2020-06-27 19:26:30 +08:00
|
|
|
// only-wasm32
|
|
|
|
|
// compile-flags: -C target-feature=-nontrapping-fptoint
|
|
|
|
|
#![crate_type = "lib"]
|
|
|
|
|
|
|
|
|
|
// CHECK-LABEL: @cast_f64_i64
|
|
|
|
|
#[no_mangle]
|
|
|
|
|
pub fn cast_f64_i64(a: f64) -> i64 {
|
rustc: Improving safe wasm float->int casts
This commit improves code generation for WebAssembly targets when
translating floating to integer casts. This improvement is only relevant
when the `nontrapping-fptoint` feature is not enabled, but the feature
is not enabled by default right now. Additionally this improvement only
affects safe casts since unchecked casts were improved in #74659.
Some more background for this issue is present on #73591, but the
general gist of the issue is that in LLVM the `fptosi` and `fptoui`
instructions are defined to return an `undef` value if they execute on
out-of-bounds values; they notably do not trap. To implement these
instructions for WebAssembly the LLVM backend must therefore generate
quite a few instructions before executing `i32.trunc_f32_s` (for
example) because this WebAssembly instruction traps on out-of-bounds
values. This codegen into wasm instructions happens very late in the
code generator, so what ends up happening is that rustc inserts its own
codegen to implement Rust's saturating semantics, and then LLVM also
inserts its own codegen to make sure that the `fptosi` instruction
doesn't trap. Overall this means that a function like this:
#[no_mangle]
pub unsafe extern "C" fn cast(x: f64) -> u32 {
x as u32
}
will generate this WebAssembly today:
(func $cast (type 0) (param f64) (result i32)
(local i32 i32)
local.get 0
f64.const 0x1.fffffffep+31 (;=4.29497e+09;)
f64.gt
local.set 1
block ;; label = @1
block ;; label = @2
local.get 0
f64.const 0x0p+0 (;=0;)
local.get 0
f64.const 0x0p+0 (;=0;)
f64.gt
select
local.tee 0
f64.const 0x1p+32 (;=4.29497e+09;)
f64.lt
local.get 0
f64.const 0x0p+0 (;=0;)
f64.ge
i32.and
i32.eqz
br_if 0 (;@2;)
local.get 0
i32.trunc_f64_u
local.set 2
br 1 (;@1;)
end
i32.const 0
local.set 2
end
i32.const -1
local.get 2
local.get 1
select)
This PR improves the situation by updating the code generation for
float-to-int conversions in rustc, specifically only for WebAssembly
targets and only for some situations (float-to-u8 still has not great
codegen). The fix here is to use basic blocks and control flow to avoid
speculatively executing `fptosi`, and instead LLVM's raw intrinsic for
the WebAssembly instruction is used instead. This effectively extends
the support added in #74659 to checked casts. After this commit the
codegen for the above Rust function looks like:
(func $cast (type 0) (param f64) (result i32)
(local i32)
block ;; label = @1
local.get 0
f64.const 0x0p+0 (;=0;)
f64.ge
local.tee 1
i32.const 1
i32.xor
br_if 0 (;@1;)
local.get 0
f64.const 0x1.fffffffep+31 (;=4.29497e+09;)
f64.le
i32.eqz
br_if 0 (;@1;)
local.get 0
i32.trunc_f64_u
return
end
i32.const -1
i32.const 0
local.get 1
select)
For reference, in Rust 1.44, which did not have saturating
float-to-integer casts, the codegen LLVM would emit is:
(func $cast (type 0) (param f64) (result i32)
block ;; label = @1
local.get 0
f64.const 0x1p+32 (;=4.29497e+09;)
f64.lt
local.get 0
f64.const 0x0p+0 (;=0;)
f64.ge
i32.and
i32.eqz
br_if 0 (;@1;)
local.get 0
i32.trunc_f64_u
return
end
i32.const 0)
So we're relatively close to the original codegen, although it's
slightly different because the semantics of the function changed where
we're emulating the `i32.trunc_sat_f32_s` instruction rather than always
replacing out-of-bounds values with zero.
There is still work that could be done to improve casts such as `f32` to
`u8`. That form of cast still uses the `fptosi` instruction which
generates lots of branch-y code. This seems less important to tackle now
though. In the meantime this should take care of most use cases of
floating-point conversion and as a result I'm going to speculate that
this...
Closes #73591
2020-07-24 03:20:42 +08:00
|
|
|
// CHECK-NOT: fptosi double {{.*}} to i64
|
|
|
|
|
// CHECK-NOT: select i1 {{.*}}, i64 {{.*}}, i64 {{.*}}
|
2021-04-23 23:59:49 +08:00
|
|
|
// CHECK: {{.*}} call {{.*}} @llvm.fptosi.sat.i64.f64{{.*}}
|
2020-06-27 19:26:30 +08:00
|
|
|
a as _
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// CHECK-LABEL: @cast_f64_i32
|
|
|
|
|
#[no_mangle]
|
|
|
|
|
pub fn cast_f64_i32(a: f64) -> i32 {
|
rustc: Improving safe wasm float->int casts
This commit improves code generation for WebAssembly targets when
translating floating to integer casts. This improvement is only relevant
when the `nontrapping-fptoint` feature is not enabled, but the feature
is not enabled by default right now. Additionally this improvement only
affects safe casts since unchecked casts were improved in #74659.
