On Wednesday 23 December 2009, Laurent Gauch wrote: > On Tuesday 22 December 2009, Alex Austin wrote: > > I doubt this is the right way to do things, but it hasn't broken > > anything yet. This is just providing framework support for an adapter > > with reset lines driven by BDBUS instead of ACBUS. > > I am a bit worry seeing your code, and the idea to mix both A channel > (JTAG) and B channel (SRST) bus. > > If you really have to mix A and B for the same functionality as the > Amontec JTAGkey or other Dongles in the driver ft2232.c, you should have > to write a new driver as ft2232_mixed_channel.c . > You only complicate the code and make SPEED regression including your > idea/code in the actual ft2232.c.
The reason for the separation is that the FT4232 does not have *CBUS ports, creating the need to separate out the reset lines to a different device. As for speed regressions, they are minimal (+just a few clock cycles) unless you actually have streams of ft2232 commands for multiple ports. On Wednesday 23 December 2009, David Brownell wrote: > Hmm, maybe splitting the ft2232.c code into more manageable > chunks would suffice? That's a mess which *does* need to > be sorted out, IMO. For the post-0.4 merge window though. I could probably do this, but supporting the 4232 is time I can bill to my company, while driver refactoring is not (yet). - Alex _______________________________________________ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
