Hello to all. For a long time I was not in this thread :) I just want to leave a link to the sensible wifi driver assembly instructions.

here it is

But how to make firelight work with touchscreen normally:


1. open about: config in firefox to set dom.w3c_touch_events.enabled = 1 (default is 2).
2. edit /etc/security/pam_env.conf and add MOZ_USE_XINPUT2 DEFAULT = 1
3. reboot and restart firefox

Post has been editedevgen-y2k - 25.02.19, 06:22

Hello to all. Read the thread from the beginning. The impression was that the piece of iron was not really overpowered. Please correct me if I am wrong in the conclusions:
1. Built-in wifi does not work in AP mode
2. Voice calls under linux do not work due to problems with sound settings.
3. The stability of any OS on this hardware (I mean linux builds) leaves much to be desired.
I am considering the option of acquiring a piece of hardware for a specific task, which includes mandatory conditions, such as voice calls and the work of the built-in wifi in the access point mode.

I checked the kernel 4.9.5. In general, very pleased. The dmesg output is pretty clean. I cursed at the attempt to download bluetooth firmware and something about acpi. Interestingly, bluetooth works fine. Everything works completely correctly except for wifi only. At the first boot after installing the kernel, for some reason, neither the touchscreen nor the Wi-Fi started up. The rtl8xxxu Wi-Fi module was distorted manually and it came to life. The touchscreen came to life on its own after a reboot. In general, so far everything starts flawlessly, but as comrade said above, the Wi-Fi works only next to the access point (it’s bad with sensitivity) and it’s in trouble with speed. Right near the access point a maximum of one and a half megabits. It will be necessary to deal with the causes of this behavior wifai.

Hello to all. I haven't been here for a long time.
Yesterday I updated the system on my stylus from my 15.10 to 16.04.1 lts. The core has left its self-assembly. The update went well, but many of my interface "buns" were expectedly dropped due to a change in the version of the gnome. Otherwise, everything is relatively normal. I do not observe any jambs, but I do not specifically look for them. I just use the tablet as usual.
Today, hands reached to restore the distorted Ainol mini pc (I wrote about it earlier). Everything turned out great. Programmer rolled kindly provided by one good man bios in the branch about Ainol. So now I have the experience of splicing a device with a BayTrail processor. If I need anything, ask what I can, I will help.
Next week I want to try different new kernels on our tablet.

Guys, does this solution support force feedback (reverse, vibration)?

I puzzle, what from non-oculus to take? Choose between OSVR and DeePoon E2. The first one is more expensive and more bells and whistles, but considering my goal of using a device under Linux, it has problems with this system. The second is less sophisticated but more trouble-free (in Linux) and cheaper. Who can say something in favor of choosing one or the other?

Strange, I have no problem

Hello to all. He looked into the topic in the hope of finding here a fresh Andrew, but alas. My device is e7ed. HDMI works, I practically do not use Windows because of the jambs on the edges of the screen. I sit mostly on android. About two weeks ago, I noticed that the Switch On cant disappeared by itself at 100% charge on the android. What is the reason - I do not know. Absolutely nothing he did. Overheating problem solved by adding thermal paste. It seems to hold. In the tanks runs with a bang and trotting does not go away.
Yes, at the expense of charging from my own power supply: I changed a few cables while it began to charge normally. Now generally I use a short cable from a bank. And it charges decently fast. Hour two is subjective.

In general, everything is very relative. As I said earlier, the system I have is installed on a microSD card, and on the SSD only the boot partition in ext4. It goes without saying that the SD card is much slower than the SSD, and this is where the use of btrfs with compression is very important. And the place saves pretty.

I recently also drew a problem here. When I exit the standby mode, I get a full hang. Quite often, about 1-2 times out of 10. Previously, this was not observed. It looks like a cant with some kind of update has arrived.

dmig

Of course it is necessary! Positive experience is priceless. And if “stepped on a rake” - be sure to announce this, so that others know “where to lay the straw”.

