On Wed, May 23, 2018 at 09:42:56AM +0100, Suzuki K Poulose wrote:
> On 03/05/18 14:20, Mark Rutland wrote:
> > +#define __ptrauth_key_install(k, v) \
> > +do { \
> > + write_sysreg_s(v.lo, SYS_ ## k ## KEYLO_EL1); \
> > + write_sysreg_s(v.hi, SYS_ ## k ## KEYHI_EL1); \
> > +} while (0)
>
> I think it might be safer to have parentheses around v, to prevent
> something like __ptrauth_key_install(APIA, *key_val) work fine.
In case v is ever an expression with side-effects, I've made this:
#define __ptrauth_key_install(k, v) \
do { \
struct ptrauth_key __pki_v = (v); \
write_sysreg_s(__pki_v.lo, SYS_ ## k ## KEYLO_EL1); \
write_sysreg_s(__pki_v.hi, SYS_ ## k ## KEYHI_EL1); \
} while (0)
... though I could just move the raw sysreg accesses into
ptrauth_keys_switch() for now.
[...]
> > +#ifdef CONFIG_ARM64_PNTR_AUTH
> > + HWCAP_CAP(SYS_ID_AA64ISAR1_EL1, ID_AA64ISAR1_APA_SHIFT, FTR_UNSIGNED,
> > 1, CAP_HWCAP, HWCAP_APIA),
> > +#endif
>
> Did you mean CONFIG_ARM64_PTR_AUTH here ?
Yes; fixed now.
Thanks,
Mark.
