mirror of
https://github.com/c64scene-ar/llvm-6502.git
synced 2025-02-22 13:29:44 +00:00
90a34679ef
The backend converts 64-bit ORs into subreg moves if the upper 32 bits of one operand and the low 32 bits of the other are known to be zero. It then tries to peel away redundant ANDs from the upper 32 bits. Since AND masks are canonicalized to exclude known-zero bits, the test ORs the mask and the known-zero bits together before checking for redundancy. The problem was that it was using the wrong node when checking for known-zero bits, so could drop ANDs that were still needed. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@196267 91177308-0d34-0410-b5e6-96231b3b80d8
181 lines
4.1 KiB
LLVM
181 lines
4.1 KiB
LLVM
; Test insertions of i32s into the low half of an i64.
|
|
;
|
|
; RUN: llc < %s -mtriple=s390x-linux-gnu | FileCheck %s
|
|
|
|
; Insertion of an i32 can be done using LR.
|
|
define i64 @f1(i64 %a, i32 %b) {
|
|
; CHECK-LABEL: f1:
|
|
; CHECK-NOT: {{%r[23]}}
|
|
; CHECK: lr %r2, %r3
|
|
; CHECK: br %r14
|
|
%low = zext i32 %b to i64
|
|
%high = and i64 %a, -4294967296
|
|
%res = or i64 %high, %low
|
|
ret i64 %res
|
|
}
|
|
|
|
; ... and again with the operands reversed.
|
|
define i64 @f2(i64 %a, i32 %b) {
|
|
; CHECK-LABEL: f2:
|
|
; CHECK-NOT: {{%r[23]}}
|
|
; CHECK: lr %r2, %r3
|
|
; CHECK: br %r14
|
|
%low = zext i32 %b to i64
|
|
%high = and i64 %a, -4294967296
|
|
%res = or i64 %low, %high
|
|
ret i64 %res
|
|
}
|
|
|
|
; Like f1, but with "in register" zero extension.
|
|
define i64 @f3(i64 %a, i64 %b) {
|
|
; CHECK-LABEL: f3:
|
|
; CHECK-NOT: {{%r[23]}}
|
|
; CHECK: lr %r2, %r3
|
|
; CHECK: br %r14
|
|
%low = and i64 %b, 4294967295
|
|
%high = and i64 %a, -4294967296
|
|
%res = or i64 %high, %low
|
|
ret i64 %res
|
|
}
|
|
|
|
; ... and again with the operands reversed.
|
|
define i64 @f4(i64 %a, i64 %b) {
|
|
; CHECK-LABEL: f4:
|
|
; CHECK-NOT: {{%r[23]}}
|
|
; CHECK: lr %r2, %r3
|
|
; CHECK: br %r14
|
|
%low = and i64 %b, 4294967295
|
|
%high = and i64 %a, -4294967296
|
|
%res = or i64 %low, %high
|
|
ret i64 %res
|
|
}
|
|
|
|
; Unary operations can be done directly into the low half.
|
|
define i64 @f5(i64 %a, i32 %b) {
|
|
; CHECK-LABEL: f5:
|
|
; CHECK-NOT: {{%r[23]}}
|
|
; CHECK: lcr %r2, %r3
|
|
; CHECK: br %r14
|
|
%neg = sub i32 0, %b
|
|
%low = zext i32 %neg to i64
|
|
%high = and i64 %a, -4294967296
|
|
%res = or i64 %high, %low
|
|
ret i64 %res
|
|
}
|
|
|
|
; ...likewise three-operand binary operations like RLL.
|
|
define i64 @f6(i64 %a, i32 %b) {
|
|
; CHECK-LABEL: f6:
|
|
; CHECK-NOT: {{%r[23]}}
|
|
; CHECK: rll %r2, %r3, 1
|
|
; CHECK: br %r14
|
|
%parta = shl i32 %b, 1
|
|
%partb = lshr i32 %b, 31
|
|
%rot = or i32 %parta, %partb
|
|
%low = zext i32 %rot to i64
|
|
%high = and i64 %a, -4294967296
|
|
%res = or i64 %low, %high
|
|
ret i64 %res
|
|
}
|
|
|
|
; Loads can be done directly into the low half. The range of L is checked
|
|
; in the move tests.
