Coffee Lake intel GPU support on android nougat x86-rc1

[YoungJoon Lee]

Hello. 

I am trying to use android nougat x86 rc1 on Coffee Lake system.

Linux kernel community said 

[*] Enable alpha quality support for new Intel hardware by default

Makes it possible to use new intel GPU. 

Q1. Can I use kernel branch 'kernel-4.14' source on Android nougat? 

Q2. Is there any modification of code is needed for HAL using new GPU?

Best regards.

[Chih-Wei Huang]

2018-01-13 23:27 GMT+08:00 YoungJoon Lee <get...@gmail.com>:
> Hello.
>
> I am trying to use android nougat x86 rc1 on Coffee Lake system.
>
> Linux kernel community said
> [*] Enable alpha quality support for new Intel hardware by default
> Makes it possible to use new intel GPU.
>
> Q1. Can I use kernel branch 'kernel-4.14' source on Android nougat?

Yes. Just do:

cd kernel/
git fetch x86 kernel-4.14
git checkout x86/kernel-4.14

(note only 64-bit defconfig is available now)

You can also refer to Mauro's repo which
has more newer branches:
https://github.com/maurossi/linux

He also released some testing iso with new kernel.
Just search the forum for it.

> Q2. Is there any modification of code is needed for HAL using new GPU?

Mesa is probably the only one needs to be updated.


--
Chih-Wei
Android-x86 project
http://www.android-x86.org

[YoungJoon Lee]

Thank you for your answer. I am testing maurossi's kernel-4.15rc8. I have to modify "./device/generic/common/tp_smapi/hdaps.c" little. 

Finally I can build it. 

Test result : 

This build work good on Sky Lake system but can't boot on Coffee Lake system. Android service is killed and restarted recursively. I attach logcat and dmesg. What I should do for fix this Error?

2018년 1월 15일 월요일 오후 12시 10분 44초 UTC+9, Chih-Wei Huang 님의 말:

[Chih-Wei Huang]

2018-01-15 23:12 GMT+08:00 YoungJoon Lee <get...@gmail.com>:
> Thank you for your answer. I am testing maurossi's kernel-4.15rc8. I have to
> modify "./device/generic/common/tp_smapi/hdaps.c" little.
>
> Finally I can build it.
> Test result :
> This build work good on Sky Lake system but can't boot on Coffee Lake
> system. Android service is killed and restarted recursively. I attach logcat
> and dmesg. What I should do for fix this Error?

I didn't see any i915 loaded message in dmesg.

What's "cat /proc/fb"?

[YoungJoon Lee]

Thank you for your answer. I will boot to debug mode  and check it is loaded or not. And then I will modprobe manually if needed. 

Best regards.

2018년 1월 16일 화요일 오전 12시 54분 34초 UTC+9, Chih-Wei Huang 님의 말:

[YoungJoon Lee]

I dump dmesg before bootstrap is not complete. I reattach new dmesg log. 

2018년 1월 16일 화요일 오전 12시 59분 40초 UTC+9, YoungJoon Lee 님의 말:

[YoungJoon Lee]

I have to change hdaps.c on ./device/generic/common/tp_smapi/ 

here is diff. 

472d471

< #if LINUX_VERSION_CODE < KERNEL_VERSION(4,15,0)

474,476d472

< #else

< static void hdaps_mousedev_poll(struct timer_list *unused)

< #endif

786d781

< #if LINUX_VERSION_CODE < KERNEL_VERSION(4,15,0)

789,791d783

< #else

<       timer_setup(&hdaps_timer, hdaps_mousedev_poll, 0);

< #endif

2018년 1월 16일 화요일 오전 1시 25분 55초 UTC+9, YoungJoon Lee 님의 말:

[Mauro Rossi]

If you are trying latest kernel, here is the two essential patches I'm using  for kernel 4.15:

device/generic/common

https://github.com/maurossi/device_generic_common/commit/eed772d7828990d64655370be6e0f02b9397f337

Some essential permission settings (needed since kernel 4.14) to get wifi working:

system/connectivity/wificond

https://github.com/maurossi/system_connectivity_wificond/commit/41e8ae0c075d6b70356e5942248f5814600d6743

