2013-10-01 19:26:28 +08:00
|
|
|
; Test high-word operations, using "h" constraints to force a high
|
|
|
|
|
; register and "r" constraints to force a low register.
|
|
|
|
|
;
|
2015-01-14 03:45:16 +08:00
|
|
|
; RUN: llc < %s -mtriple=s390x-linux-gnu -mcpu=z196 -no-integrated-as | FileCheck %s
|
2013-10-01 19:26:28 +08:00
|
|
|
|
|
|
|
|
; Test loads and stores involving mixtures of high and low registers.
|
|
|
|
|
define void @f1(i32 *%ptr1, i32 *%ptr2) {
|
|
|
|
|
; CHECK-LABEL: f1:
|
|
|
|
|
; CHECK-DAG: lfh [[REG1:%r[0-5]]], 0(%r2)
|
|
|
|
|
; CHECK-DAG: l [[REG2:%r[0-5]]], 0(%r3)
|
|
|
|
|
; CHECK-DAG: lfh [[REG3:%r[0-5]]], 4096(%r2)
|
|
|
|
|
; CHECK-DAG: ly [[REG4:%r[0-5]]], 524284(%r3)
|
|
|
|
|
; CHECK: blah [[REG1]], [[REG2]], [[REG3]], [[REG4]]
|
|
|
|
|
; CHECK-DAG: stfh [[REG1]], 0(%r2)
|
|
|
|
|
; CHECK-DAG: st [[REG2]], 0(%r3)
|
|
|
|
|
; CHECK-DAG: stfh [[REG3]], 4096(%r2)
|
|
|
|
|
; CHECK-DAG: sty [[REG4]], 524284(%r3)
|
|
|
|
|
; CHECK: br %r14
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
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
2015-02-28 03:29:02 +08:00
|
|
|
%ptr3 = getelementptr i32, i32 *%ptr1, i64 1024
|
|
|
|
|
%ptr4 = getelementptr i32, i32 *%ptr2, i64 131071
|
2013-10-01 19:26:28 +08:00
|
|
|
%old1 = load i32 *%ptr1
|
|
|
|
|
%old2 = load i32 *%ptr2
|
|
|
|
|
%old3 = load i32 *%ptr3
|
|
|
|
|
%old4 = load i32 *%ptr4
|
|
|
|
|
%res = call { i32, i32, i32, i32 } asm "blah $0, $1, $2, $3",
|
|
|
|
|
"=h,=r,=h,=r,0,1,2,3"(i32 %old1, i32 %old2, i32 %old3, i32 %old4)
|
|
|
|
|
%new1 = extractvalue { i32, i32, i32, i32 } %res, 0
|
|
|
|
|
%new2 = extractvalue { i32, i32, i32, i32 } %res, 1
|
|
|
|
|
%new3 = extractvalue { i32, i32, i32, i32 } %res, 2
|
|
|
|
|
%new4 = extractvalue { i32, i32, i32, i32 } %res, 3
|
|
|
|
|
store i32 %new1, i32 *%ptr1
|
|
|
|
|
store i32 %new2, i32 *%ptr2
|
|
|
|
|
store i32 %new3, i32 *%ptr3
|
|
|
|
|
store i32 %new4, i32 *%ptr4
|
|
|
|
|
ret void
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
; Test moves involving mixtures of high and low registers.
|
|
|
|
|
define i32 @f2(i32 %old) {
|
|
|
|
|
; CHECK-LABEL: f2:
|
|
|
|
|
; CHECK-DAG: risbhg [[REG1:%r[0-5]]], %r2, 0, 159, 32
|
|
|
|
|
; CHECK-DAG: lr %r3, %r2
|
|
|
|
|
; CHECK: stepa [[REG1]], %r2, %r3
|
|
|
|
|
; CHECK: risbhg {{%r[0-5]}}, [[REG1]], 0, 159, 0
|
|
|
|
|
; CHECK: stepb [[REG2:%r[0-5]]]
|
|
|
|
|
; CHECK: risblg %r2, [[REG2]], 0, 159, 32
|
|
|
|
|
; CHECK: br %r14
|
|
|
|
|
%tmp = call i32 asm "stepa $1, $2, $3",
|
|
|
|
|
"=h,0,{r2},{r3}"(i32 %old, i32 %old, i32 %old)
|
|
|
|
|
%new = call i32 asm "stepb $1, $2", "=&h,0,h"(i32 %tmp, i32 %tmp)
|
|
|
|
|
ret i32 %new
|
|
|
|
|
}
|
2013-10-01 20:11:47 +08:00
|
|
|
|
|
|
|
|
; Test sign-extending 8-bit loads into mixtures of high and low registers.
|
|
|
|
|
define void @f3(i8 *%ptr1, i8 *%ptr2) {
|
|
|
|
|
; CHECK-LABEL: f3:
|
|
|
|
|
; CHECK-DAG: lbh [[REG1:%r[0-5]]], 0(%r2)
|
|
|
|
|
; CHECK-DAG: lb [[REG2:%r[0-5]]], 0(%r3)
|
|
|
|
|
; CHECK-DAG: lbh [[REG3:%r[0-5]]], 4096(%r2)
|
|
|
|
|
; CHECK-DAG: lb [[REG4:%r[0-5]]], 524287(%r3)
|
|
|
|
|
; CHECK: blah [[REG1]], [[REG2]]
|
|
|
|
|
; CHECK: br %r14
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
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
2015-02-28 03:29:02 +08:00
|
|
|
%ptr3 = getelementptr i8, i8 *%ptr1, i64 4096
|
|
|
|
|
%ptr4 = getelementptr i8, i8 *%ptr2, i64 524287
|
2013-10-01 20:11:47 +08:00
|
|
|
%val1 = load i8 *%ptr1
|
|
|
|
|
%val2 = load i8 *%ptr2
|
|
|
|
|
%val3 = load i8 *%ptr3
|
|
|
|
|
%val4 = load i8 *%ptr4
|
|
|
|
|
%ext1 = sext i8 %val1 to i32
|
|
|
|
|
%ext2 = sext i8 %val2 to i32
|
|
|
|
|
%ext3 = sext i8 %val3 to i32
|
|
|
|
|
%ext4 = sext i8 %val4 to i32
|
|
|
|
|
call void asm sideeffect "blah $0, $1, $2, $3",
|
|
|
|
|
"h,r,h,r"(i32 %ext1, i32 %ext2, i32 %ext3, i32 %ext4)
|
|
|
|
|
ret void
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
; Test sign-extending 16-bit loads into mixtures of high and low registers.
