Release Candidate V2.0.0-rc1

We are happy to announce the first v2.0.0 release candidate of Black Magic Debug. This is the first major release of the v2 series.

Here are some of the highlights:

• Switched to the Meson build system
• Implemented RISC-V support
• Implemented Cortex-R support
• Implemented ARMv8-M support
• And a bunch of newly supported targets:
  • Ambiq Apollo 3
  • ArteryTek AT32F405/F403/F413/F421/F423/F437
  • Atmel SAMC21
  • GigaDevice GD32F405
  • GigaDevice GD32VF1
  • Nordic Semi nRF54L
  • NXP Kinetis MK20DX256
  • NXP S32K344
  • Puya PY32
  • Renesas RZ/A1 A1L, A1LU and A1H
  • RPi Foundation RP2350
  • ST Micro STM32F410
  • ST Micro STM32H72/H73
  • ST Micro STM32H7A/H7B
  • ST Micro STM32WB0
  • TI MSPM0
  • TI TM4C1294KCDT
  • WinChipHead CH579

For more details please refer to the ChangeLog down below.

ChangeLog V2.0.0-rc1

Core Changes

• Implemented proper IR quirk handling for Xilinx FPGAs [lethalbit]
• Fixed some preprocessor logic issues for when RISC-V Debug is disabled [dragonmux]
• Implemented support for the vCont and qAttached GDB packets [dragonmux]
• Format string fixes in the RTT support [koendv]
• Format string and documentation fixes for monitor heapinfo [koendv]
• Introduced new DEBUG_*_IS_NOOP macros to allow better logic to be built around debug output [dragonmux]
• Format string fixes across the GDB server and monitor command systems [ALTracer]
• Missing prototypes and macro usage improvements [elagil]
• Fixed some DEBUG_*() calls with bad format strings [xobs]
• Improvements to allow the GDB server to idle more [ALTracer]
• Fixed various typos and spelling mistakes across the code code base [francois-berder]
• Improved the performance of the GDB qCRC packet handling [ALTracer]
• Fixed an issue with how aliasing works under macOS breaking the build of the CRC32 code [ALTracer]
• Improved the space efficiency and correctness of the 12 character serial number logic [ALTracer]
• Format string fixes and clean up [dragonmux]
• Reorganised the semihosting support out from the Cortex-M support [dragonmux]
• Refactored the semihosting implementation and fixed some implementation errors [dragonmux]
• Fixed an issue with alias handling on macOS [amyspark]
• Warnings and nomenclature fixes across the code base [dragonmux]
• Fixed the Clang fall-through warnings as the compiler doesn't understand the magic comment used before [dragonmux]
• Laid the groundwork for supporting the AM335x family parts from TI [dragonmux]
• Fixed some issues with the handling of GDB's single register read/write commands [OmniTechnoMancer]
• Removed all usage of sscanf() in the firmware to reduce code size and improve speed [OmniTechnoMancer]
• Added the specific IR quirk for the Xilinx Artix 7 XC7A200T [LAK132]
• Fixed an error in the remaining length calculation for the vFlashWrite GDB command handling [OmniTechnoMancer]
• Cleaned up from the tooling changes [dragonmux]
• Fixed a timeout mishandling issue that would result in timeouts executing wrongly for a few seconds every 49.7 days of uptime [dragonmux]
• Converted all uses of sprintf() over to snprintf() for improved safety and code size [ALTracer]
• Fixed a bug where the result of vasprintf() was ignored in the GDB packet handling, leading to crashes and badness [ALTracer]
• Improved handling of faults in the remote protocol so BMDA can properly recover from them [dragonmux]
• Overhaulled the exception system to fix a bug in how exception frames were being handled [dragonmux]
• Fixed a bug in how F-packets are detected during semihosting response processing [OmniTechnoMancer]
• Implemented support for semihosting's SYS_ELAPSED and SYS_TICKFREQ calls [ALTracer]
• Fixed a crash from in target_list_free() caused by improper exception handling within the function [dragonmux]
• Fixed a NULL pointer dereference in the target voltage handling and made the implementations consistent [ALTracer]
• Fixed compilation with newlib 4.3.0+ by opting out of the unwinder logic it now pulls in [ALTracer]
• Implemented deinit for the SWO implementation and made the buffer for async mode heap allocated [ALTracer]
• Demarked the platform-specific commands in the monitor help output [Misaka0x2730]
• Fixed the format string warnings for all tc_printf() usage [HrMitrev]
• Fixed several bugs in the Manchester-mode SWO support that had it broken across several platforms [ALTracer]
• Fixed a NULL dereference that was possible in the vFlashDone handler when run with no target selected [ALTracer]
• Made the SWO command report back the actual baud rate achieved for UART-mode, not just the one requested [ALTracer]
• Reduced duplication by consolidating the GDB packet buffer sizing macros [Misaka0x2730]
• Implemented support for other RTT channels than 0, and improved the documentation of the implementation [xobs]
• Overhauled the SWO implementation, implementing runtime switchable decoder selection [dragonmux]
• Fixed the Git ignore configuration to properly ignore the dependency clones [desertsagebrush]
• Switched the gate macro nomenclature for several things to CONFIG_<thing> for consistency [dragonmux]
• Implemented a proper non-halting memory access marker for targets (TOPT_NON_HALTING_MEM_IO) to replace the previous heuristic that was broken [dragonmux]
• Fixed a crash caused by too short a buffer being passed to a hexify() call in the remote protocol implementation [xobs]
• Fixed several bugs in the RTT implementation and improved some code clarity [xobs]
• Fixed a crash in the tdi_low_reset monitor command when run before any JTAG scans have yet been done [dragonmux]
• Overhauled the target mass erase system, pulling lots of common logic not the target API layer [perigoso]
• Improved the GDB packet handling system, reducing allocations and improving speed a bit [perigoso]
• Fixed a missing include for string.h in the GDB packet handler [xobs]
• Allow an external buffer to be supplied to the GDB packet handler [xobs]
• Overhauled the main project README [dragonmux]
• Fixed a pile of cast-based conversions warnings [dragonmux]
• Fixed a regression in the GDB server that meant block memory reads would inexplicably fail [ALTracer]

