From: Takashi Iwai <tiwai@suse.de>
To: broonie@kernel.org
Cc: Lee Jones <lee.jones@linaro.org>,
linux-arm-kernel@lists.infradead.org,
linux-kernel@vger.kernel.org, linus.walleij@linaro.org,
alsa-devel@alsa-project.org
Subject: Re: [PATCH 02/11] ASoC: ab8500: Revert back to find_next_bit()
Date: Mon, 23 Dec 2013 11:38:05 +0100 [thread overview]
Message-ID: <s5h4n5zette.wl%tiwai@suse.de> (raw)
In-Reply-To: <s5hzjnwrbq2.wl%tiwai@suse.de>
At Thu, 19 Dec 2013 18:29:09 +0100,
Takashi Iwai wrote:
>
> At Thu, 19 Dec 2013 16:48:24 +0000,
> Lee Jones wrote:
> >
> > > > Commit '166a34d ASoC: ab8500: Fix invalid cast to long pointer'
> > > > rather carelessly converts find_next_bit() to fls() (find last
> > > > bit set), which are not the same.
> > >
> > > Does it break on the real machines?
> >
> > It does, that's how I found the bug.
> >
> > > fls() behaves differently from find_next_bit(), of course, but in this
> > > case, it should work same in the end, since there are at most two
> > > bits.
> >
> > Unfortunately it doesn't work at all. This patch brings audio back to
> > a working state. It took me the best part of a day to track down the
> > issue. :(
>
> Hmm, then isn't the original code rather buggy?
>
> Check the values set there by ffs(), fls() and the original
> find_next_bit(), especially whether find_next_bit() gives a valid
> value fitting with the mask bits.
Meanwhile, it's not good to keep the broken code in the upstream, and
the bug is obvious. Below is the fixed patch, applicable with git am
scissors option. Lee, let me know if this still doesn't fix.
thanks,
Takashi
-- 8< --
From: Takashi Iwai <tiwai@suse.de>
Subject: [PATCH] ASoC: ab8500: Back to first_find_bit() & co
ffs() and fls() give the off-by-one value in comparison with
find_*_bit() helpers, and find_first_bit() and find_next_bit() are
more intuitive in the case of two bits. So, back to find_*_bit() but
call them properly without invalid cast. This fixes the slot
assignment regression introduced by commit 166a34d27fca.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
---
sound/soc/codecs/ab8500-codec.c | 15 +++++++++------
1 file changed, 9 insertions(+), 6 deletions(-)
diff --git a/sound/soc/codecs/ab8500-codec.c b/sound/soc/codecs/ab8500-codec.c
index 1ad92cbf0b24..cf53c1e2fabc 100644
--- a/sound/soc/codecs/ab8500-codec.c
+++ b/sound/soc/codecs/ab8500-codec.c
@@ -2242,6 +2242,7 @@ static int ab8500_codec_set_dai_tdm_slot(struct snd_soc_dai *dai,
{
struct snd_soc_codec *codec = dai->codec;
unsigned int val, mask, slot, slots_active;
+ unsigned long tmp;
mask = BIT(AB8500_DIGIFCONF2_IF0WL0) |
BIT(AB8500_DIGIFCONF2_IF0WL1);
@@ -2308,21 +2309,22 @@ static int ab8500_codec_set_dai_tdm_slot(struct snd_soc_dai *dai,
dev_dbg(dai->codec->dev, "%s: Slots, active, TX: %d\n", __func__,
slots_active);
+ tmp = tx_mask;
switch (slots_active) {
case 0:
break;
case 1:
- slot = ffs(tx_mask);
+ slot = find_first_bit(&tmp, 32);
snd_soc_update_bits(codec, AB8500_DASLOTCONF1, mask, slot);
snd_soc_update_bits(codec, AB8500_DASLOTCONF3, mask, slot);
snd_soc_update_bits(codec, AB8500_DASLOTCONF2, mask, slot);
snd_soc_update_bits(codec, AB8500_DASLOTCONF4, mask, slot);
break;
case 2:
- slot = ffs(tx_mask);
+ slot = find_first_bit(&tmp, 32);
snd_soc_update_bits(codec, AB8500_DASLOTCONF1, mask, slot);
snd_soc_update_bits(codec, AB8500_DASLOTCONF3, mask, slot);
- slot = fls(tx_mask);
+ slot = find_next_bit(&tmp, 32, slot + 1);
