Carolina: Vanguard Edition Release 2 December 2014

[battleshooter]

Not a woohoo! moment yet, sorry.

Not a problem, but what happened? Dmesg, Xorg.0.log, and output of cat /proc/bus/input/devices? You should know the drill by now

But if we don't have a fix by the time you release your update I won't be too disappointed.

Oh no, I'm not letting this stop the show, I just wanted to fiddle with some ALSA stuff before the next release as well as making sure the propriety drivers will install, Billtoo mentioned an issue with the nVidia one and I haven't tested the AMD yet either. I think it would be cool if I could package the propriety drivers but I haven't figure how to do that just yet.

[Gobbi]

Hi @battleshooter , could we have the following option enabled in the kernel : CONFIG_VFIO_PCI_VGA=y ? It could be used later in qemu .

[battleshooter]

Hi @battleshooter , could we have the following option enabled in the kernel : CONFIG_VFIO_PCI_VGA=y ? It could be used later in qemu .

Sure, what's it for in qemu? It's got quite a daisy chain of other kernel option requirements.

[Gobbi]

KVM-VGA passthrough .

[battleshooter]

Ahh yes, I see. While I compiled the main KVM drivers, I have no idea what other drivers are necessary. Do you have any others you can recommend?

[Gobbi]

http://www.flaterco.com/kb/RAID-0-root.html

An interesting thing would be to enable from the kernel to autodetect and assemble RAID and LVM , if possible , at start.

Also , according to this : http://www.pugetsystems.com/labs/articl ... 4-KVM-585/ , if you take look at Step 1 - other few things should be in : pci_stub , vfio , vfio_iommu ( but , not only for intel ) for KVM-VGA passthrough ...

[yerc1]

Not a problem, but what happened? Dmesg, Xorg.0.log, and output of cat /proc/bus/input/devices?

Just from memory the first kernel 3.18.1 build compared with build2 results:

Code: Select all

# modprobe chromeos_laptop

build1 did not produce any message.
build2 result is "FATAL: Module chromeos_laptop not found."

Code: Select all

# synclient -l

same output for both kernel builds
"Couldn't find synaptics properties. No synaptics driver loaded?"

only 2 error types in build2 compared with more in build1 for dmesg
type1 relates to internal drive mount error which is fine and type2 says [drm:intel_uncore_check_errors] which I don't have a clue about.

please see attachment.
thanks

[battleshooter]

Internet was down for a day or two, but at least that gave you a break from my constant harassing yerc1

It's silly but I keep finding these "vital" kernel modules that no one seems to mention is important for Chromebook support. Here are the two new ones for today:

Code: Select all

config MFD_CROS_EC
	tristate "ChromeOS Embedded Controller"
	select MFD_CORE
	help
	  If you say Y here you get support for the ChromeOS Embedded
	  Controller (EC) providing keyboard, battery and power services.
	  You also need to enable the driver for the bus you are using. The
	  protocol for talking to the EC is defined by the bus driver.

config MFD_CROS_EC_I2C
	tristate "ChromeOS Embedded Controller (I2C)"
	depends on MFD_CROS_EC && I2C

	help
	  If you say Y here, you get support for talking to the ChromeOS
	  EC through an I2C bus. This uses a simple byte-level protocol with
	  a checksum. Failing accesses will be retried three times to
	  improve reliability.

Or maybe it's just obvious. Sounds important anyway right?

build2 result is "FATAL: Module chromeos_laptop not found."

I compiled the module as builtin as that's how Tahr does it just for kicks. This round it's back to a module, so you won't get that error.

Thanks for the diagnostics too, it would be great if that was the attached standard whenever a piece of hardware doesn't work. Hmm, should probably stick a request like that on the first post...

Also Gobbi, compiled your modules (CONFIG_VFIO_PCI_VGA and dependencies) this round if you'd like to give them a whirl to see if everything is working correctly.

Besides the usual diagnostics package yerc1, could you also test the new modules please?

Code: Select all

modprobe cros_ec
modprobe cros_ec_i2c

As a side note, and something for me to babble about while I wait for the kernel to upload, after making a phone call to figure what happened to the internet, I decided to ask if ADSL was available in my area. Apparently it is now Still, I gave him a skeptical "Really?" but placed an order anyway to find out if it's for real.

http://puppylinuxstuff.meownplanet.net/ ... _3-3.18.1/

[yerc1]

that gave you a break from my constant harassing yerc1

Hahaha... not true! I worried I might come across as the one engaging in harassment which is not.

It's silly but I keep finding these "vital" kernel modules that no one seems to mention is important for Chromebook support.

What if your findings make their way to the HOWTO thread of the forum?
Up to you entirely, with Vanguard you've given enough to the puppy community.

