IRC log for #htc-linux on 20090531

00:01.12cr2tmzt: rpc parser ;) od -A x -t x4 -w16 -j $((0x10198)) -N $((0x2048)) 1f
00:03.05cr2now i need to understand the message size
00:04.14tmztrpc parser?
00:04.53tmztxsacha: yeah, system.img is a cramfs then
00:05.08xsachaso i make a new partition?
00:05.09cr2tmzt: not a real parser.
00:05.11tmztbut you can also put it on the sd partition, it doesn't matter
00:05.22xsachaoh ok
00:05.28tmztthe pxa3xx and the msm have different arm cores
00:05.28xsachaand i extract data there too?
00:05.37tmztyou can't use code for the new one on the old one
00:05.43cr2tmzt: we may need to make some more strange things, like suspending while playing sound.
00:06.14tmztare we not able to isolate the audmgr calls?
00:10.02xsachawell i have system.img on my SD.. are you sure i wont need data.img?
00:10.14tmztjust mkdir data
00:10.27xsachabecause data has a weird issue where it gives Input/Output Error for some directories when i extract it
00:11.05xsachaso will it be able to boot without an initrd?
00:11.27tmztyou need the android init and some other stuff from the initrd
00:11.32tmztbut you don't need the initrd
00:11.41xsachaandroid init?
00:11.50tmztit's easiest to estract that initrd to the root of the card
00:12.02cr2tmzt: some magic call 0feb8f4a
00:12.28tmzttry calls again?
00:14.26cr2no
00:14.31cr2see this one in my logs
00:14.34cr2TIME_REMOTE_ATOM  0x30000048   time_*
00:14.58cr2although there is a dex call
00:15.02tmztah
00:15.17cr2ATOM means setting time
00:15.29tmztit sets it from the network
00:15.42tmztwhich direction is that?
00:15.53cr2time_tod_request_apps_bases
00:15.59cr211TO9
00:17.39cr2NetRipper: not sleeping ?
00:21.16cr2lol
00:21.25cr2pm_set_led_intensity
00:25.38xsachai see reason for initrd.. it sets up stuff (permissions) and nodes like bluetooth
00:31.07cr2tmzt: can we get a smem dump from g1 ?
00:33.40*** join/#htc-linux WyrM (n=wyrm@20158182083.user.veloxzone.com.br)
00:39.45tmztWyrM: just started gen-blob in qemu, it worked
00:40.07WyrMnice :)
00:40.14tmztbut zImage (for motoq) doesn't
00:40.28tmztyeah, I'm trying to test my bootloader
00:40.44tmztI just realized I could copy the pxafb code from genblob
00:41.15tmzthuh, it's probably waiting on spi isn't it
00:41.24tmzt(kernel)
00:41.40WyrMprobably.
00:41.41tmztI really just wanted to see if kernel would boot with the correct mtype
00:42.01tmztit looks like it is
00:42.10WyrMbut you got no messages?
00:42.18tmztfrom kernel, no
00:42.19tmztnothing
00:42.35WyrMIIRC pcap code is is only probing after pxafb.
00:43.09WyrMtry to disable it.
00:43.09tmztthat's what I thought
00:43.18tmztdisable?
00:43.23WyrMit may be some issue with memory too.
00:43.30WyrMyep, remove from .config
00:43.58tmztsince I'm here (on internet), can you tell me how to pull kernel again
00:44.27WyrMgit pull git@git.openezx.org:openezx.git
00:44.35WyrMif you want push access.
00:44.42tmztI have it
00:44.57WyrMyou also need to setup topgit.
00:44.59tmztI don't have the key on this machine though
00:45.07WyrMbut I don't remember how to do it ;)
00:45.21WyrMIIRC its just tg remove --populate origin.
00:45.31WyrMs/remove/remote/
00:45.38tmztyeah
00:45.53tmztyou ported motoq to ezx/mach/motoq?
00:53.55WyrMno.
00:54.03WyrMjust a skeleton.
00:54.14tmztah, okay
00:54.22WyrMim waiting for you to fix the basics.
00:54.29tmztonce I have it checked out I will be able to look at it
00:54.48tmztand with the stuff we did with haret today, I think I know how to add the mfp dumper
00:55.10WyrMgood. :)
00:55.40tmztharetconsole has definitions called regTwoBits() for af
00:55.57tmztso I just parse those, with my own python code
00:56.18tmztwhat I don't have is something to dump the registers directly
00:56.44tmztthe old one is in haret itself, but I think this will just use a binary or pd dump
00:56.59tmztthat's what I can't figure out
00:57.18tmztis how to initiate it from haret but do the processing with haretconsole
01:46.47parjust compiling the kernel from scratch.... its still the same tree in cvs at handhelds.org right
01:55.49tmztyeah
01:56.32tmztI guess you could try pH5's new tree, but I don't think it has ba support yet
01:59.00tmztit doesn't seem to be usable yet
01:59.05tmztor he hasn't pushed it
02:03.03*** join/#htc-linux WyrM1 (n=wyrm@20158167038.user.veloxzone.com.br)
02:20.35paroh ok
02:29.14parwheres the repo for it
02:38.03parwhat determines which partition on the physical sd card is 1 or 2?
02:39.23parlike, is the second always gonna be /dev/mmcblk0p2?
02:50.42tmztyeah
02:50.56tmztrepo is on git.linuxtogo.org
03:02.28*** join/#htc-linux BHSPitLappy (n=BHSPitLa@unaffiliated/bhspitmonkey)
03:09.02*** join/#htc-linux mrmoku|a` (n=mrmoku@ppp-93-104-34-248.dynamic.mnet-online.de)
03:33.19tmztWyrM: kernel booted to panic on qemu with your simplified motoq.c
03:33.36tmztanywhere I can put qemu source?
03:33.36WyrMgood. :)
03:33.50WyrMpanic was expected (rootfs?) ;)
03:33.51tmztwe can add pcap to it, at least a dummy one
03:33.53tmztyeah
03:34.02tmztleft out mci
03:34.03WyrMsure, put on svn.openezx
03:34.11tmztit's a git now
03:34.22tmztI can just put the patch though
03:34.24WyrMyour call.
03:34.28WyrMwe have git and svn :)
03:34.38tmztI mean qemu uses git now
03:35.12WyrMyou plan sending your changes upstream?
03:35.24tmztmaybe, if it's really supported
03:35.34tmztI was going to do this before but there was no matrix keyboard
03:35.42tmztand I couldn't figure out how to write one
03:35.46tmztnow there is
03:38.37tmztand it boots with my bootloader
03:38.48tmztlast things to try is to break the machtype in qemu
03:39.02tmztit picked up cmdline though
03:39.10tmztmmcblk0p2
03:39.29tmztmaybe
03:40.22WyrMthe builtin cmdline is this one.
03:40.33tmztyeah, but it's not in .config
03:40.35tmztfixup?
03:40.43WyrMhave you used anything before the kernel?
03:40.50tmztthis time
03:41.03tmztbut qemu has it's own bootloader, which I'm using to load mine
03:41.15tmztI could make it just load my bootloader to start of ram
03:41.32tmztor I could just change my bootloader to use a different command line
03:41.39tmztI need rootdelay=3 anyway
03:42.53WyrMplease upload the code.
03:42.54tmztwithout bootloader (-kernel zImage) it gives: Please append a correct root=
03:42.57WyrMI want to test here too ;)
03:42.58tmztwhich?
03:43.04tmztokay
03:43.35tmztget the latest qemu, you'll have to get it from softsource or something as the main download is down
03:43.53tmztwell, I should get the git version shouldn't I
03:44.37tmztand with bootloader, I get waiting 1sec, that should be 3
03:44.55tmztdoesn't seem the ATAGs are working
03:46.00tmztI'm going to have to upload this a little later
03:46.10WyrMtmzt: ok. :)
03:46.25AstainHellbringsup tmzt
04:03.12partmzt: so for angstrom from the git there at linuxtogo.com ... i need to hav oe installed
04:03.12par?
04:08.28parn/m
04:49.36*** join/#htc-linux droid001 (n=mc@p4FDCE7E2.dip.t-dialin.net)
05:16.40*** join/#htc-linux dzo (n=dzo@121-98-128-127.bitstream.orcon.net.nz)
05:22.16*** join/#htc-linux infidel206 (n=infidel2@unaffiliated/jenkempusher/x-35920)
05:40.05*** join/#htc-linux dzo (n=dzo@121-98-128-127.bitstream.orcon.net.nz)
05:41.45*** join/#htc-linux goxboxlive (n=goxboxli@237.80-202-137.nextgentel.com)
06:00.58*** join/#htc-linux dzo (n=dzo@121-98-128-127.bitstream.orcon.net.nz)
06:06.44xsachahey tmzt im back
06:09.38xsachaok ive got an idea... the initrd with kaiser has stuff i dont want (enabling gsm radio? partitioning?) so im gonna edit it and then repackage it and try booting the initrd in to my SD
06:09.55xsachai noticed it has about 5 different init files
06:18.29*** join/#htc-linux Echo31 (n=olivier@mir31-4-82-240-194-54.fbx.proxad.net)
06:28.38*** join/#htc-linux WyrM1 (n=wyrm@20158134193.user.veloxzone.com.br)
06:33.54*** join/#htc-linux Zoolooc (n=fredsiba@nrbg-4dbf8079.pool.einsundeins.de)
06:36.14xsachahmm im trying to extract that initrd with 7z, like we did with that last one.. but it says the only file in it is testkey.x509.pem?
06:40.07xsachaok this time that cat method worked but it spams a few "Cannot mknod"
06:47.56parbah, i'm giving up on oe under knoppix
06:56.08xsachainit1 is a binary? android boot?
07:02.38xsachaso is init1 the thing i need compiled as armv5?
07:23.38xsachaA N D R O I D Kernel panic
07:25.56tmztxsacha: it all has to be armv5, your cpu can only run armv5 I think, maybe pxa310 supports armv6
07:26.10xsachanaah, it is not compatible with armv6
07:26.16tmztit has to run as root
07:26.26tmztthat's why you get the mknod error
07:26.32xsachai was running as sudo
07:27.25tmztdzo: how does the qdsp4 code work, will it work in htc-msm-2.6.27 tree? cr2 says we need that for audmgr
07:29.58xsachathe SDK files.. are they compiled for ARMv5 or for X86?
07:30.51tmztwhich?
07:31.01tmztoh, those are armv5
07:31.05xsachaso i can use that?
07:31.09tmztthe ones for emulator (qemu)
07:31.10tmztmaybe
07:31.18tmztsince you don't have the 3d interface, probably
07:31.32xsachaso i can get cupcake too? :)
07:31.32tmztbut you are going to need android patches for your kernel
07:31.45xsachaoh..? what do you mean? :\
07:32.36tmztI haven't really tried to get android working on anything but raph
07:32.41*** join/#htc-linux nebi_ (n=nebi@217.142.147.19)
07:33.01tmztbut I think you need the asmem, binderipc, double framebuffer (maybe not anymore but it's still faster with it)
07:33.12xsachadamn android 1.5 is twice the size of android 1.1
07:33.21tmztit would be easier to search for pxafb android
07:33.28xsachawill it run without that stuff?
07:33.29tmztbut that's 1.1 or earlier code
07:33.35tmztas far as the animiation
07:34.05tmztI haven't done anything with android now, I'm working on efl/shr for raph
07:34.10xsachak
07:37.08xsachawell the sdk seems to come with an armv5 initrd, data and system image
07:38.11tmztokay, try that
07:38.17tmztsee how far it gets
07:41.38xsachamy main thinking is that screen, keyboard and SD work so it should be able to boot. if it's booting, it'll give a lot of encouragement for other people to help out getting all the drivers working
07:42.04tmztok, but you will probably only get as a far as a flashing android
07:42.14xsachathats ok :)
07:42.25tmztthere is a pxafb patch out there somewhere, but whether it works on pxa310 I don't know
07:43.55xsachawhats it meant to do? something to do with video?
