From: Takashi Iwai Subject: Re: [PATCH v2 3/3] topology: A API calls to directly build topology data from templates Date: Tue, 11 Aug 2015 09:57:30 +0200 Message-ID: References: <1439230429-6173-1-git-send-email-liam.r.girdwood@linux.intel.com> <1439230429-6173-4-git-send-email-liam.r.girdwood@linux.intel.com> Mime-Version: 1.0 (generated by SEMI 1.14.6 - "Maruoka") Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Return-path: Received: from mx2.suse.de (mx2.suse.de [195.135.220.15]) by alsa0.perex.cz (Postfix) with ESMTP id C48882604D3 for ; Tue, 11 Aug 2015 09:57:30 +0200 (CEST) In-Reply-To: <1439230429-6173-4-git-send-email-liam.r.girdwood@linux.intel.com> List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: alsa-devel-bounces@alsa-project.org Sender: alsa-devel-bounces@alsa-project.org To: Liam Girdwood Cc: Mengdong Lin , alsa-devel@alsa-project.org, Mark Brown List-Id: alsa-devel@alsa-project.org On Mon, 10 Aug 2015 20:13:49 +0200, Liam Girdwood wrote: > +int tplg_add_bytes(snd_tplg_t *tplg, struct snd_tplg_bytes_template *bytes_ctl, > + struct tplg_elem **e) > +{ > + struct snd_soc_tplg_bytes_control *be; > + struct tplg_elem *elem; > + int ret; > + > + tplg_dbg(" Control Bytes: %s\n", bytes_ctl->hdr.name); > + > + if (bytes_ctl->hdr.type != SND_SOC_TPLG_TYPE_BYTES) { > + SNDERR("error: invalid bytes type %d\n", bytes_ctl->hdr.type); > + return -EINVAL; > + } > + > + elem = tplg_elem_new_common(tplg, NULL, bytes_ctl->hdr.name, > + SND_TPLG_TYPE_BYTES); > + if (!elem) > + return -ENOMEM; > + > + be = elem->bytes_ext; > + be->size = elem->size; > + ret = init_ctl_hdr(&be->hdr, &bytes_ctl->hdr); > + if (ret < 0) { > + tplg_elem_free(elem); > + return ret; > + } > + > + be->max = bytes_ctl->max; > + be->mask = bytes_ctl->mask; > + be->base = bytes_ctl->base; > + be->num_regs = bytes_ctl->num_regs; > + be->ext_ops.put = bytes_ctl->ext_ops.put; > + be->ext_ops.get = bytes_ctl->ext_ops.get; > + > + if (bytes_ctl->priv != NULL) { > + be = realloc(be, > + elem->size + bytes_ctl->priv->size); > + if (!be) { > + tplg_elem_free(elem); > + return -ENOMEM; > + } > + elem->bytes_ext = be; > + elem->size += bytes_ctl->priv->size; > + > + memcpy(be->priv.data, bytes_ctl->priv->data, > + bytes_ctl->priv->size); > + > + be->priv.size = bytes_ctl->priv->size; > + } > + > + /* check on TLV bytes control */ > + if (be->hdr.access & SNDRV_CTL_ELEM_ACCESS_TLV_CALLBACK) { > + if (be->hdr.access & SNDRV_CTL_ELEM_ACCESS_TLV_READWRITE > + != SNDRV_CTL_ELEM_ACCESS_TLV_READWRITE) { > + SNDERR("error: Invalid TLV bytes control access 0x%x\n", > + be->hdr.access); > + return -EINVAL; Missing tplg_elem_free()? > +struct tplg_elem* tplg_elem_new_route(snd_tplg_t *tplg) > +{ > + struct tplg_elem *elem; > + struct snd_soc_tplg_dapm_graph_elem *line; > + > + elem = tplg_elem_new(); > + if (!elem) > + return NULL; > + > + list_add_tail(&elem->list, &tplg->route_list); > + strcpy(elem->id, "line"); > + elem->type = SND_TPLG_TYPE_DAPM_GRAPH; > + elem->size = sizeof(*line); > + > + line = calloc(1, sizeof(*line)); > + if (!line) > + return