Changes
6.00 (pre-release)- The script API is cleanup up and rewritten. All the calls from EBlink now starts with "eb".
- Add you own GDB monitor commands or override existing. Examples in the script directory.
- Script functions are combined to simplify the API.
- The exceptions are implemented in the script engine for EBlink operations which simplifies script invocation (e.g. no interf.value32 etc)
- More and extensive error feedback on script errors.
- Introducing script debugging. Break and single step through code and watch local script variables.
- Introduction of script user dialogs/questions. with yesNof(...) questions can be asked to the user (e.g. "Should I reset Readlevel protection?" see stm32H7 script as example ) with default parameters for batch/shell use.
- EBbuffer now supports indexing. Access EBbuffer as an array of unsigned. Set the index access width (un8, un16 or un32) with the EBbuffer::access(width) methode.
- any many more...
Because all the scripts are modified but not tested, it is possible that some scripts do have some minor issues. Please repost them.
To use the script debugger, place EBlink in verbose level 7 (debug info) for function level debugging or level 8 (trace info) for line level debugging. For verbose levels below 7, debugging is turned off and breakpoints ignored.
To place a breakpoint in the script use the brkPt() function, e.g.:
Code:
foreach(arch in STMICRO)
if(arch[0] == targetApi.getCpuName()){
foreach(addr in arch[1]) {
try{
devReg = intrfApi.read(addr)
brkPt() <----------- Breakpoint
if(devReg) break
}
catch(e){}
}
- CubeIDE compatible again (was broken)
- Added multiple connections to the same services like GDB. CubeIDE is not supporting native GDB live variables but only in a Tandem-GDB configuration.
So under the hood, two instances of GDB are launched, connecting to the same probe port. One GDB is used for control and the other for
live variables queries.
EmBitz is using just one GDB instance and one connection because it uses live variables in native way.
5.20- Modified flow, increased speed and bandwidth by using the cache more consistently
- Reset strategy as cube programmer, more robust initialization
- Semihost repair for EmBitz, now also printing in target window if not standalone
- Fault unwind is enhanced with an additional setting "fc" to define the fault catch vectors
- Enhanced cache control (virtual,nocache,read-through etc)
- Enhanced memory access algorithm, useful for 16 and 32 bit access (high cycle H5 memory etc)
- Bugfix: fault unwind at startup
- Bugfix: Multicore debugging unexpected fault unwind's
- Bugfix: Target control on linux
- Bugfix: unhandled exceptions
- Bugfix: don't trap on nvic_reset
- And a lot of minor bug fixes
The code is rewritten to make it more robust and the STlink initialization is copied from Cube programmer (except hotplugging)
The fault unwind levels are renamed to 0=off, 1=passive and 2=active. This reflects the real meaning better.
5.12- - Renesas M33 flashing included
- - Added sector attribute "gdb_nc". With this sector attribute, GDB will read from chip rather than cache. Used e.g. with Renesas data flash (EEProm emulation) so that any data flash
- modifications are visible in the IDE. The RV40 flash script will skip dataflash (EEprom) erase on chip erase. This is to retain config data between chip erase cycles.
- If a full chip erase is needed then comment the lines in the rv40 script at the erase function (see code)
5.11- Bugfix GDB input handler. EBlink crashes if the size of the input was exact equal to the initial allocated input buffer.
- Bugfix progress always on top
5.10- Linux nullptr bug solved (strdup with null pointer doesn't return a null pointer)
- Windows progress bar added
- Flash script function per sector added as additional XML option