Build System Changes

• Introduced a new build system based on Meson [perigoso/dragonmux]
• Added a build target summary to the Meson build system [dragonmux]
• Improved the pre-commit hook setup for clang-format [amyspark]
• Fixed an encoding issue for pre-commit under Windows [perigoso]
• Fixed a typo in one of the warnings in the Meson build system [lasutek]
• Documented a small hitch with how PowerShell consumes Meson options lines [perigoso]
• Updated the BMDA dependencies libusb and HIDAPI to address some issues, particularly on M3 devices [dragonmux]
• Added a Meson Wrap fallback for the libftdi1 BMDA dependency [dragonmux]
• Fixed the build exploding under Meson on macOS due to BMDA dependencies needing a C++ compiler [amyspark]
• Added a guard for the CH579 support to help with Flash size [ALTracer]
• Made the at32f43x support optional in both build systems [ALTracer]
• Updated the targets enabled for larger Flash platforms under Meson [ALTracer]
• Updated the libopencm3 version used to get various fixes upstream has not yet merged [dragonmux]
• Implemented the ability to select the SWO decoder built into the firmware [sidprice]
• Cleaned up the build system a bit, fixing some target family description texts [dragonmux]
• Disable some of the targets by default for the f072, stlink and native probe platforms to get things to fit Flash [esden]
• Removed the Makefile-based build system [esden]
• Implemented better control over the targets that can be enabled and created a suite of profiles for native for what is enabled out the box [dragonmux]

Script/Utility Changes

• Added nix-direnv integrations [wucke13]
• Removed deprecated tooling [dragonmux]
• Fixed a typo in the ST-Link udev rules [WeissbMa]
• Fixed an error in the clang-tidy configuration which rendered it invalid [OmniTechnoMancer]
• Fixed up the udev rules to work with udev >= 247 [dragonmux]
• Updated the swolisten utility to properly walk the descriptors for the endpoint to use [ALTracer]
• Implemented a script for building firmware for all probes under Meson [esden]

Project CI Changes

• Upgraded the pre-commit check for accidental assignment to correct handling of char literals [perigoso]
• Added MSVC and macOS jobs for PR builds to help prevent regressions [amyspark]
• Worked around a macOS 14 SDK issue making CI fail [amyspark]
• Various fixes for GCC builds on macOS [amyspark/dragonmux]
• Overhauled the CI configs to use newer OSes, compilers and have more sensible build matrices for better code coverage [dragonmux]
• Made use of make's -j option in CI for a speed boost [ALTracer]
• Fixed CI breaking due to changes in the Ubuntu images used due to Python PEP 668 [dragonmux]
• Fixed the macOS version we used becoming deprecated by upgrading the configs [dragonmux]
• Added the Black Pill platforms back into CI as they got missed [ALTracer]
• Fixed an issue in the build-and-upload flow that results in improper version information being embedded [dragonmux]

ARM (ADIv5 and ADIv6) Changes

• ARM Debug: De-duplicated the SWD error handling logic across all supported probes [perigoso]
• ARM Debug: Implemented support for discovering and debugging Cortex-R cores [dragonmux]
• ARM Debug: Unified the Cortex-A and Cortex-R handling [dragonmux]
• ARM Debug: Implemented memory I/O diagnostics for Cortex-A/R cores [dragonmux]
• ARM Debug: Extracted and de-duplicated the SWD parity calculation logic, giving it consistency [ALTracer]
• ARM Debug: Used the AP banked registers to speed up Cortex-A/R instruction launches used for register and memory access [dragonmux]
• ARM ...

Stable Release v1.10.2

We are happy to announce the v1.10.2 point release of Black Magic Debug.

In this Release

• Fixed a remote protocol UBSAN issue in BMDA [dragonmux]
• Fixed an issue for BMP (native) that in some cases would inhibit USB detach for DFU from being detected by the host [tlyu]
• Fixed a firmware bug in the SWD turnaround handling code that could trigger SWD no-response in reaction to requests on sensitive targets [tlyu]
• Fixed an intermittent issue where during sporadic momentary high current draw from a target powered by BMP, BMP would cut power [esden]

Contributors to v1.10.2

We have had 3 individuals contribute 6 commits since the v1.10.1 release.

