2015-03-05 02:43:29 +08:00
|
|
|
; RUN: opt -loop-reduce -disable-output -debug-only=loop-reduce < %s 2>&1 | FileCheck %s
|
2012-07-19 02:07:52 +08:00
|
|
|
; REQUIRES: asserts
|
|
|
|
|
;
|
|
|
|
|
; PR13361: LSR + SCEV "hangs" on reasonably sized test with sequence of loops
|
|
|
|
|
;
|
|
|
|
|
; Without limits on CollectSubexpr, we have thousands of formulae for
|
2013-01-16 21:19:59 +08:00
|
|
|
; the use that crosses loops. With limits we have five.
|
2012-07-19 02:07:52 +08:00
|
|
|
; CHECK: LSR on loop %bb221:
|
|
|
|
|
; CHECK: After generating reuse formulae:
|
|
|
|
|
; CHECK: LSR is examining the following uses:
|
|
|
|
|
; CHECK: LSR Use: Kind=Special
|
Switch the SCEV expander and LoopStrengthReduce to use
TargetTransformInfo rather than TargetLowering, removing one of the
primary instances of the layering violation of Transforms depending
directly on Target.
This is a really big deal because LSR used to be a "special" pass that
could only be tested fully using llc and by looking at the full output
of it. It also couldn't run with any other loop passes because it had to
be created by the backend. No longer is this true. LSR is now just
a normal pass and we should probably lift the creation of LSR out of
lib/CodeGen/Passes.cpp and into the PassManagerBuilder. =] I've not done
this, or updated all of the tests to use opt and a triple, because
I suspect someone more familiar with LSR would do a better job. This
change should be essentially without functional impact for normal
compilations, and only change behvaior of targetless compilations.
The conversion required changing all of the LSR code to refer to the TTI
interfaces, which fortunately are very similar to TargetLowering's
interfaces. However, it also allowed us to *always* expect to have some
implementation around. I've pushed that simplification through the pass,
and leveraged it to simplify code somewhat. It required some test
updates for one of two things: either we used to skip some checks
altogether but now we get the default "no" answer for them, or we used
to have no information about the target and now we do have some.
I've also started the process of removing AddrMode, as the TTI interface
doesn't use it any longer. In some cases this simplifies code, and in
others it adds some complexity, but I think it's not a bad tradeoff even
there. Subsequent patches will try to clean this up even further and use
other (more appropriate) abstractions.
Yet again, almost all of the formatting changes brought to you by
clang-format. =]
llvm-svn: 171735
2013-01-07 22:41:08 +08:00
|
|
|
; CHECK: {{.*reg\(\{.*\{.*\{.*\{.*\{.*\{.*\{.*\{.*\{}}
|
|
|
|
|
; CHECK: {{.*reg\(\{.*\{.*\{.*\{.*\{.*\{.*\{.*\{.*\{}}
|
|
|
|
|
; CHECK: {{.*reg\(\{.*\{.*\{.*\{.*\{.*\{.*\{.*\{.*\{}}
|
|
|
|
|
; CHECK: {{.*reg\(\{.*\{.*\{.*\{.*\{.*\{.*\{.*\{.*\{}}
|
|
|
|
|
; CHECK: {{.*reg\(\{.*\{.*\{.*\{.*\{.*\{.*\{.*\{.*\{}}
|
2012-07-19 02:07:52 +08:00
|
|
|
; CHECK-NOT:reg
|
|
|
|
|
; CHECK: Filtering for use
|
|
|
|
|
|
2015-03-05 02:43:29 +08:00
|
|
|
; Provide legal integer types.
|
|
|
|
|
target datalayout = "n8:16:32:64"
|
|
|
|
|
|
|
|
|
|
|
2012-07-19 02:07:52 +08:00
|
|
|
%struct.snork = type { %struct.fuga, i32, i32, i32, i32, i32, i32 }
|
|
|
|
|
%struct.fuga = type { %struct.gork, i64 }
|
|
|
|
|
%struct.gork = type { i8*, i32, i32, %struct.noot* }
|
|
|
|
|
%struct.noot = type opaque
|
|
|
|
|
%struct.jim = type { [5120 x i8], i32, i32, [2048 x i8], i32, [256 x i8] }
|
|
|
|
|
|
|
|
|
|
@global = external global %struct.snork, align 8
|
|
|
|
|
@global1 = external hidden unnamed_addr constant [52 x i8], align 1
|
|
|
|
|
@global2 = external hidden unnamed_addr constant [18 x i8], align 1
|
|
|
|
|
@global3 = external hidden global %struct.jim, align 32
|
|
|
|
|
@global4 = external hidden unnamed_addr constant [40 x i8], align 1
|
|
|
|
|
|
|
|
|
|
declare void @snork(...) nounwind
|
|
|
|
|
|
|
|
|
|
declare fastcc void @blarg() nounwind uwtable readonly
|
|
|
|
|
|
|
|
|
|
define hidden fastcc void @boogle() nounwind uwtable {
|
|
|
|
|
bb:
|
|
|
|
|
%tmp = trunc i64 0 to i32
|
|
|
|
|
%tmp1 = icmp slt i32 %tmp, 2047
|
|
|
|
|
%tmp2 = add i32 0, -1
|
|
|
|
|
%tmp3 = icmp ult i32 %tmp2, 255
|
|
|
|
|
%tmp4 = and i1 %tmp1, %tmp3
|
|
|
|
|
br i1 %tmp4, label %bb6, label %bb5
|
|
|
|
|
|
|
|
|
|
bb5: ; preds = %bb
|
[opaque pointer type] Add textual IR support for explicit type parameter to the call instruction
See r230786 and r230794 for similar changes to gep and load
respectively.
Call is a bit different because it often doesn't have a single explicit
type - usually the type is deduced from the arguments, and just the
return type is explicit. In those cases there's no need to change the
IR.
When that's not the case, the IR usually contains the pointer type of
the first operand - but since typed pointers are going away, that
representation is insufficient so I'm just stripping the "pointerness"
of the explicit type away.
