Ubuntu 16.04.2 trial image to play with 4.9.x LTS Kernel

Post by **odroid** » Wed Feb 22, 2017 2:04 pm

Update !!!
Hardkernel announced Linux kernel 4.9.11 support for the ODROID-XU4 and XU3 platforms in late February and we started a debugging party publicly.

After ten weeks of debugging party with our community members, we are proudly releasing a new Ubuntu image based on latest Linux kernel 4.9.27 LTS.
We've improved and added new features such UASP, UHS mode, IRQ balancing, HW mouse cursor, HW RTC Alarm, overlayfs, faster armsoc DDX, multi-touch screen and much more.
You can find over 300 commits in our GitHub history.
https://github.com/hardkernel/linux/com ... dxu4-4.9.y

We've packed all the updates into a new OS image for all XU4/3 users.
This official Ubuntu OS image is available on this link.
http://odroid.com/dokuwiki/doku.php?id= ... e_20170510
We will keep improving the Linux OS.

Note that there is no reliable way to update the Kernel to 4.9 from 3.10.
So please freshly install the new image after backup your data.

==================================================================================================
We will release a trial Ubuntu image based on 4.9.x LTS Kernel within 24 hours. (update : released)

Basic features like HDMI video/audio, Gbit Ethernet, GPU and MFC are working now.
But there should be some known/unknown issues.
So we want to fix the issues with forum members intensively for a couple of months.

Source code: https://github.com/hardkernel/linux/tre ... dxu4-4.9.y

Image:
US: http://deb.odroid.in/trial_4.9/
KR: http://dn.odroid.com/5422/ODROID-XU3/Ubuntu/trial_4.9/

Any issues or questions, please post on this thread or on this sub-forum. We'll be watching those places for this specific release.

There's a few known issues:
- USB 3.0 sometimes doesn't work.
- Kodi Playback is really laggy
- Many others.

This image has the latest 16.04.2 Ubuntu with newer Chromium 55 and Kodi 17.
Also GStreamer support for the MFC is present for video encoding/decoding.

Many kernel modules may not be enabled, you can request those to be enabled on this thread.

Kernel updates will be very frequent, so please, before posting any question make sure you have your kernel updated.

mrneilypops · Post by **mrneilypops** » Wed Feb 22, 2017 2:31 pm

Looking forward to testing...

mattrix · Post by **mattrix** » Wed Feb 22, 2017 2:52 pm

AWESOME!!!

It's here:
https://github.com/hardkernel/linux/tre ... dxu4-4.9.y

Building now

RodBelaFarin · Post by **RodBelaFarin** » Wed Feb 22, 2017 5:07 pm

Awesome - waiting for it! We will use this image in 3 weeks for an international robotics challenge. We will let you know, if it worked

Snk · Post by **Snk** » Wed Feb 22, 2017 9:47 pm

Yeah!
I'm waiting for the image!

Post by **odroid** » Wed Feb 22, 2017 10:22 pm

The image has been released. Look at the opening post.

igorpec · Post by **igorpec** » Wed Feb 22, 2017 10:38 pm

We also rebuilt images with this kernel. So far working good!

https://www.armbian.com/odroid-xu4/

rooted · Post by **rooted** » Thu Feb 23, 2017 1:25 am

I will test on the XU3 when I get home.

DarkBahamut · Post by **DarkBahamut** » Thu Feb 23, 2017 9:32 am

Just a word on XU4 NAS performance with a lightly tweaked 4.9.11 kernel.

It's bonkers

990 Mbps / 115MB/s. I don't think there's much left in that gigabit network link

Ok, it is an eMMC read, but it's still ever so slightly faster than I can get 3.10 to go, and I'm only stock clocks (1.4/2.0). The XU4 really excels for this use.

rooted · Post by **rooted** » Thu Feb 23, 2017 9:57 am

Yes it's been a long wait but there are so many benefits to 4.9.y

All the additional hardware supported and fixes since 3.10.y.

WebClaw · Post by **WebClaw** » Thu Feb 23, 2017 11:03 am

Not really a problem (per say) but may help those decide if they want to test given current state.
I use UX4 as a NAS / CP / NZB / SERVIIO etc. - so the purpose may be different.

Fresh install of ubuntu-16.04.2-mate-odroid-xu4-20170222.img.xz
Upon first boot login as root via SSH then did

Code: Select all

fstrim -v /

Took like 5 minutes to complete (to the point I thought it went non-responsive) - 3.10.y kernels took a few seconds.
Size of MMC is 64GB.