Here are the optional ones:

optional:  some hack to have Developer Settings working and possibility to enable "Stay Awake" (if needed the last two commits)

packages/apps/Settings

https://github.com/maurossi/packages_apps_Settings/commits/oreo-x86_810_hacks_v2

optional libdrm 2.4.89

https://github.com/maurossi/drm/commits/2.4.89_android-x86

optional mesa-dev

https://github.com/maurossi/mesa/tree/17.4.0devel_w02

Mauro

[Mauro Rossi]

..but...if you are building nougat, as mentioned in the title,

you should need only to patch  device/generic/common

sorry, my list was for oreo-x86

[YoungJoon Lee]

Thank you for your answers. 

I also want to try oreo. Thank you very much. I will try oreo today. Best regards.

2018년 1월 16일 화요일 오전 2시 33분 10초 UTC+9, Mauro Rossi 님의 말:

[YoungJoon Lee]

One more question, Mauro Rossi.

Is -C option on android make system removed on Oreo? Which can be used on Kernel's menuconfig? 

2018년 1월 16일 화요일 오전 2시 33분 10초 UTC+9, Mauro Rossi 님의 말:

..but...if you are building nougat, as mentioned in the title,

[Mauro Rossi]

Hi,

Il giorno martedì 16 gennaio 2018 06:11:06 UTC+1, YoungJoon Lee ha scritto:

One more question, Mauro Rossi.

Is -C option on android make system removed on Oreo? Which can be used on Kernel's menuconfig? 

I don't know about -C option uses, sorry, but it seams a make option to launch a make "in another tree",

I never had to use it

When I move to recent kernel and there is a build error I do:

lunch android_x86_64-eng

cd out/target/program/x86_64/obj/kernel

make menuconfig

[enter as many times as requested]

when I have a working defconfig file I move it in kernel/arch/x86/configs

then I use the following option to build with custom config:

example:

make -j3 iso_img TARGET_KERNEL_CONFIG=4150_64_defconfig

However by checking out latest kernel-4.15rc8, the default defconfig files are already ok

In case you introduce different defconfigs, please make sure that you have CONFIG_NO_GPU_AUTHENTICATION=y

because it's not a standard one, it should be preserved at build, but I'm not sure it will survive a make oldconfig command.

The SIGABRT due to no EGLconfig may happen for following reasons:

- not having CONFIG_NO_GPU_AUTHENTICATION=y

- with latest mesa-dev the gamut wide colour support in mesa-dev causes this,

This repo contains the latest patches/workarounds, courtesy of Tapani Palli, https://github.com/maurossi/mesa/tree/17.4.0devel_w03

- failure in dlopen() of mesa EGL/gallium_dri libraries, maybe due to part of the libraries being moved to /vendor/changes 

This repo may help also for nougat-x86, in case you to build latest mesa 17.4.0devel: https://github.com/maurossi/drm/commits/2.4.89_android-x86

Mauro

[Mauro Rossi]

Hi, a little errata corrige

Il giorno martedì 16 gennaio 2018 13:03:08 UTC, Mauro Rossi ha scritto:
> Hi,
>
> Il giorno martedì 16 gennaio 2018 06:11:06 UTC+1, YoungJoon Lee ha scritto:
> One more question, Mauro Rossi.
> Is -C option on android make system removed on Oreo? Which can be used on Kernel's menuconfig? 
>
>
>
> I don't know about -C option uses, sorry, but it seams a make option to launch a make "in another tree",
> I never had to use it
>
>
> When I move to recent kernel and there is a build error I do:
>
>
> lunch android_x86_64-eng
> cd out/target/program/x86_64/obj/kernel
> make menuconfig

Correction: it is make oldconfig

[YoungJoon Lee]

Thank you very much. I am check 3 point on your advise. 

•  On all of image I have built,  kernel setting CONFIG_NO_GPU_AUTHENTICATION is yes.
  
• mesa and drm : I am trying to use your code on my source. But not tested yet. 

I don't know detail but image working perfectly on skylake is not working on skylake system. It is very hard to boot on coffee lake system. 

