**** BEGIN LOGGING AT Wed Oct 15 02:59:57 2008 Oct 15 07:53:39 morning Oct 15 07:57:42 Sleep_Walker: morning Oct 15 08:21:11 BobOfDoom: do you have somewhere values of Treo650's power management registers? Oct 15 08:35:41 Sleep_Walker: they're all zero Oct 15 08:35:55 :O Oct 15 08:36:56 palmos doesn't set them until just before it goes to sleep Oct 15 08:37:01 and am I right you don't touch PSLR, PSTR, PVCR... Oct 15 08:37:08 oh, I see Oct 15 08:38:10 yep Oct 15 08:39:05 hmm Oct 15 08:39:16 PVCR is set by palmos to 0x14 Oct 15 08:39:32 PSTR = 0 Oct 15 08:39:37 same for me Oct 15 08:39:57 PVCR: Oct 15 08:39:59 06:00 = 0x14 - The seven-bit address of the external regulator’s I2C module. Oct 15 08:40:01 PSLR = 0xcc000000 Oct 15 08:40:18 (due to pxa27x doc) Oct 15 08:40:33 all the wake ones are zero Oct 15 08:40:46 like PWER etc Oct 15 08:40:59 hmmm, I've got it set Oct 15 08:41:24 it gave me some GPIOs to think about :) Oct 15 09:39:10 hm, on boot I've got sometimes keyboard backlight or red led switched on :( Oct 15 09:39:22 it's touched somewhere :( Oct 15 09:40:49 isn't the backlight normally on? Oct 15 09:41:04 or do you mean stuck full-on? Oct 15 09:42:39 I mean it was off but during kernel initialization it is turned on Oct 15 09:42:58 hm... that red led is behaving strange... Oct 15 09:43:11 sometimes it works, sometimes it does not Oct 15 09:45:14 now it was stuck on boot and I can't switch it Oct 15 09:51:06 hmm Oct 15 10:08:07 * Sleep_Walker needs some application to switch phone on/off - holding red button is not working since system is still busy :b Oct 15 10:10:45 hm, I've got finally dumped chip select2 region Oct 15 10:12:04 most of time only zeroes Oct 15 10:22:14 the whole lot is zeros? Oct 15 10:22:39 I guess it doesn't have ASIC6 then... maybe it has something else Oct 15 10:22:56 I dumped from 0x00 to 0xac Oct 15 10:23:10 it sometimes changes to something Oct 15 10:23:13 but rarely Oct 15 10:23:24 I have to create new dumps for that Oct 15 10:49:39 what the heck is this?! http://sleepwalker.hackndev.com/CS2REGS-tel-on-tables.txt Oct 15 10:50:08 does anyone know what communication this can be? Oct 15 10:51:16 BobOfDoom: you've got there that chars (H.T.C.-.3.0.H.8.0.0.4.9.-.0.0) in memory dump of that location, right? Oct 15 10:51:48 fafffa Oct 15 10:51:55 yeah Oct 15 10:52:00 that looks completely different Oct 15 10:52:05 fcuk Oct 15 10:52:46 BobOfDoom: but it doesn't look like assigning values into registers, right? Oct 15 10:52:56 I reckon maybe whatever is connected is off Oct 15 10:52:57 it looks like some bus?! Oct 15 10:53:04 and that fffafffa stuff is just gibberish Oct 15 10:53:27 BobOfDoom: that was dump from PalmOS during phone switching on and attaching network... Oct 15 10:53:36 oh Oct 15 10:53:50 hmm Oct 15 10:53:53 I dunno Oct 15 10:54:00 I can imagine that it is blinking red led :b Oct 15 10:54:03 that's really pretty weird Oct 15 10:55:20 BobOfDoom: dmitry said something like that bcm2133 chip is attached through USB... can this be somehow related to that? Oct 15 10:55:50 * Sleep_Walker doesn't know, how raw usb communication looks like Oct 15 11:01:12 if it was connected via USB I'd expect it to be connected to the PXA's USB ports, not through some chip select Oct 15 11:01:16 ooh Oct 15 11:01:26 well that explains why they don't need ASIC Oct 15 11:01:33 or at least not the ASIC6 Oct 15 11:01:36 if it is USB Oct 15 11:03:45 who knows if it is true :( Oct 15 11:05:06 well, I