From 3ee77f811a91350caa56783b76251cd8218e2263 Mon Sep 17 00:00:00 2001
From: Zeliang Li <lizeliang.linux@gmail.com>
Date: Sat, 15 Aug 2026 01:52:57 +0800
Subject: [PATCH] ASoC: tas2781: Add delay before block checksum retry to
 prevent deadlock

During runtime resume from power-saving states, the tas2781 amplifier
firmware block calibration checksum verification may occasionally fail
due to transient instabilities on the I2C bus or the chip's internal
power rails.

When tasdev_block_chksum() detects a checksum mismatch, it correctly
decrements block->nr_retry and returns -EAGAIN to trigger a
re-transmission loop. However, because there is no delay introduced
before repeating the loop, all available retry attempts are exhausted
almost instantly (within microseconds) before the hardware can fully
stabilize. This causes a permanent calibration failure, triggering
"ERROR_PRAM_CRCCHK", and results in a silent speaker deadlock after
pausing/resuming media on modern laptops.

Fix this by adding a 2ms usleep_range() delay before initiating a
checksum re-try loop, allowing the hardware sufficient time to
stabilize.

Signed-off-by: Zeliang Li <lizeliang.linux@gmail.com>
---
 sound/soc/codecs/tas2781-fmwlib.c | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)

diff --git a/sound/soc/codecs/tas2781-fmwlib.c
b/sound/soc/codecs/tas2781-fmwlib.c
index dcbeb9618195..09cbee7a55dc 100644
--- a/sound/soc/codecs/tas2781-fmwlib.c
+++ b/sound/soc/codecs/tas2781-fmwlib.c
@@ -1765,9 +1765,16 @@ static int tasdev_block_chksum(struct
tasdevice_priv *tas_priv,
  ret = -EAGAIN;
  block->nr_retry--;

- if (block->nr_retry <= 0)
+ if (block->nr_retry <= 0) {
  set_err_prg_cfg(block->type,
  &tas_priv->tasdevice[chn]);
+
+ } else {
+ /* Give the chip and I2C bus time to stabilize before
+ * next re-transmission attempt.
+ */
+ usleep_range(2000, 2500);
+ }
  } else
  tas_priv->tasdevice[chn].err_code &= ~ERROR_PRAM_CRCCHK;

-- 
2.53.0

-- 
KISS == Keep it simple,stupid~:-)
http://lizeliang.org

dmesg on my laptop:
[ 4024.480031]  snd_hda_codec_alc269 hdaudioC0D0: bound
i2c-TIAS2781:00 (ops tas2781_hda_comp_ops
[snd_hda_scodec_tas2781_i2c])
[ 4025.643145]  tas2781-hda i2c-TIAS2781:00: tas2781_apply_calib: V1 CRC error

李则良 <lizeliang.linux@gmail.com> 于2026年8月15日周六 02:19写道：
>
> From 3ee77f811a91350caa56783b76251cd8218e2263 Mon Sep 17 00:00:00 2001
> From: Zeliang Li <lizeliang.linux@gmail.com>
> Date: Sat, 15 Aug 2026 01:52:57 +0800
> Subject: [PATCH] ASoC: tas2781: Add delay before block checksum retry to
>  prevent deadlock
>
> During runtime resume from power-saving states, the tas2781 amplifier
> firmware block calibration checksum verification may occasionally fail
> due to transient instabilities on the I2C bus or the chip's internal
> power rails.
>
> When tasdev_block_chksum() detects a checksum mismatch, it correctly
> decrements block->nr_retry and returns -EAGAIN to trigger a
> re-transmission loop. However, because there is no delay introduced
> before repeating the loop, all available retry attempts are exhausted
> almost instantly (within microseconds) before the hardware can fully
> stabilize. This causes a permanent calibration failure, triggering
> "ERROR_PRAM_CRCCHK", and results in a silent speaker deadlock after
> pausing/resuming media on modern laptops.
>
> Fix this by adding a 2ms usleep_range() delay before initiating a
> checksum re-try loop, allowing the hardware sufficient time to
> stabilize.
>
> Signed-off-by: Zeliang Li <lizeliang.linux@gmail.com>
> ---
>  sound/soc/codecs/tas2781-fmwlib.c | 9 ++++++++-
>  1 file changed, 8 insertions(+), 1 deletion(-)
>
> diff --git a/sound/soc/codecs/tas2781-fmwlib.c
> b/sound/soc/codecs/tas2781-fmwlib.c
> index dcbeb9618195..09cbee7a55dc 100644
> --- a/sound/soc/codecs/tas2781-fmwlib.c
> +++ b/sound/soc/codecs/tas2781-fmwlib.c
> @@ -1765,9 +1765,16 @@ static int tasdev_block_chksum(struct
> tasdevice_priv *tas_priv,
>   ret = -EAGAIN;
>   block->nr_retry--;
>
> - if (block->nr_retry <= 0)
> + if (block->nr_retry <= 0) {
>   set_err_prg_cfg(block->type,
>   &tas_priv->tasdevice[chn]);
> +
> + } else {
> + /* Give the chip and I2C bus time to stabilize before
> + * next re-transmission attempt.
> + */
> + usleep_range(2000, 2500);
> + }
>   } else
>   tas_priv->tasdevice[chn].err_code &= ~ERROR_PRAM_CRCCHK;
>
> --
> 2.53.0
>
> --
> KISS == Keep it simple,stupid~:-)
> http://lizeliang.org



-- 
KISS == Keep it simple,stupid~:-)
http://lizeliang.org

On Sat, Aug 15, 2026 at 02:19:26AM +0800, 李则良 wrote:
> +++ b/sound/soc/codecs/tas2781-fmwlib.c
> @@ -1765,9 +1765,16 @@ static int tasdev_block_chksum(struct
> tasdevice_priv *tas_priv,
>   ret = -EAGAIN;
>   block->nr_retry--;
> 
> - if (block->nr_retry <= 0)
> + if (block->nr_retry <= 0) {
>   set_err_prg_cfg(block->type,
>   &tas_priv->tasdevice[chn]);
> +
> + } else {
> + /* Give the chip and I2C bus time to stabilize before
> + * next re-transmission attempt.
> + */
> + usleep_range(2000, 2500);
> + }
>   } else
>   tas_priv->tasdevice[chn].err_code &= ~ERROR_PRAM_CRCCHK;

Your mail software has mangled this patch so it's unusable - at least
all the tabs have been replaced by spaces.  You might want to look at b4
to send the patches, especially it's web submission endpoint which
avoids having to configure mail:

   https://b4.docs.kernel.org/en/latest/contributor/overview.html

> Your mail software has mangled this patch so it's unusable - at least
> all the tabs have been replaced by spaces.  You might want to look at b4
> to send the patches, especially it's web submission endpoint which
> avoids having to configure mail:
>
>    https://b4.docs.kernel.org/en/latest/contributor/overview.html

Thank you for the guidance. My mail client indeed corrupted the tabs.
I will switch to the b4 tool and submit V2 via the web endpoint
shortly.

-- 
KISS == Keep it simple,stupid~:-)
http://lizeliang.org