Plasma NM, modem Sony MD300 and more...

During the time I implemented ModemManager support in Solid I did the tests with my Sony MD300 modem. For anyone to use this modem in Linux do this as root (or use sudo):

1. Install networkmanager-0.8 and modemmanager-0.4;
2. Create the file /etc/udev/rules.d/50-md300modem.rules with the follwing contents:
   
   
   
   
   
   

   1. ACTION!="add", GOTO="3G_End"
   2. SUBSYSTEM=="usb_device", ATTRS{idProduct}=="d0cf", ATTRS{idVendor}=="0fce", PROGRAM+="md300modem.sh %p"
   3. SUBSYSTEM=="usb", ATTRS{idProduct}=="d0cf", ATTRS{idVendor}=="0fce", KERNEL=="ttyACM*", SYMLINK+="modem3G-%n"
   4. LABEL="3G_End"
   

3. Create the script /lib/udev/md300modem.sh with the following contents:
   
   
   
   
   
   

   1. #!/bin/sh
   2. echo 3 > /sys/$1/device/bConfigurationValue
   3. sleep 3
   4. # enables/activates the modem (turns radio on and registers to operator network).
   5. chat -V -s '' "AT +CFUN=1" 'OK' '' < /dev/ttyACM0 > /dev/ttyACM0

4. Run: udevadm control --reload-rules;
5. Click on Plasma NM's system tray icon -> Manage Connections -> Mobile broadband  -> Add -> GSM Connection and use the parameters:
   
   
   
   
   
   

   Number: *99#
   Username: <operator's username>
   Senha: <operator's password>
   APN: <operator's apn>
   Type: Any

   For instance my configuration is as follow:

   Number: *99#
   Nome do usuário: tim
   Senha: tim
   APN: tim.br
   Type: Any
   

   OBS: some cell phones, like my Samsung i8910, only connects if "Type" is set to "Any". The MD300 connects with any of the types listed.
   
6. Hook up the modem to the USB port, wait until it powers up and register itself to the operator network (takes about 30 seconds to do both);
7. In the Plasma NM initial window clicks on the connection you have created.

My modem connects only once in ethernet mode (default mode), to connect again I have to remove it from USB port and hook it up again. I do not know why that happens and one guy told me his MD300 does not behave like that. Oddly enough with wvdial and kppp, which uses ppp mode, it does have this problem.

To make it easier to set up a Mobile Broadband connection I have started to port the nm-applet connection wizard to KDE. It is still in early stage and I am working on it only during weekends. Probably it will be ready for KDE release 4.6 early next year. The initial window is like this:



It takes some time to understand how both nm-applet wizard and Plasma NM kcm modue work and since I can only work on it during weekends do not expected something usefull until September.

Plasma NM: testing mode

I am back to Brasil after a great time in Tampere and Helsinki. During Akademy I commited my changes to Solid and Plasma NetworkManagement to trunk, now anyone brave anough to use trunk can try them :-)

Today I have added a hack to Plasma NM to make the traffic graph work with ppp connections. The problem was that NetworkManager only provides the serial device name (e.g. /dev/ttyACM0 or /dev/ttyUSB0) and Plasma::DataEngine expects the network interface name (ppp0) used by the device. I hacked Plasma NM by changing the Plasma::DataEngine::connectSource parameter to always use ppp0 when the interface name contains ttyACM or ttyUSB. I know that is ugly and error prone, if anyone knows how to convert serial device names to network interface names let me know. The only way to do it I have found so far is monitoring /var/log/daemon (or /var/log/messages) for the line pppd[]: Connect: ppp0 <--> /dev/ttyACM0, but that is also ugly and the log file containing the needed information can be different on each platform KDE SC is installed.

I have also tested Plasma NM with my Samsung i8910 cell phone. It only connects if I set Plasma NM's tray icon -> Manage Connections -> Mobile Broadband -> -> Edit -> Type to "Any" (it was "Prefer3G" before the change). Maybe someone else also needs to change that to make his/her phone connect. The number to dial also needs to be *99# instead of *99***1# . That is strange because my Sony MD300 modem works with *99***1# and as far as I know *99# is a short name for *99***1#.

