* [PATCH] ASoC: samsung: Use IRQ safe spin lock calls
@ 2016-02-18 15:47 Charles Keepax
2016-02-19 4:49 ` Anand Moon
2016-02-19 10:59 ` [alsa-devel] " Sylwester Nawrocki
0 siblings, 2 replies; 3+ messages in thread
From: Charles Keepax @ 2016-02-18 15:47 UTC (permalink / raw)
To: broonie
Cc: lgirdwood, s.nawrocki, sbkim73, linux-samsung-soc, alsa-devel,
patches
Lockdep warns of a potential lock inversion, i2s->lock is held numerous
times whilst we are under the substream lock (snd_pcm_stream_lock). If
we use the IRQ unsafe spin lock calls, you can also end up locking
snd_pcm_stream_lock whilst under i2s->lock (if an IRQ happens whilst we
are holding i2s->lock). This could result in deadlock.
[ 18.147001] CPU0 CPU1
[ 18.151509] ---- ----
[ 18.156022] lock(&(&pri_dai->spinlock)->rlock);
[ 18.160701] local_irq_disable();
[ 18.166622] lock(&(&substream->self_group.lock)->rlock);
[ 18.174595] lock(&(&pri_dai->spinlock)->rlock);
[ 18.181806] <Interrupt>
[ 18.184408] lock(&(&substream->self_group.lock)->rlock);
[ 18.190045]
[ 18.190045] *** DEADLOCK ***
This patch changes to using the irq safe spinlock calls, to avoid this
issue.
Fixes: ce8bcdbb61d9 ("ASoC: samsung: i2s: Protect more registers with a spinlock")
Signed-off-by: Charles Keepax <ckeepax@opensource.wolfsonmicro.com>
---
sound/soc/samsung/i2s.c | 21 ++++++++++++---------
1 file changed, 12 insertions(+), 9 deletions(-)
diff --git a/sound/soc/samsung/i2s.c b/sound/soc/samsung/i2s.c
index 84d9e77..70a2559 100644
--- a/sound/soc/samsung/i2s.c
+++ b/sound/soc/samsung/i2s.c
@@ -481,10 +481,11 @@ static int i2s_set_sysclk(struct snd_soc_dai *dai,
unsigned int cdcon_mask = 1 << i2s_regs->cdclkcon_off;
unsigned int rsrc_mask = 1 << i2s_regs->rclksrc_off;
u32 mod, mask, val = 0;
+ unsigned long flags;
- spin_lock(i2s->lock);
+ spin_lock_irqsave(i2s->lock, flags);
mod = readl(i2s->addr + I2SMOD);
- spin_unlock(i2s->lock);
+ spin_unlock_irqrestore(i2s->lock, flags);
switch (clk_id) {
case SAMSUNG_I2S_OPCLK:
@@ -575,11 +576,11 @@ static int i2s_set_sysclk(struct snd_soc_dai *dai,
return -EINVAL;
}
- spin_lock(i2s->lock);
+ spin_lock_irqsave(i2s->lock, flags);
mod = readl(i2s->addr + I2SMOD);
mod = (mod & ~mask) | val;
writel(mod, i2s->addr + I2SMOD);
- spin_unlock(i2s->lock);
+ spin_unlock_irqrestore(i2s->lock, flags);
return 0;
}
@@ -590,6 +591,7 @@ static int i2s_set_fmt(struct snd_soc_dai *dai,
struct i2s_dai *i2s = to_info(dai);
int lrp_shift, sdf_shift, sdf_mask, lrp_rlow, mod_slave;
u32 mod, tmp = 0;
+ unsigned long flags;
lrp_shift = i2s->variant_regs->lrp_off;
sdf_shift = i2s->variant_regs->sdf_off;
@@ -649,7 +651,7 @@ static int i2s_set_fmt(struct snd_soc_dai *dai,
return -EINVAL;
}
- spin_lock(i2s->lock);
+ spin_lock_irqsave(i2s->lock, flags);
mod = readl(i2s->addr + I2SMOD);
/*
* Don't change the I2S mode if any controller is active on this
@@ -657,7 +659,7 @@ static int i2s_set_fmt(struct snd_soc_dai *dai,
*/
if (any_active(i2s) &&
((mod & (sdf_mask | lrp_rlow | mod_slave)) != tmp)) {
- spin_unlock(i2s->lock);
+ spin_unlock_irqrestore(i2s->lock, flags);
dev_err(&i2s->pdev->dev,
"%s:%d Other DAI busy\n", __func__, __LINE__);
return -EAGAIN;
@@ -666,7 +668,7 @@ static int i2s_set_fmt(struct snd_soc_dai *dai,
mod &= ~(sdf_mask | lrp_rlow | mod_slave);
mod |= tmp;
writel(mod, i2s->addr + I2SMOD);
- spin_unlock(i2s->lock);
+ spin_unlock_irqrestore(i2s->lock, flags);
return 0;
}
@@ -676,6 +678,7 @@ static int i2s_hw_params(struct snd_pcm_substream *substream,
{
struct i2s_dai *i2s = to_info(dai);
u32 mod, mask = 0, val = 0;
