mirror of
https://github.com/c64scene-ar/llvm-6502.git
synced 2025-02-23 20:29:30 +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
154 lines
4.3 KiB
LLVM
154 lines
4.3 KiB
LLVM
; Test 128-bit subtraction in which the second operand is variable.
|
|
;
|
|
; RUN: llc < %s -mtriple=s390x-linux-gnu -mcpu=z10 | FileCheck %s
|
|
; RUN: llc < %s -mtriple=s390x-linux-gnu -mcpu=z196 | FileCheck %s
|
|
|
|
declare i128 *@foo()
|
|
|
|
; Test register addition.
|
|
define void @f1(i128 *%ptr, i64 %high, i64 %low) {
|
|
; CHECK-LABEL: f1:
|
|
; CHECK: slgr {{%r[0-5]}}, %r4
|
|
; CHECK: slbgr {{%r[0-5]}}, %r3
|
|
; CHECK: br %r14
|
|
%a = load i128 *%ptr
|
|
%highx = zext i64 %high to i128
|
|
%lowx = zext i64 %low to i128
|
|
%bhigh = shl i128 %highx, 64
|
|
%b = or i128 %bhigh, %lowx
|
|
%sub = sub i128 %a, %b
|
|
store i128 %sub, i128 *%ptr
|
|
ret void
|
|
}
|
|
|
|
; Test memory addition with no offset.
|
|
define void @f2(i64 %addr) {
|
|
; CHECK-LABEL: f2:
|
|
; CHECK: slg {{%r[0-5]}}, 8(%r2)
|
|
; CHECK: slbg {{%r[0-5]}}, 0(%r2)
|
|
; CHECK: br %r14
|
|
%bptr = inttoptr i64 %addr to i128 *
|
|
%aptr = getelementptr i128, i128 *%bptr, i64 -8
|
|
%a = load i128 *%aptr
|
|
%b = load i128 *%bptr
|
|
%sub = sub i128 %a, %b
|
|
store i128 %sub, i128 *%aptr
|
|
ret void
|
|
}
|
|
|
|
; Test the highest aligned offset that is in range of both SLG and SLBG.
|
|
define void @f3(i64 %base) {
|
|
; CHECK-LABEL: f3:
|
|
; CHECK: slg {{%r[0-5]}}, 524280(%r2)
|
|
; CHECK: slbg {{%r[0-5]}}, 524272(%r2)
|
|
; CHECK: br %r14
|
|
%addr = add i64 %base, 524272
|
|
%bptr = inttoptr i64 %addr to i128 *
|
|
%aptr = getelementptr i128, i128 *%bptr, i64 -8
|
|
%a = load i128 *%aptr
|
|
%b = load i128 *%bptr
|
|
%sub = sub i128 %a, %b
|
|
store i128 %sub, i128 *%aptr
|
|
ret void
|
|
}
|
|
|
|
; Test the next doubleword up, which requires separate address logic for SLG.
|
|
define void @f4(i64 %base) {
|
|
; CHECK-LABEL: f4:
|
|
; CHECK: lgr [[BASE:%r[1-5]]], %r2
|
|
; CHECK: agfi [[BASE]], 524288
|
|
; CHECK: slg {{%r[0-5]}}, 0([[BASE]])
|
|
; CHECK: slbg {{%r[0-5]}}, 524280(%r2)
|
|
; CHECK: br %r14
|
|
%addr = add i64 %base, 524280
|
|
%bptr = inttoptr i64 %addr to i128 *
|
|
%aptr = getelementptr i128, i128 *%bptr, i64 -8
|
|
%a = load i128 *%aptr
|
|
%b = load i128 *%bptr
|
|
%sub = sub i128 %a, %b
|
|
store i128 %sub, i128 *%aptr
|
|
ret void
|
|
}
|
|
|
|
; Test the next doubleword after that, which requires separate logic for
|
|
; both instructions. It would be better to create an anchor at 524288
|
|
; that both instructions can use, but that isn't implemented yet.