2018년 1월 17일 수요일 오전 1시 37분 53초 UTC+9, Mauro Rossi 님의 말:

[Chih-Wei Huang]

2018-01-19 4:26 GMT+08:00 spin877 <sfuma...@gmail.com>:
>> example:
>> make -j3 iso_img TARGET_KERNEL_CONFIG=4150_64_defconfig
>

> I to speed up the build of the kernel I add this:
>
> /device/generic/common/build/tasks/kernel.mk
>
> mk_kernel := + $(hide) $(MAKE) -jX -C
>
> X=cpu

No need to add it if you have added
-jX to make iso_img.

Submake is parallel execution automatically,
as shown in 'ps axf':

29417 pts/42 Sl+ 0:00 | \_ /mnt/cwhuang/oreo-x86/out/soong_ui
--make-mode -j24 iso_img
30617 pts/42 S+ 0:09 | \_
prebuilts/build-tools/linux-x86/bin/ninja -d keepdepfile iso_img -j 24
-f out/combined-android_x86_64.ninja -w dupbuild=err
30944 pts/42 S 0:00 | \_ /bin/sh -c /bin/bash -c
"(make -j24 -C kernel
O=/mnt/cwhuang/oreo-x86/out/target/product/x86_64/obj/kernel
ARCH=x86_64 CROSS_COMPILE=\"
30945 pts/42 S 0:00 | \_ /bin/bash -c (make -j24
-C kernel O=/mnt/cwhuang/oreo-x86/out/target/product/x86_64/obj/kernel
ARCH=x86_64 CROSS_COMPILE=" /mnt/
30946 pts/42 S 0:00 | \_ make -j24 -C kernel
O=/mnt/cwhuang/oreo-x86/out/target/product/x86_64/obj/kernel
ARCH=x86_64 CROSS_COMPILE= /mnt//cwhuang

[YoungJoon Lee]

Thank you for your answer. I finally success working nougat on Coffee lake system. I left How can I do.

1. HW : https://www.msi.com/Motherboard/Z370I-GAMING-PRO-CARBON-AC.html

2. Base : e8653d3

3. /Kernel : https://github.com/maurossi/linux/tree/kernel-4.15rc8

4. /external/mesa : https://github.com/maurossi/mesa/tree/17.2.0rc4_nougat-x86

5. /external/llvm : https://github.com/maurossi/llvm/tree/nougat-x86_exports
6. /external/libdrm : Must use Default one.

7. code modification : /device/generic/common

https://github.com/maurossi/device_generic_common/commit/eed772d7828990d64655370be6e0f02b9397f337

What I have tested(Working)

• Realtek ALC1220 Codec
• Realtek ALC1220 Codec Optical out : /device/generic/common/alsa/PCH.state needed.(https://github.com/getfeus/android-oreo-x86_android-auto/blob/master/device/generic/common/alsa/PCH.state)
• Intel® Dual Band Wireless-AC 8265 card
  
• NVMe
• Intel I219-V Gigabit LAN controller
  
• enable_nativebridge : Only check command success. Not testing Glthread fatal error I have reported.(https://groups.google.com/forum/#!msg/android-x86/E8iRTfYNJ90/wbnaYCzwBwAJ;context-place=msg/android-x86/ilk69kFFcWY/WlPFjIhSAAAJ)

What is not stable.

• During boot, ACPI error is displayed. Some kernel module crashed and re-loaded. Every module is ACPI related of thermal sensor related.(Mis-configured of defconfig?) 
• I can't dump message from dmesg.(Before logging daemon start?) But, Start with this string,  ACPI Error: [_OSI] Namespace lookup failure, AE_NOT_FOUND on early boot process(Prior to android device dev partition detect). Very close to this post. https://bugzilla.redhat.com/show_bug.cgi?id=1413342
• This message is disappeared on old kernel like 4.9.x but Coffee lake GPU can only working on 4.13.x~. All Kernels I have tested have same errors. But not care.(Not displayed after boot)
• I can't boot from any Oreo build.

Best regards.

2018년 1월 19일 금요일 오후 1시 17분 42초 UTC+9, Chih-Wei Huang 님의 말:

[Mauro Rossi]

This pattern of mesa 17.2.x + nougat-x86 libdrm working suggests

that the problem with Coffelake (gen 8) was most probably due to mesa libraries moved to /vendor in mesa 17.3 and later

Also when I build nougat, with  mesa 17.3 and later, I use x86/oreo-x86 version of external/drm_gralloc

In order to confirm I'm uploading experimental test images with mesa 17.3.3 here (3 hours to go):

https://drive.google.com/drive/folders/0B_OFHiIqgpSFMlk4bkhaMnAwbDA

Mauro

[YoungJoon Lee]

Thank you Mauro Rossi. 

