Re: [PATCH 1/2] perf intel-pt-decoder: Report instruction bytes and length in sample

[Arnaldo Carvalho de Melo <acme@kernel.org>]

Em Mon, Oct 03, 2016 at 05:30:32PM -0700, Andi Kleen escreveu:
> From: Andi Kleen <ak@linux.intel.com>
> 
> Change the Intel PT decoder to pass up the length and the instruction
> bytes of the decoded or sampled instruction in the perf sample.
> 
> The decoder already knows this information, we just need to pass it
> up. Since it is only a couple of movs it is not very expensive.
> 
> Used in the next patch.

Adrian, Ack?

- Arnaldo
 
> v2: Handle instruction cache too. Make sure ilen is always initialized.
> Signed-off-by: Andi Kleen <ak@linux.intel.com>
> ---
>  tools/perf/util/event.h                                  |  3 +++
>  tools/perf/util/intel-pt-decoder/intel-pt-decoder.c      |  2 ++
>  tools/perf/util/intel-pt-decoder/intel-pt-decoder.h      |  3 +++
>  tools/perf/util/intel-pt-decoder/intel-pt-insn-decoder.h |  2 +-
>  tools/perf/util/intel-pt.c                               | 10 ++++++++++
>  5 files changed, 19 insertions(+), 1 deletion(-)
> 
> diff --git a/tools/perf/util/event.h b/tools/perf/util/event.h
> index 8d363d5e65a2..c735c53a26f8 100644
> --- a/tools/perf/util/event.h
> +++ b/tools/perf/util/event.h
> @@ -177,6 +177,8 @@ enum {
>  	PERF_IP_FLAG_TRACE_BEGIN	|\
>  	PERF_IP_FLAG_TRACE_END)
>  
> +#define MAX_INSN 16
> +
>  struct perf_sample {
>  	u64 ip;
>  	u32 pid, tid;
> @@ -193,6 +195,7 @@ struct perf_sample {
>  	u32 flags;
>  	u16 insn_len;
>  	u8  cpumode;
> +	char insn[MAX_INSN];
>  	void *raw_data;
>  	struct ip_callchain *callchain;
>  	struct branch_stack *branch_stack;
> diff --git a/tools/perf/util/intel-pt-decoder/intel-pt-decoder.c b/tools/perf/util/intel-pt-decoder/intel-pt-decoder.c
> index 8ff6c6a61291..8a5e21abb790 100644
> --- a/tools/perf/util/intel-pt-decoder/intel-pt-decoder.c
> +++ b/tools/perf/util/intel-pt-decoder/intel-pt-decoder.c
> @@ -949,6 +949,8 @@ out:
>  out_no_progress:
>  	decoder->state.insn_op = intel_pt_insn->op;
>  	decoder->state.insn_len = intel_pt_insn->length;
> +	memcpy(decoder->state.insn, intel_pt_insn->buf,
> +	       sizeof(decoder->state.insn));
>  
>  	if (decoder->tx_flags & INTEL_PT_IN_TX)
>  		decoder->state.flags |= INTEL_PT_IN_TX;
> diff --git a/tools/perf/util/intel-pt-decoder/intel-pt-decoder.h b/tools/perf/util/intel-pt-decoder/intel-pt-decoder.h
> index 02c38fec1c37..fbd7d08d97d5 100644
> --- a/tools/perf/util/intel-pt-decoder/intel-pt-decoder.h
> +++ b/tools/perf/util/intel-pt-decoder/intel-pt-decoder.h
> @@ -20,6 +20,8 @@
>  #include <stddef.h>
>  #include <stdbool.h>
>  
> +#define MAX_INSN			16
> +
>  #include "intel-pt-insn-decoder.h"
>  
>  #define INTEL_PT_IN_TX		(1 << 0)
> @@ -66,6 +68,7 @@ struct intel_pt_state {
>  	uint32_t flags;
>  	enum intel_pt_insn_op insn_op;
>  	int insn_len;
> +	char insn[MAX_INSN];
>  };
>  
>  struct intel_pt_insn;
> diff --git a/tools/perf/util/intel-pt-decoder/intel-pt-insn-decoder.h b/tools/perf/util/intel-pt-decoder/intel-pt-insn-decoder.h
> index b0adbf37323e..47e196dec224 100644