Contributor (Contributions)

dragonmux (2)
tlyu (3)
esden (1)

Sponsors

This project is sponsored in parts by:

• 1BitSquared - Design, Manufacture and distribution of open source embedded hardware
  development tools and platforms, as well as educational electronics.
  Thank you everyone who buys Black Magic Probes from 1BitSquared directly through our stores or indirectly through Adafruit. The hardware sales allow us to continue supporting the Black Magic Debug project.
• All the generous Patrons and GitHub Sponsors supporting esden’s work
• All the generous GitHub Sponsors supporting dragonmux’s work

Stable Release v1.9.3

We are happy to announce the v1.9.3 point release of Black Magic Debug.

In this Release

• Fixed a bug in the firmware-side remote protocol initialisation logic that can in certain circumstances lead to the firmware crashing during remote protocol operations [dragonmux]
• Fixed a bug in BMDA’s platform_buffer_write() for Windows which would make it wrongly skip chunks of the buffer to write and corrupt the data to send [dragonmux]
• Fixed the initial USB CDC ACM notification packets being sent from the improper endpoints in the firmware [brainsmoke]
• Fixed a remote protocol UBSAN issue in BMDA [dragonmux]
• Fixed an issue for BMP (native) that in some cases would inhibit USB detach for DFU from being detected by the host [tlyu]
• Fixed a firmware bug in the SWD turnaround handling code that could trigger SWD no-response in reaction to requests on sensitive targets [tlyu]

Contributors to v1.9.3

We have had 3 individuals contribute 9 commits since the v1.9.2 release.

Contributor (Contributions)

dragonmux (5)
tlyu (3)
brainsmoke (1)

Sponsors

This project is sponsored in parts by:

• 1BitSquared - Design, Manufacture and distribution of open source embedded hardware
  development tools and platforms, as well as educational electronics.
  Thank you everyone who buys Black Magic Probes from 1BitSquared directly through our stores or indirectly through Adafruit. The hardware sales allow us to continue supporting the Black Magic Debug project.
• All the generous Patrons and GitHub Sponsors supporting esden’s work
• All the generous GitHub Sponsors supporting dragonmux’s work

Stable Release v1.10.1

We are happy to announce the v1.10.1 point release of Black Magic Debug.

In this Release

• Fixed an issue with the usage text for the erase_ monitor commands [ALTracer]
• Fixed an issue with the BMDA J-Link backend which would result in LIBUSB_ERROR_OVERFLOW and the adaptor becoming functionally crashed [ALTracer]
• Fixed a bug in BMDA’s platform_buffer_write() for Windows which would make it wrongly skip chunks of the buffer to write and corrupt the data to send [dragonmux]
• Fixed Semihosting being broken in the firmware due to an incomplete fix after PR #1284 [ALTracer]
• Fixed the initial USB CDC ACM notification packets being sent from the improper endpoints in the firmware [brainsmoke]
• Fixed a memory leak of the AP structure for Tiva-C targets when performing scans for them [dragonmux]

Contributors to v1.10.1

We have had 3 individuals contribute 10 commits since the v1.10.0 release.

Contributor (Contributions)

ALTracer (7)
dragonmux (2)
brainsmoke (1)

Sponsors

This project is sponsored in parts by:

• 1BitSquared - Design, Manufacture and distribution of open source embedded hardware
  development tools and platforms, as well as educational electronics.
  Thank you everyone who buys Black Magic Probes from 1BitSquared directly through our stores or indirectly through Adafruit. The hardware sales allow us to continue supporting the Black Magic Debug project.
• All the generous Patrons and GitHub Sponsors supporting esden’s work
• All the generous GitHub Sponsors supporting dragonmux’s work

Stable Release v1.10.0

We are happy to announce the v1.10.0 release of Black Magic Debug. This will be the final minor release of the v1 series.

Here are some of the highlights:

• Implemented support for talking with SPI Flash directly
• The firmware now automatically binds drivers for the Trace and DFU interfaces on Windows
• Target power now has soft-start on native hardware v2.3+ to address potential brown out. This feature was also backported to all other native BMP hardware that has a digital target power control to remove power rail overshoot.
• And a bunch of newly supported targets:
  • MindMotion MM32
  • ST Micro STM32WB1X
  • GigaDevice GD32F4
  • NXP LPC43x0
  • NXP LPC55
  • ST Micro STM32C0xx
  • NXP LPC40
  • NXP/Freescale i.MXRT
  • GigaDevice GD32F2
  • GigaDevice GD32E5
  • ST Micro STM32H5
  • TI MSP432E4
  • HDSC HD32L110
  • Arterytek AT32F435
  • ST Micro STM32U5
  • Nordic Semi nRF91
• New remote protocol support for accessing SPI devices via the companion tool bmpflash

For more details please refer to the ChangeLog down below.

