Project: Kernel 3.8

Share here your ideas for new projects
luksol
Posts: 153
Joined: Tue Feb 26, 2013 5:01 am
languages_spoken: english, polish
ODROIDs: X2
Contact:

Re: Project: Kernel 3.8

Unread post by luksol » Tue May 21, 2013 12:08 am

mdrjr wrote:
luksol wrote:I hope you will feel better now. I was out for 2 weeks last month because of health issues so I know how you can feel.
Mali and HDMI done are really good news!
Is this TrustZone bug serious? Or is it nothing that next reboot wouldn't fix?
if you type reboot
when the kernel reach the state of sending the instruction to actually reboot it will kernel panic and you have to either reset it (X/X2) or remove and reconnect the power supply on U2.
It only happens if you have more them 1 CPU online..
Ah, ok. Not a show stopper for me, a bit of annoyance at most. But other people may find it a bit more difficult to accept. Anyway, I will stop pestering you, let recuperate in peace and continue your work :) Thanks for the update!

User avatar
Matt
Posts: 576
Joined: Tue Feb 26, 2013 4:57 pm
languages_spoken: English
ODROIDs: X2, U2, XU+E rev.2, Smart Power
Location: Charlottesville, VA, USA
Contact:

Re: Project: Kernel 3.8

Unread post by Matt » Tue May 21, 2013 2:08 am

mdrjr wrote:
luksol wrote:I hope you will feel better now. I was out for 2 weeks last month because of health issues so I know how you can feel.
Mali and HDMI done are really good news!
Is this TrustZone bug serious? Or is it nothing that next reboot wouldn't fix?
if you type reboot
when the kernel reach the state of sending the instruction to actually reboot it will kernel panic and you have to either reset it (X/X2) or remove and reconnect the power supply on U2.
It only happens if you have more them 1 CPU online..
It might be possible to work around the issue by modifying the reboot init script to set the scaling_governor to "powersave" right before it reboots. The powersave governor would be more likely to use only one cpu core.
ODROID Boards: U2, X2, XU+E rev.2
Accessories: VU, Smart Power, 13" LCD Kit (for the X2)

jcm
Posts: 5
Joined: Mon May 13, 2013 11:54 pm
languages_spoken: english
ODROIDs: Odroid-X, Odroid-X2
Contact:

Re: Project: Kernel 3.8

Unread post by jcm » Tue May 21, 2013 5:23 am

mdrjr wrote: - 3.8 isn't considered stable, there's are no official statement saying that it is from us. When its time and we thing its stable we'll release a image with 3.8 kernel. From that point on, you can consider it stable.
- There's no plan to move to 3.9 as it only breaks more stuff them what was broken on 3.8. (Audio, ARM GIC,.....)...
- If you are on a older -X image (aka. 3.6 kernel based) U-Boot changes a few bits and upgrading to a new version is not optional, as booting kernel 3.8 on that u-boot will probably render a non-booting board.

Thanks mdrjr for the reply. This clarifies a few things. Looking for a stable kernel supporting X and X2, it seems that it makes currently more sense for me to use the 3.0.y branch.

I did notice that the bootloader requires and update coming from the 3.6 kernel. 3.8 or 3.0 kernel is running successfully on my Odroid-X with the userland build by ptxdist. I'm going to receive the X2's soon. I was trying to find out if I can use the same u-boot for both X and X2. Doing a binary diff on the first part of an official X and X2 seems indicate that the bootloader is the same, but hard to be sure. Is there a difference in between the X and X2 u-boot? Is this documented/discussed somewhere?

mdrjr
Site Admin
Posts: 11698
Joined: Fri Feb 22, 2013 11:34 pm
languages_spoken: english, portuguese
ODROIDs: -
Location: Brazil
Contact:

Re: Project: Kernel 3.8

Unread post by mdrjr » Tue May 21, 2013 5:29 am

jcm wrote:
mdrjr wrote: - 3.8 isn't considered stable, there's are no official statement saying that it is from us. When its time and we thing its stable we'll release a image with 3.8 kernel. From that point on, you can consider it stable.
- There's no plan to move to 3.9 as it only breaks more stuff them what was broken on 3.8. (Audio, ARM GIC,.....)...
- If you are on a older -X image (aka. 3.6 kernel based) U-Boot changes a few bits and upgrading to a new version is not optional, as booting kernel 3.8 on that u-boot will probably render a non-booting board.

Thanks mdrjr for the reply. This clarifies a few things. Looking for a stable kernel supporting X and X2, it seems that it makes currently more sense for me to use the 3.0.y branch.

I did notice that the bootloader requires and update coming from the 3.6 kernel. 3.8 or 3.0 kernel is running successfully on my Odroid-X with the userland build by ptxdist. I'm going to receive the X2's soon. I was trying to find out if I can use the same u-boot for both X and X2. Doing a binary diff on the first part of an official X and X2 seems indicate that the bootloader is the same, but hard to be sure. Is there a difference in between the X and X2 u-boot? Is this documented/discussed somewhere?
We use the same u-boot for all 4412 boards. Doesn't have to change anything.
What you have to change between X and X2 is kernel.

jcm
Posts: 5
Joined: Mon May 13, 2013 11:54 pm
languages_spoken: english
ODROIDs: Odroid-X, Odroid-X2
Contact:

Re: Project: Kernel 3.8

Unread post by jcm » Tue May 21, 2013 5:43 am

mdrjr wrote: We use the same u-boot for all 4412 boards. Doesn't have to change anything.
What you have to change between X and X2 is kernel.
Thanks for the very quick reply!

marlemion
Posts: 232
Joined: Sun Mar 31, 2013 7:33 pm
languages_spoken: english
ODROIDs: Odroid U2
Contact:

Re: Project: Kernel 3.8

Unread post by marlemion » Thu May 23, 2013 6:41 pm

Tried to compile the latest sources today with g2d enabled, but it failed:

Code: Select all

  CC      drivers/gpu/drm/mali/mali_drv.o
  CC      drivers/gpu/drm/exynos/exynos_drm_g2d.o
In file included from include/linux/module.h:22:0,
                 from drivers/gpu/drm/mali/mali_drm.h:27,
                 from drivers/gpu/drm/mali/mali_drv.c:12:
/usr/src/linux-3.8.y/arch/arm/include/asm/module.h:42:0: warning: "MODULE_ARCH_VERMAGIC" redefined [enabled by default]
In file included from drivers/gpu/drm/mali/mali_drm.h:14:0,
                 from drivers/gpu/drm/mali/mali_drv.c:12:
include/linux/vermagic.h:25:0: note: this is the location of the previous definition
  CC      net/core/fib_rules.o
  CC      fs/posix_acl.o
  CC      drivers/gpu/drm/mali/mali_mm.o
  CC      fs/xattr_acl.o
In file included from include/linux/module.h:22:0,
                 from drivers/gpu/drm/mali/mali_drm.h:27,
                 from drivers/gpu/drm/mali/mali_mm.c:11:
/usr/src/linux-3.8.y/arch/arm/include/asm/module.h:42:0: warning: "MODULE_ARCH_VERMAGIC" redefined [enabled by default]
In file included from drivers/gpu/drm/mali/mali_drm.h:14:0,
                 from drivers/gpu/drm/mali/mali_mm.c:11:
include/linux/vermagic.h:25:0: note: this is the location of the previous definition
drivers/gpu/drm/mali/mali_mm.c: In function 'mali_idle':
drivers/gpu/drm/mali/mali_mm.c:199:6: warning: unused variable 'i' [-Wunused-variable]
drivers/gpu/drm/mali/mali_mm.c:198:16: warning: unused variable 'end' [-Wunused-variable]
drivers/gpu/drm/mali/mali_mm.c:197:11: warning: unused variable 'idle_reg' [-Wunused-variable]
drivers/gpu/drm/mali/mali_mm.c: At top level:
drivers/gpu/drm/mali/mali_mm.c:177:25: warning: 'mem_reg_init' defined but not used [-Wunused-function]
Is it supposed to fail, e.g. is g2d not working yet?

