[PATCH v1] ALSA: hda/tas2781: Fix the order of TAS2781 calibrated-data

A bug reported by one of my customers that the order of TAS2781 calibrated-data is incorrect, the correct way is to move R0_Low_%d and insert it between R0_%d and InvR0_%d.
Fixes: 4fe238513407 ("ALSA: hda/tas2781: Move and unified the calibrated-data getting function for SPI and I2C into the tas2781_hda lib") Signed-off-by: Shenghao Ding shenghao-ding@ti.com
--- v1: - Add varialbe csz to store cali_data->cali_dat_sz_per_dev, which is long enough to need two lines. - Add more comments on calibrated-data order --- sound/hda/codecs/side-codecs/tas2781_hda.c | 43 ++++++++++++++++++---- 1 file changed, 35 insertions(+), 8 deletions(-)
diff --git a/sound/hda/codecs/side-codecs/tas2781_hda.c b/sound/hda/codecs/side-codecs/tas2781_hda.c index f46d2e06c64f..d5913a727be4 100644 --- a/sound/hda/codecs/side-codecs/tas2781_hda.c +++ b/sound/hda/codecs/side-codecs/tas2781_hda.c @@ -33,6 +33,32 @@ const efi_guid_t tasdev_fct_efi_guid[] = { }; EXPORT_SYMBOL_NS_GPL(tasdev_fct_efi_guid, "SND_HDA_SCODEC_TAS2781");
+/* + * The order of calibrated-data writing is a bit different from the order + * in UEFI. Here is the conversion to match the order of calibrated-data + * writing. + */ +static void cali_cnv(unsigned char *data, unsigned int base, int offset) +{ + __be32 bedata[TASDEV_CALIB_N]; + int i; + + /* r0_reg */ + bedata[0] = cpu_to_be32(*(uint32_t *)&data[base]); + /* r0_low_reg */ + bedata[1] = cpu_to_be32(*(uint32_t *)&data[base + 8]); + /* invr0_reg */ + bedata[2] = cpu_to_be32(*(uint32_t *)&data[base + 4]); + /* pow_reg */ + bedata[3] = cpu_to_be32(*(uint32_t *)&data[base + 12]); + /* tlimit_reg */ + bedata[4] = cpu_to_be32(*(uint32_t *)&data[base + 16]); + + for (i = 0; i < TASDEV_CALIB_N; i++) + memcpy(&data[offset + i * 4 + 1], &bedata[i], + sizeof(bedata[i])); +} + static void tas2781_apply_calib(struct tasdevice_priv *p) { struct calidata *cali_data = &p->cali_data; @@ -46,6 +72,7 @@ static void tas2781_apply_calib(struct tasdevice_priv *p) TASDEVICE_REG(0, 0x13, 0x70), TASDEVICE_REG(0, 0x18, 0x7c), }; + unsigned int csz = cali_data->cali_dat_sz_per_dev; unsigned int crc, oft, node_num; unsigned char *buf; int i, j, k, l; @@ -86,6 +113,7 @@ static void tas2781_apply_calib(struct tasdevice_priv *p)
for (j = 0, k = 0; j < node_num; j++) { oft = j * 6 + 3; + /* Calibration registers address */ if (tmp_val[oft] == TASDEV_UEFI_CALI_REG_ADDR_FLG) { for (i = 0; i < TASDEV_CALIB_N; i++) { buf = &data[(oft + i + 1) * 4]; @@ -93,7 +121,8 @@ static void tas2781_apply_calib(struct tasdevice_priv *p) buf[2], buf[3]); } } else { - l = j * (cali_data->cali_dat_sz_per_dev + 1); + /* Calibrated data */ + l = j * (csz + 1); if (k >= p->ndev || l > oft * 4) { dev_err(p->dev, "%s: dev sum error\n", __func__); @@ -103,8 +132,7 @@ static void tas2781_apply_calib(struct tasdevice_priv *p)
data[l] = k; oft++; - for (i = 0; i < TASDEV_CALIB_N * 4; i++) - data[l + i + 1] = data[4 * oft + i]; + cali_cnv(data, 4 * oft, l); k++; } } @@ -127,12 +155,11 @@ static void tas2781_apply_calib(struct tasdevice_priv *p) dev_err(p->dev, "%s: V1 CRC error\n", __func__); return; } - + /* reverse rearrangement in case of overlap */ for (j = p->ndev - 1; j >= 0; j--) { - l = j * (cali_data->cali_dat_sz_per_dev + 1); - for (i = TASDEV_CALIB_N * 4; i > 0 ; i--) - data[l + i] = data[p->index * 5 + i]; - data[l+i] = j; + l = j * (csz + 1); + cali_cnv(data, csz * j, l); + data[l] = j; } }

On Tue, 02 Sep 2025 13:31:55 +0200, Shenghao Ding wrote:
A bug reported by one of my customers that the order of TAS2781 calibrated-data is incorrect, the correct way is to move R0_Low_%d and insert it between R0_%d and InvR0_%d.
Fixes: 4fe238513407 ("ALSA: hda/tas2781: Move and unified the calibrated-data getting function for SPI and I2C into the tas2781_hda lib") Signed-off-by: Shenghao Ding shenghao-ding@ti.com
v1:
- Add varialbe csz to store cali_data->cali_dat_sz_per_dev, which is long enough to need two lines.
- Add more comments on calibrated-data order
Is this still needed for the latest for-linus branch of sound.git tree? There have been a couple of fixes for TAS2781 already.
thanks,
Takashi
participants (2)
-
Shenghao Ding
-
Takashi Iwai