Stream: git-wasmtime

Topic: wasmtime / Issue #1342 Avoid `infer_rex` and `w` on the s...

Wasmtime GitHub notifications bot (Mar 17 2020 at 16:28):

abrown opened Issue #1342:

<!-- Please try to describe precisely what you would like to do in
Cranelift/Wasmtime and/or expect from it. You can answer the questions below if
they're relevant and delete this text before submitting. Thanks for opening an
issue! -->

Feature

Disallow infer_rex and w on the same encoding template

<!-- What is the feature or code improvement you would like to do in
Cranelift/Wasmtime? -->

Benefit

In cranelift x86 encodings, it seems unintuitive that we can specialize Templates with both infer_rex and w: if w is specified, the REX.W bit must be set so a REX prefix is required. In this case, it seems more clear to force ourselves to write rex().w()--it's more explicit and shows more clearly what cranelift will emit.
<!-- What is the value of adding this in Cranelift/Wasmtime? -->

Implementation

<!-- Do you have an implementation plan, and/or ideas for data structures or
algorithms to use? -->

• In the RecipePrefixKind::InferRex arm of Template::build, add an assertion to prevent the use of w(). E.g. assert!(!self.w_bit, "No need to infer REX prefix when REX.W is specified, use rex() instead")
• Optionally remove the runtime check for the REX.W bit in needs_rex (binemit.rs) and the size calculator functions in enc_tables.rs; this probably has some negligible runtime improvement and could be replaced by a comment indicating that the REX.W check has been performed statically.

Alternatives

<!-- Have you considered alternative implementations? If so, how are they
better or worse than your proposal? -->
Do nothing.

Wasmtime GitHub notifications bot (Mar 17 2020 at 16:29):

abrown edited Issue #1342:

<!-- Please try to describe precisely what you would like to do in
Cranelift/Wasmtime and/or expect from it. You can answer the questions below if
they're relevant and delete this text before submitting. Thanks for opening an
issue! -->

Feature

Disallow infer_rex and w on the same encoding template

<!-- What is the feature or code improvement you would like to do in
Cranelift/Wasmtime? -->

Benefit

In cranelift x86 encodings, it seems unintuitive that we can specialize Templates with both infer_rex and w: if w is specified, the REX.W bit must be set so a REX prefix is alway required--no need to infer it. In this case, it seems more clear to force ourselves to write rex().w()--it's more explicit and shows more clearly what cranelift will emit.
<!-- What is the value of adding this in Cranelift/Wasmtime? -->

Implementation

<!-- Do you have an implementation plan, and/or ideas for data structures or
algorithms to use? -->

• In the RecipePrefixKind::InferRex arm of Template::build, add an assertion to prevent the use of w(). E.g. assert!(!self.w_bit, "No need to infer REX prefix when REX.W is specified, use rex() instead")
• Optionally remove the runtime check for the REX.W bit in needs_rex (binemit.rs) and the size calculator functions in enc_tables.rs; this probably has some negligible runtime improvement and could be replaced by a comment indicating that the REX.W check has been performed statically.

Alternatives

<!-- Have you considered alternative implementations? If so, how are they
better or worse than your proposal? -->
Do nothing.

Wasmtime GitHub notifications bot (Mar 18 2020 at 17:17):

abrown labeled Issue #1342:

<!-- Please try to describe precisely what you would like to do in
Cranelift/Wasmtime and/or expect from it. You can answer the questions below if
they're relevant and delete this text before submitting. Thanks for opening an
issue! -->

Feature

Disallow infer_rex and w on the same encoding template

<!-- What is the feature or code improvement you would like to do in
Cranelift/Wasmtime? -->

Benefit

In cranelift x86 encodings, it seems unintuitive that we can specialize Templates with both infer_rex and w: if w is specified, the REX.W bit must be set so a REX prefix is alway required--no need to infer it. In this case, it seems more clear to force ourselves to write rex().w()--it's more explicit and shows more clearly what cranelift will emit.
<!-- What is the value of adding this in Cranelift/Wasmtime? -->

Implementation

<!-- Do you have an implementation plan, and/or ideas for data structures or
algorithms to use? -->

• In the RecipePrefixKind::InferRex arm of Template::build, add an assertion to prevent the use of w(). E.g. assert!(!self.w_bit, "No need to infer REX prefix when REX.W is specified, use rex() instead")
• Optionally remove the runtime check for the REX.W bit in needs_rex (binemit.rs) and the size calculator functions in enc_tables.rs; this probably has some negligible runtime improvement and could be replaced by a comment indicating that the REX.W check has been performed statically.

Alternatives

<!-- Have you considered alternative implementations? If so, how are they
better or worse than your proposal? -->
Do nothing.

Wasmtime GitHub notifications bot (Apr 02 2020 at 23:50):

abrown closed Issue #1342:

<!-- Please try to describe precisely what you would like to do in
Cranelift/Wasmtime and/or expect from it. You can answer the questions below if
they're relevant and delete this text before submitting. Thanks for opening an
issue! -->

Feature

Disallow infer_rex and w on the same encoding template

<!-- What is the feature or code improvement you would like to do in
Cranelift/Wasmtime? -->

Benefit

In cranelift x86 encodings, it seems unintuitive that we can specialize Templates with both infer_rex and w: if w is specified, the REX.W bit must be set so a REX prefix is alway required--no need to infer it. In this case, it seems more clear to force ourselves to write rex().w()--it's more explicit and shows more clearly what cranelift will emit.
<!-- What is the value of adding this in Cranelift/Wasmtime? -->

Implementation

<!-- Do you have an implementation plan, and/or ideas for data structures or
algorithms to use? -->

• In the RecipePrefixKind::InferRex arm of Template::build, add an assertion to prevent the use of w(). E.g. assert!(!self.w_bit, "No need to infer REX prefix when REX.W is specified, use rex() instead")
• Optionally remove the runtime check for the REX.W bit in needs_rex (binemit.rs) and the size calculator functions in enc_tables.rs; this probably has some negligible runtime improvement and could be replaced by a comment indicating that the REX.W check has been performed statically.

Alternatives

<!-- Have you considered alternative implementations? If so, how are they
better or worse than your proposal? -->
Do nothing.

Last updated: Dec 23 2024 at 12:05 UTC