhttps://www.gnu.org/licenses/ instead of a physical address,\n\nSigned-off-by: Benjamin Drung \u003cbdrung@ubuntu.com\u003e\n" }, { "commit": "0b5f67fbe78f61945151a4a3e620d88b7b94876a", "tree": "4472508287ed18bc6ac55bb8f9f876e25217188a", "parents": [ "c7cc0c8ef7f05cb951f47e8c003c905529913446" ], "author": { "name": "Benjamin Drung", "email": "bdrung@ubuntu.com", "time": "Wed Jan 14 01:40:08 2026 +0100" }, "committer": { "name": "Benjamin Drung", "email": "bdrung@ubuntu.com", "time": "Wed Jan 14 01:40:08 2026 +0100" }, "message": "debian: add ${python3:Depends} to rdma-core for rdma_topo\n\nThe new `rdma_topo` command is a Python script and therefore rdma-core\nmust depend on Python. Rely on the dh-python3 helper to determine the\ncorrect dependency.\n\nSigned-off-by: Benjamin Drung \u003cbdrung@ubuntu.com\u003e\n" }, { "commit": "fdb42b169ed7a774c3c73636d326450ca0960b15", "tree": "4df9fd48bb6b4686c7bf3a4052efee676d0efa6c", "parents": [ "c7cc0c8ef7f05cb951f47e8c003c905529913446" ], "author": { "name": "Benjamin Drung", "email": "bdrung@ubuntu.com", "time": "Wed Jan 14 01:36:29 2026 +0100" }, "committer": { "name": "Benjamin Drung", "email": "bdrung@ubuntu.com", "time": "Wed Jan 14 01:36:29 2026 +0100" }, "message": "debian: add libionic1 to ibverbs-providers lintian-overrides\n\nAdd `libionic1` to the lintian overrides for ibverbs-providers.\n\nSigned-off-by: Benjamin Drung \u003cbdrung@ubuntu.com\u003e\n" }, { "commit": "18d8571321d28b50da42953c61e5dded42d06a37", "tree": "a8c8a4ab4a334d246af0b28a06b7332088692909", "parents": [ "92e755566e8e0f95b0114d752e20e5f4f695ec86" ], "author": { "name": "Mazal Or Yakuel", "email": "moyakuel@nvidia.com", "time": "Thu Nov 20 14:43:07 2025 +0200" }, "committer": { "name": "Shachar Kagan", "email": "skagan@nvidia.com", "time": "Sun Jan 11 10:56:37 2026 +0200" }, "message": "tests: Fix missing port in resolve addrinfo test\n\nTest test_rdmacm_async_resolve_addrinfo was not passing ib_port to\ntwo_nodes_rdmacm_traffic(), causing it to default to port 1 regardless\nof the --port cmmand-line argument.\n\nFix by explicitly passing ib_port\u003dself.ib_port, consistent with other\ntests like test_rdmacm_async_udp_traffic.\n\nSigned-off-by: Mazal Or Yakuel \u003cmoyakuel@nvidia.com\u003e\nSigned-off-by: Shachar Kagan \u003cskagan@nvidia.com\u003e\n" }, { "commit": "92e755566e8e0f95b0114d752e20e5f4f695ec86", "tree": "7677b1580b0a97cee0a5c227b36b5b4d91364941", "parents": [ "3f5b710a2d604626f1419608c9fabd66912185ae" ], "author": { "name": "Mazal Or Yakuel", "email": "moyakuel@nvidia.com", "time": "Wed May 21 13:54:38 2025 +0300" }, "committer": { "name": "Shachar Kagan", "email": "skagan@nvidia.com", "time": "Sun Jan 11 10:56:37 2026 +0200" }, "message": "tests: Fix process exit code checking\n\nImprove the process exit code checking logic in\ntwo_nodes_rdmacm_traffic()\nfunction to properly handle test process termination and exit codes.\n\nThe function now correctly:\n- Tracks process exit codes for both passive and active sides\n- Handles SkipTest cases (exit code 5)\n- Verifies successful completion (exit code 0)\n\nSigned-off-by: Mazal Or Yakuel \u003cmoyakuel@nvidia.com\u003e\nSigned-off-by: Shachar Kagan \u003cskagan@nvidia.com\u003e\n" }, { "commit": "3f5b710a2d604626f1419608c9fabd66912185ae", "tree": "99115dd6f3c391df4f8c9689a969c02d49ed4849", "parents": [ "a019dc7973d24e85c381422ae7df389b9bcf335a" ], "author": { "name": "Mazal Or Yakuel", "email": "moyakuel@nvidia.com", "time": "Mon May 05 11:44:41 2025 +0300" }, "committer": { "name": "Shachar Kagan", "email": "skagan@nvidia.com", "time": "Sun Jan 11 10:56:37 2026 +0200" }, "message": "tests: Add resolve and query addr_info CM test\n\nAdd test support for RDMA CM address resolution and querying:\n- Add resolve_and_update_addrinfo method to CMConnection base class\n- Add get_and_update_addrinfo method to CMConnection base class\n- Add test_rdmacm_async_resolve_addrinfo test to verify\n resolve_addrinfo in async mode\n- Update CM resource creation method to support resolve_addrinfo flow\n- Add support for passive and active side address resolution in tests\n\nThis change adds comprehensive test coverage for RDMA CM address\nresolution and querying functionality, ensuring proper testing of\nboth synchronous and asynchronous connection modes.\n\nSigned-off-by: Mazal Or Yakuel \u003cmoyakuel@nvidia.com\u003e\nSigned-off-by: Shachar Kagan \u003cskagan@nvidia.com\u003e\n" }, { "commit": "a019dc7973d24e85c381422ae7df389b9bcf335a", "tree": "4fd0f142982ef3a92f952cb39b111738a0d6037e", "parents": [ "c7cc0c8ef7f05cb951f47e8c003c905529913446" ], "author": { "name": "Mazal Or Yakuel", "email": "moyakuel@nvidia.com", "time": "Sun May 04 15:31:48 2025 +0300" }, "committer": { "name": "Shachar Kagan", "email": "skagan@nvidia.com", "time": "Sun Jan 11 10:56:37 2026 +0200" }, "message": "pyverbs: Add resolve_addrinfo \u0026 query_addrinfo\n\n- Adding resolve_addrinfo and query_addrinfo methods to CMID class\n- Adding get_addrinfo to CMID class\n\nSigned-off-by: Mazal Or Yakuel \u003cmoyakuel@nvidia.com\u003e\nSigned-off-by: Shachar Kagan \u003cskagan@nvidia.com\u003e\n" }, { "commit": "a09db86ec3552408785648b30fafe3e82e51bba2", "tree": "7631794c748534d66ca9b82e7acde770150aa586", "parents": [ "c7cc0c8ef7f05cb951f47e8c003c905529913446" ], "author": { "name": "Konstantin Taranov", "email": "kotaranov@microsoft.com", "time": "Wed Dec 24 11:27:52 2025 +0100" }, "committer": { "name": "Konstantin Taranov", "email": "kotaranov@microsoft.com", "time": "Fri Jan 09 13:06:14 2026 +0100" }, "message": "providers/mana: do not check cqid on creation\n\nDo not check whether cqid is assigned as mana_ib\nover ethernet assigns it later at QP creation.\n\nSigned-off-by: Konstantin Taranov \u003ckotaranov@microsoft.com\u003e\n" }, { "commit": "c7cc0c8ef7f05cb951f47e8c003c905529913446", "tree": "54882039c145d0a25cba6805c5f7e93749275431", "parents": [ "fd865762559a7e002515eadd63da5e0c03be61e0", "5083d3af8f0e67a2d2e4ea425b87bebafc545ef0" ], "author": { "name": "Yishai Hadas", "email": "yishaih@mellanox.com", "time": "Wed Jan 07 10:25:09 2026 +0200" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Wed Jan 07 10:25:09 2026 +0200" }, "message": "Merge pull request #1674 from yishaih/verbs_misc\n\nverbs: Introduce ibv_query_port_speed() verb" }, { "commit": "5083d3af8f0e67a2d2e4ea425b87bebafc545ef0", "tree": "54882039c145d0a25cba6805c5f7e93749275431", "parents": [ "d4870bddee0005792a10476e172bdc85a4d04f9d" ], "author": { "name": "Or Har-Toov", "email": "ohartoov@nvidia.com", "time": "Thu Sep 04 16:47:24 2025 +0300" }, "committer": { "name": "Yishai Hadas", "email": "yishaih@nvidia.com", "time": "Wed Jan 07 09:53:17 2026 +0200" }, "message": "verbs: Add IBV_EVENT_DEVICE_SPEED_CHANGE to async events\n\nIntroduce IBV_EVENT_DEVICE_SPEED_CHANGE asynchronous event, which is\ntriggered when one or many of the device\u0027s port speeds change due to\nconfiguration or state updates. This is especially useful in scenarios\nsuch as VF LAG or Multiport LAG configuration, where the effective speed\nof a VF can change if one of the PFs in the LAG changes state.\n\nAfter receiving this event, applications can use ibv_query_port_speed()\non specific ports to check which ones have changed and obtain their new\nspeeds.