r253595 - Update clang tests to accomendate discriminator changes.

Steven Wu via llvm-commits llvm-commits at lists.llvm.org
Fri Nov 20 12:20:12 PST 2015 

> On Nov 20, 2015, at 12:11 PM, Duncan P. N. Exon Smith <dexonsmith at apple.com> wrote:
> 
>> 
>> On 2015-Nov-19, at 15:52, David Blaikie via llvm-commits <llvm-commits at lists.llvm.org> wrote:
>> 
>> 
>> 
>> On Thu, Nov 19, 2015 at 2:38 PM, Dehao Chen <dehao at google.com> wrote:
>> I'm not quite familiar with the pass manager, but the stack trace for
>> the AddDiscriminators call from the frontend run (with
>> -disable-llvm-optzns) is:
>> 
>> *looks around*
>> 
>> Oh, that's lame:
>> 
>>  // Handle disabling of LLVM optimization, where we want to preserve the
>>  // internal module before any optimization.
>>  if (CodeGenOpts.DisableLLVMOpts) {
>>    OptLevel = 0;
>>    Inlining = CodeGenOpts.NoInlining;
>>  }
>> 
>> It doesn't /actually/ disable LLVM optimizations... it just turns off the inliner and switches to -O0. *eyeroll*
>> 
>> I wonder what the history is here... *goes spelunking*
>> 
>> Yep, seems to have been there since it was added... in r72742
>> 
>> We really should fix that. It really shouldn't run the LLVM optimization pipeline at all. (you know, because that's what it says on the tin)
>> 
>> Hmm, OK, maybe it does make sense-ish (disable-llvm-optzns is for disabling /optimizations/? But that's still arguably... (always inlining isn't strictly an optimization))
>> 
>> But there's -disable-llvm-passes. That seems to do the trick. Could you try that? 
> 
> I think -disable-llvm-optzns at one point disabled all the passes,
> but then some of the sanitizers were skipping it.  When we included
> an IR dump in -save-temps, we noticed it wasn't working correctly
> and Steven Wu added the less muddy -disable-llvm-passes in r242565.
> 
> I wouldn't have a problem with -disable-llvm-optzns skipping all
> passes (again?), but I figured the sanitizers avoided it for a
> reason?
Most of the instrumentation passes ignore -disable-llvm-optzns. I don't know the exact reason in the beginning but my guess is that it will allow you to compile code with both -disable-llvm-optzns and sanitizer. That is why I added a new option -disable-llvm-passes instead of replace -disable-llvm-optzns. Other than that, -disable-llvm-optzns should only run immutable passes which is pretty close to no optimization.

