#qi-hardware IRC log for Tuesday, 2012-01-10

DocScrutinizermoo05:42
wpwrakhow was survival in the first days of this new year ?05:45
Action: DocScrutinizer is a fan of forbidding use of short options in scripts ;-D05:46
DocScrutinizerwell, things get a lil bit less demanding05:46
DocScrutinizerplaying with Lauterbach is way more fun than ClearCase with proprietary policy enforcement wrapper XP05:47
wpwraklauterbach ... some little known contemporary philosopher ?05:48
DocScrutinizersome ugly hw in circuit emulator05:48
wpwrakah :)05:48
DocScrutinizerhttp://www.lauterbach.com/frames.html?country=de?home_d.html05:49
wpwraka neighbour :)05:49
DocScrutinizerI guess the expensive part has to be the complex software though, not the ugly hw which I can't see how it may cause costs of 15..20k per licence/unit05:50
wpwrakit's the invisible power of ignorance :)05:51
DocScrutinizerhehe05:51
DocScrutinizerD-85635 Höhenkirchen-Siegertsbrunn  -- a neighbour? not exactly05:51
wpwrakah . that was the fair05:52
wpwrak(nuernberg, the first thing you see when you load their home page)05:52
DocScrutinizerwell, since I don't travel to TPE 6 times a year anymore, my take on what's a neighbour probably changed a bit05:53
DocScrutinizero.O05:53
DocScrutinizeraaah, embedded world - yeah05:53
wpwrakyeah, in those days, anything within a lightmonth was "pretty close" ;-)05:53
DocScrutinizerhehe05:54
DocScrutinizerindeed05:54
DocScrutinizer"as long as it pings"05:55
wpwrakhow's the internal jet-lag ?05:55
wpwrakif it answers to ping, it's not dead yet :)05:55
DocScrutinizergetting better, awoke today after 5h sleep just 30min early of my alarm clock05:55
wpwrakyour superego is strong but inaccurate05:56
DocScrutinizeractually I had a weird dream about timeslices or sth that probably made me wake up05:57
DocScrutinizeryesterday I got a new "consultant" colleague - his acconts are working better after 4h than my whole account and PC installation after 4 weeks :-S05:58
wpwrakah, being haunted by work even while your brain ought to enjoy nocturnal wish fulfillment05:58
DocScrutinizeryep05:59
wpwrak(consultant) hehe ;-)05:59
DocScrutinizerthat's when I quit my last employment, some 30 years ago05:59
DocScrutinizerrationale: nobody paying me for the thinking done when I'm asleep05:59
wpwrakwow, all because of a bad dream06:00
DocScrutinizerunless I am paid for tasks rather than time06:00
wpwrakhmm, didn't this kind of role exist in the "minority report" universe ?06:01
DocScrutinizerunpaid role, yes :-D06:01
wpwraknot sure if they were too happy about their jobs, though06:01
DocScrutinizerdefinitely they weren't06:02
DocScrutinizerthe most annoying thing: now may days are too short to keep up the voluntary work I've done so far while also doing this 9-5 job06:03
DocScrutinizer8-6 actually06:03
wpwrakyeah, that's a normal effect of "regular work"06:05
wpwrakif you truly had plenty of time left, the work day would be longer06:05
Action: DocScrutinizer ponders to have another coffee or leave for work06:05
DocScrutinizerwpwrak: yep, so true, so sad06:06
Action: DocScrutinizer decides for coffee06:07
DocScrutinizer...and a bit more IRC chatting06:07
wpwrak;-)06:07
DocScrutinizercoffee, IRC, and marzipan :-D For minutes I love my life :06:10
wpwrakuntil you crash from the sugar high :)06:10
Action: DocScrutinizer idly ponders about the ~25% of your income you have to pay for health insurance here.06:17
wpwrakyou'll see the value of it when it comes to that expensive lung cancer medication :)06:18
DocScrutinizerlung cancer medication is probably 11g of lead06:19
DocScrutinizer~1$06:19
wpwrakheh, such things tend to disappoint the optimist as well as the pessimist06:20
wpwrakthe optimist thinks nothing bad will happen, yet it does06:20
wpwrakmeanwhile, the pessimist hopes for a swift ending, yet that won't happen either, for various reasons. but then, all this just means that life is full of surprises :)06:22
DocScrutinizerhmm, 24h disconnect, missed the pessimist part :-)06:24
wpwrakmeanwhile, the pessimist hopes for a swift ending, yet that won't happen either, for various reasons. but then, all this just means that life is full of surprises :)06:24
DocScrutinizerhehe06:24
wpwrakbut i guess i should give you a less gloomy perspective for today's journey :)06:24
DocScrutinizernah, no problem06:24
DocScrutinizerthis raspberry Pi board seems so extremely cheap in money... Might be worth to have a look at it06:27
wpwrakhmm. i realize that one of the difficulties of marketing milkymist is to understand how the various effects will work with the audience, considering that a good portion of them will be on ecstasy06:28
DocScrutinizerhehe06:28
wpwrakremember the $100 laptop ? watch the pi do the same sort of "inflation" :)06:28
DocScrutinizerwhich inflation do you mean?06:29
DocScrutinizerprices generally dropping?06:29
wpwrakdropping in relation to purchase power but increasing in numeric value06:30
DocScrutinizermhm06:30
