diff --git a/crates/spirv-std/src/arch.rs b/crates/spirv-std/src/arch.rs
index df0d061c74..ead313c6ae 100644
--- a/crates/spirv-std/src/arch.rs
+++ b/crates/spirv-std/src/arch.rs
@@ -153,7 +153,7 @@ pub fn kill() -> ! {
 /// <https://htmlpreview.github.io/?https://github.com/KhronosGroup/SPIRV-Registry/blob/master/extensions/KHR/SPV_KHR_shader_clock.html>
 #[spirv_std_macros::gpu_only]
 #[doc(alias = "OpReadClockKHR")]
-pub unsafe fn read_clock_khr<const SCOPE: u32>() -> u64 {
+pub fn read_clock_khr<const SCOPE: u32>() -> u64 {
     unsafe {
         let mut result: u64;
 
@@ -175,7 +175,7 @@ pub unsafe fn read_clock_khr<const SCOPE: u32>() -> u64 {
 /// bits and the second component containing the 32 most significant bits.'
 #[spirv_std_macros::gpu_only]
 #[doc(alias = "OpReadClockKHR")]
-pub unsafe fn read_clock_uvec2_khr<V: Vector<u32, 2>, const SCOPE: u32>() -> V {
+pub fn read_clock_uvec2_khr<V: Vector<u32, 2>, const SCOPE: u32>() -> V {
     unsafe {
         let mut result = V::default();
 
diff --git a/crates/spirv-std/src/arch/barrier.rs b/crates/spirv-std/src/arch/barrier.rs
index aa7df1746c..4d08e17cb2 100644
--- a/crates/spirv-std/src/arch/barrier.rs
+++ b/crates/spirv-std/src/arch/barrier.rs
@@ -33,7 +33,7 @@ use core::arch::asm;
 #[spirv_std_macros::gpu_only]
 #[doc(alias = "OpControlBarrier")]
 #[inline]
-pub unsafe fn control_barrier<
+pub fn control_barrier<
     const EXECUTION: u32, // Scope
     const MEMORY: u32,    // Scope
     const SEMANTICS: u32, // Semantics
@@ -70,7 +70,7 @@ pub unsafe fn control_barrier<
 #[spirv_std_macros::gpu_only]
 #[doc(alias = "OpMemoryBarrier")]
 #[inline]
-pub unsafe fn memory_barrier<
+pub fn memory_barrier<
     const MEMORY: u32,    // Scope
     const SEMANTICS: u32, // Semantics
 >() {
@@ -93,16 +93,14 @@ pub unsafe fn memory_barrier<
 /// From <https://docs.microsoft.com/en-us/windows/win32/direct3dhlsl/groupmemorybarrier>
 #[spirv_std_macros::gpu_only]
 #[inline]
-pub unsafe fn workgroup_memory_barrier() {
-    unsafe {
-        memory_barrier::<
-            { crate::memory::Scope::Workgroup as u32 },
-            {
-                crate::memory::Semantics::WORKGROUP_MEMORY.bits()
-                    | crate::memory::Semantics::ACQUIRE_RELEASE.bits()
-            },
-        >();
-    }
+pub fn workgroup_memory_barrier() {
+    memory_barrier::<
+        { crate::memory::Scope::Workgroup as u32 },
+        {
+            crate::memory::Semantics::WORKGROUP_MEMORY.bits()
+                | crate::memory::Semantics::ACQUIRE_RELEASE.bits()
+        },
+    >();
 }
 
 /// Blocks execution of all threads in a group until all group shared accesses have been completed and all threads in the group have reached this call.
@@ -112,17 +110,15 @@ pub unsafe fn workgroup_memory_barrier() {
 /// From <https://docs.microsoft.com/en-us/windows/win32/direct3dhlsl/groupmemorybarrierwithgroupsync>
 #[spirv_std_macros::gpu_only]
 #[inline]
-pub unsafe fn workgroup_memory_barrier_with_group_sync() {
-    unsafe {
-        control_barrier::<
-            { crate::memory::Scope::Workgroup as u32 },
-            { crate::memory::Scope::Workgroup as u32 },
-            {
-                crate::memory::Semantics::WORKGROUP_MEMORY.bits()
-                    | crate::memory::Semantics::ACQUIRE_RELEASE.bits()
-            },
-        >();
-    }
+pub fn workgroup_memory_barrier_with_group_sync() {
+    control_barrier::<
+        { crate::memory::Scope::Workgroup as u32 },
+        { crate::memory::Scope::Workgroup as u32 },
+        {
+            crate::memory::Semantics::WORKGROUP_MEMORY.bits()
+                | crate::memory::Semantics::ACQUIRE_RELEASE.bits()
+        },
+    >();
 }
 
