1
0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2025-02-01 05:01:59 +01:00
Chandler Carruth 3e714e0291 [x86] Extend the manual ISel of add and sub with both RMW memory
operands and used flags to support matching immediate operands.

This is a bit trickier than register operands, and we still want to fall
back on a register operands even for things that appear to be
"immediates" when they won't actually select into the operation's
immediate operand. This also requires us to handle things like selecting
`sub` vs. `add` to minimize the number of bits needed to represent the
immediate, and picking the shortest immediate encoding. In order to
that, we in turn need to scan to make sure that CF isn't used as it will
get inverted.

The end result seems very nice though, and we're now generating
optimal instruction sequences for these patterns IMO.

A follow-up patch will further expand this to other operations with RMW
memory operands. But handing `add` and `sub` are useful starting points
to flesh out the machinery and make sure interesting and complex cases
can be handled.

Thanks to Craig Topper who provided a few fixes and improvements to this
patch in addition to the review!

Differential Revision: https://reviews.llvm.org/D37139

llvm-svn: 312764
2017-09-07 23:54:24 +00:00

101 lines
2.9 KiB
LLVM

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -o - %s | FileCheck %s
target triple = "i386-unknown-linux-gnu"
@a = external global i32, align 4
@d = external global i32*, align 4
@k = external global i32**, align 4
@j = external global i32***, align 4
@h = external global i32, align 4
@c = external global i32, align 4
@i = external global i32, align 4
@b = external global i32, align 4
@f = external global i64, align 8
@e = external global i64, align 8
@g = external global i32, align 4
declare i32 @fn1(i32 returned) optsize readnone
declare i32 @main() optsize
declare i32 @putchar(i32) nounwind
define void @fn2() nounwind optsize {
; CHECK-LABEL: fn2:
; CHECK: # BB#0: # %entry
; CHECK-NEXT: pushl %ebx
; CHECK-NEXT: subl $8, %esp
; CHECK-NEXT: movl $48, (%esp)
; CHECK-NEXT: calll putchar
; CHECK-NEXT: movl h, %eax
; CHECK-NEXT: movl c, %ecx
; CHECK-NEXT: movl j, %edx
; CHECK-NEXT: movl (%edx), %edx
; CHECK-NEXT: movl (%edx), %edx
; CHECK-NEXT: xorl %ebx, %ebx
; CHECK-NEXT: cmpl (%edx), %ecx
; CHECK-NEXT: setg %bl
; CHECK-NEXT: xorl %ecx, %ecx
; CHECK-NEXT: cmpl %ebx, i
; CHECK-NEXT: setg %cl
; CHECK-NEXT: movl %ecx, b
; CHECK-NEXT: xorl %edx, %edx
; CHECK-NEXT: cmpl %ecx, %eax
; CHECK-NEXT: setg %dl
; CHECK-NEXT: xorl %edx, a
; CHECK-NEXT: movl d, %eax
; CHECK-NEXT: movl (%eax), %eax
; CHECK-NEXT: andl %eax, e
; CHECK-NEXT: sarl $31, %eax
; CHECK-NEXT: andl %eax, e+4
; CHECK-NEXT: decl g
; CHECK-NEXT: addl $1, f
; CHECK-NEXT: adcl $0, f+4
; CHECK-NEXT: addl $8, %esp
; CHECK-NEXT: popl %ebx
; CHECK-NEXT: retl
entry:
%putchar = tail call i32 @putchar(i32 48)
%0 = load volatile i32, i32* @h, align 4
%1 = load i32, i32* @c, align 4, !tbaa !1
%2 = load i32***, i32**** @j, align 4
%3 = load i32**, i32*** %2, align 4
%4 = load i32*, i32** %3, align 4
%5 = load i32, i32* %4, align 4
%cmp = icmp sgt i32 %1, %5
%conv = zext i1 %cmp to i32
%6 = load i32, i32* @i, align 4
%cmp1 = icmp sgt i32 %6, %conv
%conv2 = zext i1 %cmp1 to i32
store i32 %conv2, i32* @b, align 4
%cmp3 = icmp sgt i32 %0, %conv2
%conv4 = zext i1 %cmp3 to i32
%7 = load i32, i32* @a, align 4
%or = xor i32 %7, %conv4
store i32 %or, i32* @a, align 4
%8 = load i32*, i32** @d, align 4
%9 = load i32, i32* %8, align 4
%conv6 = sext i32 %9 to i64
%10 = load i64, i64* @e, align 8
%and = and i64 %10, %conv6
store i64 %and, i64* @e, align 8
%11 = load i32, i32* @g, align 4
%dec = add nsw i32 %11, -1
store i32 %dec, i32* @g, align 4
%12 = load i64, i64* @f, align 8
%inc = add nsw i64 %12, 1
store i64 %inc, i64* @f, align 8
ret void
}
!0 = !{i32 1, !"NumRegisterParameters", i32 0}
!1 = !{!2, !2, i64 0}
!2 = !{!"int", !3, i64 0}
!3 = !{!"omnipotent char", !4, i64 0}
!4 = !{!"Simple C/C++ TBAA"}
!5 = !{!6, !6, i64 0}
!6 = !{!"any pointer", !3, i64 0}
!7 = !{!8, !8, i64 0}
!8 = !{!"long long", !3, i64 0}