On Thu, 2018-10-11 at 10:00 +0200, Takashi Iwai wrote:
> On Thu, 11 Oct 2018 09:45:07 +0200,
> Takashi Sakamoto wrote:
> >
> > On Oct 11 2018 16:31, Takashi Iwai wrote:
> > > On Wed, 10 Oct 2018 23:27:29 +0200,
> > > Guedes, Andre wrote:
> > > >
> > > > On Wed, 2018-10-10 at 12:34 +0200, Takashi Iwai wrote:
> > > > > > +static int aaf_set_hw_constraint(snd_pcm_aaf_t *aaf)
> > > > > > +{
> > > > > > + int res;
> > > > > > + snd_pcm_ioplug_t *io = &aaf->io;
> > > > > > + const unsigned int accesses[] = {
> > > > > > + SND_PCM_ACCESS_RW_INTERLEAVED,
> > > > > > + };
> > > > > > + const unsigned int formats[] = {
> > > > > > + SND_PCM_FORMAT_S16_BE,
> > > > > > + SND_PCM_FORMAT_S24_3BE,
> > > > > > + SND_PCM_FORMAT_S32_BE,
> > > > > > + SND_PCM_FORMAT_FLOAT_BE,
> > > > > > + };
> > > > > > + const unsigned int rates[] = {
> > > > > > + 8000,
> > > > > > + 16000,
> > > > > > + 24000,
> > > > > > + 32000,
> > > > > > + 44100,
> > > > > > + 48000,
> > > > > > + 88200,
> > > > > > + 96000,
> > > > > > + 176400,
> > > > > > + 192000,
> > > > > > + };
> > > > >
> > > > > Use static arrays for the above.
> > > >
> > > > It isn't clear to me what is the point in allocating these
> > > > variables in
> > > > the global scope. May I ask what is the benefit in adding the
> > > > 'static'
> > > > modifier to these local variables?
> > >
> > > Yes.
> >
> > In this case, it's not relevant to access scope.
> >
> > Just with 'const' qualifier, the symbol allocates in .text section
> > of ELF binary. With additional 'static' qualifier, the symbol
> > allocates in .rodata section.
> >
> > In loading, symbol in .rodata section locates in segments of VMA
> > with 'read-only' flag. This brings safety from several bugs such as
> > buffer over-run.
My understanding is that the arrays are allocated in the stack frame,
not in .text section, and the 'const' qualifier here has no influence
on that. But I get your point in moving the arrays to .rodata.
> OK, then scratch my comments.
I checked other plugins and it seems having 'static' to these variables
is the "norm" so I'm adding it for the sake of consistency.
Thanks,
Andre
Attachment:
smime.p7s
Description: S/MIME cryptographic signature
_______________________________________________ Alsa-devel mailing list Alsa-devel@xxxxxxxxxxxxxxxx http://mailman.alsa-project.org/mailman/listinfo/alsa-devel