Using CloudShell - also noted:

Code: Select all

echo "options fbtft_device name=hktft9340 busnum=1 rotate=270" > /etc/modprobe.d/odroid-cloudshell.conf
echo "spi_s3c64xx" >> /etc/modules
echo "fbtft_device" >> /etc/modules

No longer results in a working LCD - although odroid did say modules were missing from the kernel.

Code: Select all

[    6.065022] fbtft: module is from the staging directory, the quality is unknown, you have been warned.
[    6.075689] fbtft_device: module is from the staging directory, the quality is unknown, you have been warned.
[    6.086164] fbtft_device: spi_busnum_to_master(1) returned NULL
[    6.091092] fbtft_device: failed to register SPI device

I certainly noticed it's a lot slower compiling code (I did FFMpeg, x264, x265, etc) via SSH. Kernel 3.10.y was much faster. Perhaps the governor is set different (didn't check).

During FFMpeg build using make -j8 I did notice the fan was not spinning. That is odd - you'd think with all cores assigned to build code I'd be generating some serious heat.

Code: Select all

top - 02:40:20 up 35 min,  2 users,  load average: 1.00, 2.45, 2.59
Tasks: 178 total,   2 running, 176 sleeping,   0 stopped,   0 zombie
%Cpu(s): 12.4 us,  0.2 sy,  0.0 ni, 87.3 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
KiB Mem :  1999336 total,    46296 free,   173836 used,  1779204 buff/cache
KiB Swap:        0 total,        0 free,        0 used.  1732212 avail Mem

  PID USER      PR  NI    VIRT    RES    SHR S  %CPU %MEM     TIME+ COMMAND
17514 root      20   0   99092  96236   1076 R 100.0  4.8   5:14.16 dpkg-deb

ODROID UX4 can now transcode and stream 1080p movies without any issues.
On 3.10y kernel it would only play a few seconds then stop, then start, then stop (transcoding couldn't keep up).
While I did notice a slight decrease in playback quality - I am very excited MFC is now functional.

Side note:
As expected for anyone who uses Serviio- when you try to playback 2 1080p streams you get the stutter so I would say a single stream can be reliably transcoded at once. Good for home DLNA, but not multiple users on MediaBrowser.

Excellent work Odroid / memeka and everyone else that made this possible.
@Odroid I continue to be amazed by your development efforts on a $70 dollar product. I see great things happening for HardKernel.

If anyone wants to test x264 / x265 / FFMpeg I've uploaded packages I've compiled to assist others here. Please note, FFMpeg configured with options for use with Serviio (with additions).

http://layer3d.org/temp/faac_1.28-fixes-1-1_armhf.deb
http://layer3d.org/temp/ffmpeg_3.0.2-mi ... _armhf.deb <-- from mihailescu2m git (h264_v4l2m2m)
http://layer3d.org/temp/libmp3lame_3.99 ... _armhf.deb
http://layer3d.org/temp/rtmpdump_2.4-se ... _armhf.deb
http://layer3d.org/temp/x265_02-22-2017-1_armhf.deb

Just download / wget to /tmp do dpkg -i *.deb and run ldconfig after install (or reboot).

Code: Select all

ffmpeg version N-81312-gb83d907 Copyright (c) 2000-2016 the FFmpeg developers
  built with gcc 5.4.0 (Ubuntu/Linaro 5.4.0-6ubuntu1~16.04.4) 20160609
  configuration: --enable-libopencore-amrnb --enable-libopencore-amrwb --enable-libfaac --enable-libtheora --enable-libvorbis --enable-libx265 --enable-libxvid --enable-nonfree --enable-version3 --enable-static --disable-shared --disable-ffplay --disable-ffserver --enable-memalign-hack --enable-libmp3lame --enable-libass --enable-librtmp --enable-fontconfig --enable-libfreetype --enable-zlib --enable-libx264 --enable-libspeex --enable-runtime-cpudetect --enable-pthreads --enable-gpl --arch=arm --pkg-config-flags=--static
  libavutil      55. 28.100 / 55. 28.100
  libavcodec     57. 51.100 / 57. 51.100
  libavformat    57. 46.100 / 57. 46.100
  libavdevice    57.  0.102 / 57.  0.102
  libavfilter     6. 50.100 /  6. 50.100
  libswscale      4.  1.100 /  4.  1.100
  libswresample   2.  1.100 /  2.  1.100
  libpostproc    54.  0.100 / 54.  0.100
Hyper fast Audio and Video encoder
usage: ffmpeg [options] [[infile options] -i infile]... {[outfile options] outfile}...

