*** Martix has quit IRC | 00:13 | |
*** Sc0rpius has quit IRC | 00:23 | |
*** Sc0rpius has joined #maemo-ssu | 00:23 | |
*** Jaded has joined #maemo-ssu | 00:25 | |
*** Jaded has joined #maemo-ssu | 00:25 | |
*** Jade has quit IRC | 00:27 | |
*** Pali has quit IRC | 00:40 | |
*** Pali has joined #maemo-ssu | 00:47 | |
*** andre__ has quit IRC | 00:57 | |
*** FIQ has quit IRC | 01:11 | |
*** 14WAAQFQ8 has joined #maemo-ssu | 01:23 | |
*** Jaded has quit IRC | 01:25 | |
*** 14WAAQFQ8 is now known as jade | 01:25 | |
*** jade has quit IRC | 01:25 | |
*** jade has joined #maemo-ssu | 01:25 | |
*** jade has quit IRC | 02:06 | |
*** Jade has joined #maemo-ssu | 02:07 | |
*** Jade has joined #maemo-ssu | 02:07 | |
*** FIQ has joined #maemo-ssu | 03:06 | |
*** FIQ has quit IRC | 03:14 | |
*** Jaded has joined #maemo-ssu | 03:19 | |
*** Jaded has joined #maemo-ssu | 03:19 | |
*** Jade has quit IRC | 03:20 | |
*** nox- has quit IRC | 03:30 | |
*** ruskie has quit IRC | 03:35 | |
*** rd_ has joined #maemo-ssu | 04:02 | |
*** rd_ is now known as Guest31513 | 04:02 | |
*** _rd has quit IRC | 04:03 | |
*** ruskie has joined #maemo-ssu | 04:08 | |
*** FIQ has joined #maemo-ssu | 04:09 | |
*** jonwil has joined #maemo-ssu | 04:19 | |
*** arcean has quit IRC | 04:23 | |
*** M4rtinK2 has quit IRC | 04:37 | |
Pali | merlin1991, I suggest to downgrade pulseaudio ABI version from 16 to 15 | 04:56 |
---|---|---|
Pali | see reason on http://talk.maemo.org/showthread.php?p=1149125#post1149125 | 04:56 |
Pali | maemo version 16 is incompatible with upstream version 16 | 04:57 |
Pali | upstream pulseaudio versions are compatible (e.g. version 26 is compatible with 16) | 04:57 |
Pali | so also maemo version 16 is compatible with upstream 15 | 04:58 |
Pali | above maemo incompatibility causing that I'm not able to use maemo pulseaudio server via network | 04:59 |
Pali | see also: http://www.mail-archive.com/pulseaudio-discuss@mail.0pointer.de/msg09900.html | 04:59 |
*** rd has joined #maemo-ssu | 05:00 | |
*** rd is now known as Guest15724 | 05:00 | |
*** Guest31513 has quit IRC | 05:02 | |
*** Pali has quit IRC | 05:03 | |
*** amiconn_ has joined #maemo-ssu | 05:17 | |
*** amiconn has quit IRC | 05:17 | |
*** jonwil_ has joined #maemo-ssu | 05:35 | |
*** jonwil has quit IRC | 05:36 | |
*** jonwil_ is now known as jonwil | 05:36 | |
*** jonwil_ has joined #maemo-ssu | 05:43 | |
*** jonwil has quit IRC | 05:44 | |
*** jonwil_ is now known as jonwil | 05:44 | |
*** kolp_ has joined #maemo-ssu | 05:46 | |
*** kolp has quit IRC | 05:49 | |
*** jonwil has quit IRC | 05:54 | |
DocScrutinizer05 | ohlol | 05:56 |
DocScrutinizer05 | hail polypaudio! | 05:56 |
*** DocScrutinizer05 has quit IRC | 06:04 | |
*** DocScrutinizer05 has joined #maemo-ssu | 06:04 | |
*** joshgillies has joined #maemo-ssu | 06:19 | |
*** dhbiker has quit IRC | 06:45 | |
*** dhbiker has joined #maemo-ssu | 07:05 | |
*** jonwil has joined #maemo-ssu | 07:13 | |
*** kolp_ has quit IRC | 08:01 | |
*** joshgillies has quit IRC | 08:26 | |
*** amiconn_ is now known as amiconn | 10:35 | |
*** Pali has joined #maemo-ssu | 11:15 | |
*** Guest15724 has quit IRC | 11:23 | |
*** Guest15724 has joined #maemo-ssu | 11:23 | |
*** Martix has joined #maemo-ssu | 12:24 | |
*** NIN101 has joined #maemo-ssu | 12:25 | |
*** arcean has joined #maemo-ssu | 12:35 | |
*** Guest15724 has quit IRC | 12:40 | |
*** Guest15724 has joined #maemo-ssu | 12:58 | |
*** M4rtinK2 has joined #maemo-ssu | 13:06 | |
*** joshgillies has joined #maemo-ssu | 13:22 | |
*** arcean has quit IRC | 13:46 | |
*** arcean has joined #maemo-ssu | 13:47 | |
jonwil | pali: ping | 14:30 |
Pali | jonwil, pong | 14:30 |
jonwil | how goes it? Done anything interesting recently? :) | 14:31 |