|
|
|
|
|
define void @f4(i16 *%ptr1, i16 *%ptr2) {
|
|
|
|
|
; CHECK-LABEL: f4:
|
|
|
|
|
; CHECK-DAG: lhh [[REG1:%r[0-5]]], 0(%r2)
|
|
|
|
|
; CHECK-DAG: lh [[REG2:%r[0-5]]], 0(%r3)
|
|
|
|
|
; CHECK-DAG: lhh [[REG3:%r[0-5]]], 4096(%r2)
|
|
|
|
|
; CHECK-DAG: lhy [[REG4:%r[0-5]]], 524286(%r3)
|
|
|
|
|
; CHECK: blah [[REG1]], [[REG2]]
|
|
|
|
|
; CHECK: br %r14
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
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
2015-02-28 03:29:02 +08:00
|
|
|
%ptr3 = getelementptr i16, i16 *%ptr1, i64 2048
|
|
|
|
|
%ptr4 = getelementptr i16, i16 *%ptr2, i64 262143
|
2013-10-01 20:11:47 +08:00
|
|
|
%val1 = load i16 *%ptr1
|
|
|
|
|
%val2 = load i16 *%ptr2
|
|
|
|
|
%val3 = load i16 *%ptr3
|
|
|
|
|
%val4 = load i16 *%ptr4
|
|
|
|
|
%ext1 = sext i16 %val1 to i32
|
|
|
|
|
%ext2 = sext i16 %val2 to i32
|
|
|
|
|
%ext3 = sext i16 %val3 to i32
|
|
|
|
|
%ext4 = sext i16 %val4 to i32
|
|
|
|
|
call void asm sideeffect "blah $0, $1, $2, $3",
|
|
|
|
|
"h,r,h,r"(i32 %ext1, i32 %ext2, i32 %ext3, i32 %ext4)
|
|
|
|
|
ret void
|
|
|
|
|
}
|
2013-10-01 20:19:08 +08:00
|
|
|
|
|
|
|
|
; Test zero-extending 8-bit loads into mixtures of high and low registers.
|
|
|
|
|
define void @f5(i8 *%ptr1, i8 *%ptr2) {
|
|
|
|
|
; CHECK-LABEL: f5:
|
|
|
|
|
; CHECK-DAG: llch [[REG1:%r[0-5]]], 0(%r2)
|
|
|
|
|
; CHECK-DAG: llc [[REG2:%r[0-5]]], 0(%r3)
|
|
|
|
|
; CHECK-DAG: llch [[REG3:%r[0-5]]], 4096(%r2)
|
|
|
|
|
; CHECK-DAG: llc [[REG4:%r[0-5]]], 524287(%r3)
|
|
|
|
|
; CHECK: blah [[REG1]], [[REG2]]
|
|
|
|
|
; CHECK: br %r14
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
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
2015-02-28 03:29:02 +08:00
|
|
|
%ptr3 = getelementptr i8, i8 *%ptr1, i64 4096
|
|
|
|
|
%ptr4 = getelementptr i8, i8 *%ptr2, i64 524287
|
2013-10-01 20:19:08 +08:00
|
|
|
%val1 = load i8 *%ptr1
|
|
|
|
|
%val2 = load i8 *%ptr2
|
|
|
|
|
%val3 = load i8 *%ptr3
|
|
|
|
|
%val4 = load i8 *%ptr4
|
|
|
|
|
%ext1 = zext i8 %val1 to i32
|
|
|
|
|
%ext2 = zext i8 %val2 to i32
|
|
|
|
|
%ext3 = zext i8 %val3 to i32
|
|
|
|
|
%ext4 = zext i8 %val4 to i32
|
|
|
|
|
call void asm sideeffect "blah $0, $1, $2, $3",
|
|
|
|
|
"h,r,h,r"(i32 %ext1, i32 %ext2, i32 %ext3, i32 %ext4)
|
|
|
|
|
ret void
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
; Test zero-extending 16-bit loads into mixtures of high and low registers.
|
|
|
|
|
define void @f6(i16 *%ptr1, i16 *%ptr2) {
|
|
|
|
|
; CHECK-LABEL: f6:
|
|
|
|
|
; CHECK-DAG: llhh [[REG1:%r[0-5]]], 0(%r2)
|
|
|
|
|
; CHECK-DAG: llh [[REG2:%r[0-5]]], 0(%r3)
|
|
|
|
|
; CHECK-DAG: llhh [[REG3:%r[0-5]]], 4096(%r2)
|
|
|
|
|
; CHECK-DAG: llh [[REG4:%r[0-5]]], 524286(%r3)
|
|
|
|
|
; CHECK: blah [[REG1]], [[REG2]]
|
|
|
|
|
; CHECK: br %r14
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
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
2015-02-28 03:29:02 +08:00
|
|
|
%ptr3 = getelementptr i16, i16 *%ptr1, i64 2048
|
|
|
|
|
%ptr4 = getelementptr i16, i16 *%ptr2, i64 262143
|
2013-10-01 20:19:08 +08:00
|
|
|
%val1 = load i16 *%ptr1
|
|
|
|
|
%val2 = load i16 *%ptr2
|
|
|
|
|
%val3 = load i16 *%ptr3
|
|
|
|
|
%val4 = load i16 *%ptr4
|
|
|
|
|
%ext1 = zext i16 %val1 to i32
|
|
|
|
|
%ext2 = zext i16 %val2 to i32
|
|
|
|
|
%ext3 = zext i16 %val3 to i32
|
|
|
|
|
%ext4 = zext i16 %val4 to i32
|
|
|
|
|
call void asm sideeffect "blah $0, $1, $2, $3",
|
|
|
|
|
"h,r,h,r"(i32 %ext1, i32 %ext2, i32 %ext3, i32 %ext4)
|
|
|
|
|
ret void
|
|
|
|
|
}
|
2013-10-01 20:22:49 +08:00
|
|
|
|
|
|
|
|
; Test truncating stores of high and low registers into 8-bit memory.
|
|
|
|
|
define void @f7(i8 *%ptr1, i8 *%ptr2) {
|
|
|
|
|
; CHECK-LABEL: f7:
|
|
|
|
|
; CHECK: blah [[REG1:%r[0-5]]], [[REG2:%r[0-5]]]
|
|
|
|
|
; CHECK-DAG: stch [[REG1]], 0(%r2)
|
|
|
|
|
; CHECK-DAG: stc [[REG2]], 0(%r3)
|
|
|
|
|
; CHECK-DAG: stch [[REG1]], 4096(%r2)
|
|
|
|
|
; CHECK-DAG: stcy [[REG2]], 524287(%r3)
|
|
|
|
|
; CHECK: br %r14
|
|
|
|
|
%res = call { i32, i32 } asm "blah $0, $1", "=h,=r"()
|
|
|
|
|
%res1 = extractvalue { i32, i32 } %res, 0
|
|
|
|
|
%res2 = extractvalue { i32, i32 } %res, 1
|
|
|
|
|
%trunc1 = trunc i32 %res1 to i8
|
|
|
|
|
%trunc2 = trunc i32 %res2 to i8
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
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
2015-02-28 03:29:02 +08:00
|
|
|
%ptr3 = getelementptr i8, i8 *%ptr1, i64 4096
|
|
|
|
|
%ptr4 = getelementptr i8, i8 *%ptr2, i64 524287
|
2013-10-01 20:22:49 +08:00
|
|
|
store i8 %trunc1, i8 *%ptr1
|
|
|
|
|
store i8 %trunc2, i8 *%ptr2
|
|
|
|
|
store i8 %trunc1, i8 *%ptr3
|
|
|
|
|
store i8 %trunc2, i8 *%ptr4
|
|
|
|
|
ret void
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
; Test truncating stores of high and low registers into 16-bit memory.