compiled in ohci, mounted USBfs but I saw nothing there :/ Oct 15 11:05:30 but maybe I'm still missing something Oct 15 11:06:14 hm, but at least it makes sense to disable map_io copy&pasted from Treo650 :/ Oct 15 11:14:05 hm, I probably ommited something important in palmt680.c for USB host :B Oct 15 11:28:51 aaah Oct 15 11:28:59 UsbHostAoe.prc Oct 15 11:29:05 UsbHostCdc.prc Oct 15 11:29:09 UsbHostHcd.prc Oct 15 11:29:14 UsbHostHub.prc Oct 15 11:29:18 UsbHostUsbd.prc Oct 15 11:29:23 UsbRadioSerialBulverde.prc Oct 15 11:29:27 looks like dmitry is correct Oct 15 11:29:53 great :/ Oct 15 11:30:01 and no ASIC6 prc Oct 15 11:30:05 it will be fun Oct 15 11:30:10 yes Oct 15 11:30:15 but you know what's really cool? Oct 15 11:30:24 we can probaby access these as test points Oct 15 11:30:35 if we can find the signal Oct 15 11:30:39 but if it has a hub Oct 15 11:30:42 it should be easier Oct 15 11:30:47 cause we can trace from the cub chip Oct 15 11:30:51 hub* Oct 15 11:30:59 so you can hack on other USB devices Oct 15 11:31:34 damn Oct 15 11:31:39 I really want one now Oct 15 11:31:42 XD Oct 15 11:32:00 it'd be a fun project to try and hack a USB wifi dongle on :D Oct 15 11:32:24 hmmm Oct 15 11:32:41 I'm copy&pasting settings from spitz Oct 15 11:32:52 I hope I can't damage anything :( Oct 15 11:33:20 USB settings? Oct 15 11:33:20 hmm Oct 15 11:33:30 is there no USB AF settings? Oct 15 11:33:35 I'd rather dump UPxOCR registers first Oct 15 11:34:24 Sleep_Walker: do you have some random GPIO.txt with PXA27x registers lying around? if not don't worry, am just curious Oct 15 11:34:31 oh Oct 15 11:34:46 I wonder what the other thing connected via USB Oct 15 11:34:49 maybe bluetooth Oct 15 11:35:18 BobOfDoom: I've got plenty of them Oct 15 11:36:15 Sleep_Walker: cool can you post one? Oct 15 11:36:40 hmm hopefully BCM2133 speaks some standard USB prtoocol Oct 15 11:37:00 UsbHostCdc.prc <-- sounds like it Oct 15 11:38:34 BobOfDoom: http://sleepwalker.hackndev.com/GPIO/ Oct 15 11:39:08 some can be in older format Oct 15 11:39:16 (without GPLR3) Oct 15 11:39:37 thanks Oct 15 11:39:56 I thank you :b Oct 15 11:42:14 USBH1_PEN 89 Oct 15 11:42:34 ooh Oct 15 11:45:38 Sleep_Walker: nooo, come back.. Oct 15 11:48:03 :O Oct 15 11:48:05 Sleep_Walker: AMFG Oct 15 11:48:16 MBGNT 97 Oct 15 11:48:22 ? Oct 15 11:48:31 Memory Controller Alternate Bus Master Grant—The memory controller asserts Oct 15 11:48:35 MBGNT to allow an external device to control the memory bus. Oct 15 11:48:49 GPIO 97 = MBGNT unless my parser is screwed Oct 15 11:49:08 that means some other device can take over SDRAM Oct 15 11:50:20 hmm Oct 15 11:50:20 no Oct 15 11:50:22 actually Oct 15 11:50:28 I think my parser is screwed Oct 15 11:50:31 that doesn't make snse Oct 15 11:51:34 ahh.. Oct 15 11:51:36 that looks more ane Oct 15 11:51:37 ooh Oct 15 11:51:45 it's cause you have that CS2 data in there Oct 15 11:51:50 confusing me :) Oct 15 11:51:51 :p Oct 15 11:53:20 maybe I've got something wrong on my side Oct 15 11:54:10 nah it's ok Oct 15 11:54:17 I'm looking at the non CS2 ones now Oct 15 11:54:23 it looks normal with them Oct 15 11:58:13 oooh Oct 15 11:58:15 oooooooooooh Oct 15 11:58:19 I got the address wrong Oct 15 11:58:19 :O Oct 15 11:58:22 for the AF registers Oct 15 11:58:41 GAFR3 is wrong Oct 15 11:58:45 thats why Oct 15 11:58:52 * BobOfDoom apologises to Sleep_Walker Oct 15 12:02:47 so dumps are not correct? Oct 15 12:03:56 grrr Oct 15 12:03:58 stupid