On 3.10y kernel I did have problems with power save / auto-suspend causing disconnects on HDD connected to Cloudshell I had to manually do:

Code: Select all

sudo sh -c "echo -1 > /sys/bus/usb/devices/4-1.1/power/autosuspend_delay_ms"

I will leave my unit powered on without entering this command to see if the USB problem is still present.

crashoverride · Post by **crashoverride** » Thu Feb 23, 2017 11:06 am

What is the status of HMP with this kernel? Does it properly switch tasks between big and little cores?

Post by **odroid** » Thu Feb 23, 2017 11:16 am

@WebClaw
We will test the CloudShell LCD functionality today.

@crashoverride
The HMP works. But we are not sure if the task switches working well or not.
It is the main reason why we started this debugging party.
Please kindly join us. We need to override the crash.

DarkBahamut · Post by **DarkBahamut** » Thu Feb 23, 2017 11:20 am

On fstrim, I noticed it was a bit slower, though it was only 30 seconds for me. I'm guessing 4.9 is making the discard operation happen in real time, rather than it being async like it appears to be 3.10. It's not a big deal if you just run it as a cronjob, in fact it could even be preferred.

Rather more critical that people do not mount with discard though!

As for HMP, I thought it wasn't working myself. Performance is very variable, tasks often on the wrong cluster. Manually setting them results in better performance. If this should be working then I might have a poke at it tomorrow, I just assumed it hadn't been put in yet

crashoverride · Post by **crashoverride** » Thu Feb 23, 2017 11:30 am

odroid wrote:Please kindly join us.

I have been anxiously awaiting this, so count me in!

memeka · Post by **memeka** » Thu Feb 23, 2017 1:49 pm

crashoverride wrote:What is the status of HMP with this kernel? Does it properly switch tasks between big and little cores?

HMP patches: https://github.com/mihailescu2m/linux/c ... .9.y.patch

HMP works, but may crash under heavy load. testing need

memeka · Post by **memeka** » Thu Feb 23, 2017 1:52 pm

WebClaw wrote: Excellent work Odroid / memeka and everyone else that made this possible.
@Odroid I continue to be amazed by your development efforts on a $70 dollar product. I see great things happening for HardKernel.

If anyone wants to test x264 / x265 / FFMpeg I've uploaded packages I've compiled to assist others here. Please note, FFMpeg configured with options for use with Serviio (with additions).

http://layer3d.org/temp/faac_1.28-fixes-1-1_armhf.deb
http://layer3d.org/temp/ffmpeg_3.0.2-mi ... _armhf.deb <-- from mihailescu2m git (h264_v4l2m2m)
http://layer3d.org/temp/libmp3lame_3.99 ... _armhf.deb
http://layer3d.org/temp/rtmpdump_2.4-se ... _armhf.deb
http://layer3d.org/temp/x265_02-22-2017-1_armhf.deb

Just download / wget to /tmp do dpkg -i *.deb and run ldconfig after install (or reboot).

be aware MFC is not working full speed atm. a patch is coming soon to fix it. you might want to try transcoding more streams then.
also quality can be set for ffmpeg - after this patch: https://github.com/mihailescu2m/FFmpeg/ ... c9f4ff969a you can set bitrate to what you need - higher bitrate, better quality.

WebClaw · Post by **WebClaw** » Thu Feb 23, 2017 2:12 pm

Excellent - as always thank you.
I'll add the patch and see what happens.

emu_can · Post by **emu_can** » Thu Feb 23, 2017 2:42 pm

Aaaaargh.. XU4 is $59 now without PSU. I don't know when hardkernel guys cut the price around $10. http://www.hardkernel.com/main/products ... 3452239825
Is it a short term event or promotion to celebrate the kernel 4.9 announce? or what else?

Post by **mdrjr** » Thu Feb 23, 2017 3:58 pm

HMP patches are merged and working..
Enjoy

mad_ady · Post by **mad_ady** » Thu Feb 23, 2017 8:12 pm

emu_can wrote:Aaaaargh.. XU4 is $59 now without PSU. I don't know when hardkernel guys cut the price around $10. http://www.hardkernel.com/main/products ... 3452239825
Is it a short term event or promotion to celebrate the kernel 4.9 announce? or what else?