\n\nThe event is generated at the device level to avoid flooding the user\nwith many separate notifications when multiple ports are affected\nsimultaneously, such as when a PF with many SFs changes state or when\nall ports are part of a bonded device that updates speed. Users can then\nquery only those ports they care about for details.\n\nSigned-off-by: Or Har-Toov \u003cohartoov@nvidia.com\u003e\nReviewed-by: Patrisious Haddad \u003cphaddad@nvidia.com\u003e\nSigned-off-by: Yishai Hadas \u003cyishaih@nvidia.com\u003e\n" }, { "commit": "d4870bddee0005792a10476e172bdc85a4d04f9d", "tree": "f6bf841755ced5a1845cc3810ef08599622f291d", "parents": [ "a97fb539fdc2be4c193dd326d654334e9a18067c" ], "author": { "name": "Or Har-Toov", "email": "ohartoov@nvidia.com", "time": "Thu Nov 13 11:33:57 2025 +0200" }, "committer": { "name": "Yishai Hadas", "email": "yishaih@nvidia.com", "time": "Wed Jan 07 09:53:10 2026 +0200" }, "message": "mlx5: Implement ibv_query_port_speed() verb\n\nAdd support for the newly introduced ibv_query_port_speed() API in the\nmlx5 provider. This enables applications to obtain the effective port\nbandwidth, which for VF LAG and multiport VF configurations is\ncalculated as the sum of bandwidths of all currently active PFs that are\npart of the aggregation.\n\nSigned-off-by: Or Har-Toov \u003cohartoov@nvidia.com\u003e\nReviewed-by: Patrisious Haddad \u003cphaddad@nvidia.com\u003e\nSigned-off-by: Yishai Hadas \u003cyishaih@nvidia.com\u003e\n" }, { "commit": "a97fb539fdc2be4c193dd326d654334e9a18067c", "tree": "e080b509cb9f55cb3aa372fe8d6b9c29cc1f6a89", "parents": [ "8504597211e7982c3cf2ade04e778fe0e316017c" ], "author": { "name": "Or Har-Toov", "email": "ohartoov@nvidia.com", "time": "Thu Nov 13 11:33:30 2025 +0200" }, "committer": { "name": "Yishai Hadas", "email": "yishaih@nvidia.com", "time": "Wed Jan 07 09:53:04 2026 +0200" }, "message": "verbs: Add ioctl for ibv_query_port_speed() verb\n\nAdd ioctl support for the new ibv_query_port_speed() verb, allowing\nproviders to query the port speed from kernel via ioctl-based\ninfrastructure.\n\nSigned-off-by: Or Har-Toov \u003cohartoov@nvidia.com\u003e\nReviewed-by: Patrisious Haddad \u003cphaddad@nvidia.com\u003e\nSigned-off-by: Yishai Hadas \u003cyishaih@nvidia.com\u003e\n" }, { "commit": "8504597211e7982c3cf2ade04e778fe0e316017c", "tree": "bfaaf82462bd145bb4de1772f62f19e6d25c96f0", "parents": [ "d026300ac795bfbe365a3aba436ced72a4da8020" ], "author": { "name": "Yishai Hadas", "email": "yishaih@nvidia.com", "time": "Wed Jan 07 09:51:41 2026 +0200" }, "committer": { "name": "Yishai Hadas", "email": "yishaih@nvidia.com", "time": "Wed Jan 07 09:51:41 2026 +0200" }, "message": "verbs: Introduce ibv_query_port_speed() verb\n\nAdd a new API, ibv_query_port_speed(), which enables applications to\nquery the effective bandwidth of a port. This is useful for VF LAG and\nmultiport LAG configurations, since ibv_query_port does not report the\ncombined bandwidth of all aggregated ports. In any case, the API can be\nused on non-aggregated devices as well.\n\nSigned-off-by: Or Har-Toov \u003cohartoov@nvidia.com\u003e\nReviewed-by: Patrisious Haddad \u003cphaddad@nvidia.com\u003e\nSigned-off-by: Yishai Hadas \u003cyishaih@nvidia.com\u003e\n" }, { "commit": "d026300ac795bfbe365a3aba436ced72a4da8020", "tree": "19dd553cd1d1cdf1ec7d3c77fc001994f847274b", "parents": [ "fd865762559a7e002515eadd63da5e0c03be61e0" ], "author": { "name": "Yishai Hadas", "email": "yishaih@nvidia.com", "time": "Wed Jan 07 09:37:59 2026 +0200" }, "committer": { "name": "Yishai Hadas", "email": "yishaih@nvidia.com", "time": "Wed Jan 07 09:37:59 2026 +0200" }, "message": "Update kernel headers\n\nTo commit: aaecff5e13cd (\"RDMA/mlx5: Implement query_port_speed callback\").