User avatar
memeka
Posts: 4143
Joined: Mon May 20, 2013 10:22 am
languages_spoken: english
ODROIDs: XU rev2 + eMMC + UART
U3 + eMMC + IO Shield + UART
Contact:

Re: Project: Kernel 3.8

Unread post by memeka » Thu May 23, 2013 6:51 pm

Sorry, I have no idea what I am looking at, but I only see warnings. Where does it fail? :D

marlemion
Posts: 232
Joined: Sun Mar 31, 2013 7:33 pm
languages_spoken: english
ODROIDs: Odroid U2
Contact:

Re: Project: Kernel 3.8

Unread post by marlemion » Thu May 23, 2013 7:16 pm

You're right, here is the actual problem. :)

Code: Select all

  CC      drivers/gpu/arm/mali/linux/mali_osk_atomics.o
  CC      drivers/gpu/arm/mali/linux/mali_osk_irq.o
  CC      drivers/gpu/arm/mali/linux/mali_osk_wq.o
  CC      drivers/gpu/arm/mali/linux/mali_osk_locks.o
  CC      drivers/gpu/arm/mali/linux/mali_osk_wait_queue.o
  CC      drivers/gpu/arm/mali/linux/mali_osk_low_level_mem.o
  CC      drivers/gpu/arm/mali/linux/mali_osk_math.o
  CC      drivers/gpu/arm/mali/linux/mali_osk_memory.o
  CC      drivers/gpu/arm/mali/linux/mali_osk_misc.o
  CC      drivers/gpu/arm/mali/linux/mali_osk_mali.o
  CC      drivers/gpu/arm/mali/linux/mali_osk_notification.o
  CC      drivers/gpu/arm/mali/linux/mali_osk_time.o
  CC      drivers/gpu/arm/mali/linux/mali_osk_timers.o
  CC      drivers/gpu/arm/mali/linux/mali_ukk_mem.o
  CC      drivers/gpu/arm/mali/linux/mali_ukk_gp.o
  CC      drivers/gpu/arm/mali/linux/mali_ukk_pp.o
  CC      drivers/gpu/arm/mali/linux/mali_ukk_core.o
  CC      drivers/gpu/arm/mali/common/mali_kernel_core.o
In file included from drivers/gpu/arm/mali/common/mali_osk_profiling.h:17:0,
                 from drivers/gpu/arm/mali/common/mali_kernel_core.c:34:
drivers/gpu/arm/mali/linux/mali_profiling_events.h:15:53: fatal error: linux/mali/mali_utgard_profiling_events.h: No such file or directory
compilation terminated.
make[4]: *** [drivers/gpu/arm/mali/common/mali_kernel_core.o] Error 1
make[3]: *** [drivers/gpu/arm/mali] Error 2
make[2]: *** [drivers/gpu/arm] Error 2
make[1]: *** [drivers/gpu] Error 2
make: *** [drivers] Error 2

marlemion
Posts: 232
Joined: Sun Mar 31, 2013 7:33 pm
languages_spoken: english
ODROIDs: Odroid U2
Contact:

Re: Project: Kernel 3.8

Unread post by marlemion » Thu May 23, 2013 7:22 pm

mali_utgard_profiling_events.h seems to be completely missing. I cannot find it either locally nor on the git web frontend of odroid-3.8.y

marlemion
Posts: 232
Joined: Sun Mar 31, 2013 7:33 pm
languages_spoken: english
ODROIDs: Odroid U2
Contact:

Re: Project: Kernel 3.8

Unread post by marlemion » Thu May 23, 2013 7:29 pm

There are others, though:

Code: Select all

find . |grep utgard
./arch/arm/mach-exynos/include/mach/mali_utgard.h
./drivers/gpu/arm/mali/include/linux/mali/mali_utgard_ioctl.h
./drivers/gpu/arm/mali/include/linux/mali/mali_utgard_counters.h
./drivers/gpu/arm/mali/include/linux/mali/mali_utgard_uk_types.h
./drivers/gpu/arm/mali/include/linux/mali/mali_utgard.h

User avatar
OverSun
Posts: 1414
Joined: Mon Apr 29, 2013 5:12 pm
languages_spoken: english
Contact:

Re: Project: Kernel 3.8

Unread post by OverSun » Thu May 23, 2013 7:53 pm

Turn off mali profiling in kernel config. You don't need it.

marlemion
Posts: 232
Joined: Sun Mar 31, 2013 7:33 pm
languages_spoken: english
ODROIDs: Odroid U2
Contact:

Re: Project: Kernel 3.8

Unread post by marlemion » Thu May 23, 2013 8:14 pm

Thanks, that worked. Must have been activated when trying to enable some mali features. What is it actually good for?

User avatar
OverSun
Posts: 1414
Joined: Mon Apr 29, 2013 5:12 pm
languages_spoken: english
Contact:

Re: Project: Kernel 3.8

Unread post by OverSun » Thu May 23, 2013 8:30 pm

It's a process to measure code performance.
Checks and points will be activated inside the code (if the author put them into it) and you can then gather some stats - how long does it takes to run the code between points, how many memory was used while doing this and that, you name it, etc.

mdrjr
Site Admin
Posts: 11698
Joined: Fri Feb 22, 2013 11:34 pm
languages_spoken: english, portuguese
ODROIDs: -
Location: Brazil
Contact:

Re: Project: Kernel 3.8

Unread post by mdrjr » Thu May 23, 2013 9:32 pm

Mental note: Remove profiling from CONFIG :)

marlemion
Posts: 232
Joined: Sun Mar 31, 2013 7:33 pm
languages_spoken: english
ODROIDs: Odroid U2
Contact:

Re: Project: Kernel 3.8

Unread post by marlemion » Thu May 23, 2013 9:38 pm

I think it was my fault, as I use genkernel with external configs to build the kernels...

marlemion
Posts: 232
Joined: Sun Mar 31, 2013 7:33 pm
languages_spoken: english
ODROIDs: Odroid U2
Contact:

Re: Project: Kernel 3.8

Unread post by marlemion » Fri May 24, 2013 3:15 am

Mali doesn't work with 3.8.13

When I use the mali DDX driver, I get:

[ 329.009] (EE) MALI(0): mali_drm_open_master DRM OPEN (fd: 0x9)
[ 329.009] (EE) MALI(0): mali_drm_open_master DRM failed to interface version

When I use the sunxifb DDX driver (at the moment the standard on my system), it starts but doesn't output anything. Tried fb0, fb1 and fb6. I wonder when this will ever work. Reverting to 3.0.75 for now.

mdrjr
Site Admin
Posts: 11698
Joined: Fri Feb 22, 2013 11:34 pm
languages_spoken: english, portuguese
ODROIDs: -
Location: Brazil
Contact:

Re: Project: Kernel 3.8

Unread post by mdrjr » Fri May 24, 2013 3:45 am

I love when people ditch my work... especially because they modified the provided config and had the CONFIG_DRM and CONFIG_DRM_MALI disabled :(

Yes, mali does work on 3.8.13. Its been work since this weekend. Please refer to the provided _defconfig if you are having troubles with your custom one.

marlemion
Posts: 232
Joined: Sun Mar 31, 2013 7:33 pm
languages_spoken: english
ODROIDs: Odroid U2
Contact:

Re: Project: Kernel 3.8

Unread post by marlemion » Fri May 24, 2013 4:17 am

I don't ditch anybodies work, I just report back my experiences.

Certainly I had enabled the DRM options for mali, but I also enabled the DRM option for exynos, which might be the source of the problem. What is 'DRM Support for Samsung SoC EXYNOS Series' useful for then?

I have disabled this option and am currently rebuilding the kernel.

marlemion
Posts: 232
Joined: Sun Mar 31, 2013 7:33 pm
languages_spoken: english
ODROIDs: Odroid U2
Contact:

Re: Project: Kernel 3.8

Unread post by marlemion » Sat May 25, 2013 5:33 pm

Just to let you know that it is still not working. I tried with my custom-exynos-disabled config and the config provided with the sources. Same behavior. If you startup X, you can see it starting properly on the command line, but no output on HDMI.

Here are some pieces of output (default config):

Code: Select all

dmesg|grep s5p-tv
[    0.000000] s5p-tv: HDMI_PHY_RES=1080
[    2.925000] s5p-tv: Board is ODROID-X/X2/U2
[    2.930000] s5p-tv: ODROID-U2 or X/X2 ByPass Jumper Mode
[    2.935000] s5p-tv: Selected V4L2_DV_1080P60 via software

