#qi-hardware IRC log for Friday, 2013-11-01

whitequarkhttps://developer.android.com/about/versions/kitkat.html04:47
whitequarkyou can hate on android all you want, but they really got their shit together in the last few releases and made a great OS04:47
wpwrakcolorful :)04:54
wpwraki'd go right to cyanogenmod, though. the only real chance for long-term updates.04:54
whitequarksure, for me android pretty much equates with CM or replicant :)04:56
whitequarknexuses got a decent upgrade schedule, but CM still beats them on everything04:56
whitequark... except proprietary lock-in04:56
wpwrakdamn. not having that wuold really suck ;-)05:06
whitequarkyeah. I mean, NSA should have all my data too, right?!05:08
whitequarkheard of latest Snowden reveal? NSA tapped into Google's privately *owned* fiber between their privately *owned* DCs without Google's knowledge05:08
whitequarkbet they're pissed off now05:08
wpwrakthey shouldn't be pissed off. they should build data centers elsewhere. actually, they've been doing this for a while already ...05:13
whitequarkI doubt they could stop building datacenters in US, though05:14
whitequarkhm.05:15
whitequarkor maybe they can? what's worse, the stick of losing your customer's data or the carrot of having response time under 100ms?05:15
rohthey're talking about crypting more stuff now05:19
wpwrakthey can build the everywhere, then choose a date replication algorithm that keeps non-US things outside the US05:20
rohright direction...but only partially helpful tho ;)05:20
rohin the end... its a us company05:20
rohbut hey.. multinational companies often move if pissed off enough05:21
whitequarkroh: yeah, it's been for a while05:23
whitequarkwpwrak: I think that's pretty much what amazon does05:23
whitequarkguess that's we'll see soon05:23
qi-bot[commit] Werner Almesberger: ircstat/ML: update for 10/2013 (master) http://qi-hw.com/p/wernermisc/a804f3512:10
wpwrakDocScrutinizer05: kewl, you already passed 5k and do the design work15:44
wpwrakDocScrutinizer05: btw, the funding page is nicely done. a bit of pleasant visuals goes a long way :) plus, you have good information there, too15:45
DocScrutinizer05thanks15:48
DocScrutinizer05sebastian's work15:48
DocScrutinizer05he even couldn't resist to integrate an easter egg15:49
DocScrutinizer05wich i'll use if some user asks for pink ponies again15:50
DocScrutinizer05;-)15:50
wpwrak;-))15:52
cdeDocScrutinizer05: easter egg?15:57
DocScrutinizer05ponies16:02
wpwrakpink easter ponies16:04
DocScrutinizer05the angry blue one is scary16:09
DocScrutinizer05the magic one16:10
DocScrutinizer05well, I think they all are magic16:10
cdeI don't follow. do you have a link?16:11
DocScrutinizer05number 816:11
DocScrutinizer05#ponies16:11
DocScrutinizer05ooh, it's called the shadowbolt16:11
DocScrutinizer05err nope?16:12
cdewat16:12
DocScrutinizer05#13 is a dragon :-O16:12
wpwrakcde: wikia has an article on them: http://falloutequestria.wikia.com/wiki/Shadowbolts16:15
cdeoh wow16:16
DocScrutinizer05GOD a pinao pony16:19
DocScrutinizer05piano*16:19
DocScrutinizer05AAH, the *.gif context gives it away, I meant the nightguard16:20
DocScrutinizer05https://raw.github.com/dos1/Browser-Ponies/gh-pages/ponies/royal night guard/nightguard_left.gif16:20
DocScrutinizer05https://raw.github.com/dos1/Browser-Ponies/gh-pages/ponies/iron will/ironwill_walk_left.gif LOL16:22
DocScrutinizer05GOD! they are army  -  https://github.com/dos1/Browser-Ponies/tree/gh-pages/ponies/16:25
DocScrutinizer05a wabbit a wabbit16:30
DocScrutinizer05ok, this starts to eat significant parts of my CPU and RAM now, so time to stop it.  http://wstaw.org/m/2013/11/01/plasma-desktopyx3743.png16:33
cdethis is insane16:33
DocScrutinizer05this is what you get from Friday and holiday, after a very exhausting Thursday16:33
DocScrutinizer05with funny stuff like 1, ... 2, 3 donations a 4kEUR, then zilch since all three were hoax16:35