|
|
define i64 @f7(i64 %a, i32 *%src) {
|
|
; CHECK-LABEL: f7:
|
|
; CHECK-NOT: {{%r[23]}}
|
|
; CHECK: l %r2, 0(%r3)
|
|
; CHECK: br %r14
|
|
%b = load i32 *%src
|
|
%low = zext i32 %b to i64
|
|
%high = and i64 %a, -4294967296
|
|
%res = or i64 %high, %low
|
|
ret i64 %res
|
|
}
|
|
|
|
; ...likewise extending loads.
|
|
define i64 @f8(i64 %a, i8 *%src) {
|
|
; CHECK-LABEL: f8:
|
|
; CHECK-NOT: {{%r[23]}}
|
|
; CHECK: lb %r2, 0(%r3)
|
|
; CHECK: br %r14
|
|
%byte = load i8 *%src
|
|
%b = sext i8 %byte to i32
|
|
%low = zext i32 %b to i64
|
|
%high = and i64 %a, -4294967296
|
|
%res = or i64 %high, %low
|
|
ret i64 %res
|
|
}
|
|
|
|
; Check a case like f1 in which there is no AND. We simply know from context
|
|
; that the upper half of one OR operand and the lower half of the other are
|
|
; both clear.
|
|
define i64 @f9(i64 %a, i32 %b) {
|
|
; CHECK-LABEL: f9:
|
|
; CHECK: sllg %r2, %r2, 32
|
|
; CHECK: lr %r2, %r3
|
|
; CHECK: br %r14
|
|
%shift = shl i64 %a, 32
|
|
%low = zext i32 %b to i64
|
|
%or = or i64 %shift, %low
|
|
ret i64 %or
|
|
}
|
|
|
|
; ...and again with the operands reversed.
|
|
define i64 @f10(i64 %a, i32 %b) {
|
|
; CHECK-LABEL: f10:
|
|
; CHECK: sllg %r2, %r2, 32
|
|
; CHECK: lr %r2, %r3
|
|
; CHECK: br %r14
|
|
%shift = shl i64 %a, 32
|
|
%low = zext i32 %b to i64
|
|
%or = or i64 %low, %shift
|
|
ret i64 %or
|
|
}
|
|
|
|
; Like f9, but with "in register" zero extension.
|
|
define i64 @f11(i64 %a, i64 %b) {
|
|
; CHECK-LABEL: f11:
|
|
; CHECK: lr %r2, %r3
|
|
; CHECK: br %r14
|
|
%shift = shl i64 %a, 32
|
|
%low = and i64 %b, 4294967295
|
|
%or = or i64 %shift, %low
|
|
ret i64 %or
|
|
}
|
|
|
|
; ...and again with the operands reversed.
|
|
define i64 @f12(i64 %a, i64 %b) {
|
|
; CHECK-LABEL: f12:
|
|
; CHECK: lr %r2, %r3
|
|
; CHECK: br %r14
|
|
%shift = shl i64 %a, 32
|
|
%low = and i64 %b, 4294967295
|
|
%or = or i64 %low, %shift
|
|
ret i64 %or
|
|
}
|
|
|
|
; Like f9, but for larger shifts than 32.
|
|
define i64 @f13(i64 %a, i32 %b) {
|
|
; CHECK-LABEL: f13:
|
|
; CHECK: sllg %r2, %r2, 60
|
|
; CHECK: lr %r2, %r3
|
|
; CHECK: br %r14
|
|
%shift = shl i64 %a, 60
|
|
%low = zext i32 %b to i64
|
|
%or = or i64 %shift, %low
|
|
ret i64 %or
|
|
}
|
|
|
|
; We previously wrongly removed the upper AND as dead.
|
|
define i64 @f14(i64 %a, i64 %b) {
|
|
; CHECK-LABEL: f14:
|
|
; CHECK: risbg {{%r[0-5]}}, %r2, 6, 134, 0
|
|
; CHECK: br %r14
|
|
%and1 = and i64 %a, 144115188075855872
|
|
%and2 = and i64 %b, 15
|
|
%or = or i64 %and1, %and2
|
|
%res = icmp eq i64 %or, 0
|
|
%ext = sext i1 %res to i64
|
|
ret i64 %ext
|
|
}
|