Code: Select all

dmesg|grep Mali
[    2.360000] Mali<2>: Inserting Mali v19 device driver. 
[    2.365000] Mali<2>: Compiled: May 24 2013, time: 20:07:37.
[    2.370000] Mali<2>: Driver revision: 
[    2.375000] Mali<2>: mali_module_init() registering driver
[    2.380000] Mali<2>: mali_probe(): Called for platform device mali-utgard
[    2.385000] Mali<2>: Memory system initializing
[    2.390000] Mali<2>: Using device defined memory settings (dedicated: 0x00000000@0x00000000, shared: 0x40000000)
[    2.400000] Mali<2>: Mali OS memory allocator created with max allocation size of 0x40000000 bytes, cpu_usage_adjust 0x00000000
[    2.410000] Mali<2>: Using device defined frame buffer settings (0x00000000@0x00000000)
[    2.435000] Mali<2>: Mali PP: Creating Mali PP core: Mali_PP0
[    2.440000] Mali<2>: Mali PP: Base address of PP core: 0x13008000
[    2.445000] Mali<2>: Found Mali GPU Mali-400 MP r1p1
[    2.455000] Mali<2>: Mali L2 cache: Creating Mali L2 cache: Mali_L2
[    2.460000] Mali<2>: Mali MMU: Creating Mali MMU: Mali_GP_MMU
[    2.465000] Mali<2>: Mali GP: Creating Mali GP core: Mali_GP
[    2.470000] Mali<2>: Mali MMU: Creating Mali MMU: Mali_PP0_MMU
[    2.475000] Mali<2>: Mali PP: Creating Mali PP core: Mali_PP0
[    2.480000] Mali<2>: Mali PP: Base address of PP core: 0x13008000
[    2.485000] Mali<2>: Mali MMU: Creating Mali MMU: Mali_PP1_MMU
[    2.495000] Mali<2>: Mali PP: Creating Mali PP core: Mali_PP1
[    2.500000] Mali<2>: Mali PP: Base address of PP core: 0x1300a000
[    2.505000] Mali<2>: Mali MMU: Creating Mali MMU: Mali_PP2_MMU
[    2.510000] Mali<2>: Mali PP: Creating Mali PP core: Mali_PP2
[    2.515000] Mali<2>: Mali PP: Base address of PP core: 0x1300c000
[    2.525000] Mali<2>: Mali MMU: Creating Mali MMU: Mali_PP3_MMU
[    2.530000] Mali<2>: Mali PP: Creating Mali PP core: Mali_PP3
[    2.535000] Mali<2>: Mali PP: Base address of PP core: 0x1300e000
[    2.540000] Mali<2>: 4+0 PP cores initialized
[    2.545000] Mali<2>: Mali GPU Utilization: No utilization handler installed
[    2.550000] Mali<2>: mali_probe(): Successfully initialized driver for platform device mali-utgard
[    2.575000] Mali: Mali device driver loaded

Code: Select all

dmesg|grep s3c-fb
[    1.275000] s3c-fb exynos4-fb.0: start latency exceeded, new value 625 ns
[    1.275000] s3c-fb exynos4-fb.0: state restore latency exceeded, new value 31459 ns
[    1.305000] s3c-fb exynos4-fb.0: window 0: fb 
[  580.455000] s3c-fb: GET_UMP_SECURE_ID_BUF1 called
[  580.455000] s3c-fb: GET_UMP_SECURE_ID_BUF2 called
The last two lines appeared when starting X.

Xorg.0.log with sunxifb:

Code: Select all

[  1184.003] (II) LoadModule: "sunxifb"
[  1184.006] (II) Loading /usr/lib/xorg/modules/drivers/sunxifb_drv.so
[  1184.010] (II) Module sunxifb: vendor="X.Org Foundation"
[  1184.010]    compiled for 1.13.4, module version = 0.3.1
[  1184.011]    Module class: X.Org Video Driver
[  1184.011]    ABI class: X.Org Video Driver, version 13.1
[  1184.011] (II) SUNXIFB: driver for framebuffer: sunxifb
[  1184.012] (--) using VT number 7
[  1184.049] (WW) Falling back to old probe method for sunxifb
[  1184.049] (II) Loading sub module "fbdevhw"
[  1184.049] (II) LoadModule: "fbdevhw"
[  1184.050] (II) Loading /usr/lib/xorg/modules/libfbdevhw.so
[  1184.051] (II) Module fbdevhw: vendor="X.Org Foundation"
[  1184.051]    compiled for 1.13.4, module version = 0.0.2
[  1184.051]    ABI class: X.Org Video Driver, version 13.1
[  1184.052] (II) SUNXIFB(0): using /dev/fb0
[  1184.052] (WW) VGA arbiter: cannot open kernel arbiter, no multi-card support
[  1184.052] (II) SUNXIFB(0): Creating default Display subsection in Screen section
        "Mali-Screen" for depth/fbbpp 24/32
[  1184.052] (**) SUNXIFB(0): Depth 24, (--) framebuffer bpp 32
[  1184.052] (==) SUNXIFB(0): RGB weight 888
[  1184.053] (==) SUNXIFB(0): Default visual is TrueColor
[  1184.053] (==) SUNXIFB(0): Using gamma correction (1.0, 1.0, 1.0)
[  1184.053] (II) SUNXIFB(0): hardware:  (video memory: 4080kB)
[  1184.053] (**) SUNXIFB(0): Option "fbdev" "/dev/fb0"
[  1184.053] (**) SUNXIFB(0): Option "DRI2" "true"
[  1184.054] (**) SUNXIFB(0): Option "DRI2HWOverlay" "true"
[  1184.055] (II) SUNXIFB(0): processor: ARM Cortex-A9
[  1184.055] (II) SUNXIFB(0): checking modes against framebuffer device...
[  1184.055] (II) SUNXIFB(0): checking modes against monitor...
[  1184.055] (--) SUNXIFB(0): Virtual size is 1360x768 (pitch 1360)
[  1184.055] (**) SUNXIFB(0):  Built-in mode "current": 72.1 MHz, 47.4 kHz, 60.0 Hz
[  1184.056] (II) SUNXIFB(0): Modeline "current"x0.0   72.05  1360 1408 1440 1520  768 771 776 790 -hsync -vsync -csync (47.4 kHz b)
[  1184.056] (==) SUNXIFB(0): DPI set to (96, 96)
[  1184.056] (II) Loading sub module "fb"
[  1184.056] (II) LoadModule: "fb"
[  1184.058] (II) Loading /usr/lib/xorg/modules/libfb.so
[  1184.060] (II) Module fb: vendor="X.Org Foundation"
[  1184.060]    compiled for 1.13.4, module version = 1.0.0
[  1184.060]    ABI class: X.Org ANSI C Emulation, version 0.4
[  1184.060] (==) Depth 24 pixmap format is 32 bpp
[  1184.080] (II) SUNXIFB(0): can't load 'g2d_23' kernel module
[  1184.080] (II) SUNXIFB(0): failed to enable the use of sunxi display controller
[  1184.080] (II) SUNXIFB(0): no 2D acceleration selected via AccelMethod option
[  1184.081] (II) SUNXIFB(0): enabled NEON optimizations
[  1184.081] (==) SUNXIFB(0): Backing store disabled
[  1184.085] (==) SUNXIFB(0): DPMS enabled
[  1184.085] (II) SUNXIFB(0): failed to enable hardware cursor
[  1184.119] (II) Loading sub module "dri2"
[  1184.119] (II) LoadModule: "dri2"
[  1184.120] (II) Module "dri2" already built-in
[  1184.121] (II) SUNXIFB(0): display controller hardware overlays can't be used for DRI2
[  1184.121] (II) SUNXIFB(0): [DRI2] Setup complete
[  1184.121] (II) SUNXIFB(0): [DRI2]   DRI driver: sunxi-mali
[  1184.121] (II) SUNXIFB(0): using DRI2 integration for Mali GPU (UMP buffers)
[  1184.122] (II) SUNXIFB(0): Mali binary drivers can only accelerate EGL/GLES
[  1184.122] (II) SUNXIFB(0): so AIGLX/GLX is expected to fail or fallback to software
[  1184.122] (==) RandR enabled
Xorg.0.log with mali:

Code: Select all

[  1324.336] (II) LoadModule: "mali"
[  1324.337] (II) Loading /usr/lib/xorg/modules/extensions/mali_drv.so
[  1324.346] (II) Module mali: vendor="X.Org Foundation"
[  1324.346]    compiled for 1.13.4, module version = 0.4.2
[  1324.346]    ABI class: X.Org Video Driver, version 13.1
[  1324.346] MaliSetup
[  1324.347] MaliIdentify
[  1324.347] (II) MALI: driver for Mali Framebuffer: mali
[  1324.347] (--) using VT number 7
[  1324.384] (WW) Falling back to old probe method for mali
[  1324.384] MaliProbe
[  1324.385] MaliHWProbe
[  1324.385] (II) MALI(0): using /dev/fb0
[  1324.385] (WW) VGA arbiter: cannot open kernel arbiter, no multi-card support
[  1324.385] (II) MALI(0): MaliPreInit: ENTER
[  1324.386] (II) MALI(0): MaliGetRec: ENTER
[  1324.386] (II) MALI(0): MaliHWInit: ENTER
[  1324.386] (II) MALI(0): MaliHWGetDepth: ENTER
[  1324.386] (II) MALI(0): Creating default Display subsection in Screen section
        "Mali-Screen" for depth/fbbpp 24/32
[  1324.386] (**) MALI(0): Depth 24, (--) framebuffer bpp 32
[  1324.387] (==) MALI(0): RGB weight 888
[  1324.387] (==) MALI(0): Default visual is TrueColor
[  1324.387] (==) MALI(0): Using gamma correction (1.0, 1.0, 1.0)
[  1324.387] (II) MALI(0): MaliHWGetVidmem: ENTER
[  1324.387] (II) MALI(0): MaliHWGetName: ENTER
[  1324.387] (II) MALI(0): hardware:  (video memory: 4080kB)
[  1324.387] (**) MALI(0): Option "DRI2" "true"
[  1324.388] (**) MALI(0): Option "DRI2_PAGE_FLIP" "true"
[  1324.388] (**) MALI(0): Option "DRI2_WAIT_VSYNC" "false"
[  1324.388] (**) MALI(0): Option "UMP_CACHED" "true"
[  1324.389] (**) MALI(0): Option "UMP_LOCK" "true"
[  1324.389] (**) MALI(0): DRI Fullscreen page flip enabled
[  1324.389] (**) MALI(0): DRI Fullscreen page flip VSYNC disabled
[  1324.389] (**) MALI(0): Use cached UMP memory
[  1324.389] (**) MALI(0): Use umplock across processes
[  1324.390] (EE) MALI(0): mali_drm_open_master DRM OPEN (fd: 0x8)
[  1324.390] (**) MALI(0): DRI Fullscreen page flip enabled
[  1324.391] (**) MALI(0): DRI Fullscreen page flip VSYNC disabled
[  1324.391] (**) MALI(0): Use cached UMP memory
[  1324.391] (**) MALI(0): Use umplock across processes
[  1324.391] (II) MALI(0): MaliHWGetFD: ENTER
[  1324.391] (II) MALI(0): FBDev_crtc_config: ENTER
[  1324.391] (II) MALI(0): FBDev_crtc_config: EXIT
[  1324.391] (II) MALI(0): Output LCD has no monitor section
[  1324.392] (II) MALI(0): Mode 1360 x 768 valid
[  1324.392] (II) MALI(0): Printing probed modes for output LCD
[  1324.392] (II) MALI(0): Modeline "1360x768"x60.0   73.27  1360 1380 1400 1440  768 788 808 848 (50.9 kHz e)
[  1324.392] (II) MALI(0): Output LCD connected
[  1324.392] (II) MALI(0): Using sloppy heuristic for initial modes
[  1324.393] (II) MALI(0): Output LCD using initial mode 1360x768
[  1324.393] (II) MALI(0): Using default gamma of (1.0, 1.0, 1.0) unless otherwise stated.
[  1324.393] (II) MALI(0): MaliHWSetVideoModes: ENTER
[  1324.393] (--) MALI(0): Virtual size is 1360x768 (pitch 1360)
[  1324.393] (**) MALI(0):  Driver mode "1360x768": 73.3 MHz (scaled from 0.0 MHz), 50.9 kHz, 60.0 Hz
[  1324.393] (II) MALI(0): Modeline "1360x768"x60.0   73.27  1360 1380 1400 1440  768 788 808 848 (50.9 kHz e)
[  1324.394] (==) MALI(0): DPI set to (96, 96)
[  1324.394] (II) Loading sub module "fb"
[  1324.394] (II) LoadModule: "fb"
[  1324.396] (II) Loading /usr/lib/xorg/modules/libfb.so
[  1324.398] (II) Module fb: vendor="X.Org Foundation"
[  1324.398]    compiled for 1.13.4, module version = 1.0.0
[  1324.398]    ABI class: X.Org ANSI C Emulation, version 0.4
[  1324.398] (==) Depth 24 pixmap format is 32 bpp
[  1324.399] (II) MALI(0): MaliScreenInit: ENTER
[  1324.399] (II) MALI(0): DRI2 version: 1.2
[  1324.399] (II) MALI(0): [DRI2] Setup complete
[  1324.399] (II) MALI(0): [DRI2]   DRI driver: Mali DRI2
[  1324.399] (II) MALI(0): MaliHWMapVidmem: ENTER
[  1324.400] (II) MALI(0): MaliHWLinearOffset: ENTER
[  1324.400] (II) MALI(0): MaliHWSave: ENTER
[  1324.400] (II) MALI(0): MaliHWModeInit: ENTER
[  1324.400] (II) MALI(0): MaliHWSetMode: ENTER
[  1324.401] (II) MALI(0): MaliHWSaveScreen: ENTER
[  1324.401] (II) MALI(0): MaliHWAdjustFrame: ENTER
[  1324.402] (II) Loading sub module "exa"
[  1324.402] (II) LoadModule: "exa"
[  1324.402] (II) Loading /usr/lib/xorg/modules/libexa.so
[  1324.403] (II) Module exa: vendor="X.Org Foundation"
[  1324.403]    compiled for 1.13.4, module version = 2.6.0
[  1324.403]    ABI class: X.Org Video Driver, version 13.1
[  1324.403] (II) MALI(0): maliSetupExa: ENTER
[  1324.404] (II) MALI(0): Mali EXA driver is loaded successfully
[  1324.404] (II) MALI(0): maliSetupExa: EXIT
[  1324.404] (WW) MALI(0): Initializing EXA Driver!
[  1324.404] (II) EXA(0): Driver allocated offscreen pixmaps
[  1324.404] (II) EXA(0): Driver registered support for the following operations:
[  1324.404] (II)         Solid
[  1324.404] (II)         Copy
[  1324.405] (II)         Composite (RENDER acceleration)
[  1324.405] (==) MALI(0): Backing store disabled
[  1324.405] (==) MALI(0): Silken mouse enabled
[  1324.405] (II) MALI(0): Changing mode to 1360 768 1360 1536
[  1324.406] (II) MALI(0): RandR 1.2 enabled, ignore the following RandR disabled message.
[  1324.407] (==) MALI(0): DPMS enabled
[  1324.407] (WW) MALI(0): Failed to open umplock device!
[  1324.407] (--) RandR disabled
[  1324.480] GET_UMP_SECURE_ID_BUF1 returned 0x3 offset: 0 virt address: 0xb64d3000 fb_virt: 0xb64d3000
[  1324.484] GET_UMP_SECURE_ID_BUF2 returned 0x4 offset: 4177920 virt address: (nil) fb_virt: 0xb64d3000
[  1324.484] (II) MALI(0): Creating FRAMEBUFFER pixmap 0x1f33b8 at offset 4177920, privPixmap=0x1f3448
[  1324.484] (II) MALI(0): Creating FRAMEBUFFER pixmap 0x1f3278 at offset 0, privPixmap=0x1f3308
[  1324.485] (II) MALI(0): Setting screen physical size to 359 x 203
[  1324.485] (II) MALI(0): fbdev_crtc_config_resize: width = 1360 height = 768
[  1324.485] (II) MALI(0): Changing mode to 1360 768 1360 1536
Please note that in both cases, it does not even get the resolution right (I am booting with 1080p).

/etc/X11/xorg.d/05-odroid.conf

Code: Select all

Section "Device"
        Identifier "Mali-Fbdev"
        #Driver "fbdev"
        Driver  "mali"
        #Driver "sunxifb"
        Option  "fbdev"                 "/dev/fb0"
        Option  "DRI2"                  "true"
        # mali only
        Option  "DRI2_PAGE_FLIP"        "true"
        Option  "DRI2_WAIT_VSYNC"       "false"
        Option  "UMP_CACHED"            "true"
        Option  "UMP_LOCK"              "true"
        # sunxifb only
        Option  "DRI2HWOverlay"         "true"
        #Option  "AccelMethod"          "EXA"
EndSection

Section "Screen"
        Identifier   "Mali-Screen"
        Device       "Mali-Fbdev"
        DefaultDepth  24
EndSection

Section "DRI"
        Mode 0666
EndSection
Xorg server is 1.13.4, mali is the latest version from the ARM website, sunxifb is the latest git. EGL drivers are those taken from the odroid download page (latest, for 1.13 Xorg server).

RIght now confirmed: Changed the kernel in /boot to 3.0.75, changed /dev/fb0 to /dev/fb6 in the Xorg configuration, rebooted and the odroid went straight into XBMC.

