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
209 lines
5.7 KiB
LLVM
209 lines
5.7 KiB
LLVM
; Test zero extensions from a halfword to an i64.
|
|
;
|
|
; RUN: llc < %s -mtriple=s390x-linux-gnu | FileCheck %s
|
|
|
|
; Test register extension, starting with an i32.
|
|
define i64 @f1(i32 %a) {
|
|
; CHECK-LABEL: f1:
|
|
; CHECK: llghr %r2, %r2
|
|
; CHECK: br %r14
|
|
%half = trunc i32 %a to i16
|
|
%ext = zext i16 %half to i64
|
|
ret i64 %ext
|
|
}
|
|
|
|
; ...and again with an i64.
|
|
define i64 @f2(i64 %a) {
|
|
; CHECK-LABEL: f2:
|
|
; CHECK: llghr %r2, %r2
|
|
; CHECK: br %r14
|
|
%half = trunc i64 %a to i16
|
|
%ext = zext i16 %half to i64
|
|
ret i64 %ext
|
|
}
|
|
|
|
; Check ANDs that are equivalent to zero extension.
|
|
define i64 @f3(i64 %a) {
|
|
; CHECK-LABEL: f3:
|
|
; CHECK: llghr %r2, %r2
|
|
; CHECK: br %r14
|
|
%ext = and i64 %a, 65535
|
|
ret i64 %ext
|
|
}
|
|
|
|
; Check LLGH with no displacement.
|
|
define i64 @f4(i16 *%src) {
|
|
; CHECK-LABEL: f4:
|
|
; CHECK: llgh %r2, 0(%r2)
|
|
; CHECK: br %r14
|
|
%half = load i16 *%src
|
|
%ext = zext i16 %half to i64
|
|
ret i64 %ext
|
|
}
|
|
|
|
; Check the high end of the LLGH range.
|
|
define i64 @f5(i16 *%src) {
|
|
; CHECK-LABEL: f5:
|
|
; CHECK: llgh %r2, 524286(%r2)
|
|
; CHECK: br %r14
|
|
%ptr = getelementptr i16, i16 *%src, i64 262143
|
|
%half = load i16 *%ptr
|
|
%ext = zext i16 %half to i64
|
|
ret i64 %ext
|
|
}
|
|
|
|
; Check the next halfword up, which needs separate address logic.
|
|
; Other sequences besides this one would be OK.
|
|
define i64 @f6(i16 *%src) {
|
|
; CHECK-LABEL: f6:
|
|
; CHECK: agfi %r2, 524288
|
|
; CHECK: llgh %r2, 0(%r2)
|
|
; CHECK: br %r14
|
|
%ptr = getelementptr i16, i16 *%src, i64 262144
|
|
%half = load i16 *%ptr
|
|
%ext = zext i16 %half to i64
|
|
ret i64 %ext
|
|
}
|
|
|
|
; Check the high end of the negative LLGH range.
|
|
define i64 @f7(i16 *%src) {
|
|
; CHECK-LABEL: f7:
|
|
; CHECK: llgh %r2, -2(%r2)
|
|
; CHECK: br %r14
|
|
%ptr = getelementptr i16, i16 *%src, i64 -1
|
|
%half = load i16 *%ptr
|
|
%ext = zext i16 %half to i64
|
|
ret i64 %ext
|
|
}
|
|
|
|
; Check the low end of the LLGH range.
|
|
define i64 @f8(i16 *%src) {
|
|
; CHECK-LABEL: f8:
|
|
; CHECK: llgh %r2, -524288(%r2)
|
|
; CHECK: br %r14
|
|
%ptr = getelementptr i16, i16 *%src, i64 -262144
|
|
%half = load i16 *%ptr
|
|
%ext = zext i16 %half to i64
|
|
ret i64 %ext
|
|
}
|
|
|
|
; Check the next halfword down, which needs separate address logic.
|
|
; Other sequences besides this one would be OK.
|
|
define i64 @f9(i16 *%src) {
|
|
; CHECK-LABEL: f9:
|
|
; CHECK: agfi %r2, -524290
|
|
; CHECK: llgh %r2, 0(%r2)
|
|
; CHECK: br %r14
|
|
%ptr = getelementptr i16, i16 *%src, i64 -262145
|
|
%half = load i16 *%ptr
|
|
%ext = zext i16 %half to i64
|
|
ret i64 %ext
|
|
}
|
|
|
|
; Check that LLGH allows an index
|
|
define i64 @f10(i64 %src, i64 %index) {
|
|
; CHECK-LABEL: f10:
|
|
; CHECK: llgh %r2, 524287(%r3,%r2)
|
|
; CHECK: br %r14
|
|
%add1 = add i64 %src, %index
|
|
%add2 = add i64 %add1, 524287
|
|
%ptr = inttoptr i64 %add2 to i16 *
|
|
%half = load i16 *%ptr
|
|
%ext = zext i16 %half to i64
|
|
ret i64 %ext
|
|
}
|
|
|
|
; Test a case where we spill the source of at least one LLGHR. We want
|
|
; to use LLGH if possible.
