On 03/09/2026 1:16 pm, Bertrand Marquis wrote: > Hi Andrew, > >> On 31 Aug 2026, at 14:19, Andrew Cooper <[email protected]> wrote: >> >> PSCI v1.0 says that x4 thru x17 may be clobbered. PSCI v1.1 says they are >> strictly preserved unless they contain return values. > I think the PSCI reference here and in the rest of the commit message and the > patch > should not be used and should be SMCCC references instead (which would also be > more coherent with the fact that this code is not PSCI but SMCCC and is used > by > others than PSCI). > > PSCI does not define the preservation rules, this is defined in SMCCC spec > and we > have a way to detect the SMCCC version. > > SMCCC defines the following rules: > - 1.0: x4-x17 have an unpredictable return state. > - 1.1: x4-x17 must be preserved. > - 1.2: x4-x17 may additionally be defined as result registers, otherwise they > remain preserved. > > So the commit message and comments should consistently say SMCCC instead of > PSCI. > > Other than this the code is right i think.
My apologies, I've clearly got confused somewhere. (This is the first time I've ever looked at the SMCCC spec.) I'll adjust. ~Andrew
