Re: [OpenOCD-devel] "reset_config none" vs "reset_config srst_only srst_nogate"

[Uwe B. <bo...@el...>]

>>>>> "Paul" == Paul Fertser <fer...@gm...> writes:

    Paul> On Sun, Mar 01, 2015 at 04:06:28PM +0100, Uwe Bonnes wrote:
    >> So my original question is changed to:
    >> 
    >> Why use "reset_config srst_only" and not "reset_config none"?

    Paul> reset_config none is default. If the specific board has SRST
    Paul> connected properly, it's usually more reliable to use hardware
    Paul> reset (SRST) to reset the target CPU (and probably other
    Paul> components), especially if the target doesn't support any good way
    Paul> for software reset.

I have some F411 nucleo board that I run battery powered. The reset line and
some other lines between the stlink and the target go to none 5 volt
tolerant lines of the STLink cpu and draw substantial battery current when
the stlink is not USB powered. A nucleo config with "reset_config none" will
work with both the modified and unmodifies nucleo board, while "reset_config
srst_only" will work only for the unmodified board.

Another situation is with using the STLink from nucleo or discovery to go
off board to another CPU. "reset_config srst_only" as given with most
configs, will require another jumper wire, while "reset_config none" as
default doesn't need that wire.

In general, I think we catch more situations and we require less assumptions
with SWD and probably JTAG with "reset_config none". And the user can always
supersede with -c "reset_config srst_only" in the rare cases needed, like
remapped SWD/JTAG pins.

So I don't buy the "it's usually more reliable" argument :-)

Cheers
-- 
Uwe Bonnes                bo...@el...

Institut fuer Kernphysik  Schlossgartenstrasse 9  64289 Darmstadt
--------- Tel. 06151 162516 -------- Fax. 06151 164321 ----------