Some more background for this issue is present on #73591, but the
general gist of the issue is that in LLVM the `fptosi` and `fptoui`
instructions are defined to return an `undef` value if they execute on
out-of-bounds values; they notably do not trap. To implement these
instructions for WebAssembly the LLVM backend must therefore generate
quite a few instructions before executing `i32.trunc_f32_s` (for
example) because this WebAssembly instruction traps on out-of-bounds
values. This codegen into wasm instructions happens very late in the
code generator, so what ends up happening is that rustc inserts its own
codegen to implement Rust's saturating semantics, and then LLVM also
inserts its own codegen to make sure that the `fptosi` instruction
doesn't trap. Overall this means that a function like this:
#[no_mangle]
pub unsafe extern "C" fn cast(x: f64) -> u32 {
x as u32
}
will generate this WebAssembly today:
(func $cast (type 0) (param f64) (result i32)
(local i32 i32)
local.get 0
f64.const 0x1.fffffffep+31 (;=4.29497e+09;)
f64.gt
local.set 1
block ;; label = @1
block ;; label = @2
local.get 0
f64.const 0x0p+0 (;=0;)
local.get 0
f64.const 0x0p+0 (;=0;)
f64.gt
select
local.tee 0
f64.const 0x1p+32 (;=4.29497e+09;)
f64.lt
local.get 0
f64.const 0x0p+0 (;=0;)
f64.ge
i32.and
i32.eqz
br_if 0 (;@2;)
local.get 0
i32.trunc_f64_u
local.set 2
br 1 (;@1;)
end
i32.const 0
local.set 2
end
i32.const -1
local.get 2
local.get 1
select)
This PR improves the situation by updating the code generation for
float-to-int conversions in rustc, specifically only for WebAssembly
targets and only for some situations (float-to-u8 still has not great
codegen). The fix here is to use basic blocks and control flow to avoid
speculatively executing `fptosi`, and instead LLVM's raw intrinsic for
the WebAssembly instruction is used instead. This effectively extends
the support added in #74659 to checked casts. After this commit the
codegen for the above Rust function looks like:
(func $cast (type 0) (param f64) (result i32)
(local i32)
block ;; label = @1
local.get 0
f64.const 0x0p+0 (;=0;)
f64.ge
local.tee 1
i32.const 1
i32.xor
br_if 0 (;@1;)
local.get 0
f64.const 0x1.fffffffep+31 (;=4.29497e+09;)
f64.le
i32.eqz
br_if 0 (;@1;)
local.get 0
i32.trunc_f64_u
return
end
i32.const -1
i32.const 0
local.get 1
select)
For reference, in Rust 1.44, which did not have saturating
float-to-integer casts, the codegen LLVM would emit is:
(func $cast (type 0) (param f64) (result i32)
block ;; label = @1
local.get 0
f64.const 0x1p+32 (;=4.29497e+09;)
f64.lt
local.get 0
f64.const 0x0p+0 (;=0;)
f64.ge
i32.and
i32.eqz
br_if 0 (;@1;)
local.get 0
i32.trunc_f64_u
return
end
i32.const 0)
So we're relatively close to the original codegen, although it's
slightly different because the semantics of the function changed where
we're emulating the `i32.trunc_sat_f32_s` instruction rather than always
replacing out-of-bounds values with zero.
There is still work that could be done to improve casts such as `f32` to
`u8`. That form of cast still uses the `fptosi` instruction which
generates lots of branch-y code. This seems less important to tackle now
though. In the meantime this should take care of most use cases of
floating-point conversion and as a result I'm going to speculate that
this...
Closes #73591
2020-07-24 03:20:42 +08:00
|
|
|
// CHECK-NOT: fptosi double {{.*}} to i32
|
|
|
|
|
// CHECK-NOT: select i1 {{.*}}, i32 {{.*}}, i32 {{.*}}
|
2021-04-23 23:59:49 +08:00
|
|
|
// CHECK: {{.*}} call {{.*}} @llvm.fptosi.sat.i32.f64{{.*}}
|
2020-06-27 19:26:30 +08:00
|
|
|
a as _
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// CHECK-LABEL: @cast_f32_i64
|
|
|
|
|
#[no_mangle]
|
|
|
|
|
pub fn cast_f32_i64(a: f32) -> i64 {
|
rustc: Improving safe wasm float->int casts
This commit improves code generation for WebAssembly targets when
translating floating to integer casts. This improvement is only relevant
when the `nontrapping-fptoint` feature is not enabled, but the feature
is not enabled by default right now. Additionally this improvement only
affects safe casts since unchecked casts were improved in #74659.