+ unsigned long flags;
if (!is_secondary(i2s))
mask |= (MOD_DC2_EN | MOD_DC1_EN);
@@ -744,11 +747,11 @@ static int i2s_hw_params(struct snd_pcm_substream *substream,
return -EINVAL;
}
- spin_lock(i2s->lock);
+ spin_lock_irqsave(i2s->lock, flags);
mod = readl(i2s->addr + I2SMOD);
mod = (mod & ~mask) | val;
writel(mod, i2s->addr + I2SMOD);
- spin_unlock(i2s->lock);
+ spin_unlock_irqrestore(i2s->lock, flags);
samsung_asoc_init_dma_data(dai, &i2s->dma_playback, &i2s->dma_capture);
--
2.1.4
^ permalink raw reply related [flat|nested] 3+ messages in thread
* Re: [PATCH] ASoC: samsung: Use IRQ safe spin lock calls
2016-02-18 15:47 [PATCH] ASoC: samsung: Use IRQ safe spin lock calls Charles Keepax
@ 2016-02-19 4:49 ` Anand Moon
2016-02-19 10:59 ` [alsa-devel] " Sylwester Nawrocki
1 sibling, 0 replies; 3+ messages in thread
From: Anand Moon @ 2016-02-19 4:49 UTC (permalink / raw)
To: Charles Keepax
Cc: Mark Brown, Liam Girdwood, Sylwester Nawrocki, Kim Sangbeom,
linux-samsung-soc@vger.kernel.org, alsa-devel, patches
hi Charles,
On 18 February 2016 at 21:17, Charles Keepax
<ckeepax@opensource.wolfsonmicro.com> wrote:
> Lockdep warns of a potential lock inversion, i2s->lock is held numerous
> times whilst we are under the substream lock (snd_pcm_stream_lock). If
> we use the IRQ unsafe spin lock calls, you can also end up locking
> snd_pcm_stream_lock whilst under i2s->lock (if an IRQ happens whilst we
> are holding i2s->lock). This could result in deadlock.
>
> [ 18.147001] CPU0 CPU1
> [ 18.151509] ---- ----
> [ 18.156022] lock(&(&pri_dai->spinlock)->rlock);
> [ 18.160701] local_irq_disable();
> [ 18.166622] lock(&(&substream->self_group.lock)->rlock);
> [ 18.174595] lock(&(&pri_dai->spinlock)->rlock);
> [ 18.181806] <Interrupt>
> [ 18.184408] lock(&(&substream->self_group.lock)->rlock);
> [ 18.190045]
> [ 18.190045] *** DEADLOCK ***
>
> This patch changes to using the irq safe spinlock calls, to avoid this
> issue.
>
> Fixes: ce8bcdbb61d9 ("ASoC: samsung: i2s: Protect more registers with a spinlock")
> Signed-off-by: Charles Keepax <ckeepax@opensource.wolfsonmicro.com>
> ---
> sound/soc/samsung/i2s.c | 21 ++++++++++++---------
> 1 file changed, 12 insertions(+), 9 deletions(-)
>
> diff --git a/sound/soc/samsung/i2s.c b/sound/soc/samsung/i2s.c
> index 84d9e77..70a2559 100644
> --- a/sound/soc/samsung/i2s.c
> +++ b/sound/soc/samsung/i2s.c
> @@ -481,10 +481,11 @@ static int i2s_set_sysclk(struct snd_soc_dai *dai,
> unsigned int cdcon_mask = 1 << i2s_regs->cdclkcon_off;
> unsigned int rsrc_mask = 1 << i2s_regs->rclksrc_off;
> u32 mod, mask, val = 0;
> + unsigned long flags;
>
> - spin_lock(i2s->lock);
> + spin_lock_irqsave(i2s->lock, flags);
> mod = readl(i2s->addr + I2SMOD);
> - spin_unlock(i2s->lock);
> + spin_unlock_irqrestore(i2s->lock, flags);
>
> switch (clk_id) {
> case SAMSUNG_I2S_OPCLK:
> @@ -575,11 +576,11 @@ static int i2s_set_sysclk(struct snd_soc_dai *dai,
> return -EINVAL;
> }
>
> - spin_lock(i2s->lock);
> + spin_lock_irqsave(i2s->lock, flags);
> mod = readl(i2s->addr + I2SMOD);
> mod = (mod & ~mask) | val;
> writel(mod, i2s->addr + I2SMOD);
> - spin_unlock(i2s->lock);
> + spin_unlock_irqrestore(i2s->lock, flags);
>
> return 0;
> }
> @@ -590,6 +591,7 @@ static int i2s_set_fmt(struct snd_soc_dai *dai,
> struct i2s_dai *i2s = to_info(dai);
> int lrp_shift, sdf_shift, sdf_mask, lrp_rlow, mod_slave;
> u32 mod, tmp = 0;
> + unsigned long flags;
>
> lrp_shift = i2s->variant_regs->lrp_off;
> sdf_shift = i2s->variant_regs->sdf_off;