DocScrutinizer05FSFE zealots bitching at real life properties of WLAN chips16:35
DocScrutinizer05Brasilia no valid destination for online shopping/shipping of any goods16:36
DocScrutinizer05and 3 dozen other funny things16:36
DocScrutinizer05incl slashdot16:36
cdewell thanks to FSFE zealots we now have fully blob-free WLAN drivers16:37
DocScrutinizer05tell me more about it16:37
cdehttps://github.com/qca/open-ath9k-htc-firmware16:38
DocScrutinizer05particularly tell me how you get a blob-free hardmac running, for PSM without hogging AÜE CPU16:38
cdePCI/PCIe don't actually require a blob, the latency of the bus is low enough16:38
DocScrutinizer05APE*16:38
DocScrutinizer05well, embedded usually has no PCI/PCIe16:39
cdecorrect16:39
DocScrutinizer05and latency of a CPU in zero-clock mode is low, but not THAT low16:40
cdealso correct16:40
DocScrutinizer05tbh I prefer a blob to a firmware in some fused flashrom inside the WLAN chip16:41
DocScrutinizer05at least I can update the blob any time16:41
DocScrutinizer05and I can RE it, if I can16:41
cdewell the flash can also be updated; unless you meant an actual rom16:42
DocScrutinizer05THAT however usually requires WINDOWS blobs apps16:42
DocScrutinizer05since chip manuf doesn't bother to ship a FOSS linux tool to do such update16:43
DocScrutinizer05so the FW blob you load in chip init time is the clearly better alternative16:43
DocScrutinizer05s/ in / on 16:44
cdeok16:45
DocScrutinizer05and the heck I have no problems whatsoever with a blob of DATA on my linux system, fsck GNU and RMS16:45
DocScrutinizer05the assumption that such *data* was a *program* relevant for GNU is erratic anyway16:46
cdeso you are willing to surrender your freedom for a little bit of convenience?16:48
DocScrutinizer05was it this very cahnnel or another one where I stated just today: "why is a closed source program allowed when it runs in user land on a GNU linux system, but _not_ allowed when I offload it to a coprocessor located on a peropheral chip?"16:48
DocScrutinizer05au contraire, cde. I *insist* in my freedom to use and tamper with such blobs16:48
cdethat's good16:49
cdebut you shouldn't give your moneys to companies that sell you hw with blobs16:49
DocScrutinizer05well, I can as well keep my money and eat it16:50
DocScrutinizer05but it's not as tasty16:50
DocScrutinizer05IOW: there is no choice16:50
cdethere is a choice. for example you can choose not to buy Apple hardware16:50
DocScrutinizer05well, I can as well keep my money and eat it16:51
DocScrutinizer05oh, I already said that16:51
cdealso, you can choose not to buy on Amazon to protest their DRM policies16:51
DocScrutinizer05I can also choose to not breathe anymore16:52
cdeyes. to protest the fact that the universe is not free software16:52
DocScrutinizer05don't tell me about my choices to NOT do something16:52
DocScrutinizer05since that's a) obvious and trivial, and b) leading nowhere16:53
cdecorrect16:53
DocScrutinizer05tell me which choice I have for a WLAN chip that can do hardmac and still offers packet injection, firmware update, and monitor mode16:53
DocScrutinizer05I'd actually be happy when you come up with *any* other choice than "not do it"16:54
DocScrutinizer05but honestly, I'm not too interested in perpetuating the debate with that other guy as of yesterday16:56
DocScrutinizer05it ended in him revoking his donation since Neo900 wasn't able to offer the WLAN pink ponies he hoped for16:57
DocScrutinizer05and I wasted ~90min on that16:59
DocScrutinizer05so on the end he asked "tell me is the wlan chip sharing RAM with the CPU"16:59
DocScrutinizer05then he revoked his donation after we told him that it#s connected via SPI17:00
DocScrutinizer05futile effort to discuss with those zealots17:00
cdeany ath9k-based chip really. such as http://dx.com/p/802-11n-150mbps-wifi-wlan-wireless-network-usb-adapter-5353817:00
DocScrutinizer05fine, I'll happily check it17:01
DocScrutinizer05thanks17:01