Some more background for this issue is present on #73591, but the
general gist of the issue is that in LLVM the `fptosi` and `fptoui`
instructions are defined to return an `undef` value if they execute on
out-of-bounds values; they notably do not trap. To implement these
instructions for WebAssembly the LLVM backend must therefore generate
quite a few instructions before executing `i32.trunc_f32_s` (for
example) because this WebAssembly instruction traps on out-of-bounds
values. This codegen into wasm instructions happens very late in the
code generator, so what ends up happening is that rustc inserts its own
codegen to implement Rust's saturating semantics, and then LLVM also
inserts its own codegen to make sure that the `fptosi` instruction
doesn't trap. Overall this means that a function like this:
#[no_mangle]
pub unsafe extern "C" fn cast(x: f64) -> u32 {
x as u32
}
will generate this WebAssembly today:
(func $cast (type 0) (param f64) (result i32)
(local i32 i32)
local.get 0
f64.const 0x1.fffffffep+31 (;=4.29497e+09;)
f64.gt
local.set 1
block ;; label = @1
block ;; label = @2
local.get 0
f64.const 0x0p+0 (;=0;)
local.get 0
f64.const 0x0p+0 (;=0;)
f64.gt
select
local.tee 0
f64.const 0x1p+32 (;=4.29497e+09;)
f64.lt
local.get 0
f64.const 0x0p+0 (;=0;)
f64.ge
i32.and
i32.eqz
br_if 0 (;@2;)
local.get 0
i32.trunc_f64_u
local.set 2
br 1 (;@1;)
end
i32.const 0
local.set 2
end
i32.const -1
local.get 2
local.get 1
select)
This PR improves the situation by updating the code generation for
float-to-int conversions in rustc, specifically only for WebAssembly
targets and only for some situations (float-to-u8 still has not great
codegen). The fix here is to use basic blocks and control flow to avoid
speculatively executing `fptosi`, and instead LLVM's raw intrinsic for
the WebAssembly instruction is used instead. This effectively extends
the support added in #74659 to checked casts. After this commit the
codegen for the above Rust function looks like:
(func $cast (type 0) (param f64) (result i32)
(local i32)
block ;; label = @1
local.get 0
f64.const 0x0p+0 (;=0;)
f64.ge
local.tee 1
i32.const 1
i32.xor
br_if 0 (;@1;)
local.get 0
f64.const 0x1.fffffffep+31 (;=4.29497e+09;)
f64.le
i32.eqz
br_if 0 (;@1;)
local.get 0
i32.trunc_f64_u
return
end
i32.const -1
i32.const 0
local.get 1
select)
For reference, in Rust 1.44, which did not have saturating
float-to-integer casts, the codegen LLVM would emit is:
(func $cast (type 0) (param f64) (result i32)
block ;; label = @1
local.get 0
f64.const 0x1p+32 (;=4.29497e+09;)
f64.lt
local.get 0
f64.const 0x0p+0 (;=0;)
f64.ge
i32.and
i32.eqz
br_if 0 (;@1;)
local.get 0
i32.trunc_f64_u
return
end
i32.const 0)
So we're relatively close to the original codegen, although it's
slightly different because the semantics of the function changed where
we're emulating the `i32.trunc_sat_f32_s` instruction rather than always
replacing out-of-bounds values with zero.
There is still work that could be done to improve casts such as `f32` to
`u8`. That form of cast still uses the `fptosi` instruction which
generates lots of branch-y code. This seems less important to tackle now
though. In the meantime this should take care of most use cases of
floating-point conversion and as a result I'm going to speculate that
this...
Closes #73591
2020-07-24 03:20:42 +08:00
|
|
|
// CHECK-NOT: fptosi float {{.*}} to i64
|
|
|
|
|
// CHECK-NOT: select i1 {{.*}}, i64 {{.*}}, i64 {{.*}}
|
2021-04-23 23:59:49 +08:00
|
|
|
// CHECK: {{.*}} call {{.*}} @llvm.fptosi.sat.i64.f32{{.*}}
|
2020-06-27 19:26:30 +08:00
|
|
|
a as _
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// CHECK-LABEL: @cast_f32_i32
|
|
|
|
|
#[no_mangle]
|
|
|
|
|
pub fn cast_f32_i32(a: f32) -> i32 {
|
rustc: Improving safe wasm float->int casts
This commit improves code generation for WebAssembly targets when
translating floating to integer casts. This improvement is only relevant
when the `nontrapping-fptoint` feature is not enabled, but the feature
is not enabled by default right now. Additionally this improvement only
affects safe casts since unchecked casts were improved in #74659.