wpwrake.g. OLPC sold their units for a lot more than USD 10006:30
DocScrutinizeryep06:30
wpwrakso maybe that was "USD 100" in some frame of reference06:31
wpwrakbut ...06:31
DocScrutinizerin the end a "normal laptop" was more economic06:31
wpwrakwith a little help from intel, to exterminate that amd weed06:31
DocScrutinizer"intel! hah! I feel like starting/continuing with tizen bashing06:32
wpwrakenjoy kicking the dead ? :)06:33
wpwrakwell, stillborn even06:33
wpwrakor maybe even zombified in the womb. who knows.06:33
DocScrutinizeryeah, here I actually do. They not only wasted everybody's time but also managed to ruin maemo, and meego06:34
DocScrutinizerwell, in ruining meego Nokia had their good share06:35
wpwraki'm kinda puzzled how they succeed in being so consistent in mis-managing this06:35
pabs3the git repos tizen folk released are pretty loltastic06:35
DocScrutinizerI can't help but thinking it's intentional to follow a secret plan06:35
wpwraki mean it's intel. c'mon. if they have a halfway decent plan, they can pull it through. they don't need to add sickly partners like nokia.06:36
DocScrutinizermicrosoft couldn't have come up with a better strategy to fsck linux06:36
wpwrakindeed :)06:36
rohDocScrutinizer: lets wait how long broadcom pays for that uboot ;)06:36
wpwrakof course, google are having a merry dance on that grave :)06:37
DocScrutinizerhi roh :-D06:37
wpwrakroh: probably until it goes olpc^Wbelly up06:37
rohwpwrak: olpc is brmc now also? i meant the rasberry 06:38
wpwrakroh: i meant the price explosion06:39
wpwrakroh: usd 25 is probably too low, considering volume06:39
rohwpwrak: wouldnt say that. considering they pay nothing for the chips and have not case...06:41
wpwrakyeah. no case = no volume :)06:42
wpwrakor, if volume > noise then case :)06:42
rohor do you mean something readymade?...  and even then.. for 16-25 euros one can buy complete routers with 500mhz cpus.. including case06:43
wpwrakthey seem to be aiming at the uk educational market. kinda like acorn did millenia ago.06:43
rohmy guess is that its only possible while brcm endorces the thing by sponsoring development and part cost06:43
wpwrakeventually, they'll have to pay for their bom. and then the price inflates.06:44
wpwrakand the tcos is much higher anyway, given that they're targetting pc-less households06:45
wpwraki.e., i don' think keyboards and mice are free in the uk06:45
wpwrakthey're of course cheap. but still, they have some transaction cost06:46
wpwrakanyway, time for a nap before the evil day star scorches the land06:50
Action: DocScrutinizer waves and heads out for a slaves' day of work06:55
rohthere it is.... (daystar)06:56
larscit hides behind the clouds06:58
Aylaguys, gmenu2x is broken here13:37
Aylathe ./configure builds a test program to check if SDL has been installed, but that code loads the SDL.h header using quotes13:38
Aylaie. #include "SDL.h"13:38
Aylashouldn't it be #include <SDL.h> instead?13:38
Aylaas a result the ./configure complains that SDL has not been installed13:39
mthSDL should be located using sdl-config, which will return a -I flag for the include path13:41
mthso both "SDL.h" and <SDL.h> should work13:41
AylaAFAIK, "SDL.h" means it's on the same directory13:41
Aylaand <SDL.h> means it's on one of the header paths13:42
mthboth "" and <> are looked for in the include paths, but the priority is different13:42
mthor maybe not the priority but whether system or current dir is checked as well13:43
mthsdl-config should be picked from the toolchain though, not sure if configure does that by default13:44
Aylait does13:44
mth./configure --host=mipsel-linux --enable-platform=dingux --with-sdl-prefix=/opt/opendingux-toolchain/usr13:44
mthI'm specifying the SDL location explicitly there13:45
Aylaok, I found the problem13:46
Aylathis works:13:46
Ayla./configure --host=mipsel-linux --enable-platform=dingux13:46
Aylathis does not:13:46
Ayla./configure --host=mipsel-linux-uclibc --enable-platform=dingux13:46
mthis the legacy toolchain in your $PATH as well13:46
mth?13:46
Aylano13:49
Aylathe "legacy" toolchain is long time gone ;)13:49
Aylamth: I am modifying how the cursor moves, I would like your opinion about this13:51
Aylacurrently, when you press Right the cursor will move up to the last column of the row, then wrap to the first column of the same row13:52
AylaI propose that the cursor wraps to the first column of the next row13:52
mthI'm not sure that would be better13:54
mthwhy would someone use that feature rather than using the down button?13:54
Aylacorrect, but then why should the cursor wrap at the first column when you press Right?13:55
mthif the cursor position is preserved after you launch something (is it?), you can use it to quickly go to the first column13:56
mthwith either scheme you can do that, but it's more straightforward if it only changes the column and not the row13:56
Aylait is saved yes, but I don't see why it'd be faster13:57
