accidentally invoked return jtag_execute_queue() in the
middle of a fn. Hmm.... I would have expected gcc or
at least lint to catch this.

Signed-off-by: Øyvind Harboe <oyvind.harboe@zylin.com>

the patchup code would get false positives when checking
whether a dbgbase had to be corrected.

The solution is to have autodetect default, with manual override
in scripts.

Signed-off-by: Øyvind Harboe <oyvind.harboe@zylin.com>

Do not require unecessary roundtrips for clocking out
data.

Signed-off-by: Øyvind Harboe <oyvind.harboe@zylin.com>

Signed-off-by: Øyvind Harboe <oyvind.harboe@zylin.com>

Could this cause confusion as data sent to write would be flipped
and then if the caller subsequently used the data, e.g. a
compare mismatch might happen?

Signed-off-by: Øyvind Harboe <oyvind.harboe@zylin.com>

Importantly adapter cleanup will now happen upon startup failure.

Signed-off-by: Øyvind Harboe <oyvind.harboe@zylin.com>

Signed-off-by: Øyvind Harboe <oyvind.harboe@zylin.com>

Signed-off-by: Øyvind Harboe <oyvind.harboe@zylin.com>

this fn does not fail, it queues data.

Signed-off-by: Øyvind Harboe <oyvind.harboe@zylin.com>

Signed-off-by: Øyvind Harboe <oyvind.harboe@zylin.com>

it would *read* instead of *write* to memory
when the MMU was disabled.

Signed-off-by: Øyvind Harboe <oyvind.harboe@zylin.com>

Signed-off-by: Øyvind Harboe <oyvind.harboe@zylin.com>

Really a Cortex-A specific option, but there is no
system in place to support target specific options
currently and there has been no need for such a system
until now.

Signed-off-by: Øyvind Harboe <oyvind.harboe@zylin.com>

production processor versions increment, thus the version
bits should be ignored for future proofing. e.g.
Engineering sample version == 0x00, production version 0x01

Signed-off-by: Øyvind Harboe <oyvind.harboe@zylin.com>

Hi,

This is a more descriptive message about LPC32XX error, when the nand
chip needs 5 address cycles.

Thanks.

Hi,

This will add support for a new nand chip device.

Thanks.

The patch below fixes step <address> on mips_m4k.

Spencer Oliver <spen@spen-soft.co.uk>:

The current code is used on all other arch's - is
there a underlying issue with those aswell ?

found via valgrind, not observed/reported.

Signed-off-by: Øyvind Harboe <oyvind.harboe@zylin.com>

Signed-off-by: Øyvind Harboe <oyvind.harboe@zylin.com>

Signed-off-by: Øyvind Harboe <oyvind.harboe@zylin.com>

better to keep this in a single file.

Signed-off-by: Øyvind Harboe <oyvind.harboe@zylin.com>

Signed-off-by: Øyvind Harboe <oyvind.harboe@zylin.com>

set cur_instr to BYPASS as optimisation code will
rely on checking the cached value.

Signed-off-by: Øyvind Harboe <oyvind.harboe@zylin.com>

Move ZY1000 OpenOCD driver more in the direction of being
a client to simplify code.

Signed-off-by: Øyvind Harboe <oyvind.harboe@zylin.com>

I don't think dsp563xx_once_read_register() would ever
be called with len==0, but it would have been broken in
that case.

Signed-off-by: Øyvind Harboe <oyvind.harboe@zylin.com>

Signed-off-by: Øyvind Harboe <oyvind.harboe@zylin.com>

Targets can implement read/write_buffer to handle
alignment.

Problem is, trying to print "Hello, world!\n" just prints endless H's, because r1 is never incremented.

One way to fix it would be to add a "++" after "r1".

Fix a bunch of typos.

Most are in code comments, so nothing should break. UNKOWN_COMMAND and
CMD_UNKOWN are not used elsewhere, so correcting the spelling should
also not break anything.

Signed-off-by: Øyvind Harboe <oyvind.harboe@zylin.com>

found by inspection.

Signed-off-by: Øyvind Harboe <oyvind.harboe@zylin.com>

pthreads work under Linux and eCos.

Signed-off-by: Øyvind Harboe <oyvind.harboe@zylin.com>

only flush write queue just before waiting for more data,
rather than when fetching more data from the buffer.

Signed-off-by: Øyvind Harboe <oyvind.harboe@zylin.com>

Found by inspection: the correct thing in the context is to use
usleep() rather than jtag_sleep(). Relates to JTAG over TCP/IP
only.

Signed-off-by: Øyvind Harboe <oyvind.harboe@zylin.com>

short sleeps are handled via usleep, longer sleeps we round up
to nearest ms.

There was a bug in jtag_sleep() in that it would round *down*
to nearest ms, thus making all <1ms sleeps 0. Found by inspection
rather than symptom.

Signed-off-by: Øyvind Harboe <oyvind.harboe@zylin.com>

Signed-off-by: Øyvind Harboe <oyvind.harboe@zylin.com>

Signed-off-by: Aaron Carroll <aaronc@ok-labs.com>