Hi Niklas, On Di, 2022-03-22 at 10:32 +0000, Niklas Cassel wrote: [...] > https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c?h=v5.17#n7035 > Thank you. [...] I see your point. Many drivers should probably change assert() + udelay(x) + deassert() with a reset_control_reset(), since .reset() implementation should have the correct delay for each SoC. ..but I guess often the manual for some hardware block states the how long reset has to be held. So for that to work the delay in .reset() has to be greater equal the longest reset time needed for all hardware in that SoC? Yes, exactly. The reset-simple driver was prepared for this in commit a9701376ed0f ("reset: simple: Add reset callback"), for example. Several of the custom reset drivers also implement .reset() with a fixed delay. regards Philipp