Steven  

> 
>> 
>> - David
>> 
>> 
>> #0  (anonymous namespace)::AddDiscriminators::runOnFunction
>> (this=0x8924170, F=...) at
>> /ssd/llvm_trunk/lib/Transforms/Utils/AddDiscriminators.cpp:177
>> #1  0x0000000002769946 in llvm::FPPassManager::runOnFunction
>> (this=0x8921380, F=...) at
>> /ssd/llvm_trunk/lib/IR/LegacyPassManager.cpp:1521
>> #2  0x00000000027696d8 in llvm::legacy::FunctionPassManagerImpl::run
>> (this=0x891ee90, F=...) at
>> /ssd/llvm_trunk/lib/IR/LegacyPassManager.cpp:1470
>> #3  0x00000000027692aa in llvm::legacy::FunctionPassManager::run
>> (this=0x88e6a60, F=...) at
>> /ssd/llvm_trunk/lib/IR/LegacyPassManager.cpp:1394
>> #4  0x0000000002dcd9c2 in (anonymous
>> namespace)::EmitAssemblyHelper::EmitAssembly (this=0x7fffffffbd80,
>> Action=clang::Backend_EmitLL, OS=0x88b04a0) at
>> /ssd/llvm_trunk/tools/clang/lib/CodeGen/
>> BackendUtil.cpp:643
>> #5  0x0000000002dcdb7a in clang::EmitBackendOutput (Diags=...,
>> CGOpts=..., TOpts=..., LOpts=..., TDesc=..., M=0x88b8810,
>> Action=clang::Backend_EmitLL, OS=0x88b04a0) at /ssd/llvm_trunk/tools/c
>> lang/lib/CodeGen/BackendUtil.cpp:666
>> #6  0x00000000035f64b7 in
>> clang::BackendConsumer::HandleTranslationUnit (this=0x88b83f0, C=...)
>> at /ssd/llvm_trunk/tools/clang/lib/CodeGen/CodeGenAction.cpp:191
>> #7  0x0000000003e9aa19 in clang::ParseAST (S=..., PrintStats=false,
>> SkipFunctionBodies=false) at
>> /ssd/llvm_trunk/tools/clang/lib/Parse/ParseAST.cpp:168
>> #8  0x000000000325deec in clang::ASTFrontendAction::ExecuteAction
>> (this=0x8887530) at
>> /ssd/llvm_trunk/tools/clang/lib/Frontend/FrontendAction.cpp:539
>> #9  0x00000000035f5572 in clang::CodeGenAction::ExecuteAction
>> (this=0x8887530) at
>> /ssd/llvm_trunk/tools/clang/lib/CodeGen/CodeGenAction.cpp:797
>> #10 0x000000000325d9b1 in clang::FrontendAction::Execute
>> (this=0x8887530) at
>> /ssd/llvm_trunk/tools/clang/lib/Frontend/FrontendAction.cpp:439
>> #11 0x00000000032176e6 in clang::CompilerInstance::ExecuteAction
>> (this=0x8884410, Act=...) at
>> /ssd/llvm_trunk/tools/clang/lib/Frontend/CompilerInstance.cpp:841
>> #12 0x000000000335c64a in clang::ExecuteCompilerInvocation
>> (Clang=0x8884410) at
>> /ssd/llvm_trunk/tools/clang/lib/FrontendTool/ExecuteCompilerInvocation.cpp:222
>> #13 0x00000000014847d1 in cc1_main (Argv=..., Argv0=0x7fffffffdffd
>> "/ssd/llvm_trunk/build.debug/bin/clang", MainAddr=0x1478abe
>> <GetExecutablePath(char const*, bool)>) at /ssd/llvm_trunk/tools
>> /clang/tools/driver/cc1_main.cpp:116
>> #14 0x000000000147af89 in ExecuteCC1Tool (argv=..., Tool=...) at
>> /ssd/llvm_trunk/tools/clang/tools/driver/driver.cpp:301
>> #15 0x000000000147b938 in main (argc_=12, argv_=0x7fffffffdbf8) at
>> /ssd/llvm_trunk/tools/clang/tools/driver/driver.cpp:366
>> 
>> On Thu, Nov 19, 2015 at 1:16 PM, David Blaikie <dblaikie at gmail.com> wrote:
>>> 
>>> On Thu, Nov 19, 2015 at 1:09 PM, Dehao Chen <dehao at google.com> wrote:
>>>> 
>>>> disable-llvm-optzns does not help:
>>>> 
>>>> /ssd/llvm_trunk/build.gccopt/./bin/clang -cc1 -internal-isystem
>>>> /ssd/llvm_trunk/build.gccopt/bin/../lib/clang/3.8.0/include
>>>> -nostdsysteminc -dwarf-version=4 -debug-info-kind=line-tables-only
>>>> -disable-llvm-optzns  -emit-llvm <
>>>> /ssd/llvm_trunk/tools/clang/test/CodeGen/debug-info-scope.c
>>>> 
>>>> This still emits discriminators.
>>> 
>>> 
>>> Do you know where the discriminators are coming from in this case? Does the
>>> frontend add some of them (in which case the test case shouldn't need
>>> updating for your recent backend change still, hopefully)? Is there some
>>> other codepath that is running the AddDiscriminator transformation?
>>> 
>>>> 
>>>> 
>>>> On Thu, Nov 19, 2015 at 12:46 PM, David Blaikie <dblaikie at gmail.com>
>>>> wrote:
>>>>> 
>>>>> 
>>>>> On Thu, Nov 19, 2015 at 12:42 PM, Dehao Chen <dehao at google.com> wrote:
>>>>>> 
>>>>>> Yes, as discriminator is a backend pass, it should *not* affect
>>>>>> frontend
>>>>>> tests.
>>>>> 
>>>>> 
>>>>> Could you update these tests to not involve discriminators by passing
>>>>> -disable-llvm-optzns instead?
>>>>> 
>>>>>> 
>>>>>> 
>>>>>> On Thu, Nov 19, 2015 at 12:36 PM, David Blaikie <dblaikie at gmail.com>
>>>>>> wrote:
>>>>>>> 
>>>>>>> 
