[xnnpack] Add convolution dimension fuzzer for adversarial operator parameters - #16095
Open
smoke-wolf wants to merge 1 commit into
Open
[xnnpack] Add convolution dimension fuzzer for adversarial operator parameters#16095smoke-wolf wants to merge 1 commit into
smoke-wolf wants to merge 1 commit into
Conversation
…arameters Add fuzz_conv_dims target that exercises xnn_create_convolution2d_nhwc_qs8 with fuzzer-generated kernel dimensions, channel counts, and stride parameters. This complements the existing fuzz_model target which uses hardcoded small dimensions (3x3 kernel, 3 input channels). The new target covers: - Large kernel dimensions (up to 65536x65536) - Variable group/channel counts - Adversarial padding/stride/dilation combinations These parameter ranges exercise integer overflow paths in the operator reshape functions that the existing fixed-dimension fuzzer cannot reach.
|
smoke-wolf is a new contributor to projects/xnnpack. The PR must be approved by known contributors before it can be merged. The past contributors are: arthurscchan, hunsche |
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Add
fuzz_conv_dimstarget for the XNNPACK project that exercisesxnn_create_convolution2d_nhwc_qs8with fuzzer-generated kerneldimensions, channel counts, and stride/padding/dilation values.
The existing
fuzz_modeltarget uses hardcoded small dimensions(3×3 kernel, 3 input channels, 32 output channels). This new target
generates dimensions across the full valid range (kernel up to 65536,
channels up to 65536), covering integer overflow paths in the operator
create and reshape functions that fixed-dimension fuzzing cannot reach.
Test plan
fuzz_modeltarget