-
Notifications
You must be signed in to change notification settings - Fork 1.3k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
aarch64: Implement uextend/sextend for i128 values #3005
Merged
Merged
Changes from all commits
Commits
Show all changes
2 commits
Select commit
Hold shift + click to select a range
File filter
Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
There are no files selected for viewing
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
Original file line number | Diff line number | Diff line change |
---|---|---|
|
@@ -613,9 +613,30 @@ pub(crate) fn lower_insn_to_regs<C: LowerCtx<I = Inst>>( | |
let to_bits = ty_bits(output_ty) as u8; | ||
let to_bits = std::cmp::max(32, to_bits); | ||
assert!(from_bits <= to_bits); | ||
if from_bits < to_bits { | ||
let signed = op == Opcode::Sextend; | ||
let rd = get_output_reg(ctx, outputs[0]).only_reg().unwrap(); | ||
|
||
let signed = op == Opcode::Sextend; | ||
let dst = get_output_reg(ctx, outputs[0]); | ||
let src = | ||
if let Some(extract_insn) = maybe_input_insn(ctx, inputs[0], Opcode::Extractlane) { | ||
put_input_in_regs( | ||
ctx, | ||
InsnInput { | ||
insn: extract_insn, | ||
input: 0, | ||
}, | ||
) | ||
} else { | ||
put_input_in_regs(ctx, inputs[0]) | ||
}; | ||
|
||
let needs_extend = from_bits < to_bits && to_bits <= 64; | ||
// For i128, we want to extend the lower half, except if it is already 64 bits. | ||
let needs_lower_extend = to_bits > 64 && from_bits < 64; | ||
let pass_through_lower = to_bits > 64 && !needs_lower_extend; | ||
|
||
if needs_extend || needs_lower_extend { | ||
let rn = src.regs()[0]; | ||
let rd = dst.regs()[0]; | ||
|
||
if let Some(extract_insn) = maybe_input_insn(ctx, inputs[0], Opcode::Extractlane) { | ||
let idx = | ||
|
@@ -624,11 +645,7 @@ pub(crate) fn lower_insn_to_regs<C: LowerCtx<I = Inst>>( | |
} else { | ||
unreachable!(); | ||
}; | ||
let input = InsnInput { | ||
insn: extract_insn, | ||
input: 0, | ||
}; | ||
let rn = put_input_in_reg(ctx, input, NarrowValueMode::None); | ||
|
||
let size = VectorSize::from_ty(ctx.input_ty(extract_insn, 0)); | ||
|
||
if signed { | ||
|
@@ -654,8 +671,23 @@ pub(crate) fn lower_insn_to_regs<C: LowerCtx<I = Inst>>( | |
rn, | ||
signed, | ||
from_bits, | ||
to_bits, | ||
to_bits: std::cmp::min(64, to_bits), | ||
}); | ||
} | ||
} else if pass_through_lower { | ||
ctx.emit(Inst::gen_move(dst.regs()[0], src.regs()[0], I64)); | ||
} | ||
|
||
There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. With the above |
||
if output_ty == I128 { | ||
if signed { | ||
ctx.emit(Inst::AluRRImmShift { | ||
alu_op: ALUOp::Asr64, | ||
rd: dst.regs()[1], | ||
rn: dst.regs()[0].to_reg(), | ||
immshift: ImmShift::maybe_from_u64(63).unwrap(), | ||
}); | ||
} else { | ||
lower_constant_u64(ctx, dst.regs()[1], 0); | ||
} | ||
} | ||
} | ||
|
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
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
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,69 @@ | ||
test run | ||
target aarch64 | ||
target x86_64 machinst | ||
; TODO: Merge this file with i128-extend once the x86 legacy backend is removed | ||
|
||
function %i128_uextend_i32(i32) -> i64, i64 { | ||
block0(v0: i32): | ||
v1 = uextend.i128 v0 | ||
v2, v3 = isplit v1 | ||
return v2, v3 | ||
} | ||
; run: %i128_uextend_i32(0) == [0, 0] | ||
; run: %i128_uextend_i32(-1) == [0xffff_ffff, 0] | ||
; run: %i128_uextend_i32(0xffff_eeee) == [0xffff_eeee, 0] | ||
|
||
function %i128_sextend_i32(i32) -> i64, i64 { | ||
block0(v0: i32): | ||