07:45.31tmztmakes android work
07:45.45tmztpxafb is the framebuffer, the part of ram where the image on the lcd is stored
07:45.53*** join/#htc-linux cr2 (n=cr2@ip-77-25-89-159.web.vodafone.de)
07:46.26tmztandroid wants the framebuffer to have twice the verticle resolution it really does so it can flip screens quickly
07:46.37tmztif you don't have that it will complain and be slower
07:46.38xsachaahh ok
07:46.53tmztthe one you really need is binderrpc, android won't work without that
07:47.06xsachak
07:47.09cr2hi
07:47.26tmztcr2: hi, I have my bootloader booting motoq kernel in qemu
07:47.45cr2cpu?
07:47.50tmztpxa
07:47.58cr225x ?
07:48.03tmztdoesn't really matter, I'm just testing the ability to boot linux
07:48.05tmztpxa27x
07:48.08cr2ok
07:48.19xsachais that an armv5 processor?
07:48.29tmztnow I'm trying to verify the atags are correct
07:48.30tmztyes
07:48.35cr2im looking at the rpc msg layout
07:48.44xsachavery similar to mine
07:48.50cr263 struct phonemgr_msg {
07:48.51cr264    struct rpc_request_hdr hdr;
07:48.53cr265    int args[5];
07:48.54cr266 };
07:48.59cr2this is from vogue-hw.c
07:49.06tmztyeah
07:49.35cr2i'd like to compare it with the rpccall smem log
07:49.39tmztI didn't copy that
07:50.59cr2tmzt: it is sent like that
07:51.04cr2129         msg.args[0]=cpu_to_be32(device);
07:51.05cr2130         msg.args[1]=cpu_to_be32(earmute);
07:51.07cr2131         msg.args[2]=cpu_to_be32(micmute);
07:51.09cr2132         msg.args[3]=cpu_to_be32(3);
07:51.10cr2133         msg.args[4]=0;
07:51.11cr2134         msm_rpc_call(SND_PROG, 1, &msg, sizeof(msg),15*HZ);
07:51.40*** join/#htc-linux kiozen (n=oeichler@rgnb-5d87468d.pool.einsundeins.de)
07:52.00tmztand we know what device is?
07:52.08tmztwhat are earmute and micmute?
07:54.02cr2too much to copy.
07:54.06cr2look at
07:54.08cr2340 /** Kaiser, Polaris, Nike **/
07:54.31cr2snd_set_device(2,1,1);} // audio: headset
07:54.40cr2snd_set_device(1,1,1);} // audio: speaker
07:54.49cr2snd_set_device(0,0,0);} // earcouple
08:01.37tmztSNDPROG?
08:02.39cr2yes. 30000002 + $proc + msg_data
08:02.54tmzt?
08:03.01tmzt$proc ?
08:03.20cr2SND  0x30000002   snd_*
08:03.21cr21 snd_set_device_modem_reset
08:03.23cr22 snd_set_volume_modem_reset
08:03.29cr2<PROTECTED>
08:03.36tmztok
08:03.59cr2i'll edit the wiki
08:05.02cr2prog  proc,msg
08:08.49tmztand msg_data, is that part of the define as well?
08:09.39cr2you send header+pacmark+msg (as i understand the code)
08:09.44cr2header is
08:09.48cr240 struct rpc_request_hdr
08:09.49cr241 {
08:09.52tmztah, yeah
08:09.55cr242         uint32_t xid;
08:09.55cr243         uint32_t type;  /* 0 */
08:09.55cr244         uint32_t rpc_vers; /* 2 */
08:09.55cr245         uint32_t prog;
08:09.57cr246         uint32_t vers;
08:09.58cr247         uint32_t procedure;
08:10.00cr248         uint32_t cred_flavor;
08:10.01cr249         uint32_t cred_length;
08:10.03cr250         uint32_t verf_flavor;
08:10.20cr251         uint32_t verf_length;
08:10.20cr252 };
08:10.20cr2pacmark is murky
08:10.20cr2the msg is in vogue-hw.c
08:10.27cr2see above
08:10.41tmztI'm mean what is SND_PROG defined as?
08:10.55cr2i'd like to compare it with the smem rpc dump
08:11.06cr237  * XXX: Any cred/verif lengths > 0 not supported
08:11.06tmzt0x30000002
08:11.09tmztstill have it?
08:11.10cr2yes
08:11.23cr2yes
08:11.32cr2i have a lot of my own too.
08:11.42tmztthey are the same?
08:11.52cr2and from maejrep
08:12.03cr2it seems so
08:12.15cr2for the rpc part
08:12.40cr2it'd be nice to get the tita|kaiser version
08:12.50cr2dcordes-kais: ping
08:14.29cr2wince uses 0x48 bytes struct for rpc calls
08:15.07cr20x48/4=0x12=18 uint32_t
08:15.26cr2header has 10
08:15.34cr2then pacmark
08:15.38cr2and msg
08:15.51tmztdzo's code doesn't seem to use hdr
08:16.23cr2<PROTECTED>
08:16.38cr2hdr= prog+vers+proc
08:16.54cr2it's filled from these 3 numbers
08:16.55nathan7tmzt: Earmute==headphone mute
08:17.11cr2for our amss the vers=0
08:17.13nathan7tmzt: Micmute == microphone mute
08:17.15tmztbut is it bool?
08:17.21nathan7Also, yay.
08:17.26nathan7My magician image built =D
08:17.31cr2so it comes only to the prog+proc (documented in wiki)
08:17.47cr2and the sizeof(msg) + msg itself
08:18.00cr2which needs some more research
08:18.23tmztrpc_request_hdr seems undefined
08:18.50tmzt10 bytes, padded?
08:19.46*** join/#htc-linux xsacha2 (n=sacha@123.100.137.226)
08:20.01cr2[10:09] <cr2> 40 struct rpc_request_hdr
08:20.03cr2[10:09] <cr2> 41 {
08:20.04cr2[10:09] <cr2> 42         uint32_t xid;
08:20.06cr2[10:09] <cr2> 43         uint32_t type;  /* 0 */
08:20.07cr2[10:09] <cr2> 44         uint32_t rpc_vers; /* 2 */
08:20.09cr2[10:09] <cr2> 45         uint32_t prog;
08:20.10cr2[10:09] <cr2> 46         uint32_t vers;
08:20.12cr2[10:09] <cr2> 47         uint32_t procedure;
08:20.15cr2[10:09] <cr2> 48         uint32_t cred_flavor;
08:20.15cr2[10:09] <cr2> 49         uint32_t cred_length;
08:20.16cr2[10:09] <cr2> 50         uint32_t verf_flavor;
08:20.18cr2[10:09] <cr2> 51         uint32_t verf_length;
08:20.20cr2[10:09] <cr2> 52 };
08:20.24xsacha2i notice in the sdk there are extra files like adb, adata, etc.. these arent needed right?
08:20.44tmztnot sure about adata
08:20.52tmztadb is used on host to talk to device
08:20.59xsacha2cause in the kaiser initrd they have included adb in sbin folder
08:21.26tmztyeah, you need it there too
08:21.30tmztit's also the server
08:21.33nathan7ADB is like gdb
08:21.40tmztnot exactly
08:21.44nathan7Hm
08:21.47xsacha2i need it? aw man
08:21.51nathan7No.
08:21.56tmztthere's something else that does that
08:21.56xsacha2i hope the one in the sdk isnt x86
08:22.33cr2xid=seq,type=0,rpc_vers=2,prog,vers=0,proc,0,0,0,0,*pac,*msg
08:24.10xsacha2i found out how to extract those cpio files tmzt
08:24.23xsacha2cpio -t -F filename works
08:24.30tmztcr2: do I need to initiaize rpc_vers?
08:24.38tmztah, in the struct definition?
08:24.42tmztdoes that work?
08:24.45cr224000000 00000000 02000000 02000030 00000000 02000000 00000000 00000000 00000000 00000000 0d000000 01000000 05000000 ffffffff 00000000 00000001
08:25.18cr225000000 00000000 02000000 0a000030
08:26.06tmztnow I don't have msm_rpc_call
08:27.08cr224 ints between 2 SND full msgs
08:27.32cr2tmzt: the vogue rpc is a simpliefied version of the full rpc
08:27.46cr2we should ask dzo if it should be this way
08:29.13xsacha2what is a VMS Alpha executable? :\
08:29.21xsacha2that's what it tells me my system.img is :(
08:29.22tmztbroken file
08:29.23nathan7lol
08:29.29tmztit's a ext2
08:29.32tmztor
08:29.33nathan7It's for an ancient OS
08:29.34tmztyaffs2
08:29.37nathan7=0
08:29.42xsacha2it's the one from the SDK
08:30.09tmztif you use the generic android from android.git.kernel.org
08:30.11tmztkernel
08:30.25tmztand copy you machine, repatch Makefile and Kconfig
08:30.29tmztit should be easier
08:30.45xsacha2oh ok
08:31.10xsacha2could be my computer doesnt know what yaffs2 is
08:31.21tmztyeah
08:31.50xsacha2what do i install to be able to mount yaffs2?
08:31.59tmztnothing, enable it in .config
08:32.09tmztchange the initrd script to mount it as loop
08:32.13xsacha2i mean my desktop computer
08:32.20xsacha2i cant mount it on here
08:32.31tmztno
08:32.37tmztnot unless you patch the kernel
08:32.42xsacha2:(
08:32.51tmztthere is supposed to be a yaffs2 extract program somewhere
08:33.48tmztunyaffs2 or something
08:34.02xsacha2yep :) thx
08:34.22xsacha2http://code.google.com/p/unyaffs/
08:34.41tmztcr2: ah, it seems the rpc code is only in qdsp5 on htc-msm
08:35.34tmztcb_func
08:35.37tmztcallback_data
08:35.50tmztthe last two args?
08:36.34tmztthe first param to msm_rpc_call is ept, which is a pointer
08:36.50tmztrc = msm_rpc_call(snd->ept,
08:36.50tmztSND_SET_DEVICE_PROC,
08:36.50tmzt&dmsg, sizeof(dmsg), 5 * HZ);
08:37.51xsacha2unyaffs worked :D woot
08:38.35xsacha2sacha@X /media/ANDROID $ ls         app  build.prop  default.prop  etc    framework  init.goldfish.rc  lib    proc  sounds  system  xbin    bin  data        dev           fonts  init       init.rc           media  sbin  sys     usr
08:38.43xsacha2look right?
08:38.48tmztyeah
08:38.53tmztmaybe
08:39.04tmztare you going to extract everything?
08:39.12xsacha2yeah
08:39.36tmztyou are probably going to have to edit init.rc in that case
08:39.45tmztsee if there are any mounts in it?
08:39.51xsacha2oh, the init thinks it's on different partitions?
08:39.59tmztyes
08:40.34xsacha2init.rc:    mount yaffs2 mtd@system /system
08:40.45tmztcomment those out
08:40.57xsacha2but i didnt extract system to /system
08:41.02tmztdo that
08:41.07xsacha2o crap :(
08:41.25xsacha2so ramdisk in /, data in /data and system in /system?
08:41.27tmztand make symlinks to the bin, sbin, xbin, etc.