> @@ -649,7 +651,7 @@ static int i2s_set_fmt(struct snd_soc_dai *dai,
> return -EINVAL;
> }
>
> - spin_lock(i2s->lock);
> + spin_lock_irqsave(i2s->lock, flags);
> mod = readl(i2s->addr + I2SMOD);
> /*
> * Don't change the I2S mode if any controller is active on this
> @@ -657,7 +659,7 @@ static int i2s_set_fmt(struct snd_soc_dai *dai,
> */
> if (any_active(i2s) &&
> ((mod & (sdf_mask | lrp_rlow | mod_slave)) != tmp)) {
> - spin_unlock(i2s->lock);
> + spin_unlock_irqrestore(i2s->lock, flags);
> dev_err(&i2s->pdev->dev,
> "%s:%d Other DAI busy\n", __func__, __LINE__);
> return -EAGAIN;
> @@ -666,7 +668,7 @@ static int i2s_set_fmt(struct snd_soc_dai *dai,
> mod &= ~(sdf_mask | lrp_rlow | mod_slave);
> mod |= tmp;
> writel(mod, i2s->addr + I2SMOD);
> - spin_unlock(i2s->lock);
> + spin_unlock_irqrestore(i2s->lock, flags);
>
> return 0;
> }
> @@ -676,6 +678,7 @@ static int i2s_hw_params(struct snd_pcm_substream *substream,
> {
> struct i2s_dai *i2s = to_info(dai);
> u32 mod, mask = 0, val = 0;
> + unsigned long flags;
>
> if (!is_secondary(i2s))
> mask |= (MOD_DC2_EN | MOD_DC1_EN);
> @@ -744,11 +747,11 @@ static int i2s_hw_params(struct snd_pcm_substream *substream,
> return -EINVAL;
> }
>
> - spin_lock(i2s->lock);
> + spin_lock_irqsave(i2s->lock, flags);
> mod = readl(i2s->addr + I2SMOD);
> mod = (mod & ~mask) | val;
> writel(mod, i2s->addr + I2SMOD);
> - spin_unlock(i2s->lock);
> + spin_unlock_irqrestore(i2s->lock, flags);
>
> samsung_asoc_init_dma_data(dai, &i2s->dma_playback, &i2s->dma_capture);
>
> --
> 2.1.4
>
Thanks for fixing this bug.
Tested on Odroid U3.
Tested-by: Anand Moon <linux.amoon@gmail.com>
-Anand Moon
> --
> To unsubscribe from this list: send the line "unsubscribe linux-samsung-soc" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [alsa-devel] [PATCH] ASoC: samsung: Use IRQ safe spin lock calls
2016-02-18 15:47 [PATCH] ASoC: samsung: Use IRQ safe spin lock calls Charles Keepax
2016-02-19 4:49 ` Anand Moon
@ 2016-02-19 10:59 ` Sylwester Nawrocki
1 sibling, 0 replies; 3+ messages in thread
From: Sylwester Nawrocki @ 2016-02-19 10:59 UTC (permalink / raw)
To: Charles Keepax, broonie
Cc: alsa-devel, linux-samsung-soc, sbkim73, patches, lgirdwood
On 18/02/16 16:47, Charles Keepax wrote:
> Lockdep warns of a potential lock inversion, i2s->lock is held numerous
> times whilst we are under the substream lock (snd_pcm_stream_lock). If
> we use the IRQ unsafe spin lock calls, you can also end up locking
> snd_pcm_stream_lock whilst under i2s->lock (if an IRQ happens whilst we
> are holding i2s->lock). This could result in deadlock.
>
> [ 18.147001] CPU0 CPU1
> [ 18.151509] ---- ----
> [ 18.156022] lock(&(&pri_dai->spinlock)->rlock);
> [ 18.160701] local_irq_disable();
> [ 18.166622] lock(&(&substream->self_group.lock)->rlock);
> [ 18.174595] lock(&(&pri_dai->spinlock)->rlock);
> [ 18.181806] <Interrupt>
> [ 18.184408] lock(&(&substream->self_group.lock)->rlock);
> [ 18.190045]
> [ 18.190045] *** DEADLOCK ***
>
> This patch changes to using the irq safe spinlock calls, to avoid this
> issue.
>
> Fixes: ce8bcdbb61d9 ("ASoC: samsung: i2s: Protect more registers with a spinlock")
> Signed-off-by: Charles Keepax <ckeepax@opensource.wolfsonmicro.com>
Acked-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2016-02-19 11:01 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2016-02-18 15:47 [PATCH] ASoC: samsung: Use IRQ safe spin lock calls Charles Keepax
2016-02-19 4:49 ` Anand Moon
2016-02-19 10:59 ` [alsa-devel] " Sylwester Nawrocki
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox