fix: don't clobber saved frame pointer in arm64 assembly functions #170
+78
−0
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Outdated suggestions cannot be applied.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 arm64 neon assembly functions in this repository overwrite the frame
pointer saved by their callers, leading to crashes from the Go runtime
execution tracer and profilers which use frame pointer unwinding. For
historical reasons, on arm64 Go functions save the caller's frame
pointer register (x29) one word below their stack frame. See
go.dev/s/regabi#arm64-architecture. The assembly functions here,
translated from C compiler output, save values at the top of their
frame, and overwrite the frame pointer saved by the caller. We can fix
this by decrementing the stack pointer past where that frame pointer is
saved before saving anything on the stack.
Fixed with this sed script on my macos laptop + manual cleanup to match
indentation:
Ran the script from the root of this repository with
Then manually inspected the assembly for missing SUBs/ADDs at the
beginning of functions and prior to returns.
Fixes #150