|
|
|
|
|
define void @f8(i16 *%ptr1, i16 *%ptr2) {
|
|
|
|
|
; CHECK-LABEL: f8:
|
|
|
|
|
; CHECK: blah [[REG1:%r[0-5]]], [[REG2:%r[0-5]]]
|
|
|
|
|
; CHECK-DAG: sthh [[REG1]], 0(%r2)
|
|
|
|
|
; CHECK-DAG: sth [[REG2]], 0(%r3)
|
|
|
|
|
; CHECK-DAG: sthh [[REG1]], 4096(%r2)
|
|
|
|
|
; CHECK-DAG: sthy [[REG2]], 524286(%r3)
|
|
|
|
|
; CHECK: br %r14
|
|
|
|
|
%res = call { i32, i32 } asm "blah $0, $1", "=h,=r"()
|
|
|
|
|
%res1 = extractvalue { i32, i32 } %res, 0
|
|
|
|
|
%res2 = extractvalue { i32, i32 } %res, 1
|
|
|
|
|
%trunc1 = trunc i32 %res1 to i16
|
|
|
|
|
%trunc2 = trunc i32 %res2 to i16
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
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
2015-02-28 03:29:02 +08:00
|
|
|
%ptr3 = getelementptr i16, i16 *%ptr1, i64 2048
|
|
|
|
|
%ptr4 = getelementptr i16, i16 *%ptr2, i64 262143
|
2013-10-01 20:22:49 +08:00
|
|
|
store i16 %trunc1, i16 *%ptr1
|
|
|
|
|
store i16 %trunc2, i16 *%ptr2
|
|
|
|
|
store i16 %trunc1, i16 *%ptr3
|
|
|
|
|
store i16 %trunc2, i16 *%ptr4
|
|
|
|
|
ret void
|
|
|
|
|
}
|
2013-10-01 20:49:07 +08:00
|
|
|
|
|
|
|
|
; Test zero extensions from 8 bits between mixtures of high and low registers.
|
|
|
|
|
define i32 @f9(i8 %val1, i8 %val2) {
|
|
|
|
|
; CHECK-LABEL: f9:
|
|
|
|
|
; CHECK-DAG: risbhg [[REG1:%r[0-5]]], %r2, 24, 159, 32
|
|
|
|
|
; CHECK-DAG: llcr [[REG2:%r[0-5]]], %r3
|
|
|
|
|
; CHECK: stepa [[REG1]], [[REG2]]
|
|
|
|
|
; CHECK: risbhg [[REG3:%r[0-5]]], [[REG1]], 24, 159, 0
|
|
|
|
|
; CHECK: stepb [[REG3]]
|
|
|
|
|
; CHECK: risblg %r2, [[REG3]], 24, 159, 32
|
|
|
|
|
; CHECK: br %r14
|
|
|
|
|
%ext1 = zext i8 %val1 to i32
|
|
|
|
|
%ext2 = zext i8 %val2 to i32
|
|
|
|
|
%val3 = call i8 asm sideeffect "stepa $0, $1", "=h,0,r"(i32 %ext1, i32 %ext2)
|
|
|
|
|
%ext3 = zext i8 %val3 to i32
|
|
|
|
|
%val4 = call i8 asm sideeffect "stepb $0", "=h,0"(i32 %ext3)
|
|
|
|
|
%ext4 = zext i8 %val4 to i32
|
|
|
|
|
ret i32 %ext4
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
; Test zero extensions from 16 bits between mixtures of high and low registers.
|
|
|
|
|
define i32 @f10(i16 %val1, i16 %val2) {
|
|
|
|
|
; CHECK-LABEL: f10:
|
|
|
|
|
; CHECK-DAG: risbhg [[REG1:%r[0-5]]], %r2, 16, 159, 32
|
|
|
|
|
; CHECK-DAG: llhr [[REG2:%r[0-5]]], %r3
|
|
|
|
|
; CHECK: stepa [[REG1]], [[REG2]]
|
|
|
|
|
; CHECK: risbhg [[REG3:%r[0-5]]], [[REG1]], 16, 159, 0
|
|
|
|
|
; CHECK: stepb [[REG3]]
|
|
|
|
|
; CHECK: risblg %r2, [[REG3]], 16, 159, 32
|
|
|
|
|
; CHECK: br %r14
|
|
|
|
|
%ext1 = zext i16 %val1 to i32
|
|
|
|
|
%ext2 = zext i16 %val2 to i32
|
|
|
|
|
%val3 = call i16 asm sideeffect "stepa $0, $1", "=h,0,r"(i32 %ext1, i32 %ext2)
|
|
|
|
|
%ext3 = zext i16 %val3 to i32
|
|
|
|
|
%val4 = call i16 asm sideeffect "stepb $0", "=h,0"(i32 %ext3)
|
|
|
|
|
%ext4 = zext i16 %val4 to i32
|
|
|
|
|
ret i32 %ext4
|
|
|
|
|
}
|
2013-10-01 21:02:28 +08:00
|
|
|
|
|
|
|
|
; Test loads of 16-bit constants into mixtures of high and low registers.
|
|
|
|
|
define void @f11() {
|
|
|
|
|
; CHECK-LABEL: f11:
|
|
|
|
|
; CHECK-DAG: iihf [[REG1:%r[0-5]]], 4294934529
|
|
|
|
|
; CHECK-DAG: lhi [[REG2:%r[0-5]]], -32768
|
|
|
|
|
; CHECK-DAG: llihl [[REG3:%r[0-5]]], 32766
|
|
|
|
|
; CHECK-DAG: lhi [[REG4:%r[0-5]]], 32767
|
|
|
|
|
; CHECK: blah [[REG1]], [[REG2]], [[REG3]], [[REG4]]
|
|
|
|
|
; CHECK: br %r14
|
|
|
|
|
call void asm sideeffect "blah $0, $1, $2, $3",
|
|
|
|
|
"h,r,h,r"(i32 -32767, i32 -32768,
|
|
|
|
|
i32 32766, i32 32767)
|
|
|
|
|
ret void
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
; Test loads of unsigned constants into mixtures of high and low registers.
|
|
|
|
|
; For stepc, we expect the h and r operands to be paired by the register
|
|
|
|
|
; allocator. It doesn't really matter which comes first: LLILL/IIHF would
|
|
|
|
|
; be just as good.