They sell it without power supply now. It makes sense if you want the 6A PSU. That's where you get the price difference from

elatllat · Post by **elatllat** » Fri Feb 24, 2017 5:43 am

mdrjr wrote:HMP patches are merged and working..
Enjoy

Thanks I'll defiantly have to test HMP.
4.9.12 is out.

umiddelb · Post by **umiddelb** » Fri Feb 24, 2017 6:15 am

elatllat wrote: 4.9.12 is out.

You can apply the incremental patch without any hassle.

Code: Select all

cd .../linux
curl -sSL https://cdn.kernel.org/pub/linux/kernel/v4.x/incr/patch-4.9.11-12.xz | unxz | patch -p1

Just building 4.9.12 now.

Snk · Post by **Snk** » Fri Feb 24, 2017 6:33 am

Video locking in Kodi 17, some incompatibility with the Mali driver?

elatllat · Post by **elatllat** » Fri Feb 24, 2017 6:44 am

umiddelb wrote:
elatllat wrote: ...4.9.12 is out.
You can apply the incremental patch without any hassle.
Code: Select all
cd .../linux
curl -sSL https://cdn.kernel.org/pub/linux/kernel/v4.x/incr/patch-4.9.11-12.xz | unxz | patch -p1
Just building 4.9.12 now.

I have been using

Code: Select all

git merge remotes/kernel/linux-4.9.y

and solving the rare conflict with

Code: Select all

git mergetool -t vimdiff

because others seem to shy away form auto merge/test cron scripts.

umiddelb · Post by **umiddelb** » Fri Feb 24, 2017 7:02 am

elatllat wrote: I have been using
Code: Select all
git merge remotes/kernel/linux-4.9.y
and solving the rare conflict with
Code: Select all
git mergetool -t vimdiff
because others seem to shy away form auto merge/test cron scripts.

OK, btw. none of the patches have been rejected.

derko · Post by **derko** » Fri Feb 24, 2017 7:10 am

Hi. I want to request the following kernel module:

- fuse

apart from that: very good work, thanks !

OverSun · Post by **OverSun** » Fri Feb 24, 2017 7:20 am

MFC is working full speed, there is another issue with it that makes video laggy in Kodi. Once more, the problem is not in speed, the issue makes it very specific for playing videos in Kodi.
We are working on that, since what is wrong is clear. It could take some time, but there is at least another way to fix it in Kodi by a sad workaround. If it is too long to find what went wrong in the kernel, this workaround is going to be applied to Kodi.

umiddelb · Post by **umiddelb** » Fri Feb 24, 2017 7:35 am

derko wrote:Hi. I want to request the following kernel module:

- fuse

apart from that: very good work, thanks !

CONFIG_FUSE is mentioned here.

elatllat · Post by **elatllat** » Fri Feb 24, 2017 8:10 am

odroid wrote:...kernel modules ...you can request ... on this thread...

Please enable

CONFIG_ZRAM
CONFIG_USB_MASS_STORAGE
CONFIG_MD_RAID0
CONFIG_BTRFS_FS
CONFIG_FUSE_FS
CONFIG_F2FS_FS
CONFIG_CRYPTO_XTS
CONFIG_LZ4_COMPRESS

DarkBahamut · Post by **DarkBahamut** » Fri Feb 24, 2017 8:37 am

I would like to see F2FS enabled also. It's quite a nice FS to use with an SD card for mass storage along side the eMMC.

I tested the hperf hmp commits today. Full load for a few hours and no crashes. Seems to work as well as expected.

I also notice temps are quite a bit higher than on 3.10, even when running the same CPU voltages (about 8C higher at idle). I'm guessing there's no clock gating occurring on 4.9 yet?

WebClaw · Post by **WebClaw** » Fri Feb 24, 2017 8:43 am

I am having a problem with USB 3.0 - non-detection of a device.

Device: NexStar HX4 (a simple 4 bay SATA to USB 3.0 enclosure).
Chipset: JMS539

I did have the device connected to USB 3 on Ubuntu 14 w/3.10.y kernel (Linux odroid-server 3.10.104) and it did work.