 /// Blocks execution of all threads in a group until all device memory accesses have been completed.
@@ -132,17 +128,15 @@ pub unsafe fn workgroup_memory_barrier_with_group_sync() {
 /// From <https://docs.microsoft.com/en-us/windows/win32/direct3dhlsl/devicememorybarrier>
 #[spirv_std_macros::gpu_only]
 #[inline]
-pub unsafe fn device_memory_barrier() {
-    unsafe {
-        memory_barrier::<
-            { crate::memory::Scope::Device as u32 },
-            {
-                crate::memory::Semantics::IMAGE_MEMORY.bits()
-                    | crate::memory::Semantics::UNIFORM_MEMORY.bits()
-                    | crate::memory::Semantics::ACQUIRE_RELEASE.bits()
-            },
-        >();
-    }
+pub fn device_memory_barrier() {
+    memory_barrier::<
+        { crate::memory::Scope::Device as u32 },
+        {
+            crate::memory::Semantics::IMAGE_MEMORY.bits()
+                | crate::memory::Semantics::UNIFORM_MEMORY.bits()
+                | crate::memory::Semantics::ACQUIRE_RELEASE.bits()
+        },
+    >();
 }
 
 /// Blocks execution of all threads in a group until all device memory accesses have been completed and all threads in the group have reached this call.
@@ -152,18 +146,16 @@ pub unsafe fn device_memory_barrier() {
 /// From <https://docs.microsoft.com/en-us/windows/win32/direct3dhlsl/devicememorybarrierwithgroupsync>
 #[spirv_std_macros::gpu_only]
 #[inline]
-pub unsafe fn device_memory_barrier_with_group_sync() {
-    unsafe {
-        control_barrier::<
-            { crate::memory::Scope::Workgroup as u32 },
-            { crate::memory::Scope::Device as u32 },
-            {
-                crate::memory::Semantics::IMAGE_MEMORY.bits()
-                    | crate::memory::Semantics::UNIFORM_MEMORY.bits()
-                    | crate::memory::Semantics::ACQUIRE_RELEASE.bits()
-            },
-        >();
-    }
+pub fn device_memory_barrier_with_group_sync() {
+    control_barrier::<
+        { crate::memory::Scope::Workgroup as u32 },
+        { crate::memory::Scope::Device as u32 },
+        {
+            crate::memory::Semantics::IMAGE_MEMORY.bits()
+                | crate::memory::Semantics::UNIFORM_MEMORY.bits()
+                | crate::memory::Semantics::ACQUIRE_RELEASE.bits()
+        },
+    >();
 }
 
 /// Blocks execution of all threads in a group until all memory accesses have been completed.
@@ -173,18 +165,16 @@ pub unsafe fn device_memory_barrier_with_group_sync() {
 /// From <https://docs.microsoft.com/en-us/windows/win32/direct3dhlsl/allmemorybarrier>
 #[spirv_std_macros::gpu_only]
 #[inline]
-pub unsafe fn all_memory_barrier() {
-    unsafe {
-        memory_barrier::<
-            { crate::memory::Scope::Device as u32 },
-            {
-                crate::memory::Semantics::WORKGROUP_MEMORY.bits()
-                    | crate::memory::Semantics::IMAGE_MEMORY.bits()
-                    | crate::memory::Semantics::UNIFORM_MEMORY.bits()
-                    | crate::memory::Semantics::ACQUIRE_RELEASE.bits()
-            },
-        >();
-    }
+pub fn all_memory_barrier() {
+    memory_barrier::<
+        { crate::memory::Scope::Device as u32 },
+        {
+            crate::memory::Semantics::WORKGROUP_MEMORY.bits()
+                | crate::memory::Semantics::IMAGE_MEMORY.bits()
+                | crate::memory::Semantics::UNIFORM_MEMORY.bits()
+                | crate::memory::Semantics::ACQUIRE_RELEASE.bits()
+        },
+    >();
 }
 
