fix: Replace polynomial rolling hash with SHA-256 for prefix caching - #4158
Merged
lmcafee-nvidia merged 3 commits intoApr 25, 2026
Merged
Conversation
The polynomial rolling hash had exploitable algebraic structure where token[j] += HASH_BASE, token[j+1] -= 1 produces a zero-delta, enabling cache poisoning in multi-tenant serving. SHA-256 eliminates this class of attacks with cryptographic collision resistance. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
santhnm2
approved these changes
Apr 16, 2026
shanmugamr1992
approved these changes
Apr 24, 2026
kvareddy
approved these changes
Apr 24, 2026
svcnvidia-nemo-ci
added
Approved
Contributor
Author
|
/ok to test 46c4e6d |
Contributor
|
🔄 Merge queue validation started! You can track the progress here: https://github.com/NVIDIA/Megatron-LM/actions/runs/24912702741 |
Contributor
|
🔄 Merge queue validation started! You can track the progress here: https://github.com/NVIDIA/Megatron-LM/actions/runs/24914628506 |
Contributor
|
🔄 Merge queue validation started! You can track the progress here: https://github.com/NVIDIA/Megatron-LM/actions/runs/24916866638 |
chtruong814
added
the
core_r0.17.0
5 tasks
yangbofun
pushed a commit
to xlm-research/Megatron-LM
that referenced
this pull request
May 22, 2026
…VIDIA#4158) Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
yhgalaxy
pushed a commit
to yhgalaxy/Megatron-LM
that referenced
this pull request
Jun 17, 2026
…VIDIA#4158) Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com> Signed-off-by: yhgalaxy <yhgalaxy@outlook.com>
jon-barker
pushed a commit
to jon-barker/Megatron-LM
that referenced
this pull request
Jul 10, 2026
…VIDIA#4158) Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com> Signed-off-by: Jon Barker <jbarker@aws-cmh-slurm-1-vscode-02.cm.cluster>
terminator123
pushed a commit
to 021ai/Megatron-LM
that referenced
this pull request
Aug 3, 2026
…VIDIA#4158) Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.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
6 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.
Summary
compute_block_hashes_batched()with SHA-256 to eliminate an algebraic collision vulnerability (token[j] += 31, token[j+1] -= 1produced zero-delta in the old hash)Test plan
test_hash_collision_resistance— verifies the old V2 algebraic attack no longer produces collisionstest_hash_computation— existing test updated to removeHASH_PRIMEreference/opt/venv/bin/python -m torch.distributed.run --nproc-per-node 1 -m pytest tests/unit_tests/inference/contexts/test_dynamic_prefix_caching.py -v🤖 Generated with Claude Code