08:41.30tmztyeah
08:41.46xsacha2what is this? init.rc:    socket mountd stream 0660 root mount
08:41.58tmzteverything but default.prop init* media dev should be in system I think
08:42.01tmztand proc, sys
08:42.19tmztthe mountd daemon
08:42.24tmztdon't worry about it
08:43.40xsacha2well theres about 6 mount lines and : init.rc:    # We chown/chmod /data again so because mount is run as root + defaults
08:43.58tmztthat should be fine
08:44.04tmztand lines starting with mount?
08:44.08tmztor with mtd in them
08:44.26xsacha2init.rc:    mount tmpfs tmpfs /sqlite_stmt_journals size=4m                       init.rc:    mount rootfs rootfs / ro remount
08:44.40xsacha2init.rc:    mount yaffs2 mtd@system /system
08:44.41tmztthose are ok
08:44.41xsacha2init.rc:    mount yaffs2 mtd@system /system ro remount
08:44.47tmztnot the last two
08:45.04xsacha2init.rc:    mount yaffs2 mtd@userdata /data nosuid nodev                           init.rc:    mount yaffs2 mtd@cache /cache nosuid nodev
08:45.13xsacha2what do i do about cache?
08:45.18tmztnothing with mtd
08:45.24tmztmake a cache directory
08:47.22tmztcr2: snd->ept = msm_rpc_connect(RPC_SND_PROG, RPC_SND_VERS, MSM_RPC_UNINTERRUPTIBLE);
08:50.31xsacha2sacha@X /media/ANDROID $ ls             cache  data  default.prop  dev  init  init.goldfish.rc  init.rc  proc  sbin  sys  system
08:50.44tmztobj-$(CONFIG_MSM_ADSP) += qdsp5/
08:50.56cr239 static inline uint32_t PACMARK(uint32_t len, uint32_t mid, uint32_t last)
08:50.57cr240 {
08:50.57cr241         return (len & 0xFFFF) | ((mid & 0xFF) << 16) | ((!!last) << 31);
08:50.57cr242 }
08:51.23tmztwhat is this?
08:51.25xsacha2oh some more lines i have to get rid of.. it does mkdir /system n stuff
08:51.42tmztRPC_SND_VERS? it's based on amss define
08:52.51cr2sizeof(pacmark)=4 then
08:53.07xsacha2does it matter that i didnt make the directory with the same parameters? eg. mkdir /cache 0770 system cache
08:53.07cr2<PROTECTED>
08:53.25tmzt#if CONFIG_MSM_AMSS_VERSION == 6210
08:53.26tmzt#define RPC_SND_VERS                    0x94756085 /* 2490720389 */
08:53.26tmzt#elif (CONFIG_MSM_AMSS_VERSION == 6220) || (CONFIG_MSM_AMSS_VERSION == 6225)
08:53.26tmzt#define RPC_SND_VERS                    0xaa2b1a44 /* 2854951492 */
08:54.01tmztxsacha2: cache needs to me 0770, chmod -R 0770 cache
08:54.29tmztchown -R 1000 cache
08:54.35xsacha2and system directory wanted 0771?
08:54.46cr2tmzt: we have other amss.
08:54.59cr2398         smd_write(smd_channel, &pacmark, sizeof(pacmark));
08:55.00cr2399         smd_write(smd_channel, buffer, count);
08:55.20tmztwhere?
08:55.31tmztqdsp5 is wrong also?
08:55.54cr2no
08:56.05cr2check the vogue source.
08:56.33cr2dzo renamed his hacked qdsp5 to qdsp4 recently
08:57.28dzodid i hear my name?, hi there.
08:57.32xsacha2do these matter?     chown system system /data    chmod 0771 /data
08:57.56tmztthey do, but we need the uid from android
08:58.11xsacha2so leave it in init?
08:58.41cr2hi dzo
08:58.56tmztoh yeah
08:59.00tmztthat should work
08:59.21cr2dzo: what is the advantage of using vogue_rpc.c over the g1 version ?
08:59.25tmztdzo: how can we get rpc and qdsp5 for raph amss's?
08:59.30tmztand that
08:59.34tmztbut that's just for ril?
09:00.02tmztsorry, not vogue-hw versus snd.c
09:00.03dzothe vogue_rpc is for amss 4, it's done differently on later amss.
09:00.55cr2dzo: ok, so it's what kaiser/vogue uses ?
09:01.03dzoon amss 4 there are 4 rpc channels, 2 for te client, 2 for the server. on amss 6 they are combined into 2 channels with another header, thats what rpcrouter does,
09:01.45dzoto get pcm audio you will have to find the adsp queue numbers.
09:01.53cr2ok, i have 1 rx and 1 tx channel
09:02.09*** join/#htc-linux pH5 (n=ph5@e178248099.adsl.alicedsl.de)
09:02.15tmztso it's like an encapsulation?
09:02.23cr2hmm. sounds cryptic.
09:02.53xsacha2on property:ro.kernel.qemu=1    start adbd
09:03.31dzoan rpc asks the a9 to set up the adsp and it replies with an image number, for each image there are a set of queues to do variouse things, for audio there are 3 queues, and the numbers must be correct and the image changes depending on what has happened before. its very overcomplicated.
09:04.16xsacha2sweet everything looks good
09:04.18dzotmzt: yes, thats how the g1 works but i don't know about raph, how many rpc channels do you see.
09:04.38tmzthow do I find that out?
09:04.53tmzt[    1.043922] RPC rpcrouter_init
09:04.54tmzt[    1.049903] smd_alloc_channel() 'SMD_RPCCALL' cid=2, shared=e0110148
09:04.54tmzt[    1.051124] RPC_dev init complete!
09:04.54tmzt[    1.051215] RPC Init done
09:04.56dzocr2: ah ok, then you need to use the rpcrouter.
09:04.56tmzt[    1.051246] RPCCALL opened
09:05.30cr2dzo: ok, but then the msm_rpc_call api is different
09:05.34tmztthis?
09:05.35tmztsnd->ept = msm_rpc_connect(RPC_SND_PROG, RPC_SND_VERS, MSM_RPC_UNINTERRUPTIBLE);
09:05.44xsacha2it really does want init=/init
09:06.08dzoyes amss 4 is stateless so i got rid of the ept parameter.
09:06.30dzook got to go now, have fun....
09:06.35cr2:)
09:06.38cr2thanks
09:07.51tmztxsacha2: yeah
09:08.40xsacha2set cmdline "root=/dev/mmcblk0p2 rw init=/init rootdelay=5"   it's so small now
09:09.02tmztyeah
09:09.52xsacha2kernel panic :( unable to open a console
09:10.14tmztdid you chmod +x init?
09:10.15tmztoh
09:10.31tmztmknod dev/console c 5 1
09:10.54xsacha2oh do i need a dev directory?
09:11.42xsacha2is there other stuff i should put in dev other than console?
09:11.51Echo31Hello pH5
09:12.09tmztmmcblk0p* would be good
09:12.13tmztnull, zero, full
09:12.35tmztnull 1 3
09:12.42tmztzero 1 5
09:12.47tmztfull 1 7
09:12.48xsacha2huh?
09:12.55Echo31pH5: I want write a kernel 2.6.30 for athena
09:13.15tmztmmcblk0 b 179 1
09:13.21tmztmmcblk0p2 b 179 2
09:13.23Echo31pH5: I foresee to merge the code from htcathena (2.6.21) with the magician(2.6.30).
09:13.28tmztmmcblk0p1 b 179 1
09:13.31tmztmmcblk0p2 b 179 2
09:13.36tmztthat should enough
09:13.49xsacha2just those with console? ok
09:14.11Echo31pH5: But .21 and .30 have different htc-egpio api
09:14.15tmztEcho31: it's probably eaiser to rewrite athena based on magician than to try to directly port hh20
09:14.32tmztoh, i see
09:14.44*** join/#htc-linux pleemans (n=toi@d54C2AAB7.access.telenet.be)
09:14.46tmztxsacha2: each of those is a mknod command
09:15.12cr2rpc_clnt_lookup2(prog,vers,-1,&buf)
09:15.18xsacha2Freeing init memory: 124K Kernel panic - not syncing: Attempted to kill init!
09:15.25pH5good morning
09:15.39tmztanything before that?
09:15.56xsacha2no, it's all perfect before that... not 1 error
09:15.56pH5Echo31: yes, look at magician and also hx4700 for examples how to setup the 'new' htc-egpio with and without IRQ (I think athena has both variants).
09:16.47Echo31pH5: Can you help me to fill the correct values of htc_egpio_chip and htc_egpio_platform_data for htc athena?
09:18.31xsacha2do i need input in /dev?
09:18.46Echo31pH5: I use your code from magician
09:19.29tmztxsacha2: you shouldn't
09:19.42pH5Echo31: sure, which parts are not clear?
09:19.47xsacha2cause in the kaiser initrc all they have in block and input inside dev/
09:20.08tmztyou can add input if you want
09:20.17tmztbut /init should have that
09:20.35xsacha2i dunno.. /init is a binary module
09:21.16xsacha2yeah it has /dev/input/
09:24.10cr2tmzt: routing header before the pacmark then...
09:24.10Echo31pH5: i put the beginning of this code http://fr.pastebin.ca/1442021
09:24.45*** join/#htc-linux nathan__ (i=frozenfi@sine.cluenet.org)
09:25.36cr2Echo31: the problem is that the CPLD2 base address depends on the gpio bit from the CPLD1
09:26.22cr2Echo31: maybe it does not happen in real life, but who knows.
09:26.44pH5Echo31: cpld2 irq_num and ack_register should be 0, that's correct.
09:26.47xsacha2woah is it normal for all the google apps in system/bin to be symlinked to 'toolbox'?
09:27.03pH5cr2: then cpld2 will have to be created dynamically after filling in the iomem resource from machine_init.
09:27.14tmztxsacha2: it's like busybox
09:27.15cr2Echo31: this HACK ALERT value is for my ameo. ymmv.
09:27.29pH5if all known production models have the same boardid, maybe you can live with one fixed base address for now.
09:27.42xsacha2i dunno why it kernel panics :'(
09:27.50cr2pH5: it#s possible to temporarily ioremap the CPLD1 in machine_init ?
09:28.12pH5Echo31: cpld1's irq_ack register should be 4, I think?
09:28.26nathan7My Magician doesn't boot D=
09:28.29pH5cr2: yes, I do that on magician to read boardid.
09:28.36nathan7Something with S99finish
09:28.57cr2pH5: ok
09:29.01*** join/#htc-linux skodde (n=skodde@unaffiliated/skodde)
09:29.08tmztnathan7: look at it, if you want chmod -x it
09:29.15tmztnathan7: how old is your image?
09:29.18Echo31I took the value .start from athena code of 2.6.21
09:29.30pH5Echo31: and num_irqs: sounds like there might be 8 used bits in that register.
09:29.57cr2Echo31: i've written it for my t-mob ameo.
09:30.34tmztsame as athena?
09:31.14cr2tmzt: ameo is the first athena.
09:31.48pH5Echo31,cr2: registers A,B,C,D are output, E is the irqack, F,G are input and H is output on CPLD1?
09:31.50cr2tmzt: then there were X7500 and X7501 and something with 16GB flash hdd
09:32.00cr2pH5: http://wiki.xda-developers.com/index.php?pagename=AthenaCPLD1
09:32.13cr25 irqs
09:32.47pH5cr2: yes, but they are not packed. so I'd live with two wasted irqs and set irq_num to 7.
09:33.04cr2F and G look like input
09:33.06pH5wasn't sure if those +irqack entries are just not identified yet.
09:33.17xsacha2hey is /bin meant to be symlinked?
09:33.54cr2pH5: ok. the H is uncertain. need to look at the idb again, sometime.
