jameysharp requested cfallin for a review on PR #8449.
jameysharp requested wasmtime-compiler-reviewers for a review on PR #8449.
jameysharp opened PR #8449 from jameysharp:s390-pregs
to bytecodealliance:main
:
When OperandCollector's reg_use/reg_late_use/reg_def/reg_early_def methods are handed a Reg that refers to a physical ("real") register, they all delegate to reg_fixed_nonallocatable, ignoring the constraint kinds and positions.
In several cases, the s390x backend was calling those methods with the result of the
gpr
orwritable_gpr
functions, which return physical registers. In these cases we can be more explicit that this is a non-allocatable register.This doesn't change any behavior but is a nice change to split out of a larger PR I'm working on.
cc: @uweigand
jameysharp updated PR #8449.
jameysharp edited PR #8449:
When OperandCollector's reg_use/reg_late_use/reg_def/reg_early_def methods are handed a Reg that refers to a physical ("real") register, they all delegate to reg_fixed_nonallocatable, ignoring the constraint kinds and positions. This behavior was introduced in #5132.
In several cases, the s390x backend was calling those methods with the result of the
gpr
orwritable_gpr
functions, which return physical registers. In these cases we can be more explicit that this is a non-allocatable register.In addition, this PR reverts #4973 and #5121 because they became unecessary due, again, to #5132.
This doesn't change any behavior but is a nice change to split out of a larger PR I'm working on.
cc: @uweigand
cfallin submitted PR review:
LGTM, thanks!
jameysharp merged PR #8449.
Last updated: Dec 23 2024 at 12:05 UTC