|
|
|
|
|
define void @f12() {
|
|
|
|
|
; CHECK-LABEL: f12:
|
|
|
|
|
; CHECK-DAG: llihl [[REG1:%r[0-5]]], 32768
|
|
|
|
|
; CHECK-DAG: llihl [[REG2:%r[0-5]]], 65535
|
|
|
|
|
; CHECK-DAG: llihh [[REG3:%r[0-5]]], 1
|
|
|
|
|
; CHECK-DAG: llihh [[REG4:%r[0-5]]], 65535
|
|
|
|
|
; CHECK: stepa [[REG1]], [[REG2]], [[REG3]], [[REG4]]
|
|
|
|
|
; CHECK-DAG: llill [[REG1:%r[0-5]]], 32769
|
|
|
|
|
; CHECK-DAG: llill [[REG2:%r[0-5]]], 65534
|
|
|
|
|
; CHECK-DAG: llilh [[REG3:%r[0-5]]], 2
|
|
|
|
|
; CHECK-DAG: llilh [[REG4:%r[0-5]]], 65534
|
|
|
|
|
; CHECK: stepb [[REG1]], [[REG2]], [[REG3]], [[REG4]]
|
|
|
|
|
; CHECK-DAG: llihl [[REG1:%r[0-5]]], 32770
|
|
|
|
|
; CHECK-DAG: iilf [[REG1]], 65533
|
|
|
|
|
; CHECK-DAG: llihh [[REG2:%r[0-5]]], 4
|
|
|
|
|
; CHECK-DAG: iilf [[REG2]], 524288
|
|
|
|
|
; CHECK: stepc [[REG1]], [[REG1]], [[REG2]], [[REG2]]
|
|
|
|
|
; CHECK-DAG: iihf [[REG1:%r[0-5]]], 3294967296
|
|
|
|
|
; CHECK-DAG: iilf [[REG2:%r[0-5]]], 4294567296
|
|
|
|
|
; CHECK-DAG: iihf [[REG3:%r[0-5]]], 1000000000
|
|
|
|
|
; CHECK-DAG: iilf [[REG4:%r[0-5]]], 400000
|
|
|
|
|
; CHECK: stepd [[REG1]], [[REG2]], [[REG3]], [[REG4]]
|
|
|
|
|
; CHECK: br %r14
|
|
|
|
|
call void asm sideeffect "stepa $0, $1, $2, $3",
|
|
|
|
|
"h,h,h,h"(i32 32768, i32 65535,
|
|
|
|
|
i32 65536, i32 -65536)
|
|
|
|
|
call void asm sideeffect "stepb $0, $1, $2, $3",
|
|
|
|
|
"r,r,r,r"(i32 32769, i32 65534,
|
|
|
|
|
i32 131072, i32 -131072)
|
|
|
|
|
call void asm sideeffect "stepc $0, $1, $2, $3",
|
|
|
|
|
"h,r,h,r"(i32 32770, i32 65533,
|
|
|
|
|
i32 262144, i32 524288)
|
|
|
|
|
call void asm sideeffect "stepd $0, $1, $2, $3",
|
|
|
|
|
"h,r,h,r"(i32 -1000000000, i32 -400000,
|
|
|
|
|
i32 1000000000, i32 400000)
|
|
|
|
|
ret void
|
|
|
|
|
}
|
2013-10-01 21:10:16 +08:00
|
|
|
|
|
|
|
|
; Test selects involving high registers.
|
|
|
|
|
define void @f13(i32 %x, i32 %y) {
|
|
|
|
|
; CHECK-LABEL: f13:
|
|
|
|
|
; CHECK: llihl [[REG:%r[0-5]]], 0
|
|
|
|
|
; CHECK: cije %r2, 0
|
|
|
|
|
; CHECK: iihf [[REG]], 2102030405
|
|
|
|
|
; CHECK: blah [[REG]]
|
|
|
|
|
; CHECK: br %r14
|
|
|
|
|
%cmp = icmp eq i32 %x, 0
|
|
|
|
|
%val = select i1 %cmp, i32 0, i32 2102030405
|
|
|
|
|
call void asm sideeffect "blah $0", "h"(i32 %val)
|
|
|
|
|
ret void
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
; Test selects involving low registers.
|
|
|
|
|
define void @f14(i32 %x, i32 %y) {
|
|
|
|
|
; CHECK-LABEL: f14:
|
|
|
|
|
; CHECK: lhi [[REG:%r[0-5]]], 0
|
|
|
|
|
; CHECK: cije %r2, 0
|
|
|
|
|
; CHECK: iilf [[REG]], 2102030405
|
|
|
|
|
; CHECK: blah [[REG]]
|
|
|
|
|
; CHECK: br %r14
|
|
|
|
|
%cmp = icmp eq i32 %x, 0
|
|
|
|
|
%val = select i1 %cmp, i32 0, i32 2102030405
|
|
|
|
|
call void asm sideeffect "blah $0", "r"(i32 %val)
|
|
|
|
|
ret void
|
|
|
|
|
}
|
2013-10-01 21:18:56 +08:00
|
|
|
|
|
|
|
|
; Test immediate insertion involving high registers.
|
|
|
|
|
define void @f15() {
|
|
|
|
|
; CHECK-LABEL: f15:
|
|
|
|
|
; CHECK: stepa [[REG:%r[0-5]]]
|
|
|
|
|
; CHECK: iihh [[REG]], 4660
|
|
|
|
|
; CHECK: stepb [[REG]]
|
|
|
|
|
; CHECK: iihl [[REG]], 34661
|
|
|
|
|
; CHECK: stepc [[REG]]
|
|
|
|
|
; CHECK: br %r14
|
|
|
|
|
%res1 = call i32 asm "stepa $0", "=h"()
|
|
|
|
|
%and1 = and i32 %res1, 65535
|
|
|
|
|
%or1 = or i32 %and1, 305397760
|
|
|
|
|
%res2 = call i32 asm "stepb $0, $1", "=h,h"(i32 %or1)
|
|
|
|
|
%and2 = and i32 %res2, -65536
|
|
|
|
|
%or2 = or i32 %and2, 34661
|
|
|
|
|
call void asm sideeffect "stepc $0", "h"(i32 %or2)
|
|
|
|
|
ret void
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
; Test immediate insertion involving low registers.
|
|
|
|
|
define void @f16() {
|
|
|
|
|
; CHECK-LABEL: f16:
|
|
|
|
|
; CHECK: stepa [[REG:%r[0-5]]]
|
|
|
|
|
; CHECK: iilh [[REG]], 4660
|
|
|
|
|
; CHECK: stepb [[REG]]
|
|
|
|
|
; CHECK: iill [[REG]], 34661
|
|
|
|
|
; CHECK: stepc [[REG]]
|
|
|
|
|
; CHECK: br %r14
|
|
|
|
|
%res1 = call i32 asm "stepa $0", "=r"()
|
|
|
|
|
%and1 = and i32 %res1, 65535
|
|
|
|
|
%or1 = or i32 %and1, 305397760
|
|
|
|
|
%res2 = call i32 asm "stepb $0, $1", "=r,r"(i32 %or1)
|
|
|
|
|
%and2 = and i32 %res2, -65536
|
|
|
|
|
%or2 = or i32 %and2, 34661
|
|
|
|
|
call void asm sideeffect "stepc $0", "r"(i32 %or2)
|
|
|
|
|
ret void
|
|
|
|
|
}
|
2013-10-01 21:22:41 +08:00
|
|
|
|
|
|
|
|
; Test immediate OR involving high registers.
