perf intel-pt: Add support for emulated ptwrite

ptwrite is an Intel x86 instruction that writes arbitrary values into an
Intel PT trace. It is not supported on all hardware, so provide an
alternative that makes use of TNT packets to convey the payload data.
TNT packets encode Taken/Not-taken conditional branch information, so
taking branches based on the payload value will encode the value into
the TNT packet. Refer to the changes to the documentation file
perf-intel-pt.txt in this patch for an example.

Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Link: https://lore.kernel.org/r/20220509152400.376613-2-adrian.hunter@intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

1 files changed, 36 insertions, 1 deletions

diff --git a/tools/perf/util/intel-pt.c b/tools/perf/util/intel-pt.c
index ec43d364d0de..c7e115fefe7d 100644
--- a/tools/perf/util/intel-pt.c
+++ b/tools/perf/util/intel-pt.c
@@ -530,6 +530,7 @@ struct intel_pt_cache_entry {
 	u64				byte_cnt;
 	enum intel_pt_insn_op		op;
 	enum intel_pt_insn_branch	branch;
+	bool				emulated_ptwrite;
 	int				length;
 	int32_t				rel;
 	char				insn[INTEL_PT_INSN_BUF_SZ];
@@ -616,6 +617,7 @@ static int intel_pt_cache_add(struct dso *dso, struct machine *machine,
 	e->byte_cnt = byte_cnt;
 	e->op = intel_pt_insn->op;
 	e->branch = intel_pt_insn->branch;
+	e->emulated_ptwrite = intel_pt_insn->emulated_ptwrite;
 	e->length = intel_pt_insn->length;
 	e->rel = intel_pt_insn->rel;
 	memcpy(e->insn, intel_pt_insn->buf, INTEL_PT_INSN_BUF_SZ);
@@ -702,6 +704,28 @@ static int intel_pt_get_guest(struct intel_pt_queue *ptq)
 	return 0;
 }
 
+static inline bool intel_pt_jmp_16(struct intel_pt_insn *intel_pt_insn)
+{
+	return intel_pt_insn->rel == 16 && intel_pt_insn->branch == INTEL_PT_BR_UNCONDITIONAL;
+}
+
+#define PTWRITE_MAGIC		"\x0f\x0bperf,ptwrite  "
+#define PTWRITE_MAGIC_LEN	16
+
+static bool intel_pt_emulated_ptwrite(struct dso *dso, struct machine *machine, u64 offset)
+{
+	unsigned char buf[PTWRITE_MAGIC_LEN];
+	ssize_t len;
+
+	len = dso__data_read_offset(dso, machine, offset, buf, PTWRITE_MAGIC_LEN);
+	if (len == PTWRITE_MAGIC_LEN && !memcmp(buf, PTWRITE_MAGIC, PTWRITE_MAGIC_LEN)) {
+		intel_pt_log("Emulated ptwrite signature found\n");
+		return true;
+	}
+	intel_pt_log("Emulated ptwrite signature not found\n");
+	return false;
+}
+
 static int intel_pt_walk_next_insn(struct intel_pt_insn *intel_pt_insn,
 				   uint64_t *insn_cnt_ptr, uint64_t *ip,
 				   uint64_t to_ip, uint64_t max_insn_cnt,
@@ -764,6 +788,7 @@ static int intel_pt_walk_next_insn(struct intel_pt_insn *intel_pt_insn,
 				*ip += e->byte_cnt;
 				intel_pt_insn->op = e->op;
 				intel_pt_insn->branch = e->branch;
+				intel_pt_insn->emulated_ptwrite = e->emulated_ptwrite;
 				intel_pt_insn->length = e->length;
 				intel_pt_insn->rel = e->rel;
 				memcpy(intel_pt_insn->buf, e->insn,
@@ -795,8 +820,18 @@ static int intel_pt_walk_next_insn(struct intel_pt_insn *intel_pt_insn,
 
 			insn_cnt += 1;
 
-			if (intel_pt_insn->branch != INTEL_PT_BR_NO_BRANCH)
+			if (intel_pt_insn->branch != INTEL_PT_BR_NO_BRANCH) {
+				bool eptw;
+				u64 offs;
+
+				if (!intel_pt_jmp_16(intel_pt_insn))
+					goto out;
+				/* Check for emulated ptwrite */
+				offs = offset + intel_pt_insn->length;
+				eptw = intel_pt_emulated_ptwrite(al.map->dso, machine, offs);
+				intel_pt_insn->emulated_ptwrite = eptw;
 				goto out;
+			}
 
 			if (max_insn_cnt && insn_cnt >= max_insn_cnt)
 				goto out_no_cache;