What problems?

Post has been editedevgen-y2k - 17.04.16, 08:03

RussianNeuroMancer,

Here is a pancake, you will force to thoroughly understand!
In general, the download options I applied are the same as on one of my tablets on the Bay Trail. It was lazy to check every parameter and I just ordered what I needed.
In general, we have the following:

usbcore.autosuspend default = 2
i915.enable_rc6 default = 1, can be excluded from boot options
i915.enable_fbc = -1, leave the parameter
lvds_downclock - surprised, the latest version of the driver does not have this parameter :) We exclude from the kernel parameters as useless.


The acpi_osi = Linux parameter I applied based on previous experience on Bay Trail. Then I decompiled dsdt.dat and by itself did not find any mention of Linux there. However, the use of acpi_osi = "[mentioned version of Windows]" did not lead to anything. But the application of the not mentioned acpi_osi = Linux miraculously launched the backlight brightness control :)
Here is the same situation. Now I again decompiled dsdt.dat (cat / sys / firmware / acpi / tables / DSDT>dsdt.dat; iasl -d dsdt.dat) and in the resulting file I see only Windows 2009/2012/2013/2015. You can try to play with this parameter.


In this case, no. Rather, as insurance / guarantee :)



You're welcome :)

So, based on the above, you can bring the kernel parameters string to this form:

usbcore.autosuspend = 1 pcie_aspm = force i915.enable_fbc = 1 acpi_osi = "Windows 2015" acpi = force drm.vblankoffdelay = 1 workqueue.power_efficient
Oh how interesting it turns out!
I measured consumption with a new line of kernel parameters. All other conditions are preserved. So, 10% of the battery tablet "drank" in exactly 30 minutes!
Not difficult calculation gives us 300 minutes or 5 hours !!! And why? Okay, I restore the old parameters and repeat the test. Let's see who we have here ... :)

Advanced ->ACPI Settings ->Enable ACPI Auto Configuration = Enabled
Advanced ->Platform Misc Configuration ->Native ASPM = Enabled
I do not remember the account Advanced ->Platform Misc Configuration ->Native PCIE Enable, if disabled, enable
Advanced ->Platform Misc Configuration ->Power Sharing Manager Configuration ->Power Sharing Manager = Enabled
Advanced ->Platform Misc Configuration ->Wireless device ->Wireless device = Enabled

Chipset ->PCH-IO Configuration ->PCI Express Configuration ->DMI Link ASPM Control = Enabled
Chipset ->PCH-IO Configuration ->PCH Azalia Configuration ->Azalia PME = Enabled

On this like everything.

I would be very grateful to the person who will test all this on the system installed on the SSD.

Post has been editedevgen-y2k - 31.03.16, 13:51

RussianNeuroMancer,
I want 2 systems, and the place is not enough
Conjured a little in the field of consumption of electricity. Tested. Got a result7 hours 43 minutes continuous work. How to test:
With a fully charged battery, it loaded, launched the browser, opened 5 tabs, the active tab with constantly updated forex quotes (so that WiFi did not fall asleep and browser activity in the form of JS was). The tablet stood on its original keyboard, i.e. + keyboard and touchpad consumption. The screen brightness is 25%, the dimmer is off (always one brightness). CPU on Power Save (800 MHz). Bluetooth is disabled. Time control - the script once a minute writes uptime to the file.

Kernel Parameters:
usbcore.autosuspend = 1 pcie_aspm = force i915.enable_rc6 = 1 i915.enable_fbc = 1 i915.lvds_download = 1 acpi_osi = Linux acpi = force drm.vblankoffdelay = 1quequeue.power_efficient

Let me remind you, the system on my SD card, the consumption in this situation is more than if the system was installed on an SSD.
In the attachment file with the parameters for autoload (registered in /etc/rc.local)
Checked 4 browsers: Firefox, Opera 36, ​​Opera-develop 37, Chrome. Firefox turned out to be the worst. The most economical - Opera 36
Very pleased with the result.