ChangeLog v1.10.0

Core Changes

• Handle continue-with-signal messages from GDB [zyp]
• Restructured the GDB main loop from blocking to polling [xobs]
• Cleaned up and fixed some style issues in the RTT code [dragonmux]
• Refactored out the JEP-106 manufacturer codes into their own header [dragonmux]
• Improved the GDB register API nomenclature and organisation [dragonmux]
• Fixed some formatting issues with ternaries [dragonmux]
• Improved the correctness of JTAG bitbanging due to the generation of setup-and-hold violations upsetting RISC-V targets [dragonmux]
• Implemented automatic sizing of the Flash write buffers for targets [perigoso]
• Fixes for the setjmp header used under Windows in the exception code [sidprice]
• Restore the old GDB detach behaviour from pre-v1.8 as the auto-reattach-on-connect behaviour was breaking certain debug integrations [tlyu]
• Fixed some issues that caused several nasty linker warnings under GCC 12 [dragonmux]
• Implemented JTAG IR quirk handling to deal with parts whose IR lengths cannot be simply detected [dragonmux]
• Fixed a regression in the semihosting implementation caught by koendv [dragonmux]
• Round up reads in the generic CRC32 implementation to the nearest multiple of 4 [zyp]
• Fixed a regression in the JTAG→SWD sequence code caused by generating one too few cycles [dragonmux]
• Fixed part of how JTAG scan implements IR lengths handling [dragonmux]
• Improved the debug logging subsystem and introduced error and protocol debug levels [dragonmux]
• Fixed an issue where running load followed immediately by kill from GDB would result in a crash [dragonmux]
• Added a parameter to mon reset that allows specification of the pulse length [silbe]
• Removed the JTAG scan IR lengths mechanism [dragonmux]
• Unified the help syntax used in mon help [silbe]
• Replaced all usage of strtol with strtoul to help reduce Flash usage [ALTracer]
• Switched to the reentrant strtok_r from strtok to fix a correctness issue and reduce Flash usage [ALTracer]
• Implemented the firmware side of a new SPI remote protocol [dragonmux]
• Implemented Microsoft OS Descriptors support and provided driver binding descriptors for the Trace and DFU interfaces [dragonmux]
• Improved connect-under-reset handling reliability [jelson/dragonmux]
• Reworked the JTAG bitbanging with-delay routines to solve a bug in how they work [dragonmux]
• Decoupled the general.h and platform.h headers to reduce build churn for low performance build hosts [ALTracer]
• Increased the SysTick frequency from 100Hz to 1kHz in the firmware [ALTracer]
• Refactored the integer-only stdio definitions for the firmware [ALTracer]
• Implemented Flash state tracking for targets that need more complicated handling [dragonmux]
• Various fixes for type mistakes, missing integer suffixes and debug level usage [perigoso]
• Fixed a regression which caused SWD to hang if configured for max speed [dragonmux]
• Fixed some type mismatches in the CRC32 implementation [dragonmux]
• Fixed an issue with the fake threads implementation that can hang certain IDEs [dragonmux]
• Improved the organisation of the remote protocol packet handling logic [dragonmux]
• Fixed a pile of -fanalzyer lints highlighting bugs in target code [dragonmux]
• Fixed some return types issues in the scan logic and switched swdp_scan → swd_scan (with a fallback) as part of phasing out the “SWDP” nomenclature [perigoso]
• Fixed a missing include for stdbool.h in the GDB packet code [xobs]
• Improved the safety around target power by implementing detection of backfeeding and cutting power when this is detected [dragonmux]
• Improved the wording of the documentation around vMustReplyEmpty packets [perigoso]
• Reduced the Flash usage of the JTAG devices table [dragonmux]
• Fixed a race condition and resulting NULL dereference in the morse code subsystem [jcamdr]
• Fixed -Wpedantic warnings about forward referencing of enums [perigoso]
• Fixed -Wpedantic warnings about empty initialiser lists [perigoso]
• Fixed -Wpedantic warnings about returns in functions that return void [perigoso]
• Fixed several warnings about using characters for array subscripts (-Wchar-subscripts) [perigoso]
• Fixed several warnings about local variable shaddowing (-Wshadow=local) [perigoso]
• Fixed several warnings about bad function casts (-Wbad-function-cast) [perigoso]
• Fixed several string formatting warnings [perigoso]
• Fixed -Wpedantic warnings about using range expressions in switch-case and array definitions [perigoso]
• Removed a dangling extern variable declaration that is unused in the code base [perigoso]
• Fixed some missing braces on array of structure initializers [zyp]
• Fixed -Wpedantic warnings about binary constants [perigoso]
• Consolidated the alignment macros originally defined for ADIv5 and moved them into their own header [perigoso]
• Fixed RTT hanging the firmware when the aux serial port is not being monitored on the host [koendv]
• Fixed a bug in the firmware-side remote protocol initialisation logic that can in certain circumstances lead to the firmware crashing during remote protocol operations [dragonmux]
• Updated documentation and comments to reflect the cleaned up and improved SWD support nomenclature [perigoso]
• Cleaned up in gdb_main_loop() and added some documentation to the target_halt_reason enumeration [dragonmux]

Build System Changes

• Don’t delete version.h when building from an extracted archive [esden]
• Fixed Git detection when using BMD as a Git submodule [esden]

Script/Utility Changes

• Fixed a section heading issue in the contribution guidelines [SludgeGirl]
• Updated the clang-format pre-commit hook to v16.0.2 [lenvm]
• Added clang-tidy configuration for identifier case checking [sidprice]