Some more background for this issue is present on #73591, but the
general gist of the issue is that in LLVM the `fptosi` and `fptoui`
instructions are defined to return an `undef` value if they execute on
out-of-bounds values; they notably do not trap. To implement these
instructions for WebAssembly the LLVM backend must therefore generate
quite a few instructions before executing `i32.trunc_f32_s` (for
example) because this WebAssembly instruction traps on out-of-bounds
values. This codegen into wasm instructions happens very late in the
code generator, so what ends up happening is that rustc inserts its own
codegen to implement Rust's saturating semantics, and then LLVM also
inserts its own codegen to make sure that the `fptosi` instruction
doesn't trap. Overall this means that a function like this:
#[no_mangle]
pub unsafe extern "C" fn cast(x: f64) -> u32 {
x as u32
}
will generate this WebAssembly today:
(func $cast (type 0) (param f64) (result i32)
(local i32 i32)
local.get 0
f64.const 0x1.fffffffep+31 (;=4.29497e+09;)
f64.gt
local.set 1
block ;; label = @1
block ;; label = @2
local.get 0
f64.const 0x0p+0 (;=0;)
local.get 0
f64.const 0x0p+0 (;=0;)
f64.gt
select
local.tee 0
f64.const 0x1p+32 (;=4.29497e+09;)
f64.lt
local.get 0
f64.const 0x0p+0 (;=0;)
f64.ge
i32.and
i32.eqz
br_if 0 (;@2;)
local.get 0
i32.trunc_f64_u
local.set 2
br 1 (;@1;)
end
i32.const 0
local.set 2
end
i32.const -1
local.get 2
local.get 1
select)
This PR improves the situation by updating the code generation for
float-to-int conversions in rustc, specifically only for WebAssembly
targets and only for some situations (float-to-u8 still has not great
codegen). The fix here is to use basic blocks and control flow to avoid
speculatively executing `fptosi`, and instead LLVM's raw intrinsic for
the WebAssembly instruction is used instead. This effectively extends
the support added in #74659 to checked casts. After this commit the
codegen for the above Rust function looks like:
(func $cast (type 0) (param f64) (result i32)
(local i32)
block ;; label = @1
local.get 0
f64.const 0x0p+0 (;=0;)
f64.ge
local.tee 1
i32.const 1
i32.xor
br_if 0 (;@1;)
local.get 0
f64.const 0x1.fffffffep+31 (;=4.29497e+09;)
f64.le
i32.eqz
br_if 0 (;@1;)
local.get 0
i32.trunc_f64_u
return
end
i32.const -1
i32.const 0
local.get 1
select)
For reference, in Rust 1.44, which did not have saturating
float-to-integer casts, the codegen LLVM would emit is:
(func $cast (type 0) (param f64) (result i32)
block ;; label = @1
local.get 0
f64.const 0x1p+32 (;=4.29497e+09;)
f64.lt
local.get 0
f64.const 0x0p+0 (;=0;)
f64.ge
i32.and
i32.eqz
br_if 0 (;@1;)
local.get 0
i32.trunc_f64_u
return
end
i32.const 0)
So we're relatively close to the original codegen, although it's
slightly different because the semantics of the function changed where
we're emulating the `i32.trunc_sat_f32_s` instruction rather than always
replacing out-of-bounds values with zero.
There is still work that could be done to improve casts such as `f32` to
`u8`. That form of cast still uses the `fptosi` instruction which
generates lots of branch-y code. This seems less important to tackle now
though. In the meantime this should take care of most use cases of
floating-point conversion and as a result I'm going to speculate that
this...
Closes #73591
2020-07-24 03:20:42 +08:00
|
|
|
// CHECK-NOT: fptosi float {{.*}} to i32
|
|
|
|
|
// CHECK-NOT: select i1 {{.*}}, i32 {{.*}}, i32 {{.*}}
|
2021-04-23 23:59:49 +08:00
|
|
|
// CHECK: {{.*}} call {{.*}} @llvm.fptosi.sat.i32.f32{{.*}}
|
2020-06-27 19:26:30 +08:00
|
|
|
a as _
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// CHECK-LABEL: @cast_f64_u64
|
|
|
|
|
#[no_mangle]
|
|
|
|
|
pub fn cast_f64_u64(a: f64) -> u64 {
|
rustc: Improving safe wasm float->int casts
This commit improves code generation for WebAssembly targets when
translating floating to integer casts. This improvement is only relevant
when the `nontrapping-fptoint` feature is not enabled, but the feature
is not enabled by default right now. Additionally this improvement only
affects safe casts since unchecked casts were improved in #74659.
Some more background for this issue is present on #73591, but the
general gist of the issue is that in LLVM the `fptosi` and `fptoui`
instructions are defined to return an `undef` value if they execute on
out-of-bounds values; they notably do not trap. To implement these
instructions for WebAssembly the LLVM backend must therefore generate
quite a few instructions before executing `i32.trunc_f32_s` (for
example) because this WebAssembly instruction traps on out-of-bounds
values. This codegen into wasm instructions happens very late in the
code generator, so what ends up happening is that rustc inserts its own
codegen to implement Rust's saturating semantics, and then LLVM also
inserts its own codegen to make sure that the `fptosi` instruction
doesn't trap. Overall this means that a function like this:
#[no_mangle]
pub unsafe extern "C" fn cast(x: f64) -> u32 {
x as u32
}
will generate this WebAssembly today:
(func $cast (type 0) (param f64) (result i32)
(local i32 i32)
local.get 0
f64.const 0x1.fffffffep+31 (;=4.29497e+09;)
f64.gt
local.set 1
block ;; label = @1
block ;; label = @2
local.get 0
f64.const 0x0p+0 (;=0;)
local.get 0
f64.const 0x0p+0 (;=0;)
f64.gt
select
local.tee 0
f64.const 0x1p+32 (;=4.29497e+09;)
f64.lt
local.get 0
f64.const 0x0p+0 (;=0;)
f64.ge
i32.and
i32.eqz
br_if 0 (;@2;)
local.get 0
i32.trunc_f64_u
local.set 2
br 1 (;@1;)
end
i32.const 0
local.set 2
end
i32.const -1
local.get 2
local.get 1
select)
This PR improves the situation by updating the code generation for
float-to-int conversions in rustc, specifically only for WebAssembly
targets and only for some situations (float-to-u8 still has not great
codegen). The fix here is to use basic blocks and control flow to avoid
speculatively executing `fptosi`, and instead LLVM's raw intrinsic for
the WebAssembly instruction is used instead. This effectively extends
the support added in #74659 to checked casts. After this commit the
codegen for the above Rust function looks like:
(func $cast (type 0) (param f64) (result i32)
(local i32)
block ;; label = @1
local.get 0
f64.const 0x0p+0 (;=0;)
f64.ge
local.tee 1
i32.const 1
i32.xor
br_if 0 (;@1;)
local.get 0
f64.const 0x1.fffffffep+31 (;=4.29497e+09;)
f64.le
i32.eqz
br_if 0 (;@1;)
local.get 0
i32.trunc_f64_u
return
end
i32.const -1
i32.const 0
local.get 1
select)
For reference, in Rust 1.44, which did not have saturating
float-to-integer casts, the codegen LLVM would emit is:
(func $cast (type 0) (param f64) (result i32)
block ;; label = @1
local.get 0
f64.const 0x1p+32 (;=4.29497e+09;)
f64.lt
local.get 0
f64.const 0x0p+0 (;=0;)
f64.ge
i32.and
i32.eqz
br_if 0 (;@1;)
local.get 0
i32.trunc_f64_u
return
end
i32.const 0)
So we're relatively close to the original codegen, although it's
slightly different because the semantics of the function changed where
we're emulating the `i32.trunc_sat_f32_s` instruction rather than always
replacing out-of-bounds values with zero.