kerio | how recently? | 14:31 |
Pali | see log about pulseaudio :-) | 14:32 |
Pali | jonwil, and I wrote open source replacement for battery & profile selector menu applet | 14:32 |
Pali | both are in cssu-devel | 14:33 |
jonwil | nice | 14:33 |
Pali | profile selector which support all configured profiles (not only silent & general) | 14:33 |
Pali | jonwil and BME replacement is for testing | 14:33 |
jonwil | nice :) | 14:34 |
kerio | Pali: what's the link to the bme replacement again? | 14:35 |
Pali | https://gitorious.org/rx51-bme-replacement | 14:35 |
kerio | no, the precompiled .debs | 14:35 |
Pali | https://atrey.karlin.mff.cuni.cz/~pali/rx51-bme-replacement/ | 14:36 |
jonwil | The systemui work done by freemangordon is great too :) | 14:42 |
jonwil | If I had the time and inspiration, I would go back and work more on icd stuff or libcal or connectivity UI or whatever else | 14:46 |
jonwil | but I am too busy these days for N900 mostly | 14:46 |
*** kolp has joined #maemo-ssu | 14:52 | |
jonwil | so pali, is your bme stuff a proper drop-in replacement for Nokia BME stuff? | 14:57 |
Pali | should be | 14:57 |
Pali | but needs testing | 14:57 |
jonwil | does it actually make use of whatever BME info exists in CAL? | 14:57 |
jonwil | whatever the BME CAL data is? | 14:57 |
Pali | no, it does not use BME CAL | 14:57 |
jonwil | ok, what data is in BME cal anyway? | 14:58 |
Pali | I know that in BME CAL are constants for temperature formula | 14:58 |
Pali | seems that constants are same on all n900... | 14:58 |
Pali | so I hardcoded conversation formula for temperature into kernel driver | 14:58 |
Pali | jonwil, I do not know what other is stored in BME CAL | 14:59 |
jonwil | ok, so it contains correct temperature calculation formula? | 14:59 |
Pali | should be, at least it returns +/-1°C in range 0-30°C | 15:00 |
Pali | I think we have right formula for temperature | 15:00 |
Pali | jonwil, btw, freemangordon already RE libcal: https://gitorious.org/cal | 15:01 |
Pali | I tested that only in nandsim, and it worked | 15:01 |
Pali | jonwil, and now we have open source flasher | 15:02 |
Pali | 0xFFFF | 15:02 |
Pali | jonwil, see: http://talk.maemo.org/showthread.php?t=87996 | 15:02 |
jonwil | nice that we have 0xFFFF | 15:03 |
jonwil | except that it doesn't seem to do all things Nokia tools do :) | 15:03 |
Pali | jonwil, only mmc flashing & proper testing is missing | 15:04 |
jonwil | ok :) | 15:05 |
Pali | I rewritten 0xFFFF and added full fiasco support | 15:05 |
jonwil | great for libcal, one less binary blob needed for meego/mer/nemo :) | 15:06 |
kerio | Pali: hm, i'm not sure about the temp - i had it register 27° when there were definetely less than that | 15:10 |
kerio | (it's still precise enough) | 15:10 |
kerio | jonwil: the bme replacement works fine for me | 15:11 |
Pali | kerio, there is also some noise due to madc | 15:11 |
Pali | and in kernel is table of raw <-> °C values | 15:12 |
Pali | with precise 1°C | 15:12 |
jonwil | so you aren't using the math, you are using precomputed data tables? | 15:13 |
*** arcean has quit IRC | 15:25 | |
*** Jaded has quit IRC | 15:34 | |
*** Jade has joined #maemo-ssu | 15:35 | |
*** Jade has joined #maemo-ssu | 15:35 | |
*** joshgillies has quit IRC | 15:43 | |
*** dafox has joined #maemo-ssu | 16:07 | |
*** freemangordon has quit IRC | 16:28 | |
*** Pali has quit IRC | 16:43 | |
DocScrutinizer05 | kerio: bq27200 doesn't report ambient temp | 16:45 |
kerio | i'm talking about the temp reported by dsme-thermalobject-surface, not sure where it gets the data | 16:46 |
DocScrutinizer05 | what makes you think you know what that value should be? | 16:54 |
DocScrutinizer05 | do you know what's the surface of the thermalobject? | 16:55 |