|
|
define void @f5(i64 %base) {
|
|
; CHECK-LABEL: f5:
|
|
; CHECK: slg {{%r[0-5]}}, 0({{%r[1-5]}})
|
|
; CHECK: slbg {{%r[0-5]}}, 0({{%r[1-5]}})
|
|
; CHECK: br %r14
|
|
%addr = add i64 %base, 524288
|
|
%bptr = inttoptr i64 %addr to i128 *
|
|
%aptr = getelementptr i128, i128 *%bptr, i64 -8
|
|
%a = load i128 *%aptr
|
|
%b = load i128 *%bptr
|
|
%sub = sub i128 %a, %b
|
|
store i128 %sub, i128 *%aptr
|
|
ret void
|
|
}
|
|
|
|
; Test the lowest displacement that is in range of both SLG and SLBG.
|
|
define void @f6(i64 %base) {
|
|
; CHECK-LABEL: f6:
|
|
; CHECK: slg {{%r[0-5]}}, -524280(%r2)
|
|
; CHECK: slbg {{%r[0-5]}}, -524288(%r2)
|
|
; CHECK: br %r14
|
|
%addr = add i64 %base, -524288
|
|
%bptr = inttoptr i64 %addr to i128 *
|
|
%aptr = getelementptr i128, i128 *%bptr, i64 -8
|
|
%a = load i128 *%aptr
|
|
%b = load i128 *%bptr
|
|
%sub = sub i128 %a, %b
|
|
store i128 %sub, i128 *%aptr
|
|
ret void
|
|
}
|
|
|
|
; Test the next doubleword down, which is out of range of the SLBG.
|
|
define void @f7(i64 %base) {
|
|
; CHECK-LABEL: f7:
|
|
; CHECK: slg {{%r[0-5]}}, -524288(%r2)
|
|
; CHECK: slbg {{%r[0-5]}}, 0({{%r[1-5]}})
|
|
; CHECK: br %r14
|
|
%addr = add i64 %base, -524296
|
|
%bptr = inttoptr i64 %addr to i128 *
|
|
%aptr = getelementptr i128, i128 *%bptr, i64 -8
|
|
%a = load i128 *%aptr
|
|
%b = load i128 *%bptr
|
|
%sub = sub i128 %a, %b
|
|
store i128 %sub, i128 *%aptr
|
|
ret void
|
|
}
|
|
|
|
; Check that subtractions of spilled values can use SLG and SLBG rather than
|
|
; SLGR and SLBGR.
|
|
define void @f8(i128 *%ptr0) {
|
|
; CHECK-LABEL: f8:
|
|
; CHECK: brasl %r14, foo@PLT
|
|
; CHECK: slg {{%r[0-9]+}}, {{[0-9]+}}(%r15)
|
|
; CHECK: slbg {{%r[0-9]+}}, {{[0-9]+}}(%r15)
|
|
; CHECK: br %r14
|
|
%ptr1 = getelementptr i128, i128 *%ptr0, i128 2
|
|
%ptr2 = getelementptr i128, i128 *%ptr0, i128 4
|
|
%ptr3 = getelementptr i128, i128 *%ptr0, i128 6
|
|
%ptr4 = getelementptr i128, i128 *%ptr0, i128 8
|
|
|
|
%val0 = load i128 *%ptr0
|
|
%val1 = load i128 *%ptr1
|
|
%val2 = load i128 *%ptr2
|
|
%val3 = load i128 *%ptr3
|
|
%val4 = load i128 *%ptr4
|
|
|
|
%retptr = call i128 *@foo()
|
|
|
|
%ret = load i128 *%retptr
|
|
%sub0 = sub i128 %ret, %val0
|
|
%sub1 = sub i128 %sub0, %val1
|
|
%sub2 = sub i128 %sub1, %val2
|
|
%sub3 = sub i128 %sub2, %val3
|
|
%sub4 = sub i128 %sub3, %val4
|
|
store i128 %sub4, i128 *%retptr
|
|
|
|
ret void
|
|
}
|