>>>>>>> On Thu, Nov 19, 2015 at 11:53 AM, Dehao Chen via cfe-commits
>>>>>>> <cfe-commits at lists.llvm.org> wrote:
>>>>>>>> 
>>>>>>>> Author: dehao
>>>>>>>> Date: Thu Nov 19 13:53:16 2015
>>>>>>>> New Revision: 253595
>>>>>>>> 
>>>>>>>> URL: http://llvm.org/viewvc/llvm-project?rev=253595&view=rev
>>>>>>>> Log:
>>>>>>>> Update clang tests to accomendate discriminator changes.
>>>>>>>> 
>>>>>>>> Summary: The discriminator change in http://reviews.llvm.org/D14738
>>>>>>>> will
>>>>>>>> fail these clang tests. Update the test to accomendate the
>>>>>>>> discriminator
>>>>>>>> change.
>>>>>>> 
>>>>>>> 
>>>>>>> Do these tests need to be testing discriminators in the frontend?
>>>>>>> Could
>>>>>>> we
>>>>>>> just -mllvm -disable-llvm-optzns on these tests? (can the
>>>>>>> functionality
>>>>>>> be
>>>>>>> wholely tested in LLVM, I assume so if it's phrased as an LLVM
>>>>>>> transformation pass?)
>>>>>>> 
>>>>>>>> 
>>>>>>>> 
>>>>>>>> Reviewers: dblaikie, davidxl, dnovillo
>>>>>>>> 
>>>>>>>> Differential Revision: http://reviews.llvm.org/D14836
>>>>>>>> 
>>>>>>>> Modified:
>>>>>>>>    cfe/trunk/test/CodeGen/debug-info-scope.c
>>>>>>>>    cfe/trunk/test/CodeGenObjC/arc-linetable.m
>>>>>>>> 
>>>>>>>> Modified: cfe/trunk/test/CodeGen/debug-info-scope.c
>>>>>>>> URL:
>>>>>>>> 
>>>>>>>> 
>>>>>>>> http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGen/debug-info-scope.c?rev=253595&r1=253594&r2=253595&view=diff
>>>>>>>> 
>>>>>>>> 
>>>>>>>> 
>>>>>>>> ==============================================================================
>>>>>>>> --- cfe/trunk/test/CodeGen/debug-info-scope.c (original)
>>>>>>>> +++ cfe/trunk/test/CodeGen/debug-info-scope.c Thu Nov 19 13:53:16
>>>>>>>> 2015
>>>>>>>> @@ -10,6 +10,7 @@ int main() {
>>>>>>>> 
>>>>>>>> // GMLT-NOT: !DILexicalBlock
>>>>>>>> // GMLT: !DILexicalBlockFile({{.*}}, discriminator: 1)
>>>>>>>> +// GMLT: !DILexicalBlockFile({{.*}}, discriminator: 2)
>>>>>>>> // Make sure we don't have any more lexical blocks because we don't
>>>>>>>> need
>>>>>>>> them in
>>>>>>>> // -gmlt.
>>>>>>>> // GMLT-NOT: !DILexicalBlock
>>>>>>>> 
>>>>>>>> Modified: cfe/trunk/test/CodeGenObjC/arc-linetable.m
>>>>>>>> URL:
>>>>>>>> 
>>>>>>>> 
>>>>>>>> http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGenObjC/arc-linetable.m?rev=253595&r1=253594&r2=253595&view=diff
>>>>>>>> 
>>>>>>>> 
>>>>>>>> 
>>>>>>>> ==============================================================================
>>>>>>>> --- cfe/trunk/test/CodeGenObjC/arc-linetable.m (original)
>>>>>>>> +++ cfe/trunk/test/CodeGenObjC/arc-linetable.m Thu Nov 19 13:53:16
>>>>>>>> 2015
>>>>>>>> @@ -34,9 +34,9 @@
>>>>>>>> 
>>>>>>>> // CHECK: define {{.*}}testCleanupVoid
>>>>>>>> // CHECK: icmp ne {{.*}}!dbg ![[SKIP1:[0-9]+]]
>>>>>>>> -// CHECK: store i32 0, i32* {{.*}}, !dbg ![[RET8:[0-9]+]]
>>>>>>>> +// CHECK: store i32 0, i32* {{.*}}, !dbg ![[STORE8:[0-9]+]]
>>>>>>>> // CHECK: @objc_storeStrong{{.*}}, !dbg ![[ARC8:[0-9]+]]
>>>>>>>> -// CHECK: ret {{.*}} !dbg ![[RET8]]
>>>>>>>> +// CHECK: ret {{.*}} !dbg ![[RET8:[0-9]+]]
>>>>>>>> 
>>>>>>>> typedef signed char BOOL;
>>>>>>>> 
>>>>>>>> @@ -112,8 +112,9 @@ typedef signed char BOOL;
>>>>>>>>       [delegate testVoid :s];
>>>>>>>>     }
>>>>>>>>   }
>>>>>>>> -  // CHECK: ![[RET8]] = !DILocation(line: [[@LINE+2]], scope:
>>>>>>>> -  // CHECK: ![[ARC8]] = !DILocation(line: [[@LINE+1]], scope:
>>>>>>>> +  // CHECK: ![[STORE8]] = !DILocation(line: [[@LINE+3]], scope:
>>>>>>>> +  // CHECK: ![[ARC8]] = !DILocation(line: [[@LINE+2]], scope:
>>>>>>>> +  // CHECK: ![[RET8]] = !DILocation(line: [[@LINE+1]], scope:
>>>>>>>> }
>>>>>>>> 
>>>>>>>> 
>>>>>>>> 
>>>>>>>> 
>>>>>>>> _______________________________________________
>>>>>>>> cfe-commits mailing list
>>>>>>>> cfe-commits at lists.llvm.org
>>>>>>>> http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
>>>>>>> 
>>>>>>> 
>>>>> 
>>>>> 
>>> 
>>> 
>> 
>> _______________________________________________
>> llvm-commits mailing list
>> llvm-commits at lists.llvm.org
>> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-commits

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20151120/c0a2ee8d/attachment.html>

More information about the llvm-commits mailing list