On 20.11.2025 11:47, Grygorii Strashko wrote: > Hi Jan, > > On 20.11.25 10:56, Jan Beulich wrote: >> On 19.11.2025 20:31, Grygorii Strashko wrote: >>> From: Grygorii Strashko <[email protected]> >>> >>> The VMTRACE feature is depends on Platform/Arch HW and code support and now >>> can be used only on x86 HVM with Intel VT-x (INTEL_VMX) enabled. >>> This makes VMTRACE support optional by introducing HVM Kconfig option: >>> - CONFIG_VMTRACE to enable/disable the feature. >>> >>> Signed-off-by: Grygorii Strashko <[email protected]> >>> --- >>> changes in v4: >>> - format changes >> >> Just partly, ... >> >>> @@ -772,13 +775,24 @@ static inline int hvm_vmtrace_get_option( >>> >>> return -EOPNOTSUPP; >>> } >>> +#else >>> +/* >>> + * Function declaration(s) here are used without definition(s) to make >>> compiler >>> + * happy when VMTRACE=n, compiler DCE will eliminate unused code. >>> + */ >>> +int hvm_vmtrace_output_position(struct vcpu *v, uint64_t *pos); >>> +#endif >>> >>> static inline int hvm_vmtrace_reset(struct vcpu *v) >>> { >>> +#ifdef CONFIG_VMTRACE >>> if ( hvm_funcs.vmtrace_reset ) >>> return alternative_call(hvm_funcs.vmtrace_reset, v); >>> >>> return -EOPNOTSUPP; >>> +#else >>> + return -EOPNOTSUPP; >>> +#endif >>> } >> >> ... the #else wasn't dropped from here. If no need for a v5 arises, I guess >> I can make the adjustment while committing. But it would be really nice if >> all earlier comments were taken care of (one way or another) when submitting >> a new version. > > This is absolutely my miss. Sorry. > I can resend with fix. Just tell me how will be better for you.
Well, as said above - no need to resend unless another reason for a v5 surfaces. Jan
