Hi Tom, On Fri, 13 Feb 2026 at 11:14, Tom Rini <[email protected]> wrote: > > On Fri, Feb 13, 2026 at 10:53:26AM -0700, Simon Glass wrote: > > Hi Eric, > > > > On Tue, 10 Feb 2026 at 07:38, Eric Schikschneit > > <[email protected]> wrote: > > > > > > Hello Simon, Tom, > > > > > > 1. Tom - > > > > And we should spell out that device tree can't set these any more in the > > > > commit message, along with spelling out why we can't do that anymore > > > > safely (why can't we?). Thanks. > > > I would be happy to move the code comment into the commit message: > > > /* NOTE: This breaks u-boot ability to set board options via * > > > * device tree, but guarantees specific known values.. * > > > * you wouldnt use an uninitialized variable would you?? */ > > > These settings could be set by device tree, but every fdtdec* function > > > must > > > set a valid value. If a user has something set differently than called > > > out in the > > > FSP_MR5 Integration Guide the board stability will be compromised. Many of > > > these settings are co-dependant so we would need much more robust error > > > checking. This task would not be trivial. By using the specific values > > > called out > > > by the Integration guide we allow users to begin from a documented known > > > good configuration. Maybe we could set all the values, and then check the > > > device tree to see if the user specified a different value? If the device > > > tree does > > > not call out a specific setting we would not want to just set it to zero > > > for example > > > as this can cause undefined behavior. > > > > The device tree is actually where these values are set for all x86 > > boards. If you change any value in a device tree you risk bricking any > > board. It should be considered an integral part of the build. If you > > change the C code you can also brick a board. > > > > So I don't like the idea of changing this to C, sorry. > > > > NAK from me sorry. > > Let us not be hasty here. It sounds like we're getting some feedback > from production users of this support finally and not just people like > you and I that had reflashed our Minnowboards or similar to prove it > worked. >
OK. Regards, Simon
