Monday, 2019-11-11

rinigusMister_Magister : yes, i did. But I'm not sure it's related. It's just on my device boot + dtbo images go together, as far as i understand.05:48
pketo@Lukapanio added07:06
T42<Lukapanio> Thanks07:47
spiiroinrinigus: I've been staring at autosleep.c. If I grok the logic, suspend_work / try_to_suspend() is getting scheduled all the time while autosleep is enabled. try_to_suspend() already has rate limiting logic, but it is executed only conditionally08:10
spiiroinand further, your patch in pm_suspends() = makes about the same thing happen at lower level, i.e. kind of modifies the conditions when that rate limiting occurs08:11
spiiroinone difference is that patching pm_suspend() does the uninterrupted sleep while holding mutex at try_to_suspend() level08:12
spiiroinand I'm wondering if it would be better/worse to tune conditions at try_to_suspend() e.g. perhaps take return value of pm_suspend() into account rather than sleeping at pm_suspend()?08:14
spiiroinbut that is just speculation, and great that you found a spot that does address the issue08:15
T42<景终磊 %lastname%> (Photo, 983x551) 电报群微信群炸群推广 2、twitter、facebook增粉,非小号mytoken关注排名热度 3.国内外KYC身份材料出售  4、精准币圈菠菜棋牌粉/电报群/微信群拉币圈活人 5、国内外交易所APP实名注册认证 6、电报群客服活聊天  7、批量注册国内外交易所实名账户,,出twitter、facebook、微信带粉账号 9.pot10:29
T42<RK2019account> @景终磊 %lastname% [<reply to media>], AD!😡😡😡10:29
rinigusspiiroin: reading autosleep.c/try_to_suspend(), I am surprised how badly its written. it doesn't check at all for error in return values and, in my case, it was happily avoiding that extra delay which its expected to introduce.10:58
riniguswhen I tried yesterday, I can get it working if I use error codes on pm_suspend return. in that case, I was getting into extra delay section. as a result, I had suspend success rate 50%11:00
rinigusso, I presume that if I add the same delay unconditionally into try_to_suspend, it will work as well. maybe its better location for it, as mutex is free11:00
rinigusspiiroin: will move it there tonight11:01
Mister_Magisterrinigus: ping11:01
Mister_Magisteru there11:01
Mister_Magisterrinigus: when i add android and spl version even through mkbootimg it breaks my bootloader11:01
rinigusMister_Magister: did it work with your hack before?11:02
rinigus... patching binary directly?11:03
rinigus... or its some other device?11:04
Mister_Magisterwhen i patched it maunally it acted same11:05
Mister_Magisteri thought if i do it properly it won't break but well11:05
rinigusMister_Magister: so, its some other device that never worked before with sfos?11:06
Mister_Magisterit's same device11:06
Mister_Magisterbut yes11:06
rinigusMister_Magister: how does it break? something like your device is corrupted?11:08
Mister_Magisterfreezes bootloader11:08
rinigusMister_Magister: when you flash aosp or lineage, does it work?11:09
rinigusat least on xz2, we have dtbo and vbmeta partitions. maybe you have the same and would have to flash them accordingly.11:14
Mister_Magisterrinigus: everything works except sfos11:16
rinigusMister_Magister: well, that's good. at least there is hope :). dtbo/vbmeta on android?11:17
rinigushow do flash it? I presume, boot and some other partitions11:19
rinigusamong those other partitions, do you have vbmeta and/or dtbo?11:19
Mister_Magisterwhat are you asking even11:19
riniguslet me find an example11:19
Mister_Magisteri have vbmeta so?11:19
Mister_Magisteri don't flash vbmeta to flash twrp11:20
rinigusre twrp, I don't know. I'd suggest to build vbmeta image, on aosp/lineage and flash it with --disable-verity and --disable-verification11:23
rinigusunless you have done so already11:23
rinigusas `fastboot --disable-verity --disable-verification flash vbmeta vbmeta.img`11:24
riniguswho I am supposed to ask for legal advice regarding distribution of images? In particular, am I allowed to upload images created by and vendor to OBS? to make it possible to get CI image generation11:32
Mister_Magisterrinigus: you don't quite get it but thanks for wanting to help me11:37
Mister_Magistersfos is booting fine that isn't an issue11:37
Mister_Magisteri couldn't boot lineage without flashing vbmeta11:37
Mister_Magisterrinigus: you can ask mal to ask people i believe11:38
rinigusMister_Magister: most probably I don't get it, no problem, you are welcome11:38
Mister_Magisterrinigus: sfos is booting fine but sfos boot image breaks bootloader11:39
rinigusMister_Magister: sounds like some bootloader bug, if true. but if you have vbmeta, I would suggest to flash it once with disable options. As far as I understood, its one of the jobs of bootloader to check whether images correspond to vbmeta info. maybe, its buggy, and fails itself if its not so11:41
Mister_Magisterrinigus: i already flashed it in past is what i said ealier11:41
rinigusMister_Magister: you did not specify whether you used --disable... options. at least, not today11:43
riniguswhen flashing vbmeta11:43
Mister_Magisterwithout these options lineage wouldn't boot11:43
Mister_Magisterneither sfos would11:43
riniguswell, that's not universal. its possible to make vbmeta correctly and boot without those options.11:44
rinigusbut in that case, I'm out of options that I am aware of.11:44
Mister_Magistertho vbmeta is not issue i believe11:44
rinigusdo you have dtbo partition? if you do, you could try flashing that as well, but I don't think it will somehow freeze bootloader11:45
Mister_Magisteri mean it only freezes bootloader if i add android and spl version11:46
riniguswhat I did see on xz2 was that size of kernel could flip bootloader to consider device corrupted. but symptoms were different somewhat11:46
Mister_Magisterkernel size is same11:48