cdenp. there is also https://www.thinkpenguin.com/gnu-linux/penguin-wireless-n-usb-adapter-gnu-linux-tpe-n150usb which is FSF certified17:01
DocScrutinizer05it's not like we are averse against freedom and FOSS, you know ;-)17:01
cdeyes. we also both like ponies, but in moderate amounts17:02
DocScrutinizer05just we need a hardmac chip (aka 'cpu on chip, so main CPU can sleep and chip handles PSM')17:02
DocScrutinizer05which rules out all softmak (driver runs on APE CPU) solutions17:03
DocScrutinizer05softmac even17:03
cdeafaik all USB-based solutions are softmac due to the high latency of the bus. unfortunately you can't wake up the CPU with USB 17:04
DocScrutinizer05ideally we find a chip that can do both hardmac and softmac - iirc the prism54 been such a critter17:04
DocScrutinizer05err, that would rather speak for hardmac17:05
cdeyes, that's what I meant. the MAC runs on the stick17:05
DocScrutinizer05right17:05
cdea way to do what you need is to identify a GPIO out of the AR9271 chip that could be used to wake the CPU which would then proceed with the transfer over USB17:06
DocScrutinizer05though via USB you can control the RF lowlevel, the bandwidth is sufficient and the CPU usually strong enough (heck it even can do SDR)17:06
DocScrutinizer05USB sticks usually are used on laptops and the like, where CPU never sleeps17:07
DocScrutinizer05and I think for softmac drivers it's not relevant which interface is used to attach the WLAN radio17:08
DocScrutinizer05the prism54 softmac worked fro USB dongles17:09
DocScrutinizer05for*17:09
DocScrutinizer05it had a FOSS softmac, a blob hardmac firmware, and they worked on a FOSS hardmac firmware17:10
DocScrutinizer05the latter never got finished afaik17:10
DocScrutinizer05too many mysteries inside the chip17:10
cdeyes that is sad17:15
cdelekernel did in fact some amazing work on it17:16
lekerneland then found out about FPGAs17:17
lekernel:)17:17
lekernelwe had a bunch of things working nevertheless. UART, GPIO, USB, PCI, some of the crypto acceleration, and most of the transceiver17:18
DocScrutinizer05anyway: http://privatepaste.com/4df9cc603017:20
DocScrutinizer05is what we're confronted with17:20
DocScrutinizer05mandatory features for WLAN: PSM/low-power_always-on, 54Mb, independant operation (no constant support from main CPU), monitor mode. Almost mandatory: packet injection. Nice to have: no blobs17:24
DocScrutinizer05if any of the hardcore freedom lovers wants us to change those priorities, then they need to come up with a suggestion how to handle that in a way so the other 99.9% of users are also satisfied17:26
DocScrutinizer05ooh I forgot: small package17:27
DocScrutinizer05and a useful hw interface17:27
DocScrutinizer05and full documentation aka available drivers in linux17:28
DocScrutinizer05FOSS drivers17:28
DocScrutinizer05we're appreciating all help we can get on identifying the right chip for us17:29
DocScrutinizer05if it then is also sourcable in small quatities, all the better (euphemism!)17:29
DocScrutinizer05we'll happily use any chip you manage to come up with, as long as it satisfies the above requirements17:31
whitequarkhey folks, do you possibly know what *module* one could use for 6lowpan?17:46
whitequarknot on-chip RF.17:46
mthDocScrutinizer05: in the GCW Zero is a Realtek 8188CUS; it's far from perfect but it is usable17:47
mththere is a mainline kernel driver, but that doesn't actually work when we test it (handshaking with AP gets stuck)17:47
DocScrutinizer05o.O17:48
mththere is a driver from Realtek itself that works, but is huge and ugly code17:48
DocScrutinizer05ooh17:48
mthit has a blob in the wifi chip itself, but everything that runs on the CPU is open source17:48
mthapparently the 8188CUS is no longer available, but the 8188EUS is and it's pretty similar17:49
mthI noticed in his github that Larry Finger is cleaning up the driver code for the EUS17:49
