[Arc] Consider array/struct field alignment when computing state size #8872
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
TaoBi22
approved these changes
Aug 19, 2025
maerhart
reviewed
Aug 19, 2025
60ecc21 to
254c7bf
Compare
cec2d0e to
2afcf79
Compare
254c7bf to
cb523ee
Compare
2afcf79 to
ea7d3b0
Compare
The `StateType` in the Arc dialect currently just defers to `hw::getBitWidth` to compute the number of bits that need to be allocated in the simulator state. This is problematic, since types like structs and arrays have field and element alignment requirements, making LLVM require more bytes for these types than the summed-up bit width would suggest. As a result, the AllocateState pass would allocate 2 bytes for `struct<tag: i5, sign_ext: i1, offset: i3, size: i2>`, even though LLVM would consider the individual fields to be aligned to the nearest byte, making the struct cover 4 bytes in memory. To fix this issue, make `StateType` more carefully compute the actual number of bits required in simulator storage by requiring all elements of a array or struct to be at least byte-aligned, and to more carefully track the alignmenmt of individual fields. This allows us to predict how many bytes more accurately. This fixes a few nasty simulation mismatches between Arcilator and Verilator due to the memory corruption that would result from not allocating sufficient storage for arrays and structs.
ea7d3b0 to
cf148e4
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 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Outdated suggestions cannot be applied.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.
The
StateTypein the Arc dialect currently just defers tohw::getBitWidthto compute the number of bits that need to be allocated in the simulator state. This is problematic, since types like structs and arrays have field and element alignment requirements, making LLVM require more bytes for these types than the summed-up bit width would suggest. As a result, the AllocateState pass would allocate 2 bytes forstruct<tag: i5, sign_ext: i1, offset: i3, size: i2>, even though LLVM would consider the individual fields to be aligned to the nearest byte, making the struct cover 4 bytes in memory.To fix this issue, make
StateTypemore carefully compute the actual number of bits required in simulator storage by requiring all elements of a array or struct to be at least byte-aligned, and to more carefully track the alignmenmt of individual fields. This allows us to predict how many bytes more accurately.This fixes a few nasty simulation mismatches between Arcilator and Verilator due to the memory corruption that would result from not allocating sufficient storage for arrays and structs.