When connected to USB 2.0 Port:

Code: Select all

[   63.351946] usb 1-1: new high-speed USB device number 2 using exynos-ehci
[   63.503582] usb 1-1: New USB device found, idVendor=152d, idProduct=0551
[   63.503621] usb 1-1: New USB device strings: Mfr=1, Product=2, SerialNumber=5
[   63.503645] usb 1-1: Product: USB to ATA/ATAPI Bridge
[   63.503665] usb 1-1: Manufacturer: JMicron
[   63.503685] usb 1-1: SerialNumber: 41EAFFFFFFFF
[   63.507017] usb-storage 1-1:1.0: USB Mass Storage device detected
[   63.509929] scsi host0: usb-storage 1-1:1.0
[   64.710321] scsi 0:0:0:0: Direct-Access     ST2000DM 001-1ER164            PQ: 0 ANSI: 2 CCS
[   64.712943] sd 0:0:0:0: Attached scsi generic sg0 type 0
[   64.714685] sd 0:0:0:0: [sda] 3907029168 512-byte logical blocks: (2.00 TB/1.82 TiB)
[   64.716238] sd 0:0:0:0: [sda] Write Protect is off
[   64.716263] sd 0:0:0:0: [sda] Mode Sense: 28 00 00 00
[   64.717069] scsi 0:0:0:1: Direct-Access     ST2000DM 001-1ER164            PQ: 0 ANSI: 2 CCS
[   64.719518] sd 0:0:0:1: Attached scsi generic sg1 type 0
[   64.720261] sd 0:0:0:0: [sda] No Caching mode page found
[   64.724444] sd 0:0:0:0: [sda] Assuming drive cache: write through
[   64.731093] sd 0:0:0:1: [sdb] 3907029168 512-byte logical blocks: (2.00 TB/1.82 TiB)
[   64.732379] sd 0:0:0:1: [sdb] Write Protect is off
[   64.732406] sd 0:0:0:1: [sdb] Mode Sense: 28 00 00 00
[   64.733589] sd 0:0:0:1: [sdb] No Caching mode page found
[   64.738100] sd 0:0:0:1: [sdb] Assuming drive cache: write through
[   64.745513] scsi 0:0:0:2: Direct-Access     ST2000DM 001-1ER164            PQ: 0 ANSI: 2 CCS
[   64.748382] sd 0:0:0:2: Attached scsi generic sg2 type 0
[   64.754433] scsi 0:0:0:3: Direct-Access     ST2000DM 001-1ER164            PQ: 0 ANSI: 2 CCS
[   64.755171] sd 0:0:0:2: [sdc] 3907029168 512-byte logical blocks: (2.00 TB/1.82 TiB)
[   64.758408] sd 0:0:0:3: Attached scsi generic sg3 type 0
[   64.773913] sd 0:0:0:2: [sdc] Write Protect is off
[   64.773972] sd 0:0:0:2: [sdc] Mode Sense: 28 00 00 00
[   64.775449] sd 0:0:0:3: [sdd] 3907029168 512-byte logical blocks: (2.00 TB/1.82 TiB)
[   64.776979] sd 0:0:0:2: [sdc] No Caching mode page found
[   64.779614] sd 0:0:0:3: [sdd] Write Protect is off
[   64.779652] sd 0:0:0:3: [sdd] Mode Sense: 28 00 00 00
[   64.780845] sd 0:0:0:2: [sdc] Assuming drive cache: write through
[   64.800387] sd 0:0:0:3: [sdd] No Caching mode page found
[   64.804454] sd 0:0:0:3: [sdd] Assuming drive cache: write through
[   64.825210] sd 0:0:0:1: [sdb] Attached SCSI disk
[   64.826312] sd 0:0:0:0: [sda] Attached SCSI disk
[   64.847054] sd 0:0:0:2: [sdc] Attached SCSI disk
[   64.874418] sd 0:0:0:3: [sdd] Attached SCSI disk

There is no dmesg output when this device is connected to USB 3.0 - everything works as expected on USB 2.0 port

For diagnostics purposes I've changed the FW on the JM chip to:
JMS-539 Firmware Version NC (latest)
255.31.3.41.22 (old version)

- Same result.

memeka · Post by **memeka** » Fri Feb 24, 2017 8:45 am

OverSun wrote:MFC is working full speed, there is another issue with it that makes video laggy in Kodi. Once more, the problem is not in speed, the issue makes it very specific for playing videos in Kodi.
We are working on that, since what is wrong is clear. It could take some time, but there are at least another way to fix it in Kodi by a sad workaround. If it will be too long to find what went wrong in the kernel, this workaround is going to be applied to Kodi.

@OverSun
i can play videos fine with gstreamer (MFC) - well, i have other issues there, but related to wayland, not MFC - so it might be a kodi thing.
also, a while ago (when kernel 4.7 was the newest) i compiled and run kodi successfully: https://www.youtube.com/watch?v=o0aFehweAyk - this is from your repo, kodi 16.1 branch. i haven't tried more recently because they removed wayland support from kodi.

