#qi-hardware IRC log for Friday, 2012-01-06

whitequarkany thoughts on this? http://ospid.com/01:41
mthincoming...03:34
mthapparently qi-bot either ignored the commit or git choked on it03:36
mthanyway, I just pushed jz-3.203:37
mthit runs on Dingoo and it compiles fine for NanoNote, but it needs testing on NanoNote as well03:37
mthlarsc, xiangfu: I did not merge all patches from 3.103:38
mthsome were no longer needed03:38
mththe hw I2C driver was not merged because larsc said it might not be worth the effort since the hw is really flaky03:40
mthboth the Dingoo and the NanoNote configs use i2c-gpio instead03:40
mththe "JZ4740 cache quirks" commit was not merged, to see what happens without it03:40
mthsince larsc said we don't really know whether it helps or not03:41
mth"Framebuffer notifier: Call notifier callbacks prior to blanking the screen" was not merged since larsc pointed to an alternative patch that solves the issue in a cleaner way: http://www.spinics.net/lists/linux-fbdev/msg04503.html03:42
mththat alternative is not merged yet either though03:42
mthso that's something to do03:42
mthI took a quick look at the alternative and it seems to add new callbacks, so I guess that for it to be a solution we should register those callbacks03:43
mthso what's required is not just a merge then, it's a merge + some new code03:43
mththere were a lot of mainline cleanups in the NAND code; it might be useful to review the remaining NAND patches03:44
mthflush_kernel_dcache_page was implemented by Ralf, but the implementation does nothing except a sanity check, so I'm not sure it actually fixes the problem we once had with MMC bounce buffers (when bounce buffers were accidentally enabled in the config)03:45
mthhttp://projects.qi-hardware.com/index.php/p/qi-kernel/source/changes/jz-3.2/03:52
mthI'm going to get some sleep now03:53
xiangfumth, I will try that in nanonote. 04:27
whitequarkmth: per my personal experience, the "cache quirks" part does not do anything at all on jz4750l04:51
whitequarkmaybe it was a workaround for earlier SoCs04:51
whitequarklike the msr0 quirk which is still needed04:51
whitequark*cp004:51
whitequarkhuh04:58
whitequarkdo you remember how I talked about iteadstudio, which is a clone of seeedstudio?04:59
whitequarkjust discovered a Malaysian Sparkfun clone04:59
whitequarkhttp://www.rocketscream.com04:59
whitequarkI'm curious06:56
whitequarkif I'd drive a 2kW heater with 50hz PWM, the light in my room will blink, and my refrigerator will die06:57
whitequarkand what if I'd use 1khz (with an SSR)?06:57
wolfspraulwhitequark: interesting link about ospid.com107:43
wolfspraulthanks, will check it a bit more but looks like we should definitely add it to the Qi planet07:44
GorDonFreeManhi09:24
GorDonFreeManwazup?09:24
GorDonFreeManhey Textmode 09:42
GorDonFreeMan;>>09:42
GorDonFreeManso you are a haxx0r09:42
GorDonFreeMansoo qi-hardware is an electronic company?09:43
GorDonFreeManor a nonprofit organization supporting electronic projects?09:43
whitequarkwolfspraul: I'll definitely buy that controller when it'll be in stock09:48
whitequarkthe author says that will happen in 4-5 weeks09:48
whitequarkI'm going to use it to make a reflow oven from a roaster. I really like almost everything about that piece of hardware09:49
wolfspraulvery good09:51
wolfspraulPLEASE PLEASE keep us posted about that project09:51
wolfspraulGorDonFreeMan: qi hardware is an open source project and definitely not 'non-profit' or 'non-commercial'10:00
wolfspraulwhat we do can be used for commercial or noncommercial activities10:00
wolfspraullike any other proper free software / open source project10:00
GorDonFreeMani'm planning to do both10:06
GorDonFreeManso, can you recommend things for my FPGA project?10:06
GorDonFreeMani have chosen some new 1.8V Xilinx FPGAs10:07
GorDonFreeManbut i would need to "program" them with something10:07
GorDonFreeManis there some c00l linux tool for this?10:10
GorDonFreeManpossibly not $$$ ?10:11
GorDonFreeManand not with forced registrations10:11
GorDonFreeMantracking/spyware/logging etc...10:12
whitequarkwolfspraul: absolutely. I'll following updates on their blog, and I'll tell everything interesting on IRC10:13
whitequarkand of course I'll write about my oven, when I'll do it10:13
whitequarkGorDonFreeMan: nope, there's only Xilinx tool and it sends your netlist to Xilinx, unless you've bought it for $1000 or so.10:14
GorDonFreeManaham i see10:15
GorDonFreeManso they allow you to make things that you give them10:16
GorDonFreeManthis delays my FPGA project 10:17
GorDonFreeManbecause i'm not contributing10:17
wolfsprauldo you have a link to prior projects of yours?10:17
GorDonFreeManno i have not made a blog yet10:18
GorDonFreeManin a few month i will have one10:18
wolfspraulhave you made a project before?10:18
GorDonFreeManwith Xilinx? no10:18
GorDonFreeManbut i have used several logic ics ;/10:18
GorDonFreeManand i can make everything with 1 FPGA10:18
GorDonFreeMani surely will benefit from this10:19
whitequarksigh10:22
GorDonFreeManmy next project is a 3 axis CNC machine10:24
GorDonFreeMan:P10:24
GorDonFreeMani would use an FPGA for stepper motor control too...10:24
blogic?10:25
GorDonFreeManthen... maybe a robot arm, walking spider or whatever :)10:25
blogici owuld say a fpga to drive the stepper controller :D10:26
blogici owuld like to see you drive a stepper with a fpga10:26