Modem Manager does not report the access technology (Edge, HSDPA, etc) for my cell phone. I have received the report that Nokia N95 also connects and MM also does not report the access technology for that phone too.

There are other changes I am working on, when they are ready I will let you know :-)

Sager NP7652: finally

Yesterday, I finally got my Sager NP7652 laptop (Clevo W765CUH barebone) in my hands. The LCD was replaced and there is no dead pixels (so far). I copied my Gentoo installation from my old Acer Ferrari 4005 to it to avoid the loooooong process of installing Gentoo from the begining, also because my Internet connection is slow and it would take almost a week to download the source code for all ebuilds. The configuration (click on the image to see it completely):

More info about this laptop on Xoticpc and Notebook Review.




This report is listed at
TuxMobil - Linux compatibility guides for laptops, notebooks, PDAs, mobile phones and GPS devices.

ACPI

Battery

/proc/acpi/battery/BAT0/state does not show the "present rate", which prevents KDE's powerdevil to measure how much remaining time I still have. I had this problem with my Acer Ferrari 4005 too, a BIOS update solved the problem that time. Unfortunately Sager has discontinued this laptop, which was released less than six months ago, those notebook sellers must be kidding to discontinue a notebook so fast. So until I find a BIOS update to solve this problem I am using the kernel's CONFIG_ACPI_CUSTOM_DSDT feature to replace the BIOS's DSDT by one I have edited by hand. To use the new DSDT:

1. Download it
2. Enable  CONFIG_ACPI_CUSTOM_DSDT: make menu_config -> Power management and ACPI options -> ACPI (Advanced Configuration and Power Interface) Support -> Custom DSDT Table file to include. Select the downloaded file.
3. Compile and install the new kernel.
4. Reboot, that is it.

Here is the diff of the original DSDT.

Fans

The fans run at full speed since kernel boots up, but if you press the third hot key button (the one with a M surrounded by an elipse) the fans run quietly, this is called "silent mode". If you press it again the fans run at full speed again. OBS: the processor will *not* run at full power when in silent mode. OBS2: when using ATI proprietary drivers (ati-drivers ebuild) instead of the opensource drivers (xf86-video-ati ebuild) the fans run quietly.

Speedstep

Driver: acpi-cpufreq.ko.

I tried to use phc-intel to undervolt the CPU C-states but it did not work as expected. It seems this Core i7 620m CPU recognise the undervolt and automatically reduces CPU performance instead of allowing it to crash. Do not worry, the crash is part of the tweaking process to get the voltage values to configure the phc-intel module. I tried all values from 20 mW to 1 mV at 2667 MHz clock and the CPU did not crash but the performance decreased when compiling the kernel. Since this CPU has turbo mode capability I guess it is it that is automatically decreasing CPU performance to avoid the crash. The clock stayed at 2667 MHz, but compiling the kernel was much more slower using 1 mV than with 20 mW.

Suspend to Disk

I am using TuxOnIce for suspend to disk. To install it:

• Apply the patch against the kernel.
• Enable it  make menu_config -> Power management and ACPI options ->Enhanced Hibernation (TuxOnIce) -> File Allocator, Swap Allocator (if you have a swap partition), Userspace User Interface support and install the new kernel.
• Install pm-utils ebuild/package: emerge -v pm-utils
• Edit /etc/pm/config.d/gentoo and change the line with SLEEP_MODULE to SLEEP_MODULE="tuxonice". Now KDE's powerdevil plasmoid will use TuxOnIce to hibernate.

Sometimes some programs setfaults after resume (if one of them is the init program then system will crash). If I clean the swap space up (swapoff -a; swapon -a) before suspending then there is no segfault but that is damn slow if there several hundreds of MB in the swap file. Well, I tried a different approach and created a dedicated hibernate file that seems to also fix the problem and is pretty fast too, so do:

Configure Tux On Ice

# echo "TuxOnIce" > /hibernation-file # dd if=/dev/zero bs=1M count=2048 >> /hibernation-file # echo /hibernation-file > /sys/power/tuxonice/file/target # cat /sys/power/tuxonice/resume UUID=ea1d2e190cfe4901ac80645f47046980:0x56f8000

Change the resume kernel parameter to the new inode location:

Kernel parameter

resume=/dev/sda6:0x56f8000

Add the line below to /etc/hibernate/common.conf

/etc/hibernate/common.conf

ProcSetting file/target /hibernation-file

Bluetooth

Cambridge Silicon Radio bluetooth controller, driver: btusb.ko.