mthnot faster; what is faster depends on where the target you're going towards is13:57
mthit's simpler13:58
mthI guess it depends on whether you see the apps as a list or as a grid13:58
mthfor a list, switching to the next row on wrap makes more sense, for a grid staying on the same row makes more sense13:58
mthif an item is removed, the app are reordered though, so maybe it is more of a list actually13:59
viricDoes 'iostat' show anything for you in the nanonote, about the nand?19:32
viricmaybe I'd need some kernel options19:33
wpwrakDocScrutinizer: how was surviving the day ? :)20:40
DocScrutinizertime consuming as usual21:13
DocScrutinizerat least I managed to set my first breakpoint in lauterbach to a chosen position21:14
wpwrakso the rest of the day was spent with lavish celebration ? champagne, fireworks, ...21:14
DocScrutinizernext I probably have to learn ARM assembler, bot thumb and err whatever the other was21:14
DocScrutinizerboth21:15
wpwrakbtw, i have a mystery for you: in M1, we use the WM9707 codec: http://www.wolfsonmicro.com/products/codecs/WM9707/21:16
DocScrutinizer:nod:21:16
wpwraknow, it has line out and "line level" out. wolfson make no mention what "line level" means and how it would differ from "line"21:16
wpwrakthey don't even list the "line level" in the parameters, only "line"21:17
DocScrutinizererrr21:17
wpwrakalso, i haven't found any other wolfson codec that would have something like that21:17
wpwrakyou're my last hope. does any of this ring any bells ? :)21:17
DocScrutinizeralas not really21:18
DocScrutinizerI could only think of "line level" being a line-out with a non-matched impedance21:18
DocScrutinizera proper line-out should have ~1.5V @ 100Ohm21:19
DocScrutinizerI.E a 1.5V source with a 100R series max source impedance21:20
DocScrutinizerwhile a "line level" output might have a few kR source impedance21:20
wpwrakhmm. if it's that high, then it's useless without specification21:21
wpwrakfwiw, they define "line out" with a 10 kR load21:21
DocScrutinizer:nod:21:21
virichm21:21
viricI'm failing to use xbboot21:22
DocScrutinizer10kR load is "normal" for line-in21:22
wpwrakhttp://www.wolfsonmicro.com/documents/uploads/data_sheets/en/WM9707.pdf21:22
wpwrakpage 521:22
wpwrakokay21:22
viricError - set_addr() returned -11021:22
Last message repeated 1 time(s).21:23
larsc-110 is timeout21:23
viricmmh21:24
larscwhich means no response to a usb request21:24
viricError - get_info() returned -7121:24
viric ?21:24
larscno idea21:24
larscbut you can check errno.h21:24
viricah it's errno?21:25
wpwrakhmm. page 24 calls it "e optional stereo headphone out"21:25
larscviric: yes. 71 is EPROTO21:26
viric#define EPROTO      71  /* Protocol error */21:26
viricaha21:26
viricI wrote21:26
viric# xbboot -u 0x80600000 vmlinux.bin 21:26
viricit sid21:26
viricsaid21:26
viricInfo - found XBurst boot device.21:27
viricset_addr 80002000h21:27
viricbulk_write successfully wrote 6088 bytes.21:27
viricstart1 80002000h21:27
viricError - get_info() returned -7121:27
viricso, the numbers don't match21:27
viricwhat a big address I wrote...21:28
larsccould be the wrong firmware21:28
larscit has to load the first stage loader first21:28
larsci think21:28
virichm21:29
viricok21:29
larscor maybe something in your config is wrong21:29
larscwhich causes the jz4740 to crash21:29
viricwell, the address is wrong21:29
DocScrutinizerwpwrak: I think it's an absolutely identical auxiliary stereo out21:29
viricwhere physically are the 32MB?21:29
DocScrutinizerwpwrak: the line-out PGA is called "MASTER VOLUME" iigir21:30
wpwrak(master) yeah21:31
DocScrutinizerso "line-level" might just mean that this is _not_ controlled by "MASTER LEVEL"21:32
wpwraki think the chip may be some sort of clone for the LM4550B. but the LM seems to have a real headphone amp while the wm doesn't seem to be entirely decided on what it has21:32
wpwrakit seems to have just an independent pga21:33
wpwrak(reg 0x02 vs. reg 0x04)21:33
DocScrutinizeryep21:33
DocScrutinizer(not the (typ) in () there)21:33
DocScrutinizerwhatever that means21:33
wpwrakhmm, where's that "(typ)" ?21:34
DocScrutinizerin chip block21:35
larscviric: 80002000h is inside the icache21:35
wpwrakah, and wolfson recomment 10 uF caps, which would be 1.6 Hz with 10 kOhm. we have 1 uF. does this look like something we ought to change ? (our audio isn't supposed to be super high quality)21:35
DocScrutinizerC25 to C2921:35
DocScrutinizer10¼F21:35
DocScrutinizerOutput AC coupling caps to remove VREF DC level from outputs.21:35
DocScrutinizerhmm, I'd not worry about 4.7uF. ! though...21:36
DocScrutinizer121:36
DocScrutinizermaybe bearable21:37
wpwrakindeed, the "typ" is weird. "this it _typically_ the address of the volume control register" ? ;-)21:37
wpwrakkewl21:37
DocScrutinizerjust 10uF is a "de facto standard" for ine-out21:37
DocScrutinizerI'm not sure though about how much of non-linear distortion a small capacitor may introduce21:39
