#qi-hardware IRC log for Friday, 2012-01-13

whitequarkpabs3: it may be a good x86 tool, but what's good for x86 isn't for a particular stack VM. so I think the integration would not be efficient.00:01
pabs3seems to load arm binaries fine00:01
whitequarkinteresting00:02
whitequarkanyway... Currently Bokken is neither an hexadecimal editor nor a full featured disassembler YET, so it should not be used for deep code analysis or to try to modify files with it.00:02
whitequarkI aim exactly for "deep code analysis", and I don't see a viable infrastructure for plugging that in00:03
pabs3its definitely not a mature project, interesting anyway00:04
whitequarkyes, it's interesting, and I see that some parts of GUI are cleverly done00:05
whitequarkbut sadly not the codebase00:05
whitequark(decompiler) my thingy is already a complete disassembler, it can not assemble (because it's hard to define the operation of assembling an AS file), but it will quite certainly be a complete decompiler00:06
whitequarkbecause some transformations require a full AST representation to be composed, and when you have full AST, you have just one step to get the source00:06
whitequarkbut I think I won't do any SWF decoding00:07
pabs3flasm does both assembly/disassembly btw. and the author has a non-free decompiler (to ActionScript): http://www.nowrap.de/flare.html00:08
whitequarkas3 assembly isn't well-defined, and you don't have much use for it anyway, as it's a big pain to code directly for flash00:09
whitequarkyou need to look for variable, stack and scope compatibility at the very least00:10
whitequarkso I don't see a use for it00:10
whitequarkmaybe that was different for AS2, but AS3 VM is rather stupid00:13
pabs3ok. anyway I guess this is pretty off-topic here :)00:13
qi-bot[commit] Paul Cercueil: fbcon: add parameter to toggle bind on startup. (jz-3.2) http://qi-hw.com/p/qi-kernel/89aa70700:20
qi-bot[commit] Paul Cercueil: MIPS: JZ4740: SLCD: added a parameter to select the panel. (jz-3.2) http://qi-hw.com/p/qi-kernel/a7bcc4e00:20
qi-bot[commit] Paul Cercueil: OpenDingux: initrd: updated mininit to latest version. (jz-3.2) http://qi-hw.com/p/qi-kernel/848db1400:20
guest_3423 I am starting a company and wondering where I can buy the best card readers (for debit/credit cards ) , and barcode scanners / makers00:50
guest_3423 I.E what would be the popular companies that make these devices for whole sale00:50
guest_3423 And how much would these devices typically cost.... A link /site/or phone number would be nice. Thanks for any help as I am new to xchat and somebody told me I should try an irc channel don't really know if this is the place , first time on irc00:50
guest_3423I am excited since this is going to be my first business.... so I don't want to get ripped of on this stuff00:52
guest_3423I have the building inspect or a OK it. And now the network and computer hardware is almost setup.... All I need is the cash registers , barcode readers/writters , and credit/debit card scanners. Anybody know sorry if I am being pushy I am excited00:54
pabs3this channel is for people making copyleft hardware, I don't think they make any of the hardware you are looking for, sorry: http://qi-hardware.com/00:54
Artyomkristianpaul: hi! Are you here?05:24
kristianpaulAryom yes.13:47
kristianpaulArtyom yes !13:48
qwebirc66556anybody knows how to make opkg update all packages?15:33
rjeffriesas an example of what's possible re specs for an under $100 handheld with (resistive) touch screen and lots of useful functions there's this $90 game unit running Android16:10
rjeffriesno, it is not copyleft. but quite an interesting reference point IMO as to what is possible.16:11
rjeffrieshttp://www.isharegifts.com/jxd-s601-android-game-retro-handheld-wifi-tablet-p-1615.html16:11