*** arcean has joined #maemo-ssu | 16:56 | |
*** wumpwoast has quit IRC | 16:56 | |
DocScrutinizer05 | jonwil: not precomputed, predefined. Derived from 'lab'-test | 17:11 |
*** Pali has joined #maemo-ssu | 17:11 | |
jonwil | so this is better than using the actual math because...? | 17:11 |
Pali | jonwil, math library is not in kernel | 17:12 |
Pali | and I need log10 | 17:12 |
DocScrutinizer05 | jonwil: pali had the brilliant idea to read out raw and hald-addon-bme computed battery temp values | 17:12 |
jonwil | ok, so kernel has no log10 | 17:12 |
jonwil | makes sense | 17:12 |
DocScrutinizer05 | jonwil: and then I did a test series with my device in freezer | 17:13 |
DocScrutinizer05 | jonwil: after all it's less than 200 points in that table, so bytesize isn't that bad for kernel | 17:14 |
DocScrutinizer05 | and pali did a very clever splitting into two tables | 17:14 |
*** Jade has quit IRC | 17:15 | |
jonwil | ok, great | 17:16 |
*** Jade has joined #maemo-ssu | 17:17 | |
*** Jade has joined #maemo-ssu | 17:17 | |
Pali | merlin1991, see: https://gitorious.org/community-ssu/pulseaudio/merge_requests/1 | 17:33 |
Pali | also tell about this to luf too | 17:33 |
*** dafox has quit IRC | 17:41 | |
kerio | Pali: oh, version 15 so that newer versions of pulseaudio will use the older, working protocol when talking to our pulseaudio? | 17:43 |
Pali | yes | 17:43 |
Pali | I want this: connect my bluetooth headset to n900 and also to notebook. But connecting a2dp to two devices are not possible... So solution can be to connect n900 and notebook audio | 17:45 |
Pali | I tried to use bluetooth a2dp for n900 <-> notebook connection, but it was very slow when I connected also bluetooth headset to n900 | 17:46 |
Pali | so I'm trying to use wifi for that | 17:46 |
Pali | and easy way can be to use directly pulseaudio becuase it it network audio server | 17:46 |
Pali | but nokia crippled pulseaudio in n900 and network connection is not possible :-( | 17:47 |
Pali | kerio, do you have any other idea how to connect bluetooth headset to both devices? | 17:48 |
kerio | Pali: bluetooth audio sink on your laptop | 17:49 |
kerio | make the n900 use it as the bluetooth headset | 17:49 |
kerio | then make your laptop use your headphones | 17:49 |
Pali | kerio and there is other problem. my notebook has bad bluetooth antenna. n900 has better | 17:50 |
kerio | lol wat | 17:50 |
kerio | my n900 has an awful bluetooth antenna :( | 17:50 |
Pali | so I want to use n900 bluetooth for higher range | 17:50 |
kerio | Pali: keep the n900 close to your laptop :) | 17:50 |
kerio | oh, i see | 17:50 |
Pali | not possible, because I want to go to other room with connected bluetooth headset | 17:51 |
kerio | use the esound network protocol, i heard it works | 17:51 |
Pali | but n900 can be close to notebook | 17:51 |
kerio | and, of course, use it through usb networking | 17:51 |
Pali | kerio, yes I tried it | 17:51 |
Pali | but problem is that esound has only sink | 17:51 |
Pali | I need both sink & source | 17:51 |
kerio | you only need the sink, don't you? | 17:52 |
kerio | the sink on the n900 | 17:52 |
kerio | and then let it output normally to your BT headset | 17:52 |
Pali | and problem with esound is that there is 1s lattency | 17:53 |
Pali | no idea why | 17:53 |
Pali | pulseaudio has support for synchronization | 17:54 |
Pali | and next problem with esound is that autosuspend module not working... | 17:54 |
kerio | well, fix pulseaudio :) | 17:54 |
*** Jade has quit IRC | 17:55 | |
Pali | I'm thinking about other solution: stream audio via UDP with some synchronize protocol | 17:55 |
Pali | maybe with some realtime compression | 17:55 |
