Kernel 3.17-rc7 – Unexpected..

Kernel 3.17 was due out by now, but we have another release candidate instead - rc7..

No problems with the latest VMware (10.0.3 patched) and NVIDIA (340.32) drivers, which compile and load OK..   The latest NVIDIA ‘beta’ (343.22) compiles/loads OK, but I need to test it more thoroughly with VMware, as the previous version broke the VM display on my main system..

# uname -a
Linux rg6830l 3.17.0-rc7 #1 SMP PREEMPT Mon Sep 29 10:36:46 PDT 2014 x86_64 x86_64 x86_64 GNU/Linux

Details – and an explanation for the delay – are here:  http://lkml.iu.edu/hypermail/linux/kernel/1409.3/03393.html

Robert Gadsdon.  September 29, 2014.

Kernel 3.17-rc6 – 3.17 Final Soon?

Installed Kernel 3.17-rc6, and – as with -rc5 – the latest VMware (10.0.3 with vmnet patch) and NVIDIA (340.32) drivers compile and install/load OK..

# uname -a
Linux rg6830l 3.17.0-rc6 #1 SMP PREEMPT Mon Sep 22 09:48:38 PDT 2014 x86_64 x86_64 x86_64 GNU/Linux

Details are here:  http://lkml.iu.edu/hypermail/linux/kernel/1409.2/03201.html

From the comments, it seems that the next release may be 3.17 Final..

Robert Gadsdon.   September 23, 2014.

Kernel 3.17-rc5 – Same Comments as -rc4..

Just updated the test system to Kernel 3.17-rc5, and – as with -rc4 – the vmnet-patched VMware 10.0.3 and NVIDIA 340.32 both compile/load OK..

$ uname -a
Linux rg6830l 3.17.0-rc5 #1 SMP PREEMPT Sun Sep 14 23:17:13 PDT 2014 x86_64 x86_64 x86_64 GNU/Linux

Details are here:  http://lkml.iu.edu/hypermail/linux/kernel/1409.1/05124.html

Robert Gadsdon.   September 15, 2014.

ARM – Odroid U2 – Finally an Up-to-Date Kernel?

I had been frustrated by the apparent inability of the suppliers to provide anything other than an obsolete Ubuntu-centric 3.8+ kernel for the Odroid U2, and other kernels I had tested were somewhat ‘incomplete’…

I finally managed to get a 3.17-r3 version from https://github.com/tobiasjakobi/linux-odroid/tree/odroid-3.17.y

This now includes the necessary ‘bits’ for the U3, which is functionally identical to the U2…   After some minor config changes – including de-selecting some of the mali options, as some header files were missing/obsolete – I managed to get it to boot on my Odroid U2:

Fedora release 20 (Heisenbug)
Kernel 3.17.0-rc3 on an armv7l (ttySAC1)
..........
# uname -a
Linux rgodroid 3.17.0-rc3 #4 SMP PREEMPT Thu Sep 11 02:42:18 BST 2014 armv7l armv7l armv7l GNU/Linux

Unfortunately, further testing was impossible, after a sudden – and catastrophic – USB stick failure, which trashed the root filesystem..

More testing will follow, when that is replaced!

Robert Gadsdon.   September 12, 2014.

Kernel 3.17-rc4 – OK with (Patched) VMware, and latest NVIDIA..

Updated to Kernel 3.17-rc4, and the vmnet-patched VMware (10.0.3) and latest NVIDIA (340.32) all compile/load OK..    See previous -rc3 article for details of the patch..

$ uname -a 
Linux rg6830l 3.17.0-rc4 #1 SMP PREEMPT Sun Sep 7 22:33:23 PDT 2014 x86_64 x86_64 x86_64 GNU/Linux

Details of the changes are here:  http://lkml.iu.edu/hypermail/linux/kernel/1409.0/04463.html

Robert Gadsdon.  September 8, 2014.

Kernel 3.17-rc3 – Breaks VMNET – and a Fix..

Just tested Kernel 3.17-rc3, and it breaks VMware (10.0.3) – vmnet:

................
/tmp/modconfig-d0zqLW/vmnet-only/netif.c: In function ‘VNetNetIf_Create’:
/tmp/modconfig-d0zqLW/vmnet-only/netif.c:152:64: error: macro "alloc_netdev" requires 4 arguments, but only 3 given
    dev = alloc_netdev(sizeof *netIf, deviceName, VNetNetIfSetup);
                                                                ^
/tmp/modconfig-d0zqLW/vmnet-only/netif.c:152:10: error: ‘alloc_netdev’ undeclared (first use in this function)
    dev = alloc_netdev(sizeof *netIf, deviceName, VNetNetIfSetup);
...............

