#milkymist IRC log for Tuesday, 2011-03-08

lekernelplasma's just a toy project01:36
kristianpaulhttp://www.inmojo.com/store/videoworks/item/milkymist-one/ <-- :-)03:08
awxiangfu, after using  http://milkymist.org/msd/msd-dec2010.tar.bz2 , what's the desktop's color you have with OS RTEMS 4.10.99.0?06:13
xiangfuaw: light-blue06:14
awafter I reflashing the new msd, my color from dark blue to light blue.06:15
awxiangfu, so your color is kept with the same old one?06:15
xiangfuaw: I can not remember the default RTEMS version. after reflash the desktop goto light-blue.06:17
xiangfuaw: here is two videos before and after flash: http://www.openmobilefree.net/other/downloads/Milkymist/06:18
awxiangfu, i can pretty sure your old desk top color must be dark blue! now mine is http://en.qi-hardware.com/w/images/6/68/M1rc2_0x2c_splash_wrongbackgroundcolor.jpg06:19
xiangfuaw: I don't know where to configure the desktop.06:22
awxiangfu, me either.06:23
awxiangfu, with xilinx tool, i think i can. but with 'jtag' tool i don't know.06:25
xiangfuaw: oh.06:26
awis there anyone that tried to click "reboot" button successfully on Flicknoise 0.2 about 6 ~ 7 times then later serial/vga screen msg jumped into >BIOS shown as "No boot medium found"?07:39
awthen type cmd 'reboot' in serial console then M1 shows up normal control panel again? What possible reasons will cause it?07:41
wolfspraulaw: I think lekernel already mentioned those might be software bugs somewhere later in the bootup process. nothing to worry about for you, I think.07:41
wolfspraulof course I agree it would be better to fix it earlier, even if in software, so we know this is not something that is caused by a hardware weakness still.07:43
awwolfspraul, okay..so it there somewhere that shows all happened bugs? i don't know yet..07:44
wolfspraulbut if it doesn't fix into the software priorities at this time, maybe we can safely assume it's not something we need to worry about on the hardware side?07:44
wolfspraulI don't think so. Maybe this is a good task for xiangfu ?07:44
wolfspraulxiangfu - can you try to reproduce Adam's problem? After you can reproduce it, can you try to track it down/fix it?07:44
wolfspraul:-)07:44
wolfspraulmaybe a bit too early/difficult right now, but if it's reproducible at least it's fun to go after an actual bug...07:45
awyeah..because you know that we are not familiar with fpga s/w itself. if some somewhere records them then I can also search current status. well07:45
awseems i have to record myself first.07:46
wolfspraulit's good that you are reporting strange things you see. we are aiming for a really stable and polished product.07:46
xiangfuwolfspraul: ok. I can try to reproduce the bug. but first I would like work on compile toolchain and rtems first :)07:49
xiangfuhttps://github.com/xiangfu/compile-milkymist-one/blob/master/Makefile07:51
awthe symptom on my site is easiler to reproduce it which i am not sure that maybe the M1 board h/w has been already haven something wrong in my  frequent "fast-power-cycling" before. so after I reflash with the new steps of 'jtag' tool, it can be easily shown on my board.07:53
awso  later if xiangfu can not easily reproduce my same symptom. then I can know it's my board problem probably. :-)07:54
wolfspraulaw: no. that sounds excessive.08:11
wolfspraullet's see... The problem you have means that the m1 will not fully boot up, right?08:11
wolfspraulbut the problem does not persist, i.e. you just need to power cycle again, and then it boots?08:11
wolfspraulyou don't even need to reflash to make it boot again, just power cycle?08:12
wolfspraullet's assume that's the case. then - you are not tracking down a known or suspected hardware bug right now, or are you?08:13
awwolfspraul, no, the symptom i described above is i tried to click 'Reboot' on s/w control panel. then later it jumped into >BIOS08:13
wolfspraulyes08:13
wolfspraulbut then you just power cycle and it boots again, correct?08:14
awthen I typed 'reboot' under >BIOS then it can boot and shows up control panel...yes if later I power cycle and it boots again.08:14
wolfspraulok08:14