 /// Blocks execution of all threads in a group until all memory accesses have been completed and all threads in the group have reached this call.
@@ -194,17 +184,15 @@ pub unsafe fn all_memory_barrier() {
 /// From <https://docs.microsoft.com/en-us/windows/win32/direct3dhlsl/allmemorybarrierwithgroupsync>
 #[spirv_std_macros::gpu_only]
 #[inline]
-pub unsafe fn all_memory_barrier_with_group_sync() {
-    unsafe {
-        control_barrier::<
-            { crate::memory::Scope::Workgroup as u32 },
-            { crate::memory::Scope::Device as u32 },
-            {
-                crate::memory::Semantics::WORKGROUP_MEMORY.bits()
-                    | crate::memory::Semantics::IMAGE_MEMORY.bits()
-                    | crate::memory::Semantics::UNIFORM_MEMORY.bits()
-                    | crate::memory::Semantics::ACQUIRE_RELEASE.bits()
-            },
-        >();
-    }
+pub fn all_memory_barrier_with_group_sync() {
+    control_barrier::<
+        { crate::memory::Scope::Workgroup as u32 },
+        { crate::memory::Scope::Device as u32 },
+        {
+            crate::memory::Semantics::WORKGROUP_MEMORY.bits()
+                | crate::memory::Semantics::IMAGE_MEMORY.bits()
+                | crate::memory::Semantics::UNIFORM_MEMORY.bits()
+                | crate::memory::Semantics::ACQUIRE_RELEASE.bits()
+        },
+    >();
 }
diff --git a/crates/spirv-std/src/arch/demote_to_helper_invocation_ext.rs b/crates/spirv-std/src/arch/demote_to_helper_invocation_ext.rs
index 8be3d572a1..8f7fdbaea1 100644
--- a/crates/spirv-std/src/arch/demote_to_helper_invocation_ext.rs
+++ b/crates/spirv-std/src/arch/demote_to_helper_invocation_ext.rs
@@ -19,7 +19,7 @@ use core::arch::asm;
 /// of [demote_to_helper_invocation].
 #[spirv_std_macros::gpu_only]
 #[doc(alias = "OpDemoteToHelperInvocationEXT", alias = "discard")]
-pub unsafe fn demote_to_helper_invocation() {
+pub fn demote_to_helper_invocation() {
     unsafe {
         asm!("OpDemoteToHelperInvocationEXT");
     }
diff --git a/crates/spirv-std/src/arch/mesh_shading.rs b/crates/spirv-std/src/arch/mesh_shading.rs
index 5a06197f07..c9fadf118a 100644
--- a/crates/spirv-std/src/arch/mesh_shading.rs
+++ b/crates/spirv-std/src/arch/mesh_shading.rs
@@ -17,6 +17,11 @@ use core::arch::asm;
 /// by this instruction.
 ///
 /// This instruction is only valid in the *`MeshEXT`* Execution Model.
+///
+/// # Safety
+/// * Must be called **exactly once** in mesh shaders
+/// * Must be called in uniform control flow
+/// * Must not write any output before this instruction in invoked
 #[spirv_std_macros::gpu_only]
 #[doc(alias = "OpSetMeshOutputsEXT")]
 #[inline]
@@ -54,6 +59,10 @@ pub unsafe fn set_mesh_outputs_ext(vertex_count: u32, primitive_count: u32) {
 /// This instruction must be the last instruction in a block.
 ///
 /// This instruction is only valid in the *`TaskEXT`* Execution Model.
+///
+/// # Safety
+/// * Must be called **exactly once** in task shaders
+/// * Must be called in uniform control flow
 #[spirv_std_macros::gpu_only]
 #[doc(alias = "OpEmitMeshTasksEXT")]
 #[inline]
@@ -93,6 +102,10 @@ pub unsafe fn emit_mesh_tasks_ext(group_count_x: u32, group_count_y: u32, group_
 /// This instruction must be the last instruction in a block.
 ///
 /// This instruction is only valid in the *`TaskEXT`* Execution Model.
+///
+/// # Safety
+/// * Must be called **exactly once** in task shaders
+/// * Must be called in uniform control flow
 #[spirv_std_macros::gpu_only]
 #[doc(alias = "OpEmitMeshTasksEXT")]
 #[inline]
diff --git a/crates/spirv-std/src/arch/subgroup.rs b/crates/spirv-std/src/arch/subgroup.rs
index 0d53c85191..e9793db114 100644
--- a/crates/spirv-std/src/arch/subgroup.rs
+++ b/crates/spirv-std/src/arch/subgroup.rs
@@ -64,18 +64,16 @@ pub enum GroupOperation {
 #[doc(alias = "subgroupBarrier")]
 #[inline]
 pub fn subgroup_barrier() {
-    unsafe {
-        barrier::control_barrier::<
-            SUBGROUP,
-            SUBGROUP,
-            {
-                Semantics::ACQUIRE_RELEASE.bits()
-                    | Semantics::UNIFORM_MEMORY.bits()
-                    | Semantics::WORKGROUP_MEMORY.bits()
-                    | Semantics::IMAGE_MEMORY.bits()
-            },
-        >();
-    }
+    barrier::control_barrier::<
+        SUBGROUP,
+        SUBGROUP,
+        {
+            Semantics::ACQUIRE_RELEASE.bits()
+                | Semantics::UNIFORM_MEMORY.bits()
+                | Semantics::WORKGROUP_MEMORY.bits()
+                | Semantics::IMAGE_MEMORY.bits()
+        },
+    >();
 }
 