I found in my online search that the few linux distros catering to intel chromebooks are almost all for Acer C720's.
And there's the assumption that if things work on the Acer C720 they would on other intel chromebooks too. Maybe not.

build2 result is "FATAL: Module chromeos_laptop not found."

I compiled the module as builtin as that's how Tahr does it just for kicks. This round it's back to a module, so you won't get that error.

Yep. Confirmed.

Besides the usual diagnostics package yerc1, could you also test the new modules please?

Code: Select all

modprobe cros_ec
modprobe cros_ec_i2c

Since you asked so nicely... jokes!!! I got

Code: Select all

# modprobe cros_ec 
FATAL: Error inserting cros_ec (/lib/modules/3.18.1/kernel/drivers/mfd/cros_ec.ko): Invalid module format

and

Code: Select all

# modprobe cros_ec_i2c
WARNING: Error inserting cros_ec (/lib/modules/3.18.1/kernel/drivers/mfd/cros_ec.ko): Invalid module format
FATAL: Error inserting cros_ec_i2c (/lib/modules/3.18.1/kernel/drivers/mfd/cros_ec_i2c.ko): Invalid module format

I decided to ask if ADSL was available in my area. Apparently it is now Still, I gave him a skeptical "Really?" but placed an order anyway to find out if it's for real.

Felt exactly the same each time I was told that and yet... I hope you'll have it this time.

I still get "Couldn't find synaptics properties. No synaptics driver loaded?" from synclient -l

EDIT: The "[drm:intel_uncore_check_errors]" are still in dmesg.
They might relate to this bug https://bugs.freedesktop.org/show_bug.cgi?id=69747
So too are the internal drive mount errors, which do not bother me.

dmesg now has a single message (not 30+) for

Code: Select all

[    1.506595] find_i2c_adapter_num: i2c adapter i2c-designware-pci not found on system.
[    1.506611] platform chromeos_laptop: Driver chromeos_laptop requests probe deferral 

[battleshooter]

Hahaha... not true! I worried I might come across as the one engaging in harassment which is not.

Excellent then, let the non-harassment commence again

Build 4!

http://puppylinuxstuff.meownplanet.net/ ... _4-3.18.1/

dmesg now has a single message (not 30+) for

Code: Select all

[    1.506595] find_i2c_adapter_num: i2c adapter i2c-designware-pci not found on system.
[    1.506611] platform chromeos_laptop: Driver chromeos_laptop requests probe deferral 

For some reason the errors didn't turn up in dmesg... did you send the right dmesg? Anyway, with today's diagnostics, run dmesg last so I can see any errors that might turn up while the other tests are probing.

Today's build has loads of new kernel options compiled in. Basically what I did was compare Tahr's DOTconfig and selected all the device drivers that Tahr's kernel supported but Vanguard didn't have ticked and then some, because some kernel options looked interesting

Marv, something that may interest you, I noticed while perusing the DRM/Intel section is that Tahr uses some experimental 2D acceleration driver, I compiled that, maybe that's where the mysterious frames are coming from?

[Marv]

Marv, something that may interest you, I noticed while perusing the DRM/Intel section is that Tahr uses some experimental 2D acceleration driver, I compiled that, maybe that's where the mysterious frames are coming from?

Send me to the cellar will you Oh well, the wood stove is there along with the Bay trail. Switched back and forth between the stock kernel and build 4, Xorg petted and all else ======. Also ran LxPupTahr 14.10 and LxPupUnicorn on the same hardware with panel plugins as close as possible etc given that they are Lxde and vlina is xfce4...
First FPS and air, averaged over a minute or so:
vlina, stock kernel : 2100 FPS Xorg 1.16 3.0 Mesa 10.3.2
vlina, build4 kernel : 1810 FPS Xorg 1.16 3.0 Mesa 10.3.2

LxPupTahr 14.10 : 2922 FPS Xorg 1.15.1 3.0 Mesa 10.1.3
LxPupUnicorn : 2060 FPS Xorg 1.16 3.0 Mesa 10.3.0

Something different pertaining to gmbus? shows up in the i915 setup on build4. All the other pups above are identical. Grepped dmesg outputs attached. I'll have a look at the Pentium M and core 2 duo laptops later today.

Edit: core 2 duo laptop with intel mobile 4 series chipset (Cantiga) video:
vlina stock and build 4 : 1318 FPS
LxPupTahr 14.12.1 : 1310 FPS
LxPupUnicorn : 1360 FPS
All drm and i915 setups identical, no gmbus in build 4 i915 grep.