I will test your test image and report here.

Today I found few problems.

(Fatal) 1 : System is extremely heavy. All response is very slow(Slower than skylake). I monitor details of top command. But it looks like there is no process consume CPU 100%. I guess it is DVFS(Dynamic Frequency) Issue.(Kernel message said ACPI related driver is unstable and thermal driver also print error messages.) How can I debug performance issue? For example, is there any way to disable power management or frequency control(Working on maximum clock speed)? I remember that kerner setting on defconfig's power profile is on demand.

(Not Fatal) 2 : App crashes when keyboard input pop up.(On some apps only). I will attach logcat and left code here related to this error. It looks like it is webview related issue. So, I will attach my device mk file which is import opengapps.

01-21 13:13:56.677 16130 16118 F google-breakpad: -----END BREAKPAD MICRODUMP-----
01-21 13:13:56.800 15305 16118 F libc    : Fatal signal 11 (SIGSEGV), code -6, fault addr 0x3bc9 in tid 16118 (Chrome_InProcGp)
01-21 13:13:56.899 16133 16133 F DEBUG   : *** *** *** *** *** *** *** *** *** *** *** *** *** *** *** ***
01-21 13:13:56.899 16133 16133 F DEBUG   : Build fingerprint: 'Android-x86/android_x86_64/x86_64:7.1.2/N2G48H/root01210301:userdebug/test-keys'
01-21 13:13:56.899 16133 16133 F DEBUG   : Revision: '0'
01-21 13:13:56.899 16133 16133 F DEBUG   : ABI: 'x86'
01-21 13:13:56.899 16133 16133 F DEBUG   : pid: 15305, tid: 16118, name: Chrome_InProcGp  >>> jp.co.yahoo.android.apps.navi <<<
01-21 13:13:56.899 16133 16133 F DEBUG   : signal 11 (SIGSEGV), code -6 (SI_TKILL), fault addr 0x3bc9
01-21 13:13:56.899 16133 16133 F DEBUG   :     eax 00000000  ebx cf899d40  ecx 00008000  edx c0dbec58
01-21 13:13:56.899 16133 16133 F DEBUG   :     esi 00000000  edi 00000000
01-21 13:13:56.899 16133 16133 F DEBUG   :     xcs 00000023  xds 0000002b  xes 0000002b  xfs 0000006b  xss 0000002b
01-21 13:13:56.899 16133 16133 F DEBUG   :     eip cf66012c  ebp c0dbec78  esp c0dbe680  flags 00210282
01-21 13:13:56.899 16133 16133 F DEBUG   : 
01-21 13:13:56.899 16133 16133 F DEBUG   : backtrace:
01-21 13:13:56.899 16133 16133 F DEBUG   :     #00 pc 005d812c  /system/lib/dri/i965_dri.so
01-21 13:13:56.899 16133 16133 F DEBUG   :     #01 pc 000267b2  /system/lib/dri/i965_dri.so
01-21 13:13:56.899 16133 16133 F DEBUG   :     #02 pc 00029bee  /system/lib/dri/i965_dri.so
01-21 13:13:56.899 16133 16133 F DEBUG   :     #03 pc 0035e248  /system/lib/dri/i965_dri.so
01-21 13:13:56.899 16133 16133 F DEBUG   :     #04 pc 0000ab1e  /system/lib/libglapi.so
01-21 13:13:56.899 16133 16133 F DEBUG   :     #05 pc 00a12d91  /system/app/webview/webview.apk (offset 0x927000)

Here is mk file.