GorDonFreeMan;>>10:26
GorDonFreeMan;>>>10:26
blogicthe fpga will blow up in no time10:26
GorDonFreeManblogic<< ok not with the FPGA's ports, i'd connect mosfets to them10:26
GorDonFreeMan:P10:26
blogic?10:26
Action: whitequark looks at http://irclog.whitequark.org/qi-hardware/search?q=GorDonFreeMan10:26
blogicyou want to drive a stepper with a h bridge ?10:26
GorDonFreeManyou know, metal-oxide semiconductor10:27
whitequarkthat's... insightful.10:27
GorDonFreeManblogic<< well the simplest form is not a h-bridge10:27
GorDonFreeManit's single ended10:27
blogichave you used / seen a cnc machine before ?10:27
GorDonFreeManwhat is the meaning of that log whitequark ?10:29
whitequark4210:29
GorDonFreeManblogic<< if you ask me this i'm thinking over why am i here10:29
blogic2310:30
GorDonFreeMansqrt(-1)10:31
blogicre10:33
blogicGorDonFreeMan: dont get it10:33
blogicits a simple question really10:33
GorDonFreeManwell ou just assumed i'm retard10:34
blogic?10:34
GorDonFreeManou/you10:34
blogicwhere did i say that ?10:34
GorDonFreeManor troll10:34
blogic?10:34
GorDonFreeMan[112611] <blogic> have you used / seen a cnc machine before ? - here10:34
blogicwell10:35
blogicyou wrote you want to drive steppers with a FPGA10:35
GorDonFreeManyes10:35
blogicjust seemed to me that you have not used/built/setup a stepper controller10:35
GorDonFreeManso?10:35
GorDonFreeMani see10:35
blogicas they dont normally come with FPGA and FPGA is sort of a misplaced component10:35
blogicif you consider this to be an insult ... welll ...10:36
blogicwhat can i say10:36
blogicyou came up with the stigma of you being a troll10:36
GorDonFreeMani see FPGA as a HUGE collection of ANY logic ics10:36
blogicno one else did10:36
blogicok10:36
GorDonFreeManand i have stepper motor driver logic using logic ics10:36
blogici wish you lots of luck in your venture ;)10:36
GorDonFreeManso, i can make it like on a 20cm board10:37
GorDonFreeManand everything would fit in a tiny tqfp FPGA10:37
GorDonFreeManthis was my logic10:37
GorDonFreeManas a side effect, it would have a step rate limit of 300MHz...10:38
GorDonFreeMannot really matter for a stepper motor though10:38
whitequarkit will fit just as well in a tiny tqfp atmega10:40
whitequarkwhich is ten or fifty times cheaper, and easier to program, and does not need weird power voltage and filtered environment10:41
whitequark*voltages, of course. All Xilinx fpga I've seen use three input voltages10:41
whitequarkhave you ever tried to make a working buck converter?10:42
GorDonFreeManhm, yes, but microcontroller is not a logic block. and i can use the cheapest FPGA10:44
whitequarkit is10:44
GorDonFreeManno10:44
whitequarkproof: you can program FPGA to act like an AVR microcontroller10:45
GorDonFreeMana microcontroller executes instructions one after one10:45
whitequark(if you don't use analog peripherals)10:45
GorDonFreeManan FPGA is a HUGE logic block array10:45
whitequarkand if you can make a microcontroller out of it, then a microcontroller is a subset of a huge logic block array10:45
whitequarkergo, it is a logic block.10:46
GorDonFreeMan[114328] <whitequark> proof: you can program FPGA to act like an AVR microcontroller - ahah ok then i'll never use an atmel again10:46
GorDonFreeMan;>>10:46
GorDonFreeMantell me more about10:46
whitequarkdo you know about a thing called "Internet"? just in case10:46
whitequarkthere's a site in it no one knows about10:46
whitequarkit's called Google10:46
GorDonFreeManhmm, no please tell10:46
whitequarkhttp://google.com/search?q=fpga+avr10:46
whitequarkso hard, I understand10:47
GorDonFreeManthis is not what you said ;/10:47
GorDonFreeManthis will bring up lots of results interfacing FPGA to an AVR10:47
whitequarkjust curious, have you actually clicked on the link?10:47
GorDonFreeManno10:48
whitequarkbecause the first result is exactly what I've said.10:48
whitequarkcan you tell me your Hungarian street address? maybe I should come there and click a mouse button instead of you.10:48
GorDonFreeManahah10:49
GorDonFreeManok i clicked it10:49
whitequarkcongratulations10:49
wolfspraullet's be nice to each other :-)10:49
whitequarkI am10:50
whitequarkhave I said even one bad word to you?10:50
whitequarkno I did not10:50
GorDonFreeManhmm whitequark cool, i see someone took the time to make an FPGA act like an AVR8, but the AVR8 is still not a logic block10:51
GorDonFreeManmy point is, it executes instructions in serial, not parallel10:51
blogicwolfspraul: grouphug ?! :D10:52
GorDonFreeManin theory i can multiply arbitrary amount of floatingpoint numbers in parallel with an FPGA, do this with any atmel AVR.10:53
wolfspraulgrouphug! the qi-bot should have some feature to pass a matte cup around10:53
wolfspraulmate10:53
whitequarkGorDonFreeMan: sure, I'll just use an arbitrary number of AVRs10:53
GorDonFreeMan:) that will be inefficient, and slow.10:54
GorDonFreeManand expensive.10:54
GorDonFreeManwhitequark<< soo, since AVR is made out of logic blocks, you can make an AVR from an FPGA10:55
GorDonFreeManthis is the difference10:55
GorDonFreeManyou can make an X86 CPU from an FPGA...10:56
GorDonFreeManbut why would you ?10:56
GorDonFreeManit will be only a waste of logics.10:56
GorDonFreeMani see11:00
GorDonFreeMan"Having the core already implemented, you then only need to work with HDL for the parts of the project the AVR was unable to handle. "11:00
