Re: bug around inline asm, or expected?

["Mohamed Bamakhrama" <bamakhrama at gmail dot com>]

On 7/20/07, Luke <luke@neoclaw.net> wrote:
> I hit an odd problem recently while working with some inline assembly
> (x86).  I'm not sure if it's a bug, or if it is expected behavior.  I
> managed to repro it in a small compilable file:
>
> #include <iostream>
>
> //Compares expectedValue with the value at pInt, and if they are the
> same, sets pInt to newValue and returns true with outResultValue set to
> newValue.  If they are different then pInt is unaffected, and returns
> false with outResultValue set to the value that was found at pInt.
> inline bool AtomicCompareExchange(volatile unsigned int *pInt, unsigned
> int expectedValue, unsigned int newValue, volatile unsigned int
> &outResultValue)
> {
>    unsigned int changed=0;
>    unsigned int rval;
>    asm(
>            "lock cmpxchgl %4, (%2);"
>            "jnz AtomCmpExch_Diff%=;"
>            "movl %4, %%eax;"
>            "incl %1;"
>            "AtomCmpExch_Diff%=:;"
>            "movl %%eax, %0;"
>        :"=g"(rval), "=g"(changed)
>        :"r"(pInt), "a"(expectedValue), "r"(newValue)
>        :"memory", "cc");
>
>    outResultValue=rval;
>    return changed!=0;
> }
>
> int main()
> {
>    unsigned int value=0x100;
>    unsigned int ignored;
>    bool changed=AtomicCompareExchange(&value, 0x200, 0x300, ignored);
>      if (changed) std::cout<<"Returned true, but should have returned
> false.\n";
>    else std::cout<<"Returned false as expected.\n";
>      return 0;
> }
>
> With optimization off it works fine.  Running g++ with -O1 or -O2
> though, and it hits a problem: that the inline function always returns
> true, when it should in this case return false.  I actually did find a
> "solution" to it.  Declaring the "unsigned int changed=0;" as volatile
> prevents the problem from happening.  However, it seems a bit odd that
> that's necessary, seeing as it's given as an output parameter in the asm
> declaration.
>
> Here's the output of g++ -v:
> Using built-in specs.
> Target: i486-linux-gnu
> Configured with: ../src/configure -v
> --enable-languages=c,c++,fortran,objc,obj-c++,treelang --prefix=/usr
> --enable-shared --with-system-zlib --libexecdir=/usr/lib
> --without-included-gettext --enable-threads=posix --enable-nls
> --with-gxx-include-dir=/usr/include/c++/4.1.3 --program-suffix=-4.1
> --enable-__cxa_atexit --enable-clocale=gnu --enable-libstdcxx-debug
> --enable-mpfr --with-tune=i686 --enable-checking=release i486-linux-gnu
> Thread model: posix
> gcc version 4.1.3 20070629 (prerelease) (Debian 4.1.2-13)

You can try the following:
1) add "volatile" after the "asm" keyword so that GCC doesn't try to
move the inline asm block around during the optimization.
2) add ".set noreorder". This should instruct the assembler to emit
the instruction "as-is" without doing any reordering optimizations.

--
Mohamed