Rollup merge of #144192 - RalfJung:atomicrmw-ptr, r=nikic

atomicrmw on pointers: move integer-pointer cast hacks into backend

Conceptually, we want to have atomic operations on pointers of the form `fn atomic_add(ptr: *mut T, offset: usize, ...)`. However, LLVM does not directly support such operations (https://github.com/llvm/llvm-project/issues/120837), so we have to cast the `offset` to a pointer somewhere.

This PR moves that hack into the LLVM backend, so that the standard library, intrinsic, and Miri all work with the conceptual operation we actually want. Hopefully, one day LLVM will gain a way to represent these operations without integer-pointer casts, and then the hack will disappear entirely.

Cc ```@nikic``` -- this is the best we can do right now, right?
Fixes https://github.com/rust-lang/rust/issues/134617

1 files changed, 3 insertions, 3 deletions

diff --git a/tests/codegen-llvm/atomicptr.rs b/tests/codegen-llvm/atomicptr.rs
index 4819af40ca2..ce6c4aa0d2b 100644
--- a/tests/codegen-llvm/atomicptr.rs
+++ b/tests/codegen-llvm/atomicptr.rs
@@ -1,5 +1,5 @@
 // LLVM does not support some atomic RMW operations on pointers, so inside codegen we lower those
-// to integer atomics, surrounded by casts to and from integer type.
+// to integer atomics, followed by an inttoptr cast.
 // This test ensures that we do the round-trip correctly for AtomicPtr::fetch_byte_add, and also
 // ensures that we do not have such a round-trip for AtomicPtr::swap, because LLVM supports pointer
 // arguments to `atomicrmw xchg`.
@@ -20,8 +20,8 @@ pub fn helper(_: usize) {}
 // CHECK-LABEL: @atomicptr_fetch_byte_add
 #[no_mangle]
 pub fn atomicptr_fetch_byte_add(a: &AtomicPtr<u8>, v: usize) -> *mut u8 {
-    // CHECK: %[[INTPTR:.*]] = ptrtoint ptr %{{.*}} to [[USIZE]]
-    // CHECK-NEXT: %[[RET:.*]] = atomicrmw add ptr %{{.*}}, [[USIZE]] %[[INTPTR]]
+    // CHECK: llvm.lifetime.start
+    // CHECK-NEXT: %[[RET:.*]] = atomicrmw add ptr %{{.*}}, [[USIZE]] %v
     // CHECK-NEXT: inttoptr [[USIZE]] %[[RET]] to ptr
     a.fetch_byte_add(v, Relaxed)
 }