mirror of
https://github.com/c64scene-ar/llvm-6502.git
synced 2024-12-15 04:30:12 +00:00
198d8baafb
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@230786 91177308-0d34-0410-b5e6-96231b3b80d8
236 lines
7.5 KiB
LLVM
236 lines
7.5 KiB
LLVM
; Test memcpy using MVC.
|
|
;
|
|
; RUN: llc < %s -mtriple=s390x-linux-gnu | FileCheck %s
|
|
|
|
declare void @llvm.memcpy.p0i8.p0i8.i32(i8 *nocapture, i8 *nocapture, i32, i32, i1) nounwind
|
|
declare void @llvm.memcpy.p0i8.p0i8.i64(i8 *nocapture, i8 *nocapture, i64, i32, i1) nounwind
|
|
declare void @foo(i8 *, i8 *)
|
|
|
|
; Test a no-op move, i32 version.
|
|
define void @f1(i8 *%dest, i8 *%src) {
|
|
; CHECK-LABEL: f1:
|
|
; CHECK-NOT: %r2
|
|
; CHECK-NOT: %r3
|
|
; CHECK: br %r14
|
|
call void @llvm.memcpy.p0i8.p0i8.i32(i8 *%dest, i8 *%src, i32 0, i32 1,
|
|
i1 false)
|
|
ret void
|
|
}
|
|
|
|
; Test a no-op move, i64 version.
|
|
define void @f2(i8 *%dest, i8 *%src) {
|
|
; CHECK-LABEL: f2:
|
|
; CHECK-NOT: %r2
|
|
; CHECK-NOT: %r3
|
|
; CHECK: br %r14
|
|
call void @llvm.memcpy.p0i8.p0i8.i64(i8 *%dest, i8 *%src, i64 0, i32 1,
|
|
i1 false)
|
|
ret void
|
|
}
|
|
|
|
; Test a 1-byte move, i32 version.
|
|
define void @f3(i8 *%dest, i8 *%src) {
|
|
; CHECK-LABEL: f3:
|
|
; CHECK: mvc 0(1,%r2), 0(%r3)
|
|
; CHECK: br %r14
|
|
call void @llvm.memcpy.p0i8.p0i8.i32(i8 *%dest, i8 *%src, i32 1, i32 1,
|
|
i1 false)
|
|
ret void
|
|
}
|
|
|
|
; Test a 1-byte move, i64 version.
|
|
define void @f4(i8 *%dest, i8 *%src) {
|
|
; CHECK-LABEL: f4:
|
|
; CHECK: mvc 0(1,%r2), 0(%r3)
|
|
; CHECK: br %r14
|
|
call void @llvm.memcpy.p0i8.p0i8.i64(i8 *%dest, i8 *%src, i64 1, i32 1,
|
|
i1 false)
|
|
ret void
|
|
}
|
|
|
|
; Test the upper range of a single MVC, i32 version.
|
|
define void @f5(i8 *%dest, i8 *%src) {
|
|
; CHECK-LABEL: f5:
|
|
; CHECK: mvc 0(256,%r2), 0(%r3)
|
|
; CHECK: br %r14
|
|
call void @llvm.memcpy.p0i8.p0i8.i32(i8 *%dest, i8 *%src, i32 256, i32 1,
|
|
i1 false)
|
|
ret void
|
|
}
|
|
|
|
; Test the upper range of a single MVC, i64 version.
|
|
define void @f6(i8 *%dest, i8 *%src) {
|
|
; CHECK-LABEL: f6:
|
|
; CHECK: mvc 0(256,%r2), 0(%r3)
|
|
; CHECK: br %r14
|
|
call void @llvm.memcpy.p0i8.p0i8.i64(i8 *%dest, i8 *%src, i64 256, i32 1,
|
|
i1 false)
|
|
ret void
|
|
}
|
|
|
|
; Test the first case that needs two MVCs.
|
|
define void @f7(i8 *%dest, i8 *%src) {
|
|
; CHECK-LABEL: f7:
|
|
; CHECK: mvc 0(256,%r2), 0(%r3)
|
|
; CHECK: mvc 256(1,%r2), 256(%r3)
|
|
; CHECK: br %r14
|
|
call void @llvm.memcpy.p0i8.p0i8.i32(i8 *%dest, i8 *%src, i32 257, i32 1,
|
|
i1 false)
|
|
ret void
|
|
}
|
|
|
|
; Test the last-but-one case that needs two MVCs.
|
|
define void @f8(i8 *%dest, i8 *%src) {
|
|
; CHECK-LABEL: f8:
|
|
; CHECK: mvc 0(256,%r2), 0(%r3)
|
|
; CHECK: mvc 256(255,%r2), 256(%r3)
|
|
; CHECK: br %r14
|
|
call void @llvm.memcpy.p0i8.p0i8.i64(i8 *%dest, i8 *%src, i64 511, i32 1,
|
|
i1 false)
|
|
ret void
|
|
}
|
|
|
|
; Test the last case that needs two MVCs.
|
|
define void @f9(i8 *%dest, i8 *%src) {
|
|
; CHECK-LABEL: f9:
|
|
; CHECK: mvc 0(256,%r2), 0(%r3)
|
|
; CHECK: mvc 256(256,%r2), 256(%r3)
|
|
; CHECK: br %r14
|
|
call void @llvm.memcpy.p0i8.p0i8.i64(i8 *%dest, i8 *%src, i64 512, i32 1,
|
|
i1 false)
|
|
ret void
|
|
}
|
|
|
|
; Test an arbitrary value that uses straight-line code.
|
|
define void @f10(i8 *%dest, i8 *%src) {
|
|
; CHECK-LABEL: f10:
|
|
; CHECK: mvc 0(256,%r2), 0(%r3)
|
|
; CHECK: mvc 256(256,%r2), 256(%r3)
|
|
; CHECK: mvc 512(256,%r2), 512(%r3)
|
|
; CHECK: mvc 768(256,%r2), 768(%r3)
|
|
; CHECK: mvc 1024(255,%r2), 1024(%r3)
|
|
; CHECK: br %r14
|
|
call void @llvm.memcpy.p0i8.p0i8.i64(i8 *%dest, i8 *%src, i64 1279, i32 1,
|
|
i1 false)
|
|
ret void
|
|
}
|
|
|
|
; ...and again in cases where not all parts are in range of MVC.
|
|
define void @f11(i8 *%srcbase, i8 *%destbase) {
|
|
; CHECK-LABEL: f11:
|
|
; CHECK: mvc 4000(256,%r2), 3500(%r3)
|
|
; CHECK: lay [[NEWDEST:%r[1-5]]], 4256(%r2)
|
|
; CHECK: mvc 0(256,[[NEWDEST]]), 3756(%r3)
|
|
; CHECK: mvc 256(256,[[NEWDEST]]), 4012(%r3)
|
|
; CHECK: lay [[NEWSRC:%r[1-5]]], 4268(%r3)
|
|
; CHECK: mvc 512(256,[[NEWDEST]]), 0([[NEWSRC]])
|
|
; CHECK: mvc 768(255,[[NEWDEST]]), 256([[NEWSRC]])
|
|
; CHECK: br %r14
|
|
%dest = getelementptr i8, i8 *%srcbase, i64 4000
|
|
%src = getelementptr i8, i8* %destbase, i64 3500
|
|
call void @llvm.memcpy.p0i8.p0i8.i64(i8 *%dest, i8 *%src, i64 1279, i32 1,
|
|
i1 false)
|
|
ret void
|
|
}
|
|
|
|
; ...and again with a destination frame base that goes out of range.
|
|
define void @f12() {
|
|
; CHECK-LABEL: f12:
|
|
; CHECK: brasl %r14, foo@PLT
|
|
; CHECK: mvc 4076(256,%r15), 2100(%r15)
|
|
; CHECK: lay [[NEWDEST:%r[1-5]]], 4332(%r15)
|
|
; CHECK: mvc 0(256,[[NEWDEST]]), 2356(%r15)
|
|
; CHECK: mvc 256(256,[[NEWDEST]]), 2612(%r15)
|
|
; CHECK: mvc 512(256,[[NEWDEST]]), 2868(%r15)
|
|
; CHECK: mvc 768(255,[[NEWDEST]]), 3124(%r15)
|
|
; CHECK: brasl %r14, foo@PLT
|
|
; CHECK: br %r14
|
|
%arr = alloca [6000 x i8]
|
|
%dest = getelementptr [6000 x i8], [6000 x i8] *%arr, i64 0, i64 3900
|
|
%src = getelementptr [6000 x i8], [6000 x i8] *%arr, i64 0, i64 1924
|
|
call void @foo(i8 *%dest, i8 *%src)
|
|
call void @llvm.memcpy.p0i8.p0i8.i64(i8 *%dest, i8 *%src, i64 1279, i32 1,
|
|
i1 false)
|
|
call void @foo(i8 *%dest, i8 *%src)
|
|
ret void
|
|
}
|
|
|
|
; ...and again with a source frame base that goes out of range.
