Monday, 2024-06-03

T42_<Tailsthefoz> @NotKit lmk if there's anything else I should try01:50
T42_<Tailsthefoz> guess not11:09
T42_<Tailsthefoz> might just do the HADK method11:17
T42_<Tailsthefoz> or buy a snapdragon phone11:35
T42_<Tailsthefoz> I'll try HADK first11:36
T42_<NotKit> debug why HWC is failing (re @Tailsthefoz: @NotKit lmk if there...)11:36
T42_<NotKit> start with tes_hwcomposer code maybe11:37
T42_<Tailsthefoz> how11:37
T42_<NotKit> start with test_hwcomposer code maybe (edited)11:37
T42_<Tailsthefoz> oh alright, I don't have a compilation environment for ARM64 sailfish though11:37
T42_<Tailsthefoz> oh alright cool11:38
malwas the reason for the error 13 figured out?11:41
malwhich would suggest permission issue11:41
malwhich soc is that device using, wondering which devices needed tls-padding11:45
mal@NotKit ^11:45
malwhat is this in log: file offset for the library "/vendor/lib64/hw/" >= file size: 0 >= 011:47
malis it finding some incorrect file11:47
T42_<NotKit> that's fine if /dev/null is mounted over that file (re @SailfishFreenodeIRCBridgeBot: <mal>what is this in...)11:53
T42_<NotKit> is not supposed to be used11:53
T42_<NotKit> I guess newer libhybris with your patch avoids the need for that hack11:54
malwhy is something trying to use that gralloc? why else would it complain about it11:55
T42_<NotKit> maybe older libhybris?12:04
mal@Tailsthefoz can you check which libhybris version you are building12:17
T42_<Tailsthefoz> ok in a moment (re @SailfishFreenodeIRCBridgeBot: <mal>@Tailsthefoz ca...)12:28
T42_<Tailsthefoz> I'm using the libhybris packaged with the vidofnir rootfs12:28
T42_<Tailsthefoz> tensor g1, aka exynos 2100 (re @SailfishFreenodeIRCBridgeBot: <mal>which soc is th...)12:31
T42_<Tailsthefoz> it is, yes (re @NotKit: that's fine if /dev/...)12:33
T42_<Tailsthefoz> although I'm fairly sure it's not supposed to be12:33
T42_<Tailsthefoz> because it isn't on droidian12:33
T42_<Tailsthefoz> mal, libhybris is12:38
T42_<Tailsthefoz> ```12:38
T42_<Tailsthefoz> libhybris-```12:38
T42_<NotKit> you can install a newer one from OBS if you can set up the network via USB12:38
T42_<NotKit> you can install a newer one from OBS if you can set up the network via USB or Wi-Fi (edited)12:38
T42_<Tailsthefoz> I have sftp, so I can just send the RPM over12:38
T42_<Tailsthefoz> what's the URL12:38
T42_<Tailsthefoz> it's the latest on the repo as far as I can tell (re @NotKit: you can install a ne...)12:42
T42_<Tailsthefoz> do I compile it from here?
T42_<NotKit> will see if it builds here12:48
T42_<Tailsthefoz> oh alright12:48
T42_<NotKit> but at the same time, I don't think that gralloc message matters much12:49
T42_<Tailsthefoz> maybe12:50
T42_<NotKit> maybe try attaching strace to hwcomposer service to see if there is any hint why it fails12:50
T42_<Tailsthefoz> wtf12:50
T42_<Tailsthefoz> I just rebooted12:50
T42_<Tailsthefoz> and I have UI12:50
T42_<Tailsthefoz> ???12:50
T42_<NotKit> cool :)12:50
T42_<Tailsthefoz> I didn't even do anything lmao12:50
T42_<NotKit> maybe you masked community encryption wizard?12:50
T42_<Tailsthefoz> I didn't12:50
T42_<Tailsthefoz> I see the encryption wizard12:50
T42_<NotKit> then I guess it is some kind of race condition, maybe UI is attempting to start too early (before vendor service is properly initialized)12:51