wpwrakand we have 1 uF blockers on input while wolfson recommend 470 nF. i guess that's nothing to worry about ?21:39
DocScrutinizerby offsetting the virtual 0-point for the amp output21:39
wpwrak(amp output) hmm21:40
DocScrutinizerno, too large C on input should be OK21:40
DocScrutinizerwell, nobody but me will notice any difference I'd guess21:41
wpwrak;-)21:42
DocScrutinizerthere are still users out there that argue 1uF for headphones on GTA02 is absolutely OK21:42
wpwraknow, one more mystery: the infamous "CD" input with its own DC-blocked ground. do you remember that critter ?21:42
DocScrutinizeryep, siure21:43
DocScrutinizerDC-decoupling ground feels odd21:43
wpwrakif you had to pick an audio input for expansion, but you don't know what people will do with it, would you pick that "CD" thingy (with its odd ground) or rather just AUX, which goes to audio ground ?21:44
DocScrutinizerhard21:45
DocScrutinizerwhat purpose of the input?21:45
wpwrakmaybe that CD ground thing is meant to break ground loops ? not sure if it works that way, though21:45
DocScrutinizerwpuldn't help21:45
DocScrutinizercould break DC shorts21:45
DocScrutinizerbut no 50Hz GND-loop hum21:46
wpwrak(purpose) unknown. it's for expansion circuits. if someone wants to add something (i.e., with a expansion board), they can plug into this21:46
DocScrutinizera differential input though could21:46
DocScrutinizerand you could route both minus-in of L and R to same GND level21:46
DocScrutinizerlike on a EKG where all differential opamps have minus-in on one common electrode21:47
wpwrakadam has found a rather enigmatic comment on that CD input: http://en.qi-hardware.com/wiki/Milkymist_One_Layout_Criteria#Audio_Codec21:47
wpwrak"This can improve the input SNR for a stereo source with a good common ground but precision resistors may be needed in any external attenuators to achieve the necessary balance between the two channels."21:48
wpwraki'm not entirely sure where between school physics and metaphysics this is located :)21:48
viriclarsc: I don't know what means 'inside the icache' :)21:49
DocScrutinizeryeah, a bit enigmatic but supporting my prev comment21:49
wpwrak(differential) so you're saying our "CD" thing may be this kind of differential input ?21:49
DocScrutinizeryep21:49
DocScrutinizeragain, what purpose has the input?21:50
viricxbboot -u 80010000 vmlinux.bin    this worked fine (the command), but I don't see the kernel booting :)21:50
larscviric: the CPUs instruction cache is mapped at 0x8000000021:50
wpwrakit would go to an unpopulated header footprint21:50
viriclarsc: I imagine I've to read a bit more then, to continue21:50
DocScrutinizerusecase21:50
wpwrakthe question is if that should be CD* or AUX*21:50
wpwrakDIY expansion21:51
DocScrutinizerthat's no usecase21:51
wpwrak;-)21:51
wpwrakthe one doing the DIY defines the specific use case :)21:51
larscviric: the bootrom is loaded at 0x80000000 which is used to load the first stage loader to 0x80002000, which will initialize the peripherals and then load the actuall payload21:51
larscthe kernel in your ase21:52
DocScrutinizerwell then define it as low-fi inmput and you're fine21:52
wpwrakso, should we use the traditional AUX + AGND or the fancy CD + CDGND for this ?21:52
DocScrutinizeryou should provision footprints for antiseriel Zeners21:52
viriclarsc: but the kernel has its own start address...21:53
DocScrutinizeruse CD21:53
larscviric: yes21:53
viriclarsc: isn't it... 0xffffffff8031d150 ?21:53
wpwrakperfect21:53
larsc0x8001000021:53
viricah ok21:53
DocScrutinizerfloating GND is a good thing for a genral purpose lo-fi input21:53
viricand what is the kernel "entry point"?21:53
viricanything relevant?21:53
DocScrutinizerthink about clipping input voltage at ~4V21:54
larscthe first instruction to be executed21:54
DocScrutinizerclamp diodes to VDD, VSS after C21:54
DocScrutinizerafter all 321:54
larscviric: do you load a compressed kernel?21:54
viricno21:54
DocScrutinizerplus a series R ~1k21:55
DocScrutinizerseries to C21:55
wpwraknaw, no clamps. let's keep this simple. we don't even know if this goes outside the box.21:55
DocScrutinizerhey, I said FOOTPRINTS21:56
DocScrutinizerquite obviously you also want some minimalistic RF blocking21:57
DocScrutinizer= 47pF to GND, after the series 1kR21:57
wpwraki'd just keep the rabble we currently have21:57
viriclarsc: that'd be 'vmlinuz', no?21:57
wpwraknothing RF-ish there, though21:57
DocScrutinizerRF is *everywhere*21:58
wpwraki mean rf-aware cicuitry21:58
DocScrutinizerand that's one of the disadvantages of such a electronically balanced input21:58
DocScrutinizerit *is* RF sensitive21:59
DocScrutinizerand NO GND to shield the RF away21:59
larscviric: yes. but the compressed kernel needs a different load address21:59
wpwrakoddly enough, wolfson themselves don't suggest any rf caps either. just dc block.21:59
wpwrakmaybe there's an internal low-pass ?21:59
viriclarsc: ok. No, I'm using vmlinux.bin21:59
viricso I try:21:59
viricxbboot -u 80010000 vmlinux.bin22:00