GAPPS_VARIANT := nano

PRODUCT_PACKAGES += Chrome \

    Maps \

    KeyboardGoogle \

    GoogleTTS \

    YouTube \

    PixelIcons \

    PixelLauncher \

    Wallpapers \

    PixelLauncherIcons \

    GoogleServicesFramework \

    GoogleLoginService

#PRODUCT_PACKAGES += Hangouts

GAPPS_EXCLUDED_PACKAGES := PrebuiltGmsCoreInstantApps \

    FaceLock \

    AndroidPlatformServices

GAPPS_FORCE_BROWSER_OVERRIDES := true

PRODUCT_PROPERTY_OVERRIDES := \

    ro.kernel.android.gps=ttyUSB0 \

    ro.kernel.android.gpsttybaud=4800

$(call inherit-product,build/target/product/languages_full.mk)

$(call inherit-product,vendor/opengapps/build/opengapps-packages.mk)

# includes the base of Android-x86 platform

$(call inherit-product,device/generic/common/x86_64.mk)

# Overrides

PRODUCT_NAME := android_x86_64

PRODUCT_BRAND := Android-x86

PRODUCT_DEVICE := x86_64

PRODUCT_MODEL := Generic Android-x86_64

(Not fatal && Solve myself) 3. Cannot adjust spdif optical out's volume. I think it is normal and can adjust on RC1 is abnormal. I will attach code what I have modify.

diff --git a/audio/AudioPolicyManagerBase.cpp b/audio/AudioPolicyManagerBase.cpp

index 74ee22a..e3d16da 100644

--- a/audio/AudioPolicyManagerBase.cpp

+++ b/audio/AudioPolicyManagerBase.cpp

@@ -3248,7 +3248,7 @@ float AudioPolicyManagerBase::computeVolume(int stream,

         index != mStreams[stream].mIndexMin &&

         (device == AUDIO_DEVICE_OUT_AUX_DIGITAL ||

          device == AUDIO_DEVICE_OUT_DGTL_DOCK_HEADSET)) {

-        return 1.0;

+        // return 1.0;

     }

diff --git a/services/core/java/com/android/server/audio/AudioService.java b/services/core/java/com/android/server/audio/AudioService.java

index c70a87c..87cabe0 100644

--- a/services/core/java/com/android/server/audio/AudioService.java

+++ b/services/core/java/com/android/server/audio/AudioService.java

@@ -520,12 +520,13 @@ public class AudioService extends IAudioService.Stub {

             = new RemoteCallbackList<IAudioRoutesObserver>();

     // Devices for which the volume is fixed and VolumePanel slider should be disabled

-    int mFixedVolumeDevices = AudioSystem.DEVICE_OUT_HDMI |

-            AudioSystem.DEVICE_OUT_DGTL_DOCK_HEADSET |

-            AudioSystem.DEVICE_OUT_ANLG_DOCK_HEADSET |

-            AudioSystem.DEVICE_OUT_HDMI_ARC |

-            AudioSystem.DEVICE_OUT_SPDIF |

-            AudioSystem.DEVICE_OUT_AUX_LINE;

+    int mFixedVolumeDevices = 0;

+            //AudioSystem.DEVICE_OUT_HDMI |

+            //AudioSystem.DEVICE_OUT_DGTL_DOCK_HEADSET |

+            //AudioSystem.DEVICE_OUT_ANLG_DOCK_HEADSET |

+            //AudioSystem.DEVICE_OUT_HDMI_ARC |

+            //AudioSystem.DEVICE_OUT_SPDIF |

+            //AudioSystem.DEVICE_OUT_AUX_LINE;

     int mFullVolumeDevices = 0;

I think this modification is not good and it is not intent of designer's. Any idea's without modification of this code and how to adjust volume possible? 

 FATAL EXCEPTION: GLThread 178 issue I have reported(https://groups.google.com/forum/#!topic/android-x86/E8iRTfYNJ90) is fixed. It looks like it is related to libdrm or mesa. 

2018년 1월 21일 일요일 오후 8시 12분 2초 UTC+9, Mauro Rossi 님의 말:

[YoungJoon Lee]

I forgot to attach logcat file. Sorry. 

2018년 1월 22일 월요일 오전 1시 18분 6초 UTC+9, YoungJoon Lee 님의 말:

[Mauro Rossi]

Here is important to find the source file/line of code causing the segfault, by providing [???] (output of addr2line)

on source folder:

addr2line -Cfe out/product/target/x86_64/symbols/system/lib/dri/i965_dri.so

005d812c

[???]

