[None][fix] disagg-ctx: detach kv_cache page-index buffers on early IndexMapper release - #14423
Merged
Tabrizian merged 1 commit intoMay 21, 2026
Conversation
…ndexMapper release KVCacheManagerV2.release_index_slot recycles the IndexMapper slot for a context-only request once prefill completes, while the kv_cache itself stays alive holding pages for the upcoming KV transfer. The kv_cache's _base_page_indices memoryviews still point into the host_kv_cache_block_offsets row identified by the freed slot. When a new request takes that slot, both kv_caches alias the same row; the old kv_cache's late page unlocks (SeqBlock dtors) then write BAD_PAGE_INDEX over the new request's live page indices, causing the DSv4 compressor scatter kernel to read -1 and IMA on the CTX worker. Detach the kv_cache's base-page-index buffers before removing the IndexMapper sequence so any subsequent writes from the old kv_cache land on a private array.array copy instead of the shared host row. Signed-off-by: Iman Tabrizian <10105175+tabrizian@users.noreply.github.com>
Member
Author
|
/bot run --disable-fail-fast |
Collaborator
|
PR_Github #49741 [ run ] triggered by Bot. Commit: |
pcastonguay
approved these changes
May 21, 2026
Collaborator
|
PR_Github #49741 [ run ] completed with state |
lfr-0531
pushed a commit
to lfr-0531/TensorRT-LLM
that referenced
this pull request
Jun 10, 2026
…ndexMapper release (NVIDIA#14423) Signed-off-by: Iman Tabrizian <10105175+tabrizian@users.noreply.github.com>
lfr-0531
pushed a commit
to lfr-0531/TensorRT-LLM
that referenced
this pull request
Jun 11, 2026
…ndexMapper release (NVIDIA#14423) Signed-off-by: Iman Tabrizian <10105175+tabrizian@users.noreply.github.com>
lfr-0531
pushed a commit
to lfr-0531/TensorRT-LLM
that referenced
this pull request
Jun 12, 2026
…ndexMapper release (NVIDIA#14423) Signed-off-by: Iman Tabrizian <10105175+tabrizian@users.noreply.github.com>
jiaganc
added a commit
to jiaganc/TensorRT-LLM
that referenced
this pull request
Jun 26, 2026
…ndexMapper release (NVIDIA#14423) Source-Commit: 02ac906 Signed-off-by: Jiagan Cheng <jiaganc@nvidia.com>
jiaganc
added a commit
to jiaganc/TensorRT-LLM
that referenced
this pull request
Jun 29, 2026
…ndexMapper release (NVIDIA#14423) Source-Commit: 02ac906 Signed-off-by: Jiagan Cheng <jiaganc@nvidia.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
…ndexMapper release
KVCacheManagerV2.release_index_slot recycles the IndexMapper slot for a context-only request once prefill completes, while the kv_cache itself stays alive holding pages for the upcoming KV transfer. The kv_cache's _base_page_indices memoryviews still point into the host_kv_cache_block_offsets row identified by the freed slot. When a new request takes that slot, both kv_caches alias the same row; the old kv_cache's late page unlocks (SeqBlock dtors) then write BAD_PAGE_INDEX over the new request's live page indices, causing the DSv4 compressor scatter kernel to read -1 and IMA on the CTX worker.
Detach the kv_cache's base-page-index buffers before removing the IndexMapper sequence so any subsequent writes from the old kv_cache land on a private array.array copy instead of the shared host row.
@coderabbitai summary
Description
Test Coverage
PR Checklist
Please review the following before submitting your PR:
PR description clearly explains what and why. If using CodeRabbit's summary, please make sure it makes sense.
PR Follows TRT-LLM CODING GUIDELINES to the best of your knowledge.
Test cases are provided for new code paths (see test instructions)
If PR introduces API changes, an appropriate PR label is added - either
api-compatibleorapi-breaking. Forapi-breaking, includeBREAKINGin the PR title.Any new dependencies have been scanned for license and vulnerabilities
CODEOWNERS updated if ownership changes
Documentation updated as needed
Update tava architecture diagram if there is a significant design change in PR.
The reviewers assigned automatically/manually are appropriate for the PR.
Please check this after reviewing the above items as appropriate for this PR.
GitHub Bot Help
To see a list of available CI bot commands, please comment
/bot help.