[AutoDiff][SILGen] Do not incorrectly rotate direct inout parameters
#58437
Conversation
This file contains 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
* Update SILGenPoly.cpp * Create sr15793-rotation-of-direct-inout-parameters.swift * Add validation test for SR-15793 * Change comment to say `Double` instead of `Float`
|
@swift-ci Please test. |
|
This feels like a hacky workaround. I'm not sure whether we have truly identified the source of the problem. Has the scenario causing the bug been thought of before? |
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Outdated suggestions cannot be applied.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 there is a mutating function nested inside an encapsulating function and both are differentiated, the compiler can swizzle around components of the gradient. This first manifested in a subtle runtime crash, but caused a compiler crash when it jumbled around gradient components with different Swift types.
The fix the bug, this PR avoids rotating direct parameters if the first one is "indirect mutating". I do not know whether we should use the same tactic with indirect parameters or when a direct parameter in a position besides index 0 is "indirect mutating".
The original reproducer for SR-15793 has been added to a validation test. The regression test is a new reproducer - which crashes the compiler - that I created while finding the solution to this bug.
Fixes #58070, formerly known as SR-15793.