Hi, Geert, On 09.12.2024 15:51, Geert Uytterhoeven wrote: > Inside this block there are several return statements. > As we know DMA is not available when we get here, these do not > need to call rz_ssi_release_dma_channels() hence do not use > "goto err_release_dma_chs". > However, this may be missed when making future changes. > So perhaps it may be prudent to make this safer, by moving this inside > the failure branch of the rz_ssi_dma_request() check above? I agree! As this series is already big enough I would prefer to handle it after it is integrated. Keeping it like this doesn't impact the RZ/G3S support. Are you OK with this approach? Thank you for your review, Claudiu