Thanks to Erwin Zoer (see his comment below the earlier post for 3.17-rc1..), there is a fix available – for ~/vmware-only/netif.c

For kernel 3.17-rc3 and later, change line 152 from

dev = alloc_netdev(sizeof *netIf, deviceName, VNetNetIfSetup);
to 
dev = alloc_netdev(sizeof *netIf, deviceName, NET_NAME_UNKNOWN, VNetNetIfSetup);

Then re-TAR etc.. and re-run the VMware compile/install:

# vmware-modconfig --console --install-all

I have applied this on my test system, and it works OK..

Robert Gadsdon. September 3, 2014.

ARM – CuBox to 3.17-rc2 – 23rd Anniversary Release..

Just updated the CuBox i4-Pro to Kernel 3.17-rc2, which was – deliberately – released on the 23rd anniversary of the initial release of kernel code for ‘Linux’..

More details here:  http://lkml.iu.edu/hypermail/linux/kernel/1408.3/00773.html

And source here:  https://github.com/jmontleon/fedora-20-cubox-i_hb/tree/3.17-rc2

Fedora release 20 (Heisenbug)
Kernel 3.17.0-rc2 on an armv7l (ttymxc0)
.................
$ uname -a
Linux rgcubox 3.17.0-rc2 #1 SMP PREEMPT Tue Aug 26 01:53:28 EDT 2014 armv7l armv7l armv7l GNU/Linux

Robert Gadsdon.  August 26, 2014.

ARM – CuBox-i4 Pro to Kernel 3.17-rc1..

Just updated the CuBox to kernel 3.17-rc1, but still used the previous ‘combined zImage plus DTB’ method, which works OK..

Kernel source here:  # git clone -b 3.17-rc1 –single-branch https://github.com/jmontleon/fedora-20-cubox-i4pro

Fedora release 20 (Heisenbug)
Kernel 3.17.0-rc1 on an armv7l (ttymxc0)
................
# uname -a
Linux rgcubox 3.17.0-rc1 #1 SMP PREEMPT Tue Aug 19 13:48:28 EDT 2014 armv7l armv7l armv7l GNU/Linux

Compile workflow is:

make oldconfig 
or
make xconfig
make -j5 zImage
make dtbs
cp System.map /boot
cat arch/arm/boot/zImage arch/arm/boot/dts/imx6q-cubox-i.dtb > zImage_dtb
mkimage -A arm -O linux -T kernel -a 0x10008000 -e 0x10008000 -n cubox_linux -C none -d zImage_dtb /boot/uImage317rc1
make -j5 modules
make modules_install
cd /boot
rm -f uEnv.txt
Edit uEnv.txt.317rc1 to reference bootfile=/uImage317rc1, and softlink to uEnv.txt (in /boot)...
uEnv.txt.317rc1:
bootfile=/uImage317rc1
mmcargs=setenv bootargs root=/dev/mmcblk0p3 rootfstype=ext4 rootwait console=ttymxc0,115200n8 console=tty1 video=mxcfb0:dev=hdmi,1920x1080M@60,if=RGB24,bpp=32


Robert Gadsdon..  August 21, 2014.

Skype / Linux – When the Fix Becomes the Problem.. (Updated)

I have been running a ‘fixed’ version of Skype on my Linux systems for some time (Fedora 20, x86_64), which included ‘PULSE_LATENCY_MSEC=30′ in a skype script to run the executable (skype-bin)..

Recently, I have experienced garbled and ‘sped up’ sound, and this was easily demonstrated using the Skype Test Call facility..

The solution to this used to be the one that I was already using – to run skype with PULSE_LATENCY_MSEC=30 – but I now found that this appeared to be actually causing the latest problem, and skype worked just fine without it!

The solution is simple – to just remove the script, and rename skype-bin executable back to to skype..

It would appear that the engineers at Skype may have finally fixed the Linux problem, themselves?

Robert Gadsdon.   August 19, 2014.

Update:   It has been confirmed that – at last – the Skype executable has been fixed, to work correctly with pulseaudio..    And..  although the script caused a problem on both of my systems, it has been reported that on other systems it made no difference, and still worked OK..   (see comments below..)

RG.  Aug 26.

Kernel 3.17-rc1 – OK with VMware and NVIDIA..

Just installed Kernel 3.17-rc1, and the latest VMware (10.0.3) and NVIDIA (340.32) drivers compile and load OK…    I had previously had trouble with VMware and NVIDIA’s ‘beta’ driver 343.13, so did not test this version again..

$ uname -a
Linux rglinux-i7 3.17.0-rc1 #1 SMP PREEMPT Sat Aug 16 13:56:55 PDT 2014 x86_64 x86_64 x86_64 GNU/Linux

Brief details of -rc1 are here:  http://lkml.iu.edu/hypermail/linux/kernel/1408.2/00077.html

Robert Gadsdon.  August 16, 2014.