#qi-hardware IRC log for Saturday, 2013-06-08

uncloudeddoes the latest u-boot for NanoNote support booting from an MMC formatted as ext4?00:22
uncloudedanswering my own question: Yes.  no idea how old my u-boot was previously.  maybe even factory since I've been so scared to update it00:42
larscmth: try to at least confine the ifdef to only one place, e.g. #ifdef JZ4770 void phy_set_power(int power) { ... } #else void phy_set_power(int power) {} #endif08:35
mthlarsc: I agree09:43
mthI've got a cleaned up version now, but I guess there is no point in upstreaming this until the rest of the 4770 support is ready, right?09:44
larscatleast the core support which defines the function should be there09:55
mthlarsc: #define JZ_GPIO_PWM0 JZ_GPIO_PORTD(23)12:13
mthon the JZ4770, the pin assignment is different12:13
mthso I guess the macro has to be renamed?12:13
mth(from arch/mips/include/asm/mach-jz4740/gpio.h)12:14
larscthe macro should probably not be used outside of board code15:07
mthin that case, we'd have to insert some kind of indirection between the macro and the pwm driver15:52
larscI'm going to add pinctrl support to jz474015:55
larscthis is espcially important for anything >jz4740 since you can have multiple options to which pin you can actually route a signal15:56
larscpinctrl is pretty much the same as our jz4740_gpio_* api, except that it is more generi15:58
mththe big green block in the middle is how I currently solved the "multiple options to route a singal" for the msc on the 477016:47
mthspeaking of msc, there is slot-gpio which is a generic implementation of read-only and change detect based on gpio pins16:49
mththat could be used in the 4740 driver as well16:49
mthlarsc: you said "board code" for the pwm-to-gpio mapping, but that mapping is inside the SoC, not on the board17:02
mthyou could do it in pdata, but then every board with the same SoC would just be using the same table17:03
viricare any of you playing with 16MB systems?17:09
mthno, the smallest I have around here is 32MB17:10
viricI think openwrt deprecated all 16MB systems :(17:11
viricat backfire time (2.6.32)17:12
viricI wonder if anyone is making a distribution that will run on 16MB17:12
larscmth: since you can route it to multiple pins it is board specific17:22
larscmth: yea, I already have slot-gpio patches for the jz4740 mmc driver17:23
viricis there anything that can tell where the kernel size comes from?17:24
viricwhat part how much contributes17:24
mthlarsc: pwm routing is static, isn't it?17:25
virichm just found http://www.elinux.org/Kernel_Size_Tuning_Guide17:26
mthat least on the 4770 there is exactly one pwm pin per timer17:26
viricI guess there is no OS maintained with systems < 32MB of RAM17:36
viricfree os17:36
whitequarkFreeRTOS? :)17:38
viricon mips? :)17:39
whitequarkthe latter, why not?17:44
whitequarkI see your point, though.17:44
whitequarkI think that vxworks and similar stuff still works well on 16MB systems. maybe even smaller.17:45
whitequarkQNX? not sure17:45
wpwrakcontiki ? :)17:48
wpwrakif you have that much memory, you could probably run regular linux. else, there is uclinux17:48
viricwell, I'd like to have device drivers too ;)17:49
whitequarkugh, contiki17:49
viricwpwrak: a 3.x linux openwrt leaves little space17:49
whitequarkremind me again why exactly an OS for resource-constrained wireless sensors must have a VNC server and GUI?..17:49
viricand I have mmu. I don't want to give up on mmu17:49
viricwhitequark: it evolved17:50
whitequarkviric: I call that devolution17:50
viricwhy exactly an OS that could run fine on 8MB of RAM evolved to requiring 32MB?17:50
wpwrakwhitequark: vnc ? not sure if we're talking about the same thing :)17:51
whitequarkI would guess that it got optimized for less resource-constrained workloads17:51
whitequarkviric: ^17:51
viricmaybe also contiki17:52
whitequarkwpwrak: https://github.com/contiki-os/contiki/blob/master/apps/vnc/vnc.c17:52
whitequarkI wish I was kidding.17:52
whitequark90% of contiki is useless bloat17:52
wpwrakoh well, application17:52
wpwrakyou could add libreoffice to owrt if you wanted17:52
whitequarkand the other 10% are highly specific for AVRs17:53
whitequarkwhat is this, 2000?17:53
viric"for the fun" looks to me like a good reason, in any case17:53
