#qi-hardware IRC log for Friday, 2013-04-26

mththe bootrom checks whether OSTCNTL >= 9766 * v100:10
mthwell, it explains why it reacts different to low values of v1, but what is it doing here?00:12
mth(OSTCNTL is the lower 32 bit of the OS timer)00:12
pcercueimth: I have no idea00:36
mththere seem to be 3 instances of "goto <register>"00:45
mth"jr t4" and "jalr s4" (2x)00:45
mththe "jr t4" jumps into SRAM #2, so it's not used when executing code from the cache00:46
mththis means we're being called by jalr and therefore have an address to return to00:46
mthin both cases, the code returned to then immediately jumps to 0xbfc0128c00:47
mthwhich is in the same routine as bfc0130c is, but earlier00:47
mththe code that sets v1 is before 0xbfc0128c though00:49
larscmth: we do have the bootrom asm06:37
larscwell for the jz474006:37
mthannotated asm or just objdump output?06:43
mthbecause I got the latter for jz4770 now06:43
mthby the way, preserving v1 and a0 fixes the problem06:44
mthI'll be pushing the fix to ingenic-boot later (other changes still need cleanups)06:44
larscmth: http://projects.qi-hardware.com/index.php/p/openwrt-xburst/source/tree/master/package/boot/uboot-xburst/patches/0002-qi_lb60-add-software-usbboot-support.patch06:57
larscand yes it is annotated06:57
mthI don't think this is the same code I'm talking about07:03
larscwell, it's the bootrom, at least the usbboot protocol part07:04
mthinside the SoC itself, there is a small ROM area which at address 0x1fc0000007:04
mths/at/is at/07:04
mthbut the code you linked has a comment "Relocate code to beginning of the ram"07:05
mthwhile the code in the SoC cannot bring up the RAM07:05
larscyes, I guess it has been modified to run in the ram instead of the rom07:06
mththere is some similarity in the way the USB controller is accessed, but the overall flow is very different07:15
mthI don't know if that's because of JZ4740 and JZ4770 differences or not07:15
mththe whole timeout mechanism that was causing problems for me doesn't exist in this version07:17
mththanks for the link though, it might be useful if I want to learn more about the protocol one day07:20
DocScrutinizer05hmmm... http://www.thinkgeek.com/product/b70c/   http://www.thinkgeek.com/product/eb92/   gives me ideas07:21
DocScrutinizer05I think I already seen IR-RC made like this07:21
mththey say it's great for children, but if it takes 2 hours to dry it's not really practical for experimentation07:23
larscI think the first thing I'd try is to see what happens when you use the conductive paint on skin ;)07:30
DocScrutinizer05if they actually use carbon fibre then this could get pretty nasty07:50
DocScrutinizer05aaaah, seems there's something left over from Iraq war07:51
DocScrutinizer05didn't they short power lines with carbon fibre bombs07:52
whitequarkDocScrutinizer05: huh?07:54
DocScrutinizer05========== <-powerline   //\\/\/<- crbon fibres07:55
DocScrutinizer0550kV lines07:56
DocScrutinizer05or 100kV07:56
DocScrutinizer05they took power plants out by shorting the wires07:57
larsc¡¡¡ ¡¡¡07:57
DocScrutinizer05now they probably recycled the surplus bombs and played "does it blend?" with the carbon fibres, then mixed it with wallpaper glue07:58
whitequarkwon't that result in a huge discharge arc and failsafe switch on the plant?07:59
whitequarkand then it'll turn back on again07:59
larscif it has one07:59
whitequarklarsc: er... some of them don't?07:59
DocScrutinizer05hmm, seems the shorts been permanent07:59
whitequarkDocScrutinizer05: I'm pretty damn sure the fiber will be vaporized07:59
whitequarkyou can find lots of videos like that on 'tube07:59
DocScrutinizer05it's carbon07:59
whitequarkit's not enough carbon to dissipate several MW08:00
DocScrutinizer05the first few fibres take out the power plant and might evaporize or rather burn, the rest will happily sit on the powerless wires and short any next effort to power up the line again08:01