T42_<Tailsthefoz> maybe?12:51
T42_<Tailsthefoz> ok but new problem12:51
T42_<Tailsthefoz> as soon as I wake it up, it goes back to sleep12:51
T42_<Tailsthefoz> touch doesn't work but I know how to fix that12:51
T42_<NotKit> that sounds like mce issue, but you can use mcetool to prevent blanking for now, like `mcetool -k unlocked` (re @Tailsthefoz: as soon as I wake it...)12:52
T42_<Tailsthefoz> after doing that it still does it12:53
T42_<Tailsthefoz> as soon as I wake it up it just dims the screen to black12:53
T42_<NotKit> check brightness state too12:54
T42_<Tailsthefoz> and now it isn't waking back up hu12:54
T42_<Tailsthefoz> and now it isn't waking back up huh (edited)12:54
T42_<Tailsthefoz> how do I do that (re @NotKit: check brightness sta...)12:55
T42_<NotKit> via mcetool and manually in sysfs12:56
T42_<Tailsthefoz> which command in mcetool12:57
T42_<NotKit> just run mcetool, it will display brightness on ~10th line of output (re @Tailsthefoz: which command in mce...)12:59
T42_<Tailsthefoz> oh alright12:59
T42_<Tailsthefoz> ```12:59
T42_<Tailsthefoz> Brightness:                          100 (1-100)```12:59
T42_<NotKit> and check /sys/class/backlight/panel0-backlight/brightness13:00
T42_<Tailsthefoz> 013:00
T42_<NotKit> then probably it's writing to a wrong file or similar, try setting in sysfs manually13:00
T42_<Tailsthefoz> just echo?13:00
T42_<NotKit> yes13:00
T42_<Tailsthefoz> it set it but nothing happene13:01
T42_<Tailsthefoz> and now it's changing alot13:01
T42_<Tailsthefoz> and it's at 0 again13:02
T42_<Tailsthefoz> I think it is writing it properly13:02
T42_<Tailsthefoz> bl_power is 013:02
T42_<Tailsthefoz> if I echo 1 to that nothing happens13:04
T42_<Tailsthefoz> state says13:04
T42_<Tailsthefoz> ```13:04
T42_<Tailsthefoz> On: 1440x3120@60```13:04
T42_<Tailsthefoz> ooh I see why13:09
T42_<Tailsthefoz> it's on13:09
T42_<Tailsthefoz> but u13:09
T42_<Tailsthefoz> logcat spam13:09
T42_<Tailsthefoz> ```13:09
T42_<Tailsthefoz> nt(x:0, y:0, w:0, h:0), block(x:0, y:0, w:0, h:0)13:09
T42_<Tailsthefoz> 09-07 19:33:19.690    54    54 E display : presentDisplay:: fail to deliver win_config (-22)13:09
T42_<Tailsthefoz> 09-07 19:33:19.690     0  6461 W HwcComposer: command 0x2010000 generated error 4```13:09
T42_<Tailsthefoz> ```13:09
T42_<Tailsthefoz> 09-07 19:33:19.689    54    54 W display : configureOverlay:: ExynosCompositionInfo(1) has invalid data, handle(0x0)13:09
T42_<Tailsthefoz> 09-07 19:33:19.689    54    54 E display : [PrimaryDisplay] WIN_CONFIG error: has no buffer window```13:09
T42_<Tailsthefoz> why does it break after a couple of minutes13:10
T42_<Tailsthefoz> isn't -22 EINVAL13:11
T42_<Tailsthefoz> og yeah13:11
T42_<Tailsthefoz> ```13:11
T42_<Tailsthefoz> 09-07 19:35:48.209    54    54 D display : Invalid WIN_CONFIG13:11
T42_<Tailsthefoz> 09-07 19:35:48.211    54    89 W ComposerResources: invalid cache 0 slot 013:11
T42_<Tailsthefoz> 09-07 19:35:48.211    54    89 W display : configureOverlay:: ExynosCompositionInfo(1) has invalid data, handle(0x0)13:11
T42_<Tailsthefoz> 09-07 19:35:48.211    54    89 D display : Invalid WIN_CONFIG```13:11
malmaybe you are missing the hwc restart configs13:13
T42_<Tailsthefoz> what is that13:13
T42_<Tailsthefoz> this?
