#qi-hardware IRC log for Wednesday, 2013-10-30

whitequarklol, my android phone just BSOD'd12:40
whitequarkthe modem crashed12:40
whitequarkso it displays a blue screen with white text with instructions of how to get a memory dump12:41
kristianpaulhehe12:50
cdewhitequark: which brand/model?12:54
whitequarkcde: lg e97313:01
whitequarkit was fine on an unofficial cyanogenmod port13:02
kyakandroid phone BSODed? how is that possible?13:02
whitequarkthen I for some reason decided I should upgrade it to the official port done by the same person13:02
whitequarknothing works since13:02
whitequarkwtf13:02
whitequarkkyak: well, it's not windows' bsod, but it is blue and signifies death13:02
kyakah, not *that* BSOD! :)13:03
whitequarkwell duh13:03
whitequarkif it was winphone, then it'd be possible though.13:03
cdewhitequark: can you reproduce the blue screen and take a picture of it?13:06
cdeit is relevant to my interests13:06
whitequarkcde: I don't think so but if it repeats I can do that13:08
whitequarkwhich interests?13:08
kyakblaming lg :)13:09
cdewhitequark: professional13:25
whitequarkcde: you work for lg and collect bsods of custom firmwares? :p13:34
cdeunfortunately, I'm not allowed to say13:35
whitequarkI can take a memory dump if you want13:36
whitequarkpretty sure the required sw is leaked somewhere13:36
whitequarkyea it is13:36
kyakthe question is whether it is allowed to use memory dumps of custom firmware taken by leaked tools :)13:37
cdewhitequark: awesome! which tool?13:37
whitequarkcde: qpst13:37
cdeah cool. I have it too, but never tried on an actual phone13:38
cdeyou need to switch the phone to modem mode right?13:38
whitequarkdownload mode, basically a bootloader13:38
whitequarkbsod handler can do that on its own13:39
cdeawesome, I had no idea it could do that13:39
whitequark(also, whoever wrote it accidentally swapped the only two keys used in its UI. grr)13:39
whitequarkwell, it writes that in plain text13:39
apeletelarsc mth: Hello there21:30
apeletestill trying to understand why the udc does not seems to be detected as it should (bulk endpoint not being detected -> http://paste.debian.net/60583/, and musb->endpoints member being left initialised to O in struct musb -> http://paste.debian.net/62655/)21:49
apeleteI read something interesting in the jz4740_udc code:21:52
apelete/* Let host detect UDC:21:52
apelete * Software must write a 1 to the PWR:USB_POWER_SOFTCONN bit to turn this21:52
apelete * transistor on and pull the USBDP pin HIGH.21:52
apelete */21:52
apeleteusb_setb(dev, JZ_REG_UDC_POWER, USB_POWER_SOFTCONN);21:52
apeletelooking in programming manual (21.4.3.221:53
apeletes/looking in programming manual (21.4.3.2/looking in programming manual (21.4.3.2 POWER), bit 6 used in jz4740_udc.c is advertised as "Version specific" but undocumented/21:55
apeletebut section 21.5.1 titled "Soft connect/disconnect" explains usage of what is called "PWR:USB_POWER_SOFTCONN bit" in the jz4740_udc code21:59
apeletelarsc mth: is that SOFTCONN mechanism standard to musb controllers or do you think I should implement it specifically in the jz4740_musb driver too ?22:00
mthapelete: is there anything in the generic code that can write that particular bit?22:40
mthunsignedsoft_con:1 __deprecated; /* soft connect required */22:40
mththis exists in include/linux/usb/musb.h22:40
mthyou can use "git blame" to look up the commit in which it was deprecated22:40
mthusually the commit comment then says what it was replaced by22:41
DocScrutinizer05we started funraiser: neo900.org22:49
apeletemth: found the commit:22:49
apelete$ git show c58bfa22:49
apeletecommit c58bfa6b97731590e42cba6bd13829c4e480992f22:49
DocScrutinizer05fund, but he!22:49
apeleteAuthor: Felipe Balbi <felipe.balbi@nokia.com>22:49
apeleteDate:   Thu Jan 21 15:33:55 2010 +020022:49
apelete 22:49
apelete    USB: musb: deprecate what we don't use22:49
apelete    22:49
apelete    after 2.6.34, those fields will be removed from22:49
apelete    struct musb_hdrc_platform_data, it's expected22:49
apelete    that other architectures are fixed by then.22:50
apelete    22:50
apelete    Signed-off-by: Felipe Balbi <felipe.balbi@nokia.com>22:50
apelete    Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>22:50
DocScrutinizer05felipe balbi, UGH22:50
DocScrutinizer05KH is ok, but this balbi guy seems a good indicator for stuff that's giving you headache22:51
whitequarkDocScrutinizer05: ooooh a keyboard22:53
apeletewell, that commit message does not help finding out what should be used instead of those deprecated fields22:53
whitequarkI haven't seen a phone with a keyboard for years22:53
whitequarkoh and LTE. hm.22:53
Action: whitequark looks at his wallet22:53
whitequarkDocScrutinizer05: is there a full list of specs?22:54
