Re: [Qemu-devel] [PATCH 2/9] tcg: declare __jit_debug_descriptor to be static

[Jim Meyering <jim@meyering.net>]

Peter Maydell wrote:
> On 21 May 2012 21:10, Jim Meyering <jim@meyering.net> wrote:
>> Peter Maydell wrote:
>>> On 21 May 2012 20:51, Jim Meyering <jim@meyering.net> wrote:
>>>> From: Jim Meyering <meyering@redhat.com>
>>>>
>>>>
>>>> Signed-off-by: Jim Meyering <meyering@redhat.com>
>>>> ---
>>>>  tcg/tcg.c | 2 +-
>>>>  1 file changed, 1 insertion(+), 1 deletion(-)
>>>>
>>>> diff --git a/tcg/tcg.c b/tcg/tcg.c
>>>> index ab589c7..350fdad 100644
>>>> --- a/tcg/tcg.c
>>>> +++ b/tcg/tcg.c
>>>> @@ -2293,7 +2293,7 @@ void __jit_debug_register_code(void)
>>>>
>>>>  /* Must statically initialize the version, because GDB may check
>>>>    the version before we can set it.  */
>>>> -struct jit_descriptor __jit_debug_descriptor = { 1, 0, 0, 0 };
>>>> +static struct jit_descriptor __jit_debug_descriptor = { 1, 0, 0, 0 };
>>>>
>>>>  /* End GDB interface.  */
>>>
>>> Nak. This symbol is global so that gdb can find it by fishing around
>>> in the executable's symbol table.
>>
>> Thanks for the quick feedback.
>>
>> How does the scope of the symbol affect whether gdb can find it?
>
> If you mark it 'static' the compiler can throw it away or completely
> rearrange it if it's feeling clever enough, I think.
>
> Anyway, we're following a prescribed interface here:
> http://sourceware.org/gdb/onlinedocs/gdb/Declarations.html
>
> and I don't think we should deviate from it. As the comment says,
> "THE FOLLOWING MUST MATCH GDB DOCS.".
>
>> If declaring this variable "static" is not appropriate,
>> then the comment saying that static initialization is desired
>> should be changed.
>
> The comment means "statically initialize this variable rather than
> doing it dynamically in some function at startup".

Thanks.  I've clarified the comments and posted a V2.