v1 = sextend.i128 v0 | ||
v2, v3 = isplit v1 | ||
return v2, v3 | ||
} | ||
; run: %i128_sextend_i32(0) == [0, 0] | ||
; run: %i128_sextend_i32(-1) == [-1, -1] | ||
; run: %i128_sextend_i32(0x7fff_ffff) == [0x7fff_ffff, 0x0000_0000_0000_0000] | ||
; run: %i128_sextend_i32(0xffff_eeee) == [0xffff_ffff_ffff_eeee, 0xffff_ffff_ffff_ffff] | ||
|
||
|
||
function %i128_uextend_i16(i16) -> i64, i64 { | ||
block0(v0: i16): | ||
v1 = uextend.i128 v0 | ||
v2, v3 = isplit v1 | ||
return v2, v3 | ||
} | ||
; run: %i128_uextend_i16(0) == [0, 0] | ||
; run: %i128_uextend_i16(-1) == [0xffff, 0] | ||
; run: %i128_uextend_i16(0xffee) == [0xffee, 0] | ||
|
||
function %i128_sextend_i16(i16) -> i64, i64 { | ||
block0(v0: i16): | ||
v1 = sextend.i128 v0 | ||
v2, v3 = isplit v1 | ||
return v2, v3 | ||
} | ||
; run: %i128_sextend_i16(0) == [0, 0] | ||
; run: %i128_sextend_i16(-1) == [-1, -1] | ||
; run: %i128_sextend_i16(0x7fff) == [0x7fff, 0x0000_0000_0000_0000] | ||
; run: %i128_sextend_i16(0xffee) == [0xffff_ffff_ffff_ffee, 0xffff_ffff_ffff_ffff] | ||
|
||
|
||
function %i128_uextend_i8(i8) -> i64, i64 { | ||
block0(v0: i8): | ||
v1 = uextend.i128 v0 | ||
v2, v3 = isplit v1 | ||
return v2, v3 | ||
} | ||
; run: %i128_uextend_i8(0) == [0, 0] | ||
; run: %i128_uextend_i8(-1) == [0xff, 0] | ||
; run: %i128_uextend_i8(0xfe) == [0xfe, 0] | ||
|
||
function %i128_sextend_i8(i8) -> i64, i64 { | ||
block0(v0: i8): | ||
v1 = sextend.i128 v0 | ||
v2, v3 = isplit v1 | ||
return v2, v3 | ||
} | ||
; run: %i128_sextend_i8(0) == [0, 0] | ||
; run: %i128_sextend_i8(-1) == [-1, -1] | ||
; run: %i128_sextend_i8(0x7f) == [0x7f, 0x0000_0000_0000_0000] | ||
; run: %i128_sextend_i8(0xfe) == [0xffff_ffff_ffff_fffe, 0xffff_ffff_ffff_ffff] |
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
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -1,29 +1,26 @@ | ||
test run | ||
; target aarch64 TODO: Not yet implemented on aarch64 | ||
target aarch64 | ||
; target s390x TODO: Not yet implemented on s390x | ||
target x86_64 machinst | ||
target x86_64 legacy | ||
|
||
function %i128_uextend() -> b1 { | ||
block0: | ||
v0 = iconst.i64 0xffff_ffff_eeee_0000 | ||
function %i128_uextend_i64(i64) -> i64, i64 { | ||
block0(v0: i64): | ||
v1 = uextend.i128 v0 | ||
v2, v3 = isplit v1 | ||
v4 = icmp_imm eq v2, 0xffff_ffff_eeee_0000 | ||
v5 = icmp_imm eq v3, 0 | ||
v6 = band v4, v5 | ||
return v6 | ||
return v2, v3 | ||
} | ||
; run | ||
; run: %i128_uextend_i64(0) == [0, 0] | ||
; run: %i128_uextend_i64(-1) == [-1, 0] | ||
; run: %i128_uextend_i64(0xffff_ffff_eeee_0000) == [0xffff_ffff_eeee_0000, 0] | ||
|
||
function %i128_sextend() -> b1 { | ||
block0: | ||
v0 = iconst.i64 0xffff_ffff_eeee_0000 | ||
function %i128_sextend_i64(i64) -> i64, i64 { | ||
block0(v0: i64): | ||
v1 = sextend.i128 v0 | ||
v2, v3 = isplit v1 | ||
v4 = icmp_imm eq v2, 0xffff_ffff_eeee_0000 | ||
v5 = icmp_imm eq v3, 0xffff_ffff_ffff_ffff | ||
v6 = band v4, v5 | ||
return v6 | ||
return v2, v3 | ||
} | ||
; run | ||
; run: %i128_sextend_i64(0) == [0, 0] | ||
; run: %i128_sextend_i64(-1) == [-1, -1] | ||
; run: %i128_sextend_i64(0x7fff_ffff_ffff_ffff) == [0x7fff_ffff_ffff_ffff, 0x0000_0000_0000_0000] | ||
; run: %i128_sextend_i64(0xffff_ffff_eeee_0000) == [0xffff_ffff_eeee_0000, 0xffff_ffff_ffff_ffff] |
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.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Can we clarify the logic a bit by adding a
let pass_through_lower = to_bits > 64 && !needs_lower_extend
here, then moving the move-emission below (will comment separately)?