2014-05-30 18:09:59 +08:00
|
|
|
; RUN: llc -mtriple=arm-eabi -arm-atomic-cfg-tidy=0 %s -o - | FileCheck -check-prefix=ARM %s
|
|
|
|
|
; RUN: llc -mtriple=thumb-eabi -arm-atomic-cfg-tidy=0 %s -o - | FileCheck -check-prefix=THUMB %s
|
2017-02-25 05:19:09 +08:00
|
|
|
; RUN: llc -mtriple=thumb-eabi -arm-atomic-cfg-tidy=0 -mcpu=arm1156t2-s -mattr=+thumb2 %s -o - | FileCheck -check-prefix=T2 %s
|
2014-05-30 18:09:59 +08:00
|
|
|
; RUN: llc -mtriple=thumbv8-eabi -arm-atomic-cfg-tidy=0 %s -o - | FileCheck -check-prefix=V8 %s
|
When we look at instructions to convert to setting the 's' flag, we need to look
at more than those which define CPSR. You can have this situation:
(1) subs ...
(2) sub r6, r5, r4
(3) movge ...
(4) cmp r6, 0
(5) movge ...
We cannot convert (2) to "subs" because (3) is using the CPSR set by
(1). There's an analogous situation here:
(1) sub r1, r2, r3
(2) sub r4, r5, r6
(3) cmp r4, ...
(5) movge ...
(6) cmp r1, ...
(7) movge ...
We cannot convert (1) to "subs" because of the intervening use of CPSR.
llvm-svn: 117950
2010-11-02 04:41:43 +08:00
|
|
|
|
2017-08-02 06:20:49 +08:00
|
|
|
; FIXME: The -mtriple=thumb test doesn't change if -disable-peephole is specified.
|
2010-09-14 16:50:43 +08:00
|
|
|
|
|
|
|
|
%struct.Foo = type { i8* }
|
|
|
|
|
|
2015-04-29 06:03:55 +08:00
|
|
|
; ARM-LABEL: foo:
|
|
|
|
|
; THUMB-LABEL: foo:
|
|
|
|
|
; T2-LABEL: foo:
|
When we look at instructions to convert to setting the 's' flag, we need to look
at more than those which define CPSR. You can have this situation:
(1) subs ...
(2) sub r6, r5, r4
(3) movge ...
(4) cmp r6, 0
(5) movge ...
We cannot convert (2) to "subs" because (3) is using the CPSR set by
(1). There's an analogous situation here:
(1) sub r1, r2, r3
(2) sub r4, r5, r6
(3) cmp r4, ...
(5) movge ...
(6) cmp r1, ...
(7) movge ...
We cannot convert (1) to "subs" because of the intervening use of CPSR.
llvm-svn: 117950
2010-11-02 04:41:43 +08:00
|
|
|
define %struct.Foo* @foo(%struct.Foo* %this, i32 %acc) nounwind readonly align 2 {
|
2010-09-14 16:50:43 +08:00
|
|
|
entry:
|
[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
|
|
|
%scevgep = getelementptr %struct.Foo, %struct.Foo* %this, i32 1
|
2010-09-14 16:50:43 +08:00
|
|
|
br label %tailrecurse
|
|
|
|
|
|
|
|
|
|
tailrecurse: ; preds = %sw.bb, %entry
|
|
|
|
|
%lsr.iv2 = phi %struct.Foo* [ %scevgep3, %sw.bb ], [ %scevgep, %entry ]
|
|
|
|
|
%lsr.iv = phi i32 [ %lsr.iv.next, %sw.bb ], [ 1, %entry ]
|
|
|
|
|
%acc.tr = phi i32 [ %or, %sw.bb ], [ %acc, %entry ]
|
|
|
|
|
%lsr.iv24 = bitcast %struct.Foo* %lsr.iv2 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
|
|
|
%scevgep5 = getelementptr i8*, i8** %lsr.iv24, i32 -1
|
2015-02-28 05:17:42 +08:00
|
|
|
%tmp2 = load i8*, i8** %scevgep5
|
2010-09-14 16:50:43 +08:00
|
|
|
%0 = ptrtoint i8* %tmp2 to i32
|
|
|
|
|
|
2011-04-30 09:37:52 +08:00
|
|
|
; ARM: ands {{r[0-9]+}}, {{r[0-9]+}}, #3
|
When we look at instructions to convert to setting the 's' flag, we need to look
at more than those which define CPSR. You can have this situation:
(1) subs ...
