perf intel-pt: Add new packets for PEBS via PT

Add 3 new packets to supports PEBS via PT, namely Block Begin Packet
(BBP), Block Item Packet (BIP) and Block End Packet (BEP). PEBS data is
encoded into multiple BIP packets that come between BBP and BEP. The BEP
packet might be associated with a FUP packet. That is indicated by using
a separate packet type (INTEL_PT_BEP_IP) similar to other packets types
with the _IP suffix.

Refer to the Intel SDM for more information about PEBS via PT:

https://software.intel.com/en-us/articles/intel-sdm
May 2019 version: Vol. 3B 18.5.5.2 PEBS output to Intel® Processor Trace

Decoding of BIP packets conflicts with single-byte TNT packets. Since
BIP packets only occur in the context of a block (i.e. between BBP and
BEP), that context must be recorded and passed to the packet decoder.

Signed-off-by: Adrian Hunter <[e-mail]>
Cc: Jiri Olsa <[e-mail]>
Link: http://lkml.kernel.org/r/[e-mail]
Signed-off-by: Arnaldo Carvalho de Melo <[e-mail]>

tools/perf/util/intel-pt-decoder/intel-pt-decoder.c

···
 	int mtc_shift;
 	struct intel_pt_stack stack;
 	enum intel_pt_pkt_state pkt_state;
+	enum intel_pt_pkt_ctx pkt_ctx;
 	struct intel_pt_pkt packet;
 	struct intel_pt_pkt tnt;
 	int pkt_step;
···
 	memcpy(buf + len, decoder->buf, n);
 	len += n;
 
-	ret = intel_pt_get_packet(buf, len, &decoder->packet);
+	ret = intel_pt_get_packet(buf, len, &decoder->packet, &decoder->pkt_ctx);
 	if (ret < (int)old_len) {
 		decoder->next_buf = decoder->buf;
 		decoder->next_len = decoder->len;
···
 {
 	struct intel_pt_pkt_info pkt_info;
 	const unsigned char *buf = decoder->buf;
+	enum intel_pt_pkt_ctx pkt_ctx = decoder->pkt_ctx;
 	size_t len = decoder->len;
 	int ret;
 
···
 			if (!len)
 				return INTEL_PT_NEED_MORE_BYTES;
 
-			ret = intel_pt_get_packet(buf, len, &pkt_info.packet);
+			ret = intel_pt_get_packet(buf, len, &pkt_info.packet,
+						  &pkt_ctx);
 			if (!ret)
 				return INTEL_PT_NEED_MORE_BYTES;
 			if (ret < 0)
···
 	case INTEL_PT_MNT:
 	case INTEL_PT_PTWRITE:
 	case INTEL_PT_PTWRITE_IP:
+	case INTEL_PT_BBP:
+	case INTEL_PT_BIP:
+	case INTEL_PT_BEP:
+	case INTEL_PT_BEP_IP:
 		return 0;
 
 	case INTEL_PT_MTC:
···
 		}
 
 		ret = intel_pt_get_packet(decoder->buf, decoder->len,
-					  &decoder->packet);
+					  &decoder->packet, &decoder->pkt_ctx);
 		if (ret == INTEL_PT_NEED_MORE_BYTES && BITS_PER_LONG == 32 &&
 		    decoder->len < INTEL_PT_PKT_MAX_SZ && !decoder->next_buf) {
 			ret = intel_pt_get_split_packet(decoder);
···
 		case INTEL_PT_MWAIT:
 		case INTEL_PT_PWRE:
 		case INTEL_PT_PWRX:
+		case INTEL_PT_BBP:
+		case INTEL_PT_BIP:
+		case INTEL_PT_BEP:
+		case INTEL_PT_BEP_IP:
 			decoder->have_tma = false;
 			intel_pt_log("ERROR: Unexpected packet\n");
 			err = -EAGAIN;
···
 		case INTEL_PT_MWAIT:
 		case INTEL_PT_PWRE:
 		case INTEL_PT_PWRX:
+		case INTEL_PT_BBP:
+		case INTEL_PT_BIP:
+		case INTEL_PT_BEP:
+		case INTEL_PT_BEP_IP:
 			intel_pt_log("ERROR: Missing TIP after FUP\n");
 			decoder->pkt_state = INTEL_PT_STATE_ERR3;
 			decoder->pkt_step = 0;
···
 			decoder->state.pwrx_payload = decoder->packet.payload;
 			return 0;
 
