e6bf48d110
As of revea222be0d, LLVMs assembler will actually try to honour the "fill value" part of p2align directives. X86 printed these as 0x90, which isn't actually what it wanted: we want multi-byte nops for .text padding. Compiling via a textual assembly file produces single-byte nop padding sinceea222be0dbut the built-in assembler will produce multi-byte nops. This divergent behaviour is undesirable. To fix: don't set the byte padding field for x86, which allows the assembler to pick multi-byte nops. Test that we get the same multi-byte padding when compiled via textual assembly or directly to object file. Added same-align-bytes-with-llasm-llobj.ll to that effect, updated numerous other tests to not contain check-lines for the explicit padding.
121 lines
4.2 KiB
LLVM
121 lines
4.2 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
|
|
; RUN: llc < %s -mtriple=i686-w64-windows-gnu | FileCheck %s
|
|
|
|
@a = global i32 0, align 4
|
|
@b = global float 0.000000e+00, align 4
|
|
@d = global float 0.000000e+00, align 4
|
|
@f = global i32 0, align 4
|
|
@g = global float 0.000000e+00, align 4
|
|
@e = global i32 0, align 4
|
|
@c = global ptr null, align 4
|
|
|
|
; The FP stack should be preserved across the call to __alloca.
|
|
define void @h(float %i) {
|
|
; CHECK-LABEL: h:
|
|
; CHECK: # %bb.0: # %entry
|
|
; CHECK-NEXT: pushl %ebp
|
|
; CHECK-NEXT: .cfi_def_cfa_offset 8
|
|
; CHECK-NEXT: .cfi_offset %ebp, -8
|
|
; CHECK-NEXT: movl %esp, %ebp
|
|
; CHECK-NEXT: .cfi_def_cfa_register %ebp
|
|
; CHECK-NEXT: pushl %esi
|
|
; CHECK-NEXT: andl $-16, %esp
|
|
; CHECK-NEXT: subl $32, %esp
|
|
; CHECK-NEXT: movl %esp, %esi
|
|
; CHECK-NEXT: .cfi_offset %esi, -12
|
|
; CHECK-NEXT: flds 8(%ebp)
|
|
; CHECK-NEXT: movl _a, %ecx
|
|
; CHECK-NEXT: leal 3(%ecx), %eax
|
|
; CHECK-NEXT: andl $-4, %eax
|
|
; CHECK-NEXT: calll __alloca
|
|
; CHECK-NEXT: movl %esp, %eax
|
|
; CHECK-NEXT: andl $-16, %eax
|
|
; CHECK-NEXT: movl %eax, %esp
|
|
; CHECK-NEXT: fsts 8(%esi) # 4-byte Folded Spill
|
|
; CHECK-NEXT: fadds _b
|
|
; CHECK-NEXT: fsts _d
|
|
; CHECK-NEXT: fld1
|
|
; CHECK-NEXT: fldz
|
|
; CHECK-NEXT: testl %ecx, %ecx
|
|
; CHECK-NEXT: fld %st(0)
|
|
; CHECK-NEXT: fld %st(2)
|
|
; CHECK-NEXT: je LBB0_2
|
|
; CHECK-NEXT: # %bb.1: # %for.body.preheader
|
|
; CHECK-NEXT: fstp %st(1)
|
|
; CHECK-NEXT: fstp %st(0)
|
|
; CHECK-NEXT: movl _f, %ecx
|
|
; CHECK-NEXT: flds (%eax,%ecx,4)
|
|
; CHECK-NEXT: fld %st(3)
|
|
; CHECK-NEXT: LBB0_2: # %for.cond1.preheader
|
|
; CHECK-NEXT: movl _e, %ecx
|
|
; CHECK-NEXT: movl %ecx, 12(%esi)
