On Wed, May 31, 2023 at 1:51 AM Wolfram Sang <wsa+renesas@xxxxxxxxxxxxxxxxxxxx> wrote: > > Hi Adam, > > I am very sorry that it took so long until I am now able to respond. > First, too many other duties with higher priority came in. Second, the > card didn't fail 100% for me after all which made debugging still > challenging. But it failed often enough, so I at least have a pointer > now. I totally understand that it's an open source effort, and we have have duties to perform. I do appreciate the help. > > > From what I have been able to determine, the U-Boot code explicitly > > disables DMA during the tuning phase: > > ... > > > However, Linux does not disable the DMA, and the value of taps is zero > > and the tuning ultimately fails. > > I disabled DMA while tuning but it did not help. It finally turns out > that adding a delay after each tuning command seems to help. U-Boot does > this but Marek (the driver author for U-Boot) cannot recall why he added > the delay. I'll send a patch in some minutes. I'll give them a try today. adam > > All the best, > > Wolfram >