whitequarkGorDonFreeMan: how do you think Intel makes new CPUs? they get a huge FPGA from Xilinx and assemble it there11:00
GorDonFreeManbut still waste of logic blocks11:00
GorDonFreeMan;/11:00
GorDonFreeManok11:00
GorDonFreeMani'm not developing new CPUS11:00
GorDonFreeMani have an input data, i have a logic (black box), and i need an output11:01
GorDonFreeMani just do it using whatever method i want11:02
blogicwolfspraul: make that club mate bottles and its even better11:02
GorDonFreeManwhitequark<< so i see we have different aspects of problem solving11:03
whitequarkGorDonFreeMan: still, my point about AVR being a "logic block" (whatever do you mean by that) is avlid11:05
whitequark*valid11:05
GorDonFreeManan AVR is a huge macro of logic blocks.11:07
GorDonFreeManor rather a huge macro of macros of logic blocks11:07
blogic?11:08
blogicmay i ask something11:08
GorDonFreeManblogic<< what are your ?'s ? :)11:08
blogicwhats the aim of this discussion11:08
GorDonFreeMannothin'11:08
GorDonFreeManit's just waste of time11:08
GorDonFreeManit started with whitequark questioning me using an array of logic blocks (FPGA) for direct logic functions11:09
GorDonFreeManwhitequark<< what are your projects?11:11
GorDonFreeMando you have a blog?11:11
whitequarkhttp://github.com/whitequark; http://whitequark.org/11:11
blogicA pipelined brainfuck softcore in Verilog 11:12
blogic:D11:12
whitequarkyeah11:12
whitequarkit's quite fast11:12
whitequarkalso, very practical. As wpwrak said, "finally a device to run all of our brainfuck programs"11:13
GorDonFreeMani see so you're new to market :)11:13
blogicGorDonFreeMan: who do you mean ?11:13
GorDonFreeManwhitequark11:13
blogicwhy do you say he is new to the market ?11:13
GorDonFreeManblogic<< and what are your projects? :)11:13
blogicwww.openwrt.org dev.phrozen.org fon.com11:14
GorDonFreeManblogic<< well i looked at his website, still one more to go11:14
whitequarkNO U^W^W11:14
blogicwhyever you ask11:14
blogicok11:14
blogicthis is getting to bizarre11:15
blogicGorDonFreeMan: you scare me11:15
GorDonFreeManblogic<< just collecting intelligence, no threats11:17
whitequarkGorDonFreeMan: you know all the meanings of word "intelligence", do you?11:19
GorDonFreeManwell i don't know if you interpreted this as i intended11:19
GorDonFreeMani'm not native11:19
whitequarkoh11:21
GorDonFreeMani did not mean the basic meaning of intelligence11:21
whitequarkI have no doubt I interpreted it the right way11:21
GorDonFreeManwell currently i don't have the words to describe alternative meanings to you in english now11:22
larscmth: nice11:22
larscmth: but i think the new flush_kernel_dcache_page implementation is not sufficent11:23
GorDonFreeManwhitequark<< yet i don't know you are picking on me, or just probing me11:23
GorDonFreeManblogic<< hehe, you created the internetz on wifi ?:)11:29
blogicno11:29
blogicneither11:29
GorDonFreeMani was here http://corp.fon.com/en11:30
GorDonFreeMan"best way to share your wifi"11:30
GorDonFreeManneat  project hmmmm11:31
GorDonFreeManwhitequark<< http://www.fbi.gov/about-us/intelligence/disciplines11:32
GorDonFreeManto give you insight of the meaning i meant11:33
whitequarkoh yeah11:33
whitequarkFBI has a lot of intelligence11:33
GorDonFreeManFBI  Intelligence Collection Disciplines11:34
whitequarkGorDonFreeMan: can you give a hint why are you logged in from a machine called nude.lesbianbath.com?11:34
whitequarkI'm just curious11:34
GorDonFreeManno11:35
whitequarkanyway11:36
whitequarkI'm so proud I'm talking with an owner of a porn site11:36
GorDonFreeMan;>>11:36
pabs3anyone know if there are any FPGAs with FOSS bitstream preparers (or whatever they are called)? is this page accurate http://wiki.debian.org/FPGA#FPGA_and_the_DFSG ?11:43
GorDonFreeManwhitequark<< i see an advantage of the automatic irclog :) :"what you say can, and will be used against you"11:47
whitequarkabsolutely11:47
whitequarkpabs3: there are no FOSS tools of even alpha quality11:49
whitequarkthere are some experiments, some half-working ones, but nothing you can use for even a simple real project11:49
pabs3thats better than no tools at all11:50
whitequarkmarginally11:50
pabs3indeed11:50
whitequarkis a C compiler which cannot compile any code found in the wild better than no C compliers at all? no11:50
larsca car without tires is better than no car at all, but not if you want to get anywhere11:50
whitequark^ this.11:51
virichave you played much with 'perf' on the nanonote?11:51
GorDonFreeManyeah, you only need to make tires.11:51
pabs3whitequark: do you have any links to these projects?11:51
viric(if it works at all...)11:52
viricare there performance counters in the jz4720?11:52
larscviric: not sure, but i don't think so11:52
whitequarkpabs3: nothing atm11:53
viricok11:53
larscwe have hw timers, but thats all11:54
viriclarsc: but they can't count per-process, right?11:54
larsccorrect11:54
viricah well, at scheduling, they can be checked...11:54
larscwe use one of the timers as the clocksource for the system11:55
viricok11:55
whitequarklarsc: don't we do that already?11:55
viricwhat resolution?11:55
viriccan they be stored in a per-process variable at scheduling, to know the ticks per process?11:56
larscviric: i know nothong about perf, but i would expect that the kernel already does this11:56
viricok11:56
larscwhitequark: yes11:57
