[llvm] Allow Fast and Tail CC in UEFI (#138361)

Prabhuk · web-flow · commit 0a07e6beaf05 · 2025-05-06T15:48:05.000-07:00
Allow tail call and fastcc optimizations for UEFI X86_64 targets.
diff --git a/llvm/lib/Target/X86/X86Subtarget.h b/llvm/lib/Target/X86/X86Subtarget.h
@@ -350,9 +350,9 @@ class X86Subtarget final : public X86GenSubtargetInfo {
     switch (CC) {
     // On Win64, all these conventions just use the default convention.
     case CallingConv::C:
-      return isTargetWin64() || isTargetUEFI64();
     case CallingConv::Fast:
     case CallingConv::Tail:
+      return isTargetWin64() || isTargetUEFI64();
     case CallingConv::Swift:
     case CallingConv::SwiftTail:
     case CallingConv::X86_FastCall:
diff --git a/llvm/test/CodeGen/X86/musttail-tailcc.ll b/llvm/test/CodeGen/X86/musttail-tailcc.ll
@@ -1,5 +1,6 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown | FileCheck %s -check-prefix=X64
+; RUN: llc < %s -mtriple=x86_64-uefi | FileCheck %s -check-prefix=X64
 ; RUN: llc < %s -mtriple=i686-unknown-unknown   | FileCheck %s -check-prefix=X32
 
 ; tailcc will turn all of these musttail calls into tail calls.
diff --git a/llvm/test/CodeGen/X86/tailcall-tailcc.ll b/llvm/test/CodeGen/X86/tailcall-tailcc.ll
@@ -1,5 +1,7 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown | FileCheck %s -check-prefix=X64
+; RUN: llc < %s -mtriple=x86_64-windows-msvc | FileCheck %s -check-prefix=UEFI64
+; RUN: llc < %s -mtriple=x86_64-uefi | FileCheck %s -check-prefix=UEFI64
 ; RUN: llc < %s -mtriple=i686-unknown-unknown   | FileCheck %s -check-prefix=X32
 