User avatar
OverSun
Posts: 1414
Joined: Mon Apr 29, 2013 5:12 pm
languages_spoken: english
Contact:

Re: Project: Kernel 3.8

Unread post by OverSun » Sat May 25, 2013 10:51 pm

/dev/fb0 is an LCD output. You can be greatly surprised if you try to connect to your device via vnc (if you have x11vnc) and find out that there Is actually X11 running with this resolution. You just don't have this outputs where it is showing itself.
/dev/fb1 is HDMI

marlemion
Posts: 232
Joined: Sun Mar 31, 2013 7:33 pm
languages_spoken: english
ODROIDs: Odroid U2
Contact:

Re: Project: Kernel 3.8

Unread post by marlemion » Sat May 25, 2013 11:54 pm

Well, that did the trick, thanks. Now I will have to add everything I need back to the kernel configuration (e.g. usb-audio).

somedolphin
Posts: 4
Joined: Mon May 13, 2013 11:39 pm
languages_spoken: english
Contact:

Re: Project: Kernel 3.8

Unread post by somedolphin » Sun May 26, 2013 12:15 am

Just want to thank everyone who's been working on this port (especially Mdrjr)! I was able to get less then 2ms audio latency with the build in sound card and jackd :) It was crashing hard on the 3.0 kernel. Now I'll try to start porting the mixer controls from the 3.9 branch and we should have one hell of an effects box :) Thanks again, ike

mdrjr
Site Admin
Posts: 11698
Joined: Fri Feb 22, 2013 11:34 pm
languages_spoken: english, portuguese
ODROIDs: -
Location: Brazil
Contact:

Re: Project: Kernel 3.8

Unread post by mdrjr » Sun May 26, 2013 3:06 am

somedolphin wrote:Just want to thank everyone who's been working on this port (especially Mdrjr)! I was able to get less then 2ms audio latency with the build in sound card and jackd :) It was crashing hard on the 3.0 kernel. Now I'll try to start porting the mixer controls from the 3.9 branch and we should have one hell of an effects box :) Thanks again, ike
You are welcome :)

sboyce
Posts: 158
Joined: Thu Feb 28, 2013 7:17 am
languages_spoken: english
ODROIDs: ODROID-C2
Location: Halesowen, West Midlands, UK
Contact:

Re: Project: Kernel 3.8

Unread post by sboyce » Mon May 27, 2013 12:12 am

When I built 3.0.79 I got the same error below installing the .deb's. I had to change Makefile to EXTRAVERSION = -odroidx and 3.0.79 was happy with that.
The same happened while installing 3.8.13, so I altered Makefile the same, rebuilt as before and got the same error.

I started off with "make odroidx_defconfig" and didn't alter anything in the generated .config.

root@odroid:/2/linux-3.8.13# dpkg -i ../linux-headers-3.8.13-odroidx+_3.8.13-odroidx+-10.00.Custom_armhf.deb ../linux-image-3.8.13-odroidx+_3.8.13-odroidx+-10.00.Custom_armhf.deb
Selecting previously unselected package linux-headers-3.8.13-odroidx+.
(Reading database ... 299784 files and directories currently installed.)
Unpacking linux-headers-3.8.13-odroidx+ (from .../linux-headers-3.8.13-odroidx+_3.8.13-odroidx+-10.00.Custom_armhf.deb) ...
Selecting previously unselected package linux-image-3.8.13-odroidx+.
Unpacking linux-image-3.8.13-odroidx+ (from .../linux-image-3.8.13-odroidx+_3.8.13-odroidx+-10.00.Custom_armhf.deb) ...
Examining /etc/kernel/preinst.d/
Done.
Setting up linux-headers-3.8.13-odroidx+ (3.8.13-odroidx+-10.00.Custom) ...
Examining /etc/kernel/header_postinst.d.
Setting up linux-image-3.8.13-odroidx+ (3.8.13-odroidx+-10.00.Custom) ...
Running depmod.
Examining /etc/kernel/postinst.d.
run-parts: executing /etc/kernel/postinst.d/apt-auto-removal 3.8.13-odroidx+ /boot/vmlinuz-3.8.13-odroidx+
run-parts: executing /etc/kernel/postinst.d/initramfs-tools 3.8.13-odroidx+ /boot/vmlinuz-3.8.13-odroidx+
update-initramfs: Generating /boot/initrd.img-3.8.13-odroidx+
run-parts: executing /etc/kernel/postinst.d/pm-utils 3.8.13-odroidx+ /boot/vmlinuz-3.8.13-odroidx+
run-parts: executing /etc/kernel/postinst.d/update-notifier 3.8.13-odroidx+ /boot/vmlinuz-3.8.13-odroidx+
run-parts: executing /etc/kernel/postinst.d/zz-flash-kernel 3.8.13-odroidx+ /boot/vmlinuz-3.8.13-odroidx+
Kernel /boot/vmlinuz-3.8.13-odroidx+ does not match your subarchitecture
odroidx, therefore not writing it to flash.
root@odroid:/2/linux-3.8.13#

root@odroid:/2/linux-3.8.13# ls -l /boot/*3.8.13*
-rw-r--r-- 1 root root 90215 May 26 14:43 /boot/config-3.8.13-odroidx+
-rw-r--r-- 1 root root 3718291 May 26 15:03 /boot/initrd.img-3.8.13-odroidx+
-rw-r--r-- 1 root root 1284410 May 26 14:52 /boot/System.map-3.8.13-odroidx+
-rw-r--r-- 1 root root 3223608 May 26 14:52 /boot/vmlinuz-3.8.13-odroidx+

User avatar
OverSun
Posts: 1414
Joined: Mon Apr 29, 2013 5:12 pm
languages_spoken: english
Contact:

Re: Project: Kernel 3.8

Unread post by OverSun » Mon May 27, 2013 1:10 am

When the kernel script decides that kernel you build is not "original" one and differs from the 3.something.something forward a little it adds "+" sign at the end.
Long story short, it is too complicated to convince it that everything is ok, just edit 'scripts/setkernelversion', around line 170 there is something like
if test "${LOCALVERSION+set}" != "set"; then
scm=$(scm_version --short)
res="$res${scm:++}"
fi

comment out
# res="$res${scm:++}"

and you will be ok. just do make again.

sboyce
Posts: 158
Joined: Thu Feb 28, 2013 7:17 am
languages_spoken: english
ODROIDs: ODROID-C2
Location: Halesowen, West Midlands, UK
Contact:

Re: Project: Kernel 3.8

Unread post by sboyce » Mon May 27, 2013 4:32 am

OvverSun wrote:When the kernel script decides that kernel you build is not "original" one and differs from the 3.something.something forward a little it adds "+" sign at the end.
Long story short, it is too complicated to convince it that everything is ok, just edit 'scripts/setkernelversion', around line 170 there is something like
if test "${LOCALVERSION+set}" != "set"; then
scm=$(scm_version --short)
res="$res${scm:++}"
fi

comment out
# res="$res${scm:++}"

and you will be ok. just do make again.
There are no lines like that in Makefile. May be the problem is here.

SUBARCH := $(shell uname -m | sed -e s/i.86/i386/ -e s/sun4u/sparc64/ \
-e s/arm.*/arm/ -e s/sa110/arm/ \
-e s/s390x/s390/ -e s/parisc64/parisc/ \
-e s/ppc.*/powerpc/ -e s/mips.*/mips/ \
-e s/sh[234].*/sh/ -e s/aarch64.*/arm64/ )

root@odroid:/2/linux-3.8.13# uname -m
armv7l

root@odroid:/2/linux-3.8.13# uname -m | sed -e s/arm.*/arm/
arm

I wonder if it's a false error as it writes the files to /boot.

User avatar
OverSun
Posts: 1414
Joined: Mon Apr 29, 2013 5:12 pm
languages_spoken: english
Contact:

Re: Project: Kernel 3.8

Unread post by OverSun » Mon May 27, 2013 4:51 am

Who said Makefile?
I said 'scripts/setlocalversion'

PS. 'scripts/setkernelversion' actually, but that was just a typo, I was writing from memory.

marlemion
Posts: 232
Joined: Sun Mar 31, 2013 7:33 pm
languages_spoken: english
ODROIDs: Odroid U2
Contact:

Re: Project: Kernel 3.8

Unread post by marlemion » Mon May 27, 2013 5:14 am

Well after testing the 3.8 kernel for a few hours I can say that mali works ok, the mfc does definitely work even though it is quite useless without 2d acceleration right now.

As I am on gentoo, I compile a lot. Since using the 3.8 kernel, I get random segmentation faults when compiling. Random in the sense of failing sometimes, sometimes not, always in different stages of the compilation process. Currently, I am compiling with 4 threads, but i might go down to two and see, what will happen. I am using the ondemand governor, so may it is related to cpu throttling. This is the stock config just with snd-usb-audio enabled (and loaded). Did anybody else encounter such problems?

User avatar
OverSun
Posts: 1414
Joined: Mon Apr 29, 2013 5:12 pm
languages_spoken: english
Contact:

Re: Project: Kernel 3.8

Unread post by OverSun » Mon May 27, 2013 5:36 am

What is your impression about MFC is based on? Do you have a working code example?
Because I'm starving to get it working right.