There is still work that could be done to improve casts such as `f32` to
`u8`. That form of cast still uses the `fptosi` instruction which
generates lots of branch-y code. This seems less important to tackle now
though. In the meantime this should take care of most use cases of
floating-point conversion and as a result I'm going to speculate that
this...
Closes #73591
2020-07-24 03:20:42 +08:00
|
|
|
// CHECK-NOT: fptoui double {{.*}} to i64
|
|
|
|
|
// CHECK-NOT: select i1 {{.*}}, i64 {{.*}}, i64 {{.*}}
|
2021-04-23 23:59:49 +08:00
|
|
|
// CHECK: {{.*}} call {{.*}} @llvm.fptoui.sat.i64.f64{{.*}}
|
2020-06-27 19:26:30 +08:00
|
|
|
a as _
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// CHECK-LABEL: @cast_f64_u32
|
|
|
|
|
#[no_mangle]
|
|
|
|
|
pub fn cast_f64_u32(a: f64) -> u32 {
|
rustc: Improving safe wasm float->int casts
This commit improves code generation for WebAssembly targets when
translating floating to integer casts. This improvement is only relevant
when the `nontrapping-fptoint` feature is not enabled, but the feature
is not enabled by default right now. Additionally this improvement only
affects safe casts since unchecked casts were improved in #74659.
Some more background for this issue is present on #73591, but the
general gist of the issue is that in LLVM the `fptosi` and `fptoui`
instructions are defined to return an `undef` value if they execute on
out-of-bounds values; they notably do not trap. To implement these
instructions for WebAssembly the LLVM backend must therefore generate
quite a few instructions before executing `i32.trunc_f32_s` (for
example) because this WebAssembly instruction traps on out-of-bounds
values. This codegen into wasm instructions happens very late in the
code generator, so what ends up happening is that rustc inserts its own
codegen to implement Rust's saturating semantics, and then LLVM also
inserts its own codegen to make sure that the `fptosi` instruction
doesn't trap. Overall this means that a function like this:
#[no_mangle]
pub unsafe extern "C" fn cast(x: f64) -> u32 {
x as u32
}
will generate this WebAssembly today:
(func $cast (type 0) (param f64) (result i32)
(local i32 i32)
local.get 0
f64.const 0x1.fffffffep+31 (;=4.29497e+09;)
f64.gt
local.set 1
block ;; label = @1
block ;; label = @2
local.get 0
f64.const 0x0p+0 (;=0;)
local.get 0
f64.const 0x0p+0 (;=0;)
f64.gt
select
local.tee 0
f64.const 0x1p+32 (;=4.29497e+09;)
f64.lt
local.get 0
f64.const 0x0p+0 (;=0;)
f64.ge
i32.and
i32.eqz
br_if 0 (;@2;)
local.get 0
i32.trunc_f64_u
local.set 2
br 1 (;@1;)
end
i32.const 0
local.set 2
end
i32.const -1
local.get 2
local.get 1
select)
This PR improves the situation by updating the code generation for
float-to-int conversions in rustc, specifically only for WebAssembly
targets and only for some situations (float-to-u8 still has not great
codegen). The fix here is to use basic blocks and control flow to avoid
speculatively executing `fptosi`, and instead LLVM's raw intrinsic for
the WebAssembly instruction is used instead. This effectively extends
the support added in #74659 to checked casts. After this commit the
codegen for the above Rust function looks like:
(func $cast (type 0) (param f64) (result i32)
(local i32)
block ;; label = @1
local.get 0
f64.const 0x0p+0 (;=0;)
f64.ge
local.tee 1
i32.const 1
i32.xor
br_if 0 (;@1;)
local.get 0
f64.const 0x1.fffffffep+31 (;=4.29497e+09;)
f64.le
i32.eqz
br_if 0 (;@1;)
local.get 0
i32.trunc_f64_u
return
end
i32.const -1
i32.const 0
local.get 1
select)
For reference, in Rust 1.44, which did not have saturating
float-to-integer casts, the codegen LLVM would emit is:
(func $cast (type 0) (param f64) (result i32)
block ;; label = @1
local.get 0
f64.const 0x1p+32 (;=4.29497e+09;)
f64.lt
local.get 0
f64.const 0x0p+0 (;=0;)
f64.ge
i32.and
i32.eqz
br_if 0 (;@1;)
local.get 0
i32.trunc_f64_u
return
end
i32.const 0)
So we're relatively close to the original codegen, although it's
slightly different because the semantics of the function changed where
we're emulating the `i32.trunc_sat_f32_s` instruction rather than always
replacing out-of-bounds values with zero.