 /// The function `subgroupMemoryBarrier()` enforces the ordering of all memory
@@ -87,17 +85,15 @@ pub fn subgroup_barrier() {
 #[doc(alias = "subgroupMemoryBarrier")]
 #[inline]
 pub fn subgroup_memory_barrier() {
-    unsafe {
-        barrier::memory_barrier::<
-            SUBGROUP,
-            {
-                Semantics::ACQUIRE_RELEASE.bits()
-                    | Semantics::UNIFORM_MEMORY.bits()
-                    | Semantics::WORKGROUP_MEMORY.bits()
-                    | Semantics::IMAGE_MEMORY.bits()
-            },
-        >();
-    }
+    barrier::memory_barrier::<
+        SUBGROUP,
+        {
+            Semantics::ACQUIRE_RELEASE.bits()
+                | Semantics::UNIFORM_MEMORY.bits()
+                | Semantics::WORKGROUP_MEMORY.bits()
+                | Semantics::IMAGE_MEMORY.bits()
+        },
+    >();
 }
 
 /// The function `subgroupMemoryBarrierBuffer()` enforces the ordering of all
@@ -109,12 +105,10 @@ pub fn subgroup_memory_barrier() {
 #[doc(alias = "subgroupMemoryBarrierBuffer")]
 #[inline]
 pub fn subgroup_memory_barrier_buffer() {
-    unsafe {
-        barrier::memory_barrier::<
-            SUBGROUP,
-            { Semantics::ACQUIRE_RELEASE.bits() | Semantics::UNIFORM_MEMORY.bits() },
-        >();
-    }
+    barrier::memory_barrier::<
+        SUBGROUP,
+        { Semantics::ACQUIRE_RELEASE.bits() | Semantics::UNIFORM_MEMORY.bits() },
+    >();
 }
 
 /// The function `subgroupMemoryBarrierShared()` enforces the ordering of all
@@ -128,12 +122,10 @@ pub fn subgroup_memory_barrier_buffer() {
 #[doc(alias = "subgroupMemoryBarrierShared")]
 #[inline]
 pub fn subgroup_memory_barrier_shared() {
-    unsafe {
-        barrier::memory_barrier::<
-            SUBGROUP,
-            { Semantics::ACQUIRE_RELEASE.bits() | Semantics::WORKGROUP_MEMORY.bits() },
-        >();
-    }
+    barrier::memory_barrier::<
+        SUBGROUP,
+        { Semantics::ACQUIRE_RELEASE.bits() | Semantics::WORKGROUP_MEMORY.bits() },
+    >();
 }
 
 /// The function `subgroupMemoryBarrierImage()` enforces the ordering of all
@@ -145,12 +137,10 @@ pub fn subgroup_memory_barrier_shared() {
 #[doc(alias = "subgroupMemoryBarrierImage")]
 #[inline]
 pub fn subgroup_memory_barrier_image() {
-    unsafe {
-        barrier::memory_barrier::<
-            SUBGROUP,
-            { Semantics::ACQUIRE_RELEASE.bits() | Semantics::IMAGE_MEMORY.bits() },
-        >();
-    }
+    barrier::memory_barrier::<
+        SUBGROUP,
+        { Semantics::ACQUIRE_RELEASE.bits() | Semantics::IMAGE_MEMORY.bits() },
+    >();
 }
 