DocScrutinizer05also wire cut is likely08:01
whitequarkDocScrutinizer05: (short any next effort) isn't the only thing you need to do is to power up the line? it will remove the fiber itself. the arcing probably will be a problem, however.08:02
whitequarkI'm not sure to what extent will the wires degrade08:02
whitequark> During the Gulf War Iraq responded to the use of this type of munition by disconnecting electrical power grid circuit breakers. 08:03
DocScrutinizer05I think the more protective gear they have, the more trouble to power up a shorted line08:03
whitequarkI also heard some of that gear is both very expensive and one-time use08:03
whitequarkyeah exactly... it seems that the arcing did damage the wires08:04
Action: DocScrutinizer05 wonders of BGA can be glued instead reflowed08:16
DocScrutinizer05you for sure need perfect glue application and component positioning, since no self centering and small surface tention of glue08:18
DocScrutinizer05but hey, no more reflow, no more heat. No more water steam explosions in balls08:19
lindi-<pabs> interesting, the Broadcom Videocore reverse engineering folks have custom code running on the part of the RPi GPU responsible for booting08:23
whitequarklindi-: yeah, except they don't really know anything about videocore and can neither activate the framebuffer (not even talking about 3D) nor boot the ARM core08:38
Action: whitequark tries to get NFC working on his phone08:39
whitequarkeither bad phone or bad tags :/08:39
whitequarkthey are kinda write-only, and I think only a single attempt of several dozen succeeded08:39
whitequarkhmmm interesting09:01
whitequarkit seems that first 8 bytes of even a clean tag are non-random garbage09:02
DocScrutinizer05serial # ?09:15
whitequarkDocScrutinizer05: nope09:37
whitequarkserial is before that09:37
whitequarkthe weird thing is: it's an NXP tag and NXP tag reader (does anyone else make them? all NFC stuff I could find at my place was by NXP)09:38
whitequarksooo I clean the tag, and apart from system info it contains 01 03 A0 10 44 03 00 FE09:38
whitequarkand this is correctly recognized by the tag reader as an empty NDEF record09:39
whitequarkbut when I write the actual info to the tag, it contains 01 03 A0 10 44 44 0E and an actual NDEF record09:42
whitequarkwhich I can recognize manually09:43
whitequarknothing in the prologue seems to be a valid record start09:43
whitequarkFE seems to be an epilogue (actual info also contains it at the end) but it is not a valid record according to the NDEF standard09:44
whitequarkmaybe there's more framing going on09:44
whitequarkwhy do I have to do this at all! >_< it should be a consumer technology. easy to use yadda yadda09:44
whitequarkHWSON8: plastic thermal enhanced very very thin small outline package09:57
whitequarkindeed, there's another TLV structure there10:06
whitequarkhmmm right10:08
whitequarkinstead of "44 0E" there should've been "03 0E" for it to be a valid T/L for an NDEF record10:09
whitequarkI wonder why it writes 44 there...10:09
kuribasCan the nanonote linux kernel be used for other ingenic devices?18:19
kuribasI bought a very cheap linux ereader (http://www.walmart.com/ip/iTomic-IEBR7C-7-eBook-Reader/19887947) and I would like to put linux on it.18:20
kuribasThe processor is a jz4755.18:21
pcercueinot easily18:21
kuribasThere is little documentation on the chip, however I found a patched linux kernel from the manufacturer.18:23
larscthe jz47xx kernel tree has support for other SoCs like the jz475518:42
larscbut unless you know what you are doing you probably wont get somewhere easily18:42
wpwrakthat sounds like a universal truth :)18:50
kyakbuy ereader, try to put linux there, get too much involved, never use that ereader to read books :)18:59
whitequarkoh yes18:59
kuribasI didn't intend to read books with it :)19:02
--- Sat Apr 27 201300:00

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