MOVZXandHinata cb27582783 recoverssm: release the RecoverSSM step after CUDA graph memory profiling (port of upstream #58863, fdcc43cf3e) ...

Port of the upstream vLLM change #58863 (Apache-2.0). The step recorded during graph-capture dummy runs held the profiling KV cache
alive after profiling; it is now released at teardown.

Co-Authored-By: Hinata <hinata@idihore.id>
Signed-off-by: MOVZX <movzx@yahoo.com>

2026-10-09 20:08:55 +07:00

..

attention

[KV] Restore PP fix #54795 -- PP stages need only share a KV layout

2026-10-09 19:55:23 +07:00

core

[Scheduler] Drop the one-victim preemption guard that upstream #49675 made dead

2026-10-09 20:01:26 +07:00

engine

[Engine] Detect a fail-closed fault or death on any rank

2026-10-09 20:04:37 +07:00

executor

[Engine] Detect a fail-closed fault or death on any rank

2026-10-09 20:04:37 +07:00

fault_tolerance

[Bugfix][FT] Pass stateless process-group timeouts explicitly (#58611)

2026-09-29 03:14:02 +00:00

hisparse

[Bugfix][HiSparse] Fix MTP acceptance collapse under FULL graphs with a saturated GPU pool (#59309)

2026-10-01 23:04:21 -07:00

kv_hints

[Feature] Add first-class KV hints request envelope for programmatic KV management (#53423)

2026-09-21 09:56:11 -07:00

kv_offload

[Bugfix][Core] Exempt exactly the blocks an async KV load writes from zeroing (#59504)

2026-10-03 06:26:12 +00:00

metrics

[Metrics][P/D] Add KV-fetch stage gauges for async KV loads (#58874)

2026-10-02 01:24:33 +00:00

ple_offload

[Engine] Detect a fail-closed fault or death on any rank

2026-10-09 20:04:37 +07:00

pool

[Bugfix][Pooling] Handle multimodal cache misses without crashing (#58975)

2026-10-02 10:14:33 -04:00

sample

[CPU][Perf] Add vectorized Sampler Kernel (#53913)

2026-09-30 15:16:28 +08:00

simple_kv_offload

[Bugfix][KV Offload] Reset SimpleCPU eager-store placement state on MRV2 resume (#57816)

2026-10-05 02:00:07 +00:00

spec_decode

[PCP][DCP] Support DCP target model with non-DCP Dspark (#56723)

2026-09-26 18:12:35 -04:00

structured_output

[Rebase] Remove v0.30.0 merge artefacts; import order

2026-10-09 19:47:45 +07:00

watermarking

[Bugfix][Watermarking] Keep draft prompt lengths valid under CUDA graphs (#59779)

2026-10-02 14:32:11 -04:00

worker

recoverssm: release the RecoverSSM step after CUDA graph memory profiling (port of upstream #58863, fdcc43cf3e)

2026-10-09 20:08:55 +07:00

__init__.py

…

cudagraph_dispatcher.py

Add pydocstyle to the ruff rules (#52136)

2026-09-16 16:44:29 +00:00

e2_guards.py

[Guards] Draft identity and draft PLE guard report the drafter's own state

2026-10-09 20:03:19 +07:00

kv_cache_interface.py

[Bugfix][HiSparse] Size the KV cache from the groups HiSparse allocates (#59450)

2026-10-02 04:11:55 +00:00

kv_cache_layout.py

[CI] Ignore ruff D209, rejoin the docstrings it split, and silence incompatible-rule warnings (#57212)

2026-09-16 10:57:24 -07:00

kv_cache_spec_registry.py

Add pydocstyle to the ruff rules (#52136)

2026-09-16 16:44:29 +00:00

outputs.py

[Feature][Spec Decode] Support sampling mask replay for MRV2 MTP (#59359)

2026-10-01 13:51:39 +08:00

request.py

[Feature] Add first-class KV hints request envelope for programmatic KV management (#53423)

2026-09-21 09:56:11 -07:00

serial_utils.py

[Feature] Per-row candidate IDs for prefill token scoring (M2 of #56860) (#56984)

2026-10-03 14:17:09 +00:00

utils.py

[Rust Frontend] Expose gRPC port in Python vllm serve (#59659)

2026-10-02 08:14:37 +00:00