[alsa-devel] [PATCH 2/4] topology: Change variable type to fix gcc warning
Lin, Mengdong
mengdong.lin at intel.com
Wed Nov 18 16:14:43 CET 2015
> -----Original Message-----
> From: Takashi Iwai [mailto:tiwai at suse.de]
> Sent: Wednesday, November 18, 2015 4:19 PM
>
> On Wed, 18 Nov 2015 08:28:09 +0100,
> Lin, Mengdong wrote:
> >
> > > -----Original Message-----
> > > From: Takashi Iwai [mailto:tiwai at suse.de]
> > > Sent: Wednesday, November 18, 2015 3:16 PM
> > > To: mengdong.lin at linux.intel.com
> >
> > > On Wed, 18 Nov 2015 08:23:07 +0100,
> > > mengdong.lin at linux.intel.com wrote:
> > > >
> > > > From: Mengdong Lin <mengdong.lin at linux.intel.com>
> > > >
> > > > Fix warning: comparison between signed and unsigned integer
> > > > expressions [-Wsign-compare]
> > > >
> > > > ABI objects use type _le32, which is converted to host unsigned integer.
> > > > So the iterator 'i' in a loop as below should also be unsigned.
> > > > for (i = 0; i < pcm->num_streams; i++)
> > > > ^
> > >
> > > Using an unsigned int for the generic loop count like i is strange.
> >
> > Yes.
> >
> > > Rather compare with the original value in the template, which is actually int.
> >
> > Are you suggesting not change the code?
>
> No.
>
> > The "pcm->num_streams" here is __le32 defined in ABI:
> > struct snd_soc_tplg_pcm {
> > ...
> > __le32 num_streams; /* number of streams */
> > ...
> > } __attribute__((packed));
> >
> > It seems gcc takes it as unsigned integer and so I get the warning.
>
> The problem isn't only about the endianess.
> Conceptually, it's wrong to refer directly to le32 data, as it's not being CPU
> endian. It should have always an endian conversion. Of course, we support
> only LE32, so it's no big issue, so far.
>
> What you need to refer to is the value in the template instead. It's guaranteed
> to be CPU endianess, and it's even int.
> So, the code would look like:
>
> for (i = 0; i < pcm_tpl->num_streams; i++)
>
Okay, I'll revise the patch. Thanks for your suggestion.
> BTW, I haven't checked whether topology API would "work" on big-endian
> architecture; does it return an error properly?
Sorry, no. I'll check if an API can tell us the host endianness and let topology return an error for big-endian machines.
I had thought to use this API when generating the ABI objects from host type to _le32:
uint32_t htole32(uint32_t host_32bits);
But since I have no big-endian machines to test this, maybe it's better to return an error for them.
Thanks
Mengdong
>
> Takashi
>
>
> >
> > Thanks
> > Mengdong
> >
> > >
> > >
> > > Takashi
> > >
> > > > Signed-off-by: Mengdong Lin <mengdong.lin at linux.intel.com>
> > > >
> > > > diff --git a/src/topology/ctl.c b/src/topology/ctl.c index
> > > > 7d8787f..6dc3b3d 100644
> > > > --- a/src/topology/ctl.c
> > > > +++ b/src/topology/ctl.c
> > > > @@ -676,7 +676,8 @@ int tplg_add_mixer(snd_tplg_t *tplg, struct
> > > snd_tplg_mixer_template *mixer,
> > > > struct snd_soc_tplg_private *priv = mixer->priv;
> > > > struct snd_soc_tplg_mixer_control *mc;
> > > > struct tplg_elem *elem;
> > > > - int ret, i;
> > > > + int ret;
> > > > + unsigned int i;
> > > >
> > > > tplg_dbg(" Control Mixer: %s\n", mixer->hdr.name);
> > > >
> > > > @@ -743,7 +744,8 @@ int tplg_add_enum(snd_tplg_t *tplg, struct
> > > > snd_tplg_enum_template *enum_ctl, {
> > > > struct snd_soc_tplg_enum_control *ec;
> > > > struct tplg_elem *elem;
> > > > - int ret, i;
> > > > + int ret;
> > > > + unsigned int i;
> > > >
> > > > tplg_dbg(" Control Enum: %s\n", enum_ctl->hdr.name);
> > > >
> > > > diff --git a/src/topology/pcm.c b/src/topology/pcm.c index
> > > > 9b7e402..4b7c058 100644
> > > > --- a/src/topology/pcm.c
> > > > +++ b/src/topology/pcm.c
> > > > @@ -522,7 +522,7 @@ int tplg_add_pcm_object(snd_tplg_t *tplg,
> > > snd_tplg_obj_template_t *t)
> > > > struct snd_tplg_pcm_template *pcm_tpl = t->pcm;
> > > > struct snd_soc_tplg_pcm *pcm;
> > > > struct tplg_elem *elem;
> > > > - int i;
> > > > + unsigned int i;
> > > >
> > > > tplg_dbg("PCM: %s, DAI %s\n", pcm_tpl->pcm_name,
> > > pcm_tpl->dai_name);
> > > >
> > > > @@ -564,7 +564,7 @@ int tplg_add_link_object(snd_tplg_t *tplg,
> > > snd_tplg_obj_template_t *t)
> > > > struct snd_tplg_link_template *link = t->link;
> > > > struct snd_soc_tplg_link_config *lk;
> > > > struct tplg_elem *elem;
> > > > - int i;
> > > > + unsigned int i;
> > > >
> > > > if (t->type != SND_TPLG_TYPE_BE && t->type != SND_TPLG_TYPE_CC)
> > > > return -EINVAL;
> > > > --
> > > > 2.5.0
> > > >
> >
More information about the Alsa-devel
mailing list