[PATCH 0/6] soundwire: qcom: fixes and support for v1.7
This patchset has 2 fixes around fifo handling adds support for v1.7.0 controller found in SM8450 SoCs. Also a patch to make reset optional for v1.6 as this is not required for sc8280xp.
Thanks, Srini
Srinivas Kandagatla (6): soundwire: qcom: reinit broadcast completion soundwire: qcom: check for outanding writes before doing a read soundwire: qcom: remove unused SWRM_SPECIAL_CMD_ID soundwire: qcom: make reset optional for v1.6 controller dt-bindings: soundwire: qcom: add v1.7.0 support soundwire: qcom: add support for v1.7 Soundwire Controller
.../bindings/soundwire/qcom,sdw.txt | 1 + drivers/soundwire/qcom.c | 34 ++++++++++++++++--- 2 files changed, 30 insertions(+), 5 deletions(-)
For some reason we never reinit the broadcast completion, there is a danger that broadcast commands could be treated as completed by driver from previous complete status. Fix this by reinitializing the completion before sending a broadcast command.
Fixes: ddea6cf7b619 ("soundwire: qcom: update register read/write routine") Signed-off-by: Srinivas Kandagatla srinivas.kandagatla@linaro.org --- drivers/soundwire/qcom.c | 3 +++ 1 file changed, 3 insertions(+)
diff --git a/drivers/soundwire/qcom.c b/drivers/soundwire/qcom.c index b33d5db494a5..8f1a1eb017a2 100644 --- a/drivers/soundwire/qcom.c +++ b/drivers/soundwire/qcom.c @@ -344,6 +344,9 @@ static int qcom_swrm_cmd_fifo_wr_cmd(struct qcom_swrm_ctrl *swrm, u8 cmd_data, if (swrm_wait_for_wr_fifo_avail(swrm)) return SDW_CMD_FAIL_OTHER;
+ if (cmd_id == SWR_BROADCAST_CMD_ID) + reinit_completion(&swrm->broadcast); + /* Its assumed that write is okay as we do not get any status back */ swrm->reg_write(swrm, SWRM_CMD_FIFO_WR_CMD, val);
Reading will increase the fifo count, so check for outstanding cmd wrt. write fifo depth to avoid overflow as read will also increase write fifo cnt.
Fixes: a661308c34de ("soundwire: qcom: wait for fifo space to be available before read/write") Signed-off-by: Srinivas Kandagatla srinivas.kandagatla@linaro.org --- drivers/soundwire/qcom.c | 6 ++++++ 1 file changed, 6 insertions(+)
diff --git a/drivers/soundwire/qcom.c b/drivers/soundwire/qcom.c index 8f1a1eb017a2..cee2b2223141 100644 --- a/drivers/soundwire/qcom.c +++ b/drivers/soundwire/qcom.c @@ -380,6 +380,12 @@ static int qcom_swrm_cmd_fifo_rd_cmd(struct qcom_swrm_ctrl *swrm,
val = swrm_get_packed_reg_val(&swrm->rcmd_id, len, dev_addr, reg_addr);
+ /* + * Check for outstanding cmd wrt. write fifo depth to avoid + * overflow as read will also increase write fifo cnt. + */ + swrm_wait_for_wr_fifo_avail(swrm); + /* wait for FIFO RD to complete to avoid overflow */ usleep_range(100, 105); swrm->reg_write(swrm, SWRM_CMD_FIFO_RD_CMD, val);
Signed-off-by: Srinivas Kandagatla srinivas.kandagatla@linaro.org --- drivers/soundwire/qcom.c | 1 - 1 file changed, 1 deletion(-)
diff --git a/drivers/soundwire/qcom.c b/drivers/soundwire/qcom.c index cee2b2223141..7bc8cd0e906b 100644 --- a/drivers/soundwire/qcom.c +++ b/drivers/soundwire/qcom.c @@ -104,7 +104,6 @@ #define SWRM_REG_VAL_PACK(data, dev, id, reg) \ ((reg) | ((id) << 16) | ((dev) << 20) | ((data) << 24))
-#define SWRM_SPECIAL_CMD_ID 0xF #define MAX_FREQ_NUM 1 #define TIMEOUT_MS 100 #define QCOM_SWRM_MAX_RD_LEN 0x1
On Some Qualcomm SOCs like sc8280xp which uses v1.6 soundwire controller reset is not mandatory, so make this an optional one.
