Monday, 2019-09-16

*** [mccreary] is now known as mccreary00:10
mal@adampigg are you really using EGL_PLATFORM=fbdev on pinephone, why not EGL_PLATFORM=drm if that is native port?09:37
T42<NotKit> EGL_PLATFORM is parsed by hybris, I think it does not matter with upstream mesa09:39
T42<adampigg> @mal [@adampigg are you really using EGL_PLATFORM=fb …], config was copied from latte, likely unnescessary theretoo09:47
T42<adampigg> oops, sorry for sin!09:47
mal@NotKit @adampigg qtwayland also uses EGL_PLATFORM
malnot sure if that for the same thing09:50
malcould actually be a bit different09:50
T42<NotKit> that is a hack for hybris apps09:50
T42<NotKit> if env is not set, it defaults to wayland, it should be not in upstream09:51
malbut anyway mesa does use EGL_PLATFORM09:51
T42<NotKit> interesting, found
merbotFreedesktop bug 111529 in EGL "EGL_PLATFORM=drm doesn't expose MESA_query_driver extension" [Not set,Resolved: fixed]09:53
mal@NotKit sfos emulator uses EGL_PLATFORM=drm09:55
T42<NotKit> does it still use llvmpipe?09:55
T42<NotKit> it could greatly benefit from added VMware GPU emulation in VirtualBox with upstream mesa09:56
T42<adampigg> mal, if i reboot my tablet now, and loose it for the day, i will be less than impressed ;)09:56
T42<adampigg> well, it still works....10:03
T42<adampigg> i was hoping the browser would be magically fixed, but its the same10:04
mal@NotKit emulator uses llvmpipe10:04
mal@adampigg so provide logs, and maybe even console output from browser10:05
T42<adampigg> (Document)
T42<adampigg> already at hand!10:05
T42<adampigg> mal, does ^^ tell you anything?11:03
mal@adampigg "EGL_HYBRIS_native_buffer is not available"11:07
T42<adampigg> yes, i saw that.....11:08
T42<adampigg> reminiscent of the bug i fixed11:08
mal@adampigg hmm, have you tried the colorspace thingy?11:08
T42<adampigg> no11:08
malI think it was QT_OPENGL_NO_BGRA=1 to /var/lib/environment/nemo/99-qtscenegraph.conf11:09
T42<adampigg> mal: no change11:13
malhmm, that usually helps with failing browser11:13
T42<adampigg> that was also for gallery wasnt it?11:14
T42<adampigg> no gallery issues11:14
mal@adampigg wasn't gallery usually adreno quirks, I always forgot which was which11:18
mal@adampigg that "EGL_HYBRIS_native_buffer is not available..." comes from qtscenegraph-adaptation project, and it should be fine, that is just info that it disabled hybris support in it11:23
T42<adampigg> yeah, it used to crash in lipstick iirc when it wasnt available11:26
T42<adampigg> so, why does the browser just show a white screen..11:26
malnot sure yet11:28
T42<adampigg> is there any hybris specific code in the engine? i shouldnt think so11:29
masha11Hello. I do not have GUI. I checked mounting and firmware, updated config submodule. My device is griffin, cm-14.1. logcat: . Any ideas?11:35
malmasha11: why is there nothing about hwcomposer in logcat, have you disabled any services?11:43
malsystemd services11:43
masha11mal: no. I did not run test_hwcomposer in the previous log.Now running. logcat:
malmasha11: does test_hwcomposer also fail?12:02
malmasha11: you need to also show other logs, dmesg and/or journal12:02
Mister_Magisterlogcat is pretty useless12:04
masha11mal: yes, test_hwcomposer fails. dmesg:, journalctl:
T42<elros34> @masha11: check if you really have CONFIG_SECURITY_SELINUX_BOOTPARAM enabled in generated .config and at least for now remove audit=0 from bootparams12:22
malmasha11: is CONFIG_AUDIT=y?12:25
T42<elros34> I wonder why you have:" Yama: becoming mindful."12:25
T42<elros34> instead Security: selinux disabled at boot12:25
masha11mal: yes, CONFIG_AUDIT=y (from .config)? and I had selinux=0 audit=0 (from
malyama is optional12:27
malusually audit=0 is not used12:27
masha11mal: ok, I delete audit=012:28
masha11new logs. journalctl: ,dmesg:
malis that after trying test_hwcomposer?12:42
maljust to be sure, you use "EGL_PLATFORM=hwcomposer test_hwcomposer"?12:42
masha11mal: Yes, so. tets_hwcomposer fails also.12:43
T42<elros34> I don't know, maybe it's because you enabled YAMA but there is still not "SELinux: Disabled at boot"12:44
masha11@elros34: I can try to disable YAMA.12:45
T42<elros34> show your zcat /proc/config.gz12:46
masha11I disabled YAMA. new logs: journalctl: , dmesg:
maldoes that again include a run of test_hwcomposer?13:31