|
|
|
|
|
define void @f17() {
|
|
|
|
|
; CHECK-LABEL: f17:
|
|
|
|
|
; CHECK: stepa [[REG:%r[0-5]]]
|
|
|
|
|
; CHECK: oihh [[REG]], 4660
|
|
|
|
|
; CHECK: stepb [[REG]]
|
|
|
|
|
; CHECK: oihl [[REG]], 34661
|
|
|
|
|
; CHECK: stepc [[REG]]
|
|
|
|
|
; CHECK: oihf [[REG]], 12345678
|
|
|
|
|
; CHECK: stepd [[REG]]
|
|
|
|
|
; CHECK: br %r14
|
|
|
|
|
%res1 = call i32 asm "stepa $0", "=h"()
|
|
|
|
|
%or1 = or i32 %res1, 305397760
|
|
|
|
|
%res2 = call i32 asm "stepb $0, $1", "=h,h"(i32 %or1)
|
|
|
|
|
%or2 = or i32 %res2, 34661
|
|
|
|
|
%res3 = call i32 asm "stepc $0, $1", "=h,h"(i32 %or2)
|
|
|
|
|
%or3 = or i32 %res3, 12345678
|
|
|
|
|
call void asm sideeffect "stepd $0", "h"(i32 %or3)
|
|
|
|
|
ret void
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
; Test immediate OR involving low registers.
|
|
|
|
|
define void @f18() {
|
|
|
|
|
; CHECK-LABEL: f18:
|
|
|
|
|
; CHECK: stepa [[REG:%r[0-5]]]
|
|
|
|
|
; CHECK: oilh [[REG]], 4660
|
|
|
|
|
; CHECK: stepb [[REG]]
|
|
|
|
|
; CHECK: oill [[REG]], 34661
|
|
|
|
|
; CHECK: stepc [[REG]]
|
|
|
|
|
; CHECK: oilf [[REG]], 12345678
|
|
|
|
|
; CHECK: stepd [[REG]]
|
|
|
|
|
; CHECK: br %r14
|
|
|
|
|
%res1 = call i32 asm "stepa $0", "=r"()
|
|
|
|
|
%or1 = or i32 %res1, 305397760
|
|
|
|
|
%res2 = call i32 asm "stepb $0, $1", "=r,r"(i32 %or1)
|
|
|
|
|
%or2 = or i32 %res2, 34661
|
|
|
|
|
%res3 = call i32 asm "stepc $0, $1", "=r,r"(i32 %or2)
|
|
|
|
|
%or3 = or i32 %res3, 12345678
|
|
|
|
|
call void asm sideeffect "stepd $0", "r"(i32 %or3)
|
|
|
|
|
ret void
|
|
|
|
|
}
|
2013-10-01 22:08:44 +08:00
|
|
|
|
|
|
|
|
; Test immediate XOR involving high registers.
|
|
|
|
|
define void @f19() {
|
|
|
|
|
; CHECK-LABEL: f19:
|
|
|
|
|
; CHECK: stepa [[REG:%r[0-5]]]
|
|
|
|
|
; CHECK: xihf [[REG]], 305397760
|
|
|
|
|
; CHECK: stepb [[REG]]
|
|
|
|
|
; CHECK: xihf [[REG]], 34661
|
|
|
|
|
; CHECK: stepc [[REG]]
|
|
|
|
|
; CHECK: xihf [[REG]], 12345678
|
|
|
|
|
; CHECK: stepd [[REG]]
|
|
|
|
|
; CHECK: br %r14
|
|
|
|
|
%res1 = call i32 asm "stepa $0", "=h"()
|
|
|
|
|
%xor1 = xor i32 %res1, 305397760
|
|
|
|
|
%res2 = call i32 asm "stepb $0, $1", "=h,h"(i32 %xor1)
|
|
|
|
|
%xor2 = xor i32 %res2, 34661
|
|
|
|
|
%res3 = call i32 asm "stepc $0, $1", "=h,h"(i32 %xor2)
|
|
|
|
|
%xor3 = xor i32 %res3, 12345678
|
|
|
|
|
call void asm sideeffect "stepd $0", "h"(i32 %xor3)
|
|
|
|
|
ret void
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
; Test immediate XOR involving low registers.
|
|
|
|
|
define void @f20() {
|
|
|
|
|
; CHECK-LABEL: f20:
|
|
|
|
|
; CHECK: stepa [[REG:%r[0-5]]]
|
|
|
|
|
; CHECK: xilf [[REG]], 305397760
|
|
|
|
|
; CHECK: stepb [[REG]]
|
|
|
|
|
; CHECK: xilf [[REG]], 34661
|
|
|
|
|
; CHECK: stepc [[REG]]
|
|
|
|
|
; CHECK: xilf [[REG]], 12345678
|
|
|
|
|
; CHECK: stepd [[REG]]
|
|
|
|
|
; CHECK: br %r14
|
|
|
|
|
%res1 = call i32 asm "stepa $0", "=r"()
|
|
|
|
|
%xor1 = xor i32 %res1, 305397760
|
|
|
|
|
%res2 = call i32 asm "stepb $0, $1", "=r,r"(i32 %xor1)
|
|
|
|
|
%xor2 = xor i32 %res2, 34661
|
|
|
|
|
%res3 = call i32 asm "stepc $0, $1", "=r,r"(i32 %xor2)
|
|
|
|
|
%xor3 = xor i32 %res3, 12345678
|
|
|
|
|
call void asm sideeffect "stepd $0", "r"(i32 %xor3)
|
|
|
|
|
ret void
|
|
|
|
|
}
|
2013-10-01 22:20:41 +08:00
|
|
|
|
|
|
|
|
; Test two-operand immediate AND involving high registers.
|
|
|
|
|
define void @f21() {
|
|
|
|
|
; CHECK-LABEL: f21:
|
|
|
|
|
; CHECK: stepa [[REG:%r[0-5]]]
|
|
|
|
|
; CHECK: nihh [[REG]], 4096
|
|
|
|
|
; CHECK: stepb [[REG]]
|
|
|
|
|
; CHECK: nihl [[REG]], 57536
|
|
|
|
|
; CHECK: stepc [[REG]]
|
|
|
|
|
; CHECK: nihf [[REG]], 12345678
|
|
|
|
|
; CHECK: stepd [[REG]]
|
|
|
|
|
; CHECK: br %r14
|
|
|
|
|
%res1 = call i32 asm "stepa $0", "=h"()
|
|
|
|
|
%and1 = and i32 %res1, 268500991
|
|
|
|
|
%res2 = call i32 asm "stepb $0, $1", "=h,h"(i32 %and1)
|
|
|
|
|
%and2 = and i32 %res2, -8000
|
|
|
|
|
%res3 = call i32 asm "stepc $0, $1", "=h,h"(i32 %and2)
|
|
|
|
|
%and3 = and i32 %res3, 12345678
|
|
|
|
|
call void asm sideeffect "stepd $0", "h"(i32 %and3)
|
|
|
|
|
ret void
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
; Test two-operand immediate AND involving low registers.