whitequarklarsc: oops, I read that as "can we use".11:58
whitequarkpabs3: look at Sebastien's projects: https://github.com/sbourdeauducq11:59
larsci've recently been toying with the idea of implementing a fpga on a fpga. don't know how feasible it is though.12:04
viriclarsc: it sounds like   adding '|cat|' piping12:06
viriclarsc: just for later making an asic?12:06
larscyou'd still depend on the proprietary tools, but it allows you to implement the whole stack12:06
viricyes12:06
blogicinteresting12:09
blogica uneducated question ....12:09
blogichow do you simulate efuses in an fpga ?12:10
blogici always thought the gates are "wired" using efuses ... which in turn i thought were not really logic elements12:10
blogicor am i totally wrong on that one ?12:10
viricwell, if you can emulate an fpga in software, you can emulate it in hw12:11
blogicyes12:11
blogichence my quetion ... how would that work12:11
blogicwell ok12:12
blogicyou can simply simulate it12:12
larscfpgas use lookup tables12:12
blogicit would mean though that the resulting fpga has a lot less power and cells/gates than the the host fpga12:12
blogiclarsc: can oyu make a sentence of that so its more clear ?12:13
viricthey use memories with values12:13
viricfor the operations12:13
viricLike you had logic tables of inputs and outputs12:13
blogicok12:13
blogicdid not know that12:13
whitequarkxilinx has a pretty good explanation in their datasheets12:14
viricyes12:14
blogichehe12:14
viricthere are many memories acting as lookup tables, each with a few flipflops12:16
blogici am a noob at fpgas 12:17
blogici installed ise last week and flashed my first "code" into a cpld last night12:17
viricwell, that's more than lots of people already12:18
blogici am currently looking for a spartan3 in the AN variant12:18
blogicas in the one with the internal flash12:18
GorDonFreeMan[125517] * roland (~quassel@phobos.martem.ee) has joined ##kernel12:18
GorDonFreeMan[125552] <roland> Hi. Does linux 2.6 support RSTP ?12:18
GorDonFreeMan[125959] <_7nb> nobody uses 2.412:18
GorDonFreeMan[130024] <roland> _7nb: only really old embedded devices12:18
GorDonFreeMan[130030] <_7nb> which are of no importance..12:18
GorDonFreeMan[130042] <roland> _7nb: to you12:18
GorDonFreeMan[130108] <roland> there are a lot of 2.4 embedded devices in industrial applications12:18
GorDonFreeMan[130124] <_7nb> Quantify that.12:18
blogic... on a nice breakout board12:18
GorDonFreeMan[130136] <roland> don't understand12:18
GorDonFreeMan[130142] <roland> quantify12:18
GorDonFreeMan[130143] <roland> ?12:19
GorDonFreeMan[130158] <_7nb> I always hear people saying "a lot of 2.4", but they don't know any numbers in comparison to 2.6/3.012:19
GorDonFreeMan[130754] <GorDonFreeMan> ;>>>12:19
GorDonFreeMan[130755] <GorDonFreeMan> ok12:19
GorDonFreeMan[130811] <GorDonFreeMan> there are a lot of flying cats around the world12:19
GorDonFreeMan[130818] <_7nb> Heh.12:19
whitequarkwhy the hell should all of us read that?12:19
GorDonFreeMan[130830] <GorDonFreeMan> -where ? how many ? - i don't know...12:19
GorDonFreeMan[130840] <_7nb> GorDonFreeMan: There is even a God. Or at least, they claim...12:19
GorDonFreeMan[130855] <_7nb> some even say there are many, but nobody has scientifically observed any.12:19
GorDonFreeMan[130856] <roland> you're been lied, there more than 1 god12:19
GorDonFreeMan[130857] <GorDonFreeMan> ohh i see you mean religion12:19
viricI don't want to read that12:19
GorDonFreeMan[130916] <_7nb> Yes, Linux 2.4 is a religion.12:19
GorDonFreeMan[130921] <GorDonFreeMan>                                                                                                      cool12:19
GorDonFreeMan[130946] <_7nb> Linux is communism (others say). Religion does not belong into socialist states. Therefore, you get the picture.12:19
viricfreenode, kick that out12:19
GorDonFreeMan[131020] <GorDonFreeMan> glad we figured this out12:19
GorDonFreeMan[131028] <GorDonFreeMan> next question?12:19
GorDonFreeMan[131052] <_7nb> What's for lunch?12:19
GorDonFreeMan[131136] <GorDonFreeMan> I think i'll have fried potatoes with fried cheese slices with a bit mustar you?12:19
GorDonFreeMan[131222] <roland> thanx, now I feel hungry :D12:19
whitequarkGorDonFreeMan: please, spare me from your presence.12:19
GorDonFreeManyou said i blog here12:19
GorDonFreeMan;>>12:19
viricthank you /ignore12:20
blogicaha12:20
blogicwhitequark: first i thought you meant me12:20
larscviric: yep, just did the same12:20
whitequarkblogic: ahem12:20
blogicthen i realized i am the only one to ignore him12:20
blogic /help ignore12:20
viric:)12:20
blogicits a feature made by gods12:20
GorDonFreeManok :)12:20
whitequarkwell12:20
whitequarkthat works12:20
blogicits magic12:20
viricI imagined the irc authors realized they needed it just at the first deployment12:21
blogicno12:21
whitequarkbut it's kind of ignoring (pun intended) the original problem12:21
blogicon the 2nd12:21
blogicon the 1st gordon joined :D12:21
whitequarkand I'm idealist a bit12:21
whitequarkand, what is worse12:21
whitequarkI run a logge12:21
whitequark*logger12:21
viric:)12:21
whitequarkand all that shit is going to show up there.12:21
whitequarksigh.12:21
viricqi-bot, ignore him, when registering the logs :)12:22
whitequarkwolfspraul: can you please ban the idiot? for the sake of clean logs.12:22
whitequarkviric: irclog.whitequark.org != qi-bot12:22