+		case INTEL_PT_BBP:
+		case INTEL_PT_BIP:
+		case INTEL_PT_BEP:
+		case INTEL_PT_BEP_IP:
+			break;
+
 		default:
 			return intel_pt_bug(decoder);
 		}
···
 		case INTEL_PT_MWAIT:
 		case INTEL_PT_PWRE:
 		case INTEL_PT_PWRX:
+		case INTEL_PT_BBP:
+		case INTEL_PT_BIP:
+		case INTEL_PT_BEP:
+		case INTEL_PT_BEP_IP:
 			intel_pt_log("ERROR: Unexpected packet\n");
 			err = -ENOENT;
 			goto out;
···
 		case INTEL_PT_MWAIT:
 		case INTEL_PT_PWRE:
 		case INTEL_PT_PWRX:
+		case INTEL_PT_BBP:
+		case INTEL_PT_BIP:
+		case INTEL_PT_BEP:
+		case INTEL_PT_BEP_IP:
 		default:
 			break;
 		}
···
 static bool intel_pt_next_tsc(unsigned char *buf, size_t len, uint64_t *tsc,
 			      size_t *rem)
 {
+	enum intel_pt_pkt_ctx ctx = INTEL_PT_NO_CTX;
 	struct intel_pt_pkt packet;
 	int ret;
 
 	while (len) {
-		ret = intel_pt_get_packet(buf, len, &packet);
+		ret = intel_pt_get_packet(buf, len, &packet, &ctx);
 		if (ret <= 0)
 			return false;
 		if (packet.type == INTEL_PT_TSC) {

tools/perf/util/intel-pt-decoder/intel-pt-pkt-decoder.c

···
 	[INTEL_PT_MWAIT]	= "MWAIT",
 	[INTEL_PT_PWRE]		= "PWRE",
 	[INTEL_PT_PWRX]		= "PWRX",
+	[INTEL_PT_BBP]		= "BBP",
+	[INTEL_PT_BIP]		= "BIP",
+	[INTEL_PT_BEP]		= "BEP",
+	[INTEL_PT_BEP_IP]	= "BEP",
 };
 
 const char *intel_pt_pkt_name(enum intel_pt_pkt_type type)
···
 	return 7;
 }
 
+static int intel_pt_get_bbp(const unsigned char *buf, size_t len,
+			    struct intel_pt_pkt *packet)
+{
+	if (len < 3)
+		return INTEL_PT_NEED_MORE_BYTES;
+	packet->type = INTEL_PT_BBP;
+	packet->count = buf[2] >> 7;
+	packet->payload = buf[2] & 0x1f;
+	return 3;
+}
+
+static int intel_pt_get_bip_4(const unsigned char *buf, size_t len,
+			      struct intel_pt_pkt *packet)
+{
+	if (len < 5)
+		return INTEL_PT_NEED_MORE_BYTES;
+	packet->type = INTEL_PT_BIP;
+	packet->count = buf[0] >> 3;
+	memcpy_le64(&packet->payload, buf + 1, 4);
+	return 5;
+}
+
+static int intel_pt_get_bip_8(const unsigned char *buf, size_t len,
+			      struct intel_pt_pkt *packet)
+{
+	if (len < 9)
+		return INTEL_PT_NEED_MORE_BYTES;
+	packet->type = INTEL_PT_BIP;
+	packet->count = buf[0] >> 3;
+	memcpy_le64(&packet->payload, buf + 1, 8);
+	return 9;
+}
+
+static int intel_pt_get_bep(size_t len, struct intel_pt_pkt *packet)
+{
+	if (len < 2)
+		return INTEL_PT_NEED_MORE_BYTES;
+	packet->type = INTEL_PT_BEP;
+	return 2;
+}
+
+static int intel_pt_get_bep_ip(size_t len, struct intel_pt_pkt *packet)
+{
+	if (len < 2)
+		return INTEL_PT_NEED_MORE_BYTES;
+	packet->type = INTEL_PT_BEP_IP;
+	return 2;
+}
+
 static int intel_pt_get_ext(const unsigned char *buf, size_t len,
 			    struct intel_pt_pkt *packet)
 {
···
 		return intel_pt_get_pwre(buf, len, packet);
 	case 0xA2: /* PWRX */
 		return intel_pt_get_pwrx(buf, len, packet);
+	case 0x63: /* BBP */
+		return intel_pt_get_bbp(buf, len, packet);
+	case 0x33: /* BEP no IP */
+		return intel_pt_get_bep(len, packet);
+	case 0xb3: /* BEP with IP */
+		return intel_pt_get_bep_ip(len, packet);
 	default:
 		return INTEL_PT_BAD_PACKET;
 	}
···
 }
 
