Note that if we remove rte_eth_dev_get_reg_length() then it will break all of the drivers that implement it. Shall I remove it all leave it and modify only ethtool to use rte_eth_dev_get_regs() to get reg size? In the end the drivers will have to implement the part of setting the size in reg_info struct. rte_eth_dev_get_regs() itself wouldn't change at all. Or do you have different opinion?
On 21.06.2016 11:55, Zyta Szpak wrote: > OK, I will do the v4. > > On 17.06.2016 12:20, Thomas Monjalon wrote: >> 2016-06-13 16:51, Remy Horton: >>> On 12/06/2016 15:51, Zyta Szpak wrote: >>>> I would prefer having only one function rte_eth_dev_get_regs() >>>> which returns length and width if data is NULL. >>>> The first call is a parameter request before buffer allocation, >>>> and the second call fills the buffer. >>>> >>>> We can deprecate the old API and introduce this new one. >>>> >>>> Opinions? >>>> >>>> In my opinion as it is now it works fine. Gathering all parameters in >>>> one callback might be a good idea if the maintainer also agrees to >>>> that >>>> because as I mentioned, it interferes. >>> From my perspective changing rte_eth_dev_get_regs() isn't a >>> problem, as >>> it isn't used directly rather than through rte_ethtool_get_regs().. >> Zyta, would you like to make a v4? >