whitequarkthat's _whitelogger 12:22
viricah o12:22
virick12:22
whitequarkoh yes. http://irclog.whitequark.org/qi-hardware/2012-01-06#1325852469;12:23
whitequarkand I hope he'll never return.12:23
wolfspraulwell12:41
wolfspraulwe are lucky in having a very civilized and insightful chat and community12:41
wolfspraulsomehow those people came together12:42
wolfspraulso I am very careful about making the "open for anybody" sign any smaller or darker12:42
wolfsprauldo I think GorDonFreeMan could learn a thing or two? you bet12:42
viricwell, people ignored use to leave in a short time12:43
wolfspraulbut banning sounds a little rude to me right now, let's give him more chances to grow12:43
wolfspraulsure I understand, but there is very little to gain12:43
wolfspraulput him on a personal blacklist?12:43
wolfsprauland he is out of your view12:43
wolfsprauldon't feed the trolls12:44
viric:)12:44
wolfspraulso far in 2.5 years, I only kicked out one person once12:44
wolfspraulthat was a really insane guy on the mailing list12:44
blogicwolfspraul:  /help ignore ... a feature made by gods12:44
wolfspraulbut boy, he earned it12:44
virichehe12:44
whitequarkuh12:44
wolfspraulI do care very much for a friendly attitude though, and I hate to see people treated disrepectfully12:45
wolfsprauldisrespectfully12:45
wolfsprauland God whatever is on a nice path onto my watchlist there, definitely12:46
wolfspraulhope that's enough for now12:46
wolfspraulyes, /help ignore...12:46
whitequarkwell, maybe.12:46
whitequarkI don't know.12:46
viricso12:46
viricwhat's the kernel to run in the nanonote now?12:46
viricrtc issues... all that12:46
viricmaybe I've to keep with what I have.12:47
larscviric: xiangfu wrote a bug report to the person which introduced the rtc bug, but he hasn't responded yet :/12:50
GorDonFreeMan2012-01-06 12:22 <whitequark> wolfspraul: can you please ban the idiot? for the sake of clean logs. 12:54
GorDonFreeMan2012-01-06 12:22 <whitequark> viric: irclog.whitequark.org != qi-bot 12:54
GorDonFreeMan2012-01-06 12:22 <whitequark> that's _whitelogger 12:54
GorDonFreeMan2012-01-06 12:23 <whitequark> oh yes. http://irclog.whitequark.org/qi-hardware/2012-01-06#1325852469; 12:54
GorDonFreeMan2012-01-06 12:23 <whitequark> and I hope he'll never return. 12:55
GorDonFreeMan2012-01-06 12:44 <whitequark> uh 12:55
GorDonFreeMan2012-01-06 12:46 <whitequark> well, maybe. 12:55
GorDonFreeMan2012-01-06 12:46 <whitequark> I don't know.12:55
GorDonFreeManok12:55
viriclarsc: that happened outside of qi?12:59
viriclarsc: outside the qi branch I mean12:59
larscyes13:00
viricok13:00
viricdo you have any link about all that?13:01
larscseems to be a regression in the rtc core13:01
viriclarsc: for all mips?13:01
viricI've a mips64 here..13:01
larschttp://www.linux-mips.org/archives/linux-mips/2011-08/msg00140.html13:01
larscfor all linux I guess13:02
viricbut only on mips, no?13:03
larsci can't see a reason why it should be mips specific13:04
blogichttps://lkml.org/lkml/2011/1/20/306   <--ppc13:04
viricbut that's one year ago13:06
larscin the other thread he said he thought it got fixed13:06
viricaha13:07
viricso for 3.x, there is nothing new13:07
virichttp://lkml.org/  there is no 'search'? :)13:09
virichttp://lkml.org/lkml/2012/1/3/48513:09
viricmaybe it's not related13:10
viricI did not imagine that looking at lkml searching for 'rtc' would bring so much results :)13:14
viricouch, I can't build elfutils for the nanonote13:15
virici386_parse.y:1110:3: error: #error "bogus NMNES value"13:15
viricmy fault I imagine.13:23
virichm it's harder to cross-build than I imagined13:56
virichttp://comments.gmane.org/gmane.comp.sysutils.elfutils.devel/200513:56
wolfspraulwhitequark: I added osPID to the Qi planet - thanks a lot!14:25
whitequarkwolfspraul: that's good14:30
whitequarkwe need more oshw!14:30
wolfspraulyou bet14:33
wolfspraulalso mechanical, often overlooked14:33
viricqi planet...14:33
wolfspraulbut it is the *packaging* that makes electronics usable14:33
whitequarkoh14:34
viricI don't know it :)14:34
whitequarkhow came I forgot about that14:34
whitequarkhttp://printrbot.com/14:34
whitequarkit is an _awesome_ 3D printer. much cheaper than alternatives, and it can even upgrade itself14:35
C-Keenis it available now?14:37
whitequarkyou could backorder it back when the kickstarter project was pending14:38
whitequarkthey kind of got a lot of orders and are setting up the production now14:38
whitequark($830k out of $25k says _something_)14:38
viricgrr does openwrt have elfutils?14:39
whitequarkI can't wait to get one.14:39
C-Keenwhitequark: how does it compare to a pruja mendel?14:39
whitequarkthere is something minimalistic in it that feels exactly right14:39
whitequarkhm14:39
whitequarkthey've had a comparsion somewhere, let me find it14:39
C-Keenthe mendel seems to have the more active community it seems, although I am a bit scared by the amount of calibration time needed to get decent prints14:40
blogicviric: as a cross or host tool ?14:40
viriccross14:40
viricI've tried to make the double configure here: http://comments.gmane.org/gmane.comp.sysutils.elfutils.devel/200514:41
viricbut by now I fail a bit.14:41
viric:)14:41
whitequarkC-Keen: http://printrbot.com/storage/IMG_0626.JPG?__SQUARESPACE_CACHEVERSION=132544767692314:41
viricso maybe openwrt has a receipt14:41
whitequark(the notes read: "Prusa" "PB")14:41
