[FFmpeg-devel] [PATCH][RFC] avutil/x86/asm: rename REG_SP to REG_sp

[Ganesh Ajjanagadde]

On Fri, Aug 21, 2015 at 11:27 PM, Ganesh Ajjanagadde <gajjanag at mit.edu> wrote:
> On Fri, Aug 21, 2015 at 9:07 PM, Michael Niedermayer
> <michael at niedermayer.cc> wrote:
>> On Fri, Aug 21, 2015 at 02:21:48PM -0400, Ganesh Ajjanagadde wrote:
>>> On Fri, Aug 21, 2015 at 2:20 PM, Ganesh Ajjanagadde
>>> <gajjanagadde at gmail.com> wrote:
>>> > On Fri, Aug 21, 2015 at 2:17 PM, Ganesh Ajjanagadde
>>> > <gajjanagadde at gmail.com> wrote:
>>> >>
>>> >> REG_SP is defined by Solaris system headers.
>>> >> This fixes a sea of warnings while building on Solaris:
>>> >> http://fate.ffmpeg.org/report.cgi?time=20150820233505&slot=x86-opensolaris-gcc4.3
>>> >>
>>> >> Signed-off-by: Ganesh Ajjanagadde <gajjanagadde at gmail.com>
>>> >> ---
>>> >>  libavutil/x86/asm.h                | 5 +++--
>>> >>  libpostproc/postprocess_template.c | 2 +-
>>> >>  2 files changed, 4 insertions(+), 3 deletions(-)
>>> >>
>>> >> diff --git a/libavutil/x86/asm.h b/libavutil/x86/asm.h
>>> >> index 616ad6c..109b65e 100644
>>> >> --- a/libavutil/x86/asm.h
>>> >> +++ b/libavutil/x86/asm.h
>>> >> @@ -38,7 +38,8 @@ typedef struct ymm_reg { uint64_t a, b, c, d; } ymm_reg;
>>> >>  #    define PTR_SIZE "8"
>>> >>  typedef int64_t x86_reg;
>>> >>
>>> >> -#    define REG_SP "rsp"
>>> >> +/* REG_SP is defined in Solaris sys headers, so use REG_sp */
>>> >> +#    define REG_sp "rsp"
>>> >>  #    define REG_BP "rbp"
>>> >>  #    define REGBP   rbp
>>> >>  #    define REGa    rax
>>> >> @@ -59,7 +60,7 @@ typedef int64_t x86_reg;
>>> >>  #    define PTR_SIZE "4"
>>> >>  typedef int32_t x86_reg;
>>> >>
>>> >> -#    define REG_SP "esp"
>>> >> +#    define REG_sp "esp"
>>> >>  #    define REG_BP "ebp"
>>> >>  #    define REGBP   ebp
>>> >>  #    define REGa    eax
>>> >> diff --git a/libpostproc/postprocess_template.c b/libpostproc/postprocess_template.c
>>> >> index 5694cc5..b01be58 100644
>>> >> --- a/libpostproc/postprocess_template.c
>>> >> +++ b/libpostproc/postprocess_template.c
>>> >> @@ -1317,7 +1317,7 @@ DERING_CORE((%0, %1, 8)    ,(%%REGd, %1, 4),%%mm2,%%mm4,%%mm0,%%mm3,%%mm5,%%mm1,
>>> >>          "1:                        \n\t"
>>> >>          : : "r" (src), "r" ((x86_reg)stride), "m" (c->pQPb), "m"(c->pQPb2), "q"(tmp)
>>> >>            NAMED_CONSTRAINTS_ADD(deringThreshold,b00,b02,b08)
>>> >> -        : "%"REG_a, "%"REG_d, "%"REG_SP
>>> >> +        : "%"REG_a, "%"REG_d, "%"REG_sp
>>> >>      );
>>> >>  #else // HAVE_7REGS && (TEMPLATE_PP_MMXEXT || TEMPLATE_PP_3DNOW)
>>> >>      int y;
>>> >> --
>>> >> 2.5.0
>>> >>
>>> >
>>> > @Michael: Can you test this patch on Solaris,
>>> > assuming the michael in
>>> > http://fate.ffmpeg.org/report.cgi?time=20150820233505&slot=x86-opensolaris-gcc4.3
>>> > is you?
>>> > I do not have Solaris with me, but based patch off of what I read online.
>>>
>>> Forgot to add, could you then rerun the fate stuff and reupload?
>>> This will help me fix remaining build issues due to large reduction in noise.
>>
>> fate should rerun automatically
>
> @Michael:
> Some suggestions for removing the terrible deprecation noise.
> 1. Could you update your GCC on the solaris box to 4.6?
> This will enable recognition of the pragma,
> see e.g https://stackoverflow.com/questions/13459602/how-can-i-get-rid-of-deprecated-warnings-in-deprecated-functions-in-gcc
>
> 2. Alternatively, could you compile with -Wno-deprecated (perhaps
> extend this to all boxes that run GCC < 4.6)?
> One could object saying this will not show deprecated stuff.
> However, it is way too noisy and makes it harder to extract useful info.
> See below for a compromise.

Scratch  above 1 and 2, they are unreasonable:
we need to test old compilers and as many configurations as possible.
>
> 3. If you don't like this idea, you could have 2 fate runs:
> one with the -Wnodeprecated and one without.

Even this is mostly unreasonable,
but of all the above, it makes the most sense to me.

>
> Thanks.
>
>>
>> [...]
>> --
>> Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB
>>
>> The worst form of inequality is to try to make unequal things equal.
>> -- Aristotle
>>
>> _______________________________________________
>> ffmpeg-devel mailing list
>> ffmpeg-devel at ffmpeg.org
>> http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
>>