mirror of
https://github.com/c64scene-ar/llvm-6502.git
synced 2024-12-13 20:32:21 +00:00
7b837d8c75
This adds a second implementation of the AArch64 architecture to LLVM, accessible in parallel via the "arm64" triple. The plan over the coming weeks & months is to merge the two into a single backend, during which time thorough code review should naturally occur. Everything will be easier with the target in-tree though, hence this commit. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@205090 91177308-0d34-0410-b5e6-96231b3b80d8
150 lines
4.5 KiB
LLVM
150 lines
4.5 KiB
LLVM
; RUN: llc < %s -march=arm64 -verify-machineinstrs | FileCheck %s
|
|
; RUN: llc < %s -march=arm64 -arm64-unscaled-mem-op=true\
|
|
; RUN: -verify-machineinstrs | FileCheck -check-prefix=LDUR_CHK %s
|
|
|
|
; CHECK: ldp_int
|
|
; CHECK: ldp
|
|
define i32 @ldp_int(i32* %p) nounwind {
|
|
%tmp = load i32* %p, align 4
|
|
%add.ptr = getelementptr inbounds i32* %p, i64 1
|
|
%tmp1 = load i32* %add.ptr, align 4
|
|
%add = add nsw i32 %tmp1, %tmp
|
|
ret i32 %add
|
|
}
|
|
|
|
; CHECK: ldp_long
|
|
; CHECK: ldp
|
|
define i64 @ldp_long(i64* %p) nounwind {
|
|
%tmp = load i64* %p, align 8
|
|
%add.ptr = getelementptr inbounds i64* %p, i64 1
|
|
%tmp1 = load i64* %add.ptr, align 8
|
|
%add = add nsw i64 %tmp1, %tmp
|
|
ret i64 %add
|
|
}
|
|
|
|
; CHECK: ldp_float
|
|
; CHECK: ldp
|
|
define float @ldp_float(float* %p) nounwind {
|
|
%tmp = load float* %p, align 4
|
|
%add.ptr = getelementptr inbounds float* %p, i64 1
|
|
%tmp1 = load float* %add.ptr, align 4
|
|
%add = fadd float %tmp, %tmp1
|
|
ret float %add
|
|
}
|
|
|
|
; CHECK: ldp_double
|
|
; CHECK: ldp
|
|
define double @ldp_double(double* %p) nounwind {
|
|
%tmp = load double* %p, align 8
|
|
%add.ptr = getelementptr inbounds double* %p, i64 1
|
|
%tmp1 = load double* %add.ptr, align 8
|
|
%add = fadd double %tmp, %tmp1
|
|
ret double %add
|
|
}
|
|
|
|
; Test the load/store optimizer---combine ldurs into a ldp, if appropriate
|
|
define i32 @ldur_int(i32* %a) nounwind {
|
|
; LDUR_CHK: ldur_int
|
|
; LDUR_CHK: ldp [[DST1:w[0-9]+]], [[DST2:w[0-9]+]], [x0, #-8]
|
|
; LDUR_CHK-NEXT: add w{{[0-9]+}}, [[DST2]], [[DST1]]
|
|
; LDUR_CHK-NEXT: ret
|
|
%p1 = getelementptr inbounds i32* %a, i32 -1
|
|
%tmp1 = load i32* %p1, align 2
|
|
%p2 = getelementptr inbounds i32* %a, i32 -2
|
|
%tmp2 = load i32* %p2, align 2
|
|
%tmp3 = add i32 %tmp1, %tmp2
|
|
ret i32 %tmp3
|
|
}
|
|
|
|
define i64 @ldur_long(i64* %a) nounwind ssp {
|
|
; LDUR_CHK: ldur_long
|
|
; LDUR_CHK: ldp [[DST1:x[0-9]+]], [[DST2:x[0-9]+]], [x0, #-16]
|
|
; LDUR_CHK-NEXT: add x{{[0-9]+}}, [[DST2]], [[DST1]]
|
|
; LDUR_CHK-NEXT: ret
|
|
%p1 = getelementptr inbounds i64* %a, i64 -1
|
|
%tmp1 = load i64* %p1, align 2
|
|
%p2 = getelementptr inbounds i64* %a, i64 -2