This does make the IR a bit weird - it /sort of/ reads like the type of
the first operand: "call void () %x(" but %x is actually of type "void
()*" and will eventually be just of type "ptr". But this seems not too
bad and I don't think it would benefit from repeating the type
("void (), void () * %x(" and then eventually "void (), ptr %x(") as has
been done with gep and load.
This also has a side benefit: since the explicit type is no longer a
pointer, there's no ambiguity between an explicit type and a function
that returns a function pointer. Previously this case needed an explicit
type (eg: a function returning a void() function was written as
"call void () () * @x(" rather than "call void () * @x(" because of the
ambiguity between a function returning a pointer to a void() function
and a function returning void).
No ambiguity means even function pointer return types can just be
written alone, without writing the whole function's type.
This leaves /only/ the varargs case where the explicit type is required.
Given the special type syntax in call instructions, the regex-fu used
for migration was a bit more involved in its own unique way (as every
one of these is) so here it is. Use it in conjunction with the apply.sh
script and associated find/xargs commands I've provided in rr230786 to
migrate your out of tree tests. Do let me know if any of this doesn't
cover your cases & we can iterate on a more general script/regexes to
help others with out of tree tests.
About 9 test cases couldn't be automatically migrated - half of those
were functions returning function pointers, where I just had to manually
delete the function argument types now that we didn't need an explicit
function type there. The other half were typedefs of function types used
in calls - just had to manually drop the * from those.
import fileinput
import sys
import re
pat = re.compile(r'((?:=|:|^|\s)call\s(?:[^@]*?))(\s*$|\s*(?:(?:\[\[[a-zA-Z0-9_]+\]\]|[@%](?:(")?[\\\?@a-zA-Z0-9_.]*?(?(3)"|)|{{.*}}))(?:\(|$)|undef|inttoptr|bitcast|null|asm).*$)')
addrspace_end = re.compile(r"addrspace\(\d+\)\s*\*$")
func_end = re.compile("(?:void.*|\)\s*)\*$")
def conv(match, line):
if not match or re.search(addrspace_end, match.group(1)) or not re.search(func_end, match.group(1)):
return line
return line[:match.start()] + match.group(1)[:match.group(1).rfind('*')].rstrip() + match.group(2) + line[match.end():]
for line in sys.stdin:
sys.stdout.write(conv(re.search(pat, line), line))
llvm-svn: 235145
2015-04-17 07:24:18 +08:00
|
|
|
tail call void (...) @snork(i8* getelementptr inbounds ([52 x i8], [52 x i8]* @global1, i64 0, i64 0), i32 2021) nounwind
|
|
|
|
|
tail call void (...) @snork(i8* getelementptr inbounds (%struct.jim, %struct.jim* @global3, i64 0, i32 3, i64 1), i32 -2146631418) nounwind
|
2012-07-19 02:07:52 +08:00
|
|
|
unreachable
|
|
|
|
|
|
|
|
|
|
bb6: ; preds = %bb
|
2015-03-14 02:20:45 +08:00
|
|
|
tail call void @zot(i8* getelementptr inbounds (%struct.jim, %struct.jim* @global3, i64 0, i32 5, i64 0), i8* getelementptr inbounds (%struct.jim, %struct.jim* @global3, i64 0, i32 3, i64 1), i64 undef, i32 1, i1 false) nounwind
|
[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
|
|
|
%tmp7 = getelementptr inbounds %struct.jim, %struct.jim* @global3, i64 0, i32 5, i64 undef
|
2012-07-19 02:07:52 +08:00
|
|
|
store i8 0, i8* %tmp7, align 1
|
|
|
|
|
%tmp8 = add nsw i32 0, 1
|
|
|
|
|
%tmp9 = sext i32 %tmp8 to i64
|
|
|
|
|
%tmp10 = add i64 %tmp9, 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
|
|
|
%tmp11 = getelementptr inbounds %struct.jim, %struct.jim* @global3, i64 0, i32 3, i64 %tmp10
|
2012-07-19 02:07:52 +08:00
|
|
|
%tmp12 = sub i64 2047, %tmp9
|
|
|
|
|
%tmp13 = icmp eq i32 undef, 1
|
|
|
|
|
br i1 %tmp13, label %bb14, label %bb15
|
|
|
|
|
|
|
|
|
|
bb14: ; preds = %bb6
|
|
|
|
|
tail call fastcc void @blarg()
|
|
|
|
|
unreachable
|
|
|
|
|
|
|
|
|
|
bb15: ; preds = %bb6
|
|
|
|
|
%tmp16 = trunc i64 %tmp12 to i32
|
|
|
|
|
br label %bb17
|
|
|
|
|
|
|
|
|
|
bb17: ; preds = %bb26, %bb15
|
|
|
|
|
%tmp18 = phi i64 [ %tmp28, %bb26 ], [ 0, %bb15 ]
|
|
|
|
|
%tmp19 = phi i32 [ %tmp29, %bb26 ], [ 0, %bb15 ]
|
|
|
|
|
%tmp20 = trunc i64 %tmp18 to i32
|
|
|
|
|
%tmp21 = icmp slt i32 %tmp20, %tmp16
|
|
|
|
|
br i1 %tmp21, label %bb22, label %bb32
|
|
|
|
|
|
|
|
|
|
bb22: ; preds = %bb17
|
[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
|
|
|
%tmp23 = getelementptr inbounds %struct.jim, %struct.jim* @global3, i64 0, i32 3, i64 0
|
2015-02-28 05:17:42 +08:00
|
|
|
%tmp24 = load i8, i8* %tmp23, align 1
|
2012-07-19 02:07:52 +08:00
|
|
|
%tmp25 = icmp eq i8 %tmp24, 58
|
|
|
|
|
br i1 %tmp25, label %bb30, label %bb26
|
|
|
|
|
|
|
|
|
|
bb26: ; preds = %bb22
|
|
|
|
|
%tmp27 = icmp eq i8 %tmp24, 0
|
|
|
|
|
%tmp28 = add i64 %tmp18, 1
|
|
|
|
|
%tmp29 = add nsw i32 %tmp19, 1
|
|
|
|
|
br i1 %tmp27, label %bb32, label %bb17
|
|
|
|
|
|
|
|
|
|
bb30: ; preds = %bb22
|
|
|
|
|
%tmp31 = icmp ult i32 undef, 255
|
|
|
|
|
br i1 %tmp31, label %bb33, label %bb32
|
|
|
|
|
|
|
|
|
|
bb32: ; preds = %bb30, %bb26, %bb17
|
[opaque pointer type] Add textual IR support for explicit type parameter to the call instruction
See r230786 and r230794 for similar changes to gep and load
respectively.