Card Reader

JMicron SD/MS Host controllers, Secure Digital driver: sdhci-pci.ko, Memory Stick driver: jmb38x_ms.ko. Tested with SD and MS PRO Duo cards. Not tested but should work as well: MS, MS PRO, MS Duo, Mini-SD, MMC and RS cards.

DVD Burning

Works as expected, at least when burning an iso image to a DVD-r media (8x speed).

Finger Print Reader

Works with fingerprint-gui 1.04, just execute the command below to avoid the error message "ABSOpen() failed -1086 (An internal error occurred (pt: tfmerr -1086).)":

mkdir -m700 /var/upek_data/

Hot Key Buttons

All buttons working:

1. E-mail.
2. Internet (Web Browser).
3. Toggle Silient Mode (for power saving). When enabled, Silent Mode will reduce fan noise and save power consumption. Note this may reduce computer performance.

I have configured udev to make show a KDE notification when the wifi, bluetooth or webcam is activated / desactivated. You can do the same for the silent hot key button using khotkey section in KDE's systemsettings, no need to go for udev.

Modem

Does not work yet: Re: AW: agere softmodem HDA + Intel Ibex Peak High Definition Audio

Sata and e-Sata

Intel Ibex Peak 4 port SATA AHCI controller, driver ahci.ko. To compile the driver: make menu config -> Device Drivers ->  Serial ATA and Parallel ATA drivers ->  AHCI SATA support

This is what I do before switching my e-Sata HD off:

1. umount all partitions.
2. echo 1 > /sys/block/sdb/device/delete

When issuing the last command the following lines appears in dmesg:

[40049.014691] sd 4:0:0:0: [sdb] Synchronizing SCSI cache
[40049.014851] sd 4:0:0:0: [sdb] Stopping disk
[40049.341191] ata5.00: disabled



You configure hald and udisks to thread "e-sata" disks as removable. Udisks is the default disk manager in KDE SC 4.6.x, so it is preferable.

Sound Card

Intel Ibex Peak HDA sound card, driver: snd-hda-intel.ko. Playback and internal microphone tested with Skype, all working. I moved my old /etc/asound.conf to /etc/asound.ferrari, it is not needed anymore because this sound card multiplex audio streamings without dmix alsa plugin. I do not have any s/pdif receiver so I cannot test it.

Just a side note: when the power save mode is enabled (echo 1 > /sys/module/snd_hda_intel/parameters/power_save) every time you change the volumes with a mixer or some program starts to play something there is a short crack sound in the speakers. It sounds even when all volume channels are mute or using earphone. That is annoying, so I prefer to disable the power save mode (echo 0 > /sys/module/snd_hda_intel/parameters/power_save).

USB

Intel Ibex Peak EHCI controller, drivers: ehci-hcd.ko and uhci-hcd.ko.

Video

ATI HD4570 video card, driver: xf86-video-ati-6.13.0, firmware: radeon-ucode-20100408. Working with KMS+DRI2, Googleearth and Kwin's composite effects work together too!!!. To use the firmware enable make menuconfig -> Device Drivers -> Generic Driver Options -> Include in-kernel firmware blobs in kernel binary. Then type "radeon/R700_rlc.bin" in "External firmware blobs to build into the kernel binary", copy /lib/firmware/radeon/R700_rlc.bin to /usr/src/linux/firmware/radeon/ and compile the kernel.

Relevant ebuilds (packages) I am using:

• kernel 2.6.33.4
• xf86-video-ati-6.13.0
• radeon-ucode-20100408
• xorg-server-1.8.1-r1
• mesa-7.8.2
• libdrm-2.4.19

Proprietary Driver (fglrx.ko)

Using ati-drivers ebuild brings one big advantage: the GPU runs cooler and the fans run quieter (as quiet as in Windows 7). I think it also helps with battery time, but I have not measure it. That only works as long as Xorg is running, if you to go the console (Ctrl+Alt+F1) the GPU fans start to run at full speed. Going back to Xorg makes them run quiter again. The 3D performance is noticeably better with ati-drivers.