000267b2

[???]

00029bee

[???]

0035e248

[???]

^C

addr2line -Cfe out/product/target/x86_64/symbols/system/lib/libglapi.so

0000ab1e
[???]


Same procedure required for the other SIGSEGV

01-21 12:47:21.791  3767  3959 F libc    : Fatal signal 11 (SIGSEGV), code 1, fault addr 0x20c in tid 3959 (Binder:3767_4)
01-21 12:47:21.899  4090  4090 F DEBUG   : *** *** *** *** *** *** *** *** *** *** *** *** *** *** *** ***
01-21 12:47:21.899  4090  4090 F DEBUG   : Build fingerprint: 'Android-x86/android_x86_64/x86_64:7.1.2/N2G48H/root01210301:userdebug/test-keys'
01-21 12:47:21.899  4090  4090 F DEBUG   : Revision: '0'
01-21 12:47:21.899  4090  4090 F DEBUG   : ABI: 'x86_64'
01-21 12:47:21.899  4090  4090 F DEBUG   : pid: 3767, tid: 3959, name: Binder:3767_4  >>> com.google.android.gms.unstable <<<
01-21 12:47:21.899  4090  4090 F DEBUG   : signal 11 (SIGSEGV), code 1 (SEGV_MAPERR), fault addr 0x20c
01-21 12:47:21.899  4090  4090 F DEBUG   :     rax 0000000000000000  rbx 0000000000000000  rcx 00007c395843de0c  rdx 00007c39591d5d94
01-21 12:47:21.899  4090  4090 F DEBUG   :     rsi 0000000000000000  rdi 0000000000000000
01-21 12:47:21.899  4090  4090 F DEBUG   :     r8  0000000000000003  r9  00007c39591d7450  r10 0000000000000000  r11 0000000000000005
01-21 12:47:21.899  4090  4090 F DEBUG   :     r12 00007c395846bc00  r13 00007c3972dd1f00  r14 0000000000000020  r15 0000000000000001
01-21 12:47:21.899  4090  4090 F DEBUG   :     cs  0000000000000033  ss  000000000000002b
01-21 12:47:21.899  4090  4090 F DEBUG   :     rip 00007c39587eb741  rbp 0000000000000000  rsp 00007c39591d5ca0  eflags 0000000000010246
01-21 12:47:23.574  4090  4090 F DEBUG   : 
01-21 12:47:23.574  4090  4090 F DEBUG   : backtrace:
01-21 12:47:23.574  4090  4090 F DEBUG   :     #00 pc 0000000000064741  /system/lib64/dri/i965_dri.so
01-21 12:47:23.574  4090  4090 F DEBUG   :     #01 pc 000000000006769a  /system/lib64/dri/i965_dri.so

on source folder:

addr2line -Cfe out/product/target/x86_64/symbols/system/lib64/dri/i965_dri.so
0000000000064741
[???]

000000000006769a
[???]

[fguy]

This pattern of mesa 17.2.x + nougat-x86 libdrm working suggests

that the problem with Coffelake (gen 8) was most probably due to mesa libraries moved to /vendor in mesa 17.3 and later

Also when I build nougat, with  mesa 17.3 and later, I use x86/oreo-x86 version of external/drm_gralloc

In order to confirm I'm uploading experimental test images with mesa 17.3.3 here (3 hours to go):

https://drive.google.com/drive/folders/0B_OFHiIqgpSFMlk4bkhaMnAwbDA

I tested your latest build "nougat_x86_64_amd_dc_llvm50_mesa-17.3.3_hwc1.iso" on the skylake 6Y30. Unfortunately, the new versions of MESA and kernel did not fix the problem in 7.1.2 which appeared since November 2017. A number of applications and games (antutu 3dmark, defense zone 3) using the Unity at startup lead to a reboot of the android. This problem is not present in 8.1 and in October assemblies 7.1.2. 

[YoungJoon Lee]

Ok, I read log file after reading your comment. It seems like i965(mesa?) issue. Not opengapps issue. 