Edit: Pentium M laptop with intel 855GM video:
vlina stock : 645 FPS
vlina build 4 kernel : 662 FPS (both stock and build 4 have 'pipe underrun messages in the dmesg | grep drm)
LxPupTahr 14.12.1 : 650 FPS
LxPupUnicorn : 670 FPS
All drm and i915 setups identical, no gmbus in build 4 i915 grep.

Except for the consarn Bay-Trail, some win some lose. On the Bay-Trail (intel 4000 HD video) build 4 kernel, something (2d accelerator?) really hurts FPS. Other than that it's FPS and air. Both kernels boot and run cleanly on all these machines.

[yerc1]

For some reason the errors didn't turn up in dmesg...

Hmmm not sure what happened. The errors were somewhere at the end, but checking on the attachment, I just realised the bzip'd files came up as one huge text file. bummer

Today's build has loads of new kernel options compiled in. Basically what I did was compare Tahr's DOTconfig and selected all the device drivers that Tahr's kernel supported but Vanguard didn't have ticked and then some, because some kernel options looked interesting

Have i got good news @battleshooter...

Touchpad IS enabled! But it's behaving the same as it does in Tahr.
I can navigate around with the touchpad, select an object, and launch programs but I don't get right-click and scroll options.

I don't know a way to configure it other than via FlSynclient, which continues to give me "Couldn't find synaptics properties. No synaptics driver loaded?" and closes as soon as I get rid of the message.

Are the below outputs relevant?

Code: Select all

# modprobe chromeos_laptop
# 
# modprobe cros_ec
FATAL: Error inserting cros_ec (/lib/modules/3.18.1/kernel/drivers/mfd/cros_ec.ko): Invalid module format
# 
# modprobe cros_ec_i2c
WARNING: Error inserting cros_ec (/lib/modules/3.18.1/kernel/drivers/mfd/cros_ec.ko): Invalid module format
FATAL: Error inserting cros_ec_i2c (/lib/modules/3.18.1/kernel/drivers/mfd/cros_ec_i2c.ko): Invalid module format
#

Exciting

Edit. Later: I looked at Fatdog64-700 beta2. The chromebook touchpad works with synaptics driver loaded hence able to adjust settings. I ran our tests on it and have attached the results. I also attached the contents of etc/X11/xorg.conf.d as FD seems to use settings in that directory instead of from xorg.conf file.

[Gobbi]

Hi @battleshooter
Thank you for the KVM-VGA passthrough kernel modules - you are much faster that I thought . Today I ordered a cheap NVIDIA video card to make those tries , but it should arrive in 9 days from now ... So I'll have to wait till then to try the 3-rd or 4-th or... the last version it will be .
I tried to install the AMD (amd-14.12... for linux) module in vlina1.3 , the first edition , and it went so and so , there were some errors .
I clicked directly on the executable . After I saved the log files , I got out from X - to prompt - and after aticonfig --initial , I ran xwin . There is video (flickers on video playback) , audio , and AMD Catalyst Control Centre opens , but the video quality on my LGTV through HDMI is better in Carolina 1.2 (no flickers) where I also almost (no HDMI audio) installed the module.
In the attachment :
vlina_1... is install.log file from /usr/share/ati/ .
vlina_2... is the warning I got in terminal when I tried to run make_install.sh from /lib/modules/fglrx/ .
vlina_3... is the make.sh.log file I found in /lib/modules/fglrx/build_mod/ .
vlina_4... is the log file when I ran from the terminal make.sh from /lib/modules/fglrx/build_mod/ .

[battleshooter]

Have i got good news @battleshooter...

Touchpad IS enabled!

Whoa! Finally!

I need to still read through the reports, but felt the need to acknowledge your excellent news

Off the top of my head, I seem to remember people deleting their /etc/X11/xorg.conf to let evdev catch the touchpad and activate it correctly. Could give that a go if you'd like while waiting for me to go through the reports. I think then they either edit /usr/X11R7/share/X11/xorg.conf.d/50-synaptics.conf or add 10-synaptics.conf to the same location with the configuration desired.

Are the below outputs relevant?

Hmm, I would have thought those modules were vital for touchpad operation... well maybe they still are and that's why synaptics is still down. We'll see.

Send me to the cellar will you

Sorry for making you trudge down Marv, really thought I was on to something Oh well, at least you won't have to test again when Vanguard Edition Build 2 (or should that be Build 5 because of all the kernel Builds?... Hmmm.) is released in a few days, I think this is the kernel I'm sticking with. (Hah, I said that last time and then rebuilt the kernel 3 times over)

Thank you for the KVM-VGA passthrough kernel modules - you are much faster that I thought

Heh, thanks

I tried to install the AMD (amd-14.12... for linux) module in vlina1.3 , the first edition , and it went so and so , there were some errors .