UPD:
I almost forgot, in BIOS I turned on a number of parameters regarding ACPI and ASPM. Without this, the kernel parameter pcie_aspm = force did not work. To check performed dmesg | grep PCIe



pm.rar(356 bytes)

artyom12,
What's the point? All that is needed is in this thread. Put Ubuntu gnome 15.10, put the kernel from this branch, the screen rotation script, etc ...
Guys, share your energy efficiency information under Linux. My system at 40-50% brightness, autostart powertop - auto-tune and Power Save CPU mode with a simple fermentation on the Internet lives a little more than 5 hours. Someone like? And who uses what tricks to extend battery life?

I will add a small clarification. My system is installed on the SD card. As a result, the card reader (USB2.0-CRW) is the most devouring device, and there's nothing you can do about it. I think that if the system is installed on the SSD, then the consumption will be much lower.


Post has been editedevgen-y2k - 29.03.16, 09:33

So, I solved the problem with multitouch gestures by assembling touchegg and configurator for it touchegg-gce. The first one was to build the git version because ubuntu in turnips is a great option. The second is generally not in the repository. All gathered without problems. It turned out that the softphone is convenient and very useful. Everything is configured using a graphical utility is not easy, but very simple. Highly recommend. If anyone needs a config-speak, I'll post it.
I hasten to share a pleasant (for me) surprise. I put opera-developer 37.0.2171.0 from the repository and it turned out that everything in it works and there is no need for dancing with a tambourine! All multitouch gestures are handled correctly, the dpi is determined correctly and a factor of 124 is applied (it was in the stable opera too), with pepperfreshplugin installed, flash video is decoded using gpu hardware capabilities, but I don’t remember exactly whether it was forcibly or immediately included in hardware decoding html5 earned. Just a lot of things included in opera: // flags. But it does not matter, the main thing that works! The need for touchegg specifically for the browser has disappeared, although the gestures settings for firefox I left.
As a result, linux is now much more convenient to work in my environment than in win 10. And if you recall the incorrect operation of the on-screen win 10 on-screen keyboard with all browsers other than Microsoft's native ones, then there is nothing to compare.

Strange somehow, it's one module :)
Boasting small victories:
I started dolphin-emulator with the original wii remote connected and other body kit. Fully playable. Some games slow down, but most work fine even with the stereo mode turned on, which allowed you to connect when connected to a 3D TV and to a homemade virtual reality helmet.

I wrote down the scheme for gnome-shell so that it was convenient at our screen resolution. Small shoals of the background type of some buttons, color and size of fonts in some places remained. How to finish combing - lay out.

A bit optimized screen rotation script. Decreased sensitivity.

I painted a touchscreen disable / enable script. Actual when working with a pen. Here are just where to tie it? For now, just run if necessary.

I will post the new things later, now the tablet is discharged.
I post the updated screen rotation script and the touchscreen disable / enable script



rotate.rar(497 bytes)
ts_switcher.rar(308 bytes)

Convenient console utility for organizing groups (folders) in the gnome-shell dashboardTyts