User avatar
neagix
Posts: 271
Joined: Sun Apr 14, 2013 2:54 am
languages_spoken: english
ODROIDs: U2
Contact:

Re: Project: Kernel 3.8

Unread post by neagix » Mon May 27, 2013 5:44 am

(I know I will repent of posting this)

For those interested, the 3.8.13 uInitrd and zImage (for U2) are available here:

http://oph.mdrjr.net/neagix/uInitrd-3.8.13
http://oph.mdrjr.net/neagix/zImage-3.8.13

These have been compiled about 1 week ago with the odroidu2_defconfig kernel configuration file.

Just for testing! Remember: these are not officially provided by Hardkernel!

sboyce
Posts: 158
Joined: Thu Feb 28, 2013 7:17 am
languages_spoken: english
ODROIDs: ODROID-C2
Location: Halesowen, West Midlands, UK
Contact:

Re: Project: Kernel 3.8

Unread post by sboyce » Mon May 27, 2013 6:34 am

OvverSun wrote:Who said Makefile?
I said 'scripts/setlocalversion'

PS. 'scripts/setkernelversion' actually, but that was just a typo, I was writing from memory.
Thanks, now building.

sboyce
Posts: 158
Joined: Thu Feb 28, 2013 7:17 am
languages_spoken: english
ODROIDs: ODROID-C2
Location: Halesowen, West Midlands, UK
Contact:

Re: Project: Kernel 3.8

Unread post by sboyce » Mon May 27, 2013 6:38 am

marlemion wrote:Well after testing the 3.8 kernel for a few hours I can say that mali works ok, the mfc does definitely work even though it is quite useless without 2d acceleration right now.

As I am on gentoo, I compile a lot. Since using the 3.8 kernel, I get random segmentation faults when compiling. Random in the sense of failing sometimes, sometimes not, always in different stages of the compilation process. Currently, I am compiling with 4 threads, but i might go down to two and see, what will happen. I am using the ondemand governor, so may it is related to cpu throttling. This is the stock config just with snd-usb-audio enabled (and loaded). Did anybody else encounter such problems?
I would be interested to see if your xorg.conf for mali400 looks like mine.
# cat etc/X11/xorg.conf_mali
# X.Org X server configuration file for xfree86-video-mali

Section "Device"
Identifier "Mali FBDEV"
Driver "mali"
Option "fbdev" "/dev/fb1"
Option "DRI2" "true"
Option "DRI2" "true"
Option "DRI2_PAGE_FLIP" "true"
Option "DRI2_WAIT_VSYNC" "false"
EndSection

Section "Screen"
Identifier "Mali Screen"
Device "Mali FBDEV"
DefaultDepth 24
EndSection

Section "DRI"
Mode 0666
EndSection

User avatar
memeka
Posts: 4143
Joined: Mon May 20, 2013 10:22 am
languages_spoken: english
ODROIDs: XU rev2 + eMMC + UART
U3 + eMMC + IO Shield + UART
Contact:

Re: Project: Kernel 3.8

Unread post by memeka » Mon May 27, 2013 7:41 am

neagix wrote:(I know I will repent of posting this)

For those interested, the 3.8.13 uInitrd and zImage (for U2) are available here:

http://oph.mdrjr.net/neagix/uInitrd-3.8.13
http://oph.mdrjr.net/neagix/zImage-3.8.13

These have been compiled about 1 week ago with the odroidu2_defconfig kernel configuration file.

Just for testing! Remember: these are not officially provided by Hardkernel!

Is retroarch working with 3.8.13 hw accelerated?

User avatar
neagix
Posts: 271
Joined: Sun Apr 14, 2013 2:54 am
languages_spoken: english
ODROIDs: U2
Contact:

Re: Project: Kernel 3.8

Unread post by neagix » Mon May 27, 2013 8:06 am

marlemion wrote:Well after testing the 3.8 kernel for a few hours I can say that mali works ok, the mfc does definitely work even though it is quite useless without 2d acceleration right now.

As I am on gentoo, I compile a lot. Since using the 3.8 kernel, I get random segmentation faults when compiling. Random in the sense of failing sometimes, sometimes not, always in different stages of the compilation process. Currently, I am compiling with 4 threads, but i might go down to two and see, what will happen. I am using the ondemand governor, so may it is related to cpu throttling. This is the stock config just with snd-usb-audio enabled (and loaded). Did anybody else encounter such problems?
Did you ever get any GCC corruption message? See also this thread of mine. If yes, it could be indicator of power issues, however in that case you would certainly experience them with any version of the kernel, not specifically 3.8

User avatar
neagix
Posts: 271
Joined: Sun Apr 14, 2013 2:54 am
languages_spoken: english
ODROIDs: U2
Contact:

Re: Project: Kernel 3.8

Unread post by neagix » Mon May 27, 2013 8:07 am

memeka wrote:
neagix wrote:(I know I will repent of posting this)

For those interested, the 3.8.13 uInitrd and zImage (for U2) are available here:

http://oph.mdrjr.net/neagix/uInitrd-3.8.13
http://oph.mdrjr.net/neagix/zImage-3.8.13

These have been compiled about 1 week ago with the odroidu2_defconfig kernel configuration file.

Just for testing! Remember: these are not officially provided by Hardkernel!

Is retroarch working with 3.8.13 hw accelerated?
Keep this on topic, or mdrjr will kick us :)

I couldn't verify it, however I can't think of any reason why it should not work - I will soon compile 3.8 again (the version above was compiled before some recent UMP fix) and try

User avatar
memeka
Posts: 4143
Joined: Mon May 20, 2013 10:22 am
languages_spoken: english
ODROIDs: XU rev2 + eMMC + UART
U3 + eMMC + IO Shield + UART
Contact:

Re: Project: Kernel 3.8

Unread post by memeka » Mon May 27, 2013 8:54 am

neagix wrote:
memeka wrote:Is retroarch working with 3.8.13 hw accelerated?
Keep this on topic, or mdrjr will kick us :)
It is on topic, it's about the 3.8.y kernel :P
Thanks for the answer.

sboyce
Posts: 158
Joined: Thu Feb 28, 2013 7:17 am
languages_spoken: english
ODROIDs: ODROID-C2
Location: Halesowen, West Midlands, UK
Contact:

Re: Project: Kernel 3.8

Unread post by sboyce » Mon May 27, 2013 9:04 am

Selecting previously unselected package linux-headers-3.8.13.
(Reading database ... 311654 files and directories currently installed.)
Unpacking linux-headers-3.8.13 (from .../linux-headers-3.8.13_3.8.13-10.00.Custom_armhf.deb) ...
Selecting previously unselected package linux-image-3.8.13.
Unpacking linux-image-3.8.13 (from .../linux-image-3.8.13_3.8.13-10.00.Custom_armhf.deb) ...
Examining /etc/kernel/preinst.d/
Done.
Setting up linux-headers-3.8.13 (3.8.13-10.00.Custom) ...
Examining /etc/kernel/header_postinst.d.
Setting up linux-image-3.8.13 (3.8.13-10.00.Custom) ...
Running depmod.
Examining /etc/kernel/postinst.d.
run-parts: executing /etc/kernel/postinst.d/apt-auto-removal 3.8.13 /boot/vmlinuz-3.8.13
run-parts: executing /etc/kernel/postinst.d/initramfs-tools 3.8.13 /boot/vmlinuz-3.8.13
update-initramfs: Generating /boot/initrd.img-3.8.13
run-parts: executing /etc/kernel/postinst.d/pm-utils 3.8.13 /boot/vmlinuz-3.8.13
run-parts: executing /etc/kernel/postinst.d/update-notifier 3.8.13 /boot/vmlinuz-3.8.13
run-parts: executing /etc/kernel/postinst.d/zz-flash-kernel 3.8.13 /boot/vmlinuz-3.8.13
Kernel /boot/vmlinuz-3.8.13 does not match your subarchitecture
odroidx, therefore not writing it to flash.

From scripts/setlocalversion
------------------------------------------
else
# append a plus sign if the repository is not in a clean
# annotated or signed tagged state (as git describe only
# looks at signed or annotated tags - git tag -a/-s) and
# LOCALVERSION= is not specified
if test "${LOCALVERSION+set}" != "set"; then
scm=$(scm_version --short)
#### res="$res${scm:++}"

root@odroid:/2/linux-3.8.13# grep ODROID .config
CONFIG_ODROID_X=y
# CONFIG_ODROID_X2 is not set
# CONFIG_ODROID_U2 is not set
# CONFIG_ODROID_X_X2_BYPASS_HDMI_JUMPER is not set

KC7NOA
Posts: 843
Joined: Wed Feb 27, 2013 2:25 pm
languages_spoken: english
ODROIDs: Odroid-X2(blown HDMI output)
Odroid-C2 to replace X2
Contact:

Re: Project: Kernel 3.8

Unread post by KC7NOA » Mon May 27, 2013 10:06 am

neagix wrote:(I know I will repent of posting this)

For those interested, the 3.8.13 uInitrd and zImage (for U2) are available here:

http://oph.mdrjr.net/neagix/uInitrd-3.8.13
http://oph.mdrjr.net/neagix/zImage-3.8.13

These have been compiled about 1 week ago with the odroidu2_defconfig kernel configuration file.