09:34.20cr2pH5: what would you like to know about BA asic3_spi ?
09:34.33pH5Echo31: then you need to define 3 egpio_chip structures in that array. one for the first four outputs, one for F/G and one for H outputs, I think. or leave H out until cr2 finds out.
09:35.01pH5cr2: I'd like to know the reset procedure and how SPI_CS is handled.
09:35.29pH5does it just pull CS high and then high/low reset for a bit?
09:35.32xsacha2oh tmzt, there's a mountd.conf which has block_device /dev/block/mmcblk0
09:35.37xsacha2i guess i dont want that?
09:36.08tmztthat's fine
09:36.22cr2pH5: SPI_CS is just set to 0 on BA. on sable it is set to 0 before the transaction, and to 1 in the end. i need to look and restore the backup.
09:36.25xsacha2in kaiser they have it as /dev/block/mmcblk0p2
09:36.32xsacha2can i comment it out?
09:36.46tmztyeah, because they are booting from vfat partition
09:37.00tmzthuh, it shouldn't be blk0 this must be for superfloppy
09:37.07xsacha2lol
09:37.07tmztI don't know much about mountd
09:37.18pH5ah right cs is usually active-low. how about reset? and is there another gpio involved besides those two?
09:37.27xsacha2but i dont need to mount anything right?
09:37.28tmztif you just mount /dev/block/mmcblk0p1 /sdcard
09:37.34xsacha2since everything is on root
09:37.35tmztyou can comment out mountd in init.rc
09:37.44xsacha2k
09:37.51xsacha2so i should change it to mmcblk0p1 ?
09:37.55pH5the new asic3 code resets the gpios, so I can't just use the settings that wince had set up before.
09:38.38cr2pH5: ok, so we need to be more careful then.
09:38.57Echo31pH5: Why the old structure (2.6.21)  had only 2 egpio_pin structure,  one for cpld1 and second for cpld2 ?
09:41.40cr2tmzt: some more data than on vogue then:
09:41.44cr270 struct rr_header {
09:41.46cr271         uint32_t version;
09:41.47cr272         uint32_t type;
09:41.49cr273         uint32_t src_pid;
09:41.50cr274         uint32_t src_cid;
09:41.52cr275         uint32_t confirm_rx;
09:41.53cr276         uint32_t size;
09:41.55cr277         uint32_t dst_pid;
09:41.56cr278         uint32_t dst_cid;
09:41.58cr279 };
09:41.59cr28 more ints
09:42.09cr2before the pacmark
09:42.11tmztlooks like dmov
09:42.18tmztor dma
09:42.29pH5Echo31: the old one had a .pins array where you had to configure every non-standard pin (output default-on or input associated with irq) individually. other than that it didn't mind.
09:42.34tmztoh, rpcrouter
09:42.43pH5the new one registers the input/output type with the gpio api properly.
09:42.50tmztpH5: is it mfp now?
09:43.03pH5IIRC, of course.
09:43.06cr2tmzt: rr_header+pacmark+rpchdr+msg
09:43.24tmztgood
09:43.31tmztcid is always 0 or maybe 1?
09:43.46pH5tmzt: sorry, what is? htc cpld gpio chips don't have alternate functions on their pins.
09:43.48cr2tmzt: looking at the raw smem data
09:43.48tmztpid is adsp process from that function earlier?
09:44.21tmztpH5: the default-on, edge-detect, etc. could be implemented the same way, but I guess it make no sense without altfunctiosn/pinmux
09:44.30tmztpH5: so it's gpio-chip?
09:44.37cr2pH5: but asic3 has
09:44.57cr2pH5: i also need to check if SPI_CS is an alt func
09:45.14pH5cr2: asic3 is configured similarly to the pxa mfp setup with an array in its platform_data
09:45.33pH5the cplds are just plain gpio chips + irq, no mfp configuration.
09:45.56cr2yes
09:47.11pH5although I think the upstream is still missing the ability to configure asic3 sleep_setting.
09:47.20pH5s/upstream/upstream asic3 driver/
09:47.32cr2ok
09:47.43pH5will see when I play with suspend/resume on hx and ba.
09:48.41pH5it's a bit frustrating to redo all this stuff that was basically around since 2005/06 in some form or another.
09:51.51xsacha2tmzt: could the kernel panic due to me missing stuff in kernel that android wants?
09:52.15tmztxsacha2: the kernel panic is because init crashed, why I don't know
09:52.21pH5off, bbl
09:55.23Echo31cr2: Can you use your code from t-mob ameo device  for athena (7500) ?
09:55.58Echo31cr2: Can i use your code from t-mob ameo device  for athena (7500) ?
09:56.02cr2Echo31: i hope so. let me know if something breaks.
09:56.19cr2tmzt: 00000001 00000004 00000001 fffffffe 00000000 00000014 00000000 fffffffe
09:56.37cr2tmzt: this is the rr_header, as in tge g1 code.
09:56.57cr2tmzt: so we have the router, and can't use plain rogue code.
09:57.14cr2s/rogue/vogue/
09:57.15tmztfine, I want to embed everythin in my htc-hw for now
09:57.28tmztI was going to make snd.c a module but it's too much work for right now
09:57.37cr2ok
09:57.53tmztwhere are the structs/rpc code I need in htc-vogue?
09:58.37cr20x14=20bytes
09:58.43cr200000004 31000002 00000000 00000001 6dc3867a
09:58.47cr2must be the reply.
09:59.15cr2smd_rpcrouter.c is the original version
09:59.30tmztI see that
09:59.36tmztandroid version?
09:59.41cr2vogue_rpc.[ch] is the amss4 version according to dzo
09:59.55cr2i don't know anything about android :)
10:00.36tmztit's not in htc-msm-2.6.27
10:00.37dzoyes, check if your rpc smem dump matches smd_rpcrouter.c
10:00.47tmztI mean the google/android version
10:01.04tmztdzo: your code?
10:01.12cr2vogue is a separate branch
10:01.17tmztit's only in htc-vogue that I can find
10:01.21cr2yes
10:01.26dzono, looks like my code won't work on raph.
10:01.40tmztI mean is rpcrouter your code?
10:01.45tmztoh no
10:01.50dzothough the audio stuff should be similar.
10:01.51tmztSan
10:02.08tmztwhy are we missing it then?
10:02.13tmztin htc-msm-2.6.27
10:02.19dzono, rpcrouter is the google code.
10:02.34*** join/#htc-linux pleemans (n=toi@d54C2AAB7.access.telenet.be)
10:02.48cr2dzo: why didn't you fix snd.c, like audiomgr.c, but have vogue-hw.c hacks ?
10:03.23tmztit is here
10:03.34cr2hmm. we need to add *ept there too.
10:03.59tmztwhat are ept, and what is the search function or connect function for?
10:04.14cr2msm_rpc_call()
10:04.19cr2msm_rpc_connect()
10:04.25tmztMSM_ONCRPCROUTER?
10:04.33tmztenable that?
10:04.45cr2connect gives you the ept, which is the first parameter in _call
10:04.49cr2yes
10:04.52tmztobj-$(CONFIG_MSM_ONCRPCROUTER) += smd_rpcrouter.o
10:04.52tmztobj-$(CONFIG_MSM_ONCRPCROUTER) += smd_rpcrouter_device.o
10:04.52tmztobj-$(CONFIG_MSM_ONCRPCROUTER) += smd_rpcrouter_servers.o
10:04.55tmztok
10:04.56dzoI started to, then realised that it doesn't work for android 1.1, that uses a userspace audio router, so I left it in my ril.
10:05.14cr2dzo: ok
10:05.16tmztuserspace client?
10:05.24tmztthat might work for fso
10:05.26cr2tmzt: alsa. i want alsa :)
10:05.34dzoyes, libaudio.so
10:05.55tmztthe audiopara's are basically alsa settings I think
10:05.59tmztsimilar concept
10:06.20tmztI'm using the AudioPara.c like dzo though, converted .csv to it
10:06.29dzoThey are documented in the leaked msm doc at the end.
10:07.26cr2dzo: there are no 7201A docs
10:08.04cr2dzo: btw, i've added 8x50 clocks to http://www.htc-linux.org/wiki/index.php?title=MSM_CLK . very helpful in some cases
10:08.31cr2it seems that 8x50 is closer to 7x00 than to 7x00A in many cases
10:08.57tmzthttps://www.codeaurora.org/patches/quic/le/u-boot-2009.03_PATCH_2009_05_14.patch
10:09.01tmztfrom here
10:09.17cr2yes
10:13.35*** join/#htc-linux xsacha2 (n=sacha@123.100.137.226)
10:14.19xsacha2is initlogo.lre a picture?
10:14.21xsacha2rle*
10:14.28tmztyes
10:14.33*** join/#htc-linux BabelO (n=fcr@lun34-2-82-238-28-28.fbx.proxad.net)
10:14.43tmztyou have to use the converter program to make them
10:17.31nathan7=0
10:17.56*** join/#htc-linux StarLite (n=nnscript@s55916ca6.adsl.wanadoo.nl)
10:18.41*** join/#htc-linux Guest63738 (n=fnord@24-151-90-116.static.nwtn.ct.charter.com)
10:19.21xsacha2does init do stuff with mtd partitions?
10:19.28xsacha2<6>init: mtd partition %d, %s
10:19.57xsacha2it creates /dev/mtd/mtd%d
10:20.28nathan7=p
10:21.15tmztxsacha2: we are doing nothing with mtd partitions
10:21.24tmztand I hope you have it disabled in kernel
10:21.48tmztthat's okay, except then you still have the partition table
10:21.55tmztyou need to be very careful here
10:22.12tmztalso, have no idea how to recover a samsung device
10:22.20xsacha2CONFIG_ARCH_MTD_XIP=y ?
10:22.54xsacha2CONFIG_MTD=y?
10:23.01xsacha2CONFIG_MTD_PARITITONS=y?
10:23.22tmztdisable all of those
10:23.31xsacha2and CONFIG_MTD_CONCAT too i guess
10:23.43tmztyeah
10:23.54xsacha2could it be the cause of kernel panic?
10:23.57tmztbut it will when you CONFIG_MTD=n
10:24.00tmztprobably not
10:24.32xsacha2CONFIG_MTD_BLKDEVS
10:24.43xsacha2that one enabled too
10:25.28xsacha2there's a tonne more MTD configs set ... bank_width_8,16,32 and MTD_CGI_I1,I2
10:25.57xsacha2and the there's MTD_NAND and MTD_NAND_PXA3xx enabled
10:26.05xsacha2disbale everything to do with MTD?
10:26.12tmztyeha
10:26.20tmztMTD_NAND_PXA3xx is the bad one
10:26.31tmztbut just disable all of them for now
10:26.55tmztit's not bad, it's just you don't know your mtd partition table, and you can overwrite bootloader, ce, etc.
10:27.05xsacha2:\
10:27.11xsacha2thats not bad?
10:27.55tmztthat is bad
10:27.55tmztbut the driver itself isn't, didn't want to imply it was
10:27.58dream_killtmzt nand driver works now ?
10:28.00xsacha2so there's a chance of bricking now?
10:28.06tmztdream_kill: talk to cr2
10:28.13dream_killok
10:28.25dream_killas i can give u the exact partition table :P
10:28.36dream_killi did on the g1 manual partition table :P
10:28.39tmztxsacha2: yes, please disable that and compile your kernel, delete the zImage with it enabled
10:28.55xsacha2wait wait, chance of bricking with mtd enabled or disabled?
10:28.58tmztdream_kill: cool, some people might want that now that yaffs2 root has shrunk
10:29.03tmztdream_kill: on ion, etc.