 static int intel_pt_do_get_packet(const unsigned char *buf, size_t len,
-				  struct intel_pt_pkt *packet)
+				  struct intel_pt_pkt *packet,
+				  enum intel_pt_pkt_ctx ctx)
 {
 	unsigned int byte;
 
···
 		return INTEL_PT_NEED_MORE_BYTES;
 
 	byte = buf[0];
+
+	switch (ctx) {
+	case INTEL_PT_NO_CTX:
+		break;
+	case INTEL_PT_BLK_4_CTX:
+		if ((byte & 0x7) == 4)
+			return intel_pt_get_bip_4(buf, len, packet);
+		break;
+	case INTEL_PT_BLK_8_CTX:
+		if ((byte & 0x7) == 4)
+			return intel_pt_get_bip_8(buf, len, packet);
+		break;
+	default:
+		break;
+	};
+
 	if (!(byte & BIT(0))) {
 		if (byte == 0)
 			return intel_pt_get_pad(packet);
···
 	}
 }
 
+void intel_pt_upd_pkt_ctx(const struct intel_pt_pkt *packet,
+			  enum intel_pt_pkt_ctx *ctx)
+{
+	switch (packet->type) {
+	case INTEL_PT_BAD:
+	case INTEL_PT_PAD:
+	case INTEL_PT_TSC:
+	case INTEL_PT_TMA:
+	case INTEL_PT_MTC:
+	case INTEL_PT_FUP:
+	case INTEL_PT_CYC:
+	case INTEL_PT_CBR:
+	case INTEL_PT_MNT:
+	case INTEL_PT_EXSTOP:
+	case INTEL_PT_EXSTOP_IP:
+	case INTEL_PT_PWRE:
+	case INTEL_PT_PWRX:
+	case INTEL_PT_BIP:
+		break;
+	case INTEL_PT_TNT:
+	case INTEL_PT_TIP:
+	case INTEL_PT_TIP_PGD:
+	case INTEL_PT_TIP_PGE:
+	case INTEL_PT_MODE_EXEC:
+	case INTEL_PT_MODE_TSX:
+	case INTEL_PT_PIP:
+	case INTEL_PT_OVF:
+	case INTEL_PT_VMCS:
+	case INTEL_PT_TRACESTOP:
+	case INTEL_PT_PSB:
+	case INTEL_PT_PSBEND:
+	case INTEL_PT_PTWRITE:
+	case INTEL_PT_PTWRITE_IP:
+	case INTEL_PT_MWAIT:
+	case INTEL_PT_BEP:
+	case INTEL_PT_BEP_IP:
+		*ctx = INTEL_PT_NO_CTX;
+		break;
+	case INTEL_PT_BBP:
+		if (packet->count)
+			*ctx = INTEL_PT_BLK_4_CTX;
+		else
+			*ctx = INTEL_PT_BLK_8_CTX;
+		break;
+	default:
+		break;
+	}
+}
+
 int intel_pt_get_packet(const unsigned char *buf, size_t len,
-			struct intel_pt_pkt *packet)
+			struct intel_pt_pkt *packet, enum intel_pt_pkt_ctx *ctx)
 {
 	int ret;
 