Project CI Changes

• Implemented CI for Windows [dragonmux]
• Updated the main build CI to use the GitHub Actions actions/checkout@v3 action [gemesa]
• Spelling, punctuation and grammar fixes in the GHA configs [gemesa]
• Updated the lint CI to use the GHA actions/checkout@v3 and pre-commit/action@v3.0.0 actions [perigoso]
• Updated the ARM compilers used in the PR workflows [esden]
• Restricted the CI flows with ENABLE_RTT=1 set due to the Flash usage limitations problems [dragonmux]

ARM (ADIv5) Changes

• Cortex-A: Fixes and improvements for Cortex-R5 support
• ARM Debug: De-duplicated parts of the ADIv5 implementation and re-introduced the SWD interface header/structure [dragonmux]
• ARM Debug: Added support for the Chinese STAR-MC1 core type [koendv]
• ARM Debug: Added the Nordic Trace Buffer (NTB) to the component identification table [dragonmux]
• ARM Debug: Fixed a whole pile of JTAG protocol regressions (see #1389 for details) [dragonmux]
• ARM Debug: Improved Cortex-M halt/resume correctness [dragonmux]
• ARM Debug: Fixed an issue with the Cortex-M DWT mask calculation [silbe]
• ARM Debug: Correctness fixes for AP handling, implementing AP type dispatch, TrustZone handling and some multi-drop noise [dragonmux]
• ARM Debug: Reworked early DP bringup to improve correctness and reduce code bloat [perigoso]
• ARM Debug: Begun separating out SWD multi-drop handling and untangling it [perigoso]
• ARM Debug: Improved the reliability of debug bringup to handle parts like the Ambiq Apollo 3 which get upset when debug resetting them too much [dragonmux]
• ARM Debug: Improved how JTAG DPv0 ID codes are handled to correctly identify more parts using them [dragonmux]
• ARM Debug: Refactored the Cortex handling to pull out commonality into a new implementation file ready for proper Cortex-R support [dragonmux]
• ARM Debug: Improved the Cortex-A support and its ability to function on the same AP as a Cortex-M core [dragonmux]
• ARM Debug: Improved handling of the OS Lock registers on Cortex-A targets to handle if the lock gets set between resumption and the next halt [dragonmux]

Target Changes

• stm32l4/stm32wb: Fixed the Flash erase error handling and detection [DrZlo13]
• lpc43x0: Implemented external Flash support [dragonmux]
• nrf51: Fixed handling for the recovery access port core register faking to allow attaching again [TheyCallMeNobody]
• ch32f1: Fixed load handling with CH32F103C8T6 parts [lasutek]
• ARM Cortex-M: Fixed a missing call to sam3x_probe() [AnantaSrikar]
• ch32f1: Improved support for CH32F103C8T6 parts [lasutek]
• stm32g0: Implemented support for STM32C0 option bytes erase [fabalthazar]
• stm32f1: Improved the STM32F1 option bytes handling correctness [dragonmux]
• lmi: Fixed TM4C129 identifications [dragonmux]
• lpc: Updated Flash registration to conform to the new APIs [dragonmux]
• sam3x: Fixed an issue that ha...

Relase Candidate V1.10.0-rc1

We are happy to announce the second v1.10.0 release candidate of Black Magic Debug. This will be the final minor release of the v1 series.

In this release

• Fixed RTT hanging the firmware when the aux serial port is not being monitored on the host [koendv]
• Fixed a bug in the firmware-side remote protocol initialisation logic that can in certain circumstances lead to the firmware crashing during remote protocol operations [dragonmux]
• Updated documentation and comments to reflect the cleaned up and improved SWD support nomenclature [perigoso]
• Improved the Cortex-A support and its ability to function on the same AP as a Cortex-M core [dragonmux]
• Improved handling of the OS Lock registers on Cortex-A targets to handle if the lock gets set between resumption and the next halt [dragonmux]
• Added suport for variants other than the i.MXRT1060 [tannewt]
• Fixed handling of old SFDP basic parameter tables which lack the field for the page program size [tannewt]
• Fixed the Cortex-A cores on the STM32MP15x inducing a hang on scan and probe [ALTracer]
• Fixed a crash in the LPC40xx probe routine during probe of unrelated Cortex-M4 targets that have FPUs [desertsagebrush]
• Introduced a workaround for buggy Tiva-C devices that result in the device answering for all AP numbers, leading to 256 target entries for a device that should have just one [desertsagebrush]
• Added support for ST Micro’s STM32MP15x SoCs (Cortex-M4 coprocessor core only) [ALTracer]
• Reduced the Flash usage caused by the new soft-start feature on the native platform [dragonmux]
• Added pinout documentation to the ST-Link platform README.md [dragonmux]
• Improved the clarity and documentation of the FTDI interface state structure and usage [dragonmux]
• Fixed a bug in BMDA’s ADIv5 access decoding for the protocol log level that caused AP accesses to always be reported as being for AP1 [dragonmux]

Contributors to v1.10.0-rc1

We have had 6 individuals contribute 52 commits since the v1.10.0-rc0 release.
Contributor (Contributions)
dragonmux (22)
ALTracer (13)
Scott Shawcroft (6)
perigoso (6)
Sage Myers (3)
Koen De Vleeschauwer (2)

Sponsors

This project is sponsored in parts by:

• 1BitSquared - Design, Manufacture and distribution of open source embedded hardware development tools and platforms, as well as educational electronics. Thank you everyone who buys Black Magic Probes from 1BitSquared directly through our stores or indirectly through Adafruit. The hardware sales allow us to continue supporting the Black Magic Debug project.
• All the generous Patrons and GitHub Sponsors supporting esden’s work
• All the generous GitHub Sponsors supporting dragonmux’s work

Relase Candidate V1.10.0-rc0

We are happy to announce the first v1.10.0 release candidate of Black Magic Debug. This will be the final minor release of the v1 series.