netsplits Oct 15 12:04:07 ha Oct 15 12:04:11 it thinks I'm still here Oct 15 12:04:17 Sleep_Walker: yeah Oct 15 12:04:25 Sleep_Walker: AFs are wrong for extended GPIOs Oct 15 12:04:32 ok Oct 15 12:04:34 Sleep_Walker: that's why USB isn't showing up properly Oct 15 12:04:37 sorry about that Oct 15 12:05:29 np Oct 15 12:06:26 back Oct 15 12:06:42 that's cool.. it logs the chat and replays when it rejoins from netsplit Oct 15 12:06:43 handy Oct 15 12:09:15 oh Oct 15 12:09:17 you know what Oct 15 12:09:29 maybe the LED blinking in suspend and stuff is done by the GSM chip Oct 15 12:09:38 cause it's a full ARM chip and can do all that stuff Oct 15 12:10:56 it's possible Oct 15 12:11:03 but I think only red led is connected Oct 15 12:11:08 rest I can control Oct 15 12:13:32 I see Oct 15 12:16:05 Sleep_Walker: BtTransportUsb.bprc Oct 15 12:16:11 So it is over USB as well Oct 15 12:19:19 ok Oct 15 12:19:43 so there is hub? Oct 15 12:20:46 yep Oct 15 12:20:50 so I think it goes Oct 15 12:21:15 CPU USB host <=> HUB < GSM Oct 15 12:21:18 < BT Oct 15 12:21:30 nice Oct 15 12:24:58 BobOfDoom: can you tell which port it is attached to? Oct 15 12:28:22 hmmm, 9c000000 mapped to 4c000000 :) Oct 15 12:29:03 I'll create another gpiotrace version for dumping UHC registers :b Oct 15 12:37:24 Sleep_Walker: excellent, Linux drivers exist for both Oct 15 12:37:26 Sleep_Walker: http://ubuntuforums.org/showthread.php?t=937454 Oct 15 12:38:03 omg Oct 15 12:38:07 that was fast Oct 15 12:38:48 hmm, at least I think that's correct Oct 15 12:39:09 I found reference to some windows drivers for BCM2133 and they say something about samsung Oct 15 12:40:15 http://www.cellulari.it/notizie-news-cellulari/tecnologia/samsung-insieme-a-broadcom-per-accelerare-lo-sviluppo-3g-nel.html Oct 15 12:42:49 almost but not quite readable Oct 15 12:42:51 http://www.palminfocenter.com/images/treo-68-guts-5-L.jpg Oct 15 12:43:46 wait.. Oct 15 12:43:48 http://www.palminfocenter.com/ss.asp?f=treo-68-guts-4-L.jpg Oct 15 12:43:58 MSYSTEMS DoC G4 Oct 15 12:44:01 hmm Oct 15 12:49:56 Sleep_Walker: http://www.inkhorn.net/site/log/Entries/2007/12/16_Treo_680_ROM_hacking_3.html Oct 15 12:52:06 I probably have seen it... I have no clue why you post that... Oct 15 12:52:19 ah Oct 15 12:52:24 well you're much ahead of me then Oct 15 12:52:52 I know about bootloader Oct 15 12:53:06 I haven't tried it yet Oct 15 12:53:32 hdread (?, brahma-only) Oct 15 12:53:32 hdfill (?, brahma-only) Oct 15 12:53:35 that's interesting Oct 15 12:53:38 it's got LD stuff Oct 15 12:53:41 and T5 Oct 15 12:53:42 Angus Oct 15 12:54:09 what's especially interesting is the LifeDrive doesn't have a ROM like that Oct 15 12:54:19 I mean bootloader Oct 15 12:55:32 ah Oct 15 12:55:33 bingo Oct 15 12:55:34 http://www.inkhorn.net/site/log/Entries/2007/5/30_Interior_of_the_Treo_680.html Oct 15 12:55:37 that's what I was looking for Oct 15 12:59:07 nice Oct 15 13:00:25 hmm I'm confused Oct 15 13:00:29 he says 32mb RAM Oct 15 13:00:32 where's the rest hiding Oct 15 13:01:02 8270C5C312 = PXA270.. so not the PXA271 with builtin RAM Oct 15 13:01:28 I should try to boot linux and allocate it all :B Oct 15 13:16:23 UHC dumping gpiotrace ready, I hope I didn't made there some leaks because of that stupid string manipulation :/ Oct 15 13:20:16 ah Oct 15 13:20:20 found he extra RAM Oct 15 13:20:26 it's in the SEC chip Oct 15 13:20:40 SEC = Samsung Electronics Oct 15 13:20:43 http://www.chipmunk.nl/DRAM/Samsung.htm Oct 