10:29.09tmztxsacha2: enabled
10:29.24tmztxsacha2: mtd is memory technology device, meaning flash memory
10:29.34dream_killyes, we can redefine it in kernel and not as a parameter from hboot anymore :P
10:29.42tmzthboot?
10:29.48tmztwhat is that? is there source?
10:29.52dream_killg1 spl is called HBOOT
10:30.09dream_killHBOOT will send to the kernel the partition table
10:30.20*** join/#htc-linux cr2 (n=cr2@ip-77-25-89-159.web.vodafone.de)
10:30.22tmztyeah, from amss misc
10:30.40dream_killnope
10:30.51dream_killthe hboot has it's own partiton table alocation!
10:30.56dream_killexample
10:30.58dream_killsame radio
10:31.04dream_kill0.95.0000 hboot
10:31.09dream_killsystem partition is 70 meg
10:31.24dream_killsame radio, 1.33.20005 system partition is 91 meg
10:31.31dream_killso 200% is in HBOOT!
10:31.50dream_killflash older hboot , system is small again :P
10:31.52tmztbut we have no source for HBOOT?
10:31.54dream_killnothing else to flash :P
10:31.58dream_killnope
10:32.00dream_killbut
10:32.07dream_killin kernel i can patch the partition table
10:32.11dream_killand make it static :P
10:32.23tmztI have another way I'm working on
10:32.25dream_killso who cares about hboot anymore :P
10:32.35tmztdemo is here: people.openezx.org/tmzt/testbooter.S
10:32.42tmztbut the second ldr should be adr
10:32.50tmztand the tags are not correct
10:32.53cr2wtf is HBOOT ?
10:33.08xsacha2well disabling MTD took 200KB off my kernel
10:33.09tmztI don't know if this can replace zImage in the android boot.img
10:33.13dream_killHBOOT in G1 is same like SPL in windows mobile :P
10:33.20cr2ok
10:33.30dream_killnow
10:33.37dream_killi did some mad crasy tests on g1 :P
10:33.38cr2which is 512K. lol
10:33.47dream_killflashed diamond dump in a g1 :P
10:33.56tmztand?
10:33.56dream_killit started but no image on screen
10:34.01tmztreally
10:34.04dream_killyep
10:34.07tmztthat would be the mddi chip then
10:34.12dream_killi have jtag for g1 :P
10:34.20dream_killso i can do lots of nuts tests :P
10:34.28cr2tmzt: lcd power gpios
10:34.33cr2tmzt: and vregs
10:34.43tmztI thought you were trying radio from diam100 on drea100
10:35.08tmztis there a list of everything spl does except boot windows/linux?
10:35.13tmztall the init it does
10:35.17tmztanything with amss
10:35.32*** join/#htc-linux pleemans (n=toi@d54C2AAB7.access.telenet.be)
10:35.36cr2tmzt: spl does not boot
10:35.50tmzt?
10:36.00tmztit doesn't load xip?
10:36.49dream_killi flashed full radio!
10:36.58dream_killfull dump :P
10:37.00tmztall bands work?
10:37.20tmztif so, you are essentially confirming rf is the same
10:37.41tmztor did you have to patch in the amss drivers? I guess not with full dump
10:38.00cr2tmzt: how can it work, if the A2M/M2A communication is different.
10:38.15tmzthe flashed the amss as well
10:38.19tmztwhat do you mean
10:38.20tmzt?
10:38.25cr2tmzt: flashing g1 radio on diam/raph may have been more useful
10:38.49cr2but the kernel will need the gpio/cpld, uarts adapted
10:39.22xsacha2can i run gdb or something on init to find out why it causes kernel panic?
10:39.30dream_killwhat i did, someone sent me a full nand backup (via jtag) from diamond and flashed it in the g1
10:39.31tmztstrace
10:39.36cr2tmzt: g1 radio will save you the PITA with adsp, and rpc
10:39.41dream_killi flashed till 25000000
10:39.42tmztnot me
10:39.43xsacha2ok
10:39.44dream_kill(after spl
10:40.04dream_killbut next time i'll replace in my g1 only the radio not the htc partition, mics and mfg
10:40.08dream_killand see if works better
10:40.18tmztthat won't work
10:40.20xsacha2good thing they are both htc devices
10:40.41cr2dream_kill: where does the imgfs end ? and fat begins in your nand ?
10:41.11tmztdream_kill: the other thing I want to ask,
10:41.31tmzthow can I create an essentially empty imgfs with just xip section
10:41.36tmztof 3+mb
10:44.20dream_killimgfs ?
10:44.21xsacha2oh damn! strace not found :P
10:44.22tmztI've been reading code for different versions of imgfstools, htcflash, etc.
10:44.27tmztnot imgfs
10:44.34tmztOS.bin.payload
10:44.39dream_killg1 doesn't have imgifs
10:44.44tmztI know
10:45.05tmztI have cdma touch pro, trying to make a linux bootloader bundle to replace xip
10:45.27dream_killso
10:45.31tmztxsacha2: how are you trying it?
10:45.39dream_killSPL starts at 0x2400000
10:46.00dream_killthe rom itself starts from 0x2820000
10:46.27dream_killso imgfs should start from 0x2820000
10:46.33dream_killis same for g1 and diamond too
10:46.46xsacha2tmzt: what do you mean?
10:46.48dream_kill(is defined in the partition from radio)
10:46.53xsacha2i need to compile strace for arm
10:46.59dream_killthis is nand addresses!
10:47.03tmztyeah, it should be xbin
10:47.56dream_killso play with anything after 0x2820000 and u'll be safe as u'll not touch the spl and any other danger area :P
10:48.40tmztwhere is xip, where is code loaded from, and is this all the ce encapsulated "blocks"?
10:48.41dream_killhttp://lwn.net/Articles/135472/
10:49.01tmztnot linux xip
10:49.16tmztwon't work on nand, though the new split filesystem is cool
10:49.29tmztthe fake xip of ce, which is really ram
10:49.54dream_killso then try to flash there the boot.img and see if it starts it :P
10:50.24tmztthere's no partition table there?
10:50.27dream_killnope
10:50.30dream_killi told u!
10:50.42dream_killin android the hboot pass partition table
10:50.46tmztoh, I'm confusing things
10:50.56dream_killIN windows mobile the partition table is passed by the radio
10:51.05tmztthe question is about nbh, not the contents of nand
10:51.15tmztbecause I want to try diag.nbh first
10:51.30dream_killso if u look in the radio in partition section
10:51.39dream_killu see the apps and ext_rom localtion
10:51.47tmztso I need to know how to construct a OS.nb without any imgfs or tfat(empty) partition
10:51.50dream_killapps is the windows ce partition
10:52.02dream_killext_rom is the user space
10:52.11dream_killor extended rom
10:52.36dream_killquestion is how does the spl starts the apps
10:52.42dream_killthis u need to  check
10:52.53dream_killif check for imgfs and starts like so
10:52.58dream_killor it does a jump to it
10:53.19*** join/#htc-linux WyrM (n=wyrm@20158157247.user.veloxzone.com.br)
10:53.35dream_killgtg back in 5h
10:55.19cr2dream_kill: that's already documented http://www.htc-linux.org/wiki/index.php?title=RaphaelNAND
10:55.37cr2hehe.
10:55.55cr2there was an absolute offset there.
10:55.57cr2+0x02820000    ?
10:56.06tmztjump?
10:56.17tmztor just address, nand address?
10:56.28cr2wait.
10:56.43nathan7tmzt: Freshly built.
10:56.43cr2it comes just  from +0x00420000   0x2400000 wince imgfs
10:57.15tmztdo you have a raw dump of this?
10:57.22tmztcan you check for aa55 signature?
10:57.28cr2itutils
10:57.36tmztyeah, I'm in linux right now
10:57.43cr2but i think i've dumped it incorrectly
10:58.04cr2because i followed the itsutils PartX values
10:58.08tmztI'd like to know if the OS.nb is there with it's partition table
10:59.07cr2MSFLSH50 was visible in /dev/mtd/mtd5
10:59.31tmztthere is source that might explain the ce blocks
10:59.43tmztcode.google.com/p/htcflasher trunk
10:59.44cr2yes
11:00.17cr2tmzt: you may have xip in ext2 afair
11:00.21*** join/#htc-linux DJWillis (i=djwillis@82-46-19-72.cable.ubr02.bath.blueyonder.co.uk)
11:00.36cr2tmzt: but then you need to fix the msm_nand.c ;)
11:00.40tmztyeah, but I don't think on nand
11:00.43tmztoh
11:00.59tmztbut I'm not really concerned about that right now
11:01.00cr2maybe we can ask SanMehat about it.
11:01.10tmztand we will probably use yaffs2 I guess
11:01.39cr2btw, what is yaffs2 advantage over jffs2 ?
11:01.56tmztMSFLSH50 is imgfs, the xip should be before that
11:02.05cr2there was some talk about jffs3, but i don't know what came out.
11:02.07tmztnot sure, I could look into it
11:02.19tmztthe mtd journaling support is newer
11:02.23nathan7asplodes
11:02.25tmztit has embedded ftl/bbt
11:02.40cr2ok
11:03.33tmztthe big one might be google/android has a supported version tested for msm
11:04.18cr2yes
11:05.12cr2but it still should not barf at non-2K access
11:05.17xsacha2what happens when haret freezes during booting linux?
11:05.30cr2i see it as a bug, and not a feature
11:05.45tmztwhat is?
11:05.49xsacha2well it's not haret's fault.. my winmo was using 100% cpu when i loaded haret
11:05.50tmztxsacha2: at what point?
11:05.58xsacha2the blue line
11:06.04cr2xsacha2: you do a reset ;)
11:07.15tmztwhat is unable to handle non-4k blocks?
11:07.19xsacha2this is weird.. i didnt execute init and it gave me same kernel panic tmzt
11:07.32tmztcmdline?
11:07.59xsacha2set cmdline "root=/dev/mmcblk0p2 rw init=/initi780 rootdelay=5"
11:08.14xsacha2i have an initi780 which says some text and then runs init
11:08.23tmztpaste it here
11:08.24xsacha2except, this time i didnt run init and i got same kernel panic
11:08.26tmztif it's short
11:08.43tmztattempting to kill init?
11:08.46xsacha2yeah
11:11.57xsacha2so i need to have the original initi780 stay open forever?
11:12.25nathan7exec.
11:12.26tmztno, exec /init
11:12.30nathan7D=
11:12.32nathan7You beat me!
11:12.35nathan7beats tmzt
11:12.40xsacha2i have exec /strace /init
11:12.51nathan7...
11:12.55nathan7Remove the strace.
11:13.01xsacha2why?
11:13.13nathan7Might help.
11:13.20xsacha2i know i dont have the right options on it, but i need to strace it
11:13.22nathan7Strace spawns a subprocess
11:13.43nathan7It should be able to fork and exec though
11:13.45*** join/#htc-linux kiozen (n=oeichler@rgnb-5d87468d.pool.einsundeins.de)
11:13.59xsacha2so maybe initi780 just loads bash and then i strace from there
11:14.23nathan7NO
11:14.28nathan7Init has to be pid 1
11:14.30xsacha2lol
11:14.40xsacha2damn, how can i debug it then?
11:15.32cr2tmzt: evil bstrds qcom. i think i understand what's broken here  ;)
11:15.57cr2need to do more clk research.
11:16.15nathan7xsacha2: Make strace fork and exec
11:16.24nathan7So
11:16.32tmztcr2: which clk?
11:16.38tmztcr2: for cfi/nand?