One disadvantage in using ati-drivers is that you have to disable KMS (kernel modsetting) and the radeon kernel module. Otherwise the LCD gets blank when Xorg starts and you will have to restart the notebook to restore the video card to the correct state. Using composite (3D effects) while playing movies usually makes the movie flicker. To solve that problem DRI2 and KMS were created but since we cannot use KMS with ati-drivers I thought I would have to disable the 3D effects when using mplayer. Fortunately there is no flicker :-). In my old notebook (Ferrari 4005, ATI x700 card) the flickering is very annoying. I still own the Ferrari 4005, anyone here wants to buy a Ferrari notebook? :-) The battery still holds 72% of the original manufacturer capacity (around 3400 mAh) and everything works in Linux, accept the svideo port and the card reader, which does not read SD cards anymore, but reads Memory Stick (weird, hugh?) It comes with a bluetooth mouse too.

Unfortunately several important aticonfig commands do not work with this card:

evolucao ~ # aticonfig --adapter=0 --od-getclocks --od-gettemperature Adapter 0 - ATI Mobility Radeon HD 4500 Series Core (MHz) Memory (MHz) Current Clocks : 680 500 Current Peak : 680 500 Configurable Peak Range : [680-680] [500-500] GPU load : 0% ERROR - Get temperature failed for Adapter 0 - ATI Mobility Radeon HD 4500 Series

GPU load always stays at 0% and the thermal sensors are not compatible with aticonfig. Everest in Windows 7 is able to get the GPU temperature. As you can also notice the GPU and Memory clocks cannot be changed.

VGA Port

Works as expected. KDE has this beautiful and effecient kcm module to configure multiple monitors and it works great with the ATI HD4570:

HDMI Port

Work as expected (both video and audio). You need to configure /etc/asound.conf to route audio to the sound card that the video card (ati 4570) registers or use pulseaudio.

Webcam

BisonCam NB Pro webcam, driver: uvcvideo.ko. To compile the driver: make menu config -> Device Drivers -> Multimedia support -> Video capture adapters -> V4L USB devices -> USB Video Class (UVC)

Wired Ethernet

JMicron JMC250 gigabit ethernet PCI-e, driver jme.ko. To compile the driver: make menu config -> Device Drivers -> Network device support ->  Ethernet (1000 Mbit) ->  JMicron(R) PCI-Express Gigabit Ethernet support

Wireless

Intel WiFi Link 6000, driver: iwlagn.ko, firmware: iwl6000-ucode-9.193.4.1.

Gentoo Specific

In /etc/make.conf:

• CHOST="x86_64-pc-linux-gnu"
• CFLAGS="-O2 -march=core2 -pipe -fomit-frame-pointer -msse3 -msse4 -mcx16 -mpopcnt -msahf"
• LDFLAGS="${LDFLAGS} -Wl,-O1"
• VIDEO_CARDS="vesa radeon"
• ALSA_CARDS="snd-hda-intel usb-audio"

Symbolic link: /etc/make.profile -> ../usr/portage/profiles/default/linux/amd64/10.0/desktop/kde . It is ironic the fact that Gentoo use amd64 profile for an Intel processor.

File /etc/modules.autoload.d/kernel-2.6

microcode # for updating CPU's microcode acpi-cpufreq # for CPU frequency scaling cdc-acm # for my 3G modem fglrx # for the video card vboxdrv # for VirtualBox vboxnetflt # idem vboxnetadp # idem

All other drivers are automaticaly loaded by udev during boot initialization. Except for fglrx.ko and the three VirtualBox drivers (vboxdrv.ko, vboxnetflt.ko, vboxnetadp.ko) all other drivers are included in the vanilla Linux kernel. fglrx.ko is compiled by the x11-drivers/ati-drivers ebuild and the three VirtualBox drivers are compiled by the app-emulation/virtualbox-modules ebuild

Things yet to Test

• Express card slot
• Multi-gesture touchpad

lspci output:

00:00.0 Host bridge: Intel Corporation Arrandale DRAM Controller (rev 12)

00:01.0 PCI bridge: Intel Corporation Arrandale PCI Express x16 Root Port (rev 12)

