JTAG: simple autoprobing

This patch adds basic autoprobing support for the JTAG scan chains
which cooperate.  To use, you can invoke OpenOCD with just:

 - interface spec: "-f interface/...cfg"
 - possibly with "-c 'reset_config ...'" for SRST/TRST
 - possibly with "-c 'jtag_khz ...'" for the JTAG clock

Then set up config files matching the reported TAPs.  It doesn't
declare targets ... just TAPs.  So facilities above the JTAG and
SVF/XSVF levels won't be available without a real config; this is
almost purely a way to generate diagnostics.

Autoprobe was successful with most boards I tested, except ones
incorporating C55x DSPs (which don't cooperate with this scheme
for IR length autodetection).  Here's what one multi-TAP chip
reported, with the "Warn:" prefixes removed:

 clock speed 500 kHz
 There are no enabled taps.  AUTO PROBING MIGHT NOT WORK!!
 AUTO auto0.tap - use "jtag newtap auto0 tap -expected-id 0x2b900f0f ..."
 AUTO auto1.tap - use "jtag newtap auto1 tap -expected-id 0x07926001 ..."
 AUTO auto2.tap - use "jtag newtap auto2 tap -expected-id 0x0b73b02f ..."
 AUTO auto0.tap - use "... -irlen 4"
 AUTO auto1.tap - use "... -irlen 4"
 AUTO auto2.tap - use "... -irlen 6"
 no gdb ports allocated as no target has been specified

The patch tweaks IR setup a bit, so we can represent TAPs with
undeclared IR length.

Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>

TODO

@@ -55,8 +55,10 @@ directly in minidriver API for better embedded host performance.
 
 The following tasks have been suggested for adding new core JTAG support:
 
-- autodetect devices present on the scan chain
+- Improve autodetection of TAPs by supporting tcl escape procedures that
-  - implement 'discover_taps' command
+  can configure discovered TAPs based on IDCODE value ... they could:
+    - Remove guessing for irlen
+    - Allow non-default irmask/ircapture values
 - SPI/UART emulation:
   - (ab)use bit-banging JTAG interfaces to emulate SPI/UART
   - allow SPI to program flash, MCUs, etc.
@@ -94,6 +96,8 @@ interface support:
 - FT2232 (libftdi):
   - make performance comparable to alternatives (on Win32, D2XX is faster)
   - make usability comparable to alternatives
+- Autodetect USB based adapters; this should be easy on Linux.  If there's
+  more than one, list the options; otherwise, just select that one.
 
 The following tasks have been suggested for adding new JTAG interfaces:
 
@@ -133,6 +137,7 @@ Once the above are completed:
 
 @section thelisttargets Target Support
 
+- Many common ARM cores could be autodetected using IDCODE
 - general layer cleanup: @par
   https://lists.berlios.de/pipermail/openocd-development/2009-May/006590.html
 - regression: "reset halt" between 729(works) and 788(fails): @par

src/jtag/core.c

@@ -890,6 +890,9 @@ void jtag_sleep(uint32_t us)
  */
 #define END_OF_CHAIN_FLAG	0x000000ff
 
+/* a larger IR length than we ever expect to autoprobe */
+#define JTAG_IRLEN_MAX		60
+
 static int jtag_examine_chain_execute(uint8_t *idcode_buffer, unsigned num_idcode)
 {
 	scan_field_t field = {
@@ -1027,6 +1030,8 @@ static int jtag_examine_chain(void)
 	uint8_t idcode_buffer[JTAG_MAX_CHAIN_SIZE * 4];
 	unsigned bit_count;
 	int retval;
+	int tapcount = 0;
+	bool autoprobe = false;
 
 	/* DR scan to collect BYPASS or IDCODE register contents.
 	 * Then make sure the scan data has both ones and zeroes.
@@ -1040,11 +1045,9 @@ static int jtag_examine_chain(void)
 
 	/* point at the 1st tap */
 	jtag_tap_t *tap = jtag_tap_next_enabled(NULL);
-	if (tap == NULL)
+
-	{
+	if (!tap)
-		LOG_ERROR("JTAG: No taps enabled?");
+		autoprobe = true;
-		return ERROR_JTAG_INIT_FAILED;
-	}
 
 	for (bit_count = 0;
 			tap && bit_count < (JTAG_MAX_CHAIN_SIZE * 32) - 31;
@@ -1086,6 +1089,59 @@ static int jtag_examine_chain(void)
 		return ERROR_JTAG_INIT_FAILED;
 	}
 