viricThen:22:00
DocScrutinizerwolfson doesn't suggest a lot of things, as they are not specific for their circuit. That'S common best practice22:00
viricxbboot start2 8001000022:00
DocScrutinizerwolfson also doesn't suggest to place receptacle there22:00
viricah wait22:00
viricit takes that as decimal22:00
wpwrak:)22:00
DocScrutinizerif that's 7mm copper trace to next chip's output, you'd obviously not need any RF measures22:01
DocScrutinizernor any clamp diaodes or a series R22:01
wpwrakdo you remember the circuit ? http://milkymist.org/mmone/rc3_schematics.pdf22:01
DocScrutinizeryes, I remember it22:02
DocScrutinizerI'm DocScrutinizer 22:02
wpwrakthe input path, from source: 6.8 kOhm series, the 6.8 kOhm to ground, then DC block, and finally the codec22:02
DocScrutinizernot sure about RC3 though22:02
viriclarsc: I'd say I want to upload the kernel to 0x80010000, and execute at the entry point.22:02
viricBut the xbboot "-u" means upload at X, and execute at X22:03
larscthe entry point should be 0x80010000,22:03
wpwraknow ... do we actually need those "pull downs" ?22:03
virichm22:03
viriclarsc: that's not the entry point in case of 'mkimage'22:03
larschm22:04
wpwrakor would the footprints be better used for your 47 pF caps ?22:04
DocScrutinizeryeah!!!! and the 6k8 in GND are WRONG22:04
wpwrak*grin*22:04
wpwrakdoes that apply to LINEIN{L,R} as well ?22:05
DocScrutinizeryou can't do such a thing to a stereo bal-in, with common GND22:05
viriclarsc: KERNEL_ENTRY is the address of the 'kernel_entry' symbol22:05
larscviric: hm, yes22:05
viric80316830 T kernel_entry22:06
wpwrakah yes, combined with the floating ground, it looks even more funny :)22:06
viric(says nm)22:06
larscfor the compressed kernel load and start address are the same22:06
DocScrutinizerbo, line-in has clean direct GND22:06
wpwrakbtw, you're still joerg@openmoko.org ?22:06
viriclarsc: ah ok. Maybe I need a loader for xbboot '-u' to work22:06
viricI'll go with the compressed22:06
DocScrutinizers/bo/no/22:06
DocScrutinizersure22:06
viriclarsc: it fails to build, btw22:07
viric(vmlinuz)22:07
larscviric: 0x80600000 is the address22:07
viricarch/mips/boot/compressed/decompress.c:105:2: error: implicit declaration of function decompress22:07
wpwrakokay, so no pull-downs on CD* but keep them on LINEIN*22:07
viricah I need kernel gzip.. 22:07
DocScrutinizerreplace the pulldowns by 47pF22:07
wpwrakis there a problem with having 6k8 instead of the 1k you suggested ?22:08
larscviric: or another compresor22:08
viricok22:08
larscdecompressor22:08
viricI can't find the option :)22:08
DocScrutinizerall of them22:08
viric'F8' in nconfig says it's in General setup...22:08
DocScrutinizererr, not on the outputs though22:09
larscviric: 'Kernel compression (...)' or something like it22:09
viric"Kernel compressino choice"....22:09
viricI may be blind :)22:09
viricah it's disabled22:10
viricI could make nconfig "show all options"22:10
viricit shows 'XXX'22:10
viricdepends on HAVE_KERNEL_GZIP...22:10
viricwhat a party :)22:11
wpwrakyeah, only on input. don't want to "pop" on output :)22:11
larschm, which kernel do you use?22:11
DocScrutinizerHAH, on outputs you *got* 220pF22:11
DocScrutinizerno idea what for22:11
viriclarsc: jz-3.222:12
wpwrakregarding the differential CD ground, if the source also decides to decouple GND, we would have a problem, right ? is this something that's like to happen ?22:12
DocScrutinizerno22:12
wpwrak(output 220 pF) probably since it worked during the last zombie attack, or whatever ;-)22:13
DocScrutinizeryour outputs make nice inputs ;-P22:13
viriclarsc: I can't find the magic combination :)22:13
wpwrakthat's where cargo cult meets voodoo meets ancient rituals meets paranoia ;-)22:13
larscviric: there is a patch _somewhere_22:14
viriclarsc: oh? a patch for what?22:14
wpwrakexcept for the 10 k pull-downs, right ? :)22:14
larscfor enabling the option22:14
viric:D22:14
larscmth: pin22:14
larscg22:14
Aylawhat are you trying to achieve?22:16
DocScrutinizerwpwrak: L1 looks ODD22:16
wpwrakwith non-differential use of CD*, we wouldn't know a difference, right ?22:16
wpwrakL1 is "0" :)22:16
viricAyla: boot a kernel with xbboot22:16
larscviric: http://projects.qi-hardware.com/index.php/p/qi-kernel/source/commit/9a83f481a49cf9337a0aaed749c631572dddb950/ you need the first three hunks22:17
whitequarklol zombie attack22:17
DocScrutinizeror, depending on where from you look at it, C30 is incorrectly placed22:17
viriclarsc: why isn't that in jz-3.2?22:17
viriclarsc: I can try any other branch that works22:17
wpwrakwe have a number of dead indictors. i think this is just one more of them.22:17
Aylawhat's the difference with jzboot?22:18
viricwhat is jzboot?22:18
viric:)22:18
larscviric: nah, you need to manually port them22:18
DocScrutinizergona TV&sleep22:18
DocScrutinizero/22:18
larscviric: the better xbboot22:18
wpwrakit's the digital VDD anyway. that's supposed to be able the handle a bit of noise22:18