|
|
define void @f11(i64 *%ptr) {
|
|
; CHECK-LABEL: f11:
|
|
; CHECK: llgh {{%r[0-9]+}}, 166(%r15)
|
|
; CHECK: br %r14
|
|
%val0 = load volatile i64 *%ptr
|
|
%val1 = load volatile i64 *%ptr
|
|
%val2 = load volatile i64 *%ptr
|
|
%val3 = load volatile i64 *%ptr
|
|
%val4 = load volatile i64 *%ptr
|
|
%val5 = load volatile i64 *%ptr
|
|
%val6 = load volatile i64 *%ptr
|
|
%val7 = load volatile i64 *%ptr
|
|
%val8 = load volatile i64 *%ptr
|
|
%val9 = load volatile i64 *%ptr
|
|
%val10 = load volatile i64 *%ptr
|
|
%val11 = load volatile i64 *%ptr
|
|
%val12 = load volatile i64 *%ptr
|
|
%val13 = load volatile i64 *%ptr
|
|
%val14 = load volatile i64 *%ptr
|
|
%val15 = load volatile i64 *%ptr
|
|
|
|
%trunc0 = trunc i64 %val0 to i16
|
|
%trunc1 = trunc i64 %val1 to i16
|
|
%trunc2 = trunc i64 %val2 to i16
|
|
%trunc3 = trunc i64 %val3 to i16
|
|
%trunc4 = trunc i64 %val4 to i16
|
|
%trunc5 = trunc i64 %val5 to i16
|
|
%trunc6 = trunc i64 %val6 to i16
|
|
%trunc7 = trunc i64 %val7 to i16
|
|
%trunc8 = trunc i64 %val8 to i16
|
|
%trunc9 = trunc i64 %val9 to i16
|
|
%trunc10 = trunc i64 %val10 to i16
|
|
%trunc11 = trunc i64 %val11 to i16
|
|
%trunc12 = trunc i64 %val12 to i16
|
|
%trunc13 = trunc i64 %val13 to i16
|
|
%trunc14 = trunc i64 %val14 to i16
|
|
%trunc15 = trunc i64 %val15 to i16
|
|
|
|
%ext0 = zext i16 %trunc0 to i64
|
|
%ext1 = zext i16 %trunc1 to i64
|
|
%ext2 = zext i16 %trunc2 to i64
|
|
%ext3 = zext i16 %trunc3 to i64
|
|
%ext4 = zext i16 %trunc4 to i64
|
|
%ext5 = zext i16 %trunc5 to i64
|
|
%ext6 = zext i16 %trunc6 to i64
|
|
%ext7 = zext i16 %trunc7 to i64
|
|
%ext8 = zext i16 %trunc8 to i64
|
|
%ext9 = zext i16 %trunc9 to i64
|
|
%ext10 = zext i16 %trunc10 to i64
|
|
%ext11 = zext i16 %trunc11 to i64
|
|
%ext12 = zext i16 %trunc12 to i64
|
|
%ext13 = zext i16 %trunc13 to i64
|
|
%ext14 = zext i16 %trunc14 to i64
|
|
%ext15 = zext i16 %trunc15 to i64
|
|
|
|
store volatile i64 %val0, i64 *%ptr
|
|
store volatile i64 %val1, i64 *%ptr
|
|
store volatile i64 %val2, i64 *%ptr
|
|
store volatile i64 %val3, i64 *%ptr
|
|
store volatile i64 %val4, i64 *%ptr
|
|
store volatile i64 %val5, i64 *%ptr
|
|
store volatile i64 %val6, i64 *%ptr
|
|
store volatile i64 %val7, i64 *%ptr
|
|
store volatile i64 %val8, i64 *%ptr
|
|
store volatile i64 %val9, i64 *%ptr
|
|
store volatile i64 %val10, i64 *%ptr
|
|
store volatile i64 %val11, i64 *%ptr
|
|
store volatile i64 %val12, i64 *%ptr
|
|
store volatile i64 %val13, i64 *%ptr
|
|
store volatile i64 %val14, i64 *%ptr
|
|
store volatile i64 %val15, i64 *%ptr
|
|
|
|
store volatile i64 %ext0, i64 *%ptr
|
|
store volatile i64 %ext1, i64 *%ptr
|
|
store volatile i64 %ext2, i64 *%ptr
|
|
store volatile i64 %ext3, i64 *%ptr
|
|
store volatile i64 %ext4, i64 *%ptr
|
|
store volatile i64 %ext5, i64 *%ptr
|
|
store volatile i64 %ext6, i64 *%ptr
|
|
store volatile i64 %ext7, i64 *%ptr
|
|
store volatile i64 %ext8, i64 *%ptr
|
|
store volatile i64 %ext9, i64 *%ptr
|
|
store volatile i64 %ext10, i64 *%ptr
|
|
store volatile i64 %ext11, i64 *%ptr
|
|
store volatile i64 %ext12, i64 *%ptr
|
|
store volatile i64 %ext13, i64 *%ptr
|
|
store volatile i64 %ext14, i64 *%ptr
|
|
store volatile i64 %ext15, i64 *%ptr
|
|
|
|
ret void
|
|
}
|