> --- a/tools/perf/util/intel-pt-decoder/intel-pt-insn-decoder.h
> +++ b/tools/perf/util/intel-pt-decoder/intel-pt-insn-decoder.h
> @@ -20,7 +20,7 @@
>  #include <stdint.h>
>  
>  #define INTEL_PT_INSN_DESC_MAX		32
> -#define INTEL_PT_INSN_DBG_BUF_SZ	16
> +#define INTEL_PT_INSN_DBG_BUF_SZ	16 /* Must be >= MAX_INSN */
>  
>  enum intel_pt_insn_op {
>  	INTEL_PT_OP_OTHER,
> diff --git a/tools/perf/util/intel-pt.c b/tools/perf/util/intel-pt.c
> index b9cc353cace2..4b9d0086a383 100644
> --- a/tools/perf/util/intel-pt.c
> +++ b/tools/perf/util/intel-pt.c
> @@ -140,6 +140,7 @@ struct intel_pt_queue {
>  	u32 flags;
>  	u16 insn_len;
>  	u64 last_insn_cnt;
> +	char insn[MAX_INSN];
>  };
>  
>  static void intel_pt_dump(struct intel_pt *pt __maybe_unused,
> @@ -305,6 +306,7 @@ struct intel_pt_cache_entry {
>  	enum intel_pt_insn_branch	branch;
>  	int				length;
>  	int32_t				rel;
> +	char				insn[MAX_INSN];
>  };
>  
>  static int intel_pt_config_div(const char *var, const char *value, void *data)
> @@ -390,6 +392,7 @@ static int intel_pt_cache_add(struct dso *dso, struct machine *machine,
>  	e->branch = intel_pt_insn->branch;
>  	e->length = intel_pt_insn->length;
>  	e->rel = intel_pt_insn->rel;
> +	memcpy(e->insn, intel_pt_insn->buf, MAX_INSN);
>  
>  	err = auxtrace_cache__add(c, offset, &e->entry);
>  	if (err)
> @@ -427,6 +430,8 @@ static int intel_pt_walk_next_insn(struct intel_pt_insn *intel_pt_insn,
>  	u64 insn_cnt = 0;
>  	bool one_map = true;
>  
> +	intel_pt_insn->length = 0;
> +
>  	if (to_ip && *ip == to_ip)
>  		goto out_no_cache;
>  
> @@ -468,6 +473,7 @@ static int intel_pt_walk_next_insn(struct intel_pt_insn *intel_pt_insn,
>  				intel_pt_insn->branch = e->branch;
>  				intel_pt_insn->length = e->length;
>  				intel_pt_insn->rel = e->rel;
> +				memcpy(intel_pt_insn->buf, e->insn, MAX_INSN);
>  				intel_pt_log_insn_no_data(intel_pt_insn, *ip);
>  				return 0;
>  			}
> @@ -817,6 +823,7 @@ static void intel_pt_sample_flags(struct intel_pt_queue *ptq)
>  		if (ptq->state->flags & INTEL_PT_IN_TX)
>  			ptq->flags |= PERF_IP_FLAG_IN_TX;
>  		ptq->insn_len = ptq->state->insn_len;
> +		memcpy(ptq->insn, ptq->state->insn, MAX_INSN);
>  	}
>  }
>  
> @@ -997,6 +1004,7 @@ static int intel_pt_synth_branch_sample(struct intel_pt_queue *ptq)
>  	sample.cpu = ptq->cpu;
>  	sample.flags = ptq->flags;
>  	sample.insn_len = ptq->insn_len;
> +	memcpy(sample.insn, ptq->insn, MAX_INSN);
>  
>  	/*
>  	 * perf report cannot handle events without a branch stack when using
> @@ -1058,6 +1066,7 @@ static int intel_pt_synth_instruction_sample(struct intel_pt_queue *ptq)
>  	sample.cpu = ptq->cpu;
>  	sample.flags = ptq->flags;
>  	sample.insn_len = ptq->insn_len;
> +	memcpy(sample.insn, ptq->insn, MAX_INSN);
>  
>  	ptq->last_insn_cnt = ptq->state->tot_insn_cnt;
>  
> @@ -1120,6 +1129,7 @@ static int intel_pt_synth_transaction_sample(struct intel_pt_queue *ptq)
>  	sample.cpu = ptq->cpu;
>  	sample.flags = ptq->flags;
>  	sample.insn_len = ptq->insn_len;
> +	memcpy(sample.insn, ptq->insn, MAX_INSN);
>  
>  	if (pt->synth_opts.callchain) {
>  		thread_stack__sample(ptq->thread, ptq->chain,
> -- 
> 2.5.5