larscviric: written by whitequark 22:18
wpwrakDocScrutinizer: thanks a lot for helping to puzzle this out !22:19
DocScrutinizerwpwrak: also supposed to have quite some surges22:19
viriclarsc: is there any reason for that patch not to be in jz-3.2?22:19
DocScrutinizerthat'S why you don't wanna have inductors without buffer C on digital VDD22:19
DocScrutinizeryw22:19
DocScrutinizern822:20
larscviric: nobody ported it22:20
larscviric: it is from the jz4760 branch22:20
viriclarsc: Message:stash ? :)22:20
larscyes, and it was not even supposed to be pushed onto the servers22:21
viricaah.22:21
larscthat particular commit, just a bunch of random changes22:21
viricso I should branch from jz-3.2, and have that in my branch22:21
wpwrakah, the 6k8 series on input. should they be smaller ? or is 6k8 acceptable ?22:22
viriclarsc: (maybe I ask basic questions... I've no idea about how do you work :)22:22
DocScrutinizeris OK22:23
larscsounds fine22:23
DocScrutinizerI'd have picked 10k though22:23
viricerror: could not apply 9a83f48... stash22:23
viric:)22:23
larscyes22:24
larscyou need to manually apply the first three hunks with some fuzz22:24
larscselect SYS_SUPPORTS_ZBOOT_UART16550 needs to go in arch/mips/Kconfig under MACH_JZ474022:25
larscthe second hunk is fine as it is22:25
virichm 22:25
viricok. I'll do that slowly22:25
larscand the third one should use CONFIG_MACH_JZ4740 instead of CONFIG_MACH_JZ476022:25
larscand 0xB0030000 instead of 0xB003100022:26
Aylashouldn't those be commited?22:31
viricyes. I'll complain to the owner.22:34
viric:)22:34
viricok, got vmlinuz22:34
viriclarsc: so... vmlinuz built. I run22:35
viric# xbboot -u 0x80600000 vmlinuz22:35
viricand it does not boot :)22:35
larscand you don't have a serial right?22:36
virichm could have.22:37
viricwould it help?22:37
viricI've the console to tty0...22:37
viricthe lcd does not switch on22:37
larscthe decompressor should print some mesages to the serial console22:38
viricoh22:39
viricwhat serial console?22:39
larscthe hw serial console22:39
viricaren't there two? :)22:40
viricttyS0 and ttyS122:40
larscyes, but only one is accessible22:41
viricaccessible by who or what?22:41
viricI've that of the battery.22:41
larsci'm talking about this one http://en.qi-hardware.com/wiki/Serial_console22:42
viricah ok. yes yes22:42
viric57600...22:43
viriclarsc: nothing22:44
wolfspraulwpwrak: I'm looking at the 'solder mask clearance setting ignore' but again22:45
wolfspraulbug22:45
wolfspraulit's messy even in the GUI22:45
wolfspraulthe solder mask and paste clearance seem to be saved in the .pro, not the .brd file22:46
larscviric:  i think you need to build with DEBUG_ZBOOT22:46
wolfspraulthe menu has two different Save options (in HEAD)22:46
larsc=y22:46
viricI disabled that when asked...22:46
wolfspraulPreferences -> Dimensions -> Save22:46
wolfsprauland Preferences -> Save Preferences22:46
wolfspraulboth seem to do the same thing (not sure though)22:46
viricok, let me try again22:47
wolfspraulin my quick tests, the solder mask clearance is indeed saved in the .pro, but the solder paste clearance is lost, even when saving manually22:47
larscviric: and vmlinuz.bin22:47
larscnot vmlinuz sorry22:47
viricah oh22:47
larscvmlinuz is the elf binary, i think22:48
wolfspraulthe values are only reloaded next time if you open the .brd file via the kicad binary and .pro. if you open via pcbnew directly pointing to the .brd file, even settings saved in the .pro are not picked up, unless you manually 'load' the preferences22:48
wolfspraulso there are bugs and inconsistencies already in the GUI22:48
Aylalarsc: it's the gzipped vmlinux.bin IIRC22:48
wolfspraulwould it help if we made this a command line option?22:49
viricI don't remember seeing vmlinuz.bin22:49
viricvmlinux is elf sure...22:49
wolfspraulso we would just pass the solder mask and paste clearance in (whichever you need)22:49
Aylaanyway once it works, if you need a bootloader, you could use ubiboot :D22:49
wolfspraulthat won't help if you want them to be loaded from the .pro though22:49
larscviric: make vmlinuz.bin22:49
viricah, can do22:50
viricboots!22:50
larscparty!22:50
viric:D22:50
viriccan't open root... but well... that next :)22:50
viricdanke schun22:51
wpwraklost solder paste clearance sound like un-fun22:52
wolfspraulI am not 100% sure, there are too many inconsistencies22:52
wolfspraulone would need to unwind this one by one22:52
wolfspraulmy questions:22:52
wolfspraulwhich one of 1) solder mask clearance 2) solder paste clearance 3) solder mask ratio clearance do you actually need?22:52
wolfsprauldo you need those values to come from the .pro file (the only place they are stored, and it seems not even all of them), or is it enough to pass them in over the command line?22:53
wpwraklet's see ...22:56
wpwrakatben.brd: Pad2MaskClearance, Pad2PasteClearance, and apparently ZOptions ties into this as well. then atben.pro VEgarde22:57
viricI can't get the serial port working though...22:57
