> > + u32 intr_info = vmcs12->vm_entry_intr_info_field;
> > + u8 nr = intr_info & INTR_INFO_VECTOR_MASK;
> Nit: I would rename 'nr' to 'vector' for better readability of the code.
SGTM, will do. I used nr to be consistent with other parts of the
code. That said, I agree that vector is more explicit and clear.
> > + /* VM-entry interruption-info field: deliver error code */
> > + should_have_error_code =
> > + intr_type == INTR_TYPE_HARD_EXCEPTION &&
> > + x86_exception_has_error_code(nr, prot_mode);
> It's better to leave 'prot_mode' outside of the function as follows:
>
> should_have_error_code =
> intr_type == INTR_TYPE_HARD_EXCEPTION && prot_mode &&
> x86_exception_has_error_code(nr);
I think we should leave the helper as is because an exception can only
have an error code in protected mode.
> > + /* VM-entry exception error code */
> > + if (has_error_code &&
> > + vmcs12->vm_entry_exception_error_code & GENMASK(31, 15))
> > + return VMXERR_ENTRY_INVALID_CONTROL_FIELD;
> I would move this check above where 'delivery error code' is because
> this check is related to that.
Sure, will do.
> > +static inline bool x86_exception_has_error_code(unsigned int nr,
> > + bool protected_mode)
> > +{
> > + static u32 exception_has_error_code = BIT(DF_VECTOR) | BIT(TS_VECTOR) |
> > + BIT(NP_VECTOR) | BIT(SS_VECTOR) | BIT(GP_VECTOR) |
> > + BIT(PF_VECTOR) | BIT(AC_VECTOR);
> > +
> > + return protected_mode && ((1U << nr) & exception_has_error_code);
> Should it be (nr & exception_has_error_code) instead of ((1U << nr) &
> exception_has_error_code) ?
No, nr is a vector's absolute number, as defined in
arch/x86/include/uapi/asm/kvm.h. For example, '#define GP_VECTOR 13'.