|
|
|
|
|
define void @f22() {
|
|
|
|
|
; CHECK-LABEL: f22:
|
|
|
|
|
; CHECK: stepa [[REG:%r[0-5]]]
|
|
|
|
|
; CHECK: nilh [[REG]], 4096
|
|
|
|
|
; CHECK: stepb [[REG]]
|
|
|
|
|
; CHECK: nill [[REG]], 57536
|
|
|
|
|
; CHECK: stepc [[REG]]
|
|
|
|
|
; CHECK: nilf [[REG]], 12345678
|
|
|
|
|
; CHECK: stepd [[REG]]
|
|
|
|
|
; CHECK: br %r14
|
|
|
|
|
%res1 = call i32 asm "stepa $0", "=r"()
|
|
|
|
|
%and1 = and i32 %res1, 268500991
|
|
|
|
|
%res2 = call i32 asm "stepb $0, $1", "=r,r"(i32 %and1)
|
|
|
|
|
%and2 = and i32 %res2, -8000
|
|
|
|
|
%res3 = call i32 asm "stepc $0, $1", "=r,r"(i32 %and2)
|
|
|
|
|
%and3 = and i32 %res3, 12345678
|
|
|
|
|
call void asm sideeffect "stepd $0", "r"(i32 %and3)
|
|
|
|
|
ret void
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
; Test three-operand immediate AND involving mixtures of low and high registers.
|
|
|
|
|
define i32 @f23(i32 %old) {
|
|
|
|
|
; CHECK-LABEL: f23:
|
|
|
|
|
; CHECK-DAG: risblg [[REG1:%r[0-5]]], %r2, 28, 158, 0
|
|
|
|
|
; CHECK-DAG: risbhg [[REG2:%r[0-5]]], %r2, 24, 158, 32
|
|
|
|
|
; CHECK: stepa %r2, [[REG1]], [[REG2]]
|
|
|
|
|
; CHECK-DAG: risbhg [[REG3:%r[0-5]]], [[REG2]], 25, 159, 0
|
|
|
|
|
; CHECK-DAG: risblg %r2, [[REG2]], 24, 152, 32
|
|
|
|
|
; CHECK: stepb [[REG2]], [[REG3]], %r2
|
|
|
|
|
; CHECK: br %r14
|
|
|
|
|
%and1 = and i32 %old, 14
|
|
|
|
|
%and2 = and i32 %old, 254
|
|
|
|
|
%res1 = call i32 asm "stepa $1, $2, $3",
|
|
|
|
|
"=h,r,r,0"(i32 %old, i32 %and1, i32 %and2)
|
|
|
|
|
%and3 = and i32 %res1, 127
|
|
|
|
|
%and4 = and i32 %res1, 128
|
|
|
|
|
%res2 = call i32 asm "stepb $1, $2, $3",
|
|
|
|
|
"=r,h,h,0"(i32 %res1, i32 %and3, i32 %and4)
|
|
|
|
|
ret i32 %res2
|
|
|
|
|
}
|
2013-10-01 22:36:20 +08:00
|
|
|
|
|
|
|
|
; Test RISB[LH]G insertions involving mixtures of high and low registers.
|
|
|
|
|
define i32 @f24(i32 %old) {
|
|
|
|
|
; CHECK-LABEL: f24:
|
|
|
|
|
; CHECK-DAG: risblg [[REG1:%r[0-5]]], %r2, 28, 158, 1
|
|
|
|
|
; CHECK-DAG: risbhg [[REG2:%r[0-5]]], %r2, 24, 158, 29
|
|
|
|
|
; CHECK: stepa %r2, [[REG1]], [[REG2]]
|
|
|
|
|
; CHECK-DAG: risbhg [[REG3:%r[0-5]]], [[REG2]], 25, 159, 62
|
|
|
|
|
; CHECK-DAG: risblg %r2, [[REG2]], 24, 152, 37
|
|
|
|
|
; CHECK: stepb [[REG2]], [[REG3]], %r2
|
|
|
|
|
; CHECK: br %r14
|
|
|
|
|
%shift1 = shl i32 %old, 1
|
|
|
|
|
%and1 = and i32 %shift1, 14
|
|
|
|
|
%shift2 = lshr i32 %old, 3
|
|
|
|
|
%and2 = and i32 %shift2, 254
|
|
|
|
|
%res1 = call i32 asm "stepa $1, $2, $3",
|
|
|
|
|
"=h,r,r,0"(i32 %old, i32 %and1, i32 %and2)
|
|
|
|
|
%shift3 = lshr i32 %res1, 2
|
|
|
|
|
%and3 = and i32 %shift3, 127
|
|
|
|
|
%shift4 = shl i32 %res1, 5
|
|
|
|
|
%and4 = and i32 %shift4, 128
|
|
|
|
|
%res2 = call i32 asm "stepb $1, $2, $3",
|
|
|
|
|
"=r,h,h,0"(i32 %res1, i32 %and3, i32 %and4)
|
|
|
|
|
ret i32 %res2
|
|
|
|
|
}
|
2013-10-01 22:41:52 +08:00
|
|
|
|
|
|
|
|
; Test TMxx involving mixtures of high and low registers.
|
|
|
|
|
define i32 @f25(i32 %old) {
|
|
|
|
|
; CHECK-LABEL: f25:
|
|
|
|
|
; CHECK-DAG: tmll %r2, 1
|
|
|
|
|
; CHECK-DAG: tmlh %r2, 1
|
|
|
|
|
; CHECK: stepa [[REG1:%r[0-5]]],
|
|
|
|
|
; CHECK-DAG: tmhl [[REG1]], 1
|
|
|
|
|
; CHECK-DAG: tmhh [[REG1]], 1
|
|
|
|
|
; CHECK: stepb %r2,
|
|
|
|
|
; CHECK: br %r14
|
|
|
|
|
%and1 = and i32 %old, 1
|
|
|
|
|
%and2 = and i32 %old, 65536
|
|
|
|
|
%cmp1 = icmp eq i32 %and1, 0
|
|
|
|
|
%cmp2 = icmp eq i32 %and2, 0
|
|
|
|
|
%sel1 = select i1 %cmp1, i32 100, i32 200
|
|
|
|
|
%sel2 = select i1 %cmp2, i32 100, i32 200
|
|
|
|
|
%res1 = call i32 asm "stepa $0, $1, $2",
|
|
|
|
|
"=h,r,r"(i32 %sel1, i32 %sel2)
|
|
|
|
|
%and3 = and i32 %res1, 1
|
|
|
|
|
%and4 = and i32 %res1, 65536
|
|
|
|
|
%cmp3 = icmp eq i32 %and3, 0
|
|
|
|
|
%cmp4 = icmp eq i32 %and4, 0
|
|
|
|
|
%sel3 = select i1 %cmp3, i32 100, i32 200
|
|
|
|
|
%sel4 = select i1 %cmp4, i32 100, i32 200
|
|
|
|
|
%res2 = call i32 asm "stepb $0, $1, $2",
|
|
|
|
|
"=r,h,h"(i32 %sel3, i32 %sel4)
|
|
|
|
|
ret i32 %res2
|
|
|
|
|
}
|
2013-10-01 22:53:46 +08:00
|
|
|
|
|
|
|
|
; Test two-operand halfword immediate addition involving high registers.