There is still work that could be done to improve casts such as `f32` to
`u8`. That form of cast still uses the `fptosi` instruction which
generates lots of branch-y code. This seems less important to tackle now
though. In the meantime this should take care of most use cases of
floating-point conversion and as a result I'm going to speculate that
this...
Closes #73591
2020-07-24 03:20:42 +08:00
|
|
|
// CHECK-NOT: fptoui double {{.*}} to i32
|
|
|
|
|
// CHECK-NOT: select i1 {{.*}}, i32 {{.*}}, i32 {{.*}}
|
2021-04-23 23:59:49 +08:00
|
|
|
// CHECK: {{.*}} call {{.*}} @llvm.fptoui.sat.i32.f64{{.*}}
|
2020-06-27 19:26:30 +08:00
|
|
|
a as _
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// CHECK-LABEL: @cast_f32_u64
|
|
|
|
|
#[no_mangle]
|
|
|
|
|
pub fn cast_f32_u64(a: f32) -> u64 {
|
rustc: Improving safe wasm float->int casts
This commit improves code generation for WebAssembly targets when
translating floating to integer casts. This improvement is only relevant
when the `nontrapping-fptoint` feature is not enabled, but the feature
is not enabled by default right now. Additionally this improvement only
affects safe casts since unchecked casts were improved in #74659.
Some more background for this issue is present on #73591, but the
general gist of the issue is that in LLVM the `fptosi` and `fptoui`
instructions are defined to return an `undef` value if they execute on
out-of-bounds values; they notably do not trap. To implement these
instructions for WebAssembly the LLVM backend must therefore generate
quite a few instructions before executing `i32.trunc_f32_s` (for
example) because this WebAssembly instruction traps on out-of-bounds
values. This codegen into wasm instructions happens very late in the
code generator, so what ends up happening is that rustc inserts its own
codegen to implement Rust's saturating semantics, and then LLVM also
inserts its own codegen to make sure that the `fptosi` instruction
doesn't trap. Overall this means that a function like this:
#[no_mangle]
pub unsafe extern "C" fn cast(x: f64) -> u32 {
x as u32
}
will generate this WebAssembly today:
(func $cast (type 0) (param f64) (result i32)
(local i32 i32)
local.get 0
f64.const 0x1.fffffffep+31 (;=4.29497e+09;)
f64.gt
local.set 1
block ;; label = @1
block ;; label = @2
local.get 0
f64.const 0x0p+0 (;=0;)
local.get 0
f64.const 0x0p+0 (;=0;)
f64.gt
select
local.tee 0
f64.const 0x1p+32 (;=4.29497e+09;)
f64.lt
local.get 0
f64.const 0x0p+0 (;=0;)
f64.ge
i32.and
i32.eqz
br_if 0 (;@2;)
local.get 0
i32.trunc_f64_u
local.set 2
br 1 (;@1;)
end
i32.const 0
local.set 2
end
i32.const -1
local.get 2
local.get 1
select)
This PR improves the situation by updating the code generation for
float-to-int conversions in rustc, specifically only for WebAssembly
targets and only for some situations (float-to-u8 still has not great
codegen). The fix here is to use basic blocks and control flow to avoid
speculatively executing `fptosi`, and instead LLVM's raw intrinsic for
the WebAssembly instruction is used instead. This effectively extends
the support added in #74659 to checked casts. After this commit the
codegen for the above Rust function looks like:
(func $cast (type 0) (param f64) (result i32)
(local i32)
block ;; label = @1
local.get 0
f64.const 0x0p+0 (;=0;)
f64.ge
local.tee 1
i32.const 1
i32.xor
br_if 0 (;@1;)
local.get 0
f64.const 0x1.fffffffep+31 (;=4.29497e+09;)
f64.le
i32.eqz
br_if 0 (;@1;)
local.get 0
i32.trunc_f64_u
return
end
i32.const -1
i32.const 0
local.get 1
select)
For reference, in Rust 1.44, which did not have saturating
float-to-integer casts, the codegen LLVM would emit is:
(func $cast (type 0) (param f64) (result i32)
block ;; label = @1
local.get 0
f64.const 0x1p+32 (;=4.29497e+09;)
f64.lt
local.get 0
f64.const 0x0p+0 (;=0;)
f64.ge
i32.and
i32.eqz
br_if 0 (;@1;)
local.get 0
i32.trunc_f64_u
return
end
i32.const 0)
So we're relatively close to the original codegen, although it's
slightly different because the semantics of the function changed where
we're emulating the `i32.trunc_sat_f32_s` instruction rather than always
replacing out-of-bounds values with zero.
There is still work that could be done to improve casts such as `f32` to
`u8`. That form of cast still uses the `fptosi` instruction which
generates lots of branch-y code. This seems less important to tackle now
though. In the meantime this should take care of most use cases of
floating-point conversion and as a result I'm going to speculate that
this...