riniguskernel size is the same, agreed.11:54
rinigusanyway, as we are out of options, you may try to remove nfs and other non-vital options. its a long shot, but worth a try. in practice, you look for what's different in yours and android kernels and try to remove excess11:55
Mister_Magisterwhy would i remove nfs11:56
Mister_Magisterwhat for11:56
rinigusMister_Magister: see
rinigusMister_Magister: in short, dropping nfs and other extras allowed me to boot11:58
Mister_Magisterbut i am able to boot11:58
rinigusMister_Magister: as I mentioned, its a long shot. but, in theory, as android works and sfos doesn't when you set android/spl versions, try to make sfos boot image as close to android as possible11:59
Mister_Magistersfos works too12:00
Mister_Magistersfos is working fine12:00
riniguseasiest, is to comment as much out as you can. nfs is not needed to boot. nor many other things that we usually enable (usually towards the end of kernel config)12:01
Mister_Magisterbut what for?12:01
rinigusits not working fine if you cannot set the versions in boot image12:01
Mister_Magistersfos is booting fine12:01
Mister_Magisteryes i can12:01
riniguswell, those are the things that worked for me. not sure I can add much more to that :)12:04
Mister_Magisterbooting sfos is not a problem as sfos is working fine12:04
T42<nitin03 %lastname%> getting this12:06
T42<nitin03 %lastname%> Sailfish ofonod[7735]: Invalid version 1.22 for RIL binder transport plugin12:06
T42<nitin03 %lastname%> Sailfish ofonod[9391]: Unknown RIL transport: binder12:06
T42<nitin03 %lastname%> full log
rinigusnitin03: (would you mind to fix your telegram nick, its bad on irc/matrix, see our logs)12:11
T42<nitin03 %lastname%> @rinigus [nitin03: (would you mind to fix your telegram …], how do i do that?12:11
rinigusnitin03: I had the same and it was due to ofono on sdk having higher version than on device. as a result, you have built binder with the wrong version.12:12
rinigusso, check ofono dev installed on sdk image (sb2 ... ) and make sure its the same as on device.12:12
T42<nitin03 %lastname%> @rinigus [nitin03: I had the same and it was due to ofon …], Okk :)12:13
rinigusnitin03 - re nick, don't know. its surely covered in telegram/irc bridge instructions.12:13
T42<nitin03 %lastname%> @rinigus [nitin03 - re nick, don't know. its surely cove …], okk I'll check it12:14
rinigusnitin03: and please don't quote messages as you reply. its also not nice on irc12:17
T42<nitin03 %lastname%> okk :)12:17
rinigusgood luck!12:18
T42<nitin03 %lastname%> thank you :)12:18
T42<dontatmepls> so I wanted to build this for a android 9 based device (raphael) and I don't know where to start even after reading the hadk pdf, can someone help xd12:58
Thaodanread the guide how to build the Xperia 10 adaption that's a good start however you still need to replicate all Xperia  10 repos  for your device and have a working aosp or lineage13:01
T42<dontatmepls> I have plenty of working trees, so I hope it'll work13:02
T42<nitin_3> :)13:14
T42<dontatmepls> if your device is treble, do you add the vendor to the local manifest as well or do you not need that13:39
deathmist@dontatmepls if you're talking about vendor blobs, you'll need them. usually you always need a device/, kernel/ and vendor/ repo for a functioning final image (follow .dependencies files as needed from device/ repo onwards)14:26
deathmistactually I'm not 100% sure as I've not thought about this much, those may be needed in the build process, but before SFOS you'll flash an Android ROM (most often LineageOS matching your hybris base version), which would already supply the base with the needed vendor blobs14:29
T42<dontatmepls> yeah now im at step 5.3, the mountpoint stuff and i dont know if i have to do all of them or just some, and if some, which ones, thanks15:35
malbetter to add all, usually only less than 10 of those are needed15:42
malin case you want to check which ones are needed, you can check fstab and .rc files for any mounts15:43
T42<dontatmepls> oh god exactly 100 then15:43
deathmist@dontatmepls I can send you a script I used in a bit since I had over 70 myself and there was no way I was going to do that brainless operation so many times over and over again :D15:48
T42<dontatmepls> sure that'd be awesome!15:48
T42<hacker12455> maybe it would be handy to include that script in HADK or the HADK FAQ?15:49
malwell checking fstab and rc files is not that difficult either15:50
malit could go to faq15:50
T42<hacker12455> yeah15:50
T42<hacker12455> It would be nice for devices with a lot of mountpoints15:50
T42<hacker12455> hmm any idea why modem does not work after first stage setup but does after a reboot?16:42
T42<hacker12455> gets stuck on registering16:42
deathmist@dontatmepls there sorry for the wait lol, I got *just a bit* carried away with rewriting the whole thing as it wasn't good at all xD use it like any other bash script; you know how it goes :p19:21
deathmistit's still not perfect by any means, but A LOT better and I'm actually happy with it19:22
T42<dontatmepls> Im having probs with hybris 16 anyway, gave up for today19:24
deathmistit still has a bunch of undocumented stuff that's mostly in logs for now, what issues exactly?19:25
deathmistI assume issues building hybris-hal? => + run "hybris-patches/ --mb"19:28
T42<dontatmepls> damn deathmist you can read my mind xD, I'll try tomorrow ty :)19:28
Mister_Magistermal: 3.2 on obs when21:12
Mister_Magisterlbt: fixed obs when21:13
rinigusspiiroin: moving logic to autosleep works as well
rinigusspiiroin: probably more correct location for this patch21:17

Generated by 2.17.1 by Marius Gedminas - find it at!