(2) sub r6, r5, r4
(3) movge ...
(4) cmp r6, 0
(5) movge ...
We cannot convert (2) to "subs" because (3) is using the CPSR set by
(1). There's an analogous situation here:
(1) sub r1, r2, r3
(2) sub r4, r5, r6
(3) cmp r4, ...
(5) movge ...
(6) cmp r1, ...
(7) movge ...
We cannot convert (1) to "subs" because of the intervening use of CPSR.
llvm-svn: 117950
2010-11-02 04:41:43 +08:00
|
|
|
; ARM-NEXT: beq
|
2010-09-14 16:50:43 +08:00
|
|
|
|
2016-11-03 22:08:01 +08:00
|
|
|
; THUMB: movs r[[R0:[0-9]+]], #3
|
|
|
|
|
; THUMB-NEXT: ands r[[R0]], r
|
|
|
|
|
; THUMB-NEXT: cmp r[[R0]], #0
|
2010-09-29 18:45:43 +08:00
|
|
|
; THUMB-NEXT: beq
|
2010-09-29 18:12:08 +08:00
|
|
|
|
2016-11-03 22:08:01 +08:00
|
|
|
; T2: ands {{r[0-9]+}}, {{r[0-9]+}}, #3
|
2010-09-29 18:45:43 +08:00
|
|
|
; T2-NEXT: beq
|
2010-09-14 16:50:43 +08:00
|
|
|
|
|
|
|
|
%and = and i32 %0, 3
|
|
|
|
|
%tst = icmp eq i32 %and, 0
|
|
|
|
|
br i1 %tst, label %sw.bb, label %tailrecurse.switch
|
|
|
|
|
|
|
|
|
|
tailrecurse.switch: ; preds = %tailrecurse
|
2013-09-09 22:21:49 +08:00
|
|
|
; V8-LABEL: %tailrecurse.switch
|
|
|
|
|
; V8: cmp
|
|
|
|
|
; V8-NEXT: beq
|
|
|
|
|
; V8-NEXT: %tailrecurse.switch
|
|
|
|
|
; V8: cmp
|
|
|
|
|
; V8-NEXT: beq
|
|
|
|
|
; V8-NEXT: %tailrecurse.switch
|
|
|
|
|
; V8: cmp
|
2017-02-01 07:48:32 +08:00
|
|
|
; V8-NEXT: beq
|
|
|
|
|
; V8-NEXT: %sw.epilog
|
|
|
|
|
; V8-NEXT: bx lr
|
2010-09-14 16:50:43 +08:00
|
|
|
switch i32 %and, label %sw.epilog [
|
|
|
|
|
i32 1, label %sw.bb
|
|
|
|
|
i32 3, label %sw.bb6
|
|
|
|
|
i32 2, label %sw.bb8
|
2015-04-28 07:35:22 +08:00
|
|
|
], !prof !1
|
2010-09-14 16:50:43 +08:00
|
|
|
|
|
|
|
|
sw.bb: ; preds = %tailrecurse.switch, %tailrecurse
|
|
|
|
|
%shl = shl i32 %acc.tr, 1
|
|
|
|
|
%or = or i32 %and, %shl
|
|
|
|
|
%lsr.iv.next = add i32 %lsr.iv, 1
|
[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
|
|
|
%scevgep3 = getelementptr %struct.Foo, %struct.Foo* %lsr.iv2, i32 1
|
2010-09-14 16:50:43 +08:00
|
|
|
br label %tailrecurse
|
|
|
|
|
|
|
|
|
|
sw.bb6: ; preds = %tailrecurse.switch
|
|
|
|
|
ret %struct.Foo* %lsr.iv2
|
|
|
|
|
|
|
|
|
|
sw.bb8: ; preds = %tailrecurse.switch
|
|
|
|
|
%tmp1 = add i32 %acc.tr, %lsr.iv
|
[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
|
|
|
%add.ptr11 = getelementptr inbounds %struct.Foo, %struct.Foo* %this, i32 %tmp1
|
2010-09-14 16:50:43 +08:00
|
|
|
ret %struct.Foo* %add.ptr11
|
|
|
|
|
|
|
|
|
|
sw.epilog: ; preds = %tailrecurse.switch
|
|
|
|
|
ret %struct.Foo* undef
|
|
|
|
|
}
|
2010-10-15 11:23:44 +08:00
|
|
|
|
|
|
|
|
; Another test that exercises the AND/TST peephole optimization and also
|
|
|
|
|
; generates a predicated ANDS instruction. Check that the predicate is printed
|
|
|
|
|
; after the "S" modifier on the instruction.