Signed-off-by: Srinivas Kandagatla srinivas.kandagatla@linaro.org --- drivers/soundwire/qcom.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/soundwire/qcom.c b/drivers/soundwire/qcom.c index 7bc8cd0e906b..54d370f4b291 100644 --- a/drivers/soundwire/qcom.c +++ b/drivers/soundwire/qcom.c @@ -1330,8 +1330,8 @@ static int qcom_swrm_probe(struct platform_device *pdev) }
if (data->sw_clk_gate_required) { - ctrl->audio_cgcr = devm_reset_control_get_exclusive(dev, "swr_audio_cgcr"); - if (IS_ERR_OR_NULL(ctrl->audio_cgcr)) { + ctrl->audio_cgcr = devm_reset_control_get_optional_exclusive(dev, "swr_audio_cgcr"); + if (IS_ERR(ctrl->audio_cgcr)) { dev_err(dev, "Failed to get cgcr reset ctrl required for SW gating\n"); ret = PTR_ERR(ctrl->audio_cgcr); goto err_init;
Signed-off-by: Srinivas Kandagatla srinivas.kandagatla@linaro.org --- Documentation/devicetree/bindings/soundwire/qcom,sdw.txt | 1 + 1 file changed, 1 insertion(+)
diff --git a/Documentation/devicetree/bindings/soundwire/qcom,sdw.txt b/Documentation/devicetree/bindings/soundwire/qcom,sdw.txt index c85c25779e3f..e0faed8dceac 100644 --- a/Documentation/devicetree/bindings/soundwire/qcom,sdw.txt +++ b/Documentation/devicetree/bindings/soundwire/qcom,sdw.txt @@ -13,6 +13,7 @@ board specific bus parameters. "qcom,soundwire-v1.5.0" "qcom,soundwire-v1.5.1" "qcom,soundwire-v1.6.0" + "qcom,soundwire-v1.7.0" - reg: Usage: required Value type: <prop-encoded-array>
On 26/10/2022 07:02, Srinivas Kandagatla wrote:
Signed-off-by: Srinivas Kandagatla srinivas.kandagatla@linaro.org
Documentation/devicetree/bindings/soundwire/qcom,sdw.txt | 1 +
Acked-by: Krzysztof Kozlowski krzysztof.kozlowski@linaro.org
Best regards, Krzysztof
This patch add support for v1.7 SoundWire Controller which has support for Multi-EE (Execution Environment), resulting in a new register and extending field in BUS_CTRL register.
With these updates v1.7.0 is fully supported.
Signed-off-by: Srinivas Kandagatla srinivas.kandagatla@linaro.org --- drivers/soundwire/qcom.c | 20 ++++++++++++++++++-- 1 file changed, 18 insertions(+), 2 deletions(-)
diff --git a/drivers/soundwire/qcom.c b/drivers/soundwire/qcom.c index 54d370f4b291..335424870290 100644 --- a/drivers/soundwire/qcom.c +++ b/drivers/soundwire/qcom.c @@ -25,6 +25,8 @@
#define SWRM_COMP_SW_RESET 0x008 #define SWRM_COMP_STATUS 0x014 +#define SWRM_LINK_MANAGER_EE 0x018 +#define SWRM_EE_CPU 1 #define SWRM_FRM_GEN_ENABLED BIT(0) #define SWRM_COMP_HW_VERSION 0x00 #define SWRM_COMP_CFG_ADDR 0x04 @@ -693,7 +695,14 @@ static int qcom_swrm_init(struct qcom_swrm_ctrl *ctrl) u32p_replace_bits(&val, SWRM_DEF_CMD_NO_PINGS, SWRM_MCP_CFG_MAX_NUM_OF_CMD_NO_PINGS_BMSK); ctrl->reg_write(ctrl, SWRM_MCP_CFG_ADDR, val);
- ctrl->reg_write(ctrl, SWRM_MCP_BUS_CTRL, SWRM_MCP_BUS_CLK_START); + if (ctrl->version >= 0x01070000) { + ctrl->reg_write(ctrl, SWRM_LINK_MANAGER_EE, SWRM_EE_CPU); + ctrl->reg_write(ctrl, SWRM_MCP_BUS_CTRL, + SWRM_MCP_BUS_CLK_START << SWRM_EE_CPU); + } else { + ctrl->reg_write(ctrl, SWRM_MCP_BUS_CTRL, SWRM_MCP_BUS_CLK_START); + } + /* Configure number of retries of a read/write cmd */ if (ctrl->version > 0x01050001) { /* Only for versions >= 1.5.1 */ @@ -1518,7 +1527,13 @@ static int __maybe_unused swrm_runtime_resume(struct device *dev) } else { reset_control_reset(ctrl->audio_cgcr);