Here are some of the highlights:

• Implemented support for talking with SPI Flash directly
• The firmware now automatically binds drivers for the Trace and DFU interfaces on Windows
• And a bunch of newly supported targets:
  • MindMotion MM32
  • ST Micro STM32WB1X
  • GigaDevice GD32F4
  • NXP LPC43x0
  • NXP LPC55
  • ST Micro STM32C0xx
  • NXP LPC40
  • NXP/Freescale i.MXRT
  • GigaDevice GD32F2
  • GigaDevice GD32E5
  • ST Micro STM32H5
  • TI MSP432E4
  • HDSC HD32L110
  • Arterytek AT32F435
  • ST Micro STM32U5
  • Nordic Semi nRF91
• Target power now has soft-start on native hardware v2.3+ to address potential brown out. This feature was also backported to all other native BMP hardware that has a digital target power control to remove power rail overshoot. For more detail about this change refer to V1.9.2 release notes

For more details please refer to the ChangeLog down below.

ChangeLog v1.10.0

Core Changes

• Handle continue-with-signal messages from GDB [zyp]
• Restructured the GDB main loop from blocking to polling [xobs]
• Cleaned up and fixed some style issues in the RTT code [dragonmux]
• Refactored out the JEP-106 manufacturer codes into their own header [dragonmux]
• Improved the GDB register API nomenclature and organisation [dragonmux]
• Fixed some formatting issues with ternaries [dragonmux]
• Improved the correctness of JTAG bitbanging due to the generation of setup-and-hold violations upsetting RISC-V targets [dragonmux]
• Implemented automatic sizing of the Flash write buffers for targets [perigoso]
• Fixes for the setjmp header used under Windows in the exception code [sidprice]
• Restore the old GDB detach behaviour from pre-v1.8 as the auto-reattach-on-connect behaviour was breaking certain debug integrations [tlyu]
• Fixed some issues that caused several nasty linker warnings under GCC 12 [dragonmux]
• Implemented JTAG IR quirk handling to deal with parts whose IR lengths cannot be simply detected [dragonmux]
• Fixed a regression in the semihosting implementation caught by koendv [dragonmux]
• Round up reads in the generic CRC32 implementation to the nearest multiple of 4 [zyp]
• Fixed a regression in the JTAG→SWD sequence code caused by generating one too few cycles [dragonmux]
• Fixed part of how JTAG scan implements IR lengths handling [dragonmux]
• Improved the debug logging subsystem and introduced error and protocol debug levels [dragonmux]
• Fixed an issue where running load followed immediately by kill from GDB would result in a crash [dragonmux]
• Added a parameter to mon reset that allows specification of the pulse length [silbe]
• Removed the JTAG scan IR lengths mechanism [dragonmux]
• Unified the help syntax used in mon help [silbe]
• Replaced all usage of strtol with strtoul to help reduce Flash usage [ALTracer]
• Switched to the reentrant strtok_r from strtok to fix a correctness issue and reduce Flash usage [ALTracer]
• Implemented the firmware side of a new SPI remote protocol [dragonmux]
• Implemented Microsoft OS Descriptors support and provided driver binding descriptors for the Trace and DFU interfaces [dragonmux]
• Improved connect-under-reset handling reliability [jelson/dragonmux]
• Reworked the JTAG bitbanging with-delay routines to solve a bug in how they work [dragonmux]
• Decoupled the general.h and platform.h headers to reduce build churn for low performance build hosts [ALTracer]
• Increased the SysTick frequency from 100Hz to 1kHz in the firmware [ALTracer]
• Refactored the integer-only stdio definitions for the firmware [ALTracer]
• Implemented Flash state tracking for targets that need more complicated handling [dragonmux]
• Various fixes for type mistakes, missing integer suffixes and debug level usage [perigoso]
• Fixed a regression which caused SWD to hang if configured for max speed [dragonmux]
• Fixed some type mismatches in the CRC32 implementation [dragonmux]
• Fixed an issue with the fake threads implementation that can hang certain IDEs [dragonmux]
• Improved the organisation of the remote protocol packet handling logic [dragonmux]
• Fixed a pile of -fanalzyer lints highlighting bugs in target code [dragonmux]
• Fixed some return types issues in the scan logic and switched swdp_scan → swd_scan (with a fallback) as part of phasing out the “SWDP” nomenclature [perigoso]
• Fixed a missing include for stdbool.h in the GDB packet code [xobs]
• Improved the safety around target power by implementing detection of backfeeding and cutting power when this is detected [dragonmux]
• Improved the wording of the documentation around vMustReplyEmpty packets [perigoso]
• Reduced the Flash usage of the JTAG devices table [dragonmux]
• Fixed a race condition and resulting NULL dereference in the morse code subsystem [jcamdr]
• Fixed -Wpedantic warnings about forward referencing of enums [perigoso]
• Fixed -Wpedantic warnings about empty initialiser lists [perigoso]
• Fixed -Wpedantic warnings about returns in functions that return void [perigoso]
• Fixed several warnings about using characters for array subscripts (-Wchar-subscripts) [perigoso]
• Fixed several warnings about local variable shaddowing (-Wshadow=local) [perigoso]
• Fixed several warnings about bad function casts (-Wbad-function-cast) [perigoso]
• Fixed several string formatting warnings [perigoso]
• Fixed -Wpedantic warnings about using range expressions in switch-case and array definitions [perigoso]
• Removed a dangling extern variable declaration that is unused in the code base [perigoso]
• Fixed some missing braces on array of structure initializers [zyp]
• Fixed -Wpedantic warnings about binary constants [perigoso]
• Consolidated the alignment macros originally defined for ADIv5 and moved them into their own header [perigoso]

