kernel26-one-dev development thread

Tera · 2009-05-10 11:03:18

kernel26-one-dev 2.6.30rc5 has been released with rather large list of changes. Please read them carefully to determine if they affect your particular setup in anyway.

List of changes and improvements for version 2.6.30rc5:

New acerhdf patch and it defaults to kernel mode now. No need for rc.local tricks anymore.
ACPI video and some LCD backlight code compiled in statically due a kernel bug.
ACPI powersave, userspace and conservative built as modules. Defaults to ondemand governor.
cpufreq_stats added as module. pciehp, acpiphp compiled statically.
NTFS read/write support is now available as a module.
Removed CONFIG_LATENCYTOP, CONFIG_FRAME_POINTER and CONFIG_SYSCTL_SYSCALL_CHECK.
Removed Hardware crypto devices support completely.
Removed debugging code to make kernel slightly faster, but unfortunately this means a dropped support for the LatencyTOP tool.
Dropped support completely for the wifi led patches as they don't apply to the 2.6.30 kernel tree anymore.
        - Correction: Only the separate patch is removed. LED patches are now part of the official 2.6.30 kernel tree.
Added a patch which tweaks the deadline scheduler's fifo_batch from 16 to 1 for slightly better SSD responsiviness.
Added support for the KMS.

Kernel's boot time has increased from 3.7 seconds to 4.8 seconds on my system. Probably due the fact that 
more static code has been compiled into the kernel than on previous releases. Init seems to be working faster though(??).
Your mileage may vary depending on the external peripherals and filesystems used.

The time to suspend the laptop has decreased (tested with s2ram). It's almost immediate now.

I hope that all the future discussion of kernel26-one-dev will be done on this thread,

Last edited by Tera (2009-05-14 13:02:37)

Tera · 2009-05-10 11:08:03

Binary package is available from http://dolly.rulex.net/~giga/arch/kerne … pkg.tar.gz for those who are too lazy to compile.
MD5SUM is 1bd87bcd2ad14736c071dd275a651518

nimoworm · 2009-05-10 18:27:46

For all with integrated 3G modem (Aspire One A150X-3G)
CONFIG_USB_SERIAL_QUALCOMM=m
in config is necessary. Thanks!

Tera · 2009-05-10 21:14:24

Thanks. I will include this modification in the next release.

froli · 2009-05-11 06:46:03

GDM doesnt start!!

thats the lastest log entry in /var/log/gdm/

    Before reporting problems, check http://wiki.x.org
    to make sure that you have the latest version.
Markers: (--) probed, (**) from config file, (==) default setting,
    (++) from command line, (!!) notice, (II) informational,
    (WW) warning, (EE) error, (NI) not implemented, (??) unknown.
(==) Log file: "/var/log/Xorg.0.log", Time: Mon May 11 02:29:04 2009
(==) Using config file: "/etc/X11/xorg.conf"
(EE) intel(0): [drm] Failed to initialized agp heap manager
Xorg: intel_bufmgr_gem.c:766: drm_intel_gem_bo_unmap: Assertion `bo_gem->mem_virtual != ((void *)0)' failed.

Hope it will work, only by the booting screen I saw a improvment of the display quality

Edit: I think it's more related to Xorg than GDM though...

Last edited by froli (2009-05-11 06:47:05)

Tera · 2009-05-11 08:18:39

It looks like the Intel's driver is the problem here which is more than possible if you are trying to run a new kernel with very old driver. What is your driver version? I recommend trying the driver from testing repo. If you are already running it, try the xf86-video-intel-git with intel-dri-git AUR. With this combination all my UXA+KMS related issues were gone and I get 480fps in not-a-benchmark glxgears.

I'd be happy to hear if you get your Xorg working so that I ca make note of a minimum working driver version with this kernel.

Zom · 2009-05-11 10:52:10

How do I set the fanon, fanoff and interval stuff for acerhdf with this kernel?

kTT · 2009-05-11 13:35:37

put

acerhdf.verbose=0 acerhdf.fanon=62 acerhdf.fanoff=52 acerhdf.interval=10

in kernel line in your menu.lst file

Tera · 2009-05-11 13:40:44

Zom wrote:

How do I set the fanon, fanoff and interval stuff for acerhdf with this kernel?

Check http://piie.net/files/acerhdf_README.txt, section 2.6. Module loading does not apply on this case, so you have to use the "/sys/" interface.

EDIT: For people who are not sure where to put such commands, you can put them into /etc/rc.local and they are automatically executed on every boot.
EDIT2: seems like kTT was posting while I was writing my own. He's trick works as well.

