Øyvind Harboe wrote:
> So... finally on to working on performance.
>
> Does anyone have any good ideas on how to profile
> OpenOCD?
>
> For PC hosted OpenOCD the most crucial thing is to reduce
> the # of roundtrips.
>
> Her are my first ideas.
>
> - Count number of times the queue is flushed using a global variable.
> This variable can be "watched" in the debugger and then look at it
> change as one steps through the code. This should reveal any surprises.
> Since this is easy to do, I'll do this first.
> - Use some sort of profiling tool that will give the stack trace of
> those places where OpenOCD is waiting for the driver to flush
> a queue.
>
>
You have been told but you dont listen.
Latest patch :
Warn : 133 45987 jtag.c:1361 jtag_check_value_inner(): TAP str710.cpu:
Warn : 134 45988 jtag.c:1371 jtag_check_value_inner(): value captured
during scan didn't pass the requested check:
Warn : 135 45988 jtag.c:1373 jtag_check_value_inner(): captured: 0x00
check_value: 0x01 check_mask: 0x0F
Warn : 136 45992 jtag.c:1361 jtag_check_value_inner(): TAP str710.cpu:
Warn : 137 45993 jtag.c:1371 jtag_check_value_inner(): value captured
during scan didn't pass the requested check:
Warn : 138 45993 jtag.c:1373 jtag_check_value_inner(): captured: 0x00
check_value: 0x01 check_mask: 0x0F
User : 139 46004 command.c:629 openocd_jim_vfprintf(): Runtime error,
file "embedded:startup.tcl", line 176:
User : 140 46005 command.c:629 openocd_jim_vfprintf(): examine-fails: -4
It is time to do some basic testing before submitting patches. This is
insane. No QC whatsoever.
/Magnus
_______________________________________________
Openocd-development mailing list
[email protected]
https://lists.berlios.de/mailman/listinfo/openocd-development
