| Message ID | CA+RD57MwUfAzF2u31Ews4uT0+A6uhjwwd40O=9uyZihap4nxdA@mail.gmail.com (mailing list archive) |
|---|---|
| State | Changes Requested |
| Delegated to: | Kalle Valo |
| Headers | show |
| Series | [v2] wifi: ath10k: read qcom,coexist-support as a u32 | expand |
Vincent Tremblay <vincent@vtremblay.dev> writes: > Read qcom,coexist-support as a u32 instead of a u8 > > When we set the property to <1> in the DT (as specified in the doc), > "of_property_read_u8" read 0 instead of 1. This is because of the data format. > > By default <1> is written with 32 bits. > The problem is that the driver is trying to read a u8. > > The difference can be visualized using hexdump in a running device: > Default 32 bits output: > ======================= > 0000000 0000 0100 > 0000004 > > 8 bits output: > ============== > 0000000 0001 > 0000001 > > By changing "of_property_read_u8" by "of_property_read_u32", the driver > is aligned with the documentation and is able to read the value without > modifying the DT. > > The other solution would be to force the value in the DT to be saved as > an 8 bits value (qcom,coexist-support = /bits/ 8 <1>), > which is against the doc and less intuitive. > > Validation: > =========== > The patch was tested on a real device and we can see in the debug logs > that the feature is properly initialized: > > [ 109.102097] ath10k_ahb a000000.wifi: boot coex_support 1 coex_gpio_pin 52 > > Signed-off-by: Vincent Tremblay <vincent@vtremblay.dev> What are the changes in v2? Please always include a list of changes, more info in the wiki below.
Vincent Tremblay <vincent@vtremblay.dev> wrote: > Read qcom,coexist-support as a u32 instead of a u8 > > When we set the property to <1> in the DT (as specified in the doc), > "of_property_read_u8" read 0 instead of 1. This is because of the data format. > > By default <1> is written with 32 bits. > The problem is that the driver is trying to read a u8. > > The difference can be visualized using hexdump in a running device: > Default 32 bits output: > ======================= > 0000000 0000 0100 > 0000004 > > 8 bits output: > ============== > 0000000 0001 > 0000001 > > By changing "of_property_read_u8" by "of_property_read_u32", the driver > is aligned with the documentation and is able to read the value without > modifying the DT. > > The other solution would be to force the value in the DT to be saved as > an 8 bits value (qcom,coexist-support = /bits/ 8 <1>), > which is against the doc and less intuitive. > > Validation: > =========== > The patch was tested on a real device and we can see in the debug logs > that the feature is properly initialized: > > [ 109.102097] ath10k_ahb a000000.wifi: boot coex_support 1 coex_gpio_pin 52 > > Signed-off-by: Vincent Tremblay <vincent@vtremblay.dev> Fails to apply: error: git diff header lacks filename information when removing 1 leading pathname component (line 6) stg import: Diff does not apply cleanly Patch set to Changes Requested.
diff --git a/drivers/net/wireless/ath/ath10k/core.c b/drivers/net/wireless/ath/ath10k/core.c index 5eb131ab916f..802d6a12dc1e 100644 --- a/drivers/net/wireless/ath/ath10k/core.c +++ b/drivers/net/wireless/ath/ath10k/core.c @@ -2355,14 +2355,14 @@ static int ath10k_download_cal_data(struct ath10k *ar) static void ath10k_core_fetch_btcoex_dt(struct ath10k *ar) { struct device_node *node; - u8 coex_support = 0; + u32 coex_support = 0; int ret; node = ar->dev->of_node; if (!node) goto out; - ret = of_property_read_u8(node, "qcom,coexist-support", &coex_support); + ret = of_property_read_u32(node, "qcom,coexist-support", &coex_support); if (ret) { ar->coex_support = true; goto out;
Read qcom,coexist-support as a u32 instead of a u8 When we set the property to <1> in the DT (as specified in the doc), "of_property_read_u8" read 0 instead of 1. This is because of the data format. By default <1> is written with 32 bits. The problem is that the driver is trying to read a u8. The difference can be visualized using hexdump in a running device: Default 32 bits output: ======================= 0000000 0000 0100 0000004 8 bits output: ============== 0000000 0001 0000001 By changing "of_property_read_u8" by "of_property_read_u32", the driver is aligned with the documentation and is able to read the value without modifying the DT. The other solution would be to force the value in the DT to be saved as an 8 bits value (qcom,coexist-support = /bits/ 8 <1>), which is against the doc and less intuitive. Validation: =========== The patch was tested on a real device and we can see in the debug logs that the feature is properly initialized: [ 109.102097] ath10k_ahb a000000.wifi: boot coex_support 1 coex_gpio_pin 52 Signed-off-by: Vincent Tremblay <vincent@vtremblay.dev> --- drivers/net/wireless/ath/ath10k/core.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)