Pali | but I do not know which codec is good for realtime stream... | 17:56 |
kerio | Pali: you need pc-n900 | 17:56 |
kerio | over usb networking, you won't need compression | 17:56 |
Pali | maybe not | 17:57 |
*** Jade has joined #maemo-ssu | 17:57 | |
*** Jade has joined #maemo-ssu | 17:57 | |
Pali | but I preffer wifi as wireless solution | 17:57 |
kerio | wifi *and* bluetooth will work like absolute crap | 17:57 |
kerio | seriously | 17:58 |
Skry | Pali: how about using rtp (which usually uses udp) via gstreamer or ffmpeg | 18:03 |
Pali | Skry, do you have some howto? | 18:04 |
Skry | with gstreamer this could be a pipeline audio source -> encode with dsp -> stream with rtp, with ffmpeg it's a one-liner | 18:05 |
Skry | can't remember exact syntax but something like this ffmpeg -f alsa -i hw0:0 -acodec libmp3lame -ab 128k -ac 2 -re -f rtp rtp://ip:port | 18:08 |
*** dhbiker has quit IRC | 18:33 | |
DocScrutinizer05 | really clever headsets can pair two 2 hosts concurrently | 18:34 |
DocScrutinizer05 | via A2DP | 18:34 |
RST38h | And play each into a separate ear! | 18:35 |
DocScrutinizer05 | hahahaaa | 18:35 |
DocScrutinizer05 | I guess they either mix, or mute the input with lower prio when higher prio input starts playback | 18:35 |
DocScrutinizer05 | honestly, use *wired* headset with N900, then you can pair the N900 to arbitrary A2DP as virtual BT headset | 18:36 |
*** jonwil has quit IRC | 18:41 | |
*** Guest15724 has quit IRC | 18:41 | |
kerio | DocScrutinizer05: either g_ether sucks, or socat sucks, or the n900 sucks | 18:42 |
kerio | :( | 18:42 |
DocScrutinizer05 | or kerio sucks? ;-P | 18:43 |
DocScrutinizer05 | kerio: what's the problem? | 18:44 |
*** rd has joined #maemo-ssu | 18:45 | |
*** rd is now known as Guest48568 | 18:45 | |
kerio | DocScrutinizer05: i can only get to 10MB/s via usb networking | 18:48 |
kerio | socat tcp-listen:31337 /dev/null on one side, socat /dev/zero tcp-connect:ip:31337 on the other | 18:50 |
DocScrutinizer05 | hmm | 18:51 |
DocScrutinizer05 | ~10 * 8 | 18:51 |
infobot | 80 | 18:51 |
DocScrutinizer05 | overhead of factor 5, easily possible depending on blocksize of transmitted pkgs | 18:52 |
DocScrutinizer05 | USB2 maxes out at 400MBit/s | 18:53 |
Pali | ffmpeg or libav not working... | 18:53 |
kerio | *480, in theory | 18:53 |
Pali | no sound over wifi | 18:53 |
DocScrutinizer05 | aah, ok, 480 | 18:53 |
kerio | ok, i'm going to try with g_nokia | 18:53 |
Pali | only on localhost is ffmpeg working | 18:53 |
DocScrutinizer05 | kerio: btw you're using TCP aiui, you know this is a connection-oriented transport, with ACK and all | 18:54 |
kerio | hm, still maxes out at 10MB/s | 18:55 |
kerio | with nc, now | 18:55 |
*** dhbiker has joined #maemo-ssu | 18:56 | |
DocScrutinizer05 | netcat and socat are basically identical | 18:57 |
DocScrutinizer05 | you're still sending packages of unknown size via whole TCP stack, with complete protocol | 18:57 |
DocScrutinizer05 | use UDP for a start | 18:58 |
DocScrutinizer05 | use dd to pipe blocks of known size to socat | 18:58 |
DocScrutinizer05 | (or find the right socat parameters to define blocksize and no timeout) | 18:59 |
DocScrutinizer05 | ethernet is NOT a rs232 cable | 18:59 |
kerio | hah, second time OS X panic'd when doing mass USB transfers | 19:00 |
kerio | silly me expecting a BSD system to be good at networking | 19:00 |
RST38h | Sadly, it is not. | 19:00 |
kerio | good at networking, or a BSD system? | 19:01 |
DocScrutinizer05 | kerio: (overhead) you know each ethernet TCP packet has a header with MAC and port-addr, a IP origin and destination, a size-of-payload value, then a leadout with checksum, whatnot else | 19:02 |
kerio | hm | 19:02 |
DocScrutinizer05 | ethernet is NOT rs232 | 19:03 |