T42_<Tailsthefoz> I'll try it?13:15
T42_<Tailsthefoz> should I?13:17
malthat requires quite new mce13:17
malnot the one which is in 4.5.0 by default13:17
T42_<Tailsthefoz> ok what shpuld13:17
T42_<Tailsthefoz> ok what shoud (edited)13:17
T42_<Tailsthefoz> ok what should i do then (edited)13:17
malsomething like the opposite of this
malforgot that mce compositor config thing added in that is the one which needs new mce13:19
malthe files removed in that were the old way13:20
T42_<Tailsthefoz> wdym by opposite13:23
T42_<Tailsthefoz> oh I see13:25
T42_<Tailsthefoz> alright well that's gonna be annoying14:04
malother option is to build and install newer mce and use the simple config14:05
T42_<Tailsthefoz> ok I attempted to make all of those configs, hopefully it works14:14
T42_<Tailsthefoz> what14:22
T42_<Tailsthefoz> ```14:22
T42_<Tailsthefoz> Sep 07 20:41:47 PhoneX23 stop[9424]: Unable to stop service 'vendor.hwcomposer-214:22
T42_<Tailsthefoz> -3'```14:22
T42_<Tailsthefoz> I literally checked every file in the filesystem14:22
T42_<Tailsthefoz> pretty much14:22
T42_<Tailsthefoz> not quite though14:22
T42_<Tailsthefoz> ok found it14:22
T42_<Tailsthefoz> omg it's because the vidofnir adaptation has that line lol14:23
T42_<Tailsthefoz> alright now it should work maybe14:24
T42_<Tailsthefoz> ah I found the issue14:33
T42_<Tailsthefoz> ```14:33
T42_<Tailsthefoz> [root@PhoneX23 ~]# cat /usr/lib/systemd/system/sailfish-device-encryption-commun14:33
T42_<Tailsthefoz> ity-wizard.service.d/20-wait-hwc.conf14:34
T42_<Tailsthefoz> [Service]14:34
T42_<Tailsthefoz> ExecStartPre=/usr/bin/droid/wait_for_hwcomposer.sh14:34
T42_<Tailsthefoz> [root@PhoneX23 ~]# cat /usr/bin/droid/wait_for_hwcomposer.sh14:34
T42_<Tailsthefoz> #!/bin/sh14:34
T42_<Tailsthefoz> while [ "$(binder-list -d /dev/hwbinder |grep" == "" ]; do14:34
T42_<Tailsthefoz>         echo "Waiting for hwc"14:34
T42_<Tailsthefoz>         sleep 114:34
T42_<Tailsthefoz> done14:34
T42_<Tailsthefoz> 14:34
T42_<Tailsthefoz> exit 0```14:34
T42_<Tailsthefoz> lol, my device is 2.414:34
T42_<Tailsthefoz> no nvm14:35
T42_<Tailsthefoz> that still exists in binder-list14:35
T42_<Tailsthefoz> ok so it seems to break when I load the remaining kernel modules, it might be the proximity sensor14:42
T42_<Tailsthefoz> another thing, I can't get it to show UI everytime14:43
T42_<Tailsthefoz> sometimes it does but sometimes it doesn't14:43
T42_<Tailsthefoz> ok I stopped this service: ```15:35
T42_<Tailsthefoz> wlan-module-load```15:35
T42_<Tailsthefoz> and GUI popped up lol15:35
T42_<Tailsthefoz> touch works15:35
T42_<Tailsthefoz> oh no it panicked15:37
T42_<Tailsthefoz> why15:37
T42_<Tailsthefoz> oh I see why15:42
T42_<Tailsthefoz> @NotKit
T42_<Tailsthefoz> so it doesn't like that being /dev/null15:58
T42_<Tailsthefoz> it still happens even if I don't mount /dev/null to it16:03
T42_<Tailsthefoz> ```16:04
T42_<Tailsthefoz> 09-07 22:26:55.210     0  8782 E mali    : ==>[INIT] (booster-silica-) CDBG: In file: vendor/arm/mali/valhall/cdbg/src/mali_cdbg_env.c  line: 722 cdbgp_populate_from_system_environment16:04
T42_<Tailsthefoz> 09-07 22:26:55.210     0  8782 E mali    : Initialization of a handle to the system environment failed (3)```16:04
T42_<Tailsthefoz> that's permission denied16:05
T42_<Tailsthefoz> but to what file16:05
T42_<Tailsthefoz> that's not the cause of the panic though16:16
T42_<Tailsthefoz> ok so masking mce makes it not panic19:13
T42_<Tailsthefoz> and GUI shows, while rebooting for some reason19:13
T42_<Tailsthefoz> but only while rebooting19:14
T42_<NotKit> lipstick won't display without mce19:16
T42_<Tailsthefoz> yeah I noticed19:19
T42_<Tailsthefoz> apparently it's caused by voicecall-ui for some reason19:24
T42_<Tailsthefoz> ```19:25