|
|
|
|
|
|
|
|
|
|
%struct.S = type { i8* (i8*)*, [1 x i8] }
|
|
|
|
|
|
2015-04-29 06:03:55 +08:00
|
|
|
; ARM-LABEL: bar:
|
|
|
|
|
; THUMB-LABEL: bar:
|
|
|
|
|
; T2-LABEL: bar:
|
2013-09-09 22:21:49 +08:00
|
|
|
; V8-LABEL: bar:
|
2010-10-15 11:23:44 +08:00
|
|
|
define internal zeroext i8 @bar(%struct.S* %x, %struct.S* nocapture %y) nounwind readonly {
|
|
|
|
|
entry:
|
[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
|
|
|
%0 = getelementptr inbounds %struct.S, %struct.S* %x, i32 0, i32 1, i32 0
|
2015-02-28 05:17:42 +08:00
|
|
|
%1 = load i8, i8* %0, align 1
|
2010-10-15 11:23:44 +08:00
|
|
|
%2 = zext i8 %1 to i32
|
When we look at instructions to convert to setting the 's' flag, we need to look
at more than those which define CPSR. You can have this situation:
(1) subs ...
(2) sub r6, r5, r4
(3) movge ...
(4) cmp r6, 0
(5) movge ...
We cannot convert (2) to "subs" because (3) is using the CPSR set by
(1). There's an analogous situation here:
(1) sub r1, r2, r3
(2) sub r4, r5, r6
(3) cmp r4, ...
(5) movge ...
(6) cmp r1, ...
(7) movge ...
We cannot convert (1) to "subs" because of the intervening use of CPSR.
llvm-svn: 117950
2010-11-02 04:41:43 +08:00
|
|
|
; ARM: ands
|
2017-02-25 05:19:09 +08:00
|
|
|
; THUMB: ands
|
2010-10-15 11:23:44 +08:00
|
|
|
; T2: ands
|
2013-09-09 22:21:49 +08:00
|
|
|
; V8: ands
|
|
|
|
|
; V8-NEXT: beq
|
2010-10-15 11:23:44 +08:00
|
|
|
%3 = and i32 %2, 112
|
|
|
|
|
%4 = icmp eq i32 %3, 0
|
|
|
|
|
br i1 %4, label %return, label %bb
|
|
|
|
|
|
|
|
|
|
bb: ; preds = %entry
|
2013-09-09 22:21:49 +08:00
|
|
|
; V8-NEXT: %bb
|
[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
|
|
|
%5 = getelementptr inbounds %struct.S, %struct.S* %y, i32 0, i32 1, i32 0
|
2015-02-28 05:17:42 +08:00
|
|
|
%6 = load i8, i8* %5, align 1
|
2010-10-15 11:23:44 +08:00
|
|
|
%7 = zext i8 %6 to i32
|
When we look at instructions to convert to setting the 's' flag, we need to look
at more than those which define CPSR. You can have this situation:
(1) subs ...
(2) sub r6, r5, r4
(3) movge ...
(4) cmp r6, 0
(5) movge ...
We cannot convert (2) to "subs" because (3) is using the CPSR set by
(1). There's an analogous situation here:
(1) sub r1, r2, r3
(2) sub r4, r5, r6
(3) cmp r4, ...
(5) movge ...
(6) cmp r1, ...
(7) movge ...
We cannot convert (1) to "subs" because of the intervening use of CPSR.
