[Libre-soc-dev] versa_ecp5.py paramaterization update

[Cole Poirier]

On Mon, Oct 12, 2020 at 8:19 PM Luke Kenneth Casson Leighton
<lkcl at lkcl.net> wrote:

> On Tue, Oct 13, 2020 at 3:15 AM Cole Poirier <colepoirier at gmail.com> wrote:
>
> > Well I’m glad at least ppc64 is already gdb. So the server remote serial
> > protocol needs to be implemented for the chip?
>
> no.
>
> JTAG TAP is already done and is not a remote protocol.
>
> the *test* programs (firmware_upload.py, test_jtag_tap_srv.py) which
> connect *to* JTAG i have implemented the *openocd* jtagremote protocol
> "for convenience", by turning what was formerly a nmigen-only
> simulation test program (test_jtag_tap.py) into a *pair* of processes
> separated by the openocd jtagremote network protocol.
>
> this separation *has nothing to do with the chip*
>
> this separation *has nothing to do with JTAG TAP HDL*
>
> the only connection is that the jtagremote protocol happens to send
> bits back and forth that allows transfer of jtag bit-banging "raise
> TCK now please", "lower TDO now please", "i'm letting you know that
> TDI got lowered now".
>
>
> we do not need a gdb-server developer
>
> we do not need a gdb developer
>
> if we need anyone we need - and this is very very specific - a
> developer to add ppc64 support to openocd which implenents the
> microwatt-inspired DMI protocol over the JTAG-DMI gateway that i had
> to design a couple weeks ago.
>
> i found a patch that added or1k many years ago that shows exactly how
> that's done: it's not difficult, it's just time.
>
> in the meantime the programs such as firmware_upload.py although slow
> make it unnecessary and no longer a hard critical dependency to have
> ppc64 openocd "support"

Ah, fantastic! Thank you for clarifying, I'll raise it as a low
priority bug report then. "Add openocd ppc64 support which implements
the
microwatt-inspired DMI protocol over the JTAG-DMI gateway"?. Glad we
don't need to implement gdb.


Cole