Call is a bit different because it often doesn't have a single explicit
type - usually the type is deduced from the arguments, and just the
return type is explicit. In those cases there's no need to change the
IR.
When that's not the case, the IR usually contains the pointer type of
the first operand - but since typed pointers are going away, that
representation is insufficient so I'm just stripping the "pointerness"
of the explicit type away.
This does make the IR a bit weird - it /sort of/ reads like the type of
the first operand: "call void () %x(" but %x is actually of type "void
()*" and will eventually be just of type "ptr". But this seems not too
bad and I don't think it would benefit from repeating the type
("void (), void () * %x(" and then eventually "void (), ptr %x(") as has
been done with gep and load.
This also has a side benefit: since the explicit type is no longer a
pointer, there's no ambiguity between an explicit type and a function
that returns a function pointer. Previously this case needed an explicit
type (eg: a function returning a void() function was written as
"call void () () * @x(" rather than "call void () * @x(" because of the
ambiguity between a function returning a pointer to a void() function
and a function returning void).
No ambiguity means even function pointer return types can just be
written alone, without writing the whole function's type.
This leaves /only/ the varargs case where the explicit type is required.
Given the special type syntax in call instructions, the regex-fu used
for migration was a bit more involved in its own unique way (as every
one of these is) so here it is. Use it in conjunction with the apply.sh
script and associated find/xargs commands I've provided in rr230786 to
migrate your out of tree tests. Do let me know if any of this doesn't
cover your cases & we can iterate on a more general script/regexes to
help others with out of tree tests.
About 9 test cases couldn't be automatically migrated - half of those
were functions returning function pointers, where I just had to manually
delete the function argument types now that we didn't need an explicit
function type there. The other half were typedefs of function types used
in calls - just had to manually drop the * from those.
import fileinput
import sys
import re
pat = re.compile(r'((?:=|:|^|\s)call\s(?:[^@]*?))(\s*$|\s*(?:(?:\[\[[a-zA-Z0-9_]+\]\]|[@%](?:(")?[\\\?@a-zA-Z0-9_.]*?(?(3)"|)|{{.*}}))(?:\(|$)|undef|inttoptr|bitcast|null|asm).*$)')
addrspace_end = re.compile(r"addrspace\(\d+\)\s*\*$")
func_end = re.compile("(?:void.*|\)\s*)\*$")
def conv(match, line):
if not match or re.search(addrspace_end, match.group(1)) or not re.search(func_end, match.group(1)):
return line
return line[:match.start()] + match.group(1)[:match.group(1).rfind('*')].rstrip() + match.group(2) + line[match.end():]
for line in sys.stdin:
sys.stdout.write(conv(re.search(pat, line), line))
llvm-svn: 235145
2015-04-17 07:24:18 +08:00
|
|
|
tail call void (...) @snork(i8* getelementptr inbounds ([52 x i8], [52 x i8]* @global1, i64 0, i64 0), i32 2038) nounwind
|
|
|
|
|
tail call void (...) @snork(i8* %tmp11, i32 -2146631418) nounwind
|
2012-07-19 02:07:52 +08:00
|
|
|
unreachable
|
|
|
|
|
|
|
|
|
|
bb33: ; preds = %bb30
|
2015-03-14 02:20:45 +08:00
|
|
|
tail call void @zot(i8* getelementptr inbounds (%struct.jim, %struct.jim* @global3, i64 0, i32 5, i64 0), i8* %tmp11, i64 undef, i32 1, i1 false) nounwind
|
[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
|
|
|
%tmp34 = getelementptr inbounds %struct.jim, %struct.jim* @global3, i64 0, i32 5, i64 undef
|
2012-07-19 02:07:52 +08:00
|
|
|
store i8 0, i8* %tmp34, align 1
|
|
|
|
|
%tmp35 = add nsw i32 %tmp19, 1
|
|
|
|
|
%tmp36 = sext i32 %tmp35 to i64
|
|
|
|
|
%tmp37 = add i64 %tmp36, %tmp10
|
[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
|
|
|
%tmp38 = getelementptr inbounds %struct.jim, %struct.jim* @global3, i64 0, i32 3, i64 %tmp37
|
2012-07-19 02:07:52 +08:00
|
|
|
%tmp39 = sub i64 %tmp12, %tmp36
|
|
|
|
|
br i1 false, label %bb40, label %bb41
|
|
|
|
|
|
|
|
|
|
bb40: ; preds = %bb33
|
|
|
|
|
br label %bb41
|
|
|
|
|
|
|
|
|
|
bb41: ; preds = %bb40, %bb33
|
|
|
|
|
%tmp42 = trunc i64 %tmp39 to i32
|
|
|
|
|
br label %bb43
|
|
|
|
|
|
|
|
|
|
bb43: ; preds = %bb52, %bb41
|
|
|
|
|
%tmp44 = phi i64 [ %tmp53, %bb52 ], [ 0, %bb41 ]
|
|
|
|
|
%tmp45 = phi i32 [ %tmp54, %bb52 ], [ 0, %bb41 ]
|
|
|
|
|
%tmp46 = trunc i64 %tmp44 to i32
|
|
|
|
|
%tmp47 = icmp slt i32 %tmp46, %tmp42
|
|
|
|
|
br i1 %tmp47, label %bb48, label %bb58
|
|
|
|
|
|
|
|
|
|
bb48: ; preds = %bb43
|
|
|
|
|
%tmp49 = add i64 %tmp44, %tmp37
|
2015-02-28 05:17:42 +08:00
|
|
|
%tmp50 = load i8, i8* undef, align 1
|
2012-07-19 02:07:52 +08:00
|
|
|
%tmp51 = icmp eq i8 %tmp50, 58
|
|
|
|
|
br i1 %tmp51, label %bb55, label %bb52
|
|
|
|
|
|
|
|
|
|
bb52: ; preds = %bb48
|
|
|
|
|
%tmp53 = add i64 %tmp44, 1
|
|
|
|
|
%tmp54 = add nsw i32 %tmp45, 1
|
|
|
|
|
br i1 undef, label %bb58, label %bb43
|
|
|
|
|
|
|
|
|
|
bb55: ; preds = %bb48
|
|
|
|
|
%tmp56 = add i32 %tmp45, -1
|
|
|
|
|
%tmp57 = icmp ult i32 %tmp56, 255
|
|
|
|
|
br i1 %tmp57, label %bb59, label %bb58
|
|
|
|
|
|
|
|
|
|
bb58: ; preds = %bb55, %bb52, %bb43
|
[opaque pointer type] Add textual IR support for explicit type parameter to the call instruction
See r230786 and r230794 for similar changes to gep and load
respectively.