 /// Result is true only in the active invocation with the lowest id in the group, otherwise result is false.
diff --git a/examples/shaders/reduce/src/lib.rs b/examples/shaders/reduce/src/lib.rs
index 892e956bd2..a100ac5359 100644
--- a/examples/shaders/reduce/src/lib.rs
+++ b/examples/shaders/reduce/src/lib.rs
@@ -54,7 +54,7 @@ pub fn main(
     if subgroup_local_invocation_id == 0 {
         shared[subgroup_id as usize] = sum;
     }
-    unsafe { spirv_std::arch::workgroup_memory_barrier_with_group_sync() };
+    spirv_std::arch::workgroup_memory_barrier_with_group_sync();
     let mut sum = 0;
     if subgroup_id == 0 {
         if subgroup_local_invocation_id < num_subgroups {
diff --git a/tests/compiletests/ui/arch/subgroup/subgroup_ballot.stderr b/tests/compiletests/ui/arch/subgroup/subgroup_ballot.stderr
index d5171d95e9..ffc7660bb9 100644
--- a/tests/compiletests/ui/arch/subgroup/subgroup_ballot.stderr
+++ b/tests/compiletests/ui/arch/subgroup/subgroup_ballot.stderr
@@ -1,9 +1,9 @@
 %1 = OpFunction  %2  None %3
 %4 = OpFunctionParameter  %2
 %5 = OpLabel
-OpLine %6 377 13
+OpLine %6 367 13
 %7 = OpGroupNonUniformBallot  %8  %9 %4
-OpLine %6 416 13
+OpLine %6 406 13
 %10 = OpGroupNonUniformInverseBallot  %2  %9 %7
 OpNoLine
 OpReturnValue %10
diff --git a/tests/compiletests/ui/arch/subgroup/subgroup_ballot_bit_count.stderr b/tests/compiletests/ui/arch/subgroup/subgroup_ballot_bit_count.stderr
index 0d224f07d0..0ecffc39c4 100644
--- a/tests/compiletests/ui/arch/subgroup/subgroup_ballot_bit_count.stderr
+++ b/tests/compiletests/ui/arch/subgroup/subgroup_ballot_bit_count.stderr
@@ -1,7 +1,7 @@
 %1 = OpFunction  %2  None %3
 %4 = OpFunctionParameter  %5
 %6 = OpLabel
-OpLine %7 511 0
+OpLine %7 501 0
 %8 = OpGroupNonUniformBallotBitCount  %2  %9 Reduce %4
 OpNoLine
 OpReturnValue %8
diff --git a/tests/compiletests/ui/arch/subgroup/subgroup_broadcast_first.stderr b/tests/compiletests/ui/arch/subgroup/subgroup_broadcast_first.stderr
index 9a99f698e6..9bb737c779 100644
--- a/tests/compiletests/ui/arch/subgroup/subgroup_broadcast_first.stderr
+++ b/tests/compiletests/ui/arch/subgroup/subgroup_broadcast_first.stderr
@@ -1,7 +1,7 @@
 %1 = OpFunction  %2  None %3
 %4 = OpFunctionParameter  %2
 %5 = OpLabel
-OpLine %6 344 13
+OpLine %6 334 13
 %7 = OpGroupNonUniformBroadcastFirst  %2  %8 %4
 OpNoLine
 OpReturnValue %7
diff --git a/tests/compiletests/ui/arch/subgroup/subgroup_cluster_size_0_fail.stderr b/tests/compiletests/ui/arch/subgroup/subgroup_cluster_size_0_fail.stderr
index 6d9aa75c7a..821239dc33 100644
--- a/tests/compiletests/ui/arch/subgroup/subgroup_cluster_size_0_fail.stderr
+++ b/tests/compiletests/ui/arch/subgroup/subgroup_cluster_size_0_fail.stderr
@@ -1,27 +1,27 @@
 error[E0080]: evaluation panicked: `ClusterSize` must be at least 1
-   --> $SPIRV_STD_SRC/arch/subgroup.rs:839:1
+   --> $SPIRV_STD_SRC/arch/subgroup.rs:829:1
     |
-839 | / macro_subgroup_op_clustered!(impl Integer, "OpGroupNonUniformIAdd", subgroup_clustered_i_add; r"
-840 | | An integer add group operation of all `value` operands contributed by active invocations in the group.
-841 | |
-842 | | Result Type must be a scalar or vector of integer type.
+829 | / macro_subgroup_op_clustered!(impl Integer, "OpGroupNonUniformIAdd", subgroup_clustered_i_add; r"
+830 | | An integer add group operation of all `value` operands contributed by active invocations in the group.
+831 | |
+832 | | Result Type must be a scalar or vector of integer type.
 ...   |
-855 | | * `ClusterSize` must not be greater than the size of the group
-856 | | ");
+845 | | * `ClusterSize` must not be greater than the size of the group
+846 | | ");
     | |__^ evaluation of `spirv_std::arch::subgroup_clustered_i_add::<0, u32, u32>::{constant#0}` failed here
     |
     = note: this error originates in the macro `$crate::panic::panic_2021` which comes from the expansion of the macro `macro_subgroup_op_clustered` (in Nightly builds, run with -Z macro-backtrace for more info)
 