Just for testing! Remember: these are not officially provided by Hardkernel!

X or X2 .. i have both ..

User avatar
OverSun
Posts: 1414
Joined: Mon Apr 29, 2013 5:12 pm
languages_spoken: english
Contact:

Re: Project: Kernel 3.8

Unread post by OverSun » Mon May 27, 2013 2:11 pm

sboyce wrote: Kernel /boot/vmlinuz-3.8.13 does not match your subarchitecture
odroidx, therefore not writing it to flash.
Oh common, now set extraversion to -odroidx. This script is just waiting for the kernel name to end by "-odroidx"

marlemion
Posts: 232
Joined: Sun Mar 31, 2013 7:33 pm
languages_spoken: english
ODROIDs: Odroid U2
Contact:

Re: Project: Kernel 3.8

Unread post by marlemion » Mon May 27, 2013 2:48 pm

I use xbmc git to test the MFC. There is a huge performance improvement regarding video when using the 3.8 kernel with the firmware. However, 2d acceleration is not ready.

The mali DDX lets xbmc crash, I use sunxifb instead:

Code: Select all

Section "Device"
        Identifier "Mali-Fbdev"
        #Driver "fbdev"
        #Driver  "mali"
        Driver "sunxifb"
        Option  "fbdev"                 "/dev/fb1"
        Option  "DRI2"                  "true"
        # mali only
        Option  "DRI2_PAGE_FLIP"        "true"
        Option  "DRI2_WAIT_VSYNC"       "false"
        Option  "UMP_CACHED"            "true"
        Option  "UMP_LOCK"              "true"
        # sunxifb only
        Option  "DRI2HWOverlay"         "true"
        #Option  "AccelMethod"          "G2D"
EndSection

Section "Screen"
        Identifier   "Mali-Screen"
        Device       "Mali-Fbdev"
        DefaultDepth  24
EndSection

Section "DRI"
        Mode 0666
EndSection

User avatar
OverSun
Posts: 1414
Joined: Mon Apr 29, 2013 5:12 pm
languages_spoken: english
Contact:

Re: Project: Kernel 3.8

Unread post by OverSun » Mon May 27, 2013 2:58 pm

How do you compile it, can you post the config options?

Plus ~/.xbmc/temp/xbmc.log, if you don't mind.
I want to check is it I miss something, or is it just software decoder working much better on 3.8
Because for me MFC code in xbmc git https://github.com/huceke/xbmc/tree/exynos just crashes my kernel and don't show anything.
Otherwise 2D accel X11-EGL is working for me, because I patched it for myself.

So could you please config options and xbmc.log?

marlemion
Posts: 232
Joined: Sun Mar 31, 2013 7:33 pm
languages_spoken: english
ODROIDs: Odroid U2
Contact:

Re: Project: Kernel 3.8

Unread post by marlemion » Mon May 27, 2013 4:02 pm

My current xbmc.log:

Code: Select all

cat /home/xbmc/.xbmc/temp/xbmc.log
20:20:06 T:3035832320  NOTICE: special://profile/ is mapped to: special://masterprofile/
20:20:06 T:3035832320  NOTICE: -----------------------------------------------------------------------
20:20:06 T:3035832320  NOTICE: Starting XBMC (13.0-ALPHA4 Git:629b5c491c89e9e6757a2c88d00e50c2c0650ba1), Platform: Linux (Gentoo Base System release 2.2, 3.8.13+ armv7l). Built on May 25 2013
20:20:06 T:3035832320  NOTICE: ARM Features: Neon disabled
20:20:06 T:3035832320  NOTICE: special://xbmc/ is mapped to: /usr/share/xbmc
20:20:06 T:3035832320  NOTICE: special://xbmcbin/ is mapped to: /usr/lib/xbmc
20:20:06 T:3035832320  NOTICE: special://masterprofile/ is mapped to: /home/xbmc/.xbmc/userdata
20:20:06 T:3035832320  NOTICE: special://home/ is mapped to: /home/xbmc/.xbmc
20:20:06 T:3035832320  NOTICE: special://temp/ is mapped to: /home/xbmc/.xbmc/temp
20:20:06 T:3035832320  NOTICE: The executable running is: /usr/lib/xbmc/xbmc.bin
20:20:06 T:3035832320  NOTICE: Local hostname: odroid
20:20:06 T:3035832320  NOTICE: Log File is located: /home/xbmc/.xbmc/temp/xbmc.log
20:20:06 T:3035832320  NOTICE: -----------------------------------------------------------------------
20:20:07 T:3035832320  NOTICE: load settings...
20:20:07 T:3035832320  NOTICE: CMediaSourceSettings: loading media sources from special://masterprofile/sources.xml
20:20:07 T:3035832320  NOTICE: Loading player core factory settings from special://xbmc/system/playercorefactory.xml.
20:20:07 T:3035832320  NOTICE: Loaded playercorefactory configuration
20:20:07 T:3035832320  NOTICE: Loading player core factory settings from special://masterprofile/playercorefactory.xml.
20:20:07 T:3035832320  NOTICE: special://masterprofile/playercorefactory.xml does not exist. Skipping.
20:20:07 T:3035832320  NOTICE: LoadFromXML - unable to load:/home/xbmc/.xbmc/userdata/wakeonlan.xml
20:20:07 T:3035832320  NOTICE: No settings file to load (special://xbmc/system/advancedsettings.xml)
20:20:07 T:3035832320  NOTICE: Loaded settings file from special://profile/advancedsettings.xml
20:20:07 T:3035832320  NOTICE: Contents of special://profile/advancedsettings.xml are...
                                            <advancedsettings>
                                              <musicthumbs>
                                                <remove>folder.jpg</remove>
                                                <add>cover.jpg</add>
                                              </musicthumbs>
                                            </advancedsettings>
20:20:07 T:3035832320  NOTICE: Default DVD Player: dvdplayer
20:20:07 T:3035832320  NOTICE: Default Video Player: dvdplayer
20:20:07 T:3035832320  NOTICE: Default Audio Player: paplayer
20:20:07 T:3035832320  NOTICE: Disabled debug logging due to GUI setting. Level 0.
20:20:07 T:3035832320  NOTICE: Log level changed to 0
20:20:07 T:3035832320  NOTICE: Running database version Addons15
20:20:08 T:3031421952  NOTICE: Thread PeripheralBus start, auto delete: false
20:20:08 T:3035832320  NOTICE: Previous line repeats 1 times.
20:20:08 T:3035832320  NOTICE: Setup SDL
20:20:08 T:3035832320  NOTICE: Checking resolution 16
20:20:08 T:3035832320  NOTICE: GL_VENDOR = ARM
20:20:08 T:3035832320  NOTICE: GL_RENDERER = Mali-400 MP
20:20:08 T:3035832320  NOTICE: GL_VERSION = OpenGL ES 2.0
20:20:08 T:3035832320  NOTICE: GL_SHADING_LANGUAGE_VERSION = OpenGL ES GLSL ES 1.00
20:20:08 T:3035832320  NOTICE: GL_EXTENSIONS = GL_OES_texture_npot GL_OES_compressed_ETC1_RGB8_texture GL_OES_standard_derivatives GL_OES_EGL_image GL_OES_depth24 GL_ARM_rgba8 GL_ARM_mali_shader_binary GL_OES_depth_texture GL_OES_packed_depth_stencil GL_EXT_texture_format_BGRA8888 GL_EXT_blend_minmax GL_OES_EGL_image_external GL_OES_EGL_sync GL_EXT_multisampled_render_to_texture GL_EXT_discard_framebuffer GL_OES_get_program_binary GL_ARM_mali_program_binary GL_EXT_shader_texture_lod GL_EXT_robustness
20:20:08 T:3035832320 WARNING: EGL: Same window as before, refreshing context
20:20:08 T:3035832320  NOTICE: Running database version Addons15
20:20:08 T:3035832320  NOTICE: Running database version ViewModes5
20:20:08 T:3035832320  NOTICE: Running database version Textures13
20:20:08 T:3035832320  NOTICE: Running database version MyMusic36
20:20:08 T:3035832320  NOTICE: Running database version MyVideos75
20:20:08 T:3035832320  NOTICE: Running database version TV22
20:20:08 T:3035832320  NOTICE: Running database version Epg7
20:20:08 T:3035832320  NOTICE: start dvd mediatype detection
20:20:08 T:3035832320  NOTICE: initializing playlistplayer
20:20:08 T:3035832320  NOTICE: DONE initializing playlistplayer
20:20:08 T:3032122368  NOTICE: Thread DetectDVDMedia start, auto delete: false
20:20:09 T:2889872384  NOTICE: Thread XBPython start, auto delete: false
20:20:09 T:3035832320  NOTICE: Previous line repeats 2 times.
20:20:09 T:3035832320  NOTICE: UpdateLibraries - Starting music library startup scan
20:20:09 T:2862609408  NOTICE: Thread MusicInfoScanner start, auto delete: false
20:20:09 T:3035832320  NOTICE: initialize done
20:20:09 T:3035832320  NOTICE: Running the application...
20:20:09 T:2889872384  NOTICE: -->Python Interpreter Initialized<--
20:20:09 T:2870998016  NOTICE: -->Python Interpreter Initialized<--
20:20:09 T:2881483776  NOTICE: -->Python Interpreter Initialized<--
20:20:09 T:3035832320   ERROR: GLES: Vertical Blank Syncing unsupported
20:20:10 T:2854220800  NOTICE: Thread JobWorker start, auto delete: true
20:20:10 T:3035832320  NOTICE: Webserver: Starting...
20:20:10 T:3035832320  NOTICE: WebServer: Started the webserver
20:20:10 T:3035832320  NOTICE: ES: Starting event server
20:20:10 T:2803364864  NOTICE: Thread EventServer start, auto delete: false
20:20:10 T:2803364864  NOTICE: ES: Starting UDP Event server on 0.0.0.0:9777
20:20:10 T:2803364864  NOTICE: UDP: Listening on port 9777
20:20:10 T:3035832320   ERROR: JSONRPC Server: Failed to connect to sdpd
20:20:10 T:2794976256  NOTICE: Thread TCPServer start, auto delete: false
20:20:10 T:2786587648  NOTICE: Thread MusicFileCounter start, auto delete: false
20:20:10 T:2778199040  NOTICE: Thread AlarmClock start, auto delete: false
20:20:10 T:2862609408  NOTICE: My Music: Scanning for music info using worker thread, operation took 00:00
20:20:10 T:2870998016  NOTICE: RCB Service: Start
20:20:10 T:2870998016  NOTICE: RCB Service: launch RCB on startup = false
20:20:10 T:2870998016  NOTICE: RCB Service: scrape games on startup = false
20:20:10 T:2870998016  NOTICE: RCB Service: Done
20:20:10 T:2870998016  NOTICE: Thread JobWorker start, auto delete: true
20:20:40 T:2889872384  NOTICE: Thread XBPython start, auto delete: false
20:20:40 T:2889872384  NOTICE: -->Python Interpreter Initialized<--
22:40:12 T:2889872384  NOTICE: Thread JobWorker start, auto delete: true
22:40:12 T:2854220800  NOTICE: Thread FileCache start, auto delete: false
04:40:22 T:2889872384  NOTICE: Thread JobWorker start, auto delete: true
04:40:22 T:2854220800  NOTICE: Thread FileCache start, auto delete: false