T42_<Tailsthefoz> [  130.777702][ T8837] dbg_snapshot_do_dpm_policy: KP: Oops: Fatal exception: comm:voicecall-ui PC:gpu_dvfs_kctx_init+0x48/0x1bc [mali_kbase] LR:gpu_dvfs_kctx_init+0x48/0x1bc [mali_kbase]```19:25
T42_<Tailsthefoz> ah it's probably this same issue (re @unknown: @elros34 It does not...)19:31
T42_<Tailsthefoz> I'll try that patch19:32
T42_<Tailsthefoz> it did absolutely nothing lol19:54
T42_<Tailsthefoz> @NotKit I got lipstick somehow20:38
T42_<Tailsthefoz> I was just kinda spamming random commands20:39
T42_<Tailsthefoz> but there's nothing on screen other than background20:39
T42_<Tailsthefoz> anyone ever encountered this before? (re @Tailsthefoz: [  130.777702][ T883...)21:06
T42_<Tailsthefoz> it seems to have something to do with process ID21:14
T42_<Tailsthefoz> at least looking in my reverse engineering tool21:14
T42_<Tailsthefoz> but then this literally says it has a PID ```21:16
T42_<Tailsthefoz> CPU: 6 PID: 8969 Comm: voicecall-ui Tainted: G        W  O```21:16
T42_<Tailsthefoz> it's a different process that causes it everytime too21:19
*** tohaturok is now known as Guest850221:35
*** tohaturok_ is now known as tohaturok21:35
T42_<Tailsthefoz> ```21:45
T42_<Tailsthefoz> /* Get UID from task_struct */21:45
T42_<Tailsthefoz>   task = get_pid_task(find_get_pid(kctx->kprcs->tgid), PIDTYPE_TGID);21:45
T42_<Tailsthefoz>   uid = task->cred->uid;```21:45
T42_<Tailsthefoz> it's this line here21:45
T42_<Tailsthefoz> (the second one)21:45
tohaturokHow to rebuild droid-hal? I just rerun command make -j$(nproc --all) hybris-hal droidmedia, but got ERROR: pdx224 does not have mountpoint fixup data - see Sailfish OS HADK for details on how to fix this.
tohaturokbut I have fixup data for mountpoint for my device21:46
T42_<Tailsthefoz> I could implement a fix I guess21:50
T42_<Tailsthefoz> I could make it like21:50
T42_<Tailsthefoz> ```21:51
T42_<Tailsthefoz> if(task->cred == NULL) {return 0;}```21:51
T42_<Tailsthefoz> because after all the function isn't crucial, it's purpose is to monitor a task's GPU usage per UID21:52
T42_<Tailsthefoz> oh21:52
T42_<Tailsthefoz> oh that must be the issue21:52
T42_<Tailsthefoz> no wait nvm21:52
T42_<Tailsthefoz> i'll implement a patch and see what happens21:53
T42_<Tailsthefoz> ```21:56
T42_<Tailsthefoz> if(task == NULL) {21:56
T42_<Tailsthefoz>     return 0;21:56
T42_<Tailsthefoz>   }21:56
T42_<Tailsthefoz>   if(task->cred == NULL) {21:56
T42_<Tailsthefoz>     return 0;21:56
T42_<Tailsthefoz>   }21:56
T42_<Tailsthefoz> ```21:56
T42_<Tailsthefoz>  that should work21:56
maltohaturok: are you sure you have the fixup data, check the hybris/hybris-boot/fixup-mountpoints file in your build env21:57
tohaturokmal: sorry, my bad. I've checked out wrong branch21:58
tohaturokI built hybris-hal successfully, but I see the error in console: cpio: : No such file or directory. Log: Is it ok?22:00
tohaturokP.S. cpio is installed22:03
T42_<Tailsthefoz> no instead of doing that, I should figure out why the kernel can't get the task (re @Tailsthefoz: i'll implement a pat...)22:09
T42_<Tailsthefoz> hm ok so you can completely disable all of that in the kernel, aka I don't think the function has to run at all22:14
T42_<Tailsthefoz> so I'm going to try to make it immediately RET and see what happens22:14
T42_<elros34> tohaturok: what hybris branch do you use for android repo?22:19
tohaturokelros34: sony-aosp-1322:43
T42_<elros34> hybris-sony-aosp-13? same for hybris-patchs? It's strange that you have this cpio error which is old issue for different branch22:44
T42_<Tailsthefoz> ugh but it's so difficult to compile a kernel module lol (re @Tailsthefoz: so I'm going to try ...)23:36

Generated by 2.17.1 by Marius Gedminas - find it at!