|
|
|
|
|
define void @f26() {
|
|
|
|
|
; CHECK-LABEL: f26:
|
|
|
|
|
; CHECK: stepa [[REG:%r[0-5]]]
|
|
|
|
|
; CHECK: aih [[REG]], -32768
|
|
|
|
|
; CHECK: stepb [[REG]]
|
|
|
|
|
; CHECK: aih [[REG]], 1
|
|
|
|
|
; CHECK: stepc [[REG]]
|
|
|
|
|
; CHECK: aih [[REG]], 32767
|
|
|
|
|
; CHECK: stepd [[REG]]
|
|
|
|
|
; CHECK: br %r14
|
|
|
|
|
%res1 = call i32 asm "stepa $0", "=h"()
|
|
|
|
|
%add1 = add i32 %res1, -32768
|
|
|
|
|
%res2 = call i32 asm "stepb $0, $1", "=h,h"(i32 %add1)
|
|
|
|
|
%add2 = add i32 %res2, 1
|
|
|
|
|
%res3 = call i32 asm "stepc $0, $1", "=h,h"(i32 %add2)
|
|
|
|
|
%add3 = add i32 %res3, 32767
|
|
|
|
|
call void asm sideeffect "stepd $0", "h"(i32 %add3)
|
|
|
|
|
ret void
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
; Test two-operand halfword immediate addition involving low registers.
|
|
|
|
|
define void @f27() {
|
|
|
|
|
; CHECK-LABEL: f27:
|
|
|
|
|
; CHECK: stepa [[REG:%r[0-5]]]
|
|
|
|
|
; CHECK: ahi [[REG]], -32768
|
|
|
|
|
; CHECK: stepb [[REG]]
|
|
|
|
|
; CHECK: ahi [[REG]], 1
|
|
|
|
|
; CHECK: stepc [[REG]]
|
|
|
|
|
; CHECK: ahi [[REG]], 32767
|
|
|
|
|
; CHECK: stepd [[REG]]
|
|
|
|
|
; CHECK: br %r14
|
|
|
|
|
%res1 = call i32 asm "stepa $0", "=r"()
|
|
|
|
|
%add1 = add i32 %res1, -32768
|
|
|
|
|
%res2 = call i32 asm "stepb $0, $1", "=r,r"(i32 %add1)
|
|
|
|
|
%add2 = add i32 %res2, 1
|
|
|
|
|
%res3 = call i32 asm "stepc $0, $1", "=r,r"(i32 %add2)
|
|
|
|
|
%add3 = add i32 %res3, 32767
|
|
|
|
|
call void asm sideeffect "stepd $0", "r"(i32 %add3)
|
|
|
|
|
ret void
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
; Test three-operand halfword immediate addition involving mixtures of low
|
|
|
|
|
; and high registers. RISBHG/AIH would be OK too, instead of AHIK/RISBHG.
|
|
|
|
|
define i32 @f28(i32 %old) {
|
|
|
|
|
; CHECK-LABEL: f28:
|
|
|
|
|
; CHECK: ahik [[REG1:%r[0-5]]], %r2, 14
|
|
|
|
|
; CHECK: stepa %r2, [[REG1]]
|
|
|
|
|
; CHECK: ahik [[TMP:%r[0-5]]], [[REG1]], 254
|
|
|
|
|
; CHECK: risbhg [[REG2:%r[0-5]]], [[TMP]], 0, 159, 32
|
|
|
|
|
; CHECK: stepb [[REG1]], [[REG2]]
|
|
|
|
|
; CHECK: risbhg [[REG3:%r[0-5]]], [[REG2]], 0, 159, 0
|
|
|
|
|
; CHECK: aih [[REG3]], 127
|
|
|
|
|
; CHECK: stepc [[REG2]], [[REG3]]
|
|
|
|
|
; CHECK: risblg %r2, [[REG3]], 0, 159, 32
|
|
|
|
|
; CHECK: ahi %r2, 128
|
|
|
|
|
; CHECK: stepd [[REG3]], %r2
|
|
|
|
|
; CHECK: br %r14
|
|
|
|
|
%add1 = add i32 %old, 14
|
|
|
|
|
%res1 = call i32 asm "stepa $1, $2",
|
|
|
|
|
"=r,r,0"(i32 %old, i32 %add1)
|
|
|
|
|
%add2 = add i32 %res1, 254
|
|
|
|
|
%res2 = call i32 asm "stepb $1, $2",
|
|
|
|
|
"=h,r,0"(i32 %res1, i32 %add2)
|
|
|
|
|
%add3 = add i32 %res2, 127
|
|
|
|
|
%res3 = call i32 asm "stepc $1, $2",
|
|
|
|
|
"=h,h,0"(i32 %res2, i32 %add3)
|
|
|
|
|
%add4 = add i32 %res3, 128
|
|
|
|
|
%res4 = call i32 asm "stepd $1, $2",
|
|
|
|
|
"=r,h,0"(i32 %res3, i32 %add4)
|
|
|
|
|
ret i32 %res4
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
; Test large immediate addition involving high registers.
|
|
|
|
|
define void @f29() {
|
|
|
|
|
; CHECK-LABEL: f29:
|
|
|
|
|
; CHECK: stepa [[REG:%r[0-5]]]
|
|
|
|
|
; CHECK: aih [[REG]], -32769
|
|
|
|
|
; CHECK: stepb [[REG]]
|
|
|
|
|
; CHECK: aih [[REG]], 32768
|
|
|
|
|
; CHECK: stepc [[REG]]
|
|
|
|
|
; CHECK: aih [[REG]], 1000000000
|
|
|
|
|
; CHECK: stepd [[REG]]
|
|
|
|
|
; CHECK: br %r14
|
|
|
|
|
%res1 = call i32 asm "stepa $0", "=h"()
|
|
|
|
|
%add1 = add i32 %res1, -32769
|
|
|
|
|
%res2 = call i32 asm "stepb $0, $1", "=h,h"(i32 %add1)
|
|
|
|
|
%add2 = add i32 %res2, 32768
|
|
|
|
|
%res3 = call i32 asm "stepc $0, $1", "=h,h"(i32 %add2)
|
|
|
|
|
%add3 = add i32 %res3, 1000000000
|
|
|
|
|
call void asm sideeffect "stepd $0", "h"(i32 %add3)
|
|
|
|
|
ret void
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
; Test large immediate addition involving low registers.