whitequarkI won't even talk about buffer overflows, etc.17:54
wpwraki've heard of problem with loss of interrupts on contiki. not sure on what platform(s), though17:58
whitequarkI personally would blame lack of design :|17:59
whitequarkin unrelated news. I disassembled my notebook (proof that you can make something as thin as macbook air AND make it repairable) and discovered a peculiar feature18:00
whitequarkthe keyboard is clamped hard to the upper part of the case18:00
whitequarkhad to buy a replacement from asus for $11018:01
viricdid you break the keyboard?18:01
whitequarkthe Right button stopped working18:01
whitequarkdismantling it and blowing off the dust didn't help18:01
whitequarkweirdest part... I went to their "official service center"18:02
whitequarkthey said that replacing keyboard would take 7-10 DAYS18:02
whitequarkwhat the fuck18:02
whitequarkwill these guys literally buy a replacement from asus spare parts store and reassemble the whole critter?18:02
virican accurate task18:02
whitequarkI can do that in a hour, myself18:03
whitequarkwell, two hours18:03
whitequarkpaying some dude to be a screwdriver operator? wt18:03
viricyou are just the wrong kind of costumer.18:03
Action: whitequark shrugs18:04
whitequarkI would happily pay twice the cost of the replacement, IF they could do it "right now"18:04
whitequarkbut with this kind of... service, it's really more efficient to do that myself18:04
Action: whitequark just realized that he would likely never again pay for repairing electronics18:06
whitequark(except maybe for SMT jobs. maybe)18:07
wpwrakit's more fun that way :)18:08
whitequarkwpwrak: (that) doing it myself, that is?18:08
whitequarkit isn't even about fun. I simply cannot afford to lose my notebook for a week. not now, at least18:09
wpwrakof course, as time goes by, you'll get tired of things that you have to do over and over again. so you may get bored enough with them to outsource them.18:09
wpwrakyeah, there's also efficiency18:09
whitequarkand when I'm faced with a choice of setting up another machine for a week <> reassembling it myself... it's obvious18:10
wpwrakwhen the choice is between open-fix-close (2 h) and send/deposit, wait, receive/retrieve, test, complain, repeat ... then the selection isn't too hard18:11
whitequarkI was impressed by some guy at a flea market who fixed audio jack on my Galaxy S II. the jack is likely to be custom-made for this phone.18:12
whitequarkso, in just about three hours he found a replacement and resoldered it18:12
whitequarkI wonder if he robbed someone with the same phone and disassembled it.18:12
whitequarkwell, whatever... he can do anything if he fixes it quickly enough :)18:13
wpwrakremember, if you buy from robbers, you confirm their business model and you may be the next source of spare parts18:15
whitequarkoh yes, a well-known ethical problem18:17
whitequarkone could also say that someone else would buy it, anyway18:18
wpwrakeven then, you're increasing the demand. the invisible hand of the market will take care of adapting the supply accordingly.18:21
larscmth: hm, yes, looks like there is only one possible signal per pwm signal, so it is SoC specific18:21
larscwhitequark: here you actually be charged with buying stolen goods18:23
whitequarklarsc: hmm18:24
whitequarkthat was a joke18:24
whitequarkthough, when I think about it... well.18:25
whitequarkno idea18:25
whitequarklarsc: how would I clear myself from charges? do something to have a reasonable knowledge of the spare parts to ensure they're not stolen? or what?18:25
larscif you can't distinguish it from a legitime sale you are fine18:28
larscbut e.g. if somebody tries to sell you something at lets say 10% of it's actual value you are supposed to get suspicious18:29
whitequarkwell. it depends on what's "legitimate". the guy doesn't run an official service center (obviously). but I can totally see how someone with a soldering iron and a bag of dead phones could do his job18:29
whitequarkah I see18:29
whitequarkis $20 a lot for replacing an audio jack? :)18:29
larscI don't know18:30
larscit's then up to the judges to decide whether you should have been suspicious or not18:31
whitequarkI suppose this makes sense18:32
--- Sun Jun 9 201300:00

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