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
231 lines
5.6 KiB
LLVM
231 lines
5.6 KiB
LLVM
; Test insertions of memory into the low byte of an i32.
|
|
;
|
|
; RUN: llc < %s -mtriple=s390x-linux-gnu | FileCheck %s
|
|
|
|
; Check a plain insertion with (or (and ... -0xff) (zext (load ....))).
|
|
; The whole sequence can be performed by IC.
|
|
define i32 @f1(i32 %orig, i8 *%ptr) {
|
|
; CHECK-LABEL: f1:
|
|
; CHECK-NOT: ni
|
|
; CHECK: ic %r2, 0(%r3)
|
|
; CHECK: br %r14
|
|
%val = load i8 *%ptr
|
|
%ptr2 = zext i8 %val to i32
|
|
%ptr1 = and i32 %orig, -256
|
|
%or = or i32 %ptr1, %ptr2
|
|
ret i32 %or
|
|
}
|
|
|
|
; Like f1, but with the operands reversed.
|
|
define i32 @f2(i32 %orig, i8 *%ptr) {
|
|
; CHECK-LABEL: f2:
|
|
; CHECK-NOT: ni
|
|
; CHECK: ic %r2, 0(%r3)
|
|
; CHECK: br %r14
|
|
%val = load i8 *%ptr
|
|
%ptr2 = zext i8 %val to i32
|
|
%ptr1 = and i32 %orig, -256
|
|
%or = or i32 %ptr2, %ptr1
|
|
ret i32 %or
|
|
}
|
|
|
|
; Check a case where more bits than lower 8 are masked out of the
|
|
; register value. We can use IC but must keep the original mask.
|
|
define i32 @f3(i32 %orig, i8 *%ptr) {
|
|
; CHECK-LABEL: f3:
|
|
; CHECK: nill %r2, 65024
|
|
; CHECK: ic %r2, 0(%r3)
|
|
; CHECK: br %r14
|
|
%val = load i8 *%ptr
|
|
%ptr2 = zext i8 %val to i32
|
|
%ptr1 = and i32 %orig, -512
|
|
%or = or i32 %ptr1, %ptr2
|
|
ret i32 %or
|
|
}
|
|
|
|
; Like f3, but with the operands reversed.
|
|
define i32 @f4(i32 %orig, i8 *%ptr) {
|
|
; CHECK-LABEL: f4:
|
|
; CHECK: nill %r2, 65024
|
|
; CHECK: ic %r2, 0(%r3)
|
|
; CHECK: br %r14
|
|
%val = load i8 *%ptr
|
|
%ptr2 = zext i8 %val to i32
|
|
%ptr1 = and i32 %orig, -512
|
|
%or = or i32 %ptr2, %ptr1
|
|
ret i32 %or
|
|
}
|
|
|
|
; Check a case where the low 8 bits are cleared by a shift left.
|
|
define i32 @f5(i32 %orig, i8 *%ptr) {
|
|
; CHECK-LABEL: f5:
|
|
; CHECK: sll %r2, 8
|
|
; CHECK: ic %r2, 0(%r3)
|
|
; CHECK: br %r14
|
|
%val = load i8 *%ptr
|
|
%ptr2 = zext i8 %val to i32
|
|
%ptr1 = shl i32 %orig, 8
|
|
%or = or i32 %ptr1, %ptr2
|
|
ret i32 %or
|
|
}
|
|
|
|
; Like f5, but with the operands reversed.
|
|
define i32 @f6(i32 %orig, i8 *%ptr) {
|
|
; CHECK-LABEL: f6:
|
|
; CHECK: sll %r2, 8
|
|
; CHECK: ic %r2, 0(%r3)
|
|
; CHECK: br %r14
|
|
%val = load i8 *%ptr
|
|
%ptr2 = zext i8 %val to i32
|
|
%ptr1 = shl i32 %orig, 8
|
|
%or = or i32 %ptr2, %ptr1
|
|
ret i32 %or
|
|
}
|
|
|
|
; Check insertions into a constant.
|
|
define i32 @f7(i32 %orig, i8 *%ptr) {
|
|
; CHECK-LABEL: f7:
|
|
; CHECK: lhi %r2, 256
|
|
; CHECK: ic %r2, 0(%r3)
|
|
; CHECK: br %r14
|
|
%val = load i8 *%ptr
|
|
%ptr2 = zext i8 %val to i32
|
|
%or = or i32 %ptr2, 256
|
|
ret i32 %or
|
|
}
|
|
|
|
; Like f7, but with the operands reversed.
|
|
define i32 @f8(i32 %orig, i8 *%ptr) {
|
|
; CHECK-LABEL: f8:
|
|
; CHECK: lhi %r2, 256
|
|
; CHECK: ic %r2, 0(%r3)
|
|
; CHECK: br %r14
|
|
%val = load i8 *%ptr
|
|
%ptr2 = zext i8 %val to i32
|
|
%or = or i32 256, %ptr2
|
|
ret i32 %or
|
|
}
|
|
|
|
; Check the high end of the IC range.
|
|
define i32 @f9(i32 %orig, i8 *%src) {
|
|
; CHECK-LABEL: f9:
|
|
; CHECK: ic %r2, 4095(%r3)
|
|
; CHECK: br %r14
|
|
%ptr = getelementptr i8, i8 *%src, i64 4095
|
|
%val = load i8 *%ptr
|
|
%src2 = zext i8 %val to i32
|
|
%src1 = and i32 %orig, -256
|
|
%or = or i32 %src2, %src1
|
|
ret i32 %or
|
|
}
|
|
|
|
; Check the next byte up, which should use ICY instead of IC.
