Hi Mark, Oliver, On 10/18/23 14:16, Mark Brown wrote: > On Wed, Oct 18, 2023 at 01:57:31PM +0200, Eric Auger wrote: >> On 10/11/23 21:57, Oliver Upton wrote: > >>> #define set_pstate_pan(x) asm volatile(SET_PSTATE_PAN(x)) >>> #define set_pstate_uao(x) asm volatile(SET_PSTATE_UAO(x)) >>> #define set_pstate_ssbs(x) asm volatile(SET_PSTATE_SSBS(x)) >>> +#define set_pstate_dit(x) asm volatile(SET_PSTATE_DIT(x)) > >> could you comment on the *DIT* addictions, what is it for? > > DIT is data independent timing, this tells the processor to ensure that > instructions take a constant time regardless of the data they are > handling. > > Note that this file is just a copy of arch/arm64/include/asm/gpr-num.h, > the main purpose here is to sync with the original. Ah thanks. that's helpful for me to understand where this gpr-num.h comes from. This could be documented in the commit msg though. Something like: adding tools/arch/arm64/include/asm/gpr-num.h matching linux arch/arm64/include/asm/gpr-num.h and syncing tools/arch/arm64/include/asm/sysreg.h with the fellow header in the linux tree. tbh I did not initially understand that all this diffstat was aimed to match the linux arch/arm64/include/asm/sysreg.h. Now diffing both I have some diffs. Doesn't it need a refresh? Thanks Eric > >>> +/* >>> + * Automatically generated definitions for system registers, the >>> + * manual encodings below are in the process of being converted to >>> + * come from here. The header relies on the definition of sys_reg() >>> + * earlier in this file. >>> + */ >>> +#include "asm/sysreg-defs.h" > >> strange to have this include in the middle of the file > > It relies on defines from earlier in the header.