-	ret = intel_pt_do_get_packet(buf, len, packet);
+	ret = intel_pt_do_get_packet(buf, len, packet, *ctx);
 	if (ret > 0) {
 		while (ret < 8 && len > (size_t)ret && !buf[ret])
 			ret += 1;
+		intel_pt_upd_pkt_ctx(packet, ctx);
 	}
 	return ret;
 }
···
 		return snprintf(buf, buf_len, "%s 0x%llx IP:0", name, payload);
 	case INTEL_PT_PTWRITE_IP:
 		return snprintf(buf, buf_len, "%s 0x%llx IP:1", name, payload);
+	case INTEL_PT_BEP:
 	case INTEL_PT_EXSTOP:
 		return snprintf(buf, buf_len, "%s IP:0", name);
+	case INTEL_PT_BEP_IP:
 	case INTEL_PT_EXSTOP_IP:
 		return snprintf(buf, buf_len, "%s IP:1", name);
 	case INTEL_PT_MWAIT:
···
 				(unsigned int)((payload >> 4) & 0xf),
 				(unsigned int)(payload & 0xf),
 				(unsigned int)((payload >> 8) & 0xf));
+	case INTEL_PT_BBP:
+		return snprintf(buf, buf_len, "%s SZ %s-byte Type 0x%llx",
+				name, packet->count ? "4" : "8", payload);
+	case INTEL_PT_BIP:
+		return snprintf(buf, buf_len, "%s ID 0x%02x Value 0x%llx",
+				name, packet->count, payload);
 	default:
 		break;
 	}

tools/perf/util/intel-pt-decoder/intel-pt-pkt-decoder.h

···
 	INTEL_PT_MWAIT,
 	INTEL_PT_PWRE,
 	INTEL_PT_PWRX,
+	INTEL_PT_BBP,
+	INTEL_PT_BIP,
+	INTEL_PT_BEP,
+	INTEL_PT_BEP_IP,
 };
 
 struct intel_pt_pkt {
···
 	uint64_t		payload;
 };
 
+/*
+ * Decoding of BIP packets conflicts with single-byte TNT packets. Since BIP
+ * packets only occur in the context of a block (i.e. between BBP and BEP), that
+ * context must be recorded and passed to the packet decoder.
+ */
+enum intel_pt_pkt_ctx {
+	INTEL_PT_NO_CTX,	/* BIP packets are invalid */
+	INTEL_PT_BLK_4_CTX,	/* 4-byte BIP packets */
+	INTEL_PT_BLK_8_CTX,	/* 8-byte BIP packets */
+};
+
 const char *intel_pt_pkt_name(enum intel_pt_pkt_type);
 
 int intel_pt_get_packet(const unsigned char *buf, size_t len,
-			struct intel_pt_pkt *packet);
+			struct intel_pt_pkt *packet,
+			enum intel_pt_pkt_ctx *ctx);
+
+void intel_pt_upd_pkt_ctx(const struct intel_pt_pkt *packet,
+			  enum intel_pt_pkt_ctx *ctx);
 
 int intel_pt_pkt_desc(const struct intel_pt_pkt *packet, char *buf, size_t len);
 

tools/perf/util/intel-pt.c

···
 	int ret, pkt_len, i;
 	char desc[INTEL_PT_PKT_DESC_MAX];
 	const char *color = PERF_COLOR_BLUE;
+	enum intel_pt_pkt_ctx ctx = INTEL_PT_NO_CTX;
 
 	color_fprintf(stdout, color,
 		      ". ... Intel Processor Trace data: size %zu bytes\n",
 		      len);
 
 	while (len) {
-		ret = intel_pt_get_packet(buf, len, &packet);
+		ret = intel_pt_get_packet(buf, len, &packet, &ctx);
 		if (ret > 0)
 			pkt_len = ret;
 		else