|
|
; CHECK-NEXT: fildl 12(%esi)
|
|
; CHECK-NEXT: movl _c, %edx
|
|
; CHECK-NEXT: jmp LBB0_3
|
|
; CHECK-NEXT: .p2align 4
|
|
; CHECK-NEXT: LBB0_5: # %for.inc
|
|
; CHECK-NEXT: # in Loop: Header=BB0_3 Depth=1
|
|
; CHECK-NEXT: fxch %st(5)
|
|
; CHECK-NEXT: fadd %st(4), %st
|
|
; CHECK-NEXT: fxch %st(5)
|
|
; CHECK-NEXT: LBB0_3: # %for.cond1
|
|
; CHECK-NEXT: # =>This Inner Loop Header: Depth=1
|
|
; CHECK-NEXT: fld %st(5)
|
|
; CHECK-NEXT: fmul %st(4), %st
|
|
; CHECK-NEXT: fdiv %st(2), %st
|
|
; CHECK-NEXT: fadd %st(3), %st
|
|
; CHECK-NEXT: fsts _g
|
|
; CHECK-NEXT: fxch %st(1)
|
|
; CHECK-NEXT: fucom %st(1)
|
|
; CHECK-NEXT: fstp %st(1)
|
|
; CHECK-NEXT: fnstsw %ax
|
|
; CHECK-NEXT: # kill: def $ah killed $ah killed $ax
|
|
; CHECK-NEXT: sahf
|
|
; CHECK-NEXT: jbe LBB0_5
|
|
; CHECK-NEXT: # %bb.4: # %if.then
|
|
; CHECK-NEXT: # in Loop: Header=BB0_3 Depth=1
|
|
; CHECK-NEXT: flds 8(%esi) # 4-byte Folded Reload
|
|
; CHECK-NEXT: fstps (%edx,%ecx,4)
|
|
; CHECK-NEXT: jmp LBB0_5
|
|
entry:
|
|
%0 = load i32, ptr @a, align 4
|
|
%1 = alloca i8, i32 %0, align 16
|
|
%2 = load float, ptr @b, align 4
|
|
%add = fadd float %2, %i
|
|
store float %add, ptr @d, align 4
|
|
%tobool.not = icmp eq i32 %0, 0
|
|
br i1 %tobool.not, label %for.cond1.preheader, label %for.body.preheader
|
|
|
|
for.body.preheader: ; preds = %entry
|
|
%3 = load i32, ptr @f, align 4
|
|
%arrayidx.le = getelementptr inbounds float, ptr %1, i32 %3
|
|
%4 = load float, ptr %arrayidx.le, align 4
|
|
br label %for.cond1.preheader
|
|
|
|
for.cond1.preheader: ; preds = %for.body.preheader, %entry
|
|
%k.0.lcssa = phi float [ %4, %for.body.preheader ], [ undef, %entry ]
|
|
%l.0.lcssa = phi float [ %add, %for.body.preheader ], [ 1.000000e+00, %entry ]
|
|
%5 = load i32, ptr @e, align 4
|
|
%conv = sitofp i32 %5 to float
|
|
%6 = load ptr, ptr @c, align 4
|
|
%arrayidx4 = getelementptr inbounds float, ptr %6, i32 %5
|
|
br label %for.cond1
|
|
|
|
for.cond1: ; preds = %for.inc, %for.cond1.preheader
|
|
%m.0 = phi float [ %add5, %for.inc ], [ %add, %for.cond1.preheader ]
|
|
%mul = fmul float %m.0, 0.000000e+00
|
|
%div = fdiv float %mul, %l.0.lcssa
|
|
%add2 = fadd float %k.0.lcssa, %div
|
|
store float %add2, ptr @g, align 4
|
|
%cmp = fcmp olt float %add2, %conv
|
|
br i1 %cmp, label %if.then, label %for.inc
|
|
|
|
if.then: ; preds = %for.cond1
|
|
store float %i, ptr %arrayidx4, align 4
|
|
br label %for.inc
|
|
|
|
for.inc: ; preds = %if.then, %for.cond1
|
|
%add5 = fadd float %m.0, 1.000000e+00
|
|
br label %for.cond1
|
|
}
|