mirror of
https://github.com/RPCS3/llvm-mirror.git
synced 2024-10-23 04:52:54 +02:00
0d99339102
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
llvm-svn: 230786
200 lines
5.4 KiB
LLVM
200 lines
5.4 KiB
LLVM
; Test sign extensions from a byte to an i32.
|
|
;
|
|
; RUN: llc < %s -mtriple=s390x-linux-gnu | FileCheck %s
|
|
|
|
; Test register extension, starting with an i32.
|
|
define i32 @f1(i32 %a) {
|
|
; CHECK-LABEL: f1:
|
|
; CHECK: lbr %r2, %r2
|
|
; CHECK: br %r14
|
|
%byte = trunc i32 %a to i8
|
|
%ext = sext i8 %byte to i32
|
|
ret i32 %ext
|
|
}
|
|
|
|
; ...and again with an i64.
|
|
define i32 @f2(i64 %a) {
|
|
; CHECK-LABEL: f2:
|
|
; CHECK: lbr %r2, %r2
|
|
; CHECK: br %r14
|
|
%byte = trunc i64 %a to i8
|
|
%ext = sext i8 %byte to i32
|
|
ret i32 %ext
|
|
}
|
|
|
|
; Check LB with no displacement.
|
|
define i32 @f3(i8 *%src) {
|
|
; CHECK-LABEL: f3:
|
|
; CHECK: lb %r2, 0(%r2)
|
|
; CHECK: br %r14
|
|
%byte = load i8 *%src
|
|
%ext = sext i8 %byte to i32
|
|
ret i32 %ext
|
|
}
|
|
|
|
; Check the high end of the LB range.
|
|
define i32 @f4(i8 *%src) {
|
|
; CHECK-LABEL: f4:
|
|
; CHECK: lb %r2, 524287(%r2)
|
|
; CHECK: br %r14
|
|
%ptr = getelementptr i8, i8 *%src, i64 524287
|
|
%byte = load i8 *%ptr
|
|
%ext = sext i8 %byte to i32
|
|
ret i32 %ext
|
|
}
|
|
|
|
; Check the next byte up, which needs separate address logic.
|
|
; Other sequences besides this one would be OK.
|
|
define i32 @f5(i8 *%src) {
|
|
; CHECK-LABEL: f5:
|
|
; CHECK: agfi %r2, 524288
|
|
; CHECK: lb %r2, 0(%r2)
|
|
; CHECK: br %r14
|
|
%ptr = getelementptr i8, i8 *%src, i64 524288
|
|
%byte = load i8 *%ptr
|
|
%ext = sext i8 %byte to i32
|
|
ret i32 %ext
|
|
}
|
|
|
|
; Check the high end of the negative LB range.
|
|
define i32 @f6(i8 *%src) {
|
|
; CHECK-LABEL: f6:
|
|
; CHECK: lb %r2, -1(%r2)
|
|
; CHECK: br %r14
|
|
%ptr = getelementptr i8, i8 *%src, i64 -1
|
|
%byte = load i8 *%ptr
|
|
%ext = sext i8 %byte to i32
|
|
ret i32 %ext
|
|
}
|
|
|
|
; Check the low end of the LB range.
|
|
define i32 @f7(i8 *%src) {
|
|
; CHECK-LABEL: f7:
|
|
; CHECK: lb %r2, -524288(%r2)
|
|
; CHECK: br %r14
|
|
%ptr = getelementptr i8, i8 *%src, i64 -524288
|
|
%byte = load i8 *%ptr
|
|
%ext = sext i8 %byte to i32
|
|
ret i32 %ext
|
|
}
|
|
|
|
; Check the next byte down, which needs separate address logic.
|
|
; Other sequences besides this one would be OK.
|
|
define i32 @f8(i8 *%src) {
|
|
; CHECK-LABEL: f8:
|
|
; CHECK: agfi %r2, -524289
|
|
; CHECK: lb %r2, 0(%r2)
|
|
; CHECK: br %r14
|
|
%ptr = getelementptr i8, i8 *%src, i64 -524289
|
|
%byte = load i8 *%ptr
|
|
%ext = sext i8 %byte to i32
|
|
ret i32 %ext
|
|
}
|
|
|
|
; Check that LB allows an index
|
|
define i32 @f9(i64 %src, i64 %index) {
|
|
; CHECK-LABEL: f9:
|
|
; CHECK: lb %r2, 524287(%r3,%r2)
|
|
; CHECK: br %r14
|
|
%add1 = add i64 %src, %index
|
|
%add2 = add i64 %add1, 524287
|
|
%ptr = inttoptr i64 %add2 to i8 *
|
|
%byte = load i8 *%ptr
|
|
%ext = sext i8 %byte to i32
|
|
ret i32 %ext
|
|
}
|
|
|
|
; Test a case where we spill the source of at least one LBR. We want
|
|
; to use LB if possible.
