From 8f9d76550bc15e50372020a8270ecc1b969c151f Mon Sep 17 00:00:00 2001 From: Caleb Zulawski Date: Sat, 23 Nov 2024 14:31:20 -0500 Subject: [PATCH 1/4] Add simd_relaxed_fma intrinsic --- src/intrinsics/simd.rs | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/src/intrinsics/simd.rs b/src/intrinsics/simd.rs index f787b8a6fd9..e0ebe30752a 100644 --- a/src/intrinsics/simd.rs +++ b/src/intrinsics/simd.rs @@ -415,7 +415,8 @@ pub(super) fn codegen_simd_intrinsic_call<'tcx>( }); } - sym::simd_fma => { + // FIXME: simd_relaxed_fma doesn't relax to non-fused multiply-add + sym::simd_fma | sym::simd_relaxed_fma => { intrinsic_args!(fx, args => (a, b, c); intrinsic); if !a.layout().ty.is_simd() { From b9e2bdd050f1e076fc3a215e2a02e19b3f9a3fc1 Mon Sep 17 00:00:00 2001 From: Taiki Endo Date: Mon, 25 Nov 2024 00:36:22 +0900 Subject: [PATCH 2/4] Fix clobber_abi in RV32E and RV64E inline assembly --- src/inline_asm.rs | 11 ++++++++--- 1 file changed, 8 insertions(+), 3 deletions(-) diff --git a/src/inline_asm.rs b/src/inline_asm.rs index 70176754f33..73ee47b2c63 100644 --- a/src/inline_asm.rs +++ b/src/inline_asm.rs @@ -472,9 +472,14 @@ impl<'tcx> InlineAssemblyGenerator<'_, 'tcx> { let mut new_slot = |x| new_slot_fn(&mut slot_size, x); // Allocate stack slots for saving clobbered registers - let abi_clobber = InlineAsmClobberAbi::parse(self.arch, &self.tcx.sess.target, sym::C) - .unwrap() - .clobbered_regs(); + let abi_clobber = InlineAsmClobberAbi::parse( + self.arch, + &self.tcx.sess.target, + &self.tcx.sess.unstable_target_features, + sym::C, + ) + .unwrap() + .clobbered_regs(); for (i, reg) in self.registers.iter().enumerate().filter_map(|(i, r)| r.map(|r| (i, r))) { let mut need_save = true; // If the register overlaps with a register clobbered by function call, then From e5bc7e38c3c9d75a993435e4dfdcc15644e2262b Mon Sep 17 00:00:00 2001 From: Taiki Endo Date: Tue, 26 Nov 2024 03:10:22 +0900 Subject: [PATCH 3/4] Fix handling of x18 in AArch64 inline assembly on ohos/trusty or with -Zfixed-x18 --- src/inline_asm.rs | 11 ++++++++--- 1 file changed, 8 insertions(+), 3 deletions(-) diff --git a/src/inline_asm.rs b/src/inline_asm.rs index 7bc500b1814..d74c366a87f 100644 --- a/src/inline_asm.rs +++ b/src/inline_asm.rs @@ -476,9 +476,14 @@ impl<'tcx> InlineAssemblyGenerator<'_, 'tcx> { let mut new_slot = |x| new_slot_fn(&mut slot_size, x); // Allocate stack slots for saving clobbered registers - let abi_clobber = InlineAsmClobberAbi::parse(self.arch, &self.tcx.sess.target, sym::C) - .unwrap() - .clobbered_regs(); + let abi_clobber = InlineAsmClobberAbi::parse( + self.arch, + &self.tcx.sess.target, + &self.tcx.sess.unstable_target_features, + sym::C, + ) + .unwrap() + .clobbered_regs(); for (i, reg) in self.registers.iter().enumerate().filter_map(|(i, r)| r.map(|r| (i, r))) { let mut need_save = true; // If the register overlaps with a register clobbered by function call, then From a2e9aac9057a99ba772860c3b6e2850172b67877 Mon Sep 17 00:00:00 2001 From: clubby789 Date: Wed, 27 Nov 2024 17:52:57 +0000 Subject: [PATCH 4/4] Replace `Symbol::intern` calls with preinterned symbols --- src/driver/jit.rs | 9 ++------- src/lib.rs | 9 ++------- 2 files changed, 4 insertions(+), 14 deletions(-) diff --git a/src/driver/jit.rs b/src/driver/jit.rs index 0d62a13b472..ae9578eeffb 100644 --- a/src/driver/jit.rs +++ b/src/driver/jit.rs @@ -74,7 +74,7 @@ fn create_jit_module( jit_builder.symbol("__clif_jit_fn", clif_jit_fn as *const u8); let mut jit_module = UnwindModule::new(JITModule::new(jit_builder), false); - let cx = crate::CodegenCx::new(tcx, jit_module.isa(), false, Symbol::intern("dummy_cgu_name")); + let cx = crate::CodegenCx::new(tcx, jit_module.isa(), false, sym::dummy_cgu_name); crate::allocator::codegen(tcx, &mut jit_module); @@ -276,12 +276,7 @@ fn jit_fn(instance_ptr: *const Instance<'static>, trampoline_ptr: *const u8) -> jit_module.module.prepare_for_function_redefine(func_id).unwrap(); - let mut cx = crate::CodegenCx::new( - tcx, - jit_module.isa(), - false, - Symbol::intern("dummy_cgu_name"), - ); + let mut cx = crate::CodegenCx::new(tcx, jit_module.isa(), false, sym::dummy_cgu_name); codegen_and_compile_fn(tcx, &mut cx, &mut Context::new(), jit_module, instance); assert!(cx.global_asm.is_empty()); diff --git a/src/lib.rs b/src/lib.rs index e6f6ae30581..cac9975f04c 100644 --- a/src/lib.rs +++ b/src/lib.rs @@ -189,18 +189,13 @@ impl CodegenBackend for CraneliftCodegenBackend { // FIXME return the actually used target features. this is necessary for #[cfg(target_feature)] if sess.target.arch == "x86_64" && sess.target.os != "none" { // x86_64 mandates SSE2 support - vec![Symbol::intern("fxsr"), sym::sse, Symbol::intern("sse2")] + vec![sym::fsxr, sym::sse, sym::sse2] } else if sess.target.arch == "aarch64" { match &*sess.target.os { "none" => vec![], // On macOS the aes, sha2 and sha3 features are enabled by default and ring // fails to compile on macOS when they are not present. - "macos" => vec![ - sym::neon, - Symbol::intern("aes"), - Symbol::intern("sha2"), - Symbol::intern("sha3"), - ], + "macos" => vec![sym::neon, sym::aes, sym::sha2, sym::sha3], // AArch64 mandates Neon support _ => vec![sym::neon], }