Closes #73591
2020-07-24 03:20:42 +08:00
|
|
|
// CHECK-NOT: fptoui float {{.*}} to i64
|
|
|
|
|
// CHECK-NOT: select i1 {{.*}}, i64 {{.*}}, i64 {{.*}}
|
2021-04-23 23:59:49 +08:00
|
|
|
// CHECK: {{.*}} call {{.*}} @llvm.fptoui.sat.i64.f32{{.*}}
|
2020-06-27 19:26:30 +08:00
|
|
|
a as _
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// CHECK-LABEL: @cast_f32_u32
|
|
|
|
|
#[no_mangle]
|
|
|
|
|
pub fn cast_f32_u32(a: f32) -> u32 {
|
rustc: Improving safe wasm float->int casts
This commit improves code generation for WebAssembly targets when
translating floating to integer casts. This improvement is only relevant
when the `nontrapping-fptoint` feature is not enabled, but the feature
is not enabled by default right now. Additionally this improvement only
affects safe casts since unchecked casts were improved in #74659.
Some more background for this issue is present on #73591, but the
general gist of the issue is that in LLVM the `fptosi` and `fptoui`
instructions are defined to return an `undef` value if they execute on
out-of-bounds values; they notably do not trap. To implement these
instructions for WebAssembly the LLVM backend must therefore generate
quite a few instructions before executing `i32.trunc_f32_s` (for
example) because this WebAssembly instruction traps on out-of-bounds
values. This codegen into wasm instructions happens very late in the
code generator, so what ends up happening is that rustc inserts its own
codegen to implement Rust's saturating semantics, and then LLVM also
inserts its own codegen to make sure that the `fptosi` instruction
doesn't trap. Overall this means that a function like this:
#[no_mangle]
pub unsafe extern "C" fn cast(x: f64) -> u32 {
x as u32
}
will generate this WebAssembly today:
(func $cast (type 0) (param f64) (result i32)
(local i32 i32)
local.get 0
f64.const 0x1.fffffffep+31 (;=4.29497e+09;)
f64.gt
local.set 1
block ;; label = @1
block ;; label = @2
local.get 0
f64.const 0x0p+0 (;=0;)
local.get 0
f64.const 0x0p+0 (;=0;)
f64.gt
select
local.tee 0
f64.const 0x1p+32 (;=4.29497e+09;)
f64.lt
local.get 0
f64.const 0x0p+0 (;=0;)
f64.ge
i32.and
i32.eqz
br_if 0 (;@2;)
local.get 0
i32.trunc_f64_u
local.set 2
br 1 (;@1;)
end
i32.const 0
local.set 2
end
i32.const -1
local.get 2
local.get 1
select)
This PR improves the situation by updating the code generation for
float-to-int conversions in rustc, specifically only for WebAssembly
targets and only for some situations (float-to-u8 still has not great
codegen). The fix here is to use basic blocks and control flow to avoid
speculatively executing `fptosi`, and instead LLVM's raw intrinsic for
the WebAssembly instruction is used instead. This effectively extends
the support added in #74659 to checked casts. After this commit the
codegen for the above Rust function looks like:
(func $cast (type 0) (param f64) (result i32)
(local i32)
block ;; label = @1
local.get 0
f64.const 0x0p+0 (;=0;)
f64.ge
local.tee 1
i32.const 1
i32.xor
br_if 0 (;@1;)
local.get 0
f64.const 0x1.fffffffep+31 (;=4.29497e+09;)
f64.le
i32.eqz
br_if 0 (;@1;)
local.get 0
i32.trunc_f64_u
return
end
i32.const -1
i32.const 0
local.get 1
select)
For reference, in Rust 1.44, which did not have saturating
float-to-integer casts, the codegen LLVM would emit is:
(func $cast (type 0) (param f64) (result i32)
block ;; label = @1
local.get 0
f64.const 0x1p+32 (;=4.29497e+09;)
f64.lt
local.get 0
f64.const 0x0p+0 (;=0;)
f64.ge
i32.and
i32.eqz
br_if 0 (;@1;)
local.get 0
i32.trunc_f64_u
return
end
i32.const 0)
So we're relatively close to the original codegen, although it's
slightly different because the semantics of the function changed where
we're emulating the `i32.trunc_sat_f32_s` instruction rather than always
replacing out-of-bounds values with zero.
There is still work that could be done to improve casts such as `f32` to
`u8`. That form of cast still uses the `fptosi` instruction which
generates lots of branch-y code. This seems less important to tackle now
though. In the meantime this should take care of most use cases of
floating-point conversion and as a result I'm going to speculate that
this...