Last edited by Tera (2009-05-11 14:01:39)

Zom · 2009-05-11 13:48:00

Just what I was looking for, since I didn't want to clutter my menu.lst. Thank you.

Though, shouldn't I be able to use sysctl.conf for this?

Oh, and I suppose I could say this too, suspend-to-ram still doesn't work. I've tried with both s2ram -f and pm-suspend, and both suspend fine but completely locks the system on resume. Only a blank screen (Or in s2ram, the console login prompt) with no response from keys, not even Caps Lock.

I'll continue investigating this issue.

Last edited by Zom (2009-05-11 13:50:20)

Tera · 2009-05-11 14:05:40

Zom wrote:

Though, shouldn't I be able to use sysctl.conf for this?

I've been in touch with the developer himself and he said that it does not YET has support for sysctl but it might be coming after it hits the mainline kernel.

Zom wrote:

Oh, and I suppose I could say this too, suspend-to-ram still doesn't work. I've tried with both s2ram -f and pm-suspend, and both suspend fine but completely locks the system on resume. Only a blank screen (Or in s2ram, the console login prompt) with no response from keys, not even Caps Lock.
I'll continue investigating this issue.

Sounds like something is borked because it works lovely on my machine and I've had succesfull reports from other people as well. When you've tried to suspend, have you had wifi on or off? What's your Intel driver version? Can you try to update it?

EDIT: Can you try to suspend while Xorg is not running at all and report back the results.

Last edited by Tera (2009-05-11 14:09:16)

froli · 2009-05-11 16:39:17

Tera wrote:

It looks like the Intel's driver is the problem here which is more than possible if you are trying to run a new kernel with very old driver. What is your driver version? I recommend trying the driver from testing repo. If you are already running it, try the xf86-video-intel-git with intel-dri-git AUR. With this combination all my UXA+KMS related issues were gone and I get 480fps in not-a-benchmark glxgears.
I'd be happy to hear if you get your Xorg working so that I ca make note of a minimum working driver version with this kernel.

My xorg.conf is good, I'm using it on the 2.29.rc8git right now.

Heres my xorg.conf:

Section "ServerLayout"
       Identifier     "Default Layout"
       Screen      0  "Screen0" 0 0
       InputDevice    "Synaptics Mouse" "AlwaysCore"
       InputDevice    "Keyboard0" "CoreKeyboard"
       InputDevice    "USB Mouse" "CorePointer"
EndSection

Section "ServerFlags"
       Option "AllowMouseOpenFail"  "true"
       Option "AutoAddDevices" "False"
EndSection

Section "Module"
       Load "synaptics"
EndSection

Section "Files"
       ModulePath   "/usr/lib/xorg/modules"
       FontPath     "/usr/share/fonts/misc:unscaled"
       FontPath     "/usr/share/fonts/misc"
       FontPath     "/usr/share/fonts/75dpi:unscaled"
       FontPath     "/usr/share/fonts/75dpi"
       FontPath     "/usr/share/fonts/100dpi:unscaled"
       FontPath     "/usr/share/fonts/100dpi"
       FontPath     "/usr/share/fonts/PEX"
# Additional fonts: Locale, Gimp, TTF...
       FontPath     "/usr/share/fonts/cyrillic"
#       FontPath     "/usr/share/lib/X11/fonts/latin2/75dpi"
#       FontPath     "/usr/share/lib/X11/fonts/latin2/100dpi"
# True type and type1 fonts are also handled via xftlib, see /etc/X11/XftConfig!
       FontPath     "/usr/share/fonts/Type1"
       FontPath     "/usr/share/fonts/ttf/western"
       FontPath     "/usr/share/fonts/ttf/decoratives"
       FontPath     "/usr/share/fonts/truetype"
       FontPath     "/usr/share/fonts/truetype/openoffice"
       FontPath     "/usr/share/fonts/truetype/ttf-bitstream-vera"
       FontPath     "/usr/share/fonts/latex-ttf-fonts"
       FontPath     "/usr/share/fonts/defoma/CID"
       FontPath     "/usr/share/fonts/defoma/TrueType"
       FontPath     "/usr/share/fonts/artwiz-fonts"
       FontPath     "/usr/share/fonts/local"
EndSection

Section "InputDevice"
       Identifier  "Keyboard0"
       Driver      "keyboard"
       Option      "CoreKeyboard"
       Option "XkbRules" "xorg"
       Option "XkbModel" "pc105"
       Option "XkbLayout" "ca"
       Option "XkbVariant" "fr"
#       Option "XkbVariant" "nodeadkeys"
EndSection

