Understanding the Problem
When debugging multi-core systems using Lauterbach Trace32, configuring the debugger for all cores in a synchronized manner can present several challenges. These issues typically arise due to incorrect synchronization, inconsistent core configurations, or improper handling of shared resources. To effectively debug such systems, a precise configuration is vital.
Synchronize Start-up Sequence
Each core needs to be initialized correctly and possibly synchronized before starting the debug session. Ensure you have appropriate scripts to automate and verify the startup:
; Sample Start-Up Script for each core
SYStem.CPU <core_id>
SYStem.Up
SYStem.Mode Attach
Data.LOAD.Elf <core_program.elf> /NoCODE /NoFILE
Ensure synchronization where required:
; Synchronize all cores at a specific point
Break.Set E:0xADDRESS /Program ; Set breakpoint at synchronization point
Configuring Core-Specific Debug Interfaces
Each core might have a different debug interface; thus, configure them correctly to avoid conflicts:
; Set interface for Core 0
SYStem.CPU CORE0
SYStem.JtagConfig 0x1
; Set interface for Core 1
SYStem.CPU CORE1
SYStem.JtagConfig 0x2
Ensure that each core's debug data and instruction paths are configured separately, if applicable.
Utilizing the AUTOSYNC Feature
AUTOSYNC in Trace32 ensures instructions are synchronized across cores. Enable this feature to maintain consistent execution states across cores.
; Auto synchronization across cores
SYStem.Option AUTOSYNC on
Shared Resources Considerations
Multi-core systems often allow cores to share resources like memory segments. Ensure these are handled correctly to prevent race conditions:
- Use specific memory access controls to avoid conflicts:
; Lock access to specific memory segment during operation
Data.AccessLock <address_range>
- Use debugger macros to verify that shared resources are properly initialized:
; Debugger macro for resource verification
&verify_shared_resources
{
IF (Data.LOAD.CHECK <start_address> <end_address> == FAILED)
(
PRINT "Mismatch detected in shared memory initialization!"
)
}
Core-Specific Breakpoints and Watchpoints
Multi-core debugging requires using different breakpoints for each core:
; Core 0 breakpoints
SYStem.CPU CORE0
Break.Set 0xADDRESS
; Core 1 breakpoints
SYStem.CPU CORE1
Break.Set 0xADDRESS
Ensure breakpoints are set appropriately for different cores to avoid unwanted behavior during debugging sessions.
Logging and Diagnostic Outputs
Capture diagnostic data separately for each core to analyze execution flow and resource usage:
; Enable logging for Core 0
Trace.STREAM.BUFFER &log_buffer0
Trace.List AUTOCORE.RC(core0) &log_buffer0
; Enable logging for Core 1
Trace.STREAM.BUFFER &log_buffer1
Trace.List AUTOCORE.RC(core1) &log_buffer1
Testing Configuration Robustness
Simulate stress tests to ensure the configuration can handle concurrent processing loads without breakdown:
- Develop test cases that emphasize shared resource access and synchronization.
- Continuously monitor CPU thread utilization statistics.
Documentation and Versioning
Maintain comprehensive documentation of your multi-core settings and sync it with version control for ease of troubleshooting and historical reference. Ensure all scripts and configurations are backed up regularly.
Conclusion
By tackling configuration challenges methodically and using the full suite of features provided by Lauterbach Trace32, multi-core debugging can be efficient and effective. Address each core's needs individually but keep the global state in mind to maintain consistency. Regular revision of these configurations as the firmware evolves is recommended to adapt to new developments.