 ; With -tailcallopt, CodeGen guarantees a tail call optimization
@@ -16,6 +18,14 @@ define dso_local tailcc i32 @tailcaller(i32 %in1, i32 %in2) nounwind {
 ; X64-NEXT:    popq %rax
 ; X64-NEXT:    jmp tailcallee # TAILCALL
 ;
+; UEFI64-LABEL: tailcaller:
+; UEFI64:       # %bb.0: # %entry
+; UEFI64-NEXT:    subq	$40, %rsp
+; UEFI64-NEXT:    movl %ecx, %r8d
+; UEFI64-NEXT:    movl %edx, %r9d
+; UEFI64-NEXT:    addq	$40, %rsp
+; UEFI64-NEXT:    jmp tailcallee # TAILCALL
+;
 ; X32-LABEL: tailcaller:
 ; X32:       # %bb.0: # %entry
 ; X32-NEXT:    subl $16, %esp
@@ -39,6 +49,10 @@ define tailcc noalias ptr @noalias_caller() nounwind {
 ; X64-NEXT:    popq %rax
 ; X64-NEXT:    jmp alias_callee # TAILCALL
 ;
+; UEFI64-LABEL: noalias_caller:
+; UEFI64:       # %bb.0:
+; UEFI64-NEXT:    jmp alias_callee # TAILCALL
+;
 ; X32-LABEL: noalias_caller:
 ; X32:       # %bb.0:
 ; X32-NEXT:    jmp alias_callee # TAILCALL
@@ -55,6 +69,10 @@ define dso_local tailcc ptr @alias_caller() nounwind {
 ; X64-NEXT:    popq %rax
 ; X64-NEXT:    jmp noalias_callee # TAILCALL
 ;
+; UEFI64-LABEL: alias_caller:
+; UEFI64:       # %bb.0:
+; UEFI64-NEXT:    jmp noalias_callee # TAILCALL
+;
 ; X32-LABEL: alias_caller:
 ; X32:       # %bb.0:
 ; X32-NEXT:    jmp noalias_callee # TAILCALL
@@ -71,6 +89,10 @@ define dso_local tailcc i32 @ret_undef() nounwind {
 ; X64-NEXT:    popq %rax
 ; X64-NEXT:    jmp i32_callee # TAILCALL
 ;
+; UEFI64-LABEL: ret_undef:
+; UEFI64:       # %bb.0:
+; UEFI64-NEXT:    jmp i32_callee # TAILCALL
+;
 ; X32-LABEL: ret_undef:
 ; X32:       # %bb.0:
 ; X32-NEXT:    jmp i32_callee # TAILCALL
@@ -87,6 +109,10 @@ define dso_local tailcc i32 @noret() nounwind {
 ; X64-NEXT:    popq %rax
 ; X64-NEXT:    jmp does_not_return # TAILCALL
 ;
+; UEFI64-LABEL: noret:
+; UEFI64:       # %bb.0:
+; UEFI64-NEXT:    jmp does_not_return # TAILCALL
+;
 ; X32-LABEL: noret:
 ; X32:       # %bb.0:
 ; X32-NEXT:    jmp does_not_return # TAILCALL
@@ -103,6 +129,16 @@ define dso_local tailcc void @void_test(i32, i32, i32, i32) {
 ; X64-NEXT:    .cfi_def_cfa_offset 8
 ; X64-NEXT:    jmp void_test # TAILCALL
 ;
+; UEFI64-LABEL: void_test:
+; UEFI64:       # %bb.0: # %entry
+; UEFI64-NEXT:   subq	$40, %rsp
+; UEFI64-NEXT:   .seh_stackalloc 40
+; UEFI64-NEXT:   .seh_endprologue
+; UEFI64-NEXT:   .seh_startepilogue
+; UEFI64-NEXT:   addq	$40, %rsp
+; UEFI64-NEXT:   .seh_endepilogue
+; UEFI64-NEXT:   jmp void_test # TAILCALL
+;
 ; X32-LABEL: void_test:
 ; X32:       # %bb.0: # %entry
 ; X32-NEXT:    pushl %esi
@@ -133,6 +169,16 @@ define dso_local tailcc i1 @i1test(i32, i32, i32, i32) {
 ; X64-NEXT:    .cfi_def_cfa_offset 8
 ; X64-NEXT:    jmp i1test # TAILCALL
 ;
+; UEFI64-LABEL: i1test:
+; UEFI64:       # %bb.0: # %entry
+; UEFI64-NEXT:   subq	$40, %rsp
+; UEFI64-NEXT:   .seh_stackalloc 40
+; UEFI64-NEXT:   .seh_endprologue
+; UEFI64-NEXT:   .seh_startepilogue
+; UEFI64-NEXT:   addq	$40, %rsp
+; UEFI64-NEXT:   .seh_endepilogue
+; UEFI64-NEXT:   jmp i1test # TAILCALL
+;
 ; X32-LABEL: i1test:
 ; X32:       # %bb.0: # %entry
 ; X32-NEXT:    pushl %esi
diff --git a/llvm/test/CodeGen/X86/tailcc-fastcc.ll b/llvm/test/CodeGen/X86/tailcc-fastcc.ll
@@ -1,5 +1,6 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc -tailcallopt < %s -mtriple=x86_64-unknown-unknown | FileCheck %s -check-prefix=X64
+; RUN: llc -tailcallopt < %s -mtriple=x86_64-uefi | FileCheck %s -check-prefix=UEFI64
 ; RUN: llc -tailcallopt < %s -mtriple=i686-unknown-unknown   | FileCheck %s -check-prefix=X32
 