*** Guest48568 has quit IRC | 19:03 | |
kerio | now i'm wondering if my router has the appropriate rndis ethernet driver | 19:03 |
kerio | and i'm also wondering why libicd-network-usb doesn't take down the usb0 interface properly | 19:03 |
kerio | oh well | 19:03 |
DocScrutinizer05 | for packet sizes of 16kBytes, overhead is almost negligible. For transfering single chars as ethernet packages, it's insane | 19:04 |
RST38h | AppleTalk was however. | 19:04 |
*** Jaded has joined #maemo-ssu | 19:05 | |
*** Jaded has joined #maemo-ssu | 19:05 | |
DocScrutinizer05 | kerio: so basically you're probing some random stuff with your socat setup | 19:05 |
kerio | there should be a way to specify the packet size, right? | 19:05 |
kerio | in theory, at least | 19:05 |
DocScrutinizer05 | the result is meaningless, unless you understand what's really going on | 19:05 |
DocScrutinizer05 | yes, there is indeed a plethora of ways to handle package size | 19:06 |
DocScrutinizer05 | e.g for ssh input, you can define timeouts after which package gets sent anyway, and also you can define termination chars which make package get sent, like CR/NL | 19:06 |
*** Jade has quit IRC | 19:07 | |
DocScrutinizer05 | too large chunks are segmented to max transmission size of channel by TCP stack | 19:08 |
DocScrutinizer05 | it's not exactly trivial | 19:08 |
DocScrutinizer05 | see ifconfig|grep MTU | 19:09 |
DocScrutinizer05 | and a lot of other parameters | 19:09 |
DocScrutinizer05 | man socat: >> | 19:10 |
DocScrutinizer05 | -b<size> | 19:10 |
DocScrutinizer05 | Sets the data transfer block <size> [size_t]. At most <size> bytes are transferred per step. Default is 8192 bytes. | 19:10 |
DocScrutinizer05 | but now you're at your own, since I'm afk | 19:11 |
DocScrutinizer05 | anyway, read "man socat", and when you're done with that and *understood* all of it, you're a step further to proper testing of g_ether (and actually a step further than I am) | 19:18 |
kerio | ok, i've gotten 20MB/s | 19:20 |
kerio | telling socat to use 32k or 64k as the block size | 19:21 |
kerio | actually no, it's irrelevant | 19:21 |
kerio | apparently something is seriously weird | 19:22 |
kerio | n900->pc is 20mb/s | 19:22 |
kerio | pc->n900 is 10mb/s | 19:22 |
kerio | i'll try with a different pc, so we can rule out os x's funkyness | 19:22 |
DocScrutinizer05 | kerio: WTF are you trying to proove? | 19:23 |
kerio | i want to see how much data i can push through usb networking | 19:23 |
kerio | how much data over time, i mean | 19:23 |
DocScrutinizer05 | that's an incomplete question, unless you define the means you use to transfer that data as a part of the answer to find | 19:24 |
DocScrutinizer05 | obviously the amount of data/time is massively depending on the transport and tools and parameters you use | 19:25 |
RST38h | Doc: Sorry to interrupt this lively conversation, but he is on my /ignore list, if it tells you anything | 19:25 |
DocScrutinizer05 | and the transport, tools, parameters define your error-recovery you get, and a number of other not so irrelevant properties | 19:26 |
kerio | ...wtf, what did i do? | 19:26 |
DocScrutinizer05 | RST38h: hehe | 19:26 |
*** FIQ has quit IRC | 19:27 | |
*** FIQ has joined #maemo-ssu | 19:27 | |
kerio | DocScrutinizer05: anyway, this started just as a way to test libicd-network-usb | 19:31 |
DocScrutinizer05 | mhm | 19:31 |
kerio | it works fine, apart for the fact that the usb0 interface is kept up even after you disconnect | 19:32 |
DocScrutinizer05 | hmm, it implements a virtual NIC I guess | 19:33 |
DocScrutinizer05 | you have to modprobe -r to remove that NIC, I guess | 19:33 |
DocScrutinizer05 | best indication that you got a virtual NIC is the fact it's using a MAC addr | 19:34 |