wpwrakscreen's too tiny. for milkymist, it would be nice to find a cheap 7"-10" screen, can be extra-wide if 10", with capacitative touch. as a finished module that doesn't need too many I/Os pins. ideally operating on a 5 V or 3.3 V supply. if you come across something like that, this may be interesting16:19
rjeffrieswpwrak so you are thinking of a controller for Milkymist? that would make sense in the VJ conext. I understand iPads are used in that scene (not w/MM)16:55
rjeffrieswhen OLPC ver 3 actually ships, the specs will be close to what you need. Linux already on board. although there are plenty of OLPC haters. they may have gotten it right this time.16:57
rjeffries"they" referring to OLPC, *not* the OLPC haters. ;)16:59
wpwrakrjeffries: yeah, just like they use iPads today (btw, M1 does actually support OSC, which is a common protocol for carrying MIDI and similar over TCP/IP)17:02
wpwrakwe probably only need the display, not an entire device. but yes, some rock bottom cost droidblet may do, too. i'm not entirely sure what would be the less painful path from an engineering point of view - build up from scratch or defeat all the obstacles the sw/fw/hw of a ready product imposes17:04
rjeffrieswpwrak finding something that's already packaged would be a win. no harm if the owner can also use it for non-MM tasks when he/she doesn't need it for MM17:21
rjeffriesmaybe you go client server, not unlike a VNC approach?17:23
rjeffriesor is that too clumsy?17:23
wpwrakthe harm would be more in making it work with our environment. e.g., finding a means of communication that works for both side. supply power. and so on.17:23
rjeffrieswell all tablets are battery operated, or can be charged over USB17:24
rjeffriesMM has USB and many tablets have USB17:24
wpwrakyou basically have to go up a few layers on one side, then connect on the usually complex point of compatibility, then go down again. if you just get the naked component, you save all that. and it's the same for software.17:24
wpwrakyes, if power over usb works, that's good. if communication over usb works with, say, OSC, even better.17:25
rjeffriesI wonder if MM does feed out 5V on USB. you'd thinks so...17:25
wpwrakyou'd still have a log of duplication of functionality, though17:25
wpwrakit does17:26
rjeffriess/log/lot/ ;)17:26
wpwrakup to 500 mA17:26
rjeffriesunderstood17:26
wpwrakyeah :)17:26
rjeffriesneed to look up OSC protocol17:27
rjeffriesok, simple enough protocol. what is use case for this touch pad controller for MM?17:31
wpwrakrjeffries: same as midi. plus some extra control functions. e.g., image selection, patch selection. basically array of images and you pick one. basic stuff that's currently absurdly messy.17:34
viricanyone using F8 in the nanonote?18:51
viricIt looks to me like linux considers SYSRQ and F8 two different key values18:52
viricand sysreq happens always on leftalt+sysrq.18:52
viricI can't make it happen on leftalt+f8(value). I've to change the value of the f8 key to sysrq.18:52
kristianpauli dont use any F keys*19:03
kristianpaulArtyom: hi 19:04
kristianpauls/F/Function19:04
Artyomhello Kristianpaul :)19:04
Artyomhow is your work with namuru?19:05
kristianpaulhaving some issues getting the new data..19:05
kristianpauli mean the flag19:05
ArtyomAnd I have some good news :)19:06
kristianpauloh 19:06
kristianpaultell us :)19:06
ArtyomI had to switch to verilog19:06
kristianpauloh !19:07
kristianpaulwhy? :)19:07
ArtyomI connected namuru to wishbone19:07
Artyomwrote a test19:07
Artyomand it was successful19:08
ArtyomAnd finally I used the program that I have already published and it worked19:08
kristianpaulbut all this was with verilog version?19:09
ArtyomSo single channel acquisition, confirmation and pull-in is working on MM SoC (Digilent s3e500 port)19:09
kristianpauli get lost19:09
kristianpaulwonderfull !19:09
Artyomyes all is in verilog19:09
ArtyomI need some time to clean the code19:10
Artyomand then I will publish it19:10
kristianpaulwhat happened with vhdl?19:10
kristianpaulso the test program was ok?19:10