Call is a bit different because it often doesn't have a single explicit
type - usually the type is deduced from the arguments, and just the
return type is explicit. In those cases there's no need to change the
IR.
When that's not the case, the IR usually contains the pointer type of
the first operand - but since typed pointers are going away, that
representation is insufficient so I'm just stripping the "pointerness"
of the explicit type away.
This does make the IR a bit weird - it /sort of/ reads like the type of
the first operand: "call void () %x(" but %x is actually of type "void
()*" and will eventually be just of type "ptr". But this seems not too
bad and I don't think it would benefit from repeating the type
("void (), void () * %x(" and then eventually "void (), ptr %x(") as has
been done with gep and load.
This also has a side benefit: since the explicit type is no longer a
pointer, there's no ambiguity between an explicit type and a function
that returns a function pointer. Previously this case needed an explicit
type (eg: a function returning a void() function was written as
"call void () () * @x(" rather than "call void () * @x(" because of the
ambiguity between a function returning a pointer to a void() function
and a function returning void).
No ambiguity means even function pointer return types can just be
written alone, without writing the whole function's type.
This leaves /only/ the varargs case where the explicit type is required.
Given the special type syntax in call instructions, the regex-fu used
for migration was a bit more involved in its own unique way (as every
one of these is) so here it is. Use it in conjunction with the apply.sh
script and associated find/xargs commands I've provided in rr230786 to
migrate your out of tree tests. Do let me know if any of this doesn't
cover your cases & we can iterate on a more general script/regexes to
help others with out of tree tests.
About 9 test cases couldn't be automatically migrated - half of those
were functions returning function pointers, where I just had to manually
delete the function argument types now that we didn't need an explicit
function type there. The other half were typedefs of function types used
in calls - just had to manually drop the * from those.
import fileinput
import sys
import re
pat = re.compile(r'((?:=|:|^|\s)call\s(?:[^@]*?))(\s*$|\s*(?:(?:\[\[[a-zA-Z0-9_]+\]\]|[@%](?:(")?[\\\?@a-zA-Z0-9_.]*?(?(3)"|)|{{.*}}))(?:\(|$)|undef|inttoptr|bitcast|null|asm).*$)')
addrspace_end = re.compile(r"addrspace\(\d+\)\s*\*$")
func_end = re.compile("(?:void.*|\)\s*)\*$")
def conv(match, line):
if not match or re.search(addrspace_end, match.group(1)) or not re.search(func_end, match.group(1)):
return line
return line[:match.start()] + match.group(1)[:match.group(1).rfind('*')].rstrip() + match.group(2) + line[match.end():]
for line in sys.stdin:
sys.stdout.write(conv(re.search(pat, line), line))
llvm-svn: 235145
2015-04-17 07:24:18 +08:00
|
|
|
tail call void (...) @snork(i8* getelementptr inbounds ([52 x i8], [52 x i8]* @global1, i64 0, i64 0), i32 2055) nounwind
|
|
|
|
|
tail call void (...) @snork(i8* %tmp38, i32 -2146631418) nounwind
|
2012-07-19 02:07:52 +08:00
|
|
|
br label %bb247
|
|
|
|
|
|
|
|
|
|
bb59: ; preds = %bb55
|
|
|
|
|
%tmp60 = sext i32 %tmp45 to i64
|
2015-03-14 02:20:45 +08:00
|
|
|
tail call void @zot(i8* getelementptr inbounds (%struct.jim, %struct.jim* @global3, i64 0, i32 5, i64 0), i8* %tmp38, i64 %tmp60, i32 1, i1 false) nounwind
|
[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
|
|
|
%tmp61 = getelementptr inbounds %struct.jim, %struct.jim* @global3, i64 0, i32 5, i64 %tmp60
|
2012-07-19 02:07:52 +08:00
|
|
|
store i8 0, i8* %tmp61, align 1
|
|
|
|
|
%tmp62 = add nsw i32 %tmp45, 1
|
|
|
|
|
%tmp63 = sext i32 %tmp62 to i64
|
|
|
|
|
%tmp64 = add i64 %tmp63, %tmp37
|
|
|
|
|
%tmp65 = sub i64 %tmp39, %tmp63
|
|
|
|
|
%tmp66 = icmp eq i32 undef, 2
|
|
|
|
|
br i1 %tmp66, label %bb67, label %bb68
|
|
|
|
|
|
|
|
|
|
bb67: ; preds = %bb59
|
|
|
|
|
tail call fastcc void @blarg()
|
|
|
|
|
unreachable
|
|
|
|
|
|
|
|
|
|
bb68: ; preds = %bb59
|
|
|
|
|
switch i32 undef, label %bb71 [
|
|
|
|
|
i32 0, label %bb74
|
|
|
|
|
i32 -1, label %bb69
|
|
|
|
|
]
|
|
|
|
|
|
|
|
|
|
bb69: ; preds = %bb68
|
[opaque pointer type] Add textual IR support for explicit type parameter to the call instruction
See r230786 and r230794 for similar changes to gep and load
respectively.
Call is a bit different because it often doesn't have a single explicit
type - usually the type is deduced from the arguments, and just the
return type is explicit. In those cases there's no need to change the
IR.