|
|
%tmp2 = load i64* %p2, align 2
|
|
%tmp3 = add i64 %tmp1, %tmp2
|
|
ret i64 %tmp3
|
|
}
|
|
|
|
define float @ldur_float(float* %a) {
|
|
; LDUR_CHK: ldur_float
|
|
; LDUR_CHK: ldp [[DST1:s[0-9]+]], [[DST2:s[0-9]+]], [x0, #-8]
|
|
; LDUR_CHK-NEXT: add s{{[0-9]+}}, [[DST2]], [[DST1]]
|
|
; LDUR_CHK-NEXT: ret
|
|
%p1 = getelementptr inbounds float* %a, i64 -1
|
|
%tmp1 = load float* %p1, align 2
|
|
%p2 = getelementptr inbounds float* %a, i64 -2
|
|
%tmp2 = load float* %p2, align 2
|
|
%tmp3 = fadd float %tmp1, %tmp2
|
|
ret float %tmp3
|
|
}
|
|
|
|
define double @ldur_double(double* %a) {
|
|
; LDUR_CHK: ldur_double
|
|
; LDUR_CHK: ldp [[DST1:d[0-9]+]], [[DST2:d[0-9]+]], [x0, #-16]
|
|
; LDUR_CHK-NEXT: add d{{[0-9]+}}, [[DST2]], [[DST1]]
|
|
; LDUR_CHK-NEXT: ret
|
|
%p1 = getelementptr inbounds double* %a, i64 -1
|
|
%tmp1 = load double* %p1, align 2
|
|
%p2 = getelementptr inbounds double* %a, i64 -2
|
|
%tmp2 = load double* %p2, align 2
|
|
%tmp3 = fadd double %tmp1, %tmp2
|
|
ret double %tmp3
|
|
}
|
|
|
|
; Now check some boundary conditions
|
|
define i64 @pairUpBarelyIn(i64* %a) nounwind ssp {
|
|
; LDUR_CHK: pairUpBarelyIn
|
|
; LDUR_CHK-NOT: ldur
|
|
; LDUR_CHK: ldp [[DST1:x[0-9]+]], [[DST2:x[0-9]+]], [x0, #-256]
|
|
; LDUR_CHK-NEXT: add x{{[0-9]+}}, [[DST2]], [[DST1]]
|
|
; LDUR_CHK-NEXT: ret
|
|
%p1 = getelementptr inbounds i64* %a, i64 -31
|
|
%tmp1 = load i64* %p1, align 2
|
|
%p2 = getelementptr inbounds i64* %a, i64 -32
|
|
%tmp2 = load i64* %p2, align 2
|
|
%tmp3 = add i64 %tmp1, %tmp2
|
|
ret i64 %tmp3
|
|
}
|
|
|
|
define i64 @pairUpBarelyOut(i64* %a) nounwind ssp {
|
|
; LDUR_CHK: pairUpBarelyOut
|
|
; LDUR_CHK-NOT: ldp
|
|
; Don't be fragile about which loads or manipulations of the base register
|
|
; are used---just check that there isn't an ldp before the add
|
|
; LDUR_CHK: add
|
|
; LDUR_CHK-NEXT: ret
|
|
%p1 = getelementptr inbounds i64* %a, i64 -32
|
|
%tmp1 = load i64* %p1, align 2
|
|
%p2 = getelementptr inbounds i64* %a, i64 -33
|
|
%tmp2 = load i64* %p2, align 2
|
|
%tmp3 = add i64 %tmp1, %tmp2
|
|
ret i64 %tmp3
|
|
}
|
|
|
|
define i64 @pairUpNotAligned(i64* %a) nounwind ssp {
|
|
; LDUR_CHK: pairUpNotAligned
|
|
; LDUR_CHK-NOT: ldp
|
|
; LDUR_CHK: ldur
|
|
; LDUR_CHK-NEXT: ldur
|
|
; LDUR_CHK-NEXT: add
|
|
; LDUR_CHK-NEXT: ret
|
|
%p1 = getelementptr inbounds i64* %a, i64 -18
|
|
%bp1 = bitcast i64* %p1 to i8*
|
|
%bp1p1 = getelementptr inbounds i8* %bp1, i64 1
|
|
%dp1 = bitcast i8* %bp1p1 to i64*
|
|
%tmp1 = load i64* %dp1, align 1
|
|
|
|
%p2 = getelementptr inbounds i64* %a, i64 -17
|
|
%bp2 = bitcast i64* %p2 to i8*
|
|
%bp2p1 = getelementptr inbounds i8* %bp2, i64 1
|
|
%dp2 = bitcast i8* %bp2p1 to i64*
|
|
%tmp2 = load i64* %dp2, align 1
|
|
|
|
%tmp3 = add i64 %tmp1, %tmp2
|
|
ret i64 %tmp3
|
|
}
|