This is probably because we were "faking" long doubles as 64-bits, because we didn't have large integer support.
I just checked in some changes which should improve the situation some, this is continuing work on http://llvm.org/PR4295 There is still more work to be done, but if your test case still doesn't work, please attach the test case to that bugzilla and I will look at it. - Daniel On Fri, Jun 26, 2009 at 10:26 AM, Liviu Ciortea<liviu.ciortea at epfl.ch> wrote: > Hi, > > That's great, thanks for sharing the newest Uclibc! I was actually using > some old code in runtime/POSIX that used 'klee_make_symbolic_name', so I > updated to the latest versions of LLVM and KLEE. However, now I'm running > into some other problem when running coreutils, linking uclibc and using the > POSIX runtime. It seems to be an inconsistency between LLVM and llvm-gcc. > The problem is when trying to initialize a 80-bit constant supposedly on 64 > bits: > > "uint64_t klee::ConstantExpr::getZExtValue(unsigned int) const: Assertion > `getWidth() <= bits && "Value may be out of range!"' failed." > > > Have you experienced this? > > Thanks, > Liviu > > > Daniel Dunbar wrote: >> >> Hi, >> >> We renamed klee_make_symbolic_name to klee_make_symbolic, to shorten >> the function name and have the extra name argument always be present. >> >> If you would like to try our current version of uclibc, I have put it >> here: >> ?http://t1.minormatter.com/~ddunbar/klee-uclibc-0.01.tgz >> >> ?- Daniel >> >> On Tue, Jun 23, 2009 at 8:34 AM, Ciortea Liviu<liviu.ciortea at epfl.ch> >> wrote: >> >>> >>> Hi Daniel, >>> >>> I am trying to run coreutils, but I am getting some errors. I am using >>> the same version of uclibc as with the older version of KLEE, compiled with >>> the new llvm-gcc. Also, I am using the POSIX runtime, in the lines of: >>> >>> klee ... --libc=uclibc --posix-runtime --init-env ... >>> >>> I am getting the error "failed external call" for >>> klee_make_symbolic_name. I guess I am missing something, but I cannot >>> pinpoint exactly what. Any help would be greatly appreciated. >>> >>> Thanks, >>> Liviu >>> >>> >>> ________________________________________ >>> From: daniel.dunbar at gmail.com [daniel.dunbar at gmail.com] On Behalf Of >>> Daniel Dunbar [daniel at zuster.org] >>> Sent: Thursday, June 11, 2009 5:50 PM >>> To: Ciortea Liviu >>> Cc: Daniel Dunbar >>> Subject: Re: [klee-dev] Compile error when compiling KLEE with gcc 4.2.4 >>> >>> On Thursday, June 11, 2009, Liviu Ciortea <liviu.ciortea at epfl.ch> wrote: >>> >>>> >>>> Hi Daniel, >>>> >>>> Thanks for your reply. Indeed, this how I build it. I would still be >>>> interested in knowing what GCC version you are using. >>>> >>> >>> On my Mac I use gcc-4.2, and I think Cristi uses gcc-4.0. On my Linux >>> box I use gcc 4.3.3. >>> >>> -Daniel >>> >>> >>>> >>>> Cheers, >>>> Liviu >>>> >>>> >>>> Daniel Dunbar wrote: >>>> >>>> Hi Liviu, >>>> >>>> I moved the declaration out of the function, does this fix the build for >>>> you? >>>> >>>> ?- Daniel >>>> >>>> On Wed, Jun 10, 2009 at 8:19 AM, Ciortea Liviu<liviu.ciortea at epfl.ch> >>>> wrote: >>>> >>>> >>>> Hi guys, >>>> >>>> What version of GCC are you using when building? When compiling with GCC >>>> 4.2.4 I get the following compile error: >>>> >>>> Expr.cpp: In destructor ?klee::Array::~Array()?: >>>> Expr.cpp:359: error: expected unqualified-id before string constant >>>> Expr.cpp:363: error: ?::vc_DeleteExpr? has not been declared >>>> >>>> The problem here is related to the extern "C" declaration inside the >>>> destructor. >>>> >>>> Thanks, >>>> Liviu >>>> _______________________________________________ >>>> klee-dev mailing list >>>> klee-dev at keeda.stanford.edu >>>> http://keeda.Stanford.EDU/mailman/listinfo/klee-dev >>>> >>>> >>>> >>>> >>>> >>>> > >