00:16.0 Communication controller: Intel Corporation Ibex Peak HECI Controller (rev 06)

00:1a.0 USB Controller: Intel Corporation Ibex Peak USB2 Enhanced Host Controller (rev 06)

00:1b.0 Audio device: Intel Corporation Ibex Peak High Definition Audio (rev 06)

00:1c.0 PCI bridge: Intel Corporation Ibex Peak PCI Express Root Port 1 (rev 06)

00:1c.1 PCI bridge: Intel Corporation Ibex Peak PCI Express Root Port 2 (rev 06)

00:1c.2 PCI bridge: Intel Corporation Ibex Peak PCI Express Root Port 3 (rev 06)

00:1c.3 PCI bridge: Intel Corporation Ibex Peak PCI Express Root Port 4 (rev 06)

00:1d.0 USB Controller: Intel Corporation Ibex Peak USB2 Enhanced Host Controller (rev 06)

00:1e.0 PCI bridge: Intel Corporation 82801 Mobile PCI Bridge (rev a6)

00:1f.0 ISA bridge: Intel Corporation Ibex Peak LPC Interface Controller (rev 06)

00:1f.2 SATA controller: Intel Corporation Ibex Peak 4 port SATA AHCI Controller (rev 06)

00:1f.3 SMBus: Intel Corporation Ibex Peak SMBus Controller (rev 06)

02:00.0 VGA compatible controller: ATI Technologies Inc M92 [Mobility Radeon HD 4500 Series]

02:00.1 Audio device: ATI Technologies Inc R700 Audio Device [Radeon HD 4000 Series]

06:00.0 Network controller: Intel Corporation WiFi Link 6000 Series (rev 35)

07:00.0 System peripheral: JMicron Technology Corp. SD/MMC Host Controller (rev 80)

07:00.2 SD Host controller: JMicron Technology Corp. Standard SD Host Controller (rev 80)

07:00.3 System peripheral: JMicron Technology Corp. MS Host Controller (rev 80)

07:00.5 Ethernet controller: JMicron Technology Corp. JMC250 PCI Express Gigabit Ethernet (rev 03)

ff:00.0 Host bridge: Intel Corporation QuickPath Architecture Generic Non-core Registers (rev 02)

ff:00.1 Host bridge: Intel Corporation QuickPath Architecture System Address Decoder (rev 02)

ff:02.0 Host bridge: Intel Corporation QPI Link 0 (rev 02)

ff:02.1 Host bridge: Intel Corporation QPI Physical 0 (rev 02)

ff:02.2 Host bridge: Intel Corporation Device 2d12 (rev 02)

ff:02.3 Host bridge: Intel Corporation Device 2d13 (rev 02)

lsusb output:

Bus 002 Device 006: ID 5986:0343 Acer, Inc -> webcam

Bus 002 Device 003: ID 147e:1000 Upek -> fingerprint reader

Bus 002 Device 002: ID 8087:0020 

Bus 002 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub

Bus 001 Device 005: ID 0a12:0001 Cambridge Silicon Radio, Ltd Bluetooth Dongle (HCI mode)

Bus 001 Device 003: ID 03eb:0902 Atmel Corp. 4-Port Hub

Bus 001 Device 002: ID 8087:0020 

Bus 001 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub

Full support to Modem Manager 0.4 in Solid

I have finally finished to implement Modem Manager 0.4 specification in solid. I still need to add more comment in the source code and implement more caching to prevent stress in the DBus bus. Now I need help to test the implementation because my Sony MD300 modem supports less than 1/3 of the specification. Nothing from interfaces Location, Cdma, Gsm.Contacts, Gsm.SMS, Gsm.Hso and Gsm.Ussd work with my modem, but I have implemented them anyway. I am going to commit the patch after trunk is opened for new implementations, which AFAIK is after the first KDE SC 4.5.0 rc release next week.

For those who likes statistics, the patch adds 128 new files, changes other 23 files in trunk, is 400 KB in size and took me eight weeks to reach this stage.