- ctrl->reg_write(ctrl, SWRM_MCP_BUS_CTRL, SWRM_MCP_BUS_CLK_START); + if (ctrl->version >= 0x01070000) { + ctrl->reg_write(ctrl, SWRM_LINK_MANAGER_EE, SWRM_EE_CPU); + ctrl->reg_write(ctrl, SWRM_MCP_BUS_CTRL, + SWRM_MCP_BUS_CLK_START << SWRM_EE_CPU); + } else { + ctrl->reg_write(ctrl, SWRM_MCP_BUS_CTRL, SWRM_MCP_BUS_CLK_START); + } ctrl->reg_write(ctrl, SWRM_INTERRUPT_CLEAR, SWRM_INTERRUPT_STATUS_MASTER_CLASH_DET);
@@ -1582,6 +1597,7 @@ static const struct of_device_id qcom_swrm_of_match[] = { { .compatible = "qcom,soundwire-v1.3.0", .data = &swrm_v1_3_data }, { .compatible = "qcom,soundwire-v1.5.1", .data = &swrm_v1_5_data }, { .compatible = "qcom,soundwire-v1.6.0", .data = &swrm_v1_6_data }, + { .compatible = "qcom,soundwire-v1.7.0", .data = &swrm_v1_5_data }, {/* sentinel */}, };
@@ -1582,6 +1597,7 @@ static const struct of_device_id qcom_swrm_of_match[] = { { .compatible = "qcom,soundwire-v1.3.0", .data = &swrm_v1_3_data }, { .compatible = "qcom,soundwire-v1.5.1", .data = &swrm_v1_5_data }, { .compatible = "qcom,soundwire-v1.6.0", .data = &swrm_v1_6_data },
- { .compatible = "qcom,soundwire-v1.7.0", .data = &swrm_v1_5_data },
is this line intentional, it looks odd that 1.7 is compatible with 1.5, but 1.6 isn't?
Hi Pierre,
On 26/10/2022 21:14, Pierre-Louis Bossart wrote:
@@ -1582,6 +1597,7 @@ static const struct of_device_id qcom_swrm_of_match[] = { { .compatible = "qcom,soundwire-v1.3.0", .data = &swrm_v1_3_data }, { .compatible = "qcom,soundwire-v1.5.1", .data = &swrm_v1_5_data }, { .compatible = "qcom,soundwire-v1.6.0", .data = &swrm_v1_6_data },
- { .compatible = "qcom,soundwire-v1.7.0", .data = &swrm_v1_5_data },
is this line intentional, it looks odd that 1.7 is compatible with 1.5, but 1.6 isn't?
This is more around the data rather than compatible, 1.6 data is marked with sw_clk_gate_required = true which is not the case with 1.7.
from 1.5 we use same frame shape info for 1.6, 1.7, except that 1.6 had this sw_clk_gate_required flag set.
--srini
@@ -1582,6 +1597,7 @@ static const struct of_device_id qcom_swrm_of_match[] = { { .compatible = "qcom,soundwire-v1.3.0", .data = &swrm_v1_3_data }, { .compatible = "qcom,soundwire-v1.5.1", .data = &swrm_v1_5_data }, { .compatible = "qcom,soundwire-v1.6.0", .data = &swrm_v1_6_data }, + { .compatible = "qcom,soundwire-v1.7.0", .data = &swrm_v1_5_data },
is this line intentional, it looks odd that 1.7 is compatible with 1.5, but 1.6 isn't?
This is more around the data rather than compatible, 1.6 data is marked with sw_clk_gate_required = true which is not the case with 1.7.
from 1.5 we use same frame shape info for 1.6, 1.7, except that 1.6 had this sw_clk_gate_required flag set.
probably worthy of a comment to show intent and explain the differences?
participants (4)
-
Krzysztof Kozlowski -
Pierre-Louis Bossart -
Srinivas Kandagatla -
Vinod Koul