|
|
define i32 @f10(i32 %orig, i8 *%src) {
|
|
; CHECK-LABEL: f10:
|
|
; CHECK: icy %r2, 4096(%r3)
|
|
; CHECK: br %r14
|
|
%ptr = getelementptr i8, i8 *%src, i64 4096
|
|
%val = load i8 *%ptr
|
|
%src2 = zext i8 %val to i32
|
|
%src1 = and i32 %orig, -256
|
|
%or = or i32 %src2, %src1
|
|
ret i32 %or
|
|
}
|
|
|
|
; Check the high end of the ICY range.
|
|
define i32 @f11(i32 %orig, i8 *%src) {
|
|
; CHECK-LABEL: f11:
|
|
; CHECK: icy %r2, 524287(%r3)
|
|
; CHECK: br %r14
|
|
%ptr = getelementptr i8, i8 *%src, i64 524287
|
|
%val = load i8 *%ptr
|
|
%src2 = zext i8 %val to i32
|
|
%src1 = and i32 %orig, -256
|
|
%or = or i32 %src2, %src1
|
|
ret i32 %or
|
|
}
|
|
|
|
; Check the next byte up, which needs separate address logic.
|
|
; Other sequences besides this one would be OK.
|
|
define i32 @f12(i32 %orig, i8 *%src) {
|
|
; CHECK-LABEL: f12:
|
|
; CHECK: agfi %r3, 524288
|
|
; CHECK: ic %r2, 0(%r3)
|
|
; CHECK: br %r14
|
|
%ptr = getelementptr i8, i8 *%src, i64 524288
|
|
%val = load i8 *%ptr
|
|
%src2 = zext i8 %val to i32
|
|
%src1 = and i32 %orig, -256
|
|
%or = or i32 %src2, %src1
|
|
ret i32 %or
|
|
}
|
|
|
|
; Check the high end of the negative ICY range.
|
|
define i32 @f13(i32 %orig, i8 *%src) {
|
|
; CHECK-LABEL: f13:
|
|
; CHECK: icy %r2, -1(%r3)
|
|
; CHECK: br %r14
|
|
%ptr = getelementptr i8, i8 *%src, i64 -1
|
|
%val = load i8 *%ptr
|
|
%src2 = zext i8 %val to i32
|
|
%src1 = and i32 %orig, -256
|
|
%or = or i32 %src2, %src1
|
|
ret i32 %or
|
|
}
|
|
|
|
; Check the low end of the ICY range.
|
|
define i32 @f14(i32 %orig, i8 *%src) {
|
|
; CHECK-LABEL: f14:
|
|
; CHECK: icy %r2, -524288(%r3)
|
|
; CHECK: br %r14
|
|
%ptr = getelementptr i8, i8 *%src, i64 -524288
|
|
%val = load i8 *%ptr
|
|
%src2 = zext i8 %val to i32
|
|
%src1 = and i32 %orig, -256
|
|
%or = or i32 %src2, %src1
|
|
ret i32 %or
|
|
}
|
|
|
|
; Check the next byte down, which needs separate address logic.
|
|
; Other sequences besides this one would be OK.
|
|
define i32 @f15(i32 %orig, i8 *%src) {
|
|
; CHECK-LABEL: f15:
|
|
; CHECK: agfi %r3, -524289
|
|
; CHECK: ic %r2, 0(%r3)
|
|
; CHECK: br %r14
|
|
%ptr = getelementptr i8, i8 *%src, i64 -524289
|
|
%val = load i8 *%ptr
|
|
%src2 = zext i8 %val to i32
|
|
%src1 = and i32 %orig, -256
|
|
%or = or i32 %src2, %src1
|
|
ret i32 %or
|
|
}
|
|
|
|
; Check that IC allows an index.
|
|
define i32 @f16(i32 %orig, i8 *%src, i64 %index) {
|
|
; CHECK-LABEL: f16:
|
|
; CHECK: ic %r2, 4095({{%r4,%r3|%r3,%r4}})
|
|
; CHECK: br %r14
|
|
%ptr1 = getelementptr i8, i8 *%src, i64 %index
|
|
%ptr2 = getelementptr i8, i8 *%ptr1, i64 4095
|
|
%val = load i8 *%ptr2
|
|
%src2 = zext i8 %val to i32
|
|
%src1 = and i32 %orig, -256
|
|
%or = or i32 %src2, %src1
|
|
ret i32 %or
|
|
}
|
|
|
|
; Check that ICY allows an index.
|
|
define i32 @f17(i32 %orig, i8 *%src, i64 %index) {
|
|
; CHECK-LABEL: f17:
|
|
; CHECK: icy %r2, 4096({{%r4,%r3|%r3,%r4}})
|
|
; CHECK: br %r14
|
|
%ptr1 = getelementptr i8, i8 *%src, i64 %index
|
|
%ptr2 = getelementptr i8, i8 *%ptr1, i64 4096
|
|
%val = load i8 *%ptr2
|
|
%src2 = zext i8 %val to i32
|
|
%src1 = and i32 %orig, -256
|
|
%or = or i32 %src2, %src1
|
|
ret i32 %or
|
|
}
|