<kristianpaul>
lekernel: all this changes about the new wm chip are great, but i guess it break some compatibillity with the other chips, isnt?
<kristianpaul>
chip*
<kristianpaul>
so.. well i still having the old-one and other boards/people as well. will be nice to keep supporting this some how (by git?) or wait until the next bug-fix partie ;)
<kristianpaul>
(I'm aware this commets are too early, as adam dint tested the changes with rc2, but i wanted to comment it, before i forgot :-))
<roh>
kristianpaul: i see no reason to break support for the lm
<kristianpaul>
sure, me either but i'm curios about the current state of that, well i dint follow all commits sebastien did, so i may be talking bullshit, so i just wanted manifest some ramdon toughts about that topic
<kristianpaul>
compiles xiangfu last patches
<kristianpaul>
recompiled rtems for other xiangfu and noticed ac97's too
<kristianpaul>
so lets see :_)
<roh>
so far i understand it the ac97 code should be compatible with both codecs.
<roh>
no deviations besides detecting both not only one.
<roh>
the fixes for 'not garbling content of auxilary channels' shouldnt hurt the lm.
<kristianpaul>
indeed
<kristianpaul>
oh hell.. (No rx buffer left in the pool! )
<xiangfu>
I meet this error a lot after update the last rtems-milkymist.git.
<kristianpaul>
so you memory card allow you get the files i guess
<kristianpaul>
but this error was not before..
<kristianpaul>
hmm
<kristianpaul>
xiangfu: this may sound no-sense, but had you tried a 1GB kingstone memory card on the mm1?
<xiangfu>
kristianpaul: just send out.  too bad. I didn't find my 1GB card.
<xiangfu>
kristianpaul: maybe I goto buy one now :)Â Â back in 40 mins
<kristianpaul>
oh no, well just asking, i still need to understand how this memocard protocols works ..
<xiangfu>
kristianpaul: maybe I can help on memory card driver. since I work on nanonote memory card before.
<kristianpaul>
:-)
<kristianpaul>
I think mine also stoped boot process, but now seems like just were not there, well i test with scope and i confirmed CLK and some CMD signals, but still debug deeper
<kristianpaul>
i'm off for now
<xiangfu>
kristianpaul: forget taking money. he don't borrow me the 1GB memcard .
<kristianpaul>
;-)
<xiangfu>
Hi what is the meaning of "Milkymist" ?
<Fallenou>
hehe
<Fallenou>
Milky - like milk , mist
<Fallenou>
I think it comes from Milkdrop :)
<xiangfu>
thanks
<lekernel>
xiangfu: when do you get the "no rx buffer left" error? immediately after booting?
<lekernel>
it seems there is a reset bug of the ethernet PHY that causes it to continuously flood the soc with garbage RX packets
<kristianpaul>
i got mine when pulling data from mm1
<lekernel>
hmm... it's not this bug then
<kristianpaul>
when booting i just experienced, the old  message about fifo overflow just after a pld load soc.fpg
<lekernel>
yeah, that one is the reset bug
<kristianpaul>
ah :-)
<lekernel>
'pld load' seems to be a sure way to trigger it
<lekernel>
and it goes away when you do a soft reset of the system by pressing the three buttons
<kristianpaul>
yeap
<lekernel>
for some reason, it's pretty infrequent when the fpga boots from flash
<lekernel>
maybe I can try to make the FPGA assert the PHY reset for a bit more time after configuration
<kristianpaul>
i was about to ask if the same reset bug, i experienced this rx fifo flood in the past something when doing cold booting or just a "system reboot"
<kristianpaul>
is this a problem with un-initialized registers at boot? not been cleared by the sys_rst?
<lekernel>
I don't know. you can investigate this if you want :)
<kristianpaul>
;-)
<lekernel>
but I'd rather think the PHY either comes up in a "factory test" mode
<lekernel>
or just in a completely broken state
<kristianpaul>
he, well time to use MDIO at boot? :-)
<lekernel>
wolfspraul: hi
<wolfspraul>
lekernel: hi
<wolfspraul>
great results on wm9707, Adam will get a sample and confirm, then we switch to wm9707 in rc3, unless we misunderstood your message...
<lekernel>
what are your thoughts for rc3. go ahead with the wm codec? (after some more tests by Adam)
<wolfspraul>
yes sure
<wolfspraul>
what do you suggest?
<lekernel>
same
<wolfspraul>
we are already moving in that direction, Adam is getting sample already
<lekernel>
do you like the new GUI look better btw?
<wolfspraul>
haven't tried it on my m1 yet
<wolfspraul>
yes sure, the screenshots look much better
<wolfspraul>
lekernel: how do you feel about the transparent-blue color case?
<lekernel>
why not
<lekernel>
I think we can go ahead with that one for rc3
<wolfspraul>
ok great, Jon will be happy
<wolfspraul>
lekernel: how about engraving/lasering the Milkymist name and/or logo onto the bottom acrylic?
<wolfspraul>
have you and roh ever talked about that?