wolfspraulthat means the impact of the bug to the end user is very small, we have a very easy 'workaround' (simple power cycle)08:15
wolfspraulin addition, you are not currently tracing down any suspected hardware bug, or are you?08:15
wolfspraulso it's just a behavior (bug) you noticed, and reported (which is good).08:15
wolfsprauldo you have a suspicion or idea how this could relate to something we can fix on the hardware/electronic/layout side?08:15
wolfspraulif not - just hand the bug over to xiangfu to reproduce and maybe fix, and forget about it :-)08:16
wolfspraulif xiangfu cannot reproduce it - no problem, not worth further investigations on your side, UNLESS you have a suspicion or idea how to fix it in hardware by yourself08:16
wolfspraulotherwise we get stuck with multiple people on problems of very little or no value08:17
awhmm..i am not working on suspected h/w ...i just curious that trying to see if any reset signals from fpga when I tried to click more time on "Reboot" button...08:18
wolfspraulyes, exactly. so you have no suspicion you are tracing/following right now.08:18
wolfspraulthat means - you hand the bug to xiangfu. but - whether xiangfu can or cannot reproduce it is not so important now, if he cannot reproduce it we just leave it sitting there.08:18
awsince I am going to do the patch on reversed polarity work, so meanwhile I played it for a while. :-)08:19
wolfspraulotherwise I feel we are hunting ghosts, and spending our valuable resources on that while we have far more valuable things that are totally open08:19
awwolfspraul, yeah..okay08:19
wolfsprauldefinitely, it's good that you report anything strange you see.08:19
wolfsprauleven if you just see it once, and then it goes away.08:19
wolfspraulbut if xiangfu cannot reproduce, that doesn't mean it goes back to you for further investigation.08:20
wolfspraulif he cannot reproduce, that means we leave it sitting just like that.08:20
wolfspraulUNLESS you have independent idea of your own what you think a hardware cause might be. If you don't have such an idea, leave the unreproducible bug sitting there.08:20
aw_yeah~leave it sitting there now. :-)08:22
wolfspraulyes. this is just my thinking, let's see what lekernel says...08:27
Fallenouhi xiangfu13:54
Fallenouwhat is your Makefile for ? to compile the lm32 toolchain ?13:54
xiangfuFallenou: yes.13:55
Fallenouok nice :)13:55
Fallenouis it working ?13:55
kristianpaulnice mm1 case feedback from wolfgang :-)14:07
Fallenousure14:08
kristianpaulxiangfu: hi !!14:11
xiangfukristianpaul: hi14:11
kristianpaulwow i was thiking about you right now ;)14:11
kristianpaulis adam using last firmware?14:11
kristianpauli read the backlog, but i just could not ask him..14:12
lekernelhi xiangfu14:12
xiangfulekernel: Hi14:13
lekernelwell, the different background is something to be expected and is totally normal :)14:13
lekerneldon't worry about such details...14:13
kristianpaul:-)14:13
lekernelthe app tries to load /flash/wallpaper.png and /memcard/wallpaper.png (in this order)14:14
xiangfulekernel: ok. one thing. can you please setup a account on milkymist wiki :http://www.milkymist.org/wiki/index.php?title=Main_Page14:14
xiangfulekernel: I would like username "xiangfu" :)14:14
lekerneland 0.2 uses the yaffs2 filesystem and cannot read the old flash files... therefore cannot load the previous background14:14
lekernelxiangfu: you can create an account now14:16
lekerneljust tell me when it's done so I can disable it again14:16
lekernelhttp://www.milkymist.org/wiki/index.php?title=Special:UserLogin&type=signup14:16
lekernelsame for others on this chan... if you need accounts14:16
xiangfulekernel: done . thanks. "Your account has been created. Do not forget to change your Milkymist Wiki preferences. " :)14:18
xiangfukristianpaul: thanks for the URL about compile lm32-rtems toolchain shell file: http://home.gwu.edu/~cssmith/LuaRtems/RTEMS_Tools.html14:41
kristianpauli wonder if it works..14:43
kristianpauli dint try it yet14:43
xiangfukristianpaul: me ether. but my Makefile work fine in my notebook :) now I got those commands: http://pastebin.com/uBBqshQG14:44