kristianpaulyou said last time had some issues with wrong dealays or such19:10
kristianpaulthose are great news anyway ! :)19:11
ArtyomThere was one stange thing when I tried to connect vhdl-namuru-port to wishbone. I had to work on falling edge of clock (otherwise there were errors)19:12
kristianpaulyou just rewrote the whole vhdl code to verilog? including reset registers?19:12
kristianpaulerros.. hmm may be ise was very happy mixing vhdl and verilog ;-)19:13
ArtyomAnd I decided that it would be better to have everything in verilog (as MM SoC is written in verilog). If I would find time to test gaisler's LEON3 then I would use vhdl ;)19:13
Artyommay be ISE, may be my weak knowledge of HDL ;)19:13
ArtyomI just took namuru code, rewrote my test of each module from vhdl to verilog and I also studied VPI (or PLI) extension of verilog. 19:14
kristianpaulhow are you using VPI?19:15
kristianpaulor with wich porpuse?19:15
kristianpaullooks interesting  i dint knew it about it19:16
Artyomverilog's vpi is very interesting thing. It allows to call C-functions from verilog code. So I could debug at the same time verilog and C-code. It's very powerful in my opinion. The main disadvantage is the speed. 3 seconds are simulating 1.5 hours19:16
viricCan't linux be built with "-O0" ?19:16
virichttp://lkml.indiana.edu/hypermail/linux/kernel/0809.1/0711.html19:17
viricCompiling with -O0 never worked, and is not likely to ever work.19:17
viricnice.19:19
kristianpaulArtyom: what do you do after read the accumulator, i mean you re-initialice wich registers?19:20
kristianpauli think i missing one, i was getting new data to work now is not.. :-/19:21
kristianpaulwell i havent looked too much the code i was bit away last week..19:21
kristianpaulbut now i think i should :)19:21
ManuZzhey guys i have s super offer for all of you - but only one will get it! check this!! --> http://www.ebay.de/itm/OCZ-Vertex-3-240-GB-Intern-2-5-Zoll-/110808896734?pt=DE_Elektronik_Computer_Computer_Festplatten&hash=item19ccb994de     It is pretty CHEAP!19:21
Artyomkristianpaul: after reading 6 accumulators, I make some calculations to generate new code_frequency and carrier_frequency (during pull-in or tracking). And during acquisition I make next code_slew if threshold is not passed.19:24
viricthat kernel thread is a must-read for kernel beginners like me, I think.19:25
kristianpaulArtyom: but no the prn code right?19:25
Artyomkristianpaul: And I start reading 6 accumulators only after status[1]=1 and new_data=1. prn code is set only once (next time only after you want to start to search for new satellite)19:27
kristianpaulArtyom: in your HDL namuru do you load ch0_epoch_enable bit by a separate register or just after ch0_epoch_load is loaded?19:29
kristianpauli think was idea of mine to add separaate enable register for h0_epoch_enable,ch0_slew_enable,ch0_prn_key_enable19:30
kristianpauland separate register to clear new_data_read,status_read flags19:30
kristianpaulor what do you think?19:30
Artyomkristianpaul: Why do wnat ot use separate enable registers: ch0_epoch_enable, ch0_slew_enable, ch0_prn_key_enable? And why do you need separate register to clear new_data_read and status_read flags?19:40
kristianpaulwhy.. well19:40
viriclarsc: after some time not touching any key, my screen got 'white' instead of switched off or black19:40
viriclarsc: time to attach a gdb, if I can reproduce it again. I hope then I'll be able to check some registers19:41
kristianpauli dont remenber i dint like how it was implemeted in namuru upstream19:41
kristianpaulhow do you implemented it? :-)19:41
kristianpaulthw enable after write and clear after read :)19:42
kristianpauldo you get my point?19:42
Artyomkristianpaul: I previously used some additional commands (like clear_status and clear_new_data) because I had difficulties with connecting namuru-correlator to ARM through asynchronous memory bus (SRAM-like interface). But now I use MM SoC with synchronous wishbone bus. And now I got rid from these commands (status_clean and new_data_clean)19:44