When that's not the case, the IR usually contains the pointer type of
the first operand - but since typed pointers are going away, that
representation is insufficient so I'm just stripping the "pointerness"
of the explicit type away.
This does make the IR a bit weird - it /sort of/ reads like the type of
the first operand: "call void () %x(" but %x is actually of type "void
()*" and will eventually be just of type "ptr". But this seems not too
bad and I don't think it would benefit from repeating the type
("void (), void () * %x(" and then eventually "void (), ptr %x(") as has
been done with gep and load.
This also has a side benefit: since the explicit type is no longer a
pointer, there's no ambiguity between an explicit type and a function
that returns a function pointer. Previously this case needed an explicit
type (eg: a function returning a void() function was written as
"call void () () * @x(" rather than "call void () * @x(" because of the
ambiguity between a function returning a pointer to a void() function
and a function returning void).
No ambiguity means even function pointer return types can just be
written alone, without writing the whole function's type.
This leaves /only/ the varargs case where the explicit type is required.
Given the special type syntax in call instructions, the regex-fu used
for migration was a bit more involved in its own unique way (as every
one of these is) so here it is. Use it in conjunction with the apply.sh
script and associated find/xargs commands I've provided in rr230786 to
migrate your out of tree tests. Do let me know if any of this doesn't
cover your cases & we can iterate on a more general script/regexes to
help others with out of tree tests.
About 9 test cases couldn't be automatically migrated - half of those
were functions returning function pointers, where I just had to manually
delete the function argument types now that we didn't need an explicit
function type there. The other half were typedefs of function types used
in calls - just had to manually drop the * from those.
import fileinput
import sys
import re
pat = re.compile(r'((?:=|:|^|\s)call\s(?:[^@]*?))(\s*$|\s*(?:(?:\[\[[a-zA-Z0-9_]+\]\]|[@%](?:(")?[\\\?@a-zA-Z0-9_.]*?(?(3)"|)|{{.*}}))(?:\(|$)|undef|inttoptr|bitcast|null|asm).*$)')
addrspace_end = re.compile(r"addrspace\(\d+\)\s*\*$")
func_end = re.compile("(?:void.*|\)\s*)\*$")
def conv(match, line):
if not match or re.search(addrspace_end, match.group(1)) or not re.search(func_end, match.group(1)):
return line
return line[:match.start()] + match.group(1)[:match.group(1).rfind('*')].rstrip() + match.group(2) + line[match.end():]
for line in sys.stdin:
sys.stdout.write(conv(re.search(pat, line), line))
llvm-svn: 235145
2015-04-17 07:24:18 +08:00
|
|
|
tail call void (...) @snork(i8* getelementptr inbounds ([52 x i8], [52 x i8]* @global1, i64 0, i64 0), i32 2071) nounwind
|
2015-03-14 02:20:45 +08:00
|
|
|
%tmp70 = load i32, i32* getelementptr inbounds (%struct.snork, %struct.snork* @global, i64 0, i32 2), align 4
|
2012-07-19 02:07:52 +08:00
|
|
|
unreachable
|
|
|
|
|
|
|
|
|
|
bb71: ; preds = %bb68
|
2015-03-14 02:20:45 +08:00
|
|
|
%tmp72 = load i32, i32* getelementptr inbounds (%struct.snork, %struct.snork* @global, i64 0, i32 4), align 4
|
2012-07-19 02:07:52 +08:00
|
|
|
%tmp73 = icmp eq i32 undef, 0
|
|
|
|
|
br i1 %tmp73, label %bb247, label %bb74
|
|
|
|
|
|
|
|
|
|
bb74: ; preds = %bb71, %bb68
|
|
|
|
|
%tmp75 = trunc i64 %tmp65 to i32
|
|
|
|
|
br label %bb76
|
|
|
|
|
|
|
|
|
|
bb76: ; preds = %bb82, %bb74
|
|
|
|
|
%tmp77 = phi i64 [ %tmp84, %bb82 ], [ 0, %bb74 ]
|
|
|
|
|
%tmp78 = phi i32 [ %tmp85, %bb82 ], [ 0, %bb74 ]
|
|
|
|
|
%tmp79 = trunc i64 %tmp77 to i32
|
|
|
|
|
%tmp80 = icmp slt i32 %tmp79, %tmp75
|
|
|
|
|
br i1 %tmp80, label %bb81, label %bb87
|