\n\nSigned-off-by: Yishai Hadas \u003cyishaih@nvidia.com\u003e\n" }, { "commit": "fd865762559a7e002515eadd63da5e0c03be61e0", "tree": "9a64d117f36e4bb4ec76924563a01558a39e382b", "parents": [ "902dce1cee1fe74123a854e22858d604ee3ff4f7", "9adb2c8bb2142f01093148aa2a0b0d9cfc5f38c3" ], "author": { "name": "Leon Romanovsky", "email": "leon@kernel.org", "time": "Tue Jan 06 15:26:11 2026 +0200" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Tue Jan 06 15:26:11 2026 +0200" }, "message": "Merge pull request #1677 from bdrung/fix-libibverbs-symbols\n\ndebian: correct symbol version of new ibverbs 1.15 functions" }, { "commit": "9adb2c8bb2142f01093148aa2a0b0d9cfc5f38c3", "tree": "9a64d117f36e4bb4ec76924563a01558a39e382b", "parents": [ "902dce1cee1fe74123a854e22858d604ee3ff4f7" ], "author": { "name": "Benjamin Drung", "email": "bdrung@ubuntu.com", "time": "Tue Jan 06 13:07:04 2026 +0100" }, "committer": { "name": "Benjamin Drung", "email": "bdrung@ubuntu.com", "time": "Tue Jan 06 13:07:04 2026 +0100" }, "message": "debian: correct symbol version of new ibverbs 1.15 functions\n\nThe functions `ibv_alloc_dmah`, `ibv_dealloc_dmah`, and `ibv_reg_mr_ex`\nwere introduced in rdma-core version 59 and not version 37.\n\nFixes: c2ba4c10bf1b (\"verbs: Introduce verbs for DMA handle allocation and deallocation\")\nFixes: d2391095ed42 (\"verbs: Introduce the ibv_reg_mr_ex() verb\")\nSigned-off-by: Benjamin Drung \u003cbdrung@ubuntu.com\u003e\n" }, { "commit": "902dce1cee1fe74123a854e22858d604ee3ff4f7", "tree": "28020dfb5e9b925c929e70a45047f46ad928c152", "parents": [ "ed0476488e9ef725b1b181e0158f99093ae37c78", "61d4e59f58de0ee75a335b0be7175a56c620c83d" ], "author": { "name": "Leon Romanovsky", "email": "leon@kernel.org", "time": "Wed Dec 31 09:51:01 2025 +0200" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Wed Dec 31 09:51:01 2025 +0200" }, "message": "Merge pull request #1675 from hginjgerx/fix\n\nlibhns: Two fixes" }, { "commit": "4a05c56368f3a3600f680b05127d3e4af2e731f7", "tree": "d82c45586cae1ffc31ede23b6048474f2287458f", "parents": [ "ed0476488e9ef725b1b181e0158f99093ae37c78" ], "author": { "name": "Abhijit Gangurde", "email": "abhijit.gangurde@amd.com", "time": "Tue Dec 30 11:56:15 2025 +0530" }, "committer": { "name": "Abhijit Gangurde", "email": "abhijit.gangurde@amd.com", "time": "Wed Dec 31 12:01:21 2025 +0530" }, "message": "ionic: Reserve bytes in CQE format for future use\n\nReserve a few bytes from wqe_id in the CQE format for future use.\n\nSigned-off-by: Abhijit Gangurde \u003cabhijit.gangurde@amd.com\u003e\n" }, { "commit": "61d4e59f58de0ee75a335b0be7175a56c620c83d", "tree": "28020dfb5e9b925c929e70a45047f46ad928c152", "parents": [ "20147f9abd65690b594544d10ccd1601741de2f4" ], "author": { "name": "Lianfa Weng", "email": "wenglianfa@huawei.com", "time": "Tue Dec 30 23:54:22 2025 +0800" }, "committer": { "name": "Junxian Huang", "email": "huangjunxian6@hisilicon.com", "time": "Tue Dec 30 23:59:01 2025 +0800" }, "message": "libhns: Fix CQ clean when CQ record doorbell enabled\n\nThe update of CQ record doorbell is missing in hns_roce_v2_cq_clean().\nHandle both record doorbell and the HW doorbell in update_cq_db() to\navoid any omissions.\n\nFixes: 3bb4f045752e (\"libhns: Support cq record doorbell\")\nSigned-off-by: Lianfa Weng \u003cwenglianfa@huawei.com\u003e\nSigned-off-by: Junxian Huang \u003chuangjunxian6@hisilicon.com\u003e\n" } ], "next": "20147f9abd65690b594544d10ccd1601741de2f4" }