DocScrutinizer05 | basically for USB networking MAC are rather meaningless or at least not needed, but since any NIC has a MAC, the USB-networking also has one (actually one per end) | 19:36 |
*** FIQ has quit IRC | 19:36 | |
kerio | it does, but i'd still expect the disconnect to do the equivalent of ifconfig usb0 down | 19:36 |
DocScrutinizer05 | why should it? | 19:36 |
kerio | because it does the same for every other interface | 19:38 |
kerio | wlan0 goes down, gprs0 goes down | 19:38 |
DocScrutinizer05 | send patches ;-) | 19:41 |
DocScrutinizer05 | btw I don't think wlan0 goes "state:down" just for broken RF connection. It needs action from ICD to make WLAN IF go down (and that seems to fail for me sometimes, which results in WLAN still up for AP that's already miles away) | 19:43 |
DocScrutinizer05 | once or twice I might've encountered same for GPRS roaming | 19:44 |
DocScrutinizer05 | wasn't exactly easy to investigate | 19:45 |
*** rd has joined #maemo-ssu | 20:12 | |
*** rd is now known as Guest25449 | 20:12 | |
*** FIQ has joined #maemo-ssu | 20:29 | |
*** Guest25449 is now known as _rd | 21:00 | |
Pali | kerio, due to that musb working only if some gadget is loaded (no idea why) and charger detection is in musb, some gadget driver is still loaded | 21:20 |
Pali | so when you disconnect usb cable (from nokia pc suite mode), g_nokia is still loaded | 21:21 |
kerio | Pali: huh? | 21:21 |
Pali | and this is reason why usb0 is still connected | 21:21 |
kerio | the phone detects the charging even without a gadget module i think | 21:21 |
Pali | no | 21:21 |
Pali | you can try it | 21:21 |
kerio | Pali: connected and up are different things | 21:21 |
Pali | also for usb host mode some gadget driver must be loaded | 21:22 |
kerio | cat /sys/class/power_supply/bq24150a-0/mode -> "auto (host)" | 21:22 |
kerio | lsmod | grep g_ -> "" | 21:22 |
Pali | connect wallcharger | 21:23 |
kerio | oh, it'll stay as host? | 21:23 |
kerio | i see | 21:23 |
Pali | btw, g_nokia is composite gadget driver | 21:24 |
Pali | what about adding also file storage component to nokia gadget? | 21:24 |
kerio | just don't make it necessarily start softupd and phonet-at and i'm cool with it | 21:24 |
Pali | and we do not need to switch between mass storage & pc suite | 21:24 |
Pali | problem is that mass storage kernel code is modular only in upstream kernel | 21:25 |
Pali | so we need some 3.x for that | 21:25 |
kerio | can't you just "magically" merge g_file_storage and g_nokia? | 21:25 |
Pali | and g_file_storage (non composite) was renamed to g_mass_storage (comosite) | 21:26 |
kerio | i see | 21:26 |
Pali | you cannot merge non composite to composite driver | 21:26 |
kerio | so you'd need to backport g_mass_storage at the very least | 21:26 |
Pali | yes | 21:26 |
kerio | hmm, look into it and see if you feel like doing it | 21:26 |
kerio | (if it's even possible at all ofc) | 21:27 |
Pali | but there can be other problem.... there are a lot of parts which loading g_nokia or g_file_storage driver | 21:27 |
Pali | then we need to patch every script which loading gadget driver | 21:27 |
kerio | the UI is a different problem i suppose | 21:27 |
Pali | yes that stupid usb dialog... | 21:29 |
*** _rd has quit IRC | 21:31 | |
DocScrutinizer05 | musb_core eats some 60mA when not in suspend | 21:37 |
DocScrutinizer05 | that's why you want to wakeup musb_core when VBUS gets detected, so you can talk to 1707 PHY via ULPI interface, to detect D+- short | 21:38 |
DocScrutinizer05 | then the latest when VBUS gets removed, you might want to suspend musb_core to save those 60mA it eats | 21:38 |