I tried ubuntu 15.10 with a gnome. With the native core (I have 4.2.0-27 generic), it was necessary to assemble modules for Wi-Fi, Bluetooth and accelerometer. I found a 4.4 core kernel patched to our hardware (SOURCE). I assembled the core, packed it in .deb packages (attached). Out of the box, everything works. We put the core, test, unsubscribe by results.
To auto rotate the screen, you need to put the rotate.py file in autoload. This is a slightly different script from me.
I have a system installed on an SD card. The btrfs file system. The / boot partition on the built-in SSD is 200Mb in size, but it's probably worth a bit to increase it to 300Mb. / boot as a partition on the SSD is required to boot the system. Our tablet does not know how to boot directly from the SD card. I chose the btrfs file system for the root partition based on two considerations: 1 - saving space by compressing files on the fly, and 2 - increasing the speed of working with an SD card due to compression / decompression on the fly :)
Now let's talk about the comfortable value of DPI. On our diagonal for me this is 150. GNOME unfortunately cannot flexibly adapt to the required DPI. Scaling is specified only by integer values ​​of type 1, 2, etc. This value is a multiplier for a rigidly embedded gnome DPI = 96. Thus, if we set scaling to 2, it will give us 96x2 = 192, which is clearly a bust. There is another parameter, the scaling factor of the text. I put it in 1.12 and it slightly improved the situation, but did not decide radically.
Chromium browser does not want to scale perfectly, but has a plus normal multitouch support. Still looks too small. Therefore, I use firefox with installed extensions for convenience. Here is a list of extensions:
Grab and drag - convenient finger control.
Tab kit 2nd edition - a handy customizable tabs organization and the ability to change the DPI as a whole for the browser. DPI set to 150.
Zoom page - zoom button. Because The usual two-finger zoom does not work, I use the button when I need it.
Now about the problems.
When installing the tablet on its native hardware keyboard, the touchscreen is periodically disabled. Who is knocking him down - until he figured out, but it seems that this dwarf is bizarre. In the xinput output, the touchscreen is present and to enable it either you need to overload the i2c_hid module (modprobe -r i2c_hid; modprobe i2c_hid), or activate it from xinput
(xinput enable "FTSC1000: 00 2808: 5012"). But the second option for some reason does not always work.
Also sometimes when you restart the touchscreen does not start. To fix this, you need to load the i2c_hid module with one of the last. I haven’t decided on this beautiful problem yet.
I use onboard as an on-screen keyboard. Also installed a gnome-shell extension called onboard integration. In theory, this extension was supposed to turn off the built-in flawed Caribou keyboard and completely replace it, but this did not happen. Caribou climbs out in the welcome window and when searching for applications in gnome-shell. In principle, it does not cause discomfort, but I would still like to disable the Caribou completely. Who is in the subject - I ask for a solution.
Now a little more details on the extensions for gnome-shell. They are all available athttps://extensions.gnome.orgset in one click. In firefox, the default firefox extension is pre-installed. Of the useful extensions, I note the following:
Battery status - replacement of the default indicator of the battery;
CPU Power Manager - CPU frequency control. Useful to save battery
Display Switcher is an add-on for xrandr that controls on / off on / off displays. Conveniently if you connect the tablet to a TV or external monitor
Net Speed ​​- displays the current network speed
Slide for Keyboard - open the on-screen keyboard via swipe from bottom to top
QuickLaunch - launch your .desktop files. Convenient to add your own functionality scripts.
GRUB Reboot - allows you to reboot into any of the available grub menu items. It did not work for me while I was using the grub-reboot command for target overload.

update 1:
I noticed an interesting thing, the touchscreen independently began to activate with a slight delay. Those. Put the tablet on the keyboard - the touch fell off. Potapal on the screen - seconds after 20-30 earned. I did not observe such behavior earlier in my native Yubunt core.

update 2:
Touchscreen starts if you touch the screen and hold it for a bit, like a long tap.


linux-headers-4.4.0-cube-i7_4.4.0-cube-i7-10.00.Custom_amd64.deb(7.63 MB)
linux-image-4.4.0-cube-i7_4.4.0-cube-i7-10.00.Custom_amd64.deb(45.14 MB)
rotate.rar(497 bytes)


Post has been editedevgen-y2k - 02.03.16, 09:57

Question to those who use my kernel from the header:
1. Is there a jerk in 3D? For the test, run glmark2 or play some kind of 3D shny. I have fading at the same time interval.
2. Does bluetooth work? If not, then you need to attach the firmware file.
I found the reason for the friezes of 3D graphics, which I mentioned earlier. The culprit script screen rotation. The error migrated from the original script. I post the corrected version of the screen rotation script.



rotate.rar(497 bytes)

Guys, does anyone know why multitouch gestures from the touchpad are practiced everywhere and always, but not from the touchscreen? Where to dig?