Re: [Qemu-devel] "Enable _FORTIFY_SOURCE=2" result in building failure for qemu-img.c

[Markus Armbruster]

Jamie Lokier <address@hidden> writes:

> Sheng Yang wrote:
>>     printf("qemu-img version " QEMU_VERSION ", Copyright (c) 2004-2008 
>> Fabrice 
>> Bellard\n"
>>            "usage: qemu-img command [command options]\n"
>>            "QEMU disk image utility\n"
>>            "\n"
>>            "Command syntax:\n"
>> #define DEF(option, callback, arg_string)        \
>>            "  " arg_string "\n"
>> #include "qemu-img-cmds.h"
>> #undef DEF
>> #undef GEN_DOCS
>> ....
>> 
>> Seems gcc take "printf" as a marco. I added a "#undef printf" before the 
>> line, 
>> then it works...
>> 
>> So any clue on what's happened and how to fix?
>
> You can't have preprocessor directives inside the arguments of a macro
> call.  Yes it's occasionally annoying like this.
>
> You can prevent the macro call without #undef by writing:
>
>     (printf)("qemu-img version " ...etc)

Yup.

> I'm not sure if Glibc is compliant with ISO C by
> making printf into a macro that takes arguments.

It is.  7.1.4 "Use of library functions" says "Any function declared in
a header may be additionally implemented as a function-like macro
defined in the header".

> Certain functions such as putchar() are specified as being allowed to
> be macros, which implies the other standard functions aren't.

You mean putc(), don't you?  7.19.7.8:

    The putc function is equivalent to fputc, except that if it is
    implemented as a macro, it may evaluate stream more than once, so
    that argument should never be an expression with side effects.

The general license-to-macro in 7.1.4 wouldn't allow a macro that
evaluates the argument more than once, so we need a special license
here.