blogicowrt has no receipt14:41
blogicwe have makefiles14:41
blogic:D14:41
blogiclet me look14:41
viricwhatever14:41
blogicfeeds/packages/libs/elfutils/14:42
viricI've seen...14:42
blogicis not whatever14:42
viricit builds only libdw and libelf114:42
viricmaybe I should try that. I only want those.14:42
whitequarkC-Keen: the build area is 5x5x5 in, it has a .5mm tip14:43
whitequarkactually, you can just read the FAQ on kickstarter: http://www.kickstarter.com/projects/printrbot/printrbot-your-first-3d-printer14:43
blogicwhitequark: nice14:44
blogicthe problem is the firmware14:44
blogicwe used the kubichek FW and imho its insane14:44
blogichunted lots of bugs and last night we setup a machine with emc2 and only use a atmel for the heater14:44
blogicfirst of all the motion planner uses the a axis as input for the acceleration of xyz14:45
blogicbasically someone took ghbl and patched it to be ready for 3d printing14:45
blogicand did so by simply adding A to every piece of code that references xyz14:45
blogicand thus totally breaking the setup14:46
blogicapart from the fact that GHBL's motionplanner is already code written while on crack14:46
blogicnonetheless .... once you get your setup printing its a nice experience14:46
C-Keenwhitequark: ah interesting14:47
whitequarkblogic: google does not know anything related to "kubichek firmware" 14:48
blogicwhitequark: let me get the real name14:49
blogicits the FW inside makerbot14:49
blogic90% of all atmega run 3d printers use it14:49
whitequarkI slowly start to think that atmegas aren't the best uC for small projects anymore14:52
whitequarkSTM32 are cheaper, much more powerful in processing power and have comparable, often wider, set of peripherals14:53
whitequarkthe main difference is 3.3V14:53
whitequarkbut they're 5V-tolerant14:53
whitequarkand that's not that hard14:53
blogicyes14:54
blogicthe thing is that for average joe there is lots of docs around on the web to get started with atmega14:54
blogic... imho14:54
whitequarkoh, do you mean the crappy examples and equivalently crappy half-done arduino libraries?14:55
whitequarkyes, there exists a lot of that. rather unfortunately, I'd say.14:56
blogicno14:56
blogici did not mean arduino14:56
whitequarkarduino is the php of electronics14:57
blogicthat is a c++ wrapper built on processing which in an ideal world is an abstraction layer above the actual silicon14:57
blogicmight be14:57
blogici am talking about the 1 trilion atmega related sites on the web14:57
blogicthe fact that a programmer takes 4 resistors and that its all documentd14:57
blogicmost other uC vendors dont have that support14:57
blogicand atmel has had this since early 200014:58
blogicwhen pic still sold the icd for 200¬14:58
blogics/pic/microchip/14:58
whitequarkwell you have wiggler14:58
whitequarkwhich is basically the same14:58
blogicyes14:59
blogicimho atmel is the easiest to get going14:59
blogicwhich imho is also important14:59
blogicof course eventually you start thinking about he sense and meaning of patterns and so on14:59
blogicbut if you have no clue and want to get tsrated you dont care for patterns15:00
whitequarkwell, I tend to agree15:00
blogicwhat i dont like is arm715:00
whitequarkafter all, it's extremely tolerant, and it has DIP, and that sites, and so on15:00
blogicthey tned to take a kloc to even come up15:00
whitequarkSTM32 is mostly Cortex-M3, which is thumb15:00
blogicyes15:00
blogicthey tend to be a pain to get booted15:00
whitequarkand also peripherals are really like atmel ones15:00
whitequarknope15:00
whitequarknothing like that.15:00
whitequarkmy boilerplate code to make a blinkie is... I dunno, two lines?15:01
blogici played with nxp lpc arm7 units and they liturally took 500.1000 lines of c before anything worked15:01
blogicah cool15:01
blogicthey improved it then15:01
whitequarkI even wrote a libc when I got bored15:01
whitequark10 lines of asm15:01
blogicthe lpc2318 was a nightmare15:01
whitequarkand some struct definitions15:01
whitequarkthat's all15:01
whitequarkyou struggle a bit trying to understand that you actually need to enable clock to GPIO module15:02
whitequarkbut after that, it's actually often easier15:02
whitequark32 bit registers allow for a lot15:02
blogichehe15:02
whitequarkperipherals are much more powerful and often much more sane15:02
whitequarkno bit-stuffing15:02
blogicand faster as the core clock is higher15:03
whitequarkyou've got 128 MiB of address space just for bitbanging15:03
whitequarkyes15:03
blogicdoes cortex have dma ?15:03
whitequarkyes15:03
blogicok +1 then15:03
whitequarkand it has a nested VIC15:03
whitequarkit's really good, and it is much much better than a conventional ARM IC15:03
whitequarkalso15:03
whitequarkon-chip debugging15:03
whitequarkthat has saved me a lot of time15:03
blogicusing jtag ?15:03
whitequarkyou don't have that with atmel.15:03
whitequarkSWD works too15:03
blogicok15:03
blogici hav a bdi300015:04
whitequarkI have a STM32 developer board15:04
blogicjtag always worked for me with that :D15:04
whitequarkit's $10, and I got it for free15:04
whitequarkit has two STM32's on it15:04
whitequarkone is for experimenting15:04
blogicah ok15:04
blogici have seen that unit actually15:04
whitequarkother one (which is like four times more powerful, which is funny) is the programmer15:04
whitequarkSTM32VLDISCOVERY?15:04
whitequarkyes15:04
whitequarkI like it15:04
