Auto merge of #120718 - saethlin:reasonable-fast-math, r=nnethercote

Add "algebraic" fast-math intrinsics, based on fast-math ops that cannot return poison

Setting all of LLVM's fast-math flags makes our fast-math intrinsics very dangerous, because some inputs are UB. This set of flags permits common algebraic transformations, but according to the [LangRef](https://llvm.org/docs/LangRef.html#fastmath), only the flags `nnan` (no nans) and `ninf` (no infs) can produce poison.

And this uses the algebraic float ops to fix https://github.com/rust-lang/rust/issues/120720

cc `@orlp`

1 files changed, 25 insertions, 0 deletions

diff --git a/compiler/rustc_codegen_gcc/src/builder.rs b/compiler/rustc_codegen_gcc/src/builder.rs
index 42e61b3ccb5..5f1e4538376 100644
--- a/compiler/rustc_codegen_gcc/src/builder.rs
+++ b/compiler/rustc_codegen_gcc/src/builder.rs
@@ -705,6 +705,31 @@ impl<'a, 'gcc, 'tcx> BuilderMethods<'a, 'tcx> for Builder<'a, 'gcc, 'tcx> {
         self.frem(lhs, rhs)
     }
 
+    fn fadd_algebraic(&mut self, lhs: RValue<'gcc>, rhs: RValue<'gcc>) -> RValue<'gcc> {
+        // NOTE: it seems like we cannot enable fast-mode for a single operation in GCC.
+        lhs + rhs
+    }
+
+    fn fsub_algebraic(&mut self, lhs: RValue<'gcc>, rhs: RValue<'gcc>) -> RValue<'gcc> {
+        // NOTE: it seems like we cannot enable fast-mode for a single operation in GCC.
+        lhs - rhs
+    }
+
+    fn fmul_algebraic(&mut self, lhs: RValue<'gcc>, rhs: RValue<'gcc>) -> RValue<'gcc> {
+        // NOTE: it seems like we cannot enable fast-mode for a single operation in GCC.
+        lhs * rhs
+    }
+
+    fn fdiv_algebraic(&mut self, lhs: RValue<'gcc>, rhs: RValue<'gcc>) -> RValue<'gcc> {
+        // NOTE: it seems like we cannot enable fast-mode for a single operation in GCC.
+        lhs / rhs
+    }
+
+    fn frem_algebraic(&mut self, lhs: RValue<'gcc>, rhs: RValue<'gcc>) -> RValue<'gcc> {
+        // NOTE: it seems like we cannot enable fast-mode for a single operation in GCC.
+        self.frem(lhs, rhs)
+    }
+
     fn checked_binop(&mut self, oop: OverflowOp, typ: Ty<'_>, lhs: Self::Value, rhs: Self::Value) -> (Self::Value, Self::Value) {
         self.gcc_checked_binop(oop, typ, lhs, rhs)
     }