 note: erroneous constant encountered
-   --> $SPIRV_STD_SRC/arch/subgroup.rs:839:1
+   --> $SPIRV_STD_SRC/arch/subgroup.rs:829:1
     |
-839 | / macro_subgroup_op_clustered!(impl Integer, "OpGroupNonUniformIAdd", subgroup_clustered_i_add; r"
-840 | | An integer add group operation of all `value` operands contributed by active invocations in the group.
-841 | |
-842 | | Result Type must be a scalar or vector of integer type.
+829 | / macro_subgroup_op_clustered!(impl Integer, "OpGroupNonUniformIAdd", subgroup_clustered_i_add; r"
+830 | | An integer add group operation of all `value` operands contributed by active invocations in the group.
+831 | |
+832 | | Result Type must be a scalar or vector of integer type.
 ...   |
-855 | | * `ClusterSize` must not be greater than the size of the group
-856 | | ");
+845 | | * `ClusterSize` must not be greater than the size of the group
+846 | | ");
     | |__^
     |
     = note: this note originates in the macro `macro_subgroup_op_clustered` (in Nightly builds, run with -Z macro-backtrace for more info)
diff --git a/tests/compiletests/ui/arch/subgroup/subgroup_cluster_size_non_power_of_two_fail.stderr b/tests/compiletests/ui/arch/subgroup/subgroup_cluster_size_non_power_of_two_fail.stderr
index ed0fbbf35f..8ef6ee7cb1 100644
--- a/tests/compiletests/ui/arch/subgroup/subgroup_cluster_size_non_power_of_two_fail.stderr
+++ b/tests/compiletests/ui/arch/subgroup/subgroup_cluster_size_non_power_of_two_fail.stderr
@@ -1,27 +1,27 @@
 error[E0080]: evaluation panicked: `ClusterSize` must be a power of 2
-   --> $SPIRV_STD_SRC/arch/subgroup.rs:839:1
+   --> $SPIRV_STD_SRC/arch/subgroup.rs:829:1
     |
-839 | / macro_subgroup_op_clustered!(impl Integer, "OpGroupNonUniformIAdd", subgroup_clustered_i_add; r"
-840 | | An integer add group operation of all `value` operands contributed by active invocations in the group.
-841 | |
-842 | | Result Type must be a scalar or vector of integer type.
+829 | / macro_subgroup_op_clustered!(impl Integer, "OpGroupNonUniformIAdd", subgroup_clustered_i_add; r"
+830 | | An integer add group operation of all `value` operands contributed by active invocations in the group.
+831 | |
+832 | | Result Type must be a scalar or vector of integer type.
 ...   |
-855 | | * `ClusterSize` must not be greater than the size of the group
-856 | | ");
+845 | | * `ClusterSize` must not be greater than the size of the group
+846 | | ");
     | |__^ evaluation of `spirv_std::arch::subgroup_clustered_i_add::<5, u32, u32>::{constant#0}` failed here
     |
     = note: this error originates in the macro `$crate::panic::panic_2021` which comes from the expansion of the macro `macro_subgroup_op_clustered` (in Nightly builds, run with -Z macro-backtrace for more info)
 