|
|
|
|
|
|
|
|
|
bb81: ; preds = %bb76
|
|
|
|
|
br i1 false, label %bb86, label %bb82
|
|
|
|
|
|
|
|
|
|
bb82: ; preds = %bb81
|
|
|
|
|
%tmp83 = icmp eq i8 0, 0
|
|
|
|
|
%tmp84 = add i64 %tmp77, 1
|
|
|
|
|
%tmp85 = add nsw i32 %tmp78, 1
|
|
|
|
|
br i1 %tmp83, label %bb87, label %bb76
|
|
|
|
|
|
|
|
|
|
bb86: ; preds = %bb81
|
|
|
|
|
br i1 undef, label %bb88, label %bb87
|
|
|
|
|
|
|
|
|
|
bb87: ; preds = %bb86, %bb82, %bb76
|
|
|
|
|
unreachable
|
|
|
|
|
|
|
|
|
|
bb88: ; preds = %bb86
|
|
|
|
|
%tmp89 = add nsw i32 %tmp78, 1
|
|
|
|
|
%tmp90 = sext i32 %tmp89 to i64
|
|
|
|
|
%tmp91 = add i64 %tmp90, %tmp64
|
|
|
|
|
%tmp92 = sub i64 %tmp65, %tmp90
|
|
|
|
|
br i1 false, label %bb93, label %bb94
|
|
|
|
|
|
|
|
|
|
bb93: ; preds = %bb88
|
|
|
|
|
unreachable
|
|
|
|
|
|
|
|
|
|
bb94: ; preds = %bb88
|
|
|
|
|
%tmp95 = trunc i64 %tmp92 to i32
|
|
|
|
|
br label %bb96
|
|
|
|
|
|
|
|
|
|
bb96: ; preds = %bb102, %bb94
|
|
|
|
|
%tmp97 = phi i64 [ %tmp103, %bb102 ], [ 0, %bb94 ]
|
|
|
|
|
%tmp98 = phi i32 [ %tmp104, %bb102 ], [ 0, %bb94 ]
|
|
|
|
|
%tmp99 = trunc i64 %tmp97 to i32
|
|
|
|
|
%tmp100 = icmp slt i32 %tmp99, %tmp95
|
|
|
|
|
br i1 %tmp100, label %bb101, label %bb106
|
|
|
|
|
|
|
|
|
|
bb101: ; preds = %bb96
|
|
|
|
|
br i1 undef, label %bb105, label %bb102
|
|
|
|
|
|
|
|
|
|
bb102: ; preds = %bb101
|
|
|
|
|
%tmp103 = add i64 %tmp97, 1
|
|
|
|
|
%tmp104 = add nsw i32 %tmp98, 1
|
|
|
|
|
br i1 false, label %bb106, label %bb96
|
|
|
|
|
|
|
|
|
|
bb105: ; preds = %bb101
|
|
|
|
|
br i1 undef, label %bb107, label %bb106
|
|
|
|
|
|
|
|
|
|
bb106: ; preds = %bb105, %bb102, %bb96
|
|
|
|
|
br label %bb247
|
|
|
|
|
|
|
|
|
|
bb107: ; preds = %bb105
|
|
|
|
|
%tmp108 = add nsw i32 %tmp98, 1
|
|
|
|
|
%tmp109 = sext i32 %tmp108 to i64
|
|
|
|
|
%tmp110 = add i64 %tmp109, %tmp91
|
|
|
|
|
%tmp111 = sub i64 %tmp92, %tmp109
|
|
|
|
|
br i1 false, label %bb112, label %bb113
|
|
|
|
|
|
|
|
|
|
bb112: ; preds = %bb107
|
|
|
|
|
unreachable
|
|
|
|
|
|
|
|
|
|
bb113: ; preds = %bb107
|
|
|
|
|
%tmp114 = trunc i64 %tmp111 to i32
|
|
|
|
|
br label %bb115
|
|
|
|
|
|
|
|
|
|
bb115: ; preds = %bb121, %bb113
|
|
|
|
|
%tmp116 = phi i64 [ %tmp122, %bb121 ], [ 0, %bb113 ]
|
|
|
|
|
%tmp117 = phi i32 [ %tmp123, %bb121 ], [ 0, %bb113 ]
|
|
|
|
|
%tmp118 = trunc i64 %tmp116 to i32
|
|
|
|
|
%tmp119 = icmp slt i32 %tmp118, %tmp114
|
|
|
|
|
br i1 %tmp119, label %bb120, label %bb125
|
|
|
|
|
|
|
|
|
|
bb120: ; preds = %bb115
|
|
|
|
|
br i1 undef, label %bb124, label %bb121
|
|
|
|
|
|
|
|
|
|
bb121: ; preds = %bb120
|
|
|
|
|
%tmp122 = add i64 %tmp116, 1
|
|
|
|
|
%tmp123 = add nsw i32 %tmp117, 1
|
|
|
|
|
br i1 false, label %bb125, label %bb115
|
|
|
|
|
|
|
|
|
|
bb124: ; preds = %bb120
|
|
|
|
|
br i1 false, label %bb126, label %bb125
|
|
|
|
|
|
|
|
|
|
bb125: ; preds = %bb124, %bb121, %bb115
|
|
|
|
|
unreachable
|
|
|
|
|
|
|
|
|
|
bb126: ; preds = %bb124
|
|
|
|
|
%tmp127 = add nsw i32 %tmp117, 1
|
|
|
|
|
%tmp128 = sext i32 %tmp127 to i64
|
|
|
|
|
%tmp129 = add i64 %tmp128, %tmp110
|
|
|
|
|
%tmp130 = sub i64 %tmp111, %tmp128
|
|
|
|
|
tail call fastcc void @blarg()
|
|
|
|
|
br i1 false, label %bb132, label %bb131
|
|
|
|
|
|
|
|
|
|
bb131: ; preds = %bb126
|
|
|
|
|
unreachable
|
|
|
|
|
|
|
|
|
|
bb132: ; preds = %bb126
|
|
|
|
|
%tmp133 = trunc i64 %tmp130 to i32
|
|
|
|
|
br label %bb134
|
|
|
|
|
|
|
|
|
|
bb134: ; preds = %bb140, %bb132
|
|
|
|
|
%tmp135 = phi i64 [ %tmp141, %bb140 ], [ 0, %bb132 ]
|
|
|
|
|
%tmp136 = phi i32 [ %tmp142, %bb140 ], [ 0, %bb132 ]
|
|
|
|
|
%tmp137 = trunc i64 %tmp135 to i32
|
|
|
|
|
%tmp138 = icmp slt i32 %tmp137, %tmp133
|
|
|
|
|
br i1 %tmp138, label %bb139, label %bb144
|
|
|
|
|
|
|
|
|
|
bb139: ; preds = %bb134
|
|
|
|
|
br i1 false, label %bb143, label %bb140
|
|
|
|
|
|
|
|
|