11:16.44xsacha2./strace -ff -F -tt -s 200 -o /tmp/strace init ?
11:17.03cr2tmzt: no. msm_clk
11:17.03xsacha2exec that?
11:17.19tmztwhy? you mean the scaling is breaking it?
11:17.34nathan7Dunno
11:18.06nathan7if(fork()){exec('/init');}
11:18.22xsacha2where does the strace come in to it?
11:19.56nathan7That's what should happen in strace kinda
11:20.09xsacha2oh
11:21.26nathan7/away bbl
11:22.29tmztmaybe strace -p 1 could work
11:22.41tmztif it was started with waitfor
11:23.05tmztbut there is already a 1, so maybe not
11:23.09tmztunless we don't care
11:23.33tmzt(while true; do strace -p 1; done) &
11:23.36tmztexec /init
11:23.45tmztah, but not tty
11:24.05tmzt(while true; do strace -p 1 2>/dev/tty0; done) &
11:24.24tmztbefore that: mknod /dev/tty0 c 4 0
11:24.30tmztor do that on the card
11:27.07xsacha2strace -p 1.... init?
11:27.38xsacha2dont you call strace with init?
11:27.41tmztno, what I typed
11:27.54tmztadd that line to your /initi780
11:28.01tmzt(while true; do strace -p 1 2>/dev/tty0; done) &
11:28.20xsacha2exec /init             mknod /dev/tty0 c 4 0           (while true; do strace -p 1 init 2>/dev/tty0;done) &
11:28.47tmztno
11:28.49xsacha2oops, no init in there
11:30.52xsacha2exec /init             mknod /dev/tty0 c 4 0            (while true; do strace -p 1 2>/dev/tty0;done) &               right?
11:32.06xsacha2ok ill try that (with exec init after the while loop)
11:33.22xsacha2mm can't open /dev/null
11:34.40xsacha2the whole loop gives error that it can't open /dev/null.. im guessing i need to mknod for /dev/null so it scan throw out stuff it doesnt need (2>(
11:35.50xsacha2mknod /dev/null c 1 3
11:36.44xsacha2beautiful strace!!
11:36.50xsacha2thanks! :D:D
11:37.56xsacha2it seg faulted while reading default.prop
11:39.03*** join/#htc-linux zycho_ (n=zycho@dslb-088-070-073-162.pools.arcor-ip.net)
11:56.36*** join/#htc-linux jeanseb (n=jeanseb@gazypan.dyndns.org)
11:56.43jeansebHi
12:17.20*** join/#htc-linux MLM (n=mlvdmeid@meide.xs4all.nl)
12:26.15*** join/#htc-linux wirelessdreamer (n=dreamer@chrobd01.vailsys.com)
12:52.35*** join/#htc-linux Tinyboom (n=nahh@206.84-49-89.nextgentel.com)
13:08.39*** join/#htc-linux MethoS- (n=clemens@host-091-096-213-249.ewe-ip-backbone.de)
13:09.08cr2omg. now i see it
13:10.46cr273
13:11.20cr20x49, clk_79 offset e0
13:12.02cr2hehe. that's why it hangs on uart1
13:12.07cr2really evil
13:19.21*** join/#htc-linux JuNu_KN (n=chatzill@kons-5f714bbe.pool.einsundeins.de)
13:23.38*** join/#htc-linux miknix (n=miknix@gentoo/developer/miknix)
13:27.20*** join/#htc-linux MethoS-- (n=clemens@host-091-097-245-151.ewe-ip-backbone.de)
13:28.06*** join/#htc-linux AntiXpucT (n=Skim@77.106.108.232)
13:28.25*** join/#htc-linux pH5 (n=ph5@p5485FD24.dip.t-dialin.net)
13:36.50*** join/#htc-linux pleemans (n=toi@d54C2AAB7.access.telenet.be)
13:47.39xsacha2yay, finished my applied maths 3 assignment
13:53.37Echo31pH5: good afternoon. I trie to follow your suggestion http://fr.pastebin.ca/1442294
14:06.11*** join/#htc-linux Guest63738 (n=fnord@24-151-90-116.static.nwtn.ct.charter.com)
14:07.34*** join/#htc-linux xsacha2 (n=sacha@123.100.137.226)
14:11.07pH5Echo31: I'd suggest to put CPLD1 GPIOs directly after PXA270 ones (starting at NR_BUILTIN_GPIO) and to put CPLD2 ones directly thereafter.
14:11.43pH5I think PXA has the default of max. 256 GPIOs set in include/asm-generic/gpio.h
14:12.27pH5also, iomem resource end should be the last used byte, not the first unused one
14:12.37pH5i.e. .end = XYZ_BASE + size - 1,
14:12.42xsacha2so many people with pxa :)
14:15.02pH5and of course you need to give an IRQ resource for CPLD1 (the gpio irq on which the cpld irqs are multiplexed)
14:28.24Echo31ph5: Do you suggest to create one egpio_resource structure with 3 items ?
14:28.24Echo31[0] for pxa270 (PXA_CS3_PHYS)
14:28.24Echo31[1] for cpld1
14:28.24Echo31[2] for cpld2
14:45.16xsacha2is it true i can take the source from android.git.kernel.org and add my machine and it'll be patched to work with android?
14:48.42Echo31pH5:I don't find a GPIO CPLD1 irq,  perhaps the GPIO 14 =  'CPLD chain' from http://wiki.xda-developers.com/index.php?pagename=AthenaGPIO
15:00.15dcordes-kaisxsacha2 kernel source?
15:00.31xsacha2yeah
15:00.41xsacha2apparently i need to add android patches to get it to work with android
15:00.58xsacha2so i was wondering if android.git.kernel.org already has those
15:01.01dcordes-kaisah got you
15:01.14dcordes-kaisthe binder stuff etc.
15:01.17xsacha2yes
15:01.47dcordes-kaisprobably. there was one person here who patched some pxa kernel for androed use
15:02.30xsacha2oo i need :)
15:02.32xsacha2im using pxa
15:03.16dcordes-kaisthey had some kind of blog website where all the info was put
15:03.35*** join/#htc-linux zycho (n=zycho@dslb-088-070-073-162.pools.arcor-ip.net)
15:04.10dcordes-kaiscan't recall the nickname unfortunately. which device are you using?
15:05.56xsacha2i780
15:06.50cr2dcordes-kais: we have some rpc progress. it's different from vogue and more close to g1
15:08.48xsacha2ok im gonna get android.git.kernel.org source and just add my machine
15:11.27xsacha2which kernel version is recommended? latest?
15:12.23*** join/#htc-linux pleemans (n=toi@d54C2AAB7.access.telenet.be)
15:18.23*** join/#htc-linux dcordes (n=luke@unaffiliated/dcordes)
15:19.51xsacha2ok i figured i have a very common problem
15:19.56xsacha2seg fault after reading config files
15:22.20*** join/#htc-linux timebomb (n=tb@f054023160.adsl.alicedsl.de)
15:33.23*** join/#htc-linux xsacha (n=sacha@123.100.137.226)
16:03.36*** join/#htc-linux hollo (n=hollo@3e6b7b2c.rev.stofanet.dk)
16:06.00dcordes-kaisdcordes-kais, go away, punk!
16:06.31dcordesdcordes-kais, you talking to yourself? wow
16:06.49dcordes-kaislol
16:07.32cr20x38/4=14
16:07.44cr214-10=4
16:08.48cr20x3c/4=15
16:08.51*** join/#htc-linux WyrM (n=wyrm@20158136090.user.veloxzone.com.br)
16:08.52cr215-10=5
16:08.59cr2looks very good.
16:09.14dcordes-kaiswtf is that?
16:09.50cr2coming patchez
16:10.36cr2dcordes-kais: i've found out why enabling uart1 clk crashed raph100
16:11.10AstainHellbringwhy did it crash cr2?
16:12.01cr2AstainHellbring: because the uart_clk is at a different location. looking at oemsbl helps ;)
16:13.24AstainHellbringlol yep that can help
16:13.48AstainHellbringcr2 curious I saw the check in for enabling BT on the device will BT audio work now?
16:15.05cr27200 and 8x50 have it at c0, and 7x0xA at e0
16:15.19cr2not, the ttyHS1 oopses
16:15.38cr2need to track it down too
16:30.51cr2mickey|sun: why did you disable the accelerometer in defconfig ?
16:33.47*** join/#htc-linux miknix_ghost (n=miknix@bl4-91-2.dsl.telepac.pt)
16:34.35*** join/#htc-linux miknix (n=miknix@gentoo/developer/miknix)
16:34.37cr20x2c/4=11, -10=1
16:34.38*** join/#htc-linux wdslbr (n=asa@dslb-088-069-208-087.pools.arcor-ip.net)
16:45.17cr20x40/4=16, -10=6
17:19.44Echo31pH5: I wrote this code on the same egpio resource  (http://fr.pastebin.ca/1442494). Is it the way?
17:34.11*** join/#htc-linux wdslbr (n=asa@dslb-088-069-208-087.pools.arcor-ip.net)
18:23.20*** join/#htc-linux MLM (n=mlvdmeid@meide.xs4all.nl)
18:30.21*** join/#htc-linux fnord_ (n=fnord@24-151-90-116.static.nwtn.ct.charter.com)
18:41.17*** join/#htc-linux infernix (i=nix@unaffiliated/infernix)
18:42.47*** join/#htc-linux AstainMBL (n=AstainHe@unaffiliated/astainhellbring)
19:12.44*** join/#htc-linux lama (i=lama@netbsd.pl)
19:18.59*** join/#htc-linux WyrM (n=wyrm@189.13.124.134)
19:19.09*** join/#htc-linux marex (n=marex@vasut.kolej.mff.cuni.cz)
19:24.01*** join/#htc-linux Zoolooc (n=fredsiba@nrbg-4dbf8079.pool.einsundeins.de)
19:27.36*** join/#htc-linux lucxxx (n=o_lucian@89-115-128-35.cl.ipv4ilink.net)
19:30.54*** join/#htc-linux fnord_ (n=fnord@24-151-90-116.static.nwtn.ct.charter.com)
19:34.30*** join/#htc-linux nebi_ (n=nebi@217.142.147.19)
19:46.43cr2#if CONFIG_MSM_AMSS_VERSION < 6220
19:49.14cr2audmgr ok
19:50.20tmzthey
19:54.50*** join/#htc-linux tre9 (n=ubuntu@lepke.dreamhost.com)
19:56.19cr2struct rpc_adsp_rtos_app_to_modem_args_t {
19:56.20cr2<PROTECTED>
19:56.22cr2<PROTECTED>
19:56.24cr2<PROTECTED>
19:56.25cr2<PROTECTED>
19:56.26cr2<PROTECTED>
19:56.28cr2};
19:57.27cr2< 6220 means == 6210
19:58.32cr2tmzt: can you just enable qdsp and rpc* (without time*) Kconfig options, and try to compile ?
19:59.17tmztDisconnect: here.
20:00.06tmztI only have ADSP and ONCRPCROUTER
20:00.11tmztin .config
20:01.41cr2CONFIG_MSM_RPCSERVERS ?
20:02.19*** join/#htc-linux MLM (n=mlvdmeid@meide.xs4all.nl)
20:02.19tmztyes, that was enabled
20:02.27cr2ok, battery is CONFIG_TROUT_BATTCHG, so it's not important
20:02.47tmztwhat about AMSS version, do we need to change that?
20:03.49cr2already 5200
20:04.01cr2we will get compile errors ad will fix them.