Plasma Network Management and ModemManager

Well, new update, this time I have implemented org.freedesktop.ModemManager.Modem and org.freedesktop.ModemManager.Modem.Gsm.Card interfaces from specification and added some information it provides to the interface details tab:

There is more information to add to interface details tab, I have not done it yet because of lack of time. The ModemManager patch is getting bigger (it touches or adds 86 files to trunk) and the Plasma NM patch adds/changes other 28 files. Managing all those files outside svn makes me loose time, but since trunk is closed for new features I will have to wait until KDE SC 4.5.0 is released.

Of course I will have to change interface details window, it becomes too big when all information is displayed. Maybe I will add a new "Hardware details" tab and move some things there. When I have more time I am going to implement selecting which information to display too. The MM implementation is taking more time than I have thought because MM specification does some things differently from NM specification and I had to think about how to solve the problems that changes brought up.

Solid ModemManager backend

Hi there, after two days of hard working now we have a real ModemManager backend for solid. The first patches I created implement MM support too tied to GsmNetworkInterface class, which means you had to instantiate a NetworkManager class to use ModemManager, that is not necessary anymore. The bad news is that the translate team has a couple of .desktop files to translate besides the source code.

This night was long, I spent several hours trying to fix a difficult to trace bug. I inspected the program and it should work but somehow the kded module crashed when instantiating a ModemGsmNetworkInterface class, in fact it instantiated the class but when it tried to use one of its methods the program crashed. The problem was that ModemGsmNetworkInterface inherited from ModemInterface and declared a d_ptr pointer, which it should not, only ModemInterface class should declare d_ptr. That problem fixed I found an infinite loop in the program hehe, that was easy to fix. After that I had to fight with the signal and slots strings to make them match. If you see a lot of "No such slot" (or signal) messages in ~/.xsession-errors you know what I am talking about. This time I saw other messages too, like "incompatible signal and slot types" (because of full nested namespace names in some places but not in other) and "impossible to connect" (because the slot's object was null). After some try and error rounds I fixed that too and finally my Plasma NM is working again. Only one bug remains: when removing the modem from USB port the kded module crashes because the ModemGsmNetworkinterface object becomes invalid and the program does not detect that. I will fix that next weekend, now I need rest and I also have other work to do during the week. Keeping this pace I think I can finish the patch by the end of the month.

Plasma NM: more updates

My first MM implementation to solid added the MM dbus interface to GsmNetworkInterface class, MM can run independently from NM, then it is more logical to implement solid MM as a separated backend, that is what I started to do last Saturday. Well, the work is much more complex than I have thought. I spent the whole Saturday reorganizing the code, creating new files, and eliminating some namespaces. The solid interface is ready (I think), now I need to reorganize the NM backend and make everything work again. If the first implementation was an old car I may say I am pimping my code hehe, I completely dismantled the code, now I am reassembling everything together with a nice look. Probably tomorrow I will try to finish the backend.

Last Friday I did some experiments with using html-like tables to implement interface details using a single Plasma::Label, that simplifies the code and, most important, reduces memory usage. I almost succeed, the result is this:

Another advantage of using html tables is that now we can select which detail to show per interface type. In the current implementation the entries Operator, Signal Quality and Access Technology are only shown for Gsm connections. Monolithic Knm uses a similar code to showing tooltips for the interfaces. One other thing Knm does is let the user select which detail and in which order to present them, that is more flexible than what I have implemented so far. I plan to port that code to Plasma NM in the future as more I progress in MM implementation the more information will be added to interface details, so it would be good to select which detail I want to see. One more thing to the TODO list.

Sometimes I get very stubborn to finish one thing. If you look at the screenshot above you can see the access technology and signal quality beside the connection name at the right side of Plasma NM. I thought I could do that in a couple of hours at most, what naive of me. Yesterday morning I started to dig the code finding how Plasma NM does that for wireless connections. Man, it looks like a labirint, only today morning I finally have got everything working, well almost everything, there are some edges to polish. What took me so much time was the fact the wireless component gets the signal strength througth org.kded.networkmanagement (a kded module), so I had to implemented one interface for org.kded.networkmanagement too, more precisely org.kded.networkmanagement.GsmInterfaceConnection. The implementation adds 13 new files to the source code :-/ and charnges several other files. Well, at least now I can see the signal quality and access technology without having to open the interface details window :-) Yes!!!