Commit Graph

12 Commits

Author SHA1 Message Date
78f6bb9b34 digilent: add driver skeleton + CMake option BS_ENABLE_DIGILENT
Stub Digilent JTAG-SMT backend, off by default. Wiring only: option,
conditional sources, dl link, drivers_list registration. Detect()
returns 0 for now; dlopen + real implementation in follow-up commits.
2026-05-23 11:06:43 +02:00
3c1e5f987e script: show drv/probe index in jtag_get_probes_list output
The probe ID printed by jtag_get_probes_list is the hex value to pass
verbatim to jtag_open_probe (parsed as base 16), but reading
"ID 0x00000000" and typing "1" as a 1-based index is a natural mistake
— and jtag_open_probe will accept 1, fail with a misleading
"FT_DEVICE_NOT_FOUND" since (drv=0, probe=1) does not exist.

Append explicit [drv N, probe M] decomposition so the value to copy is
unambiguous.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-23 10:49:16 +02:00
0bd109c209 fix: bsdl folder error printed uninitialized buffer
In cmd_autoinit, when find_first_file fails to open ./bsdl_files/ the
error path printed `filename` — which is only populated inside the
directory-walk loop. Outside that loop it is uninitialized stack
content, leading to garbage in the error message (and a confusing
diagnostic when bs is launched from a directory without a bsdl_files/
subfolder, e.g. build/).

Print `scanfolder` (the actual path that was attempted) instead.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-23 10:48:46 +02:00
dec0d14a06 phase 2.5: add bscan_spi/ — BSCAN proxy infrastructure
Low-level JTAG primitives operating directly on jc->io_functions
(single-device chain assumed), independent of jtag_core:
- bscan_set_ir
- bscan_shift_dr (TDI/TDO, LSB-first packing)
- bscan_idle_cycles

High-level operations driven by an fpga_target descriptor:
- bscan_load_bitstream: JPROGRAM -> CFG_IN -> shift (bit-reversed for
  Xilinx) -> JSTART -> idle -> BYPASS
- bscan_load_bitstream_file: parses the Xilinx .bit container header
  (sections a/b/c/d/e), falls back to raw .bin

bscan_spi_xfer is stubbed: the quartiq jtagspi protocol details will
be wired once we have a proxy .bit to validate against (OpenOCD
src/flash/nor/jtagspi.c is the host-side reference).

Three new script commands:
- bscan_set_ir <opcode_hex> <ir_length>
- bscan_shift_dr <nbits>  (writes zeros, prints captured TDO)
- bscan_load_bitstream <device> <path>

The sanity check for a healthy primitive on KU15P:
  jtag_init_scan; bscan_set_ir 9 6; bscan_shift_dr 32  ->  04 A5 60 93

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-20 23:10:54 +02:00
545fe09fd5 phase 2: add fpga/ module — per-target descriptor & registry
modules/fpga/ holds an fpga_target struct (IDCODE/mask, family, IR
length and private opcodes, proxy bitstream path, quirks) and a
compile-time registry. Initial entry: Xilinx Kintex UltraScale+
XCKU15P, populated from bsdl_files/xcku15p_ffve1517.bsd (IDCODE
0x04A56093, IR 6, USER1=0x02, CFG_IN=0x05, JPROGRAM=0x0B, JSTART=0x0C,
JSHUTDOWN=0x0D, ISC_DISABLE=0x16, quirk CCLK_VIA_STARTUP).

Two new script commands:
- fpga_list: enumerate the registry
- fpga_info: match each device on the JTAG chain against the registry
  and surface known quirks

Adding another FPGA = one entry in fpga_registry[] + its .bsd in
bsdl_files/. Proxy .bit will be wired in phase 2.5 (bscan_spi/).

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-20 22:52:52 +02:00
François Dausseur
9e0ca10a71 commands doc corrections.
Added a "script" directory for examples.
2025-02-18 15:40:58 +01:00
François Dausseur
5dfe5b123e compiles and works 2025-02-18 11:36:18 +01:00
b66e82da87 init and deinit 2025-02-16 19:43:48 +01:00
c8bda25d90 code refactoring 2025-02-16 19:32:01 +01:00
f3c2569a30 app dir changed and src dir removed 2025-02-16 17:37:56 +01:00
4f6cd20130 dir refactoring done. 2025-02-16 12:39:24 +01:00
a61fe778e6 dirs refactoring 2025-02-16 12:38:13 +01:00