HOWEVER - you can see in the kodi video - also the same thing happens in gstreamer - when decoding, one core is always used at 100% - maybe there's some kind of polling now on the MFC for frames? I did not like this behaviour, as ffmpeg decoding was able to keep up with 1080p videos without maxing out any core.

EDIT: the weird MFC behaviour is obvious when i run gstreamer output to fakesink (so just decode, and don't do anything to the decoded video) - i get 1 big core 100% (fan starts too), and the rest of the cores are 2-4%. i think mfc should keep total cores usage to <10%.

crashoverride · Post by **crashoverride** » Fri Feb 24, 2017 8:57 am

memeka wrote:when decoding, one core is always used at 100%

I see the same behavior in the XU4VideoCube sample. The cause in my case is the X11 DDX. The surface (GLES) is blitted by the CPU. The evidence for this conclusion is running "top" or "htop" which shows the process is "xorg" consuming a large core.

memeka · Post by **memeka** » Fri Feb 24, 2017 10:00 am

crashoverride wrote:
memeka wrote:when decoding, one core is always used at 100%
I see the same behavior in the XU4VideoCube sample. The cause in my case is the X11 DDX. The surface (GLES) is blitted by the CPU. The evidence for this conclusion is running "top" or "htop" which shows the process is "xorg" consuming a large core.

was it the same in 3.10?

crashoverride · Post by **crashoverride** » Fri Feb 24, 2017 10:20 am

memeka wrote:was it the same in 3.10?

Its been a long time since I tested it on 3.10. The CPU usage was always high for the blit. I did an experimental armsoc with G2D blitting; however, due to the kernel driver it was never stable. I hope to find time to revisit this.

The odd thing is that when a X11 window is fullscreen (like Kodi), the armsoc DDX should do buffer flipping instead of blitting. Its still early in my testing to find what does and does not work.

Post by **mdrjr** » Fri Feb 24, 2017 10:53 am

New updates today:
Docker and LXC stuff enabled.
Improved Mali support on 4.9
Improved performance by tweaking tmu values.

elatllat · Post by **elatllat** » Fri Feb 24, 2017 12:24 pm

Thanks.

Please disable ads7846 until it's fixed (currently it uses 4.6% all the time when it should be sleeping)

might be helpful to set CONFIG_LOCALVERSION_AUTO=y

Also can we have the trivial compile warnings cleaned up?

Code: Select all

kernel/sched/fair.c: In function ‘select_task_rq_fair’:
kernel/sched/fair.c:5988:6: warning: suggest explicit braces to avoid ambiguous ‘else’ [-Wparentheses]
   if (sd_flag & SD_BALANCE_WAKE) /* XXX always ? */
      ^
drivers/bluetooth/rtk_coex.c: In function ‘rtk_notify_info_to_wifi’:
drivers/bluetooth/rtk_coex.c:1254:21: warning: unused variable ‘report’ [-Wunused-variable]
  struct rtl_btinfo *report = (struct rtl_btinfo *)report_info;
                     ^
drivers/bluetooth/rtk_coex.c: In function ‘rtk_notify_regester_to_wifi’:
drivers/bluetooth/rtk_coex.c:1291:24: warning: unused variable ‘reg’ [-Wunused-variable]
  hci_mailbox_register *reg = (hci_mailbox_register *) reg_value;
                        ^
drivers/gpu/drm/exynos/exynos_hdmi.c: In function ‘hdmi_hpd_enable’:
drivers/gpu/drm/exynos/exynos_hdmi.c:174:1: warning: no return statement in function returning non-void [-Wreturn-type]
 }
 ^
drivers/usb/dwc3/core.c: In function ‘dwc3_core_init’:
drivers/usb/dwc3/core.c:736:1: warning: label ‘err1’ defined but not used [-Wunused-label]
 err1:
 ^
drivers/usb/dwc3/core.c: At top level:
drivers/usb/dwc3/core.c:341:12: warning: ‘dwc3_alloc_scratch_buffers’ defined but not used [-Wunused-function]
 static int dwc3_alloc_scratch_buffers(struct dwc3 *dwc)
            ^

linuxest · Post by **linuxest** » Fri Feb 24, 2017 12:37 pm

My realtek wifi 3 adapter is working now with 4.9.11-11 update today.
Thank you for the quick fix! I really love this upstream kernel.

Post by **mdrjr** » Fri Feb 24, 2017 2:23 pm

elatllat wrote:Thanks.