|
|
|
|
|
define void @f30() {
|
|
|
|
|
; CHECK-LABEL: f30:
|
|
|
|
|
; CHECK: stepa [[REG:%r[0-5]]]
|
|
|
|
|
; CHECK: afi [[REG]], -32769
|
|
|
|
|
; CHECK: stepb [[REG]]
|
|
|
|
|
; CHECK: afi [[REG]], 32768
|
|
|
|
|
; CHECK: stepc [[REG]]
|
|
|
|
|
; CHECK: afi [[REG]], 1000000000
|
|
|
|
|
; CHECK: stepd [[REG]]
|
|
|
|
|
; CHECK: br %r14
|
|
|
|
|
%res1 = call i32 asm "stepa $0", "=r"()
|
|
|
|
|
%add1 = add i32 %res1, -32769
|
|
|
|
|
%res2 = call i32 asm "stepb $0, $1", "=r,r"(i32 %add1)
|
|
|
|
|
%add2 = add i32 %res2, 32768
|
|
|
|
|
%res3 = call i32 asm "stepc $0, $1", "=r,r"(i32 %add2)
|
|
|
|
|
%add3 = add i32 %res3, 1000000000
|
|
|
|
|
call void asm sideeffect "stepd $0", "r"(i32 %add3)
|
|
|
|
|
ret void
|
|
|
|
|
}
|
2013-10-01 22:56:23 +08:00
|
|
|
|
|
|
|
|
; Test large immediate comparison involving high registers.
|
|
|
|
|
define i32 @f31() {
|
|
|
|
|
; CHECK-LABEL: f31:
|
|
|
|
|
; CHECK: stepa [[REG1:%r[0-5]]]
|
|
|
|
|
; CHECK: cih [[REG1]], 1000000000
|
|
|
|
|
; CHECK: stepb [[REG2:%r[0-5]]]
|
|
|
|
|
; CHECK: clih [[REG2]], 1000000000
|
|
|
|
|
; CHECK: br %r14
|
|
|
|
|
%res1 = call i32 asm "stepa $0", "=h"()
|
|
|
|
|
%cmp1 = icmp sle i32 %res1, 1000000000
|
|
|
|
|
%sel1 = select i1 %cmp1, i32 0, i32 1
|
|
|
|
|
%res2 = call i32 asm "stepb $0, $1", "=h,r"(i32 %sel1)
|
|
|
|
|
%cmp2 = icmp ule i32 %res2, 1000000000
|
|
|
|
|
%sel2 = select i1 %cmp2, i32 0, i32 1
|
|
|
|
|
ret i32 %sel2
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
; Test large immediate comparison involving low registers.
|
|
|
|
|
define i32 @f32() {
|
|
|
|
|
; CHECK-LABEL: f32:
|
|
|
|
|
; CHECK: stepa [[REG1:%r[0-5]]]
|
|
|
|
|
; CHECK: cfi [[REG1]], 1000000000
|
|
|
|
|
; CHECK: stepb [[REG2:%r[0-5]]]
|
|
|
|
|
; CHECK: clfi [[REG2]], 1000000000
|
|
|
|
|
; CHECK: br %r14
|
|
|
|
|
%res1 = call i32 asm "stepa $0", "=r"()
|
|
|
|
|
%cmp1 = icmp sle i32 %res1, 1000000000
|
|
|
|
|
%sel1 = select i1 %cmp1, i32 0, i32 1
|
|
|
|
|
%res2 = call i32 asm "stepb $0, $1", "=r,r"(i32 %sel1)
|
|
|
|
|
%cmp2 = icmp ule i32 %res2, 1000000000
|
|
|
|
|
%sel2 = select i1 %cmp2, i32 0, i32 1
|
|
|
|
|
ret i32 %sel2
|
|
|
|
|
}
|
2013-10-01 23:00:44 +08:00
|
|
|
|
|
|
|
|
; Test memory comparison involving high registers.
|
|
|
|
|
define void @f33(i32 *%ptr1, i32 *%ptr2) {
|
|
|
|
|
; CHECK-LABEL: f33:
|
|
|
|
|
; CHECK: stepa [[REG1:%r[0-5]]]
|
|
|
|
|
; CHECK: chf [[REG1]], 0(%r2)
|
|
|
|
|
; CHECK: stepb [[REG2:%r[0-5]]]
|
|
|
|
|
; CHECK: clhf [[REG2]], 0(%r3)
|
|
|
|
|
; CHECK: br %r14
|
|
|
|
|
%res1 = call i32 asm "stepa $0", "=h"()
|
|
|
|
|
%load1 = load i32 *%ptr1
|
|
|
|
|
%cmp1 = icmp sle i32 %res1, %load1
|
|
|
|
|
%sel1 = select i1 %cmp1, i32 0, i32 1
|
|
|
|
|
%res2 = call i32 asm "stepb $0, $1", "=h,r"(i32 %sel1)
|
|
|
|
|
%load2 = load i32 *%ptr2
|
|
|
|
|
%cmp2 = icmp ule i32 %res2, %load2
|
|
|
|
|
%sel2 = select i1 %cmp2, i32 0, i32 1
|
|
|
|
|
store i32 %sel2, i32 *%ptr1
|
|
|
|
|
ret void
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
; Test memory comparison involving low registers.
|
|
|
|
|
define void @f34(i32 *%ptr1, i32 *%ptr2) {
|
|
|
|
|
; CHECK-LABEL: f34:
|
|
|
|
|
; CHECK: stepa [[REG1:%r[0-5]]]
|
|
|
|
|
; CHECK: c [[REG1]], 0(%r2)
|
|
|
|
|
; CHECK: stepb [[REG2:%r[0-5]]]
|
|
|
|
|
; CHECK: cl [[REG2]], 0(%r3)
|
|
|
|
|
; CHECK: br %r14
|
|
|
|
|
%res1 = call i32 asm "stepa $0", "=r"()
|
|
|
|
|
%load1 = load i32 *%ptr1
|
|
|
|
|
%cmp1 = icmp sle i32 %res1, %load1
|
|
|
|
|
%sel1 = select i1 %cmp1, i32 0, i32 1
|
|
|
|
|
%res2 = call i32 asm "stepb $0, $1", "=r,r"(i32 %sel1)
|
|
|
|
|
%load2 = load i32 *%ptr2
|
|
|
|
|
%cmp2 = icmp ule i32 %res2, %load2
|
|
|
|
|
%sel2 = select i1 %cmp2, i32 0, i32 1
|
|
|
|
|
store i32 %sel2, i32 *%ptr1
|
|
|
|
|
ret void
|
|
|
|
|
}
|