Section "InputDevice"
       Identifier "Synaptics Mouse"
       Driver     "synaptics"
       Option     "Device" "/dev/psaux"
       Option     "Protocol" "auto-dev"
       Option     "LeftEdge"  "1700"
       Option  "RightEdge"     "5300"
       Option  "TopEdge"       "1700"
       Option  "BottomEdge"    "4200"
       Option  "FingerLow"     "25"
       Option  "FingerHigh"    "30"
       Option  "MaxTapTime"    "180"
       Option  "MaxTapMove"    "220"
       Option  "VertScrollDelta" "100"
       Option  "MinSpeed"      "0.09"
       Option  "MaxSpeed"      "0.15"
       Option  "AccelFactor"   "0.0015"
       Option  "SHMConfig"     "on"
# new in synaptics 0.99
       Option  "ClickFinger1"  "1"
       Option  "ClickFinger2"  "0"
       Option  "ClickFinger3"  "0"
       Option  "HorizTwoFingerScroll"  "0"
       Option  "VertTwoFingerScroll"   "0"
       Option  "HorizScrollDelta"      "100"
       Option  "PressureMotionMinZ"    "10"
       Option  "FingerPress"   "256"
       Option  "PalmDetect"    "0"
       Option  "PalmMinWidth"  "10"
       Option  "PalmMinZ"      "200"
       Option  "MaxTapMove"    "220"
       Option  "MaxTapTime"    "180"
       Option  "MaxDoubleTapTime"      "150"
       Option  "TapButton1"    "1"
       Option  "TapButton2"    "0"
       Option  "TapButton3"    "0"
       Option  "RTCornerButton"        "2"
       Option  "RBCornerButton"        "3"
       Option  "LTCornerButton"        "0"
       Option  "LBCornerButton"        "0"
# Circular scrolling is uber-cool, but it's not for everyone. Check out "gsynaptics" as well.
       Option  "CircularScrolling"     "0"
# Scrolling with the right and bottom side can be fun... or incredibly annoying. Use "1" to enable.
       Option  "HorizEdgeScroll"       "1"
       Option  "VertEdgeScroll"        "1"
EndSection

Section "InputDevice"
       Identifier      "USB Mouse"
       Driver          "mouse"
       Option          "Device"                "/dev/input/mice"
       Option          "SendCoreEvents"        "true"
       Option          "Protocol"              "IMPS/2"
       Option          "ZAxisMapping"          "4 5"
       Option          "Buttons"               "5"
EndSection

Section "Monitor"
       Identifier  "Monitor0"
       Modeline  "1024x600" 48.96 1024 1064 1168 1312 600 601 604 622 -HSync +VSync
       DisplaySize 346 203 # 75 DPI @ 1024x600
EndSection

Section "Device"
       Identifier  "Videocard0"
       Driver      "intel"
       Option      "Clone" "true"
       Option      "MonitorLayout"     "LVDS,VGA"
       BusID       "PCI:0:2:0"
       Option      "MigrationHeuristic" "greedy"
       Option      "AccelMethod" "EXA"
       Option      "NoDCC"
EndSection

Section "Screen"
       Identifier "Screen0"
       Device     "Videocard0"
       Monitor     "Monitor0"
       DefaultDepth     24
       SubSection "Display"
               Viewport   0 0
               Depth     24
               Modes    "1024x600" "800x600" "640x480"
               Virtual 1920 1800
       EndSubSection
EndSection

Section "DRI"
       Mode 0666
EndSection

Will try with the testing version of the driver then I'll put results here.

EDIT: I just got an update of extra/xf86-video-intel. From version 2.6.3 to version 2.7.0. Hope It'll be enough...

Last edited by froli (2009-05-11 16:43:04)

froli · 2009-05-11 17:10:58

It works now with extra/xf86-video-intel-2.7.0

Zom · 2009-05-12 09:42:01

Tera wrote:

I've been in touch with the developer himself and he said that it does not YET has support for sysctl but it might be coming after it hits the mainline kernel.

Sweet.

Tera wrote:

Sounds like something is borked because it works lovely on my machine and I've had succesfull reports from other people as well. When you've tried to suspend, have you had wifi on or off? What's your Intel driver version? Can you try to update it?
EDIT: Can you try to suspend while Xorg is not running at all and report back the results.

When I tried to suspend, wifi was active, but I wasn't connected to any network. I'm using whatever version that comes with the kernel.

I will, later on.