wpwrak(ben-wpan commit daaac58f885cdc017bad91fa98c36dba37c3c26a)22:57
virichm I connected for a while TX/GND inversed22:58
viriccan that break the serial port?22:58
wpwrakthey're part of the design, so they should be in the file22:58
wolfspraulhmm, but where do Pad2MaskClearance and Pad2PasteClearance in the .brd file come from in the GUI?22:58
larscviric: not really22:59
wolfspraulif I open pcbnew, and change the values in Preferences -> Dimensions -> Pads Mask Clearance, those values in the .brd file are not affected22:59
wpwrakmaybe ask on the list ? :)22:59
wpwraki think there are several places where you can and have to save preferences22:59
larscviric: the serial port is a mystery for tomorrow night ;)22:59
viric:) yes22:59
larscat least for me22:59
viricme too.22:59
viricthank you!23:00
wolfspraulmidnight, lars is becoming civilized23:00
wolfspraulthe system gets you!23:00
larscwolfspraul: well, i have to be at the office at 8am23:00
wolfsprauloh I can imagine23:00
wpwraki don't remember how the clearances (radio vs. absolute) get mixed in the end. i remember having to track it down in the sources before i could figure out how to do it in the gui, though.23:00
wolfspraulI miraculously always had the freedom to do whatever I wanted23:00
wolfspraulshow up at lunch time :-)23:01
wolfspraulbut infantilization rules23:01
wolfspraulyou get a free soda (yuhuuuu), but you have to be there at 9am sharp...23:01
wolfspraulnothing much you can do, hope you survive :-)23:02
larscso far nobody has complained when i showed up 9:30, but i'm trying to not making it the norm23:02
Aylawhat's your job?23:02
larsclinux :)23:03
wpwraki remember getting funny looks when leaving at 8 am :)23:03
Aylalarsc: you're paid to work on Linux?23:04
wolfspraulwpwrak: ok here's what I do: I will leave a comment in TODO, and try to get this committed first23:04
wolfspraulthere are still many details, your patches need to be up-leveled as well, etc.23:04
wolfspraulI just wait until the bug reports come in, then look at this again23:04
larscAyla: yes! unbelievable isn't it?23:04
Aylayes23:04
Aylayou take applicants? :D23:05
larscI just started myself a few month ago23:05
wpwrakwolfspraul: sounds good. it's better to get feedback from upstream early on anyway. no point in working three years on the perfect patch set, just to have it summarily rejected23:05
wolfsprauloh of course23:06
wolfspraulI am just focused to get our stuff back in shape23:06
wolfspraulthere's quite a number of little things in there23:06
Aylathat's the sort of job I want to have. When it'll be time for me to seek a 'real' job, I'll ask you some advices23:06
mthlarsc: pong (reading backlog now)23:07
larscmth: problem solved23:07
larscAyla: make sure to get your patches upstream.23:09
Aylayep, I believe it helps23:09
wpwrakwolfspraul: (little things) yeah, they have that tendency of piling up :)23:10
AylaI also plan to do the GSoC23:10
larscthere are a lot of people who apply for a linux job, but have actually no much of an idea how the upstream development process works23:11
mthis there anything to be merged into jz-.3.2 to simplify things for other people?23:13
mthI don't know if what viric is doing is standard or non-standard23:14
Aylathe defconfig should contain CONFIG_SYS_SUPPORTS_ZBOOT_UART1655023:16
mthwe actually have a patch forcing VMLINUZ_LOAD_ADDRESS:=0x80600000 already for OpenDingux23:16
AylaI've had that issue when dealing with ubiboot23:16
mthI regenerate the defconfig every kernel release, so if the flag isn't there, it is because none of the Kconfigs set it23:16
mthSYS_SUPPORT flags are not set by the user afaik23:17
AylaAFAIK it requires a change on arch/mips/boot/compressed/uart-16550.c too23:17
Aylaotherwise the PORT() macro is not defined23:18
larscviric: can you send the patch to the mailinglist?23:18
wpwraklarsc: (inexperienced applicants) so you're helping out in HR now ? ;-)23:19
larscwpwrak: just something i've heard23:21
larscbut i guess it always depends for kind of position you are hiring someone23:28
qi-bot[commit] Wolfgang Spraul: uplevel cmdline patches (entire set won't build right now) (master) http://qi-hw.com/p/eda-tools/a27777923:28
wolfspraulwpwrak: alright, first step. do you actually want to move to kicad head?23:29
wolfspraulif so I need to go through your patches as well23:29
wpwraklarsc: yeah, there's probably a lot of demand for upstream-agnostic developers ...23:30
wpwrakmoving to kicad head would be great23:30
wolfspraulI took the liberty to break backward compatibility in the cmdline options23:30
wolfspraulin favor of staying close to the GUI23:30
wolfspraulhere's what it looks like now http://pastebin.com/4xucpRLa23:33
wpwraknice ! a test script23:33
wolfspraula very basic one, just for myself23:33
wolfspraulI type fast, but don't like to repeat23:33
wpwrakhehe :)23:33
wolfspraulthe test script just calls the major options one by one, and lists the output files23:34
