JTAG: simple autoprobing
authorDavid Brownell <dbrownell@users.sourceforge.net>
Mon, 26 Oct 2009 07:36:03 +0000 (00:36 -0700)
committerDavid Brownell <dbrownell@users.sourceforge.net>
Mon, 26 Oct 2009 07:36:03 +0000 (00:36 -0700)
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
src/jtag/core.c

diff --git a/TODO b/TODO
index 6f9c749414accee8c46a9c20f5b9f15565c3e184..f567a82f9cba00dc09af8b1596505a2ff597be77 100644 (file)
--- a/TODO
+++ b/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
-  - implement 'discover_taps' command
+- Improve autodetection of TAPs by supporting tcl escape procedures that
+  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
index 08cfe436b236a41b94a009ef2a72b3abc6bfc404..8bb45bcdf53dd791cbd9843cf44f787f8054bf91 100644 (file)
@@ -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)
-       {
-               LOG_ERROR("JTAG: No taps enabled?");
-               return ERROR_JTAG_INIT_FAILED;
-       }
+
+       if (!tap)
+               autoprobe = true;
 
        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)
-               continue;
+                       total_ir_length += tap->ir_length) {
+               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);
-
-       ir_len_bits = tap->ir_length;
+       /* if we're autoprobing, cope with potentially huge ir_length */
+       ir_len_bits = tap->ir_length ? : JTAG_IRLEN_MAX;
        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?");
-               return ERROR_JTAG_INIT_FAILED;
+               /* Once JTAG itself is properly set up, and the scan chain
+                * 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();

Linking to existing account procedure

If you already have an account and want to add another login method you MUST first sign in with your existing account and then change URL to read https://review.openocd.org/login/?link to get to this page again but this time it'll work for linking. Thank you.

SSH host keys fingerprints

1024 SHA256:YKx8b7u5ZWdcbp7/4AeXNaqElP49m6QrwfXaqQGJAOk gerrit-code-review@openocd.zylin.com (DSA)
384 SHA256:jHIbSQa4REvwCFG4cq5LBlBLxmxSqelQPem/EXIrxjk gerrit-code-review@openocd.org (ECDSA)
521 SHA256:UAOPYkU9Fjtcao0Ul/Rrlnj/OsQvt+pgdYSZ4jOYdgs gerrit-code-review@openocd.org (ECDSA)
256 SHA256:A13M5QlnozFOvTllybRZH6vm7iSt0XLxbA48yfc2yfY gerrit-code-review@openocd.org (ECDSA)
256 SHA256:spYMBqEYoAOtK7yZBrcwE8ZpYt6b68Cfh9yEVetvbXg gerrit-code-review@openocd.org (ED25519)
+--[ED25519 256]--+
|=..              |
|+o..   .         |
|*.o   . .        |
|+B . . .         |
|Bo. = o S        |
|Oo.+ + =         |
|oB=.* = . o      |
| =+=.+   + E     |
|. .=o   . o      |
+----[SHA256]-----+
2048 SHA256:0Onrb7/PHjpo6iVZ7xQX2riKN83FJ3KGU0TvI0TaFG4 gerrit-code-review@openocd.zylin.com (RSA)