<terpstra>
mwalle, hey -- sorry i didn't respond earlier. i had gone home for the night. :) re: the break instruction, yes it is a debug command. it seems somewhere along the line i ended up introducing an extra register of buffering the jtag interface. i intend to debug/fix this now. as for openocd, i need to figure out how to put an sld hub between the usb_blaster driver and the lm32 driver.
<wpwrak>
lekernel: your competitor. shall we pity them ?
<Fallenou>
humm nice
<lekernel>
competitor? not exactly :)
<Fallenou>
they are doing good stuff at microsoft :)
<lekernel>
ok license is crappy anyway
<lekernel>
You can use this Software for any non-commercial purpose, including
<lekernel>
distributing derivatives. Running your business operations would not
<lekernel>
be considered non-commercial.
<lekernel>
For commercial purposes, you can reference this software solely to
<lekernel>
assist in developing and testing your own software and hardware for
<lekernel>
the Microsoft eMIPS platform. You may not distribute this software
<lekernel>
in source or object form for commercial purposes under any circumstances.
<Fallenou>
Too bad they do such a work and then put a crappy license that makes it non usuable
<scrts`>
but it is available to buy?
<Fallenou>
dunno
<Fallenou>
seems like research stuff
<Fallenou>
maybe one day
<lekernel>
"if you sue anyone over patents that you think may apply to
<lekernel>
    the Software for a person's use of the Software, your license to
<lekernel>
    the Software ends automatically."
<lekernel>
lol?
<Fallenou>
I don't even understand this sentence
<wolfspraul>
that sounds like a legal trick to stop someone from claiming patent infringement
<wolfspraul>
a hardcore patent lawyer will probably not be worried, but who knows, I'm sure ms legal team also spent some time on this
<wolfspraul>
if you don't have a license, you cannot demonstrate the infringement :-)
<lekernel>
hehe, if this works, open hardware licenses should have that too
<terpstra>
mwalle, how do i get the openocd to startup in a valid state?
<terpstra>
this openocd project is a mess :-/
<terpstra>
the debug rom for the lm32 is quite nice, though
<terpstra>
maybe just write a new tool to talk to it instead of trying to fix openocd
<lekernel>
I have never touched openocd but it seems to be the de-facto "standard" for embedded debuggers
<terpstra>
it can't even maintain sync with gdb!
<terpstra>
comparing it to the grmon tool for the leon3 is laughable.
<lekernel>
which, unfortunately, does not mean it's technically good
<terpstra>
it nicely led me to libftdi, however.
<wpwrak>
(openocd) yeah, at openmoko it was also often a source of trouble, or at least an eager participant in trouble. i think one big problem is weak error recovery, so the daemon is often quite short-lived ...
<lekernel>
damn, handling relatively complex and big databases structures in C is such a tar pit...
<terpstra>
indeed
<terpstra>
i am thinking of a nice c++ library that does this properly
<terpstra>
it's not like jtag is a complex protocol!
<lekernel>
it's for representing the FPGA structure
<terpstra>
ah
<terpstra>
openocd also has a 'tar pit' of a database ;)
<lekernel>
we're dealing with multi-gigabyte text dumps to parse, and the resulting representation should enable relatively fast place and route algorithms to be made
<lekernel>
also, said text dumps have the peculiarity to refer to the same wire with a different name depending on what area of the fpga it is crossing, which adds to the mess
<larsc>
wpwrak: i think openocd has been rewritten several times since the ancient openmoko days ;)
<wpwrak>
larsc: oh, interesting :)
<larsc>
at least all the configs don't work anymore
<wpwrak>
larsc: a sure sign of progress ;-))
<mwalle>
terpstra: yeah openocd is a mess :) thats one reason i'm not very motivated to develop the lm32 port for openocd
<mwalle>
and you still see its roots in being an arm debugging tool, there are still hardcoded opcodes in some files (like the flash subsystem)
<kristianpaul>
mwalle: what about urjtag? i guess it can be use it for something else than flashing mm1 with fjmem
<mwalle>
yeah flashing but not debugging
<kristianpaul>
ah :/
<mwalle>
terpstra: "target reset halt", "target reset run" should work, you could try the lowlevel functions, "lm32 dp .." and "lm32 juart .." too