| Message ID | 1421081776-2350-1-git-send-email-andre.przywara@arm.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On 01/12/2015 10:56 AM, Andre Przywara wrote: > Currently the trace printk talks about "wfi" only, though the trace > point triggers both on wfi and wfe traps. > Add a parameter to differentiate between the two. > > Signed-off-by: Andre Przywara <andre.przywara@arm.com> Given that this patch is related to arm64 counter-part: Reviewed-by: Wei Huang <wei@redhat.com> -Wei > --- > Hi, > > this replaces the patch I sent earlier and now just contains the ARM > parts of it. The arm64 part is now merged into Wei's patch (which > isn't a prerequisite for this one anymore). > > Cheers, > Andre. > > arch/arm/kvm/handle_exit.c | 8 +++++--- > arch/arm/kvm/trace.h | 11 +++++++---- > 2 files changed, 12 insertions(+), 7 deletions(-) > > diff --git a/arch/arm/kvm/handle_exit.c b/arch/arm/kvm/handle_exit.c > index a96a804..95f12b2 100644 > --- a/arch/arm/kvm/handle_exit.c > +++ b/arch/arm/kvm/handle_exit.c > @@ -87,11 +87,13 @@ static int handle_dabt_hyp(struct kvm_vcpu *vcpu, struct kvm_run *run) > */ > static int kvm_handle_wfx(struct kvm_vcpu *vcpu, struct kvm_run *run) > { > - trace_kvm_wfi(*vcpu_pc(vcpu)); > - if (kvm_vcpu_get_hsr(vcpu) & HSR_WFI_IS_WFE) > + if (kvm_vcpu_get_hsr(vcpu) & HSR_WFI_IS_WFE) { > + trace_kvm_wfx(*vcpu_pc(vcpu), true); > kvm_vcpu_on_spin(vcpu); > - else > + } else { > + trace_kvm_wfx(*vcpu_pc(vcpu), false); > kvm_vcpu_block(vcpu); > + } > > kvm_skip_instr(vcpu, kvm_vcpu_trap_il_is32bit(vcpu)); > > diff --git a/arch/arm/kvm/trace.h b/arch/arm/kvm/trace.h > index b1d640f..f741449 100644 > --- a/arch/arm/kvm/trace.h > +++ b/arch/arm/kvm/trace.h > @@ -140,19 +140,22 @@ TRACE_EVENT(kvm_emulate_cp15_imp, > __entry->CRm, __entry->Op2) > ); > > -TRACE_EVENT(kvm_wfi, > - TP_PROTO(unsigned long vcpu_pc), > - TP_ARGS(vcpu_pc), > +TRACE_EVENT(kvm_wfx, > + TP_PROTO(unsigned long vcpu_pc, bool is_wfe), > + TP_ARGS(vcpu_pc, is_wfe), > > TP_STRUCT__entry( > __field( unsigned long, vcpu_pc ) > + __field( bool, is_wfe ) > ), > > TP_fast_assign( > __entry->vcpu_pc = vcpu_pc; > + __entry->is_wfe = is_wfe; > ), > > - TP_printk("guest executed wfi at: 0x%08lx", __entry->vcpu_pc) > + TP_printk("guest executed wf%c at: 0x%08lx", > + __entry->is_wfe ? 'e' : 'i', __entry->vcpu_pc) > ); > > TRACE_EVENT(kvm_unmap_hva, >
On Mon, Jan 12, 2015 at 04:56:16PM +0000, Andre Przywara wrote: > Currently the trace printk talks about "wfi" only, though the trace > point triggers both on wfi and wfe traps. > Add a parameter to differentiate between the two. > > Signed-off-by: Andre Przywara <andre.przywara@arm.com> > --- Thanks, applied! -Christoffer
diff --git a/arch/arm/kvm/handle_exit.c b/arch/arm/kvm/handle_exit.c index a96a804..95f12b2 100644 --- a/arch/arm/kvm/handle_exit.c +++ b/arch/arm/kvm/handle_exit.c @@ -87,11 +87,13 @@ static int handle_dabt_hyp(struct kvm_vcpu *vcpu, struct kvm_run *run) */ static int kvm_handle_wfx(struct kvm_vcpu *vcpu, struct kvm_run *run) { - trace_kvm_wfi(*vcpu_pc(vcpu)); - if (kvm_vcpu_get_hsr(vcpu) & HSR_WFI_IS_WFE) + if (kvm_vcpu_get_hsr(vcpu) & HSR_WFI_IS_WFE) { + trace_kvm_wfx(*vcpu_pc(vcpu), true); kvm_vcpu_on_spin(vcpu); - else + } else { + trace_kvm_wfx(*vcpu_pc(vcpu), false); kvm_vcpu_block(vcpu); + } kvm_skip_instr(vcpu, kvm_vcpu_trap_il_is32bit(vcpu)); diff --git a/arch/arm/kvm/trace.h b/arch/arm/kvm/trace.h index b1d640f..f741449 100644 --- a/arch/arm/kvm/trace.h +++ b/arch/arm/kvm/trace.h @@ -140,19 +140,22 @@ TRACE_EVENT(kvm_emulate_cp15_imp, __entry->CRm, __entry->Op2) ); -TRACE_EVENT(kvm_wfi, - TP_PROTO(unsigned long vcpu_pc), - TP_ARGS(vcpu_pc), +TRACE_EVENT(kvm_wfx, + TP_PROTO(unsigned long vcpu_pc, bool is_wfe), + TP_ARGS(vcpu_pc, is_wfe), TP_STRUCT__entry( __field( unsigned long, vcpu_pc ) + __field( bool, is_wfe ) ), TP_fast_assign( __entry->vcpu_pc = vcpu_pc; + __entry->is_wfe = is_wfe; ), - TP_printk("guest executed wfi at: 0x%08lx", __entry->vcpu_pc) + TP_printk("guest executed wf%c at: 0x%08lx", + __entry->is_wfe ? 'e' : 'i', __entry->vcpu_pc) ); TRACE_EVENT(kvm_unmap_hva,
Currently the trace printk talks about "wfi" only, though the trace point triggers both on wfi and wfe traps. Add a parameter to differentiate between the two. Signed-off-by: Andre Przywara <andre.przywara@arm.com> --- Hi, this replaces the patch I sent earlier and now just contains the ARM parts of it. The arm64 part is now merged into Wei's patch (which isn't a prerequisite for this one anymore). Cheers, Andre. arch/arm/kvm/handle_exit.c | 8 +++++--- arch/arm/kvm/trace.h | 11 +++++++---- 2 files changed, 12 insertions(+), 7 deletions(-)