 ; llc -tailcallopt should not enable tail calls from fastcc to tailcc or vice versa
@@ -15,6 +16,14 @@ define fastcc i32 @tailcaller1(i32 %in1, i32 %in2) nounwind {
 ; X64-NEXT:    callq tailcallee1@PLT
 ; X64-NEXT:    retq $8
 ;
+; UEFI64-LABEL: tailcaller1:
+; UEFI64:       # %bb.0: # %entry
+; UEFI64-NEXT:    subq	$40, %rsp
+; UEFI64-NEXT:    movl %ecx, %r8d
+; UEFI64-NEXT:    movl %edx, %r9d
+; UEFI64-NEXT:    callq tailcallee1
+; UEFI64-NEXT:    retq $40
+;
 ; X32-LABEL: tailcaller1:
 ; X32:       # %bb.0: # %entry
 ; X32-NEXT:    pushl %edx
@@ -37,6 +46,14 @@ define tailcc i32 @tailcaller2(i32 %in1, i32 %in2) nounwind {
 ; X64-NEXT:    callq tailcallee2@PLT
 ; X64-NEXT:    retq $8
 ;
+; UEFI64-LABEL: tailcaller2:
+; UEFI64:       # %bb.0: # %entry
+; UEFI64-NEXT:    subq	$40, %rsp
+; UEFI64-NEXT:    movl %ecx, %r8d
+; UEFI64-NEXT:    movl %edx, %r9d
+; UEFI64-NEXT:    callq tailcallee2
+; UEFI64-NEXT:    retq $40
+;
 ; X32-LABEL: tailcaller2:
 ; X32:       # %bb.0: # %entry
 ; X32-NEXT:    pushl %edx
diff --git a/llvm/test/CodeGen/X86/tailcc-fastisel.ll b/llvm/test/CodeGen/X86/tailcc-fastisel.ll
@@ -1,11 +1,13 @@
-; RUN: llc < %s -mtriple=x86_64-apple-darwin -fast-isel -fast-isel-abort=1 | FileCheck %s
+; RUN: llc < %s -mtriple=x86_64-apple-darwin -fast-isel -fast-isel-abort=1 | FileCheck %s  -check-prefix=DARWIN64
+; RUN: llc < %s -mtriple=x86_64-uefi -fast-isel -fast-isel-abort=1 | FileCheck %s -check-prefix=UEFI64
 
 %0 = type { i64, i32, ptr }
 
 define tailcc ptr @"visit_array_aux<`Reference>"(%0 %arg, i32 %arg1) nounwind {
 fail:                                             ; preds = %entry
   %tmp20 = tail call tailcc ptr @"visit_array_aux<`Reference>"(%0 %arg, i32 undef) ; <ptr> [#uses=1]
-; CHECK: jmp "_visit_array_aux<`Reference>" ## TAILCALL
+; DARWIN64: jmp "_visit_array_aux<`Reference>" ## TAILCALL
+; UEFI64:   jmp	"visit_array_aux<`Reference>"  # TAILCALL
   ret ptr %tmp20
 }
 
diff --git a/llvm/test/CodeGen/X86/tailccstack64.ll b/llvm/test/CodeGen/X86/tailccstack64.ll
@@ -1,5 +1,6 @@
 ; RUN: llc < %s -mcpu=generic -mtriple=x86_64-linux -post-RA-scheduler=true | FileCheck %s
 ; RUN: llc < %s -mcpu=generic -mtriple=x86_64-win32 -post-RA-scheduler=true | FileCheck %s
+; RUN: llc < %s -mcpu=generic -mtriple=x86_64-uefi -post-RA-scheduler=true | FileCheck %s
 
 ; FIXME: Redundant unused stack allocation could be eliminated.
 ; CHECK: subq  ${{24|72|80}}, %rsp
diff --git a/llvm/test/CodeGen/X86/uefi-fastcc.ll b/llvm/test/CodeGen/X86/uefi-fastcc.ll
@@ -0,0 +1,19 @@
+; RUN: llc < %s -mtriple=x86_64-uefi | FileCheck %s -check-prefix=UEFIFAST64
+; RUN: llc < %s -mtriple=x86_64-windows-msvc | FileCheck %s -check-prefix=UEFIFAST64
+
+declare fastcc i32 @fastcallee1(i32 %a1, i32 %a2, i32 %a3, i32 %a4)
+
+define fastcc i32 @fastcaller1(i32 %in1, i32 %in2) nounwind {
+;; Test that the caller allocates stack space for callee to spill the register arguments.
+; UEFIFAST64-LABEL: fastcaller1:
+; UEFIFAST64:       # %bb.0: # %entry
+; UEFIFAST64-NEXT:    subq	$40, %rsp
+; UEFIFAST64-NEXT:    movl %ecx, %r8d
+; UEFIFAST64-NEXT:    movl %edx, %r9d
+; UEFIFAST64-NEXT:    callq fastcallee1
+; UEFIFAST64-NEXT:    addq	$40, %rsp
+; UEFIFAST64-NEXT:    retq
+entry:
+  %tmp11 = call fastcc i32 @fastcallee1(i32 %in1, i32 %in2, i32 %in1, i32 %in2)
+  ret i32 %tmp11
+}
