Rewrite read_u32v_be() and write_u32_be() to make sure that all memory access is properly aligned #16191
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
…y access is properly aligned
bors
added a commit
that referenced
this pull request
Aug 3, 2014
When I originally wrote the read_u32v_be() and write_u32_be() functions, I didn't consider memory alignment requirements of various architectures. Unfortunately, the current implementations may result in unaligned reads and writes. This doesn't impact x86 / x86_64, but it can cause a compiler crash on ARM. This pull requests rewrites those functions to make sure that all memory access is always correctly aligned. This fix is a little bit academic - due to the way that LLVM aligns the structures that are passed as arguments to these functions, I believe that the end result is that all memory access happens to be aligned anyway. However, there is nothing in that code that actually enforces that, at least not explicitly. The new implementations are definitely slower than the existing ones. However, I don't believe that these functions are all that significant when looking at the overall performance of the compiler. I think getting rid of some unsafe code and removing a potential portability landmine justifies a very slight decrease in raw performance.
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Outdated suggestions cannot be applied.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.
When I originally wrote the read_u32v_be() and write_u32_be() functions, I didn't consider memory alignment requirements of various architectures. Unfortunately, the current implementations may result in unaligned reads and writes. This doesn't impact x86 / x86_64, but it can cause a compiler crash on ARM. This pull requests rewrites those functions to make sure that all memory access is always correctly aligned.
This fix is a little bit academic - due to the way that LLVM aligns the structures that are passed as arguments to these functions, I believe that the end result is that all memory access happens to be aligned anyway. However, there is nothing in that code that actually enforces that, at least not explicitly. The new implementations are definitely slower than the existing ones. However, I don't believe that these functions are all that significant when looking at the overall performance of the compiler. I think getting rid of some unsafe code and removing a potential portability landmine justifies a very slight decrease in raw performance.