15 13:22:07 fine :) Oct 15 13:23:08 k4n51163pc Oct 15 13:23:13 32mB Oct 15 13:23:19 yep Oct 15 13:23:20 that's it Oct 15 13:23:22 okay Oct 15 13:23:29 mystery solved :) Oct 15 13:23:54 so 64Mbs! Oct 15 13:24:06 and 128 of flash Oct 15 13:24:14 with possibility to add USB wifi Oct 15 13:24:17 hmmm Oct 15 13:24:31 that was real bargain :b Oct 15 13:25:46 in fact it's got more flash Oct 15 13:25:51 it's 196mb Oct 15 13:25:59 cause the spansion chip has flash too Oct 15 13:26:00 hmm Oct 15 13:26:03 unless Oct 15 13:26:13 oh perhaps some of that RAM/flash is used by the GSM Oct 15 13:26:57 yes... Oct 15 13:27:14 BCM2133 PB says "External NAND, NOR, SRAM, SDRAM at 1.8V or 3.0V" Oct 15 13:27:52 ah Oct 15 13:28:01 and the spansion chip is right next to the BCM2133 Oct 15 13:28:21 so not mystery solved Oct 15 13:28:27 unless they're sharing RAM Oct 15 13:29:09 oh wait Oct 15 13:29:15 Samsung chip has 64mb Oct 15 13:29:19 I misread it Oct 15 13:29:23 it was talking about banks of 32mbit Oct 15 13:29:25 right Oct 15 13:29:27 so all solved now :D Oct 15 13:29:47 96 Mb of RAM, 196 Mb of Flash, dual ARM device Oct 15 13:29:53 hahaha! Oct 15 13:30:09 phone is just side effect Oct 15 13:31:29 :p Oct 15 13:31:52 dang Oct 15 13:32:02 that means all the flash is M-System DoC G4 Oct 15 13:32:04 no driver :( Oct 15 13:40:37 so, I can touch PalmOS, but I can't touch system data, right? Oct 15 13:41:17 I must have some typo somewhere - I got all that registers 0xffffffff :( Oct 15 13:43:26 yaya msystems Oct 15 13:43:34 * raster hugs msystems Oct 15 13:43:38 gotta lov 'em Oct 15 13:44:30 imagine even msystem chip with vegemite! ;) Oct 15 13:52:52 yes! got dumps :) Oct 15 13:53:51 :) Oct 15 13:54:16 can be seen here http://sleepwalker.hackndev.com/GPIO/ Oct 15 13:54:20 UHC*txt Oct 15 13:56:50 gotta go to school now Oct 15 13:56:53 gn bob :) Oct 15 13:56:57 see you today Oct 15 13:56:59 gn Oct 15 13:57:04 and I need sleep Oct 15 13:57:05 cya Oct 15 20:19:19 good morning, bob :b Oct 15 20:20:54 * marex keeps discovering new possibilities of his shiny new logitech g15 keyboard Oct 15 20:21:09 eg. I have hotkey for kick, ban, etc. :) Oct 15 20:21:48 * Sleep_Walker is discovering usb host capabilities of Treo680 :) Oct 15 20:21:56 * mwester resolves to behave then ;) Oct 15 20:23:19 mwester: keep looking straight! Oct 15 20:23:55 8) Oct 15 20:24:10 omg Oct 15 20:24:10 just perfect :) Oct 15 20:24:40 macromode3 -> G3 -> type Sleep_Walker -> press enter ... how simple :) Oct 15 20:25:16 G3? :D Oct 15 20:25:31 next row to F3? :b Oct 15 20:25:41 it's on the left side ;) Oct 15 20:30:53 Sleep_Walker, ^ a victim ... I can test the new functionality right away 8) Oct 15 20:31:37 :b Oct 15 20:33:46 marex: I've got serious problem with booting :( Oct 15 20:34:00 wanna help? :B Oct 15 20:34:03 lucky it's not my problem Oct 15 20:35:54 it's cocoboot problem Oct 15 20:37:15 works for me ;) Oct 15 20:37:26 Sleep_Walker, spit it out ... :/ Oct 15 20:41:18 :( Oct 15 20:44:43 Sleep_Walker, so ? Oct 15 20:45:28 marex: it can't recognize initrd or kernel Oct 15 20:45:36 most of tries Oct 15 20:46:23 it fails to get to framebuffer or it get to kernel panic Oct 15 20:46:51 No filesystem could mount root... Oct 15 20:46:58 so no initrd Oct 15 20:47:41 I've suspect that it has something different with memory management :( Oct 15 20:48:34 so the file is broken after it's loaded to memory ? Oct 15 20:48:57 maybe Oct 15 20:49:27 like it loads into memory and it's