20:04.32cr2all  #if CONFIG_MSM_AMSS_VERSION < 6220
20:04.47cr2should be changed to  #if CONFIG_MSM_AMSS_VERSION == 6210
20:05.10cr2dcordes-kais: ping
20:06.42tmztarch/arm/mach-msm/qdsp5/adsp.c:890:2: error: #error "Unknown AMSS version"
20:08.31cr2yes. ok, i'll try myself.
20:11.31cr2let's hope that adsp_6210 is useful for us
20:12.02cr2maybe diff with the version of dzo
20:24.32*** join/#htc-linux WyrM1 (n=wyrm@189.13.124.134)
20:30.02*** join/#htc-linux MLM_ (n=mlvdmeid@meide.xs4all.nl)
20:35.27cr2tmzt: warning ?
20:35.31cr2arch/arm/mach-msm/qdsp5/adsp.c: In function ‘msm_adsp_probe’:
20:35.32cr2arch/arm/mach-msm/qdsp5/adsp.c:805: warning: assignment makes integer from pointer without a cast
20:35.33cr2arch/arm/mach-msm/qdsp5/adsp.c:806: warning: assignment makes integer from pointer without a cast
20:35.35cr2arch/arm/mach-msm/qdsp5/adsp.c:807: warning: assignment makes integer from pointer without a cast
20:35.47tmztI usually ignore those
20:37.44cr2but what is the reason ?
20:37.58tmztno idea, hold on
20:38.34tmztadsp_info.send_irq += MSM_AD5_BASE ?
20:39.05cr2yes
20:39.24cr2is adsp_info.send_irq  a pointer ?
20:39.36tmztWyrM: gvim  -k adsp_info is that right?
20:40.20tmzt-t
20:40.22tmztsorry
20:40.32tmztworked
20:40.56tmztunit32_t send_irq
20:41.15tmztIOMEM(0xE0300000)
20:41.22tmztI guess that's a pointer?
20:41.38cr2should not be
20:41.42cr2what is IOMEM ?
20:41.45tmztno
20:43.15tmzt#define IOMEM(x) x
20:43.57tmztmaybe adsp_info?
20:43.59cr2ok
20:44.42tmztstruct adsp_info *info;
20:45.44tmztwhere does adsp_info come from?
20:45.56tmztin msm_adsp_probe
20:46.00tmztshould it be info?
20:46.05tmztthis is google/android code
20:46.29tmztstatic struct adsp_info adsp_info;
20:48.07*** join/#htc-linux WyrM1 (n=wyrm@20158130106.user.veloxzone.com.br)
20:51.49goxboxlivecr2, are you there?
20:55.45j0b0tmzt, #define IOMEM(x)((void __force __iomem *)(x))  is a pointer
20:56.42tmztyeah, if __ASSEMBLY__ is defined
20:57.07j0b0no. if its not defined.
20:59.47goxboxlivecr2, anyway.I am going to Munich again (late october though). Meet you at the Hofbrähaus :-) ... I'll be sitting on a bench, a little drunk with a large beer an humming ompa-ompa music :-D
21:00.05goxboxlivetalk to you later on. night ..
21:00.39tmztj0b0: where?
21:01.02j0b0tmzt in msm_iomap.h
21:01.19tmztoh sorry
21:01.27tmztthat's it
21:01.57j0b0the proper way to fix those warning is probably to change these struct members to void __iomem *
21:02.27j0b0as that is also what readl() and writel() take
21:02.57cr2oh, i missed him.
21:03.05cr2j0b0: do you have umts ?
21:03.35j0b0in my sim, yes. working in linux, no.
21:03.59cr2j0b0: that means you can init a video call ?
21:04.20cr2j0b0: umts should work in linux (!=android)
21:04.27tmztyou can't cast to int?
21:04.39tmztisn't the point of the struct to define the wire format?
21:05.24cr2tmzt: booted with rpc and qdsp
21:05.38cr2don't know what to test :)
21:05.41tmztcool
21:05.52tmztwe need a little ioctl program
21:05.55tmztto set device
21:06.05cr2this oob is fscking annoying.
21:06.14j0b0that struct doesnt look like its used to serialize anything (or didnt you mean that)as it contains other pointers. but just casting to int would also work
21:06.14cr2how can we reach San ?
21:06.20tmztSND_SET_DEVICE
21:06.23tmzttry #android
21:06.40cr2tmzt: since the patch boots we can commit it.
21:07.20cr2j0b0: i'd like to get a smem dump of  video call to look for the rpcs. i don't know any other way to init the vga cam
21:08.07j0b0cr2 idont know about video calls. it does mean i have high speed internet access.
21:08.33j0b0cr2 do i need to patch haret to trace smem, or do you just need a dump, no trace
21:09.08tmztcr2: at least on kais I think it's not connected to mddi, but to the older qualcomm interface
21:09.10cr2j0b0: press the 'call' button, and then 'videocall'
21:09.20tmztcr2: just from looking at htc-vogue tree
21:09.27cr2tmzt: yes, it's not an mdc
21:09.36cr2tmzt: mdc is 3.2Mpix cam
21:09.49j0b0cr2 hold on .. booting into windows. which area do you need dumped afterwards?
21:09.51cr2j0b0: no, dump the full smem
21:10.10cr2pwf smem-vcall 0x01f00000 0x100000
21:10.14j0b0is that 1M from 0x01f00000
21:10.18cr2yes
21:10.20j0b0ok, so it is
21:10.22cr2the smem
21:10.41cr2<PROTECTED>
21:10.51cr2but the dmesg is spammed by oob ;)
21:11.18j0b0the 3G network is not verfügbar right now :/
21:11.29cr2hehe
21:11.34cr2j0b0: same here.
21:11.48cr2j0b0: but i used another card for it.
21:12.02cr2because i don't have voice on my umts card
21:12.21j0b0nicht anquatschfähig... i think it will take 3g if i force it to. the gsm signal is just stronger here
21:12.40cr2rpc 'a' is adsp. the snd and audmgr are not visible.
21:12.48cr2weird
21:13.44cr2tmzt: ok, so how do i play sound ?
21:13.48*** join/#htc-linux fnord_ (n=fnord@24-151-90-116.static.nwtn.ct.charter.com)
21:14.07cr2root@htcraphael:/sys# ls -als /dev/dsp
21:14.14cr2ls: /dev/dsp: No such file or directory
21:14.37tmztI don't know, I will look at audioflinger and see if I can figure it out
21:14.43cr2root@htcraphael:/sys# ls -1 /dev/ | grep msm
21:14.45cr2msm_mp3
21:14.47cr2msm_pcm_ctl
21:14.48cr2msm_pcm_in
21:14.50cr2msm_pcm_out
21:15.30cr2cat < /dev/msm_pcm_in
21:15.37cr2[  723.463031] audmgr: failed to connect to audmgr svc
21:15.53tmztah
21:16.00tmztyou changed the PROG number?
21:16.18cr2root@htcraphael:/sys# cat /tmp/fso-monitor.log > /dev/msm_pcm_out
21:16.19cr2-sh: can't create /dev/msm_pcm_out: No route to host
21:16.25tmztfun
21:16.27cr2i think yes.
21:16.33cr2yeah,
21:16.56cr2root@htcraphael:/sys# cat /tmp/fso-monitor.log > /dev/msm_mp3
21:16.57cr2-sh: can't create /dev/msm_mp3: No route to host
21:17.24cr2root@htcraphael:/sys# cat /dev/msm_pcm_ctl
21:17.26cr2cat: read error: Invalid argument
21:17.39cr2[  819.705737] audmgr: failed to connect to audmgr svc
21:17.53cr2yes, i need to check why these rpcs are not registered
21:18.13cr2i'll disable nand now.
21:19.14j0b0cr2 ok i dont get 3g right here right now
21:20.17cr2j0b0: ok. but if you want the vga cam, we need to get such log. i can't do it.
21:21.26j0b0is it a log youre loking for? (so can i make the dump after the fact, when im back in gprs)
21:21.57cr2yes. the 2K buffer is big enough to keep many rpc calls
21:22.21cr2realtime tracing is not really needed.
21:25.09cr2[    6.641070] smd_alloc_channel() 'SMD_RPCCALL' cid=2, shared=e0110198
21:25.10cr2[    6.642108] RPC_dev init complete!
21:25.12cr2[    6.642199] RPC Init done
21:25.13cr2[    6.642260] RPCCALL opened
21:26.51cr2oncrpc_router
21:26.57cr2rs3000000a:00000000
21:28.04cr2msm_snd
21:37.00parj0b0: are you going to ph-neutral party?
21:37.56j0b0par i dont know what that is
21:38.08paroh, thought you were in berlin
21:38.23j0b0no. i'm in groningen .nl
21:38.43parcr2: did you email san?
21:41.26*** join/#htc-linux cr2_ (n=cr2@ip-90-187-78-200.web.vodafone.de)
21:51.56cr2_tmzt: https://privatepaste.com/95xHKAFS4U
21:53.06tmztgreat, I'll save what I have as a git diff, reset and apply :0
21:55.13tmztcr2: just from looking at htc-vogue treederror: patch failed: arch/arm/mach-msm/qdsp5/Makefile:1
21:55.19tmzthah
21:55.24tmztyeah, didn't work
22:06.25*** join/#htc-linux pH5 (n=ph5@p5485F13B.dip.t-dialin.net)
22:06.30tmztcr2_: didn't apply, and git doesn't say why exacly
22:07.10tmztcould it be the blank line?
22:09.28tmzthttp://nopaste.info/3ff8a29e66.html
22:10.05tmztls /sys/platform/drivers from g1
22:10.12tmztstill need smem?
22:10.25tmztcan you give a range without private data?
22:11.00tmztrs3000000a:71d1094b
22:11.01cr2_tmzt: we don't enable dog and time, so it's ok
22:11.16tmztrd30000048:0da5b528
22:11.19tmztoh
22:11.23tmztthat's in the paste
22:11.31tmztwhat do you mean?
22:12.03cr2_48 is |   TIME_REMOTE_ATOM              || 0x30000048   || || time_*
22:12.41tmztis one of those audmgr?
22:13.13cr2_no.
22:13.17cr2_audmgr is 13
22:14.57tmztso g1 doesn't use it?
22:15.24tmztok, but why won't the patch apply?
22:16.23cr2_apply manually, it's not that big
22:17.11*** join/#htc-linux marcin_ (n=marcin@chello089078134143.chello.pl)
22:20.10*** join/#htc-linux Shinto (n=John@f048129043.adsl.alicedsl.de)
22:22.01parwhat repo?
22:22.29parlinuxtogo
22:22.30par?
22:33.43tmztcr2_: make[2]: *** No rule to make target `arch/arm/mach-msm/qdsp5/adsp_5200.o', needed by `arch/arm/mach-msm/qdsp5/built-in.o'.  Stop.
22:33.54tmztpar: doesn't apply to vogue, this stuff already works there
22:34.30tmztI don't havea adsp_5200.c, should I comment that out in Makefile?
22:35.11tmztor did I just not change the AMSS in .config
22:35.13cr2_tmzt: copy adsp_6210.c to adsp_5200.c
22:35.16tmztwhat should I change that to?
22:35.52cr2_the exact values are not known yet
22:36.40tmztI mean what are we testing it with now
22:36.51tmztarch/arm/mach-msm/qdsp5/adsp.h:130:2: error: #error "Unknown AMSS version"
22:36.51tmztmake[2]: *** [arch/arm/mach-msm/qdsp5/adsp_5200.o] Error 1
22:36.52tmztmake[1]: *** [arch/arm/mach-msm/qdsp5] Error 2
22:37.16tmztCONFIG_MSM_AMSS_VERSION=7000
22:37.24tmztshould that be 5200 or 6210?
22:37.45tmztdid you patch Kconfig?
