[PATCH 0/1 v2] KVM: nVMX: nSVM: Show guest mode in kvm_{entry,exit} tracepoints

[PATCH 0/1 v2] KVM: nVMX: nSVM: Show guest mode in kvm_{entry,exit} tracepoints

[Krish Sadhukhan]

v1 -> v2:
	1. The tracepoints will print "nested" if the VCPU is running
	   a nested guest.
	2. The commit message has been modified to reflect the change.


[PATCH v2 1/1] KVM: nVMX: nSVM: Show guest mode in kvm_{entry,exit}

 arch/x86/kvm/trace.h | 16 +++++++++++-----
 1 file changed, 11 insertions(+), 5 deletions(-)

Krish Sadhukhan (1):
      KVM: nVMX: nSVM: Show guest mode in kvm_{entry,exit} tracepoints

[PATCH v2 1/1] KVM: nVMX: nSVM: Show guest mode in kvm_{entry,exit} tracepoints

[Krish Sadhukhan]

From debugging point of view, KVM entry and exit tracepoints are important
in that they indicate when a guest starts and exits. When L1 runs L2, there
is no way we can determine from KVM tracing when L1 starts/exits and when
L2 starts/exits as there is no marker in place today in those tracepoints.
Therefore, showing guest mode in the entry and exit tracepoints
will make debugging much easier.
With this patch KVM entry and exit tracepoints will show "nested" if the
vCPU is running a nested guest.

Signed-off-by: Krish Sdhukhan <krish.sadhukhan@oracle.com>
---
 arch/x86/kvm/trace.h | 16 +++++++++++-----
 1 file changed, 11 insertions(+), 5 deletions(-)

diff --git a/arch/x86/kvm/trace.h b/arch/x86/kvm/trace.h
index 03ebe368333e..091415870ae2 100644
--- a/arch/x86/kvm/trace.h
+++ b/arch/x86/kvm/trace.h
@@ -21,14 +21,17 @@ TRACE_EVENT(kvm_entry,
 	TP_STRUCT__entry(
 		__field(	unsigned int,	vcpu_id		)
 		__field(	unsigned long,	rip		)
+		__field(        bool,           nested		)
 	),
 
 	TP_fast_assign(
 		__entry->vcpu_id        = vcpu->vcpu_id;
 		__entry->rip		= kvm_rip_read(vcpu);
+		__entry->nested		= is_guest_mode(vcpu);
 	),
 
-	TP_printk("vcpu %u, rip 0x%lx", __entry->vcpu_id, __entry->rip)
+	TP_printk("vcpu %u, rip 0x%lx%s", __entry->vcpu_id,
+		  __entry->rip, __entry->nested ? ", (nested)" : "")
 );
 
 /*
@@ -300,6 +303,7 @@ TRACE_EVENT(name,							     \
 		__field(	u32,	        intr_info	)	     \
 		__field(	u32,	        error_code	)	     \
 		__field(	unsigned int,	vcpu_id         )	     \
+		__field(        bool,           nested		)	     \
 	),								     \
 									     \
 	TP_fast_assign(							     \
@@ -310,15 +314,17 @@ TRACE_EVENT(name,							     \
 		static_call(kvm_x86_get_exit_info)(vcpu, &__entry->info1,    \
 					  &__entry->info2,		     \
 					  &__entry->intr_info,		     \
-					  &__entry->error_code);	     \
+					  &__entry->error_code),	     \
+		  __entry->nested	= is_guest_mode(vcpu);		     \
 	),								     \
 									     \
 	TP_printk("vcpu %u reason %s%s%s rip 0x%lx info1 0x%016llx "	     \
-		  "info2 0x%016llx intr_info 0x%08x error_code 0x%08x",	     \
-		  __entry->vcpu_id,					     \
+		  "info2 0x%016llx intr_info 0x%08x error_code 0x%08x "	     \
+		  "%s", __entry->vcpu_id,				     \
 		  kvm_print_exit_reason(__entry->exit_reason, __entry->isa), \
 		  __entry->guest_rip, __entry->info1, __entry->info2,	     \
-		  __entry->intr_info, __entry->error_code)		     \
+		  __entry->intr_info, __entry->error_code,                   \
+		  __entry->nested ? "(nested)" : "")      		     \
 )
 
 /*
-- 
2.18.4