#milkymist IRC log for Saturday, 2011-04-23

kristianpaulhmm some missing (stall_ , err_ , lock_ , rty_, sel_, tga_ ) signla in milkymist wishbone-like soc00:32
kristianpaulerr_ looks like a should help no?00:34
kristianpaulat least current conbus is inmune to errors, by detect those ;)00:35
kristianpaulwell so ack is the only way of end bus cycle..00:52
kristianpaulah well, err_ will require async support in conbus.00:57
kristianpauland rule 3.40 defines the minimun MUST wich is what milkymist uses..00:58
kristianpaulnot bad 8-)00:59
wolfspraulOK I'm moving to Ethernet land, and of course first it won't work :-) Can't get any connection to my notebook08:02
wolfspraulon the m1 side, both the green and yellow lights are always on, starting from somewhere in the boot process08:02
wolfsprauldmesg says 'eth0: link is not ready'08:04
wolfspraul(this is with xiangfu's latest snapshot)08:05
lekernelphy problem08:32
wolfspraulis it normal that on the m1 side, both the green and yellow lights are on even without a cable inserted?08:36
lekernelno. as I said this is a symptom of the phy problem08:37
lekernel(the reset bug)08:37
lekernelreboot the board until it works08:37
wolfspraulhmm. looks like the reset bug is more serious than I thought. let's hope those things are gone in rc3. ok trying now, rebooting...08:38
lekernelI don't think this has anything to do with the 'no configuration' bug08:38
lekernelbtw, if you get the rtems shell, you can reset the PHY alone from there08:39
wolfspraulreset bug and configuration bug are two different things?08:42
wolfspraulare both fixed in our current rc3 plan?08:42
wolfspraulhow can I reset PHY in the rtems shell?08:42
lekernelwrite 1 to the minimac phy rst register, then 008:43
lekernelconfiguration reset bug is fixed, phy reset bug is fixable in the bitstream by making the FPGA continuously reset that piece of crap PHY until it works08:45
lekernel(if nothing else...)08:45
wolfsprauloh I see, the PHY reset bug is fixable in software, even on rc2?08:48
lekernelyes, should be08:50
lekernelfor some reason it didn't manifest itself as often with the previous minimac btw08:50
lekernelwolfspraul: does it work now?11:05
wolfspraulI just booted and the lights stayed off, need to try (give me 10min or so)11:06
wolfspraulyes sure11:13
wolfspraulit got a lot better before and since the Olympics though11:13
wolfspraulbefore the Olympics it would have been like on the right side > 50% of days11:13
wolfspraulmaybe even 80%11:13
wolfspraulbut now it's like that only 10% of days11:13
wolfspraulif even that11:13
wolfspraulthat's not to say that air quality is good, in no way. it just improved from horrendous to very bad.11:14
wolfspraullekernel: wow yes, I can ftp into the box now :-)11:18
wolfspraulanother bug to remember - Ethernet PHY reset bug...11:18
wolfspraulthanks a lot11:18
wolfspraulhow do I get to the rtems shell when m1 is running?12:26
wolfspraulI connected my jtag-serial usb cable, but unfortunately my kernel doesn't even recognize the serial port12:26
wolfspraulso I have to tell it about the serial port, then that's where the rtems shell lives?12:26
lekernelyeah, or you can telnet in (but that needs working ethernet *g*)12:33
wolfspraulcool, telnet work - great!12:39
CIA-48milkymist: Sebastien Bourdeauducq master * ra026067 / (3 files): BIOS: report rescue mode to the application using R4 - http://bit.ly/gX5ayA13:18
lekernelhmm what's that http://pastebin.com/WdEC0hWG13:23
kristianpaulsimulation compatibillity code?13:30
lekernelyeah but for what simulator?13:31
lekernelthe GDB thingy doesn't work, and qemu-lm32 wasn't there (afaik) when that code was written13:31
lekerneli'll remove it, it's not portable anyway (e.g. won't work with qemu-lm32 -M milkymist)13:32
kristianpaulhi trem_13:58
trem_hi kristianpaul13:58
lekernelxiangfu: just do away with keymap.c in one pass15:29
lekernelotherwise you're writing translation tables for keycodes that immediately become obsolete after15:29
CIA-48rtems-milkymist: Sebastien Bourdeauducq master * r853139d / (2 files in 2 dirs): Ethernet: minimac2 driver - http://bit.ly/fv8G7y15:39
CIA-48rtems-milkymist: Sebastien Bourdeauducq master * rfeab54e / c/src/lib/libbsp/lm32/shared/start/start.S : CRT: detect rescue mode through R4 and set BSP command line - http://bit.ly/ex63Xi15:39
CIA-48rtems-milkymist: Sebastien Bourdeauducq master * r3890d16 / c/src/lib/libbsp/lm32/shared/start/start.S : remove CRT cruft - http://bit.ly/ijYKsT15:39
CIA-48flickernoise: Sebastien Bourdeauducq master * r2c9f5e7 / src/sysconfig.c : Do not load configuration in rescue mode - http://bit.ly/hGgpyx15:47
kristianpaulah, cti_ is intended register for feedback_ , looks intereresting16:33
kristianpaul:o slaves may also support some burst types with cti, nice16:46
kristianpaulha, cti is no implemented on the milkymist soc.. i just see it asigned to masters including the lm32 core wich seems the only able to support it?..16:56
kristianpaulso classic cycle for now :-)16:57
guyzmofor flterm, I tried to use a pipe, without luck, and I found a poll-emulation library in the macports, but either I didn't succeed in compiling against it, or it is non working (I'm not sure, ldd does not show me the libpoll I compiled against)17:30
CIA-48milkymist: Sebastien Bourdeauducq master * re39899e / cores/minimac2/rtl/minimac2_ctlif.v : Do not assert PHY reset for too long - http://bit.ly/hHdW2I17:34
guyzmoanyway, I'm not using linux' ldd, but darwin's equivalent that is 'otool -L' and it shows me only one lib : "libSystem" ...17:35
lekernelguyzmo: i'm sure you'd have spent less overall time using system()17:57
tremnite all, sweet dreams21:34
--- Sun Apr 24 201100:00

Generated by irclog2html.py 2.9.2 by Marius Gedminas - find it at mg.pov.lt!