whitequarkafter you get past the "ohh, that's not DIP" and "what programmer?.." stages15:05
blogichttp://www.mikrocontroller.net/attachment/86962/STM32Discovery.jpg15:05
whitequarkSTM32s are juts as easy as atmels15:05
whitequark*just15:05
whitequarkyes15:05
whitequarkthis one15:05
blogicand its still solderable by hand15:05
whitequarknot by a common person, I'd say15:06
whitequarkand not by a beginner15:06
whitequarkbut yes, it is15:06
blogicsoldering is a matter of patience and experience really15:06
blogicand the right tools15:06
whitequark(I have not seen a lot of people who can solder TQFP by hand. I can do up to LQFP, but that has taken quite a bit of practicising)15:06
whitequarkyes, a good soldering station helps a lot15:06
whitequarkalso, avoiding RoHS solder too15:06
rohhm.. btw.. i would wait on the pb thing. its not that liked in the foss 3d printing community15:06
blogicflux15:06
blogic:D15:07
whitequarkdefinitely15:07
whitequarkRMA has worked just fine for me15:07
blogichehe15:07
whitequarkat least all other things were worse15:07
whitequarklike: it's accidentally conductive (yick!)15:07
blogichehe15:07
whitequarkor: does not wash out.15:07
whitequark_and_ accidentally conductive15:07
blogici milled a atmega2u2 pcb last night15:07
blogicsoldered it and lfashed it just now15:08
blogicthat is the atmega with usb device in it15:08
blogicnice chip15:08
blogicrunning LUFA on it with the dual serial firmware15:08
rohtqfp is my limit too.. can solder it but below that there isnt a real chance.. cant get slimmer tips also15:08
whitequark(soldering by hand) also, I think half of arduinos being made now are actually with some kind of *qf*15:08
blogicroh: the trick is to solder and then remove solder for small structures15:08
whitequarkthere's a lot of arduino-like breakouts for STM32, too15:09
viricwolfspraul: the atom of the qi planet has 498 posts!15:09
blogicdoes it have usb device built in ?15:09
viricincredible15:09
rohblogic: i know... enough flux. doesnt help if you need to add small wires to tap something. then you really need to heat single pads15:09
blogicadding a ftdi is always too expensive15:09
whitequarkblogic: there are stm32s with usb devices15:09
whitequarkand usb OTG too15:09
blogicbrb wife fired a nmi15:09
blogic:D15:09
whitequarkeven more, they're actually pin-compatible if they're in the same case15:10
whitequarkregardless of family/model15:10
whitequarkgreat stuff15:10
whitequarkand hw multiplier15:10
whitequarkand so on.15:10
whitequark32-bit fixed point arithmetics in a couple of cycles per operation. I like that.15:11
wolfspraulviric: yes I increased it recently from 60 to 500 :-)15:12
whitequarkto summarize, I don't see a point of using atmel chips in any of my (new) projects anymore15:12
wolfspraulthat's because I found browser searching in the debian planet to be so nice15:12
wolfsprauland the debian planet has a huge long 1-page listing as well15:12
wolfspraulthe important idea behind the Qi planet is to look for high quality sources, feeds that post few but good articles15:13
wolfsprauland then combine them together into a stream that is still readable, let's say at most 3-5 posts per day15:13
wolfspraulbetter only 2 or so per day, but good ones15:13
wolfsprauleverything about open hardware and the people behind open hardware15:13
viricwolfspraul: 'browser searching'? What is that15:13
viric?15:13
wolfspraulctrl-f, as opposed to google15:14
viricah!15:14
viricI just put that into my offrss15:14
viricand it took a while to pick the posts :)15:14
viricwolfspraul: so, good with in-page search, but a pain for the browser :)15:15
wolfspraullet the browser work a little15:15
wolfspraul:-)15:15
wolfspraulthe limit is 500, not 50,000 :-)15:15
viricok ok :)15:15
wolfspraulif it's too bad, we can go down to 300, but I think it's ok15:16
wolfspraulnobody will read that many top-down, it's more for searching15:16
viricWell, I meant the 'rss'15:16
wolfspraulso people may read a bit at the top, page-down until they find an interesting picture or term that catches their attention15:16
wolfspraulor they search with ctrl-f15:16
virica list of post titles wouldn't be enough?15:17
viricwolfspraul: I'd like to see a loongson2f with firefox show that web :)15:17
wolfspraulwhat do you mean with post titles? you mean short rss with just the first paragraph or so?15:18
viricah15:18
viricwell, you talked about the web page15:19
viricI talked about rss at the beginning15:19
wolfspraulI have yet to find someone who would prefer that over the full text download, but if there are (maybe you?), then it's a feature to look for in planetplanet (which is mostly unmaintained unfortunately, I think)15:19
viricI'm always for full-text download - don't worry :)15:19
viricSo, I think I talked about rss, while you talked about the amount of posts in the plantet *webpage*15:20
viricI like a lot the contents of the planet, btw. :) Great job15:21
viricAnd what I said about 'wouldn't post titles be enough?', was about.. instead of showing 500 full articles in a single web page in the browser, you could have a place (another web page) where to show only post-titles, for in-browser searching.15:23
viricMaybe other browsers handle those 500 full text posts in a single web, but firefox 9.0.1 in my computer does not.15:24
wolfspraulok understood, thanks!15:26
viricin any case...15:26