 note: erroneous constant encountered
-   --> $SPIRV_STD_SRC/arch/subgroup.rs:839:1
+   --> $SPIRV_STD_SRC/arch/subgroup.rs:829:1
     |
-839 | / macro_subgroup_op_clustered!(impl Integer, "OpGroupNonUniformIAdd", subgroup_clustered_i_add; r"
-840 | | An integer add group operation of all `value` operands contributed by active invocations in the group.
-841 | |
-842 | | Result Type must be a scalar or vector of integer type.
+829 | / macro_subgroup_op_clustered!(impl Integer, "OpGroupNonUniformIAdd", subgroup_clustered_i_add; r"
+830 | | An integer add group operation of all `value` operands contributed by active invocations in the group.
+831 | |
+832 | | Result Type must be a scalar or vector of integer type.
 ...   |
-855 | | * `ClusterSize` must not be greater than the size of the group
-856 | | ");
+845 | | * `ClusterSize` must not be greater than the size of the group
+846 | | ");
     | |__^
     |
     = note: this note originates in the macro `macro_subgroup_op_clustered` (in Nightly builds, run with -Z macro-backtrace for more info)
diff --git a/tests/compiletests/ui/arch/subgroup/subgroup_elect.stderr b/tests/compiletests/ui/arch/subgroup/subgroup_elect.stderr
index 10c5624049..d18849b6f6 100644
--- a/tests/compiletests/ui/arch/subgroup/subgroup_elect.stderr
+++ b/tests/compiletests/ui/arch/subgroup/subgroup_elect.stderr
@@ -1,6 +1,6 @@
 %1 = OpFunction  %2  None %3
 %4 = OpLabel
-OpLine %5 174 13
+OpLine %5 164 13
 %6 = OpGroupNonUniformElect  %2  %7
 OpNoLine
 OpReturnValue %6
diff --git a/tests/compiletests/ui/arch/subgroup/subgroup_i_add_clustered.stderr b/tests/compiletests/ui/arch/subgroup/subgroup_i_add_clustered.stderr
index d1d520364d..4d06f8f632 100644
--- a/tests/compiletests/ui/arch/subgroup/subgroup_i_add_clustered.stderr
+++ b/tests/compiletests/ui/arch/subgroup/subgroup_i_add_clustered.stderr
@@ -1,7 +1,7 @@
 %1 = OpFunction  %2  None %3
 %4 = OpFunctionParameter  %2
 %5 = OpLabel
-OpLine %6 839 0
+OpLine %6 829 0
 %7 = OpGroupNonUniformIAdd  %2  %8 ClusteredReduce %4 %9
 OpNoLine
 OpReturnValue %7
diff --git a/tests/compiletests/ui/arch/subgroup/subgroup_i_add_exclusive_scan.stderr b/tests/compiletests/ui/arch/subgroup/subgroup_i_add_exclusive_scan.stderr
index 00a742fc98..4890551e39 100644
--- a/tests/compiletests/ui/arch/subgroup/subgroup_i_add_exclusive_scan.stderr
+++ b/tests/compiletests/ui/arch/subgroup/subgroup_i_add_exclusive_scan.stderr
@@ -1,7 +1,7 @@
 %1 = OpFunction  %2  None %3
 %4 = OpFunctionParameter  %2
 %5 = OpLabel
-OpLine %6 826 0
+OpLine %6 816 0
 %7 = OpGroupNonUniformIAdd  %2  %8 ExclusiveScan %4
 OpNoLine
 OpReturnValue %7
diff --git a/tests/compiletests/ui/arch/subgroup/subgroup_i_add_inclusive_scan.stderr b/tests/compiletests/ui/arch/subgroup/subgroup_i_add_inclusive_scan.stderr
index 694495fb73..a41a806a40 100644
--- a/tests/compiletests/ui/arch/subgroup/subgroup_i_add_inclusive_scan.stderr
+++ b/tests/compiletests/ui/arch/subgroup/subgroup_i_add_inclusive_scan.stderr
@@ -1,7 +1,7 @@
 %1 = OpFunction  %2  None %3
 %4 = OpFunctionParameter  %2
 %5 = OpLabel
-OpLine %6 826 0
+OpLine %6 816 0
 %7 = OpGroupNonUniformIAdd  %2  %8 InclusiveScan %4
 OpNoLine
 OpReturnValue %7
diff --git a/tests/compiletests/ui/arch/subgroup/subgroup_i_add_reduce.stderr b/tests/compiletests/ui/arch/subgroup/subgroup_i_add_reduce.stderr
index 4776bf1903..2e8d77e479 100644
--- a/tests/compiletests/ui/arch/subgroup/subgroup_i_add_reduce.stderr
+++ b/tests/compiletests/ui/arch/subgroup/subgroup_i_add_reduce.stderr
@@ -1,7 +1,7 @@
 %1 = OpFunction  %2  None %3
 %4 = OpFunctionParameter  %2
 %5 = OpLabel
-OpLine %6 826 0
+OpLine %6 816 0
 %7 = OpGroupNonUniformIAdd  %2  %8 Reduce %4
 OpNoLine
 OpReturnValue %7
diff --git a/tests/difftests/tests/arch/atomic_ops/atomic_ops-rust/src/lib.rs b/tests/difftests/tests/arch/atomic_ops/atomic_ops-rust/src/lib.rs
index f80b63d41f..0a59047970 100644
--- a/tests/difftests/tests/arch/atomic_ops/atomic_ops-rust/src/lib.rs
+++ b/tests/difftests/tests/arch/atomic_ops/atomic_ops-rust/src/lib.rs
@@ -31,9 +31,7 @@ pub fn main_cs(
     // Thread 0 stores the final values after all operations complete
     if tid == 0 {
         // Use atomic loads to ensure we read the final values
-        unsafe {
-            spirv_std::arch::workgroup_memory_barrier_with_group_sync();
-        }
+        spirv_std::arch::workgroup_memory_barrier_with_group_sync();
         output[0] = counters[0]; // Should be initial + 32
         output[1] = counters[1]; // Should be initial - 32
         output[2] = counters[2]; // Should be min(initial, 0)
diff --git a/tests/difftests/tests/arch/memory_barriers/memory_barriers-rust/src/lib.rs b/tests/difftests/tests/arch/memory_barriers/memory_barriers-rust/src/lib.rs
index 63d75c3aba..4a3979bf03 100644
--- a/tests/difftests/tests/arch/memory_barriers/memory_barriers-rust/src/lib.rs
+++ b/tests/difftests/tests/arch/memory_barriers/memory_barriers-rust/src/lib.rs
@@ -19,9 +19,7 @@ pub fn main_cs(
         shared[lid] = input[tid];
 