mthDocScrutinizer05: I can't judge his code quality, but he did respond quickly and friendly when I mailed him once22:54
DocScrutinizer05whitequark: click on "specs" - though "subject to change"22:55
whitequarkresistive display, huh22:55
mthapelete: that sounds like the generic code ignores it, so then the glue would have to take care of it22:56
apeletemth: unsigned  softconnect:1;23:00
apeletethat's in struct musb in musb_core.h23:00
apeleteso it seems it has been moved there23:00
mthah, it still exists but it is no longer platform data?23:01
apeletewell, found it in musb_core.h, I don't know how it is supposed to be used23:02
apeletemth: do you think the glue should write that field to use soft connect capability ?23:03
mthonly if the generic code doesn't23:03
apeletegeneric code does not seem to handle it, but I may be wrong here23:03
mthlook for writes to that particular register offset23:04
mththe generic code might use different names for it, but the address will be unique23:04
DocScrutinizer05whitequark: http://neo900.org/faq#touch23:08
whitequarkthat's... wishful thinking23:09
whitequarkfor one, it's not multitouch. completely different pattern of interaction23:09
whitequarkthough it's not going to matter, considering all FOSS shells are stuck in 90's anyway23:10
DocScrutinizer05well, we don't have multitouch in any of it's ancestors either23:10
whitequarksure. I mainly wanted to know whether it's still in the "it's way inferior but yeah FOSS" category23:10
whitequarkgot better, but still there.23:11
Action: whitequark shrugs23:11
whitequarkpersonally, I'm somewhat torn between "a foss phone" and "fuck phones completely"23:11
wpwrakboth at least have the "f" in common :)23:12
wpwraka third option would be "feature phone"23:12
whitequarkwpwrak: well, I use my current android phone, in order of importance: GPS nav, SIP, web, SMS23:12
whitequarkso, uh, feature phone ~ no phone at all23:13
whitequarkit's great in emergencies, but not as something I'd use daily.23:13
wpwrakah, my car nav can't make phone calls :)23:14
wpwrakand my feature phone does SMS and voice. that's all.23:14
whitequarkSIP & web are good enough on the notebook23:15
whitequarkGPS's kinda inconvenient23:15
apeletemth: you're right, it is handled in generic code indeed:23:17
apelete$ find drivers/usb/musb/ -type f -exec grep -Hin "MUSB_POWER_SOFTCONN" {} \;23:17
apeletedrivers/usb/musb/blackfin.c:265:val &= ~MUSB_POWER_SOFTCONN;23:17
apeletedrivers/usb/musb/blackfin.c:270:val |= MUSB_POWER_SOFTCONN;23:17
apeletedrivers/usb/musb/musb_gadget.c:1616:power |= MUSB_POWER_SOFTCONN;23:17
apeletedrivers/usb/musb/musb_gadget.c:1618:power &= ~MUSB_POWER_SOFTCONN;23:17
apeletedrivers/usb/musb/musb_regs.h:46:#define MUSB_POWER_SOFTCONN0x4023:17
apeletelooks like set_vbus() call in musb glue will need fixing23:18
apeleteDocScrutinizer05: just read http://neo900.org/faq, I was thinking the device would ship with a phone OS (no matter how limited), ie. could make phone calls and/or sms out of the box23:36
DocScrutinizer05we will ship a rudimentary dialer23:36
DocScrutinizer05actually I guess by the time we're shipping SHR should be ported and ready to use23:37
DocScrutinizer05after all it's working on GTA04 already afaik23:38
apeleteDocScrutinizer05: great. so the device would be more of a foss dev phone than an actual ready to use phone, like the GTA04, right ?23:41
DocScrutinizer05right23:41
DocScrutinizer05GolDeliCo doesn't ship a ral OS - that's up to commuity23:42
DocScrutinizer05we're busy porting fremantle23:42
DocScrutinizer05see23:42
DocScrutinizer05http://talk.maemo.org/showthread.php?t=91142 23:42
DocScrutinizer05link on post#123:42
apeleteDocScrutinizer05: that's what I was going to ask: "who's working on it and how far has it gone ?" :)23:45
apeleteit's great to see an attempt at building a phone with a keyboard, really23:45
DocScrutinizer05see FPTF thread, though the devels are not really updating that one frequently23:45
DocScrutinizer05well, hw kbd been one of the must-have's23:46
DocScrutinizer05otherwise I wouldn't have left my bed for this23:46
apeleteI'm still using a Nokia N9, but wish a phone with a keyboard will be available by the time the N9 isn't working anymore23:50
apeleteDocScrutinizer05: going to keep an eye on this, let's see if I can give a hand when that little musb project of mine is ready :)23:57
Action: apelete looks at his wallet too...23:57
DocScrutinizer05:-)23:57
DocScrutinizer05much appreciated23:57
whitequarkfrankly it's tempting23:58
whitequarkbut I just don't want to have yet another phone that doesn't work, for the cost of iphone this time23:58
--- Thu Oct 31 201300:00

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