+	/* if autoprobing, the tap list is still empty ... populate it! */
+	while (autoprobe && bit_count < (JTAG_MAX_CHAIN_SIZE * 32) - 31) {
+		uint32_t idcode;
+		char buf[12];
+
+		/* Is there another TAP? */
+		idcode = buf_get_u32(idcode_buffer, bit_count, 32);
+		if (jtag_idcode_is_final(idcode))
+			break;
+
+		/* Default everything in this TAP except IR length.
+		 *
+		 * REVISIT create a jtag_alloc(chip, tap) routine, and
+		 * share it with jim_newtap_cmd().
+		 */
+		tap = calloc(1, sizeof *tap);
+		if (!tap)
+			return ERROR_FAIL;
+
+		sprintf(buf, "auto%d", tapcount++);
+		tap->chip = strdup(buf);
+		tap->tapname = strdup("tap");
+
+		sprintf(buf, "%s.%s", tap->chip, tap->tapname);
+		tap->dotted_name = strdup(buf);
+
+		/* tap->ir_length == 0 ... signifying irlen autoprobe */
+		tap->ir_capture_mask = 0x03;
+		tap->ir_capture_value = 0x01;
+
+		tap->enabled = true;
+
+		if ((idcode & 1) == 0) {
+			bit_count += 1;
+			tap->hasidcode = false;
+		} else {
+			bit_count += 32;
+			tap->hasidcode = true;
+			tap->idcode = idcode;
+
+			tap->expected_ids_cnt = 1;
+			tap->expected_ids = malloc(sizeof(uint32_t));
+			tap->expected_ids[0] = idcode;
+		}
+
+		LOG_WARNING("AUTO %s - use \"jtag newtap "
+				"%s %s -expected-id 0x%8.8" PRIx32 " ...\"",
+				tap->dotted_name, tap->chip, tap->tapname,
+				tap->idcode);
+
+		jtag_tap_init(tap);
+	}
+
 	/* After those IDCODE or BYPASS register values should be
 	 * only the data we fed into the scan chain.
 	 */
@@ -1120,10 +1176,13 @@ static int jtag_validate_ircapture(void)
 	int chain_pos = 0;
 	int retval;
 
+	/* when autoprobing, accomodate huge IR lengths */
 	for (tap = NULL, total_ir_length = 0;
 			(tap = jtag_tap_next_enabled(tap)) != NULL;
-			total_ir_length += tap->ir_length)
+			total_ir_length += tap->ir_length) {
-		continue;
+		if (tap->ir_length == 0)
+			total_ir_length += JTAG_IRLEN_MAX;
+	}
 
 	/* increase length to add 2 bit sentinel after scan */
 	total_ir_length += 2;
@@ -1156,6 +1215,25 @@ static int jtag_validate_ircapture(void)
 			break;
 		}
 
+		/* If we're autoprobing, guess IR lengths.  They must be at
+		 * least two bits.  Guessing will fail if (a) any TAP does
+		 * not conform to the JTAG spec; or (b) when the upper bits
+		 * captured from some conforming TAP are nonzero.
+		 *
+		 * REVISIT alternative approach: escape to some tcl code
+		 * which could provide more knowledge, based on IDCODE; and
+		 * only guess when that has no success.
+		 */
+		if (tap->ir_length == 0) {
+			tap->ir_length = 2;
+			while ((val = buf_get_u32(ir_test, chain_pos,
+						tap->ir_length + 1)) == 1) {
+				tap->ir_length++;
+			}
+			LOG_WARNING("AUTO %s - use \"... -irlen %d\"",
+					jtag_tap_name(tap), tap->ir_length);
+		}
+
 		/* Validate the two LSBs, which must be 01 per JTAG spec.
 		 *
 		 * Or ... more bits could be provided by TAP declaration.
@@ -1207,9 +1285,8 @@ void jtag_tap_init(jtag_tap_t *tap)
 	unsigned ir_len_bits;
 	unsigned ir_len_bytes;
 
-	assert(0 != tap->ir_length);
+	/* if we're autoprobing, cope with potentially huge ir_length */
-
+	ir_len_bits = tap->ir_length ? : JTAG_IRLEN_MAX;
-	ir_len_bits = tap->ir_length;
 	ir_len_bytes = CEIL(ir_len_bits, 8);
 
 	tap->expected = calloc(1, ir_len_bytes);
@@ -1302,8 +1379,21 @@ int jtag_init_inner(struct command_context_s *cmd_ctx)
 
 	tap = jtag_tap_next_enabled(NULL);
 	if (tap == NULL) {
-		LOG_ERROR("There are no enabled taps?");
+		/* Once JTAG itself is properly set up, and the scan chain
-		return ERROR_JTAG_INIT_FAILED;
+		 * isn't absurdly large, IDCODE autoprobe should work fine.
+		 *
+		 * But ... IRLEN autoprobe can fail even on systems which
+		 * are fully conformant to JTAG.  Also, JTAG setup can be
+		 * quite finicky on some systems.
+		 *
+		 * REVISIT: if TAP autoprobe works OK, then in many cases
+		 * we could escape to tcl code and set up targets based on
+		 * the TAP's IDCODE values.
+		 */
+		LOG_WARNING("There are no enabled taps.  "
+				"AUTO PROBING MIGHT NOT WORK!!");
+
+		/* REVISIT default clock will often be too fast ... */
 	}
 
 	jtag_add_tlr();