Yeah, a lot of patching is required for newer kernels. I don't think AMD's updated since kernels 3.0.x. I have a script that patches everything up so make will run cleanly. I'll get it up later today, see if it helps.

There is video (flickers on video playback) , audio , and AMD Catalyst Control Centre opens , but the video quality on my LGTV through HDMI is better in Carolina 1.2 (no flickers) where I also almost (no HDMI audio) installed the module.

I'm surprised the module is even useable, that error looks pretty show stopping.

Edit: I realized I use the AMD legacy driver package. I have to borrow a laptop to see if the non legacy drivers patch the same way, so no patches today

[Marv]

Off the top of my head, I seem to remember people deleting their /etc/X11/xorg.conf to let evdev catch the touchpad and activate it correctly. Could give that a go if you'd like while waiting for me to go through the reports. I think then they either edit /usr/X11R7/share/X11/xorg.conf.d/50-synaptics.conf or add 10-synaptics.conf to the same location with the configuration desired.

Attached is what I have used in older Carolinas. It is located at:

Code: Select all

usr/X11R7/share/X11/xorg.conf.d/ 

where the X11 is a symlink to usr/X11R7/lib/X11/xorg.conf.d.

Edit: I think the links are there in vlina but I get dizzy following them so I'll stop. I assume X is responsible for this scattergun of stuff

[partsman]

Wonderful Job battleshooter !!
Thank you !!
Manual frugal install ext3 partition to hard drive on my acer aspire 5732z laptop
All good so far !! Very well done !!
Thank you !!

[yerc1]

Off the top of my head, I seem to remember people deleting their /etc/X11/xorg.conf to let evdev catch the touchpad and activate it correctly. Could give that a go if you'd like while waiting for me to go through the reports. I think then they either edit /usr/X11R7/share/X11/xorg.conf.d/50-synaptics.conf or add 10-synaptics.conf to the same location with the configuration desired.

Attached is what I have used in older Carolinas. It is located at:

Code: Select all

usr/X11R7/share/X11/xorg.conf.d/ 

where the X11 is a symlink to usr/X11R7/lib/X11/xorg.conf.d.

Edit: I think the links are there in vlina but I get dizzy following them so I'll stop. I assume X is responsible for this scattergun of stuff

Thanks battleshooter and Marv.
I'll try this method when I'm back on the chromebook.

[battleshooter]

Edit: I think the links are there in vlina but I get dizzy following them so I'll stop. I assume X is responsible for this scattergun of stuff

I'll agree there. It should be straight forward, just dump everything in /usr/X11R7, but nooo, there's stuff in /usr/local/lib/X11, /usr/X11R7, /usr/X11R6, /usr/X11, /etc/X11, symlinks crossing everywhere... good times, good times.

Regarding the touchpad saga, according to the reports, Fatdog loads cros_ec and cros_ec_i2c, so they are the designated solution at the moment. I've PMed jamesbond asking (nicely of course ) if he could be troubled to send me the Fatdog DOTconfig so I can compare and figure why the modules aren't loading.

Then again if just deleting Xorg.conf fixes things I'd be happy with that solution too

Wonderful Job battleshooter !!
Thank you !! Cool
Manual frugal install ext3 partition to hard drive on my acer aspire 5732z laptop Very Happy
All good so far !! Very well done !!
Thank you !!

Thanks partsman! I have to say I appreciate the detailed report in addition to your generous words. I like them to compare dmesgs and what nots to see if everything is working as it should and if doesn't, the detailed report makes the problem easier to identify. I do get my kicks in weird ways... at least I'm in the right hobby...

On the other hand... *forehead slap* I've been making yerc1 compile these reports by hand over and over again when he could have done most of them with one click using sys-info! Sorry yerc....

[partsman]

Wonderful Job battleshooter !!
Thank you !! Cool
Manual frugal install ext3 partition to hard drive on my acer aspire 5732z laptop Very Happy
All good so far !! Very well done !!
Thank you !!

Thanks partsman! I have to say I appreciate the detailed report in addition to your generous words. I like them to compare dmesgs and what nots to see if everything is working as it should

battleshooter I just noticed that I skipped Billtoo's message about needing two reboots to get Broadcom wireless to work and I loaded a pet for Broadcom wireless so you may not want take that into consideration when looking at my detailed report
However ! I will retry tonight and resend another detailed report

I just wanted you to know sorry for any confusion my bad !

[yerc1]

On the other hand... *forehead slap* I've been making yerc1 compile these reports by hand over and over again when he could have done most of them with one click using sys-info! Sorry yerc....

I don't mind. Good getting my hands dirty with the terminal after a long long time