Code: Select all

./configure --prefix=/usr --build=armv7a-hardfloat-linux-gnueabi --host=armv7a-hardfloat-linux-gnueabi --mandir=/usr/share/man --infodir=/usr/share/info --datadir=/usr/share --sysconfdir=/etc --localstatedir=/var/lib --disable-silent-rules --disable-dependency-tracking --docdir=/usr/share/doc/xbmc-9999 --disable-ccache --enable-optimizations --enable-gles --disable-airplay --disable-avahi --disable-libbluray --disable-libcap --enable-libcec --enable-dvdcss --disable-debug --enable-fishbmc --enable-gles --enable-goom --disable-hal --enable-joystick --disable-mid --enable-mysql --enable-neon --enable-nfs --disable-gl --disable-profiling --enable-projectm --enable-pulse --enable-mythtv --enable-rsxs --disable-rtmp --enable-samba --enable-sdl --enable-ssh --enable-libusb --disable-upnp --disable-vaapi --disable-vdpau --enable-webserver --enable-x11 --enable-xrandr
I might try your git branch tonight and report back to you. Maybe you are right and the MFC is just working better in software in 3.8.

User avatar
OverSun
Posts: 1414
Joined: Mon Apr 29, 2013 5:12 pm
languages_spoken: english
Contact:

Re: Project: Kernel 3.8

Unread post by OverSun » Mon May 27, 2013 4:20 pm

1. Your XBMC is using 2D acceleration X11-GLES
2. Could you play something and then post your xbmc.log? Because how the file is played is printed out only when the file is played. But I'm 99% sure there will be no MFC support, because in original XBMC git (not huceke) there is no MFC support at all.

marlemion
Posts: 232
Joined: Sun Mar 31, 2013 7:33 pm
languages_spoken: english
ODROIDs: Odroid U2
Contact:

Re: Project: Kernel 3.8

Unread post by marlemion » Mon May 27, 2013 4:25 pm

Maybe xbmc uses it, but it is definitely not enabled in the driver. I get sluggish ui and tearing. I know that G2D is possible via sunxifb and it is enabled via the s5p-g2d driver in the kernel, but it is not enabled via the DDX driver. I tried to get it working (it is exposed via /dev/video5) by linking to /dev/g2d or /dev/disp, but that wouldn't work.

Tonight I will play something and repost the xbmc.log.

User avatar
OverSun
Posts: 1414
Joined: Mon Apr 29, 2013 5:12 pm
languages_spoken: english
Contact:

Re: Project: Kernel 3.8

Unread post by OverSun » Mon May 27, 2013 4:39 pm

Hint: kernel 3.8 in git before yesterday had a glitch which caused to GLES be slow (15 - 20 fps), and it was fixed (probably) only yesterday. I didn't try it by myself still, was too busy, but GLES was not fully working before yesterday thats 100%.

marlemion
Posts: 232
Joined: Sun Mar 31, 2013 7:33 pm
languages_spoken: english
ODROIDs: Odroid U2
Contact:

Re: Project: Kernel 3.8

Unread post by marlemion » Mon May 27, 2013 5:15 pm

Thanks for the hint. My compilation problems seem to be related to overheating. I have no fan yet, but when I compile with two threads only, it seems to be stable. The 3.8 kernel produces slightly more heat and apparently has higher max frequency default values. This will be hopefully sorted out the next days as I have ordered a fan.

marlemion
Posts: 232
Joined: Sun Mar 31, 2013 7:33 pm
languages_spoken: english
ODROIDs: Odroid U2
Contact:

Re: Project: Kernel 3.8

Unread post by marlemion » Mon May 27, 2013 5:28 pm

Well - apparently it is not related to overheating. I managed to overheat the odroid by compiling xbmc with just two threads enabled. So the segmentation faults must be caused by something else. I don't understand these issues, the 3.0.x kernel never had problems like these. I could compile with even five threads for hours without any overheating or segmentation fault problems. Any suggestions? As for now, this kernel is pretty much unusable for me.

User avatar
OverSun
Posts: 1414
Joined: Mon Apr 29, 2013 5:12 pm
languages_spoken: english
Contact:

Re: Project: Kernel 3.8

Unread post by OverSun » Mon May 27, 2013 5:51 pm

Did you install the headers of new kernel into the system?

Code: Select all

make headers_check
make INSTALL_HDR_PATH=dest headers_install
find dest/include \( -name .install -o -name ..install.cmd \) -delete
cp -rv dest/include/* /usr/include

marlemion
Posts: 232
Joined: Sun Mar 31, 2013 7:33 pm
languages_spoken: english
ODROIDs: Odroid U2
Contact:

Re: Project: Kernel 3.8

Unread post by marlemion » Mon May 27, 2013 6:22 pm

I can't check it right now as it locked up during compilation. But generally, Gentoo has quite recent headers by default especially when having a global keyword enabled. But I will check and let you guys know.

sboyce
Posts: 158
Joined: Thu Feb 28, 2013 7:17 am
languages_spoken: english
ODROIDs: ODROID-C2
Location: Halesowen, West Midlands, UK
Contact:

Re: Project: Kernel 3.8

Unread post by sboyce » Mon May 27, 2013 9:03 pm

KC7NOA wrote:
neagix wrote:(I know I will repent of posting this)

For those interested, the 3.8.13 uInitrd and zImage (for U2) are available here:

http://oph.mdrjr.net/neagix/uInitrd-3.8.13
http://oph.mdrjr.net/neagix/zImage-3.8.13

These have been compiled about 1 week ago with the odroidu2_defconfig kernel configuration file.

Just for testing! Remember: these are not officially provided by Hardkernel!

X or X2 .. i have both ..
Won't boot on X.

Post Reply

Return to “The Ideas”

Who is online

Users browsing this forum: No registered users and 1 guest