llvm-svn: 117950
2010-11-02 04:41:43 +08:00
|
|
|
; ARM: andsne
|
2010-10-15 11:23:44 +08:00
|
|
|
; THUMB: ands
|
|
|
|
|
; T2: andsne
|
2013-09-09 22:21:49 +08:00
|
|
|
; V8: ands
|
|
|
|
|
; V8-NEXT: beq
|
2010-10-15 11:23:44 +08:00
|
|
|
%8 = and i32 %7, 112
|
|
|
|
|
%9 = icmp eq i32 %8, 0
|
|
|
|
|
br i1 %9, label %return, label %bb2
|
|
|
|
|
|
|
|
|
|
bb2: ; preds = %bb
|
2013-09-09 22:21:49 +08:00
|
|
|
; V8-NEXT: %bb2
|
|
|
|
|
; V8-NEXT: cmp
|
|
|
|
|
; V8-NEXT: it ne
|
|
|
|
|
; V8-NEXT: cmpne
|
|
|
|
|
; V8-NEXT: bne
|
2010-10-15 11:23:44 +08:00
|
|
|
%10 = icmp eq i32 %3, 16
|
|
|
|
|
%11 = icmp eq i32 %8, 16
|
|
|
|
|
%or.cond = or i1 %10, %11
|
|
|
|
|
br i1 %or.cond, label %bb4, label %return
|
|
|
|
|
|
|
|
|
|
bb4: ; preds = %bb2
|
|
|
|
|
%12 = ptrtoint %struct.S* %x to i32
|
|
|
|
|
%phitmp = trunc i32 %12 to i8
|
|
|
|
|
ret i8 %phitmp
|
|
|
|
|
|
|
|
|
|
return: ; preds = %bb2, %bb, %entry
|
|
|
|
|
ret i8 1
|
|
|
|
|
}
|
2015-04-28 07:35:22 +08:00
|
|
|
|
2015-04-29 06:03:55 +08:00
|
|
|
|
|
|
|
|
; We were looking through multiple COPY instructions to find an AND we might
|
|
|
|
|
; fold into a TST, but in doing so we changed the register being tested allowing
|
|
|
|
|
; folding of unrelated tests (in this case, a TST against r1 was eliminated in
|
|
|
|
|
; favour of an AND of r0).
|
|
|
|
|
|
[ARM] don't transform an add(ext Cond), C to select unless there's a setcc of the condition
The transform in question claims to be doing:
// fold (add (select cc, 0, c), x) -> (select cc, x, (add, x, c))
...starting in PerformADDCombineWithOperands(), but it wasn't actually checking for a setcc node
for the sext/zext patterns.
This is exactly the opposite of a transform I'd like to add to DAGCombiner's foldSelectOfConstants(),
so I was seeing infinite loops with my draft of a patch applied.
The changes in select_const.ll look positive (less instructions). The change in arm-and-tst-peephole.ll
is unrelated. We're changing the input IR in that test to preserve the intent of the test, but that's
not affected by this code change.
Differential Revision:
https://reviews.llvm.org/D30355
llvm-svn: 296389
2017-02-28 05:30:54 +08:00
|
|
|
define i32 @test_tst_assessment(i32 %a, i32 %b) {
|
2015-04-29 06:03:55 +08:00
|
|
|
; ARM-LABEL: test_tst_assessment:
|
2017-12-05 01:18:51 +08:00
|
|
|
; ARM: @ %bb.0:
|
2017-02-25 05:19:09 +08:00
|
|
|
; ARM-NEXT: and r0, r0, #1
|
|
|
|
|
; ARM-NEXT: tst r1, #1
|
|
|
|
|
; ARM-NEXT: subne r0, r0, #1
|
|
|
|
|
; ARM-NEXT: mov pc, lr
|
|
|
|
|
;
|
2015-04-29 06:03:55 +08:00
|
|
|
; THUMB-LABEL: test_tst_assessment:
|
2017-12-05 01:18:51 +08:00
|
|
|
; THUMB: @ %bb.0:
|
2017-03-07 17:38:16 +08:00
|
|
|
; THUMB-NEXT: movs r2, r0
|
[ARM] don't transform an add(ext Cond), C to select unless there's a setcc of the condition
The transform in question claims to be doing:
// fold (add (select cc, 0, c), x) -> (select cc, x, (add, x, c))
...starting in PerformADDCombineWithOperands(), but it wasn't actually checking for a setcc node
for the sext/zext patterns.
This is exactly the opposite of a transform I'd like to add to DAGCombiner's foldSelectOfConstants(),
so I was seeing infinite loops with my draft of a patch applied.
The changes in select_const.ll look positive (less instructions). The change in arm-and-tst-peephole.ll
is unrelated. We're changing the input IR in that test to preserve the intent of the test, but that's
not affected by this code change.