DocScrutinizer05 | I hope this might shed some light on the problems we face with USB and musb and loading kernel modules and charger detection and whatnot | 21:39 |
*** Sicelo900 has joined #maemo-ssu | 21:43 | |
*** Sicelo900 has quit IRC | 22:14 | |
*** zogg__ has quit IRC | 22:19 | |
*** Sicelo900 has joined #maemo-ssu | 22:20 | |
*** Sicelo900 has joined #maemo-ssu | 22:20 | |
*** luf has joined #maemo-ssu | 22:30 | |
luf | Pali: ping | 22:30 |
*** Sicelo900 has quit IRC | 22:33 | |
Pali | luf, pong | 22:33 |
luf | The merge request seems ok. BTW do we know nothing else needs the cripled ABI? | 22:34 |
Pali | no idea | 22:34 |
Pali | needs testing | 22:34 |
luf | I have to merge it using my local computer or should I just click somewhere? | 22:35 |
*** arcean has quit IRC | 22:39 | |
Pali | luf, git pull from my repo & git push to cssu | 22:40 |
Pali | but I can do it too | 22:40 |
Pali | every cssu member can push to cssu repos | 22:40 |
luf | I know. | 22:40 |
luf | Feel free to push it. I ack it. | 22:41 |
kerio | Pali: going back to the gadget module thing... technically only things we control should unload/load those modules | 22:51 |
kerio | the UI when you do mass storage/pc suite - really some scripts, /etc/event.d/bme, possibly a usb host mode script or two | 22:52 |
kerio | Pali: a grep shows that the string "g_file_storage" only appears in a handful of files | 22:58 |
kerio | osso-usb-mass-storage-{en|dis}able.sh, pcsuite-enable.sh, bme, softupd, testserver (wtf is this?), ke-recv | 22:59 |
kerio | also, wtf is g_softupd? and g_zero? | 23:00 |
Pali | osso-usb-* pcsuite* is in ke-recv | 23:01 |
Pali | bme is going to be replaced | 23:01 |
Pali | softupd is server application for flashing | 23:01 |
Pali | (flasher on n900 is connecting to softupd and sending kernel image to softupd) | 23:02 |
kerio | Pali: ooh, g_zero is what we should be really using when in no mode i think | 23:02 |
Pali | softupd is also responsible for emmc flashing | 23:02 |
kerio | Pali: softupd is also launched as a daemon when in pc suite mode | 23:02 |
kerio | i think to do pc suite stuff | 23:02 |
kerio | to do stuff with pcsuite/ovisuite i mean | 23:02 |
Pali | yes, because in pc suite mode, you can flash kernel/mmc | 23:02 |
Pali | also nolo, xloader & cmt | 23:03 |
Pali | via Mk II protocol | 23:03 |
kerio | Pali: that makes no sense really | 23:03 |
kerio | it's a live system | 23:03 |
Pali | rootfs is not possible | 23:03 |
Pali | but why kernel, nolo & xloader? | 23:03 |
Pali | because it does not break anything running | 23:03 |
kerio | hm, g_zero doesn't even negotiate for power actually | 23:03 |
Pali | testserver is propably server application for phoenix testing & cal access | 23:03 |
Pali | it is started only in TEST bootmode | 23:04 |
Pali | so never on normal n900 | 23:04 |
kerio | so i can remove it? | 23:04 |
Pali | yes | 23:04 |
Pali | g_softupd is gadget for emmc flashing via softupd | 23:05 |
Pali | g_nokia has also g_softupd component | 23:05 |
Pali | I bet that nokia wanted only zero copy driver without usb networking (which is in g_nokia) so they created stupid g_softupd | 23:05 |
Pali | g_zero is zero gadget driver which doing nothing | 23:06 |
Pali | kerio, testserver is good candidate which will be removed from some possible CSSU fiasco image | 23:06 |
Pali | it only store space on rootfs | 23:06 |
Pali | and 99,99% users and non developers did not used it | 23:07 |
Pali | s/non/non nokia/ | 23:07 |
infobot | Pali meant: and 99,99% users and non nokia developers did not used it | 23:07 |
kerio | how do you even enter the TEST bootmode? | 23:07 |
kerio | DocScrutinizer05: what good is a usb gadget module that doesn't even negotiate the maximum current? :s | 23:08 |
Pali | kerio, legal way only via phoenix | 23:10 |
