On Wed, Apr 22, 2026 at 03:00:13AM +0000, Stanislav Kinsburskii wrote:
> The trace event was checking for HV_INTERCEPT_ACCESS_READ twice in the
> ternary operator chain instead of checking for EXECUTE, which would
> cause execute accesses to be incorrectly displayed as '?' in traces.
>
> Move the access type conversion to the TP_fast_assign section where
> the raw value is available, rather than performing the conversion
> during formatting. This eliminates the redundant checks and ensures
> all access types are correctly identified and displayed.
>
> Fixes: 03f7d01f699010 ("mshv: Add tracepoint for GPA intercept handling")
This is not yet in Linus' tree.
I folded this into the original patch.
Wei
> Signed-off-by: Stanislav Kinsburskii <[email protected]>
> ---
> drivers/hv/mshv_trace.h | 8 ++++----
> 1 file changed, 4 insertions(+), 4 deletions(-)
>
> diff --git a/drivers/hv/mshv_trace.h b/drivers/hv/mshv_trace.h
> index 6b8fa477fa3bf..e7280c47e579a 100644
> --- a/drivers/hv/mshv_trace.h
> +++ b/drivers/hv/mshv_trace.h
> @@ -524,16 +524,16 @@ TRACE_EVENT(mshv_handle_gpa_intercept,
> __entry->partition_id = partition_id;
> __entry->vp_index = vp_index;
> __entry->gfn = gfn;
> - __entry->access_type = access_type;
> + __entry->access_type = access_type ==
> HV_INTERCEPT_ACCESS_READ ? 'R' :
> + (access_type ==
> HV_INTERCEPT_ACCESS_WRITE ? 'W' :
> + (access_type ==
> HV_INTERCEPT_ACCESS_EXECUTE ? 'X' : '?'));
> __entry->handled = handled;
> ),
> TP_printk("partition_id=%llu vp_index=%u gfn=0x%llx access_type=%c
> handled=%d",
> __entry->partition_id,
> __entry->vp_index,
> __entry->gfn,
> - __entry->access_type == HV_INTERCEPT_ACCESS_READ ? 'R' :
> - (__entry->access_type ==
> HV_INTERCEPT_ACCESS_WRITE ? 'W' :
> - (__entry->access_type ==
> HV_INTERCEPT_ACCESS_READ ? 'X' : '?')),
> + __entry->access_type,
> __entry->handled
> )
> );
>
>