Closes #73591
2020-07-24 03:20:42 +08:00
|
|
|
// CHECK-NOT: fptoui float {{.*}} to i32
|
|
|
|
|
// CHECK-NOT: select i1 {{.*}}, i32 {{.*}}, i32 {{.*}}
|
2021-04-23 23:59:49 +08:00
|
|
|
// CHECK: {{.*}} call {{.*}} @llvm.fptoui.sat.i32.f32{{.*}}
|
2020-06-27 19:26:30 +08:00
|
|
|
a as _
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// CHECK-LABEL: @cast_f32_u8
|
|
|
|
|
#[no_mangle]
|
|
|
|
|
pub fn cast_f32_u8(a: f32) -> u8 {
|
2021-04-23 23:59:49 +08:00
|
|
|
// CHECK-NOT: fptoui float {{.*}} to i8
|
|
|
|
|
// CHECK-NOT: select i1 {{.*}}, i8 {{.*}}, i8 {{.*}}
|
|
|
|
|
// CHECK: {{.*}} call {{.*}} @llvm.fptoui.sat.i8.f32{{.*}}
|
2020-06-27 19:26:30 +08:00
|
|
|
a as _
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// CHECK-LABEL: @cast_unchecked_f64_i64
|
|
|
|
|
#[no_mangle]
|
|
|
|
|
pub unsafe fn cast_unchecked_f64_i64(a: f64) -> i64 {
|
2020-07-23 05:51:12 +08:00
|
|
|
// CHECK: {{.*}} call {{.*}} @llvm.wasm.trunc.signed.{{.*}}
|
2020-06-27 19:26:30 +08:00
|
|
|
// CHECK-NEXT: ret i64 {{.*}}
|
|
|
|
|
a.to_int_unchecked()
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// CHECK-LABEL: @cast_unchecked_f64_i32
|
|
|
|
|
#[no_mangle]
|
|
|
|
|
pub unsafe fn cast_unchecked_f64_i32(a: f64) -> i32 {
|
2020-07-23 05:51:12 +08:00
|
|
|
// CHECK: {{.*}} call {{.*}} @llvm.wasm.trunc.signed.{{.*}}
|
2020-06-27 19:26:30 +08:00
|
|
|
// CHECK-NEXT: ret i32 {{.*}}
|
|
|
|
|
a.to_int_unchecked()
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// CHECK-LABEL: @cast_unchecked_f32_i64
|
|
|
|
|
#[no_mangle]
|
|
|
|
|
pub unsafe fn cast_unchecked_f32_i64(a: f32) -> i64 {
|
2020-07-23 05:51:12 +08:00
|
|
|
// CHECK: {{.*}} call {{.*}} @llvm.wasm.trunc.signed.{{.*}}
|
2020-06-27 19:26:30 +08:00
|
|
|
// CHECK-NEXT: ret i64 {{.*}}
|
|
|
|
|
a.to_int_unchecked()
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// CHECK-LABEL: @cast_unchecked_f32_i32
|
|
|
|
|
#[no_mangle]
|
|
|
|
|
pub unsafe fn cast_unchecked_f32_i32(a: f32) -> i32 {
|
2020-07-23 05:51:12 +08:00
|
|
|
// CHECK: {{.*}} call {{.*}} @llvm.wasm.trunc.signed.{{.*}}
|
2020-06-27 19:26:30 +08:00
|
|
|
// CHECK-NEXT: ret i32 {{.*}}
|
|
|
|
|
a.to_int_unchecked()
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// CHECK-LABEL: @cast_unchecked_f64_u64
|
|
|
|
|
#[no_mangle]
|
|
|
|
|
pub unsafe fn cast_unchecked_f64_u64(a: f64) -> u64 {
|
2020-07-23 05:51:12 +08:00
|
|
|
// CHECK: {{.*}} call {{.*}} @llvm.wasm.trunc.unsigned.{{.*}}
|
2020-06-27 19:26:30 +08:00
|
|
|
// CHECK-NEXT: ret i64 {{.*}}
|
|
|
|
|
a.to_int_unchecked()
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// CHECK-LABEL: @cast_unchecked_f64_u32
|
|
|
|
|
#[no_mangle]
|
|
|
|
|
pub unsafe fn cast_unchecked_f64_u32(a: f64) -> u32 {
|
2020-07-23 05:51:12 +08:00
|
|
|
// CHECK: {{.*}} call {{.*}} @llvm.wasm.trunc.unsigned.{{.*}}
|
2020-06-27 19:26:30 +08:00
|
|
|
// CHECK-NEXT: ret i32 {{.*}}
|
|
|
|
|
a.to_int_unchecked()
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// CHECK-LABEL: @cast_unchecked_f32_u64
|
|
|
|
|
#[no_mangle]
|
|
|
|
|
pub unsafe fn cast_unchecked_f32_u64(a: f32) -> u64 {
|
2020-07-23 05:51:12 +08:00
|
|
|
// CHECK: {{.*}} call {{.*}} @llvm.wasm.trunc.unsigned.{{.*}}
|
2020-06-27 19:26:30 +08:00
|
|
|
// CHECK-NEXT: ret i64 {{.*}}
|
|
|
|
|
a.to_int_unchecked()
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// CHECK-LABEL: @cast_unchecked_f32_u32
|
|
|
|
|
#[no_mangle]
|
|
|
|
|
pub unsafe fn cast_unchecked_f32_u32(a: f32) -> u32 {
|
2020-07-23 05:51:12 +08:00
|
|
|
// CHECK: {{.*}} call {{.*}} @llvm.wasm.trunc.unsigned.{{.*}}
|
2020-06-27 19:26:30 +08:00
|
|
|
// CHECK-NEXT: ret i32 {{.*}}
|
|
|
|
|
a.to_int_unchecked()
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// CHECK-LABEL: @cast_unchecked_f32_u8
|
|
|
|
|
#[no_mangle]
|
|
|
|
|
pub unsafe fn cast_unchecked_f32_u8(a: f32) -> u8 {
|
|
|
|
|
// CHECK-NOT: {{.*}} call {{.*}} @llvm.wasm.trunc.{{.*}}
|
|
|
|
|
// CHECK: fptoui float {{.*}} to i8
|
|
|
|
|
// CHECK-NEXT: ret i8 {{.*}}
|
|
|
|
|
a.to_int_unchecked()
|
|
|
|
|
}
|