kerio | is it a special NOLO thing? | 23:10 |
Pali | kerio, other way is to patch 0xFFFF | 23:10 |
Pali | I think | 23:10 |
Pali | or patching U-Boot | 23:10 |
kerio | i was going to say that :) | 23:10 |
Pali | you need to add test into omap atag bootreason | 23:11 |
Pali | btw, last uboot has variable where can be sotred bootreson and bootmode | 23:11 |
Pali | there is NOLO command for booting kernel (used when you want to exit flashing without disconnecting cable) | 23:11 |
Pali | and also used if you want to boot loaded kernel via USB | 23:12 |
Pali | and if you change some param from 0 to 1, then it boot into UPDATE mode which is used for flashing eMMC | 23:12 |
Pali | so I think if I change 0 to 2 or 3 it can enter to LOCAL or TEST mode :-) | 23:12 |
Pali | kerio, or by patching getbootstate | 23:13 |
Pali | LOCAL & TEST mode is selected from BSI... | 23:13 |
*** arcean_ has joined #maemo-ssu | 23:14 | |
DocScrutinizer05 | kerio: sorry, I don't know how to answer | 23:20 |
DocScrutinizer05 | ((<kerio> how do you even enter the TEST bootmode?)) afaik with a special TEST 'battery' that has a certain special BSI value | 23:24 |
kerio | i wonder why nokia left all this cruft around, considering that they were trying to save as much space as they could on the rootfs | 23:25 |
DocScrutinizer05 | well, the knowhow became sparse and spread all over the team, nobody really had a comprehensive overview | 23:26 |
Pali | DocScrutinizer05, getbootstate enter to test mode if there is special BSI | 23:26 |
Pali | you are right | 23:27 |
DocScrutinizer05 | I know ;-) | 23:27 |
DocScrutinizer05 | I know about test battery since ~2 years | 23:27 |
Pali | https://bazaar.launchpad.net/~pali/+junk/maemo_getbootstate/view/head:/getbootstate.c#L494 | 23:27 |
Pali | RE getbootstate ^^^^ | 23:27 |
Pali | DocScrutinizer05, where did you found test battery?? | 23:28 |
DocScrutinizer05 | and there's an explicit test for that somewhere in preinit or sth | 23:28 |
Pali | it it in getbootstate, see source :-) | 23:28 |
DocScrutinizer05 | iirc stskeeps pointed me to some details about BSI | 23:28 |
Pali | hm, do you know something more about BSI? | 23:29 |
DocScrutinizer05 | yep, the test in preinit is using getbootstate | 23:29 |
Pali | so is test mode really used in phoenix? | 23:29 |
DocScrutinizer05 | (more BSI) alas nope | 23:29 |
DocScrutinizer05 | it's used in phoenix with service fixture | 23:30 |
Pali | from phoenix GUI which I saw, I think that phoenix can write to CAL only from test mode | 23:30 |
Pali | there is button to boot into normal, test or local mode... | 23:30 |
DocScrutinizer05 | plausible | 23:30 |
DocScrutinizer05 | yep, that changes BSI resistor value | 23:31 |
DocScrutinizer05 | afaik | 23:31 |
Pali | and only application which doing something with libcal is testserver | 23:31 |
Pali | and testserver event.d script has line start in testmode only | 23:31 |
Pali | so then it make sense | 23:31 |
DocScrutinizer05 | o/ bbl ->dinner | 23:32 |
DocScrutinizer05 | ooh one thing: I'm quite sure NOLO also checks for BSI TEST value, and possibly behaves differently when TEST battery detected | 23:36 |
DocScrutinizer05 | and even cmt does, afaik | 23:36 |
DocScrutinizer05 | it seems Nokia is using this scheme on a lot of their phones | 23:37 |
DocScrutinizer05 | after all phoenix isn't N900-specific | 23:37 |
*** NIN101 has quit IRC | 23:40 | |
*** nox- has joined #maemo-ssu | 23:47 | |
Pali | DocScrutinizer05, but NOLO and fiasco yes | 23:53 |
*** joshgillies has joined #maemo-ssu | 23:53 | |
*** joshgillies has quit IRC | 23:58 | |
*** joshgillies has joined #maemo-ssu | 23:58 |
Generated by irclog2html.py 2.15.1 by Marius Gedminas - find it at mg.pov.lt!