DocScrutinizer05yeah, I'm sure we had that one on our radar17:49
mthbut I don't have the EUS hardware yet, so I can't test it17:49
DocScrutinizer05oops, nope, W2CBW003  868617:50
DocScrutinizer05W2CBW0015 878717:50
DocScrutinizer05TiWi 5/6/7/817:51
DocScrutinizer05jorJin17:52
DocScrutinizer05I'll have a look into the realtek 818817:52
Action: DocScrutinizer05 wonders if we should set up a wiki page based on the feasibility study17:57
DocScrutinizer05you're aware of the feasibility study, are you?17:58
DocScrutinizer05http://projects.goldelico.com/p/neo900/downloads/53/18:01
rohworking hardmac wifi? bwahahahahaha18:20
rohsorry.. but you already lost.18:21
rohdoesnt exist atheros usb chipsets are as near as it gets18:21
rohmost important: n support. in front of anything else on wifi.18:23
rohloads of people disable b/g support nowadays and i've seen routers delivered that way in the stock firmware18:24
mthI hadn't seen that study yet18:24
rohme neither.. nice work18:25
rohmy best guess would be ar9170 or something near that series18:27
rohbut i havent checked module avail or how big these are18:28
rohanyhow.. whichever vendor makes the race.. there needs to be some way to flash/ramload that hardmac to be able to update its firmware (closed or not) and a proper process to get these. i got quite some devices around which basically arent useable anymore due to 'too old wifi'18:32
wpwrakblargh. such crap weather. 98% humidity all day, rains almost non-stop.18:35
wpwrakDocScrutinizer05: what's PSM ?18:37
cdeindeed, what's PSM?18:37
roh*sigh* .. it seems the only hardmac i find which does 'proper' ap mode now is ar6kl18:40
rohno monitor mode tho18:41
wpwraki've received some somewhat enigmatic hints that it may be possible to adapt the brcm80211 driver also for the BCM43362. this would be a very embedded-friendly chip (apart from its current closedness)18:42
cdewhat's wrong with ath9k? seems the most foss friendly wifi18:42
wpwrakit has binary-only firmware, potentially with redistribution issues. but i guess these could be worked around. with modules available.18:44
cdethe feasibility study is cool but it doesn't change the core problem which is the lack of open-source baseband18:44
wpwrakmth: and yes, the Realtek is an old but apparently viable alternative. olimex use it, too.18:45
wpwrakwhitequark: why do you need a module ? it's easy enough to build with chips18:46
whitequarkwpwrak: let's say a novice in electronics wants to build one18:47
whitequarkwell, a system on top of one18:47
whitequarkeasy enough for you with how much decades of experience? three? :)18:48
rohcde: softmac, not intended for embedded18:48
wpwrakwhitequark: does the result pass certifications ? or is it sufficient if it just works ?18:48
whitequarkjust works18:48
rohatleast not for stuff such low-power18:48
cderoh: not the wifi, the gsm/umts part (option)18:48
DocScrutinizer05wpwrak: PowerSavingMode18:48
cdeit's orders of magnitude more complex18:48
wpwrakwhitequark: then he could just copy any of my rf230 designs. there are now three tested ones to choose from: atben, atusb, and antorcha (function tested but rf performance not quantified). anelok probably works, too.18:49
wpwrakwhitequark: if he really really doesn't want to touch chips, he could find a GPIO-programmable uSD slot and toss in an atben.18:50
whitequarkwpwrak: yup, sounds doable18:50
rohcde: ah. hm... is it thin (the module)18:50
roh?18:50
rohbtw.. http://wireless.kernel.org/en/users/Drivers seems to be the best table ive seen so far about what exists and how the support is arm18:51
cdeI'm not following18:51
whitequarkwpwrak: that's what I thought18:51
DocScrutinizer05http://wiki.maemo.org/Wifi_Power_Saving_Mode_(PSM)18:51
wpwrakwhitequark: here's for example an atben in an STM32-E407: https://github.com/frtos-wpan/frtos-wpan/blob/master/doc/evb/e407-atben.jpg18:52
whitequarkwpwrak: how complex is the bitbang iface?18:52