22:38.50cr2_5200
22:39.02cr2_your git does not look very uptodate
22:40.01tmztAuto-merged arch/arm/mach-msm/smd.c
22:40.02tmztCONFLICT (content): Merge conflict in arch/arm/mach-msm/smd.c
22:40.35cr2_http://git.linuxtogo.org/?p=groups/mobile-linux/kernel.git;a=commitdiff;h=a8675d204a8789c2620778f83f2b6f0adc0cf9d5
22:40.39tmzt<PROTECTED>
22:40.40tmzt++=======
22:40.40tmzt+       if ((chn > 28) || (chn == 1))
22:40.55tmztif that's commited I can checkout the file
22:41.17cr2_yes
22:41.20tmztoh, I committed it locally
22:41.29tmztI guess I have to rebase?
22:41.47*** join/#htc-linux dcordes (n=luke@unaffiliated/dcordes)
22:41.50tmztor I'll just reapply my patches later
22:42.02dcordeshi
22:42.13cr2_hi dcordes
22:42.29cr2_dcordes: can you apply the rpc/qdsp ptch ?
22:42.36*** join/#htc-linux tsdogs (n=tsdogs@net203-187-146.mclink.it)
22:42.48cr2_it deos not do much, but i still can boot with it :)
22:43.02cr2_<PROTECTED>
22:43.04dcordesyes hand it over
22:43.30tmztdiff ../git-diff-tmp-a570f01443873251f0b481b400e7f591a7c99b25-d26173847fbb83776bd971947e3f065f17d19325.diff ../git-diff-absmsmts-fbdev-1.diff
22:43.33tmztcool
22:43.51cr2_dcordes: it enables rpc and qdsp, and provides the right version/prog/proc for the AMSS version maze
22:44.11tmztmy question is, we have CONFIG_MSM_AMSS_VERSION now
22:44.18tmztbut we have to variants
22:44.29tmztthe = 5200 and the _5200=y
22:44.37tmztand we don't have the Kconfig changes for that
22:44.44tmztto set it automatically
22:45.00tmztnot that we should, linux should be fine with multiple versions set
22:45.07tmztat least things like adsp
22:45.08cr2_you can sed WINCE to 5200, but it's not really needed.
22:45.24dcordescr2_, does it depend on the other changes you gave me? I did not apply them because mickey|sun told me he was going to do that
22:46.56dcordesok nvm
22:47.49cr2_dcordes: no, it's a completely separate patch
22:50.31cr2_the rpc looks like windows programming
22:50.44dcordesthe new A devices are compatbile with htc-msm-2.6.27 ?
22:50.57cr2_magic undocumented values to functions, cross the fingers, and hope it will work
22:51.16cr2_dcordes: we will see. but in big parts - yes.
22:52.18parits the audio mgr rpc for android for thr raphael?
22:54.34cr2_par: yes.
22:55.01cr2_par: umts works, so i need to make gps work, bt and wifi :)
22:55.03dcordes~seen alfadir
22:55.04aptalfadir <n=alfadir@e180018088.adsl.alicedsl.de> was last seen on IRC in channel #htc-linux, 24d 1h 39m 25s ago, saying: 'are there a maillist or forum that are active or where is the latest chatter taking place ?'.
22:55.13cr2_the rest is not much worth to me :)
22:57.05tmztno anything about the trout errors?
23:04.14cr2_|                 ||    || 7,1 || clk_regime_sec_enable (x48,x28,x33,x29)
23:04.22tmztI just switched to msm_defconfig after pulling
23:04.26tmztand it started
23:04.41tmztarch/arm/mach-msm/board-trout-panel.o:(.data+0x734): multiple definition of `toshiba_client_data'
23:04.44tmztarch/arm/mach-msm/board-htcraphael-panel.o:(.data+0x500): first defined here
23:04.49tmztah, disable TROUT
23:05.45dcordescr2_, enable rpc and qdsp, provide the right AMSS version in procfs
23:06.42tmztbuilding now
23:07.16dcordespushed
23:07.39tmztpushed what?
23:07.51dcordeshttps://privatepaste.com/95xHKAFS4U
23:08.34cr2_dcordes: procfs ?
23:08.48cr2_http://www.htc-linux.org/wiki/index.php?title=MSM_RPC
23:08.55tmztarch/arm/mach-msm/built-in.o: In function `battery_show':
23:08.56tmzt/home/tmzt/src/kernel/arch/arm/mach-msm/htc_hw.c:70: undefined reference to `msm_proc_comm_wince'
23:09.00dcordesoh too late
23:09.05cr2_looks much better now, with the msg size and sample traced values
23:10.24cr2_wow
23:10.33tmztnand won't do anything without the ATAGs, right?
23:10.54cr2_pdsm_get_location returns 28 ints
23:11.14cr2_tmzt: the partition table is missing, and the writes below 36MB are trapped
23:11.23tmztok
23:11.27tmztthat includes spl?
23:11.53tmztarch/arm/mach-msm/built-in.o: In function `halibut_init':
23:11.53tmzt/home/tmzt/src/kernel/arch/arm/mach-msm/board-htcdiamond.c:269: undefined reference to `msm_proc_comm_wince_init'
23:11.56tmzt/home/tmzt/src/kernel/arch/arm/mach-msm/board-htcdiamond.c:288: undefined reference to `msm_proc_comm_wince'
23:11.59tmzt/home/tmzt/src/kernel/arch/arm/mach-msm/board-htcdiamond.c:291: undefined reference to `msm_proc_comm_wince'
23:13.01tmztwhy is it still called halibut_init?
23:15.21cr2_blame NetRipper
23:15.34cr2_|                 ||    || 0xc,28 || pdsm_get_position (a,0,1,1,1,ffc99a3b,1,0,0,0,14*0,1,32,2,da3)
23:15.49cr2_nice call
23:16.14tmztso gps can work?
23:16.14*** join/#htc-linux AstainHellbring (n=AstainHe@unaffiliated/astainhellbring)
23:16.26*** join/#htc-linux marcin__ (n=marcin@chello089077185047.chello.pl)
23:16.49NetRipperno blaming me!
23:16.56cr2_:)
23:17.30tmztthey all have references to msm_proc_comm_wince_init
23:17.43tmztso msm_defconfig is broken
23:17.45tmztoh yeah
23:17.49tmztit's htcraphael now
23:18.30cr2_NetRipper: do you have videocall working in wince ?
23:19.49NetRipperi had once when i tried with another winmo user
23:19.54NetRipperhavent tried since a long time
23:20.40cr2_NetRipper: i'd like to get the smem dump after the videocall.
23:21.06NetRipperi can do that
23:21.11NetRippertuesday at work
23:21.15cr2_thanks
23:21.26NetRipperjust after the call ends?
23:21.29NetRipperor during?
23:21.48NetRipperah i'll just do both
23:21.57cr2_yes
23:21.58NetRipperwhat was the haret command?
23:22.16cr2_pwf mysmemdump 0x01f00000 0x100000
23:22.20NetRipperok
23:23.08cr2_NetRipper: instead of just providing nmea datastream on gpio enabled, msm is doing crazy things for gps ;)
23:23.40NetRipperwith the RPC you mean?
23:23.44NetRipperor even more weird things?
23:24.05cr2_not a direct access to the I/Q correlator data, or DSP. but still something like that.
23:24.05cr2_RPC
23:24.15cr2_pdsm_get_position (a,0,1,1,1,ffc99a3b,1,0,0,0,14*0,1,32,2,da3)
23:24.50cr2_da3 is probably some memory location
23:25.02NetRipperand a?
23:25.04tmztcompiled
23:25.04cr2_<PROTECTED>
23:25.23NetRipperhm
23:25.29NetRippermaybe where to place the data?
23:25.34cr2_don't know what is a
23:25.54NetRipperthat would also explain why ARM9 crashes
23:25.57NetRipperwhen linux boots
23:25.57cr2_algorithm/data path
23:26.16NetRipperMMU is reset, virtual memory addressing gone, any references the ARM9 still has are fucked
23:26.19*** join/#htc-linux zycho_ (n=zycho@dslb-088-070-073-162.pools.arcor-ip.net)
23:26.26cr2_NetRipper: http://www.htc-linux.org/wiki/index.php?title=MSM_RPC
23:26.36NetRipperoh
23:26.39NetRipperbrand new site
23:27.41*** join/#htc-linux [01dot11] (n=anonymou@cpc2-acto6-0-0-cust99.brnt.cable.ntl.com)
23:27.50cr2_exists for ages
23:28.27NetRipperbrand new wiki then
23:28.39NetRipperdid the wiki move?
23:28.45NetRipperfrom xda
23:30.10cr2_only for raph
23:30.30cr2_the research pages. not android
23:30.42NetRipperok
23:30.54cr2_i think athena will follow
23:33.01cr2_NetRipper: nand works
23:33.21cr2_NetRipper: and umts works, but i think you know it.
23:33.35NetRipperi saw the commit yes
23:33.57NetRipperif that if statement is the only change then i still doubt it'll work for me
23:34.04NetRipperi fiddled with that
23:36.28cr2_NetRipper: you need a haret version that provides partition ATAGs for nand to work
23:36.52NetRipperyes or we hard-code it in the kernel
23:37.01cr2_NetRipper: and i'm not 100% sure where imgfs ends, and fat begins
23:37.24cr2_NetRipper: it depends on the device.
23:37.28NetRipperi think i have the fat offset somewhere
23:37.35NetRipperdream_kill spilled it
23:37.41tmzthttp://code.google.com/p/htcflasher , the code for generating the ce oob is there somewhat
23:37.44tmztat least the size
23:38.10NetRipper"the ce oob"?
23:38.33tmztout of band
23:38.36tmztdata
23:38.54NetRipperah
23:40.47cr2_The requested URL /p/htcflasher/ was not found on this server.
23:41.59*** join/#htc-linux WyrM1 (n=wyrm@20158130106.user.veloxzone.com.br)
23:42.04tmzthttp://code.google.com/p/htc-flasher
23:49.47*** join/#htc-linux cr2 (n=cr2@ip-90-187-78-200.web.vodafone.de)
23:50.40*** join/#htc-linux zycho (n=zycho@dslb-088-070-073-162.pools.arcor-ip.net)
23:54.33cr2pdsm_get_position(*cbfunc,0,...
23:54.56tmztcr2: why doesn't debian initrd like it?
23:55.14*** part/#htc-linux [01dot11] (n=anonymou@cpc2-acto6-0-0-cust99.brnt.cable.ntl.com)
23:55.19*** join/#htc-linux zycho_ (n=zycho@dslb-088-070-073-162.pools.arcor-ip.net)
23:55.23cr2because it reads 0x6 bytes from superblock
23:55.41tmztwhat does?
23:55.42tmztoh
23:55.45tmztdiscover
23:55.47tmzt?
23:56.14cr2there is no block layer in between
23:56.19tmztis there any reason the driver can't read ahead behind the scenes?
23:56.20tmztyeah
23:56.29tmztyou have to use mtdblock I think
23:56.53cr2yes
23:57.31tmztso what in debian is probing mtd randomly?
23:57.35cr2then we need to patch msm_nand
23:57.43cr2don't know.
23:57.54cr2the real question is
23:58.12cr2do we want to have ext2+dos partition table on nand ?
23:58.19tmztI don't think so
23:58.32tmztdid you get the concept I gave you a few days ago?
23:58.35tmztdid that make any sense
23:58.39tmztor was that not you
23:58.57cr2ok
23:59.29cr2then i really would like to know how to find the fat offset (in a portable way)

Generated by irclog2html.pl Modified by Tim Riker to work with infobot.