Kernel 3.18 – Another Lockup – Back to 3.16.7!

Yet another ‘lockup’ – this time with 3.18..    So, as the problem also exists with 3.17.x, I have taken the unusual step of reverting to Kernel 3.16.7, for the time being!

Some of the latest correspondence on the issue, here:

Robert Gadsdon.   December 12, 2014.

ARM – CuBox-i4 Pro to 3.18, and full Device Tree..

Updated the CuBox-i4 Pro to Kernel 3.18, and now use the complete ‘device tree’ zImage and dtb boot method:

The source can be found here:

– and the u-boot boot commands are:

setenv bootargs "root=/dev/mmcblk0p3 rootfstype=ext4 rootwait console=ttymxc0,115200n8 console=tty1 video=mxcfb0:dev=hdmi,1920x1080M@60,if=RGB24,bpp=32"
setenv bootcmd "load mmc 0:1 0x10800000 zImage; load mmc 0:1 0x18000000 imx6q-cubox-i.dtb; bootz 0x10800000 - 0x18000000"
Fedora release 21 (Twenty One)
Kernel 3.18.0 on an armv7l (ttymxc0)
# uname -a
Linux rgcubox 3.18.0 #1 SMP PREEMPT Wed Dec 10 20:32:22 EST 2014 armv7l armv7l armv7l GNU/Linux

Robert Gadsdon.   December 11, 2014.


Kernel 3.18 – Small Changes from -rc7..

Installed kernel 3.18 Final, and as expected the latest NVIDIA (346.16) and VMware (11.0) drivers compile and load/run OK.     The older NVIDIA (340.58) and VMware (10.0.4 + vmnet patch) drivers should also be OK..

$ uname -a
Linux rglinux-i7 3.18.0 #4 SMP PREEMPT Sun Dec 7 18:54:19 PST 2014 x86_64 x86_64 x86_64 GNU/Linux

The obscure ‘lockup’ bug is still unfixed, but it appears this only affected a very small number of users, so was not determined to be a reason to hold things up..

More details of the changes from -rc7 here:

Robert Gadsdon.   December 7, 2014.


ARM – Odroid U3 and Kernel 3.18-rc7 – Still Unstable..

I have managed to get Kernel 3.18-rc7 to boot on my Odroid U3, but it is still very unstable..

Fedora release 21 (Twenty One)
Kernel 3.18.0-rc7 on an armv7l (ttySAC1)
rgodroid login:
$ uname -a
Linux rgodroid 3.18.0-rc7 #3 SMP PREEMPT Wed Dec 3 18:41:29 EST 2014 armv7l armv7l armv7l GNU/Linux

Sometimes the device will boot without detecting the USB-connected Ethernet port (smsc95xx), and at other times it will be detected – and nothing has been changed between boot attempts..

The system still ‘tombstones’ at irregular intervals:

[ 69.610957] BUG: sleeping function called from invalid context at kernel/locking/mutex.c:97
[ 69.613715] in_atomic(): 1, irqs_disabled(): 128, pid: 79, name: mmcqd/0
[ 69.620379] Preemption disabled at:[< (null)>] (null)

[ 69.627151] CPU: 0 PID: 79 Comm: mmcqd/0 Tainted: G W 3.18.0-rc7 #3
[ 69.634463] [<c001991c>] (unwind_backtrace) from [<c00140d8>] (show_stack+0x20/0x24)
[ 69.642176] [<c00140d8>] (show_stack) from [<c06f7ec8>] (dump_stack+0x90/0xd0)
[ 69.649387] [<c06f7ec8>] (dump_stack) from [<c004f4e0>] (__might_sleep+0x16c/0x174)
[ 69.657019] [<c004f4e0>] (__might_sleep) from [<c06fc2b8>] (mutex_lock+0x2c/0x4c)
[ 69.664479] [<c06fc2b8>] (mutex_lock) from [<c055f1a8>] (clk_prepare_lock+0x7c/0x104)
[ 69.672291] [<c055f1a8>] (clk_prepare_lock) from [<c05611cc>] (clk_set_rate+0x24/0xdc)
[ 69.680189] [<c05611cc>] (clk_set_rate) from [<c05409c8>] (sdhci_cmu_set_clock+0x78/0x150)
[ 69.688446] [<c05409c8>] (sdhci_cmu_set_clock) from [<c053c3f4>] (sdhci_do_set_ios+0x148/0x63c)
[ 69.697116] [<c053c3f4>] (sdhci_do_set_ios) from [<c053c1fc>] (sdhci_set_ios+0x38/0x48)
[ 69.705107] [<c053c1fc>] (sdhci_set_ios) from [<c05288a0>] (__mmc_set_clock+0xd4/0x118)
[ 69.713093] [<c05288a0>] (__mmc_set_clock) from [<c0528950>] (mmc_ungate_clock+0x34/0x38)
[ 69.721251] [<c0528950>] (mmc_ungate_clock) from [<c052b9dc>] (mmc_host_clk_hold+0x5c/0xc0)
[ 69.729582] [<c052b9dc>] (mmc_host_clk_hold) from [<c0526af8>] (mmc_start_request+0x218/0x250)
[ 69.738174] [<c0526af8>] (mmc_start_request) from [<c052822c>] (mmc_start_req+0x2fc/0x388)
[ 69.746420] [<c052822c>] (mmc_start_req) from [<c05383d4>] (mmc_blk_issue_rw_rq+0x3f4/0xa6c)
[ 69.754840] [<c05383d4>] (mmc_blk_issue_rw_rq) from [<c0538e78>] (mmc_blk_issue_rq+0x42c/0x50c)
[ 69.763519] [<c0538e78>] (mmc_blk_issue_rq) from [<c0539f74>] (mmc_queue_thread+0xbc/0x168)
[ 69.771863] [<c0539f74>] (mmc_queue_thread) from [<c0048168>] (kthread+0xec/0xf4)
[ 69.779327] [<c0048168>] (kthread) from [<c000fa58>] (ret_from_fork+0x14/0x20)

