[cfe-dev] scan-build with -analyzer-ipa=dynamic options

[Anna Zaks]

Byoungyoung,

Thanks for uncovering the issue! We've fixed the default ipa setting for clang --analyze in r164543.

We'd prefer to keep analyzer-ipa option hidden from the users of scan-build as we consider it to be an internal option.

Does this work for you?

Cheers,
Anna.

On Sep 24, 2012, at 7:25 AM, Byoungyoung Lee <blee at gatech.edu> wrote:

> I've changed scan-build script to control IPA options as follows, and
> it works fine. It seems to me that scan-build was not *truly*
> supproting interprocedural analysis?
> 
> ===================================================================
> --- scan-build  (revision 163967)
> +++ scan-build  (working copy)
> @@ -1431,6 +1431,11 @@
>     push @AnalysesToRun, "-analyzer-disable-checker", shift @ARGV;
>     next;
>   }
> +  if ($arg eq "-analyzer-ipa") {
> +    shift @ARGV;
> +    push @AnalysesToRun, "-analyzer-ipa", shift @ARGV;
> +    next;
> +  }
>   if ($arg eq "-load-plugin") {
>     shift @ARGV;
>     push @PluginsToLoad, "-load", shift @ARGV;
> 
> 
> Thanks,
> Byoungyoung
> 
> On Mon, Sep 24, 2012 at 12:01 AM, Byoungyoung Lee <blee at gatech.edu> wrote:
>> By spending some time with scan-build and ccc-analyzer, I found that
>> the options provided to the analyzer are set by parsing the output
>> from "clang -### ...". As far as I know, the default IPA option is
>> dynamic-bifurcate, but clang shows -dynamic-ipa=inlining.
>> 
>> -----------------------------------------------------------------------
>> $ clang++ -### --analyze -x c++ testCode/ShapeTable.cpp
>> clang version 3.2 (trunk 163967)
>> Target: x86_64-unknown-linux-gnu
>> Thread model: posix
>> "/usr/local/bin/clang" "-cc1" "-triple" "x86_64-unknown-linux-gnu"
>> "-analyze" "-disable-free" "-main-file-name" "ShapeTable.cpp"
>> "-analyzer-store=region" "-analyzer-opt-\
>> analyze-nested-blocks" "-analyzer-eagerly-assume"
>> "-analyzer-ipa=inlining" "-analyzer-checker=core"
>> "-analyzer-checker=unix" "-analyzer-checker=deadcode"
>> "-analyzer-check\
>> er=security.insecureAPI.UncheckedReturn"
>> "-analyzer-checker=security.insecureAPI.getpw"
>> "-analyzer-checker=security.insecureAPI.gets"
>> "-analyzer-checker=security.insecure\
>> API.mktemp" "-analyzer-checker=security.insecureAPI.mkstemp"
>> "-analyzer-checker=security.insecureAPI.vfork" "-analyzer-output"
>> "plist" "-w" "-mrelocation-model" "static" \
>> "-mdisable-fp-elim" "-fmath-errno" "-masm-verbose"
>> "-mconstructor-aliases" "-munwind-tables" "-target-cpu" "x86-64"
>> "-target-linker-version" "2.22" "-momit-leaf-frame-poi\
>> nter" "-resource-dir" "/usr/local/bin/../lib/clang/3.2"
>> "-fmodule-cache-path" "/var/tmp/clang-module-cache"
>> "-internal-isystem" "/usr/lib/gcc/x86_64-linux-gnu/4.6/../../.\
>> ./../include/c++/4.6" "-internal-isystem"
>> "/usr/lib/gcc/x86_64-linux-gnu/4.6/../../../../include/c++/4.6/x86_64-linux-gnu"
>> "-internal-isystem" "/usr/lib/gcc/x86_64-linux-\
>> gnu/4.6/../../../../include/c++/4.6/backward" "-internal-isystem"
>> "/usr/local/include" "-internal-isystem"
>> "/usr/local/bin/../lib/clang/3.2/include" "-internal-externc-is\
>> ystem" "/usr/include/x86_64-linux-gnu" "-internal-externc-isystem"
>> "/include" "-internal-externc-isystem" "/usr/include"
>> "-fdeprecated-macro" "-fdebug-compilation-dir" "/\
>> home/super/llvm/llvm/tools/clang/examples/InputMixPointer"
>> "-ferror-limit" "19" "-fmessage-length" "171" "-mstackrealign"
>> "-fobjc-runtime=gcc" "-fcxx-exceptions" "-fexcep\
>> tions" "-fdiagnostics-show-option" "-o" "ShapeTable.plist" "-x" "c++"
>> "testCode/ShapeTable.cpp"
>> 
>> Thanks,
>> Byoungyoung
>> 
>> 
>> On Sun, Sep 23, 2012 at 5:14 PM, Byoungyoung Lee <blee at gatech.edu> wrote:
>>> Hello Clang,
>>> 
>>> I'm trying to run scan-build with my own checker, but it seems like
>>> scan-build does not support to control "-analyzer-ipa" options? I
>>> haven't checked the details of scan-build scripts, but -analyzer-ipa
>>> options are hardcoded somewhere and scan-build only uses
>>> "-analyzer-ipa=inlining" options. Is this because IPA analysis is too
>>> expensive to be used for scan-build, or simply it's not implemented
>>> yet?
>>> 
>>> Thanks,
>>> Byoungyoung
> _______________________________________________
> cfe-dev mailing list
> cfe-dev at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/cfe-dev