|
|
define void @f10(i32 *%ptr) {
|
|
; CHECK-LABEL: f10:
|
|
; CHECK: lb {{%r[0-9]+}}, 16{{[37]}}(%r15)
|
|
; CHECK: br %r14
|
|
%val0 = load volatile i32 *%ptr
|
|
%val1 = load volatile i32 *%ptr
|
|
%val2 = load volatile i32 *%ptr
|
|
%val3 = load volatile i32 *%ptr
|
|
%val4 = load volatile i32 *%ptr
|
|
%val5 = load volatile i32 *%ptr
|
|
%val6 = load volatile i32 *%ptr
|
|
%val7 = load volatile i32 *%ptr
|
|
%val8 = load volatile i32 *%ptr
|
|
%val9 = load volatile i32 *%ptr
|
|
%val10 = load volatile i32 *%ptr
|
|
%val11 = load volatile i32 *%ptr
|
|
%val12 = load volatile i32 *%ptr
|
|
%val13 = load volatile i32 *%ptr
|
|
%val14 = load volatile i32 *%ptr
|
|
%val15 = load volatile i32 *%ptr
|
|
|
|
%trunc0 = trunc i32 %val0 to i8
|
|
%trunc1 = trunc i32 %val1 to i8
|
|
%trunc2 = trunc i32 %val2 to i8
|
|
%trunc3 = trunc i32 %val3 to i8
|
|
%trunc4 = trunc i32 %val4 to i8
|
|
%trunc5 = trunc i32 %val5 to i8
|
|
%trunc6 = trunc i32 %val6 to i8
|
|
%trunc7 = trunc i32 %val7 to i8
|
|
%trunc8 = trunc i32 %val8 to i8
|
|
%trunc9 = trunc i32 %val9 to i8
|
|
%trunc10 = trunc i32 %val10 to i8
|
|
%trunc11 = trunc i32 %val11 to i8
|
|
%trunc12 = trunc i32 %val12 to i8
|
|
%trunc13 = trunc i32 %val13 to i8
|
|
%trunc14 = trunc i32 %val14 to i8
|
|
%trunc15 = trunc i32 %val15 to i8
|
|
|
|
%ext0 = sext i8 %trunc0 to i32
|
|
%ext1 = sext i8 %trunc1 to i32
|
|
%ext2 = sext i8 %trunc2 to i32
|
|
%ext3 = sext i8 %trunc3 to i32
|
|
%ext4 = sext i8 %trunc4 to i32
|
|
%ext5 = sext i8 %trunc5 to i32
|
|
%ext6 = sext i8 %trunc6 to i32
|
|
%ext7 = sext i8 %trunc7 to i32
|
|
%ext8 = sext i8 %trunc8 to i32
|
|
%ext9 = sext i8 %trunc9 to i32
|
|
%ext10 = sext i8 %trunc10 to i32
|
|
%ext11 = sext i8 %trunc11 to i32
|
|
%ext12 = sext i8 %trunc12 to i32
|
|
%ext13 = sext i8 %trunc13 to i32
|
|
%ext14 = sext i8 %trunc14 to i32
|
|
%ext15 = sext i8 %trunc15 to i32
|
|
|
|
store volatile i32 %val0, i32 *%ptr
|
|
store volatile i32 %val1, i32 *%ptr
|
|
store volatile i32 %val2, i32 *%ptr
|
|
store volatile i32 %val3, i32 *%ptr
|
|
store volatile i32 %val4, i32 *%ptr
|
|
store volatile i32 %val5, i32 *%ptr
|
|
store volatile i32 %val6, i32 *%ptr
|
|
store volatile i32 %val7, i32 *%ptr
|
|
store volatile i32 %val8, i32 *%ptr
|
|
store volatile i32 %val9, i32 *%ptr
|
|
store volatile i32 %val10, i32 *%ptr
|
|
store volatile i32 %val11, i32 *%ptr
|
|
store volatile i32 %val12, i32 *%ptr
|
|
store volatile i32 %val13, i32 *%ptr
|
|
store volatile i32 %val14, i32 *%ptr
|
|
store volatile i32 %val15, i32 *%ptr
|
|
|
|
store volatile i32 %ext0, i32 *%ptr
|
|
store volatile i32 %ext1, i32 *%ptr
|
|
store volatile i32 %ext2, i32 *%ptr
|
|
store volatile i32 %ext3, i32 *%ptr
|
|
store volatile i32 %ext4, i32 *%ptr
|
|
store volatile i32 %ext5, i32 *%ptr
|
|
store volatile i32 %ext6, i32 *%ptr
|
|
store volatile i32 %ext7, i32 *%ptr
|
|
store volatile i32 %ext8, i32 *%ptr
|
|
store volatile i32 %ext9, i32 *%ptr
|
|
store volatile i32 %ext10, i32 *%ptr
|
|
store volatile i32 %ext11, i32 *%ptr
|
|
store volatile i32 %ext12, i32 *%ptr
|
|
store volatile i32 %ext13, i32 *%ptr
|
|
store volatile i32 %ext14, i32 *%ptr
|
|
store volatile i32 %ext15, i32 *%ptr
|
|
|
|
ret void
|
|
}
|