viricdon't worry about me. I use my own rss reader :)15:26
pabs3whitequark: thanks, added to the wiki page15:46
viricok, cross-built setfont2...19:32
viricI imagine http://amadeus.dist.ro/setfont/setfont.tar.bz2 is not updated at all though19:33
virickyak: I run setfont2 with your cyrilic font, and I get only some horizontal small lines19:54
virickyak: the rest in black19:54
viricthat can mean I need a patch? I thought I'd have the kernel patched already20:02
viricwhat do I need to load properly a font with setfont2?20:11
viricaha, I don't have the patch20:13
viriclarsc: I can't find the setfont2 patches. neither any 3.x kernel patches.20:14
viricnot in git://projects.qi-hardware.com/openwrt-xburst.git20:14
viricnot in openwrt-trunk20:14
viricoh found. in 2.6.3720:16
virictarget/linux/xburst/patches-2.6.37/450-fbcon-color-fonts.patch20:16
kyakviric: there is the same patch for 3 series20:29
kyakyou can find nearby20:29
virichm20:34
viriccan't20:34
viricouch!20:34
viricI wasn't looking at master!20:34
kyak:)20:35
viricpatches-3.020:35
viricok20:35
viricI'm running 2.6.3620:35
viricmaybe I should take that of 2.6.3720:35
kyakthis patch is so small, it shouldn't take long to port even if it doesn't apply20:36
virichm not that small20:40
viricand openwrt-xburst.git is huge20:44
viric(to clone)20:44
viricto update the kernel.. should I use usbboot, or I can do that writing to something in /dev?21:04
viricI'd need the mtdtools or so I imagine, to first erase...21:04
larscyou can reflash the kernel using mtd.write21:18
viricwhat is that?21:20
virica mtdtools command?21:20
larscyes. but the name is nandwrite, sory21:22
viricah ok21:22
viricI'll try :)21:23
larschttp://en.qi-hardware.com/wiki/Updating_Ben_with_an_SD#To_re-flash_the_kernel21:23
larscAlthough the title says "updating with an SD" this also works from nand if you only update the kernel21:24
larscor uboot21:24
viricflash_eraseall is important too21:32
viricright?21:32
larscyes21:33
viricok, I built all22:20
viricmtdblock0 => uboot, mtdblock1 => kernel22:23
viricor I need "mtd"...22:24
viricok written...22:27
viricfailed to boot.22:28
viricperfect.22:28
viricoh22:29
viricflash_erase did not erase all 8 eraseblocks22:29
viriclarsc:  http://sprunge.us/ZfBK22:29
viricusbboot now22:34
viricok. now fails at 'verifying checksum'22:34
larscviric: the command is flash_eraseall not flash_erase22:35
viricouch22:35
viric:)22:35
viricand now, failed checksum! grr22:35
viricafter this  http://sprunge.us/WfaS22:35
viriclooks like the comparision worked22:37
larscviric: i would re-try at least once. usbtool tends to be unreliable sometimes22:39
virichm I retried once22:39
viricsame.22:39
larscalternativly you could boot from mmc card if you have one around22:39
viric http://sprunge.us/SLOM22:39
larscon the other hand, what's the size of your kernel image?22:40
viricslightly over 2MB22:40
viric2MiB22:40
viricData Size:    2126402 Bytes = 2076.56 kB = 2.03 MB22:40
larschm22:40
viricoh, uboot loads only 2MiB!?22:40
viricIt says:    D read: device 0 offset 0x400000 size 0x20000022:41
larscdo you have serial access to your nanonote?22:41
viricouch. Right. 2MiB.22:41
viricnot really... but I can build another uboot22:41
viricit's just about changing the load, isn't it?22:42
larsceither that or try to squeeze some bytes from the kernel22:42
viricI better load more.22:42
viric3.1 or 3.2 will be bigger I imagine22:42
viriclarsc: I have the uboot output on screen, not on tty :)22:43
viric#define CONFIG_BOOTCOMMAND  "nand read 0x80600000 0x400000 0x200000;bootm"22:44
larscchaning that should work22:44
viric0x300000 ?22:44
viricor I write 4? the partition is 4MiB big...22:45
viricthere we go with 422:46
larsci think there is even a command which reads the uImage header and extracts the size from it and than loads the remaining bytes22:46
viricah... is it?22:46
viricthat would be nicer22:46
viricI'll go with 0x400000 by now22:46
viric Checking 454656 bytes... no check! End at Page: 11122:48
virickernel booting, and... can't find rootfs22:49
viricas if it lacked ubi22:49
viricgrr22:50
viriclarsc: is 2.6.37 with the openwrt-xburst patches supposed to boot?22:50
larscyes22:51
larscalthough I haven't tried it in a whole22:52
larscwhile22:52
virichm22:53
viricsomething may be failing about ubi22:53
viricI hope nothing overwrote blocks...22:54
viricwasn't there a nread command for usbboot?22:55
larscyes22:55
viricI can't find the docs...22:56
larscah, right there is only ndump22:58
viricndump?22:58
viricno qiwiki page talks about ndump22:58
viricoh, "usboot -c help"22:59
larscno wait22:59
larscthere is nread23:00
larscbut all it does is prints the contents to the screen23:00
viricI see23:00
viricI just tried it23:00
viricreading 2GB that way... may hurt :)23:00
viric(I'm building the kernel 2.6.36 I was using, though)23:01
viricI put back 2.6.36, and it boots23:08
viricso it may be some new kernel option..grrr23:09
viricI have the exact ubi options between both23:11
viricoh # CONFIG_CRYPTO_ZLIB is not set23:18
viricI might be using zlib in my ubifs...23:19
viricsilly. I've UBIFS_FS_ZLIB enabled, and CRYPTO_ZLIB disabled though. The kernel could have complained.23:23
viricunless ubifs comes with its own zlib implementation23:23
viricah there is both deflate and zlib. weird.23:24
--- Sat Jan 7 201200:00

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