Differential Revision:
https://reviews.llvm.org/D30355
llvm-svn: 296389
2017-02-28 05:30:54 +08:00
|
|
|
; THUMB-NEXT: movs r0, #1
|
|
|
|
|
; THUMB-NEXT: ands r0, r2
|
2017-02-25 05:19:09 +08:00
|
|
|
; THUMB-NEXT: lsls r1, r1, #31
|
[ARM] don't transform an add(ext Cond), C to select unless there's a setcc of the condition
The transform in question claims to be doing:
// fold (add (select cc, 0, c), x) -> (select cc, x, (add, x, c))
...starting in PerformADDCombineWithOperands(), but it wasn't actually checking for a setcc node
for the sext/zext patterns.
This is exactly the opposite of a transform I'd like to add to DAGCombiner's foldSelectOfConstants(),
so I was seeing infinite loops with my draft of a patch applied.
The changes in select_const.ll look positive (less instructions). The change in arm-and-tst-peephole.ll
is unrelated. We're changing the input IR in that test to preserve the intent of the test, but that's
not affected by this code change.
Differential Revision:
https://reviews.llvm.org/D30355
llvm-svn: 296389
2017-02-28 05:30:54 +08:00
|
|
|
; THUMB-NEXT: beq .LBB2_2
|
2017-12-05 01:18:51 +08:00
|
|
|
; THUMB-NEXT: @ %bb.1:
|
2018-10-27 03:32:24 +08:00
|
|
|
; THUMB-NEXT: subs r0, r0, #1
|
2017-02-25 05:19:09 +08:00
|
|
|
; THUMB-NEXT: .LBB2_2:
|
|
|
|
|
; THUMB-NEXT: bx lr
|
|
|
|
|
;
|
2015-04-29 06:03:55 +08:00
|
|
|
; T2-LABEL: test_tst_assessment:
|
2017-12-05 01:18:51 +08:00
|
|
|
; T2: @ %bb.0:
|
2017-02-25 05:19:09 +08:00
|
|
|
; T2-NEXT: and r0, r0, #1
|
2018-10-30 23:04:40 +08:00
|
|
|
; T2-NEXT: lsls r1, r1, #31
|
2017-02-25 05:19:09 +08:00
|
|
|
; T2-NEXT: it ne
|
|
|
|
|
; T2-NEXT: subne r0, #1
|
|
|
|
|
; T2-NEXT: bx lr
|
|
|
|
|
;
|
2015-04-29 06:03:55 +08:00
|
|
|
; V8-LABEL: test_tst_assessment:
|
2017-12-05 01:18:51 +08:00
|
|
|
; V8: @ %bb.0:
|
2017-02-25 05:19:09 +08:00
|
|
|
; V8-NEXT: and r0, r0, #1
|
2017-06-28 15:07:03 +08:00
|
|
|
; V8-NEXT: lsls r1, r1, #31
|
2017-02-25 05:19:09 +08:00
|
|
|
; V8-NEXT: it ne
|
|
|
|
|
; V8-NEXT: subne r0, #1
|
|
|
|
|
; V8-NEXT: bx lr
|
[ARM] don't transform an add(ext Cond), C to select unless there's a setcc of the condition
The transform in question claims to be doing:
// fold (add (select cc, 0, c), x) -> (select cc, x, (add, x, c))
...starting in PerformADDCombineWithOperands(), but it wasn't actually checking for a setcc node
for the sext/zext patterns.
This is exactly the opposite of a transform I'd like to add to DAGCombiner's foldSelectOfConstants(),
so I was seeing infinite loops with my draft of a patch applied.
The changes in select_const.ll look positive (less instructions). The change in arm-and-tst-peephole.ll
is unrelated. We're changing the input IR in that test to preserve the intent of the test, but that's
not affected by this code change.
Differential Revision:
https://reviews.llvm.org/D30355
llvm-svn: 296389
2017-02-28 05:30:54 +08:00
|
|
|
%and1 = and i32 %a, 1
|
|
|
|
|
%sub = sub i32 %and1, 1
|
|
|
|
|
%and2 = and i32 %b, 1
|
|
|
|
|
%cmp = icmp eq i32 %and2, 0
|
|
|
|
|
%sel = select i1 %cmp, i32 %and1, i32 %sub
|
|
|
|
|
ret i32 %sel
|
2015-04-29 06:03:55 +08:00
|
|
|
}
|
|
|
|
|
|
2015-04-28 07:35:22 +08:00
|
|
|
!1 = !{!"branch_weights", i32 1, i32 1, i32 3, i32 2 }
|