wpwrakhere with an adapter board in a WM09 eval board: https://github.com/frtos-wpan/frtos-wpan/blob/master/doc/evb/wm09-dev9.jpg18:52
wpwrakthey all "work", though RF not quantified18:53
wpwrakwhitequark: pretty easy. you implement SPI and then it's a few register reads and writes. complexity is more in the higher layers.18:54
whitequarkwhat does it implement? PHY/MAC?18:54
wpwrakPHY and a bit of MAC, yes (CSMA, auto-ACK, etc.)18:55
whitequarkcool18:55
whitequarkthanks18:55
DocScrutinizer05wpwrak: http://www.google.de/search?q=wifi+PSM18:59
DocScrutinizer05cde: ^^^18:59
dos1[19:44] <cde> the feasibility study is cool but it doesn't change the core problem which is the lack of open-source baseband19:01
cdethanks DocScrutinizer05 19:01
dos1and that problem will persist for very long time19:01
cdeunless we tackle it19:01
dos1as it's pretty impossible to do properly and legally19:01
cdeit is actually. use a faraday cage19:02
dos1unless you don't want to use it :P19:02
cdeyes true19:02
wpwrakdos1: doing it properly alone would be a challenge. legally, well, you could probably always sell it as "lab equipment" :)19:02
wpwrakelse, pretend its purpose it to spy on friends. that seems to be a good enough excuse19:03
DocScrutinizer05good luck with UMTS, even more you'll need it with LTE19:05
DocScrutinizer05in ST-E a team of ~500 worked for ~1 year to make the UMTS&LTE stack work so that we could do first tests in carrier networks, and even then a 10min without breakdown of the connection were a huge success19:08
cdeyep. I know someone who used to work there19:08
cdethey sold everything to Broadcom, very sad19:08
whitequarkDocScrutinizer05: 500!!!19:11
whitequarkfuck.19:11
whitequarkdid you rewrite the damn thing from scratch?19:11
DocScrutinizer05and compared to LTE the CMU200 you need to check what your homegrown GSM stack does OTA is just the spare money19:12
wpwrakmaking something work can sometimes take many times the effort of doing it from scratch ...19:12
DocScrutinizer05dunno what you refer to, but ST baked the chip and ST & Ericsson built the UMTS and LTE radio stack from scratch and  made it work19:13
DocScrutinizer05google for NovaThor19:13
whitequarkDocScrutinizer05: I thought vendors just pull through the decades-old code19:13
wpwrakseems that they were smarter than to do that :)19:14
larscuntil things collapse over them19:14
DocScrutinizer05actually we didn't re-use the code of the previous chip generation19:14
DocScrutinizer05we peaked a lot19:15
DocScrutinizer05peeked19:15
wpwrakpeeked ?19:15
wpwrakah :)19:15
wpwrakthat's usually the most efficient approach19:15
wpwrakdo it your way but let existing proven code help you to find it19:15
larscbut sometimes it's just NIH19:16
DocScrutinizer05sometimes, yeah19:16
DocScrutinizer05when ST contributed ugly code that been replaced by other ugly code from Ericsson ;-P19:17
cdeDocScrutinizer05, NovaThor uses a modem from renesas right? (previously nokia)19:18
cdeafaik their new lte stuff was only used by Samsung19:19
DocScrutinizer05anyway, I see the probability to see a community-driven FOSS LTE stack any time soon to be == 0.000000000000001%/year19:19
cdeit was to be used in the new LTE NovaThor which sadly never came to be19:19
cdewell we have openlte which is already somewhat advanced19:19
DocScrutinizer05the modem is from ST19:19
DocScrutinizer05ST micro19:19
DocScrutinizer05and customers were all the big names19:20
DocScrutinizer05even Nokia, before they canceled that19:20
cdeDocScrutinizer05, are you talking about samsung? they used ST-E chips in a couple of phones19:20
DocScrutinizer05I'm talking about ST-E NovaThor chip19:21
cdeI meant the customers19:21
DocScrutinizer05yes, Samsung used them19:21
DocScrutinizer05rim, LG, dunno whom19:21
DocScrutinizer05nokia19:21
cdebut they failed, ultimately. now Qualcomm has all the market19:22
DocScrutinizer05possible19:23
DocScrutinizer05i left there a year ago19:23
