Re: [PATCH] trace: adjust code layout in get_recursion_context

[Jesper Dangaard Brouer]

On Tue, 22 Aug 2017 19:00:39 +0200
Jesper Dangaard Brouer <brouer@xxxxxxxxxx> wrote:

> On Tue, 22 Aug 2017 17:20:25 +0200
> Peter Zijlstra <peterz@xxxxxxxxxxxxx> wrote:
> 
> > On Tue, Aug 22, 2017 at 05:14:10PM +0200, Peter Zijlstra wrote:  
> > > On Tue, Aug 22, 2017 at 04:40:24PM +0200, Jesper Dangaard Brouer wrote:    
> > > > In an XDP redirect applications using tracepoint xdp:xdp_redirect to
> > > > diagnose TX overrun, I noticed perf_swevent_get_recursion_context()
> > > > was consuming 2% CPU. This was reduced to 1.6% with this simple
> > > > change.    
> > > 
> > > It is also incorrect. What do you suppose it now returns when the NMI
> > > hits a hard IRQ which hit during a Soft IRQ?    
> > 
> > Does this help any? I can imagine the compiler could struggle to CSE
> > preempt_count() seeing how its an asm thing.  
> 
> Nope, it does not help (see assembly below, with perf percentages).
> 
> But I think I can achieve that I want by a simple unlikely(in_nmi()) annotation.

Like:

diff --git a/kernel/events/internal.h b/kernel/events/internal.h
index 486fd78eb8d5..e1a7ac7bd686 100644
--- a/kernel/events/internal.h
+++ b/kernel/events/internal.h
@@ -208,7 +208,7 @@ static inline int get_recursion_context(int *recursion)
 {
        int rctx;
 
-       if (in_nmi())
+       if (unlikely(in_nmi()))
                rctx = 3;
        else if (in_irq())
                rctx = 2;

Testing this show I get the expected result. Although, the 2% is
reduced to 1.85% (and not 1.6% as before).  


perf_swevent_get_recursion_context  /proc/kcore
       â
       â    Disassembly of section load0:
       â
       â    ffffffff811465c0 <load0>:
  4.94 â      push   %rbp
  2.56 â      mov    $0x14d20,%rax
 14.81 â      mov    %rsp,%rbp
  3.47 â      add    %gs:0x7eec3b5d(%rip),%rax
  0.91 â      lea    0x34(%rax),%rdx
  1.46 â      mov    %gs:0x7eec5db2(%rip),%eax
  8.04 â      test   $0x100000,%eax
       â    â jne    59
  3.11 â      test   $0xf0000,%eax
       â    â jne    4d
  0.37 â      test   $0xff,%ah
  1.83 â      setne  %cl
  9.87 â      movzbl %cl,%eax
  2.01 â      movzbl %cl,%ecx
  1.65 â      shl    $0x2,%rcx
  4.39 â3c:   add    %rcx,%rdx
 29.62 â      mov    (%rdx),%ecx
  2.93 â      test   %ecx,%ecx
       â    â jne    65
  0.55 â      movl   $0x1,(%rdx)
  2.56 â      pop    %rbp
  4.94 â    â retq
       â4d:   mov    $0x8,%ecx
       â      mov    $0x2,%eax
       â    â jmp    3c
       â59:   mov    $0xc,%ecx
       â      mov    $0x3,%eax
       â    â jmp    3c
       â65:   mov    $0xffffffff,%eax
       â      pop    %rbp
       â    â retq



> > ---
> >  kernel/events/internal.h | 7 ++++---
> >  1 file changed, 4 insertions(+), 3 deletions(-)
> > 
> > diff --git a/kernel/events/internal.h b/kernel/events/internal.h
> > index 486fd78eb8d5..e0b5b8fa83a2 100644
> > --- a/kernel/events/internal.h
> > +++ b/kernel/events/internal.h
> > @@ -206,13 +206,14 @@ perf_callchain(struct perf_event *event, struct pt_regs *regs);
> >  
> >  static inline int get_recursion_context(int *recursion)
> >  {
> > +	unsigned int pc = preempt_count();
> >  	int rctx;
> >  
> > -	if (in_nmi())
> > +	if (pc & NMI_MASK)
> >  		rctx = 3;
> > -	else if (in_irq())
> > +	else if (pc & HARDIRQ_MASK)
> >  		rctx = 2;
> > -	else if (in_softirq())
> > +	else if (pc & SOFTIRQ_OFFSET)  
> 
> Hmmm... shouldn't this be SOFTIRQ_MASK?
> 
> >  		rctx = 1;
> >  	else
> >  		rctx = 0;  
> 
> perf_swevent_get_recursion_context  /proc/kcore
>        â
>        â
>        â    Disassembly of section load0:
>        â
>        â    ffffffff811465c0 <load0>:
>  13.32 â      push   %rbp
>   1.43 â      mov    $0x14d20,%rax
>   5.12 â      mov    %rsp,%rbp
>   6.56 â      add    %gs:0x7eec3b5d(%rip),%rax
>   0.72 â      lea    0x34(%rax),%rdx
>   0.31 â      mov    %gs:0x7eec5db2(%rip),%eax
>   2.46 â      mov    %eax,%ecx
>   6.86 â      and    $0x7fffffff,%ecx
>   0.72 â      test   $0x100000,%eax
>        â    â jne    40
>        â      test   $0xf0000,%eax
>   0.41 â    â je     5b
>        â      mov    $0x8,%ecx
>        â      mov    $0x2,%eax
>        â    â jmp    4a
>        â40:   mov    $0xc,%ecx
>        â      mov    $0x3,%eax
>   2.05 â4a:   add    %rcx,%rdx
>  16.60 â      mov    (%rdx),%ecx
>   2.66 â      test   %ecx,%ecx
>        â    â jne    6d
>   1.33 â      movl   $0x1,(%rdx)
>   1.54 â      pop    %rbp
>   4.51 â    â retq
>   3.89 â5b:   shr    $0x8,%ecx
>   9.53 â      and    $0x1,%ecx
>   0.61 â      movzbl %cl,%eax
>   0.92 â      movzbl %cl,%ecx
>   4.30 â      shl    $0x2,%rcx
>  14.14 â    â jmp    4a
>        â6d:   mov    $0xffffffff,%eax
>        â      pop    %rbp
>        â    â retq
>        â      xchg   %ax,%ax
> 
> 
> 



-- 
Best regards,
  Jesper Dangaard Brouer
  MSc.CS, Principal Kernel Engineer at Red Hat
  LinkedIn: http://www.linkedin.com/in/brouer