While probing, AC97 codec drivers and soc-core generically execute the following sequence: snd_soc_new_ac97_codec -> snd_soc_new_pcms -> reset ac-link/read AC97 ID to detect ->... -> set platform_data to ac97 by soc-core

commit 474828a40f6ddab6e2a3475a19c5c84aa3ec7d60 adds platform_data to snd_ac97 instance. But ac97 platform data hasn't given to snd_ac97 before actual ac97 operations. Then while ac97_read access platform_data of snd_ac97 for detecting, NULL pointer oops will fire. That means old platform_data patch doesn't work in real-life cases.

This patch moves the operation of setting ac97 platform_data earlier than ac97 reading/writing operations. Then it makes platform_data of AC97 become practically useful.

Signed-off-by: Barry Song 21cnbao@gmail.com cc: Marek Vasut marek.vasut@gmail.com --- sound/soc/soc-core.c | 9 +++++---- 1 files changed, 5 insertions(+), 4 deletions(-)

diff --git a/sound/soc/soc-core.c b/sound/soc/soc-core.c index e2b6d75..ef8f282 100644 --- a/sound/soc/soc-core.c +++ b/sound/soc/soc-core.c @@ -1083,11 +1083,8 @@ static void snd_soc_instantiate_card(struct snd_soc_card *card) continue; } } - if (card->dai_link[i].codec_dai->ac97_control) { + if (card->dai_link[i].codec_dai->ac97_control) ac97 = 1; - snd_ac97_dev_add_pdata(codec->ac97, - card->dai_link[i].cpu_dai->ac97_pdata); - } }

snprintf(codec->card->shortname, sizeof(codec->card->shortname), @@ -1510,6 +1507,10 @@ int snd_soc_new_pcms(struct snd_soc_device *socdev, int idx, const char *xid) mutex_unlock(&codec->mutex); return ret; } + if (card->dai_link[i].codec_dai->ac97_control) { + snd_ac97_dev_add_pdata(codec->ac97, + card->dai_link[i].cpu_dai->ac97_pdata); + } }

mutex_unlock(&codec->mutex);