not there correctly when executing.. Oct 15 20:50:07 try loading it to different address ... Oct 15 20:50:26 I was thinking about kernel size, but 1.3 MB kernel+initrd is not that much Oct 15 20:50:41 on T3 I was able to boot even 5 or more megs :) Oct 15 20:54:17 that shouldnt be a problem Oct 15 20:54:28 agreed Oct 15 20:54:39 try loading it to different address Oct 15 20:54:42 how? Oct 15 20:54:56 load it twice and use the second load? :b Oct 15 20:55:01 (well palmtt had issues with sice, it could load only ~800k kernel, but that's different) Oct 15 20:55:09 Sleep_Walker, just change the load offset Oct 15 20:55:16 arm/somewhere.c Oct 15 20:55:28 it's 0x8000 for all != T3 Oct 15 20:55:35 it's 0x800000 for T3 Oct 15 20:55:45 if noone changed it which someone iirc did Oct 15 20:57:04 I want that for Treo680 - that has that problem Oct 15 21:00:40 it's not that hard to add ... Oct 15 21:27:26 wtf?! Oct 15 21:32:16 Hi, all Oct 15 21:32:18 http://lists.arm.linux.org.uk/lurker/message/20081015.182538.23a70d17.en.html Oct 15 21:32:33 Did anybody looked at it ? Oct 15 21:32:57 no, why ? Oct 15 21:33:21 marex: hm, it seems that it was good idea :) Oct 15 21:33:37 Sleep_Walker, yea ... that happens Oct 15 21:35:08 marex, I wonder if this can make power management better Oct 15 21:35:23 that's PWM ... it has nothing to do with PM Oct 15 21:36:27 ma bad Oct 15 21:36:37 old eyes Oct 15 21:52:58 marex: false alarm, same bad Oct 15 22:23:10 this really sucks! Oct 15 22:23:32 I've got prepared kernel for two hours but I simply can't boot :E Oct 15 22:24:19 revolutional Treo680 kernel with OHCI support which should prove that GSM and BT is connected through USB ;E Oct 15 22:34:10 BT might be connected through usb, yes Oct 15 22:34:26 due to PalmOS filenames - it seems so Oct 15 22:34:34 it is done that way on hw6515d, but there is csr bluecore3 or something Oct 15 22:34:52 bluecore3? Oct 15 22:35:03 I think there is Blutonium BCM2045 :) Oct 15 22:35:08 codename of that chip Oct 15 22:35:19 in 6515s Oct 15 22:35:30 well that can be connected over usb too Oct 15 22:38:08 there are probably already done drivers Oct 15 22:38:42 but I have to boot first :b Oct 15 22:40:37 there are generic drivers for most of those usb dongles Oct 15 22:40:51 s/are/is/;s/s// Oct 15 22:42:52 yes I know Oct 15 22:43:22 BobOfDoom: are you alive already? Oct 15 23:19:46 finally booting! Oct 15 23:21:57 Sleep_Walker, what was the fix ? Oct 15 23:23:17 no fix Oct 15 23:23:19 random boot Oct 15 23:23:24 it sometimes boots Oct 15 23:32:07 Sleep_Walker, try dumping the ram after loading kernel and see if it's corrupted Oct 15 23:32:33 hmmm Oct 15 23:53:41 hmm, I wonder how much there is inside treo 680 Oct 15 23:53:55 could rig a USB memory stick up to the USB host :D Oct 15 23:54:19 then you can have tons of (somewhat slow) storage Oct 16 00:12:14 12 MB/s Oct 16 00:12:21 (theoretically) Oct 16 00:12:29 is your SD faster? :b Oct 16 00:13:08 BobOfDoom: that cocoboot problem is awfull Oct 16 00:13:25 I spent 3 hours fighting with that to make it booting and it resisted... Oct 16 00:14:39 interesting is that if I make it booting once, it well probably boot next time when running cocoboot from RAM Oct 16 00:15:09 btw. gm Oct 16 00:17:53 12 Mbits/sec Oct 16 00:17:58 MB = MegaBytes / sec Oct 16 00:19:42 ok Oct 16 00:22:46 last boot before sleep Oct 16 00:28:44 hmm Oct 16 00:28:46 gm Oct 16 00:29:01 and gn :b **** ENDING LOGGING AT Thu Oct 16 02:59:58 2008