NULL; Missing free of elem in the error path. Also, tplg->route_list still contains the stray elem. I guess this was a bug in the original code, so you can address it before this patch. > +int tplg_add_widget_object(snd_tplg_t *tplg, snd_tplg_obj_template_t *t) > +{ > + struct snd_tplg_widget_template *wt = t->widget; > + struct snd_soc_tplg_dapm_widget *w; > + struct tplg_elem *elem; > + int i, ret = 0; > + > + tplg_dbg("Widget: %s\n", wt->name); > + > + elem = tplg_elem_new_common(tplg, NULL, wt->name, > + SND_TPLG_TYPE_DAPM_WIDGET); > + if (!elem) > + return -ENOMEM; > + > + /* init new widget */ > + w = elem->widget; > + w->size = elem->size; > + > + w->id = wt->id; > + elem_copy_text(w->name, wt->name, SNDRV_CTL_ELEM_ID_NAME_MAXLEN); > + if (wt->sname) > + elem_copy_text(w->sname, wt->sname, > + SNDRV_CTL_ELEM_ID_NAME_MAXLEN); Indentation problem. > + /* add controls to the widget's reference list */ > + for (i = 0 ; i < wt->num_ctls; i++) { > + struct snd_tplg_ctl_template *ct = wt->ctl[i]; > + struct tplg_elem *elem_ctl; > + struct snd_tplg_mixer_template *mt; > + struct snd_tplg_bytes_template *bt; > + struct snd_tplg_enum_template *et; > + > + if (!ct) { > + tplg_elem_free(elem); > + return -EINVAL; > + } > + > + switch (ct->type) { > + case SND_SOC_TPLG_TYPE_MIXER: > + mt = container_of(ct, struct snd_tplg_mixer_template, hdr); > + ret = tplg_add_mixer(tplg, mt, &elem_ctl); > + break; > + > + case SND_SOC_TPLG_TYPE_BYTES: > + bt = container_of(ct, struct snd_tplg_bytes_template, hdr); > + ret = tplg_add_bytes(tplg, bt, &elem_ctl); > + break; > + > + case SND_SOC_TPLG_TYPE_ENUM: > + et = container_of(ct, struct snd_tplg_enum_template, hdr); > + ret = tplg_add_enum(tplg, bt, &elem_ctl); et instead of bt? > +int snd_tplg_build(snd_tplg_t *tplg, const char *outfile) > +{ > + int err; > + > + /* delete any old output files */ > + unlink(outfile); > + > + tplg->out_fd = > + open(outfile, O_RDWR | O_CREAT, S_IRWXU | S_IRWXG | S_IRWXO); > + if (tplg->out_fd < 0) { > + SNDERR("error: failed to open %s err %d\n", > + outfile, -errno); > + return -errno; > + } > + > + err = tplg_build_integ(tplg); > + if (err < 0) { > + SNDERR("error: failed to check topology integrity\n"); > + goto out; > + } > + > + err = tplg_write_data(tplg); > + if (err < 0) { > + SNDERR("error: failed to write data %d\n", err); > + goto out; > + } > + > +out: > + close(tplg->out_fd); It's a question whether we should unlink automatically at error or not. If we leave the file as is, it should be documented. > diff --git a/src/topology/tplg_local.h b/src/topology/tplg_local.h > index 953d822..97ee33e 100644 > --- a/src/topology/tplg_local.h > +++ b/src/topology/tplg_local.h > @@ -34,6 +34,10 @@ > #define ALSA_TPLG_DIR ALSA_CONFIG_DIR "/topology" > #define ARRAY_SIZE(x) (sizeof(x) / sizeof(x[0])) > > +#define container_of(ptr, type, member) ({ \ > + const typeof( ((type *)0)->member ) *__mptr = (ptr); \ > + (type *)( (char *)__mptr - offsetof(type,member) );}) This is a pretty common macro, so I prefer having it in include/local.h. thanks, Takashi