| 09:58 | Sleep_Walker | anyone working on HTC Desire Z support? |
| 15:55 | Auwia | anyone for htc mega? |
| 19:05 | detule | hey WisTilt2 |
| 19:09 | detule | to answer your question from yesterday, 3.0 is missing the timer commit, and apparently still has the usb_cable_status business in pm.c (i'll update these when i get home)....in 3.1 I switched to pm.c that is pretty much stock code-aurora (via emwe's tree) for testing purposes....it seems pretty flawless so far, any idle related glitches are gone, and it seems to still go into idle PC, if proc/msm_pm_stats is to be believed.... |
| 19:10 | detule | and yes, gbx0b has the updated camera lib and also incorporates a hack to leave mic bias on |
| 19:14 | WisTilt2 | hi detule: that answers my next question about the timer. i did notice that 3.0 has some weird quirks also. sometimes in sleep it wont wake up with power button unless you hit it a dozen times or so. .39 is rock solid so you might want to take some of those commits over to 3.x eventually. Also, getting much more battery drain on both 3.x's than .39 |
| 19:14 | detule | WisTilt2, 3.0 still has scbs enabled |
| 19:14 | detule | that might have something to do with the wake quirks |
| 19:15 | WisTilt2 | scbs isnt running on the public rootfs though but yes i think you're right about it causing the wake probs. they dont happen on .39 though with all the other things ive tweaked. |
| 19:16 | detule | yeah the daemon isn't there but there's still the suspend part "do_scbs" that runs without the daemon i think... |
| 19:16 | detule | someone needs to get a hold of jonpry to tell him 3.2 is out |
| 19:18 | WisTilt2 | in my PM on .39 i changed scbs_do so its not called when its an idle collapse also since it was preventing idle from actually happening. |
| 19:18 | WisTilt2 | 3.2 change much from 3.1? |
| 19:19 | WisTilt2 | if proc/msm_pm_stats is to be believed.... <- they report each condition pretty accurately from what i tested |
| 19:21 | detule | can you point me to the scbs change on .39 |
| 19:22 | WisTilt2 | http://pastebin.com/z7apHsKU |
| 19:23 | detule | oh yeah i saw that thanks |
| 19:24 | detule | so these panel changes are improving batter life you are saying? |
| 19:24 | WisTilt2 | thats my commit # b3c930e981cb14a99ec6963bbd2b9a67e524e236 |
| 19:26 | detule | need to query jonpry what happens if scbs can't poll in idle, i was under the impression that this polling interval has to be kept constant at all costs... |
| 19:26 | WisTilt2 | the panel changes i made were just commenting out all the vreg enable/disable stuff. the gpu0/1 are ok power wise but those new ones for gsm cause nasty battery drain in sleep |
| 19:27 | WisTilt2 | scbs still polls in idle, it was redundant calling it inside of idle collapse/suspend only and caused idle to exit from any power saving is all. |
| 19:27 | WisTilt2 | it polls from the timer still i mean |
| 19:28 | detule | those gpu vregs that's fine for all rhods? wasn't there some issue with data, or were those different registers |
| 19:29 | WisTilt2 | the vregs from [acl] are only for gsm and the gp2/4 used to be ok on all rhods but emwe said something about those killing data on cdma so i just disabled all of them. no big diff in battery without gp2/4 at all that i can see. |
| 19:30 | WisTilt2 | without scbs on .39 im consistently seeing .58% hr in sleep with sync off and radio still on. with scbs its around 1% so thats pretty darn good. |
| 19:30 | WisTilt2 | thats without any of the vregs being touched also |
| 19:33 | WisTilt2 | detule on other thing i noticed in this new GB is netloc is broke again. did the ril get changed? |
| 19:33 | detule | no it's something to do with the gapps that's bundled |
| 19:33 | rpierce99 | no there's an updated userland apk |
| 19:34 | detule | http://forum.xda-developers.com/showpost.php?p=21053860&postcount=63 |
| 19:34 | WisTilt2 | guess they'll get that corrected apk added in the next release then. no biggie, just glad its not something else we need to track down. |
| 19:36 | WisTilt2 | another fix thats needed is that issue when you get a call but dont answer it. from that point on the rear speaker stays on with static and all audio is gone until reboot. not sure if thats userland logic or something else in acoustic, haven't looked into it yet. |
| 19:36 | rpierce99 | ah so that's the trigger, couldn't put my finger on what caused that |
| 19:36 | rpierce99 | had to reboot to fix it |
| 19:38 | WisTilt2 | yeah looks like when ringtone plays the logic doesn't turn off speaker after and the sound state never gets touched again. im blaming it on userland for now. |
| 19:44 | rpierce99 | hm, i can't get mine to open up the rear speaker to ring at all right now, just vibrates and ringer volume is all the way up but won't play the beeps either |
| 19:44 | detule | alright WisTilt2 from what i see in the .39 commits you currently have all vregs enable/disable commented out, and i guess this would only affect gsm rhods from where rhodium-panel.c is in 3.1 |
| 19:47 | detule | good call on moving scbs in the if(!from_idle) bracket |