First, I will check i965_dri.so. I don't know it is right. 

root@UbuntuPV:/mnt/nfs/data/src/android-x86-nougat-x86-rc2# addr2line -Cfe ./out/target/product/x86_64/symbols/system/lib/dri/i965_dri.so

005d812c

addr2line: Dwarf Error: Info pointer extends beyond end of attributes

addr2line: Dwarf Error: Info pointer extends beyond end of attributes

addr2line: Dwarf Error: Info pointer extends beyond end of attributes

addr2line: Dwarf Error: Info pointer extends beyond end of attributes

addr2line: Dwarf Error: Info pointer extends beyond end of attributes

addr2line: Dwarf Error: Info pointer extends beyond end of attributes

addr2line: Dwarf Error: Info pointer extends beyond end of attributes

addr2line: Dwarf Error: Info pointer extends beyond end of attributes

addr2line: Dwarf Error: Info pointer extends beyond end of attributes

addr2line: Dwarf Error: Info pointer extends beyond end of attributes

addr2line: Dwarf Error: Info pointer extends beyond end of attributes

addr2line: Dwarf Error: Info pointer extends beyond end of attributes

addr2line: Dwarf Error: Info pointer extends beyond end of attributes

addr2line: Dwarf Error: Info pointer extends beyond end of attributes

addr2line: Dwarf Error: Info pointer extends beyond end of attributes

addr2line: Dwarf Error: Info pointer extends beyond end of attributes

addr2line: Dwarf Error: Info pointer extends beyond end of attributes

get_fast_clear_rect

/proc/self/cwd/external/mesa/src/intel/blorp/blorp_clear.c:196

000267b2

do_single_blorp_clear

/proc/self/cwd/external/mesa/src/mesa/drivers/dri/i965/brw_blorp.c:859

00029bee

brw_clear

/proc/self/cwd/external/mesa/src/mesa/drivers/dri/i965/brw_clear.c:233

0035e248

clear

/proc/self/cwd/external/mesa/src/mesa/main/clear.c:221

Next, libglapi.so

root@UbuntuPV:/mnt/nfs/data/src/android-x86-nougat-x86-rc2# addr2line -Cfe ./out/target/product/x86_64/symbols/system/lib/libglapi.so

0000ab1e

shared_dispatch_stub_203

/proc/self/cwd/out/target/product/x86_64/obj_x86/SHARED_LIBRARIES/libglapi_intermediates/shared-glapi/glapi_mapi_tmp.h:18408

Next, i965_dri.so

root@UbuntuPV:/mnt/nfs/data/src/android-x86-nougat-x86-rc2# addr2line -Cfe ./out/target/product/x86_64/symbols/system/lib64/dri/i965_dri.so

0000000000064741

do_single_blorp_clear

/proc/self/cwd/external/mesa/src/mesa/drivers/dri/i965/brw_blorp.c:789

000000000006769a

brw_clear

/proc/self/cwd/external/mesa/src/mesa/drivers/dri/i965/brw_clear.c:233

I will check each lines of code.

 

2018년 1월 22일 월요일 오전 1시 54분 13초 UTC+9, Mauro Rossi 님의 말:

[YoungJoon Lee]

I found that mesa I have used is Not rc4. rc2 is used.

/external/mesa : https://github.com/maurossi/mesa/tree/17.2.0rc2_nougat-x86

2018년 1월 22일 월요일 오전 10시 32분 58초 UTC+9, YoungJoon Lee 님의 말:

2. Base : <a href="https://osdn.net/projects/android-x86/scm/git/manifest/commits/e8653d3a463c6894daff37073b35b2f880e59a0e" title="e8653d3a463c6894daff37073b35b2f880e59a0e" style="color:rgb(0,85,128);outline:0px;font-family:"Helvetica Neue",Helvetica,Arial,sans-serif;font-size:14.4px" rel="nofollow" target="_blank" onmousedo

[fguy]

I tested your latest build "nougat_x86_64_amd_dc_llvm50_mesa-17.3.3_hwc1.iso" on the skylake 6Y30. Unfortunately, the new versions of MESA and kernel did not fix the problem in 7.1.2 which appeared since November 2017. A number of applications and games (antutu 3dmark, defense zone 3) using the Unity at startup lead to a reboot of the android. This problem is not present in 8.1 and in October assemblies 7.1.2. 

This error occurs if "ro.hardware.sensor=kbd", if you assign "hdaps", then these applications run without problems.