         // Workgroup barrier to ensure all threads have loaded their data
-        unsafe {
-            workgroup_memory_barrier_with_group_sync();
-        }
+        workgroup_memory_barrier_with_group_sync();
 
         // Perform operations on shared memory
         let mut result = shared[lid];
@@ -53,17 +51,13 @@ pub fn main_cs(
         }
 
         // Another barrier before writing back
-        unsafe {
-            workgroup_memory_barrier_with_group_sync();
-        }
+        workgroup_memory_barrier_with_group_sync();
 
         // Write result back to shared memory
         shared[lid] = result;
 
         // Memory barrier to ensure writes are visible
-        unsafe {
-            workgroup_memory_barrier_with_group_sync();
-        }
+        workgroup_memory_barrier_with_group_sync();
 
         // Final read and output
         output[tid] = shared[lid];
diff --git a/tests/difftests/tests/arch/workgroup_memory/workgroup_memory-rust/src/shader.rs b/tests/difftests/tests/arch/workgroup_memory/workgroup_memory-rust/src/shader.rs
index d743a1cea8..45bc392497 100644
--- a/tests/difftests/tests/arch/workgroup_memory/workgroup_memory-rust/src/shader.rs
+++ b/tests/difftests/tests/arch/workgroup_memory/workgroup_memory-rust/src/shader.rs
@@ -14,9 +14,7 @@ pub fn main_cs(
     shared[lid] = input[lid];
 
     // Synchronize to ensure all threads have loaded
-    unsafe {
-        workgroup_memory_barrier_with_group_sync();
-    }
+    workgroup_memory_barrier_with_group_sync();
 
     // Each thread sums its value with its neighbor (reduction step)
     if lid < 32 {
@@ -24,49 +22,37 @@ pub fn main_cs(
     }
 
     // Synchronize again
-    unsafe {
-        workgroup_memory_barrier_with_group_sync();
-    }
+    workgroup_memory_barrier_with_group_sync();
 
     if lid < 16 {
         shared[lid] += shared[lid + 16];
     }
 
-    unsafe {
-        workgroup_memory_barrier_with_group_sync();
-    }
+    workgroup_memory_barrier_with_group_sync();
 
     if lid < 8 {
         shared[lid] += shared[lid + 8];
     }
 
-    unsafe {
-        workgroup_memory_barrier_with_group_sync();
-    }
+    workgroup_memory_barrier_with_group_sync();
 
     if lid < 4 {
         shared[lid] += shared[lid + 4];
     }
 
-    unsafe {
-        workgroup_memory_barrier_with_group_sync();
-    }
+    workgroup_memory_barrier_with_group_sync();
 
     if lid < 2 {
         shared[lid] += shared[lid + 2];
     }
 
-    unsafe {
-        workgroup_memory_barrier_with_group_sync();
-    }
+    workgroup_memory_barrier_with_group_sync();
 
     if lid < 1 {
         shared[lid] += shared[lid + 1];
     }
 
-    unsafe {
-        workgroup_memory_barrier_with_group_sync();
-    }
+    workgroup_memory_barrier_with_group_sync();
 
     // Write final result
     if lid == 0 {