<wolfspraul>
I thought I remembered something, that was like 'doesn't look good' or so
<wolfspraul>
if that's the case I don't need to spend more time on it...
<lekernel>
yeah we didn't want to add it to the top
<lekernel>
to the bottom, maybe
<lekernel>
but not sure if it's worth the trouble
<lekernel>
we'll already have a sticker there
<lekernel>
and the ESD shield
<wolfspraul>
I think Jon meant some sort of engraving, like a shadow or so
<wolfspraul>
maybe only the 'M' logo
<wolfspraul>
on the acrylic
<wolfspraul>
at the bottom, definitely
<wolfspraul>
lekernel: OK I leave this with you. If nothing happens, fine with me.
<wolfspraul>
if roh wants to experiment, I am willing to pay for material and work.
<wolfspraul>
I don't have to have an engraved 'M' there, but I can roughly imagine what Jon has in mind
<lekernel>
wolfspraul: do you have an estimate when the rc3's will be out?
<wolfspraul>
end of may?
<wolfspraul>
I think we can target May, I mean we are working on details everywhere, I think it's coming together nicely.
<wolfspraul>
as of today, Milkymist One is not out of stock, don't forget
<wolfspraul>
I still have one unit, and I am sure there are more at Bearstech/Tuxbrain/kd85
<wolfspraul>
then there are some at potential distributors like Pulster, freedomincluded.com
<wolfspraul>
it adds up
<wolfspraul>
so I want to make sure we can generate a bit more interest
<mwalle>
mh i guess hardware breakpoint never works with the original gdb stub..
<mwalle>
lekernel: do you have acccess to the binutils (esp. gas) repo?
<lekernel>
no
<mwalle>
mh ok :)
<mwalle>
rcsr r1,BP0 should be an illegal instruction..
<lekernel>
maybe I can ask for it
<mwalle>
i have no idea how other assembler handle instruction combination which are not valid
<lekernel>
atm rtems is shipping 4.6 toolchains for all archs except lm32 and h8300
<lekernel>
the C++ compiler still wouldn't build...
<lekernel>
what a mess
<lekernel>
anyway, it didn't work before, so maybe they can simply remove it from the packages
<mwalle>
btw a reset through sysctl does a full soc reset, right?
<lekernel>
yes
<mwalle>
i needed a storage which is cleared on a reset. atm i'm using a uart scratchpad register, any other ideas?
<lekernel>
why uart?
<kristianpaul>
mwalle: a internal sram memory?
<mwalle>
doesnt really belong to uart, i should move that to sysctl
<mwalle>
kristianpaul: isnt cleared on reset, afaik
<kristianpaul>
well, of course you need implement the proper clear on reset
<lekernel>
i'd say add a scratchpad to the sysctl
<mwalle>
without any additional logic at least
<kristianpaul>
oh yes i agree with lekernel (sysctl)
<mwalle>
lekernel: yeah, sounds good :)
<mwalle>
btw enabled both breakpoints and watchpoint and it synthesized without any problems
<Fallenou>
pong kristianpaul
<kristianpaul>
Fallenou: hi !
<kristianpaul>
Fallenou: well, false alarm, i need check ac97 code again, i had some doubts but i discover the code was too old
<kristianpaul>
dunno why
<Fallenou>
I know nothing about ac97 btw
<kristianpaul>
yeah also realize that ;-)
<kristianpaul>
when i saw the sebastien name in last version
<CIA-43>
flickernoise: Sebastien Bourdeauducq master * rd391188 / (8 files): Use new filedialog API everywhere - http://bit.ly/eLXWI3
<lekernel>
#~#[|! ethernet is always crashing
<lekernel>
i'll move debugging that pesky crap upper in the priority list
<lekernel>
lots of time wasted
<lekernel>
rebooting etc. every time you want a file transfer
<CIA-43>
mtk: Sebastien Bourdeauducq master * r1f7e937 / (lib/edit.c lib/entry.c): edit+entry: delete selected text on backspace/del - http://bit.ly/hDCu3M
<lekernel>
kristianpaul: the DEL and backspace keys now behave as you expect in text boxes
<kristianpaul>
:D
<kristianpaul>
so i can swich back to screen!
<kristianpaul>
yipiee!
<lekernel>
screen?
<kristianpaul>
gnu screen :p
<lekernel>
huh?
<kristianpaul>
ah sorry
<kristianpaul>
sorry
<kristianpaul>
yes,you're talking about flicernoise patch editor
<kristianpaul>
:-)
<lekernel>
i'm talking about del/backspace not deleting selected text in flickernoise
<kristianpaul>
i missundertood with uart , yes yes
<kristianpaul>
sorry again
<kristianpaul>
ah ok
<lekernel>
UART backspace/del/editing, I will never fix. I'm fed up with that immense amount of cruft accumulated from the typewriter era you have to deal with every time you need a serial console
<kristianpaul>
is okay with his hacked version of flterm
<lekernel>
it works with gtkterm too
<lekernel>
screw the rest
<mwalle>
and gdb works too ;)
<kristianpaul>
:D
<Fallenou>
mwalle: so now we can put breakpoints ?