malyou always need to try to run it before showing logs13:32
masha11I tried. but nothing new.13:32
malbut was it included in those logs13:33
malso the logs are essentially useless13:37
masha11test_hwcomposer run logs : . Ater starting test_hwcomposer, the device reboots also.13:40
malif the device reboots then you probably should have log catching running in several terminals to get all logs, one for dmesg, one for logcat and one for journal and last one for running test_hwcomposer13:42
maldmesg -w should continue catching the logs, logcat does that automatically and journalctl -f13:42
masha11Yes, I look. Now I will do logs of all terminals.13:44
masha11test_hwcomposer + journalctl -f: ,logcat:, dmesg -w:
masha11Error from logcat: GetAndroidProperty: property_get debug.egl.swapinterval failed (err = 0)13:51
masha11Error from journalctl and dmesg: devfreq soc:qcom,kgsl-busmon: Couldn't update frequency transition information13:52
masha11I do not know what to do next13:53
T42<elros34> dmesg -w might be to slow to catch everything so you can always check /sys/fs/pstore/console-ramoops, also journalctl have option -b -1 to show log from previous boot13:54
masha11@elros34: thanks, I try13:56
masha11@elros34: Please take a look at my log, I may not notice something. console-ramoops:, journalctl -b -1:
mal@elros34 wondering if that device has broken hwcomposer again14:03
T42<elros34> Could it be because of missing QCOM_BSP or it crash to early?14:06
T42<elros34> @masha11: just to be sure. Check if you have QCOM_BSP in android-config.h in droid-hal-devel rpm14:08
mal@elros34 that is also possible, worth a try14:09
masha11abranson ok, I'm checking14:10
masha11@elros34, ok, I'm checking14:10
masha11@elros34: Yes, I have. :%define android_config \    #define QCOM_BSP 1\    #define QTI_BSP 1\%{nil}14:16
malmasha11: in spec I assume14:18
malwhich should then add those #define lines to android-config.h14:19
T42<elros34> what? Have you added it to spec and rebuild everything? Does android-config.h really missing it?14:19
masha11This lines from rpm/droid-hal-griffin.spec14:20
masha11I already had these lines14:21
malmasha11: check the android-config.h in droid-hal devel rpm manually14:23
masha11But I do not find android-config.h file.14:23
malmasha11: inside the built droid-hal devel rpm14:23
T42<elros34> show your droid-hal-griffin.spec14:28
malmasha11: android-config.h is always in that rpm14:28
malmasha11: are you sure you are checking it correctly?14:28
masha11mal: I thynk yes. ls -la hadk/rpm and hadk/rpm/dhd:
T42<elros34> in $ANDROID_ROOT/droid-local-repo/griffin/droid-hal-griffin there is droid-hal-devel rpm package, unpack it and search for android-config.h file14:34
masha11@elros34: ok14:35
masha11@elros34: sorry, android-config.h:
masha11Do you need to put this file somewhere?14:38
masha11To hadk/rpm/dhd?14:40
masha11or hadk/rpm14:42
malmasha11: THE BUILT rpm, NOT rpm folder14:42
malthe actual rpm archive in droid-local-repo14:42
T42<elros34> as you can see QCOM_BSP was already added so you shoudn't add it to spec14:42
malso you never needed the defines in spec14:42
mallooks more like the hwcomposer bug14:43
masha11ok, thanks14:43
masha11can hwcomposer error be fixed in some way?14:44
T42elvarr was added by: elvarr17:22
piggzkrnlyng: hi, any thoughs on my bluebinder PR?18:00
krnlyngpiggz, lgtm except maybe i would write droid instead of port in the script name18:07
krnlynglike we use in other scripts18:08
piggzkrnlyng: ok, i'll change that18:42
piggzkrnlyng: mal: ?18:44
krnlyngpiggz, or or something like that18:45
krnlyngaddress* not firmware18:45
piggzPR updated18:48
piggzi think this way makes sense, so all ports dont have to create their own services etc to do the same thing18:49
malpiggz: also it still has the fallback so old ports don't have to be changed18:49
piggzmal: for latte and pro1, i set a random address as ive found nothing so far that contains a bt address!19:01
piggzon latte, its all bodged together, but this will be much neater on the pro119:01
piggzive ran out of flipping USB ports on my laptop .... on one side i have the pinephone, taking 2 ports, one for power and one for the serial ... on the other i only have 1 port, and need both mido and my nitrokey!19:08

Generated by 2.17.1 by Marius Gedminas - find it at!