On Mon, Nov 17, 2025 at 05:37:49PM +0100, Maciej Fijalkowski wrote: > This revision is much more clear to me. Only thing that might be bothering > someone is doubled i40e_rx_bi() call in i40e_get_rx_buffer(). Not sure if > we can do about it though as we need to use ntp from before potential > increment. > > ...maybe pass rx_buffer to i40e_get_rx_buffer() ?
Surely the compiler isn't going to actually reload here, but yeah not great code wise. How about I pass it the buffer and rename to i40e_prepare_rx_buffer to better match what's happening now?
