Anilkumar Kolli <akolli@xxxxxxxxxxxxxx> writes:
> IPQ6018 has one 5G and one 2G radio with 2x2,
> shares ipq8074 configurations.
>
> Signed-off-by: Anilkumar Kolli <akolli@xxxxxxxxxxxxxx>
[...]
> @@ -890,9 +995,15 @@ static int ath11k_ahb_probe(struct platform_device *pdev)
> goto err_hal_srng_deinit;
> }
>
> - ab->hw_params.svc_to_ce_map_len =
> + if (ab->hw_rev == ATH11K_HW_IPQ8074) {
> + ab->hw_params.svc_to_ce_map_len =
> ARRAY_SIZE(target_service_to_ce_map_wlan_ipq8074);
> - ab->hw_params.svc_to_ce_map = target_service_to_ce_map_wlan_ipq8074;
> + ab->hw_params.svc_to_ce_map = target_service_to_ce_map_wlan_ipq8074;
> + } else if (ab->hw_rev == ATH11K_HW_IPQ6018) {
> + ab->hw_params.svc_to_ce_map_len =
> + ARRAY_SIZE(target_service_to_ce_map_wlan_ipq6018);
> + ab->hw_params.svc_to_ce_map = target_service_to_ce_map_wlan_ipq6018;
> + }
The idea of having ab->hw_params is to avoid the need for hw_rev
sprinkled all over the code, this here defeats the purpose. I recommend
checking if my proposal from an earlier version is implementable:
"You could copy the map arrays to hw.c, as an example see how
ath11k_hw_ring_mask_ipq8074 is implemented. That way we don't need any
new hw_rev checks."
--
https://wireless.wiki.kernel.org/en/developers/documentation/submittingpatches
