Add dimension check to prevent out-of-bounds access in example 05_bmg_gemm_with_epilogue_splitk #529
Conversation
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
Signed-off-by: Chen, Xi2 <[email protected]>
tdeng5
approved these changes
Sep 26, 2025
taozha2
reviewed
Sep 29, 2025
Comment on lines
+222
to
+224
| std::cout << "Error: n < num_head * (nope_dim + rope_dim). Please set a sufficiently large value for n. Skipping the check." << std::endl; | ||
| return true; | ||
| } |
There was a problem hiding this comment.
put this parameters checking before launching the kernel?
There was a problem hiding this comment.
And, should we skip the validation of these out-of-range data or never let the kernel launching?
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.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.
Adds a safety check in the 05_bmg_gemm_with_epilogue_splitk.cpp to handle cases where the N dimension is insufficient for the split-K fusion logic.
Without this check, small N values can lead to out-of-bounds memory access in this line :
This is because N is at least NUM_HEAD * (NOPE_DIM + ROPE_DIM) to properly split the output into D1 and D2 tensors. If a user specifies a smaller N (e.g., via command-line arguments like --n=128), the loop attempts to access indices beyond the bounds of the D array, resulting in a segmentation fault (core dump).