Please disable ads7846 until it's fixed (currently it uses 4.6% all the time when it should be sleeping)

might be helpful to set CONFIG_LOCALVERSION_AUTO=y

Also can we have the trivial compile warnings cleaned up?

Code: Select all

kernel/sched/fair.c: In function ‘select_task_rq_fair’:
kernel/sched/fair.c:5988:6: warning: suggest explicit braces to avoid ambiguous ‘else’ [-Wparentheses]
   if (sd_flag & SD_BALANCE_WAKE) /* XXX always ? */
      ^
drivers/bluetooth/rtk_coex.c: In function ‘rtk_notify_info_to_wifi’:
drivers/bluetooth/rtk_coex.c:1254:21: warning: unused variable ‘report’ [-Wunused-variable]
  struct rtl_btinfo *report = (struct rtl_btinfo *)report_info;
                     ^
drivers/bluetooth/rtk_coex.c: In function ‘rtk_notify_regester_to_wifi’:
drivers/bluetooth/rtk_coex.c:1291:24: warning: unused variable ‘reg’ [-Wunused-variable]
  hci_mailbox_register *reg = (hci_mailbox_register *) reg_value;
                        ^
drivers/gpu/drm/exynos/exynos_hdmi.c: In function ‘hdmi_hpd_enable’:
drivers/gpu/drm/exynos/exynos_hdmi.c:174:1: warning: no return statement in function returning non-void [-Wreturn-type]
 }
 ^
drivers/usb/dwc3/core.c: In function ‘dwc3_core_init’:
drivers/usb/dwc3/core.c:736:1: warning: label ‘err1’ defined but not used [-Wunused-label]
 err1:
 ^
drivers/usb/dwc3/core.c: At top level:
drivers/usb/dwc3/core.c:341:12: warning: ‘dwc3_alloc_scratch_buffers’ defined but not used [-Wunused-function]
 static int dwc3_alloc_scratch_buffers(struct dwc3 *dwc)
            ^

Sure, just send a PR or a patch and I'll apply it

Post by **mdrjr** » Fri Feb 24, 2017 2:24 pm

More stuff:
Merged v4.9.12
Rtl8812/rtl8821 (Wifi Module 5) now working

strause · Post by **strause** » Fri Feb 24, 2017 10:20 pm

Please enable to use Logitech rumblepad 2 cordless. I have this problem https://retropie.org.uk/forum/topic/320 ... blepad-2/2
There is solution https://git.kernel.org/cgit/linux/kerne ... id=2078665

Thank you

elatllat · Post by **elatllat** » Sat Feb 25, 2017 12:08 am

strause wrote:Please enable to use Logitech rumblepad 2 cordless. I have this problem https://retropie.org.uk/forum/topic/320 ... blepad-2/2
There is solution https://git.kernel.org/cgit/linux/kerne ... id=2078665

Thank you

Please test before you request: That patch was applied 2 years ago.

Code: Select all

> grep -B 1 iface_num drivers/hid/hid-lg.c | tail -n 3
	if ((hdev->product == USB_DEVICE_ID_LOGITECH_G29_WHEEL) &&
	    (iface_num != 0)) {
		dbg_hid("%s: ignoring ifnum %d\n", __func__, iface_num);

strause · Post by **strause** » Sat Feb 25, 2017 5:24 pm

elatllat wrote:
strause wrote:Please enable to use Logitech rumblepad 2 cordless. I have this problem https://retropie.org.uk/forum/topic/320 ... blepad-2/2
There is solution https://git.kernel.org/cgit/linux/kerne ... id=2078665

Thank you
Please test before you request: That patch was applied 2 years ago.
Code: Select all
> grep -B 1 iface_num drivers/hid/hid-lg.c | tail -n 3
	if ((hdev->product == USB_DEVICE_ID_LOGITECH_G29_WHEEL) &&
	    (iface_num != 0)) {
		dbg_hid("%s: ignoring ifnum %d\n", __func__, iface_num);

But I'm still getting same error, when I connect my controller "logitech: probe of 0003:046D:C219.0004 failed with error -1"

scroom · Post by **scroom** » Sat Feb 25, 2017 7:44 pm

Great to read that Kernel 4.9 is coming to the ODROID XU4. Will it be possible to install snap packages with the new kernel version? And when it is ready will it be possbile to update the kernel just via apt-get?

derko · Post by **derko** » Sat Feb 25, 2017 11:29 pm

@ Odroid: Kernel 4.9.12-13 is working great. Thanks for the requested modules, the transparency in GitHub and your very fast updates. I will test Kodi now

@scroom: Yes you can install the kernel just by:

Code: Select all

apt-get install linux-image-xu3

Vasco · Post by **Vasco** » Sat Feb 25, 2017 11:46 pm

Hi there,

for testing I booted the 16.0.4.2 image from an SD card.
On first run /etc/rc.local was executed which called a script named /aafirstboot.