|
bb140: ; preds = %bb139
|
|
|
|
|
%tmp141 = add i64 %tmp135, 1
|
|
|
|
|
%tmp142 = add nsw i32 %tmp136, 1
|
|
|
|
|
br i1 false, label %bb144, label %bb134
|
|
|
|
|
|
|
|
|
|
bb143: ; preds = %bb139
|
|
|
|
|
br i1 false, label %bb145, label %bb144
|
|
|
|
|
|
|
|
|
|
bb144: ; preds = %bb143, %bb140, %bb134
|
|
|
|
|
br label %bb247
|
|
|
|
|
|
|
|
|
|
bb145: ; preds = %bb143
|
|
|
|
|
%tmp146 = add nsw i32 %tmp136, 1
|
|
|
|
|
%tmp147 = sext i32 %tmp146 to i64
|
|
|
|
|
%tmp148 = add i64 %tmp147, %tmp129
|
|
|
|
|
%tmp149 = sub i64 %tmp130, %tmp147
|
|
|
|
|
switch i32 0, label %bb152 [
|
|
|
|
|
i32 0, label %bb150
|
|
|
|
|
i32 16, label %bb150
|
|
|
|
|
i32 32, label %bb150
|
|
|
|
|
i32 48, label %bb150
|
|
|
|
|
i32 64, label %bb150
|
|
|
|
|
i32 256, label %bb150
|
|
|
|
|
i32 4096, label %bb150
|
|
|
|
|
]
|
|
|
|
|
|
|
|
|
|
bb150: ; preds = %bb145, %bb145, %bb145, %bb145, %bb145, %bb145, %bb145
|
|
|
|
|
%tmp151 = trunc i64 %tmp149 to i32
|
|
|
|
|
br label %bb153
|
|
|
|
|
|
|
|
|
|
bb152: ; preds = %bb145
|
|
|
|
|
unreachable
|
|
|
|
|
|
|
|
|
|
bb153: ; preds = %bb160, %bb150
|
|
|
|
|
%tmp154 = phi i64 [ %tmp161, %bb160 ], [ 0, %bb150 ]
|
|
|
|
|
%tmp155 = phi i32 [ %tmp162, %bb160 ], [ 0, %bb150 ]
|
|
|
|
|
%tmp156 = trunc i64 %tmp154 to i32
|
|
|
|
|
%tmp157 = icmp slt i32 %tmp156, %tmp151
|
|
|
|
|
br i1 %tmp157, label %bb158, label %bb166
|
|
|
|
|
|
|
|
|
|
bb158: ; preds = %bb153
|
|
|
|
|
%tmp159 = add i64 %tmp154, %tmp148
|
|
|
|
|
br i1 false, label %bb163, label %bb160
|
|
|
|
|
|
|
|
|
|
bb160: ; preds = %bb158
|
|
|
|
|
%tmp161 = add i64 %tmp154, 1
|
|
|
|
|
%tmp162 = add nsw i32 %tmp155, 1
|
|
|
|
|
br i1 false, label %bb166, label %bb153
|
|
|
|
|
|
|
|
|
|
bb163: ; preds = %bb158
|
|
|
|
|
%tmp164 = add i32 %tmp155, -1
|
|
|
|
|
%tmp165 = icmp ult i32 %tmp164, 255
|
|
|
|
|
br i1 %tmp165, label %bb167, label %bb166
|
|
|
|
|
|
|
|
|
|
bb166: ; preds = %bb163, %bb160, %bb153
|
|
|
|
|
unreachable
|
|
|
|
|
|
|
|
|
|
bb167: ; preds = %bb163
|
|
|
|
|
%tmp168 = add nsw i32 %tmp155, 1
|
|
|
|
|
%tmp169 = sext i32 %tmp168 to i64
|
|
|
|
|
%tmp170 = add i64 %tmp169, %tmp148
|
|
|
|
|
%tmp171 = sub i64 %tmp149, %tmp169
|
|
|
|
|
br i1 false, label %bb173, label %bb172
|
|
|
|
|
|
|
|
|
|
bb172: ; preds = %bb167
|
|
|
|
|
unreachable
|
|
|
|
|
|
|
|
|
|
bb173: ; preds = %bb167
|
|
|
|
|
%tmp174 = trunc i64 %tmp171 to i32
|
|
|
|
|
br label %bb175
|
|
|
|
|
|
|
|
|
|
bb175: ; preds = %bb181, %bb173
|
|
|
|
|
%tmp176 = phi i64 [ %tmp183, %bb181 ], [ 0, %bb173 ]
|
|
|
|
|
%tmp177 = phi i32 [ %tmp184, %bb181 ], [ 0, %bb173 ]
|
|
|
|
|
%tmp178 = trunc i64 %tmp176 to i32
|
|
|
|
|
%tmp179 = icmp slt i32 %tmp178, %tmp174
|
|
|
|
|
br i1 %tmp179, label %bb180, label %bb186
|
|
|
|
|
|
|
|
|
|
bb180: ; preds = %bb175
|
|
|
|
|
br i1 false, label %bb185, label %bb181
|
|
|
|
|
|
|
|
|
|
bb181: ; preds = %bb180
|
|
|
|
|
%tmp182 = icmp eq i8 0, 0
|
|
|
|
|
%tmp183 = add i64 %tmp176, 1
|
|
|
|
|
%tmp184 = add nsw i32 %tmp177, 1
|
|
|
|
|
br i1 %tmp182, label %bb186, label %bb175
|
|
|
|
|
|
|
|
|
|
bb185: ; preds = %bb180
|
|
|
|
|
br i1 false, label %bb187, label %bb186
|
|
|
|
|
|
|
|
|
|
bb186: ; preds = %bb185, %bb181, %bb175
|
|
|
|
|
unreachable
|
|
|
|
|
|
|
|
|
|
bb187: ; preds = %bb185
|
|
|
|
|
%tmp188 = add nsw i32 %tmp177, 1
|
|
|
|
|
%tmp189 = sext i32 %tmp188 to i64
|
|
|
|
|
%tmp190 = sub i64 %tmp171, %tmp189
|
|
|
|
|
br i1 false, label %bb192, label %bb191
|
|
|
|
|
|
|
|
|
|
bb191: ; preds = %bb187
|
|
|
|
|
unreachable
|
|
|
|
|
|
|
|
|
|
bb192: ; preds = %bb187
|
|
|
|
|
%tmp193 = trunc i64 %tmp190 to i32
|
|
|
|
|
br label %bb194
|
|
|
|
|
|
|
|
|
|
bb194: ; preds = %bb200, %bb192
|
|
|
|
|
%tmp195 = phi i64 [ %tmp201, %bb200 ], [ 0, %bb192 ]
|
|
|
|
|
%tmp196 = phi i32 [ %tmp202, %bb200 ], [ 0, %bb192 ]