And…  The most annoying thing is that the fan is running constantly!    I have pwm-fan configured in, but it appears to have no effect..

I will keep testing, and hopefully 3.19-rc will bring some improvements..    Patches for this device seem to be around, now, but appear to be slow in acceptance into the actual kernel codebase..

Robert Gadsdon.   December 4, 2014.

VMware 11 – OK with Kernel 3.17 and 3.18..

Just upgraded to VMware Workstation 11.0, and – as expected – it compiles and loads/runs OK with Kernel 3.17 (3.17.4) and 3.18 (3.18-rc7) without the need for any patches..

Release notes are here:

I have not – yet – had a chance to test functionality with a Linux guest, especially with regard to shared folders..

Robert Gadsdon.   December 1, 2014..

Kernel 3.18-rc7 – Still No Lockup Fix..

Just installed Kernel 3.17-rc7, and the usual ‘latest’ drivers – NVIDIA 340.58 and 346.16, and VMware 10.0.4 with the vmnet patch – compile and load OK..

$ uname -a
Linux rg6830l 3.18.0-rc7 #1 SMP PREEMPT Mon Dec 1 13:52:29 PST 2014 x86_64 x86_64 x86_64 GNU/Linux

Details of changes etc. are here:

It seems that – despite the ‘lockup’ issue – 3.18 Final may be released fairly soon..

VMware 11 has just been released, and hopefully that will be compatible with – at least – Kernel 3.17, and probably 3.18 as well, as the version 10 patch (vmnet) worked with both kernel versions..    I will be testing that soon..

Robert Gadsdon.   December 1, 2014..

Kernel 3.18 – and 3.17 – Beware of Lockups..

A problem that still plagued the 3.18-rc series of kernel releases now seems to affect 3.17 as well..

I have had two complete system hangs with 3.17.3 and 3.17.4 on my main system in the last two weeks, and thought at first that these were caused by KDE and/or NVIDIA hangs, but then found that I could not ssh into the system remotely to kill the affected processes..     In true IT Crowd mode, I actually had to ‘try turning it off and on again’..

More details here: and and

If I get another lockup, I’ll have to seriously think about reverting to 3.16.7 on my main system, for the time being..

Robert Gadsdon.   November 29, 2014.

Kernel 3.18-rc6 – Still Needs Work..

Just installed Kernel 3.18-rc6, and the latest NVIDIA drivers – 340.58 and 346.16 – compile and load OK…

Although VMware have just released version 10.0.4, this still needs the ‘vmnet patch’ to compile OK, with kernels 3.17 and 3.18..

# uname -a
Linux rg6830l 3.18.0-rc6 #1 SMP PREEMPT Sun Nov 23 23:41:40 PST 2014 x86_64 x86_64 x86_64 GNU/Linux

Details of changes, and outstanding ‘issues’, are here:  and here:

The US ‘Thanksgiving’ holiday could cause some delays, but if all goes well, then 3.18 Final may be out soon..

Robert Gadsdon.  November 24, 2014.

VMware 10.0.4 Released – Still Needs VMNET Patch…

VMware have released VMware Workstation version 10.0.4, but it still doesn’t compile with Kernel 3.17 or later..    In fact, the only ‘fix’ appears to be for ‘memory problems with Windows 8.1 and Server 2012′, according to the release notes:

It would appear that up-to-date compatibility is being kept for the next major release, which is due in December…

So, the vmnet patch still needs to be applied…   See previous article, from early September..

Robert Gadsdon.   November 21, 2014.

Kernel 3.18-r5 – Bigger than -rc4..

Just installed Kernel 3.18-rc5 on the test system, and the latest NVIDIA drivers – 340.58 and 346.16 – and VMware 10.0.3 – with the vmnet patch – compile and load OK..

$ uname -a
Linux rg6830l 3.18.0-rc5 #1 SMP PREEMPT Sun Nov 16 23:23:41 PST 2014 x86_64 x86_64 x86_64 GNU/Linux

Details can be found here:

Robert Gadsdon.   November 17, 2014.