Then this script deleted the partitions on /dev/mmcblk0 - eMMC flash - without warning and without confirmation (!) and repartitions it probably for resizing (at which it failed).

On my eMMC flash is (was?) a running Gentoo distro with kernel 4.10.12 installed...

This is NOT what I expect!

[ 19.262029] rc.local[1189]: Welcome to fdisk (util-linux 2.27.1).
[ 19.280678] rc.local[1189]: Changes will remain in memory only, until you decide to write them.
[ 19.300615] rc.local[1189]: Be careful before using the write command.
[ 19.320651] rc.local[1189]: Command (m for help): Disk /dev/mmcblk0: 58.2 GiB, 62537072640 bytes, 122142720 sectors
[ 19.345564] rc.local[1189]: Units: sectors of 1 * 512 = 512 bytes
[ OK [ 19.365738] ] rc.localStarted LSB: Start NTP daemon.[1189]:
Sector size (logical/physical): 512 bytes / 512 bytes
[ 19.390829] rc.local[1189]: I/O size (minimum/optimal): 512 bytes / 512 bytes
[ 19.410715] rc.local[1189]: Disklabel type: dos
[ 19.426646] rc.local[1189]: Disk identifier: 0xa9e3e2c7
[ 19.451527] rc.local[1189]: Device Boot Start End Sectors Size Id Type
[ 19.471442] rc.local[1189]: /dev/mmcblk0p1 2048 122142719 122140672 58.2G 83 Linux
[ 19.490930] rc.local[1189]: Command (m for help): Selected partition 1
[ 19.510844] rc.local[1189]: Partition 1 has been deleted.
[ 19.530900] rc.local[1189]: Command (m for help): 2: unknown command
[ 19.550895] rc.local[1189]: Command (m for help): Partition type
[ 19.570957] rc.local[1189]: p primary (0 primary, 0 extended, 4 free)
[ 19.590856] rc.local[1189]: e extended (container for logical partitions)
[ 19.610851] rc.local[1189]: Select (default p): Partition number (1-4, default 1): First sector (2048-122142719, default 2048): Last sector, +sectors or +size{K,M,G,T,P} (2048-122142719, default 122142719):
[ 19.640815] rc.local[1189]: Created a new partition 2 of type 'Linux' and of size 58.2 GiB.
[ 19.660750] rc.local[1189]: Command (m for help): Disk /dev/mmcblk0: 58.2 GiB, 62537072640 bytes, 122142720 sectors
[ 19.685788] rc.local[1189]: Units: sectors of 1 * 512 = 512 bytes
[ 19.705752] rc.local[1189]: Sector size (logical/physical): 512 bytes / 512 bytes
[ 19.725724] rc.local[1189]: I/O size (minimum/optimal): 512 bytes / 512 bytes
[ 19.745703] rc.local[1189]: Disklabel type: dos
[ 19.765854] rc.local[1189]: Disk identifier: 0xa9e3e2c7
[ 19.786563] rc.local[1189]: Device Boot Start End Sectors Size Id Type
[ 19.806186] rc.local[1189]: /dev/mmcblk0p2 2048 122141696 122139649 58.2G 83 Linux
[ 21.128456] rc.local[1189]: Command (m for help): The partition table has been altered.
[ 21.151856] rc.local[1189]: Calling ioctl() to re-read partition table.
[ 21.172184] rc.local[1189]: Syncing disks.

Who came up with this brilliant idea to erase /dev/mmcblk0?
The SD card is /dev/mmcblk1 (!)

Installation and first-time configuration scripts must not erase data - especially partitions - without confirmation.

Cheers,
Vasco

johnvick · Post by **johnvick** » Sun Feb 26, 2017 6:24 am

I installed this kernel on XU4 running DietPi v144 - it installed and booted but no LAN was available.

Any suggestions to fix this?

Thanks, John

crashoverride · Post by **crashoverride** » Sun Feb 26, 2017 7:03 am

johnvick wrote:Any suggestions to fix this?

Make sure you also installed the modules in addition to the kernel. The ethernet on XU4 is a USB device.

johnvick · Post by **johnvick** » Sun Feb 26, 2017 7:18 am

Ah thanks - how do I do this?