|
|
define void @f13() {
|
|
; CHECK-LABEL: f13:
|
|
; CHECK: brasl %r14, foo@PLT
|
|
; CHECK: mvc 200(256,%r15), 3826(%r15)
|
|
; CHECK: mvc 456(256,%r15), 4082(%r15)
|
|
; CHECK: lay [[NEWSRC:%r[1-5]]], 4338(%r15)
|
|
; CHECK: mvc 712(256,%r15), 0([[NEWSRC]])
|
|
; CHECK: mvc 968(256,%r15), 256([[NEWSRC]])
|
|
; CHECK: mvc 1224(255,%r15), 512([[NEWSRC]])
|
|
; CHECK: brasl %r14, foo@PLT
|
|
; CHECK: br %r14
|
|
%arr = alloca [6000 x i8]
|
|
%dest = getelementptr [6000 x i8], [6000 x i8] *%arr, i64 0, i64 24
|
|
%src = getelementptr [6000 x i8], [6000 x i8] *%arr, i64 0, i64 3650
|
|
call void @foo(i8 *%dest, i8 *%src)
|
|
call void @llvm.memcpy.p0i8.p0i8.i64(i8 *%dest, i8 *%src, i64 1279, i32 1,
|
|
i1 false)
|
|
call void @foo(i8 *%dest, i8 *%src)
|
|
ret void
|
|
}
|
|
|
|
; Test the last case that is done using straight-line code.
|
|
define void @f14(i8 *%dest, i8 *%src) {
|
|
; CHECK-LABEL: f14:
|
|
; CHECK: mvc 0(256,%r2), 0(%r3)
|
|
; CHECK: mvc 256(256,%r2), 256(%r3)
|
|
; CHECK: mvc 512(256,%r2), 512(%r3)
|
|
; CHECK: mvc 768(256,%r2), 768(%r3)
|
|
; CHECK: mvc 1024(256,%r2), 1024(%r3)
|
|
; CHECK: mvc 1280(256,%r2), 1280(%r3)
|
|
; CHECK: br %r14
|
|
call void @llvm.memcpy.p0i8.p0i8.i64(i8 *%dest, i8 *%src, i64 1536, i32 1,
|
|
i1 false)
|
|
ret void
|
|
}
|
|
|
|
; Test the first case that is done using a loop.
|
|
define void @f15(i8 *%dest, i8 *%src) {
|
|
; CHECK-LABEL: f15:
|
|
; CHECK: lghi [[COUNT:%r[0-5]]], 6
|
|
; CHECK: [[LABEL:\.L[^:]*]]:
|
|
; CHECK: pfd 2, 768(%r2)
|
|
; CHECK: mvc 0(256,%r2), 0(%r3)
|
|
; CHECK: la %r2, 256(%r2)
|
|
; CHECK: la %r3, 256(%r3)
|
|
; CHECK: brctg [[COUNT]], [[LABEL]]
|
|
; CHECK: mvc 0(1,%r2), 0(%r3)
|
|
; CHECK: br %r14
|
|
call void @llvm.memcpy.p0i8.p0i8.i64(i8 *%dest, i8 *%src, i64 1537, i32 1,
|
|
i1 false)
|
|
ret void
|
|
}
|
|
|
|
; ...and again with frame bases, where the base must be loaded into a
|
|
; register before the loop.
|
|
define void @f16() {
|
|
; CHECK-LABEL: f16:
|
|
; CHECK: brasl %r14, foo@PLT
|
|
; CHECK-DAG: lghi [[COUNT:%r[0-5]]], 6
|
|
; CHECK-DAG: la [[BASE:%r[0-5]]], 160(%r15)
|
|
; CHECK: [[LABEL:\.L[^:]*]]:
|
|
; CHECK: pfd 2, 2368([[BASE]])
|
|
; CHECK: mvc 1600(256,[[BASE]]), 0([[BASE]])
|
|
; CHECK: la [[BASE]], 256([[BASE]])
|
|
; CHECK: brctg [[COUNT]], [[LABEL]]
|
|
; CHECK: mvc 1600(1,[[BASE]]), 0([[BASE]])
|
|
; CHECK: brasl %r14, foo@PLT
|
|
; CHECK: br %r14
|
|
%arr = alloca [3200 x i8]
|
|
%dest = getelementptr [3200 x i8], [3200 x i8] *%arr, i64 0, i64 1600
|
|
%src = getelementptr [3200 x i8], [3200 x i8] *%arr, i64 0, i64 0
|
|
call void @foo(i8 *%dest, i8 *%src)
|
|
call void @llvm.memcpy.p0i8.p0i8.i64(i8 *%dest, i8 *%src, i64 1537, i32 1,
|
|
i1 false)
|
|
call void @foo(i8 *%dest, i8 *%src)
|
|
ret void
|
|
}
|