snd_soc_update_bits(codec, AB8500_DASLOTCONF2, mask, slot);
snd_soc_update_bits(codec, AB8500_DASLOTCONF4, mask, slot);
break;
@@ -2349,22 +2351,23 @@ static int ab8500_codec_set_dai_tdm_slot(struct snd_soc_dai *dai,
dev_dbg(dai->codec->dev, "%s: Slots, active, RX: %d\n", __func__,
slots_active);
+ tmp = rx_mask;
switch (slots_active) {
case 0:
break;
case 1:
- slot = ffs(rx_mask);
+ slot = find_first_bit(&tmp, 32);
snd_soc_update_bits(codec, AB8500_ADSLOTSEL(slot),
AB8500_MASK_SLOT(slot),
AB8500_ADSLOTSELX_AD_OUT_TO_SLOT(AB8500_AD_OUT3, slot));
break;
case 2:
- slot = ffs(rx_mask);
+ slot = find_first_bit(&tmp, 32);
snd_soc_update_bits(codec,
AB8500_ADSLOTSEL(slot),
AB8500_MASK_SLOT(slot),
AB8500_ADSLOTSELX_AD_OUT_TO_SLOT(AB8500_AD_OUT3, slot));
- slot = fls(rx_mask);
+ slot = find_next_bit(&tmp, 32, slot + 1);
snd_soc_update_bits(codec,
AB8500_ADSLOTSEL(slot),
AB8500_MASK_SLOT(slot),
--
1.8.5.2
next prev parent reply other threads:[~2013-12-23 10:38 UTC|newest]
Thread overview: 35+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-12-19 15:54 [PATCH 00/10] ASoC: ARM: ux500: Better prepare for full Device Tree Lee Jones
2013-12-19 15:54 ` [PATCH 01/11] ARM: ux500: Don't use enums for MSP IDs - for easy DT conversion Lee Jones
2014-01-07 14:03 ` Linus Walleij
2014-01-07 14:11 ` Lee Jones
2014-01-07 14:19 ` Mark Brown
2014-01-07 14:44 ` Lee Jones
2014-01-07 15:36 ` Mark Brown
2013-12-19 15:54 ` [PATCH 02/11] ASoC: ab8500: Revert back to find_next_bit() Lee Jones
2013-12-19 16:06 ` Takashi Iwai
2013-12-19 16:48 ` Lee Jones
2013-12-19 17:29 ` Takashi Iwai
2013-12-20 14:02 ` Lee Jones
2013-12-20 14:23 ` Takashi Iwai
2013-12-23 10:38 ` Takashi Iwai [this message]
2013-12-19 15:55 ` [PATCH 03/11] ASoC: ux500: Provide better checking for Device Tree and/or Platform Data Lee Jones
2014-01-07 15:37 ` Mark Brown
2013-12-19 15:55 ` [PATCH 04/11] ASoC: Ux500: Match platform by device node when booting Device Tree Lee Jones
2014-01-07 15:37 ` Mark Brown
2013-12-19 15:55 ` [PATCH 05/11] ASoC: ux500_pcm: Stop pretending that we support varying address widths Lee Jones
2014-01-07 15:38 ` Mark Brown
2013-12-19 15:55 ` [PATCH 06/11] ASoC: ux500_pcm: Expect different saved DMA data when obtaining from DAI store Lee Jones
2014-01-07 15:38 ` Mark Brown
2013-12-19 15:55 ` [PATCH 07/11] ASoC: ux500_pcm: Extend Device Tree support to deal with DMA data Lee Jones
2014-01-07 15:38 ` Mark Brown
2013-12-19 15:55 ` [PATCH 08/11] ASoC: ux500: Store DMA data in the DAI differently in the pdata and DT case Lee Jones
2014-01-07 15:38 ` Mark Brown
2013-12-19 15:55 ` [PATCH 09/11] ASoC: ux500_pcm: Take out pointless dev_dbg() call Lee Jones
2014-01-07 15:38 ` Mark Brown
2013-12-19 15:55 ` [PATCH 10/11] ASoC: ux500_pcm: Differentiate between pdata and DT initialisation Lee Jones
2014-01-07 15:39 ` Mark Brown
2013-12-19 15:55 ` [PATCH 11/11] ASoC: ux500: Dramatically reduce the size of the DAI driver data struct Lee Jones
2014-01-07 15:39 ` Mark Brown
2013-12-21 14:29 ` [PATCH 00/10] ASoC: ARM: ux500: Better prepare for full Device Tree Mark Brown
2013-12-21 17:19 ` Lee Jones
2014-01-07 14:06 ` Linus Walleij
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=s5h4n5zette.wl%tiwai@suse.de \
--to=tiwai@suse.de \
--cc=alsa-devel@alsa-project.org \
--cc=broonie@kernel.org \
--cc=lee.jones@linaro.org \
--cc=linus.walleij@linaro.org \
--cc=linux-arm-kernel@lists.infradead.org \
--cc=linux-kernel@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®