|
|
|
|
|
%tmp197 = trunc i64 %tmp195 to i32
|
|
|
|
|
%tmp198 = icmp slt i32 %tmp197, %tmp193
|
|
|
|
|
br i1 %tmp198, label %bb199, label %bb204
|
|
|
|
|
|
|
|
|
|
bb199: ; preds = %bb194
|
|
|
|
|
br i1 false, label %bb203, label %bb200
|
|
|
|
|
|
|
|
|
|
bb200: ; preds = %bb199
|
|
|
|
|
%tmp201 = add i64 %tmp195, 1
|
|
|
|
|
%tmp202 = add nsw i32 %tmp196, 1
|
|
|
|
|
br i1 false, label %bb204, label %bb194
|
|
|
|
|
|
|
|
|
|
bb203: ; preds = %bb199
|
|
|
|
|
br i1 undef, label %bb205, label %bb204
|
|
|
|
|
|
|
|
|
|
bb204: ; preds = %bb203, %bb200, %bb194
|
|
|
|
|
unreachable
|
|
|
|
|
|
|
|
|
|
bb205: ; preds = %bb203
|
|
|
|
|
%tmp206 = add nsw i32 %tmp196, 1
|
|
|
|
|
%tmp207 = sext i32 %tmp206 to i64
|
|
|
|
|
%tmp208 = add i64 %tmp207, 0
|
|
|
|
|
%tmp209 = sub i64 %tmp190, %tmp207
|
|
|
|
|
br i1 %tmp13, label %bb210, label %bb211
|
|
|
|
|
|
|
|
|
|
bb210: ; preds = %bb205
|
|
|
|
|
unreachable
|
|
|
|
|
|
|
|
|
|
bb211: ; preds = %bb205
|
|
|
|
|
%tmp212 = trunc i64 %tmp209 to i32
|
|
|
|
|
%tmp213 = icmp slt i32 0, %tmp212
|
|
|
|
|
br i1 false, label %bb215, label %bb214
|
|
|
|
|
|
|
|
|
|
bb214: ; preds = %bb211
|
|
|
|
|
unreachable
|
|
|
|
|
|
|
|
|
|
bb215: ; preds = %bb211
|
|
|
|
|
%tmp216 = add i64 undef, %tmp208
|
|
|
|
|
%tmp217 = sub i64 %tmp209, undef
|
|
|
|
|
br i1 false, label %bb218, label %bb219
|
|
|
|
|
|
|
|
|
|
bb218: ; preds = %bb215
|
|
|
|
|
br label %bb219
|
|
|
|
|
|
|
|
|
|
bb219: ; preds = %bb218, %bb215
|
|
|
|
|
%tmp220 = trunc i64 %tmp217 to i32
|
|
|
|
|
br label %bb221
|
|
|
|
|
|
|
|
|
|
bb221: ; preds = %bb230, %bb219
|
|
|
|
|
%tmp222 = phi i64 [ %tmp231, %bb230 ], [ 0, %bb219 ]
|
|
|
|
|
%tmp223 = phi i32 [ %tmp232, %bb230 ], [ 0, %bb219 ]
|
|
|
|
|
%tmp224 = trunc i64 %tmp222 to i32
|
|
|
|
|
%tmp225 = icmp slt i32 %tmp224, %tmp220
|
|
|
|
|
br i1 %tmp225, label %bb226, label %bb234
|
|
|
|
|
|
|
|
|
|
bb226: ; preds = %bb221
|
|
|
|
|
%tmp227 = add i64 %tmp222, %tmp216
|
[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
|
|
|
%tmp228 = getelementptr inbounds %struct.jim, %struct.jim* @global3, i64 0, i32 3, i64 %tmp227
|
2015-02-28 05:17:42 +08:00
|
|
|
%tmp229 = load i8, i8* %tmp228, align 1
|
2012-07-19 02:07:52 +08:00
|
|
|
br i1 false, label %bb233, label %bb230
|
|
|
|
|
|
|
|
|
|
bb230: ; preds = %bb226
|
|
|
|
|
%tmp231 = add i64 %tmp222, 1
|
|
|
|
|
%tmp232 = add nsw i32 %tmp223, 1
|
|
|
|
|
br i1 undef, label %bb234, label %bb221
|
|
|
|
|
|
|
|
|
|
bb233: ; preds = %bb226
|
|
|
|
|
br i1 undef, label %bb235, label %bb234
|
|
|
|
|
|
|
|
|
|
bb234: ; preds = %bb233, %bb230, %bb221
|
|
|
|
|
br label %bb247
|
|
|
|
|
|
|
|
|
|
bb235: ; preds = %bb233
|
|
|
|
|
%tmp236 = add nsw i32 %tmp223, 1
|
|
|
|
|
%tmp237 = sext i32 %tmp236 to i64
|
|
|
|
|
%tmp238 = sub i64 %tmp217, %tmp237
|
|
|
|
|
br i1 %tmp66, label %bb239, label %bb240
|
|
|
|
|
|
|
|
|
|
bb239: ; preds = %bb235
|
|
|
|
|
unreachable
|
|
|
|
|
|
|
|
|
|
bb240: ; preds = %bb235
|
|
|
|
|
switch i32 0, label %bb244 [
|
|
|
|
|
i32 0, label %bb241
|
|
|
|
|
i32 1, label %bb241
|
|
|
|
|
i32 4, label %bb241
|
|
|
|
|
i32 6, label %bb241
|
|
|
|
|
i32 9, label %bb241
|
|
|
|
|
]
|
|
|
|
|
|
|
|
|
|
bb241: ; preds = %bb240, %bb240, %bb240, %bb240, %bb240
|
|
|
|
|
%tmp242 = trunc i64 %tmp238 to i32
|
|
|
|
|
%tmp243 = icmp slt i32 0, %tmp242
|
|
|
|
|
br i1 false, label %bb246, label %bb245
|
|
|
|
|
|
|
|
|
|
bb244: ; preds = %bb240
|
|
|
|
|
unreachable
|
|
|
|
|
|
|
|
|
|
bb245: ; preds = %bb241
|
|
|
|
|
unreachable
|
|
|
|
|
|
|
|
|
|
bb246: ; preds = %bb241
|
|
|
|
|
unreachable
|
|
|
|
|
|
|
|
|
|
bb247: ; preds = %bb234, %bb144, %bb106, %bb71, %bb58
|
|
|
|
|
ret void
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
declare void @zot(i8* nocapture, i8* nocapture, i64, i32, i1) nounwind
|