kristianpaulgreat !14:48
kristianpaultime to compile rtems bsp fot the milkymist one :-)14:49
xiangfuhow about add a default RTEMS_MAKEFILE_PATH value: +RTEMS_MAKEFILE_PATH?=/opt/rtems-4.11/lm32-rtems4.11/milkymist/15:13
xiangfukristianpaul: next step maybe try to write a makefile  include those http://www.milkymist.org/wiki/index.php?title=Flickernoise_build_instructions15:17
kristianpaul:D15:18
kristianpaulI think Fallenou and i will be really happy with the flicernoise makefile :-)15:18
kristianpaulflickernoise**15:18
kristianpaulxiangfu: (rtme_makefile_path) for now i guess is okay, but rtems version should be a variable easy to change15:19
kristianpaulbut there is long time until that i think15:20
Fallenouoh sure15:32
Fallenoubuilding the toolchain is not a problem for me15:32
Fallenoubut the Makefile is a good idea15:32
Fallenoubut building flickernoise is really a pain in the ass15:32
Fallenoutoo much dependencies15:32
Fallenoua Makefile is REALLY appreciated :)15:33
xiangfukristianpaul: how about this one: http://pastebin.com/Wh6YR9Qe15:35
Fallenoua makefile fetching all the flickernoise dependencies and then building them15:36
Fallenouand installing them15:36
Fallenouwould be awesome15:36
Fallenoumakefile or shell script, whatever15:36
xiangfuFallenou: I like makefile :)15:37
FallenouI'm fine with it :)15:37
Fallenouas long as I can type one command, go drink a coffee15:37
Fallenouand when I come back everything is built15:37
xiangfuFallenou: yes. that is the plan.15:37
Fallenouthen it's wonderful :) I am looking forward to it15:38
CIA-37milkymist: Sebastien Bourdeauducq master * r01b2c9e / (4 files in 2 dirs): Add GPL headers - http://bit.ly/fTvIsG15:39
Fallenoulekernel: thanks !15:58
kristianpaulxiangfu: yeah, looks good16:01
CIA-37flickernoise: Sebastien Bourdeauducq master * r95557d9 / src/Makefile : add default RTEMS_MAKEFILE_PATH (Xiangfu) - http://bit.ly/feqCqY17:48
CIA-37flickernoise: Sebastien Bourdeauducq master * r6bb8c45 / (4 files): clean up flash.sh a little, remove some duplicate code (Xiangfu) - http://bit.ly/dNfgBy17:48
Fallenoui've put the RTEMS_MAKEFILE_PATH in my .bashrc17:55
Fallenouso it's not really a problem17:55
Fallenoucompiling flickernoise dependencies is17:55
kristianpaul;-)17:57
Fallenou(sorry to insist :p)18:02
kristianpaulThe dma in mm1 is based on a bus mastering (i donk think so) or there is an dma controller (fmlarb) ?18:38
kristianpaulSorry if i'm not clear i still a bit confused undersand dma implementation18:39
Fallenouactually I don't know how the dma works18:40
Fallenouthat's a good question18:40
Fallenoulet me go back home18:40
FallenouI will try to figure that out18:40
Fallenou*going back home*18:41
kristianpaulFallenou: how works in sofware for you?18:41
kristianpaulis it already maped i guess19:01
kristianpaulsdsdsdsdsdsdsddsaaaa19:12
kristianpauldamit19:12
kristianpaulhmm xbar19:12
kristianpaulthe other bus was the fml19:13
kristianpauldefinittelly it is, i wonder now why stop checking it on sunday.. :/19:16
Fallenoukristianpaul: to do memory -> ethernet, I just give the start address, and the size to the ethernet core19:23
FallenouSo maybe it's the ethernet code itself which is doing the dma19:23
Fallenousame thing for ethernet -> memory transfers19:23
kristianpauls/dma/swich  :-) ?19:31
kristianpaulswitch19:31
Action: Fallenou is back19:40
Fallenouno no dma19:40
kristianpaulxbar.v is from sep 2010 so before..19:43
kristianpaulhmm :/19:43
mwallemh?19:43
mwalleethernet is a bus master19:44
mwalleso it does dma19:44
mwalleitself19:44
kristianpaulah19:44
mwallehave a look at system.v (iirC)19:44
mwallethere are all bus master listed19:44
kristianpaulyes, i'm just trying do understand the dma internals a bit19:45
Action: kristianpaul need add another core to hist local branch with dma support19:45
Fallenouis it Joachim Steiger who did the two case designs for Milkymist One ?22:54
larscyes, his nick is roh23:01
Fallenouok thanks :)23:03
Fallenoukristianpaul: your blog is down, isn't it ?23:13
--- Wed Mar 9 201100:00

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