I think the trouble might be coming from the sdhc-card itself. Sometime ago when I tried to suspend, it borked the filesystem. After restoring it using fdisk, it appears as if I have two partitions, one on mmblk0 and one on mmblk0p1. I was meaning to remove the partition table and reformat the whole stick, but I forgot and just formatted mmblk0p1 to xfs.

Now, it might sound far-fetched, but I guess the problem could be that mmblk0 seems to have ext4 as a filesystem. Either way, I'll try to fix this first and then see if suspend still doesn't work.

tom_g · 2009-05-12 11:33:23

Tera,

Thank you for all the work you have done on the one-dev kernel. This version has fixed a few problems I was having.

One request I have for future versions is to include CONFIG_BT_RFCOMM_TTY=y for rfcomm serial connections (eg. for accessing a dial-up service (GPRS/HSDPA) through a bluetooth enabled phone).

Tera · 2009-05-12 14:22:51

Zom wrote:

Now, it might sound far-fetched, but I guess the problem could be that mmblk0 seems to have ext4 as a filesystem. Either way, I'll try to fix this first and then see if suspend still doesn't work.

I use ext4 with journal on my SDHC, which is mounted as /home. Everything works fine.

Zom wrote:

I'm using whatever version that comes with the kernel.

What do you mean? Intel driver does not come with the kernel, only some low level kernel code. You can determine your driver version with command "pacman -Qs xf86-video-intel". Please check that you are using at least version 2.7.0, from the [extra] repo.

tom_g wrote:

Thank you for all the work you have done on the one-dev kernel. This version has fixed a few problems I was having.
One request I have for future versions is to include CONFIG_BT_RFCOMM_TTY=y for rfcomm serial connections (eg. for accessing a dial-up service (GPRS/HSDPA) through a bluetooth enabled phone).

You're welcome. I will add it to the next release which I will probably release at the time of rc6.

Zom · 2009-05-12 19:39:47

Tera wrote:

I use ext4 with journal on my SDHC, which is mounted as /home. Everything works fine.

Weird.

Tera wrote:

What do you mean? Intel driver does not come with the kernel, only some low level kernel code. You can determine your driver version with command "pacman -Qs xf86-video-intel". Please check that you are using at least version 2.7.0, from the [extra] repo.

Oh, sorry, was thinking of wireless cards with low blood-sugar. Yeah, I'm using version 2.7.0 of the intel driver.

What's strange about it is that iit worked with the stock kernel when it was mounted by HAL in /media and unmounted by a script in /etc/pm/sleep.d/, so I haven't got a clue what the difference is really. Well, except the mount point. Perhaps it's remounted too late in the resume process, and since I'm running KDE it makes X hard lock for some reason or other?

I'll give it some try in openbox later.

kaola_linux · 2009-05-13 03:33:19

Thanks for updating your kernel tera...Suspend and Hibernate worked flawlessly on my AA1 with kdemod 4.2.3..:)

Vyazovoi · 2009-05-13 15:20:59

I missed the control of cpu freq after updating =/

Tera · 2009-05-13 17:31:57

Vyazovoi wrote:

I missed the control of cpu freq after updating =/

ACPI ondemand is default. There are several others to choose from. As the announcement states, other ACPI governers are compiled now as modules. You need to put the ACPI module into "modules" array in rc.conf and configure your laptop accordinly if you are not satisfied with the default choice.

Tera · 2009-05-13 17:37:21

Zom wrote:

What's strange about it is that iit worked with the stock kernel when it was mounted by HAL in /media and unmounted by a script in /etc/pm/sleep.d/, so I haven't got a clue what the difference is really. Well, except the mount point. Perhaps it's remounted too late in the resume process, and since I'm running KDE it makes X hard lock for some reason or other?
I'll give it some try in openbox later.

Okey, this time i'm not sure what you mean by "remounting". Do you have a script which unmounts your SDHC before suspend? Because I don't. Try taking off your "scripts" of any kind and trying with plain suspend. If that doesn't work, can you try doing a suspend while Xorg is NOT running at all? This would at least give a clue if it's an Xorg issue, or not.

Zom · 2009-05-13 21:08:27

Tera wrote:

Okey, this time i'm not sure what you mean by "remounting". Do you have a script which unmounts your SDHC before suspend? Because I don't. Try taking off your "scripts" of any kind and trying with plain suspend. If that doesn't work, can you try doing a suspend while Xorg is NOT running at all? This would at least give a clue if it's an Xorg issue, or not.