Build System Changes

• Don’t delete version.h when building from an extracted archive [esden]
• Fixed Git detection when using BMD as a Git submodule [esden]

Script/Utility Changes

• Fixed a section heading issue in the contribution guidelines [SludgeGirl]
• Updated the clang-format pre-commit hook to v16.0.2 [lenvm]
• Added clang-tidy configuration for identifier case checking [sidprice]

Project CI Changes

• Implemented CI for Windows [dragonmux]
• Updated the main build CI to use the GitHub Actions actions/checkout@v3 action [gemesa]
• Spelling, punctuation and grammar fixes in the GHA configs [gemesa]
• Updated the lint CI to use the GHA actions/checkout@v3 and pre-commit/action@v3.0.0 actions [perigoso]
• Updated the ARM compilers used in the PR workflows [esden]
• Restricted the CI flows with ENABLE_RTT=1 set due to the Flash usage limitations problems [dragonmux]

ARM (ADIv5) Changes

• Cortex-A: Fixes and improvements for Cortex-R5 support
• ARM Debug: De-duplicated parts of the ADIv5 implementation and re-introduced the SWD interface header/structure [dragonmux]
• ARM Debug: Added support for the Chinese STAR-MC1 core type [koendv]
• ARM Debug: Added the Nordic Trace Buffer (NTB) to the component identification table [dragonmux]
• ARM Debug: Fixed a whole pile of JTAG protocol regressions (see #1389 for details) [dragonmux]
• ARM Debug: Improved Cortex-M halt/resume correctness [dragonmux]
• ARM Debug: Fixed an issue with the Cortex-M DWT mask calculation [silbe]
• ARM Debug: Correctness fixes for AP handling, implementing AP type dispatch, TrustZone handling and some multi-drop noise [dragonmux]
• ARM Debug: Reworked early DP bringup to improve correctness and reduce code bloat [perigoso]
• ARM Debug: Begun seperating out SWD multi-drop handling and untangling it [perigoso]
• ARM Debug: Improved the reliability of debug bringup to handle parts like the Ambiq Apollo 3 which get upset when debug resetting them too much [dragonmux]
• ARM Debug: Improved how JTAG DPv0 ID codes are handled to correctly identify more parts using them [dragonmux]
• ARM Debug: Refactored the Cortex handling to pull out commonality into a new implementation file ready for proper Cortex-R support [dragonmux]

Target Changes

• stm32l4/stm32wb: Fixed the Flash erase error handling and detection [DrZlo13]
• lpc43x0: Implemented external Flash support [dragonmux]
• nrf51: Fixed handling for the recovery access port core register faking to allow attaching again [TheyCallMeNobody]
• ch32f1: Fixed load handling with CH32F103C8T6 parts [lasutek]
• ARM Cortex-M: Fixed a missing call to sam3x_probe() [AnantaSrikar]
• ch32f1: Improved support for CH32F103C8T6 parts [lasutek]
• stm32g0: Implemented support for STM32C0 option bytes erase [fabalthazar]
• stm32f1: Improved the STM32F1 option bytes handling correctness [dragonmux]
• lmi: Fixed TM4C129 identifications [dragonmux]
• lpc: Updated Flash registration to conform to the new APIs [dragonmux]
• sam3x: Fixed an issue that had Flash programming broken due to nRST interactions [dragonmux]
• stm32l0: Provided an implementation for mass erase for STM32L0 and STM32L1 [dragonmux]
• lpc55xx: Fixed a memory leak in lpc55x_add_flash() [francois-berder]
• target: Fixed an out of bounds array access in jtag_scan() [francois-berder]
• renesas: General documentation and cleanup [perigoso]
• rp: Fixed the Flash erase and write reliability issues caused by the ROM table calls hanging under certain circumstances [dragonmux]
• lpc55xx: Implemented support for LPC551x and LPC55S1x devices [dragonmux]
• lpc: Improved the correctness of the LPC IAP call mechanism and fixed some undefined behaviour [dragonmux]
• imxrt: Corrected some mistakes in debug formatting specifiers [xobs]
• ARM Cortex-M...

Stable Release v1.9.2

We are happy to announce the v1.9.2 point release of Black Magic Debug.

In this release

• Fixed building BMDA due to alloca not being defined. [hydra]
• Fixed an issue with the fake threads implementation that can hang certain IDEs [dragonmux]
• Soft TPWR ON: Mitigates target power enable brownout issue on v2.3 hardware [dragonmux] [esden]

Note: The brownout mitigation was also implemented for all older native hardware versions. Read below for more information.

