UCT: Add outstanding operation extraction support (#11580)
* UCT: Add outstanding operation extraction support * UCT/MLX5: fix format * UCT/MLX5: add msn meta * UCT/IB/MLX5: Defer failed send ownership to failover Arm RC mlx5 endpoints before peer failure so error handling leaves outstanding sends available for extraction. Suppress user completions while releasing CQ resources, and let endpoint destruction hand the failed QP to asynchronous garbage collection. * GTEST/UCT: Cover outstanding operation extraction Verify that extraction reports posted operations in order and preserves the metadata needed by failover replay. * UCT/IB/MLX5: address comments * UCT/FAILOVER: Harden mlx5 outstanding extraction * UCT/RMA: Add completion-backed PUT bcopy * UCT/RMA: Rename completion-backed PUT bcopy API Use the failover-specific _ft suffix for the completion-backed PUT bcopy entry point and transport operation. Remove the redundant capability bit since availability follows QUERY_TOKEN support. * UCT/API: Refine outstanding operation metadata Organize extracted operation metadata by protocol parameter group, align field-mask ordering with the structure layout, and distinguish AM flags from flush parameters. Also align internal operation initializers touched by outstanding extraction support. * UCT/IB/MLX5: Guard MSN metadata updates * UCT: Add endpoint failover enable state * UCT/IB/MLX5: Remove redundant extract wrapper * UCT/API: fix doc build * UCT: fix format * UCT/RMA: Add completion-backed PUT short * UCT: cleanup * UCT: address comments * UCT/IB/MLX5: add mlx5 priv * TEST/UCT: add plugin test * TEST/UCT: fix CI * UCT: fix format * UCT/API: keep API and stubs on failover_ext_uct Move mlx5 WQE private metadata tracking to failover_ext_uct_impl and retain the public outstanding-purge API with plugin dispatch stubs and unit tests on this branch. * UCT/API: Keep api only changes. * UCT: fix format * UCT: address comments * UCT/API: address comments * UCT: cleanup changes and fix param check * UCT: address comments * UCT/API: address comments * UCT/BASED: address comments * UCT/BASED: address comments * UCT/PLUGIN: address comments --------- Co-authored-by: Zihao Zhao <zizhao@nvidia.com>
Z
Zihao Zhao committed
d4148b16734f11563f4b96df32c6ea149ee3f05a
Parent: b68a434
Committed by GitHub <noreply@github.com>
on 7/16/2026, 11:32:14 AM