Re: [PATCH] TRACING: Fix a copmile warning

[Arnaud Lacombe <lacombar at gmail dot com>]

Hi,

[adding gcc-help@ to the Cc: list]

On Mon, Jul 25, 2011 at 6:38 PM, Arnaud Lacombe <lacombar@gmail.com> wrote:
> Hi,
>
> On Mon, Jul 25, 2011 at 4:19 PM, Steven Rostedt <rostedt@goodmis.org> wrote:
>> On Mon, 2011-07-25 at 15:43 -0400, Arnaud Lacombe wrote:
>>
>>> Actually, we have a special uninitialized_var(x) macro to handle such
>>> false positive. From include/linux/compiler-gcc.h:
>>>
>>> /*
>>> ?* A trick to suppress uninitialized variable warning without generating any
>>> ?* code
>>> ?*/
>>> #define uninitialized_var(x) x = x
>>
>> I'm aware of that too, but I think that is inappropriate as well. As I
>> said, some versions of gcc report it, others don't. Seems that gcc 4.6.0
>> says this is an error where 4.5.1 does not (I just tried both).
>>
gcc will only emits the warning at -Os. It seems to me that the
resulting code clearly ends-up testing an uninitialized value, ie.
assuming the following test-case:

extern void *e(void);
extern void *f(void);
extern void g(void);

void fn(void)
{
        void *b, *a;

        a = e();
        if (a != 0)
                b = f();
        if (a != 0 && b != 0)
                g();
}

gcc 4.5.1 will generates the following x86-32 assembly:

% gcc -m32 -Wall -Os -c -S -o - kernel/trace/trace_printk.c
        .file   "trace_printk.c"
kernel/trace/trace_printk.c: In function 'fn':
kernel/trace/trace_printk.c:7:8: warning: 'b' may be used
uninitialized in this function
        .text
.globl fn
        .type   fn, @function
fn:
        pushl   %ebp
        movl    %esp, %ebp
        pushl   %esi
        pushl   %ebx
        call    e
        testl   %eax, %eax
        movl    %eax, %ebx
        je      .L2
        call    f
        movl    %eax, %esi
.L2:
        testl   %esi, %esi
        je      .L1
        testl   %ebx, %ebx
        je      .L1
        popl    %ebx
        popl    %esi
        popl    %ebp
        jmp     g
.L1:
        popl    %ebx
        popl    %esi
        popl    %ebp
        ret
        .size   fn, .-fn
        .ident  "GCC: (GNU) 4.5.1 20100924 (Red Hat 4.5.1-4)"
        .section        .note.GNU-stack,"",@progbits

It seems gcc transforms the conditional from:

if (a != NULL && b != NULL) ...

to

if (b != NULL && a != NULL) ...

In which case the warning is fully valid. I'm not sure what's the C
standard guarantee in term of conditional test order. gcc 4.7.0 has
the same behavior.

 - Arnaud