We had reports from some users that the V2.3 hardware was resetting/crashing when the user enabled target power using the monitor tpwr enable command. During the investigation it turned out that the BMP was browning out. The GD32 found on the V2.3 hardware has higher minumum voltage requirement than the STM32. This means it will brown out more likely if the 3.3v rail dips. The solution to this problem was to slowly turn on the target power using PWM instead of instantly turning it on.

We also tested the solution on older hardware versions that use the STM32. Those were also experiencing a significant voltage drop. Additionally after the voltage drops the rail voltage overshoots by a significant amount. Because the overshoot we decided that it is a significant issue, as it can have adverse effects on the connected user targets. We decided to backport this fix to older firmware releases (1.8 & 1.9) as well as older hardware (all native hardware versions except for hardware version 0 as it does not have TPWR control).

To illustrate the problem and solution here are some oscilloscope traces:

HW4 without Soft TWPR ON:

HW4 with Soft TPWR ON:

Contributors to v1.9.2

We have had 2 individuals contribute 6 commits since the v1.9.1 release.

Contributor (Contributions)

dragonmux (5)
Dominic Clifton (1)

Sponsors

This project is sponsored in parts by:

• 1BitSquared - Design, Manufacture and distribution of open source embedded hardware development tools and platforms, as well as educational electronics. Thank you everyone who buys Black Magic Probes from 1BitSquared directly through our stores or indirectly through Adafruit. The hardware sales allow us to continue supporting the Black Magic Debug project.
• All the generous Patrons and GitHub Sponsors supporting esden’s work
• All the generous GitHub Sponsors supporting dragonmux’s work

Stable Release v1.8.3

We are happy to announce the v1.8.3 point release of Black Magic Debug.

In this release

• Fixed an issue with the fake threads implementation that can hang certain IDEs [dragonmux]
• Soft TPWR ON: Mitigates target power enable brownout issue on v2.3 hardware [dragonmux] [esden]

Note: The brownout mitigation was also implemented for all older native hardware versions. Read below for more information.

We had reports from some users that the V2.3 hardware was resetting/crashing when the user enabled target power using the monitor tpwr enable command. During the investigation it turned out that the BMP was browning out. The GD32 found on the V2.3 hardware has higher minumum voltage requirement than the STM32. This means it will brown out more likely if the 3.3v rail dips. The solution to this problem was to slowly turn on the target power using PWM instead of instantly turning it on.

We also tested the solution on older hardware versions that use the STM32. Those were also experiencing a significant voltage drop. Additionally after the voltage drops the rail voltage overshoots by a significant amount. Because the overshoot we decided that it is a significant issue, as it can have adverse effects on the connected user targets. We decided to backport this fix to older firmware releases (1.8 & 1.9) as well as older hardware (all native hardware versions except for hardware version 0 as it does not have TPWR control).

To illustrate the problem and solution here are some oscilloscope traces:

HW4 without Soft TWPR ON:

HW4 with Soft TPWR ON:

Contributors to v1.8.3

We have had 1 individual contribute 5 commits since the v1.8.2 release.

Contributor (Contributions)

dragonmux (5)

Sponsors

This project is sponsored in parts by:

• 1BitSquared - Design, Manufacture and distribution of open source embedded hardware development tools and platforms, as well as educational electronics. Thank you everyone who buys Black Magic Probes from 1BitSquared directly through our stores or indirectly through Adafruit. The hardware sales allow us to continue supporting the Black Magic Debug project.
• All the generous Patrons and GitHub Sponsors supporting esden’s work
• All the generous GitHub Sponsors supporting dragonmux’s work

Stable Release v1.9.1

We are happy to announce the v1.9.1 point release of Black Magic Debug.

In this release

• Fixed an issue with an include path making BMDA fail to build on Windows [sidprice]
• Fixed a missing probe call for ATSAM3x parts which caused a scan regression [AnantaSrikar]
• Fixed an issue where detaching from the firmware and then reattaching would reattach to the target, causing subtle breakage [tlyu]
• Fixed some pinout issues in the firmware for the native hardware [dragonmux]
• Fixed several build issues with BMDA on Windows which caused the build to be broken in some configurations of MSYS2 [dragonmux]
• Fixed some issues building BMDA on Windows under cygwin [hydra]
• Fixed make clean issue deleting version.h when building in a source archive [esden]

Contributors to v1.9.1

We have had 6 individuals contribute 15 commits since the v1.9.0 release.

Contributor (Contributions)

dragonmux (9)
Sid Price (2)
Taylor Yu (1)
Piotr Esden-Tempski (1)
Dominic Clifton (1)
AnantaSrikar (1)

Sponsors

This project is sponsored in parts by:

• 1BitSquared - Design, Manufacture and distribution of open source embedded hardware development tools and platforms, as well as educational electronics. Thank you everyone who buys Black Magic Probes from 1BitSquared directly through our stores or indirectly through Adafruit. The hardware sales allow us to continue supporting the Black Magic Debug project.
• All the generous Patrons and GitHub Sponsors supporting esden’s work
• All the generous GitHub Sponsors supporting dragonmux’s work