I had a script that unmounted the card when suspending and used a kde-plasmoid to auto-mount it again on resume. I disabled this when I switched to your kernel, since it was only neccessary for the stock one.

I'll give suspending without X a try now.

E: Both s2ram -f and pm-suspend work flawlessly in only console. Going to try it in openbox now.

E2: Neither work in openbox. Both just cause the computer to hang. The LED-indicator doesn't even respond to switching capslock on and off. How odd.

Oh, I'm running KMS and UXA. This worked before, do you suppose it could make any difference with your kernel?
I'm using neiter an initrd nor splashy, so can't be those either I suppose.

Last edited by Zom (2009-05-13 21:26:59)

Vyazovoi · 2009-05-14 03:57:47

Tera wrote:

ACPI ondemand is default

But cat /proc/cpuinfo | grep MHz always:

cpu MHz : 1595.881
cpu MHz : 1595.881

cpufreq-info
no or unknown cpufreq driver is active on this CPU

psch · 2009-05-14 07:41:29

I do have trouble with cpufreq, too.

According to:

$ zcat /proc/config.gz | grep -i cpufreq
# CPUFreq processor drivers
# CONFIG_X86_ACPI_CPUFREQ is not set
# CONFIG_X86_CPUFREQ_NFORCE2 is not set

It seems to simply be not enabled. Unless of course I'm looking in the wrong spot.
Anyway, any pointers would be appreciated.

Add: I did compile the kernel myself, via the PKGBUILD on my desktop system, in case that could be relevant.

Add2: recompiling the kernel with

CONFIG_X86_ACPI_CPUFREQ=y

set in the config resolved the cpufreq issue. Seems like the uploaded config in the tarball wasn't correct.

Edit: something weird i just noticed: The announcement says, WiFi LED support has completely been dropped, yet they are functional for me.
uname does report the proper kernel, as well as zcat /proc/config.gz | grep -i localversion.
I'm confused.

Last edited by psch (2009-05-14 08:19:26)

RayNbow · 2009-05-14 08:41:27

psch wrote:

I do have trouble with cpufreq, too.
Edit: something weird i just noticed: The announcement says, WiFi LED support has completely been dropped, yet they are functional for me.

I haven't upgraded to this one-dev version, but I think Tera means he has removed the LED patches since 2.6.30 now supports the WiFi LED out of the box:

Kernel Log: What's coming in 2.6.30 - Network: New Wi-Fi drivers and other network novelties
...
Wi-Fi
* ...
* ath5k: support LEDs on Acer Aspire One netbook
* ...
...

Arch Linux

#1 2009-05-10 11:03:18

kernel26-one-dev development thread

#2 2009-05-10 11:08:03

Re: kernel26-one-dev development thread

#3 2009-05-10 18:27:46

Re: kernel26-one-dev development thread

#4 2009-05-10 21:14:24

Re: kernel26-one-dev development thread

#5 2009-05-11 06:46:03

Re: kernel26-one-dev development thread

#6 2009-05-11 08:18:39

Re: kernel26-one-dev development thread

#7 2009-05-11 10:52:10

Re: kernel26-one-dev development thread

#8 2009-05-11 13:35:37

Re: kernel26-one-dev development thread

#9 2009-05-11 13:40:44

Re: kernel26-one-dev development thread

#10 2009-05-11 13:48:00

Re: kernel26-one-dev development thread

#11 2009-05-11 14:05:40

Re: kernel26-one-dev development thread

#12 2009-05-11 16:39:17

Re: kernel26-one-dev development thread

#13 2009-05-11 17:10:58

Re: kernel26-one-dev development thread

#14 2009-05-12 09:42:01

Re: kernel26-one-dev development thread

#15 2009-05-12 11:33:23

Re: kernel26-one-dev development thread

#16 2009-05-12 14:22:51

Re: kernel26-one-dev development thread

#17 2009-05-12 19:39:47

Re: kernel26-one-dev development thread

#18 2009-05-13 03:33:19

Re: kernel26-one-dev development thread

#19 2009-05-13 15:20:59

Re: kernel26-one-dev development thread

#20 2009-05-13 17:31:57

Re: kernel26-one-dev development thread

#21 2009-05-13 17:37:21

Re: kernel26-one-dev development thread

#22 2009-05-13 21:08:27

Re: kernel26-one-dev development thread

#23 2009-05-14 03:57:47

Re: kernel26-one-dev development thread

#24 2009-05-14 07:41:29

Re: kernel26-one-dev development thread

#25 2009-05-14 08:41:27

Re: kernel26-one-dev development thread

Board footer