DocScrutinizer05and probably for a reason19:23
cdeoh, you worked at ST-E? very cool! were you at the Rennes site?19:24
DocScrutinizer05Nuernberg19:24
DocScrutinizer05datacom19:24
cdeI see. well you were right to leave I guess. did you also work on the baseband?19:24
DocScrutinizer05nope, datacom19:25
DocScrutinizer05Nuernberg radio stack been 2 floors below my desk19:25
DocScrutinizer05I had the joy to deal with all the interfaces19:25
DocScrutinizer05HSI, UART, you name it19:25
DocScrutinizer05shared RAM :-o19:26
DocScrutinizer05but 19:29
DocScrutinizer05but Nuernberg had only a part of the radio stack team19:30
DocScrutinizer05and part of the chip makers19:30
cdebtw fun fact: NovaThor uses the plan9 filesystem :)19:30
larscSmartIron: http://www.bbc.co.uk/news/blogs-news-from-elsewhere-2470733719:31
DocScrutinizer05err, sounds somewhat familiar19:31
DocScrutinizer05larsc: ohmy19:33
DocScrutinizer05a little pcb with a microphone that infects all comouters with unprotected wifi in 600m radius, suuuure19:34
DocScrutinizer05ooops 600ft19:35
DocScrutinizer05silly enough19:35
dos1very clever idea though :)19:37
DocScrutinizer05I don't see how that flies19:37
dos1I mean, including wifi chipsets in devices like irons to send spam19:38
dos1brilliant, who would guess it? :D19:38
DocScrutinizer05maybe used as actual eavesdropping devices and sending the recorded audio via unprotected WiFi they find19:38
DocScrutinizer05ooh, that. Yeah that's probably actually feasible19:39
larscthe mic is probably in there because it was on the pcb anyway19:39
DocScrutinizer05honestly you'd need more than a internet connection to send out spam19:40
DocScrutinizer05doesn't sound plausible to me19:40
dos1replace sending spam with any other malware activity19:41
DocScrutinizer05why should you use a hijacked internet via open AP, when all you get is... internet. Which is dirt cheap anyway19:41
DocScrutinizer05it's about abusing the IT infra, thus about infecting PCs19:42
DocScrutinizer05you possibly could do that better from inside a WLAN rather than from outside, but if that really accounts for the expense and effort of that approach19:43
DocScrutinizer05sounds like a hoax to me19:43
dos1possibly19:43
dos1but still, neat idea19:44
dos1and scary19:44
dos1if you'd have to carefully check what sits in every electric device you own to just feel safe, that certainly wouldn't be fun19:46
DocScrutinizer05I think hijacked routermodems are waaaay more scary19:46
DocScrutinizer05and we seen that only a month ago19:46
dos1:nod:19:46
DocScrutinizer05with a lot of D-Link19:46
DocScrutinizer05I'd get really scared when the blackhats manage to hijack the DSL-multiplexers19:48
DocScrutinizer05or the real routers, like cisco19:48
DocScrutinizer05but I guess NSA would kick them off their lawn in a millisecond19:49
whitequarkDocScrutinizer05: they do20:23
whitequarkfor a long time20:23
DocScrutinizer05I bet on that20:23
DocScrutinizer05particularly in china et al20:23
DocScrutinizer05;-P20:24
DocScrutinizer05was it micro-VAX where they actually implemented hw-backdoor?20:24
whitequarkbtw talking about modems20:25
DocScrutinizer05or was it DECstation?20:25
whitequarkhttp://internetcensus2012.bitbucket.org/paper.html20:25
whitequarksomeone made a botnet of several 100k's of vulnerable modems (think admin/admin) and portscanned /020:25
DocScrutinizer05hehehehe20:26
DocScrutinizer05which brand/make?20:26
whitequarkvarious20:27
whitequarkI think they hijacked an existing botnet with a vulnerability20:27
whitequarkread the paper, it's very fascinating20:29
DocScrutinizer05indeed20:34
DocScrutinizer05HEHE  >>We would also like to mention that building and running a gigantic botnet and then watching it as it scans nothing less than the whole Internet at rates of billions of IPs per hour over and over again is really as much fun as it sounds like.<<21:25
--- Sat Nov 2 201300:00

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