wolfspraulno fancy checks with the generated files23:34
wolfspraulcan you look over the pastebin - let me know if you spot anything wrong23:35
wolfspraulso I will work on the rest of the set now23:35
wpwrakwhy cryptic abbreviations like --plot-gb-ex-pcb-edge ?23:36
wolfspraulon one hand I try to stay close to the text in the GUI23:36
wolfspraulon the other hand I don't want the option to be too long to --help still looks good23:36
wolfspraulthere's some space with that one, we can make it longer23:37
wolfspraulgb-exclude-edges ?23:37
wolfspraul--plot-gb-exclude-edges23:38
wpwraks/gb/gerber/ :)23:38
wpwrakgrr23:38
wolfspraul:-)23:38
wolfspraulgets too long, --help won't look good23:38
wpwrakyou can put the option and the beginning of the description on separate lines23:39
mth--plot-gerber-noedges?23:39
wolfspraulI try to follow the text and structure of the GUI23:39
wpwrak--very-very-long-option=with-parameters,even-a-list-of-them23:39
wpwrak\tthis does absolutely nothing23:39
wolfspraulthe GUI is our baseline I believe23:39
wpwrak\t(default: enabled)23:40
wpwrakor such23:40
wpwrakusing the same terminology as the GUI is good23:40
wolfspraulI don't have that much room to influence how --help looks like because this is the wx framework system23:40
wpwraktoo easy to make a confusing mess otherwise23:40
wpwrakoh, i see23:40
wolfspraullet me see what looks good :-)23:40
wpwrakdrop the "plot-" prefix ?23:41
wolfspraulI actually can look at --help now, before was just a mountain of characters23:41
wpwrak;-)23:41
wolfspraulyes we could drop plot-23:41
wolfspraulwe risk an overlap with another dialog then, but so what23:41
wolfspraulwe can wait until we have such a conflict23:41
wolfspraulnamespaces can never be perfect...23:41
wolfspraul--plot-gerber-exclude-edge and --plot-gerber-aux-origin still looks good23:43
wolfspraulshould we remove all plot- ?23:43
wolfspraulactuall the aux-origin exists in the drill and plot dialogs23:44
wolfspraulso we have --drill-origin-aux and --plot-gerber-aux-origin right now23:44
wpwrakcan they be merged ?23:44
wolfspraulthey could, but that's what I want to avod23:44
wolfspraulavoid23:44
wolfspraultwo different dialogs23:44
wpwrak:)23:44
wolfspraullet's map the dialogs23:44
wolfspraulI am certain the dialogs will change more23:45
wolfsprauloptions are moving around, dialogs get merged, etc.23:45
wolfspraulso there are only 2 paths23:45
wolfspraulone is that the cmdline options completely have their own design23:45
wolfspraulgiven that we only offer 5% of what's available now I don't think that's a good idea at all23:45
wpwrak--plot-exclude-pcb-edge          exclude PCB edge (Gerber only)23:45
wolfspraulor the second one is to follow the GUI, even if that means some redundancy and changes/breaks in the cmdline options23:46
wpwrak--plot-aux-origin           Use aux axis as origin (default: abs; Gerber only)23:46
wolfspraulanything in between will be a total mess23:46
wolfspraulthanks to the wx command line system, I cannot repeat options in the --help output23:47
wpwrak-drill-aux-origin             Use aux axis as origin (def: abs)23:47
wolfspraulI think I cannot23:47
wolfspraulyes aux-origin, better23:47
wpwrak(use aux-origin or origin-aux but better don't mix)23:47
wolfspraulaux-origin23:47
wpwrak(wx) the joy of overbearing frameworks :)23:48
wolfspraulpcbnew --help http://pastebin.com/aiSAb6ZC23:49
wolfspraulI think this is ok for now23:50
wpwrakah, we can probably drop the "ERC pin exceptions" patches. that doesn't quite work anyway and it's incompatible with the kicad belief system23:50
wpwrakyes, looks good23:51
wolfspraulthen there is only streamline-erc.patch left23:51
wolfspraultwo are about aux-origin for DXF (will do that)23:52
wolfsprauland two are about the erc exceptions you want to drop23:52
wolfspraulerc-exceptions.patch and fix-pinedit-collision.patch23:52
wolfspraulcorrect?23:52
wpwrakno, i want to keep fix-pinedit-collision.patch (if it's still needed)23:52
wolfspraulok23:53
wolfspraulso both fix-pinedit-collisions and streamline-erc23:53
wpwrakstreamline-erc.patch is merely cleanup. not needed by us if there's no erc-exceptions.patch23:53
wolfspraulok. then only fix-pinedit-collisions (plus DXF aux origin)23:54
wpwrakyup23:56
wpwrakthese two are hopefully not too messy23:56
wolfspraulshould not. OK I hope to get it all to build tomorrow then23:57
wolfsprauldo you want installable packages or do you rather build from source anyway?23:57
wolfspraulthere will definitely be bugs in the cmdline patches, it was quite a lot of lifting around and *lots* of changes in the KiCad sources23:58
wpwraki build from source23:58
wolfspraulok23:58
wolfspraulthen I will ask xiangfu to upstream the whole thing23:58
wolfsprauladd it to his nearly infinite (tm) todo list23:59
wolfspraulI'm adding about 1-2 items / day to that list :-)23:59
wolfspraulluckily xiangfu ignores most of it23:59
--- Wed Jan 11 201200:00

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