| 19:53 | WisTilt2 | rpierce99 which kernel is that happening on? |
| 19:53 | rpierce99 | i'm on the bundled .27 ATM, but it's not normal, i'm sure it would work after a reboot |
| 19:54 | detule | anything in logcat |
| 19:55 | rpierce99 | fails at life |
| 19:55 | rpierce99 | already rebooted |
| 19:55 | WisTilt2 | lol. .27, well there's your problem:) |
| 19:56 | rpierce99 | yeah, i figured i should test it as bundled to see issues that others were reporting |
| 19:59 | rpierce99 | pm, there are "wince" entries in logcat this time |
| 20:00 | detule | there should always be wince entries in logcat |
| 20:00 | detule | so this is when the issue occured? |
| 20:00 | rpierce99 | and i do have static in the back even though the ringer didn't go off |
| 20:00 | rpierce99 | only vibed |
| 20:00 | rpierce99 | yep, booted it up, called it, it didn't ring but the back speaker is left open |
| 20:01 | detule | 01-06 12:56:44.170 W/AudioTrack( 1269): obtainBuffer timed out (is the CPU pegged?) 0x3a328 user=000023b6, server=00000f4e |
| 20:01 | detule | someone needs to squash that |
| 20:02 | detule | it's left open i think because it keeps trying to write to it |
| 20:06 | rpierce99 | so it's definitely a state the hardware gets into, not a software bug, because kexec doesn't fix the problem, i have to reboot so winmo can reset the hw state |
| 20:16 | WisTilt2 | rpierce99 that's not good. Sounds like we're not initing everything we should be if winmo has to do some of it. |
| 20:18 | detule | is it too much to ask in the new year |
| 20:18 | detule | to only see dmesg output from the modified dmesg binary that displays proper timestamps |
| 20:18 | rpierce99 | get arrrghhh to change the bundled dmesg |
| 20:20 | detule | i am not sure if it's because this is .27 but I am not used to seeing this [ 110.444396] SMD: ch 7 CLOSED -> CLOSING |
| 20:20 | detule | [ 110.444488] SMD: ch 7 CLOSING -> OPENING |
| 20:20 | detule | [ 110.444549] SMD: ch 7 OPENING -> OPENED |
| 20:20 | detule | [ 122.681091] audmgr_enable: ARM9 did not reply to RPC am->state = 2 |
| 20:20 | detule | [ 122.681152] audmgr: unexpected state 2 while enabling?! |
| 20:20 | rpierce99 | i can try again on .39 or one of the other kernels |
| 20:23 | detule | if it's something you think you can reproduce |
| 20:23 | rpierce99 | apparently you just have to miss a call |
| 20:23 | rpierce99 | we'll see i guess |
| 20:27 | mgross029 | WisTilt2: was able get scbs fully set up, but when I went to analyze logs it completed normally but still says not generated, so I'll have to give it another try. Battery usage is much better with scbs though. |
| 20:28 | detule | i don't get how battery usage can be better |
| 20:31 | rpierce99 | usage is not better, meter accuracy is better |
| 20:31 | rpierce99 | and when meter accuracy is better, charging accuracy is better |
| 20:32 | mgross029 | detule: WisTilt2: had been working on a .39 kernel with better power consumption which I was testing. Without scbs the battery meter was all over the place. Once I enable scbs the meter was more consistant, so as rpierce99 just stated maybe that is why it appeared better to me. |
| 20:32 | detule | i am just saying battery usage can hardly be better |
| 20:34 | mgross029 | I really had to run a lot of apps to get the battery done to 30% so I could create the battery model yesterday though. |
| 20:35 | mgross029 | I ran waze (GPS app) for over an hour without connecting to the charger and it only dropped about form 84% to 60%. |
| 20:36 | mgross029 | Normally it would have appeared to drain the battery to 30% or less. |
| 20:37 | WisTilt2 | mgross029: did it take some time to do the calcs on creating the model and gave you the finished screen with all the completed values screen that you click "use these" on? that should have created a scbs.conf file with those values. |
| 20:38 | mgross029 | WisTilt2: Yep I clicked that and it must have not taken it... It dropped back to the first screen and Battery Model still said not created. |
| 20:39 | WisTilt2 | might have to start with a preset scbs.conf file im not sure on that. let me upload mine so you have a starting point. |
| 20:39 | WisTilt2 | or is that file present in your sdcard root now? |
| 20:40 | detule | i keep wanting to go back to manekineko zoombar patch |
| 20:40 | detule | s/manekineko/manekineko's/ |
| 20:40 | apt | detule meant: i keep wanting to go back to manekineko's zoombar patch |
| 21:11 | detule | rpierce99, tried it with the other kernels? |
| 21:15 | rpierce99 | trying to get the speaker to freak out, but on initial boot im still not getting any ringing, just vibe |
| 21:17 | rpierce99 | my tweeters play fine though, hm |
| 21:20 | detule | you still get that audmgr_enable: ARM9 did not reply to RPC am->state = 2 in dmesg on booy |
| 21:20 | detule | *t |
| 21:21 | rpierce99 | i don't have that in my dmesg from this last boot, no |
| 21:21 | rpierce99 | [ 160.320037] Sending arm9_low_speed 2 |
| 21:21 | detule | yeah that's normal |
| 21:38 | detule | oh wow getevent is pretty cool |