kristianpaulI see19:45
Artyomkristianpaul: Now I use exactly namuru-code (the only difference is that I have implemented wishbone bus instead of avalon-bus)19:46
kristianpaulhmmm, dont do it for some reason but i cant remenber.. well if it works for you19:49
Artyomkristianpaul: I think i remember... There ware couple of bugs with reset-signal. May be something else. I've corrected them and forgot immediatly ;)19:50
kristianpaullol19:50
kristianpaulbut i guess is avalin was ok, this way should work https://gist.github.com/160835719:51
kristianpauls/avalin/avalon19:51
kristianpaulyeah reset signal was buggy indeed19:51
kristianpaulactually i guess as a result of implement a software reset and system reset.. dont remenber now either ..19:52
kristianpauloops19:53
Artyomyes, you are right... After sending software reset everything continued to be in reset state forever19:53
kristianpaulhttps://gist.github.com/1608357 fixed19:54
kristianpaulArtyom: ^ this does look okay for you?19:54
kristianpaulokay i have some todo, just for been sturnbon and dont migrate the avalon as it should ;)19:55
Artyomwait a second...19:55
kristianpaulk19:55
kristianpaulis load not enable, my head is a mess with all this names..19:56
kristianpaulsorry confusion (in the gist)19:56
kristianpaulah no is ok19:58
Artyomkristianpaul: http://pastebin.com/zEKgi3KR19:59
kristianpaulhmm i tought stb was optional20:00
kristianpaulunless the write cycle begin, but yes i need confirm for this the transfer cycle as well20:01
kristianpaulargh20:01
Artyomkristianpaul: I've copied this code from some bram-core20:02
kristianpaulhe20:02
kristianpaulokay20:02
kristianpaulbtw you feek the s3e500 is getting short in resources now?20:03
kristianpauls/feek/noticed20:04
ArtyomI tried to use only two of three (don't remember which exactly) in my vhdl code, like in the description of wishbone bus. And it didn't work. So I just copied the code from the bram-core20:04
larscviric: if you ware using 3.2 this is expected behaviour since it misses the patch which fixes it20:06
kristianpaulArtyom: yup, ram0we (from bram), you always right !20:06
kristianpaulok i'll do some fixing20:06
Artyoms3e500 is rather small. I use about 68% of it for lm32 without cache + single channel correlator. (no vga, ethernet, memorycard and all other unneccesary cores). But it's just a check of concept20:06
kristianpaulArtyom: ok now i need to implement the other arms on the milkymist and make sofyware support it :)20:07
kristianpaulplus migrate the osgps part that get the fix..20:07
kristianpaulbtw now the late and early arms are behaving right?20:07
kristianpaulBTW check this http://sdr.osmocom.org/ when have some time, the IF tunner supports a wide range of freq including GPS L1  ! :)20:08
Artyomno, still some unussual behaviour. I noticed it in the testbench of namuru-core. That's a question for the future ;)20:09
kristianpaulhmm20:09
ArtyomI didn't run simulation for a long time. May be it's just a transient and after several seconds everything will be fine20:12
kristianpaulgota go i need do some herrands before 18:00 here, sorry..20:13
kristianpaulbut thanks for the update and head up, now me behind ;) need move again..20:14
Artyomno problem, bye! It's time to sleep for me ;)20:14
kristianpaulgood sleep then20:14
Artyomthanks :)20:14
kristianpauls/herrands/errands20:15
viriclarsc: ah, where is the patch?20:22
viriclarsc: something to be ported? Maybe I could try to do that20:23
larscviric: patches are here http://www.spinics.net/lists/linux-fbdev/msg04503.html21:45
larscthose two have to be added to the kernel tree and then our lcd driver needs to be adjusted to use the new callbacks21:46
--- Sat Jan 14 201200:00

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