Kernel 4.12RC (Release Candidate) series

[1fallen]

About our failing to boot issue, I found this.
I'm going "git pull"s a couple of times a day, and should notice when it gets there.

Thanks Doug your hard work is very much appreciated!
Kind Regards

[mrfelker]

Kernel 4.12rc3 fails to boot on a physical drive or in a VMware WS 12.5.6 VM. Since rc2 works in both instances, rc3 seems to be a regression. Anybody else had different result?.

[mrfelker]

4.12rc3 did boot in a Virtualbox (testbuild 5.23). On my physical machine, the compiler was complaining about possible missing firm for 4.12rc3 but not rc2. Then it would not boot rc3 but will boot rc2.

My host is openSUSE Leap 42.3.

[Doug S]

For the -rc3 will not boot issue, originally found by zika, the fix is now in the mainline tree:

$ git show c08d517
commit c08d517480ea342cc43acdacc5cf4a795e18151d
Author: Ingo Molnar <mingo@kernel.org>
Date:   Thu Jun 1 15:52:23 2017 +0200

    Revert "x86/PAT: Fix Xorg regression on CPUs that don't support PAT"

    This reverts commit cbed27cdf0e3f7ea3b2259e86b9e34df02be3fe4.

I'm busy with something else at the moment and so can not make a kernel to try it right now. Pretty much tested by many people, including me, already though.

EDIT:

cod/tip/daily/2017-06-02 mainline build
These binary packages represent builds of the mainline or stable Linux kernel tree at the commit below:

  cod/tip/daily/2017-06-02 (3b1e342be265f7df915c68c7c6e450956d8865aa)

that daily commit is after the reversion commit, so the daily should boot fine.

[zika]

For the -rc3 will not boot issue, originally found by zika, the fix is now in the mainline tree:

$ git show c08d517
commit c08d517480ea342cc43acdacc5cf4a795e18151d
Author: Ingo Molnar <mingo@kernel.org>
Date:   Thu Jun 1 15:52:23 2017 +0200

    Revert "x86/PAT: Fix Xorg regression on CPUs that don't support PAT"

    This reverts commit cbed27cdf0e3f7ea3b2259e86b9e34df02be3fe4.

I'm busy with something else at the moment and so can not make a kernel to try it right now. Pretty much tested by many people, including me, already though.

EDIT:

cod/tip/daily/2017-06-02 mainline build
These binary packages represent builds of the mainline or stable Linux kernel tree at the commit below:

  cod/tip/daily/2017-06-02 (3b1e342be265f7df915c68c7c6e450956d8865aa)

that daily commit is after the reversion commit, so the daily should boot fine.

999 works now, writing this from it. Nice. Thank You.

[Doug S]

kernel 4.12-rc4 is available. It boots fine on my test server.

$ uname -a
Linux s15 4.12.0-041200rc4-generic #201706042031 SMP Mon Jun 5 00:32:36 UTC 2017 x86_64 x86_64 x86_64 GNU/Linux

[1fallen]

Dang these low-lat kernels are just getting better and better. (Much bigger also )
Ran it pretty hard yesterday Encoding some DVD's with Handbrake (with 2 pass) and CPU seldom ran higher than 50%. (Kudo's)
With Arch I run at least 80 to 96% and of course the heat goes up>>>but not bad, well under my comfort level.

inxi -C && uname -r
CPU: Quad core Intel Core i5-6400 (-MCP-) cache: 6144 KB 
        clock speeds: max: 3300 MHz 1: 2294 MHz 2: 911 MHz 3: 1033 MHz 4: 1512 MHz
        4.12.0-041200rc4-lowlatency

But still no luck with the bfq issue.
Greetz to zika!

EDIT: BTW has any one had any luck with:

sudo grep -i switcheroo /boot/config-*
[sudo] password for me: 
/boot/config-4.10.0-22-generic:CONFIG_VGA_SWITCHEROO=y
/boot/config-4.12.0-041200rc3-lowlatency:CONFIG_VGA_SWITCHEROO=y
/boot/config-4.12.0-041200rc4-lowlatency:CONFIG_VGA_SWITCHEROO=y

[Rustan]

How to enable BFQ on Kernel v.4.12 series?
CONFIG_SCSI_MQ_DEFAULT=y *or* scsi_mod.use_blk_mq=1

cat /sys/block/sd*/queue/scheduler
[mq-deadline] kyber bfq none
[mq-deadline] kyber bfq none

[Doug S]

How to enable BFQ on Kernel v.4.12 series?
CONFIG_SCSI_MQ_DEFAULT=y *or* scsi_mod.use_blk_mq=1

cat /sys/block/sd*/queue/scheduler
[mq-deadline] kyber bfq none
[mq-deadline] kyber bfq none

If I set CONFIG_SCSI_MQ_DEFAULT=y in my kernel configuration, it doesn't work for me. First, the unchanged Ubuntu kernel:

$ cat /sys/block/sd?/queue/scheduler
noop deadline [cfq]
noop deadline [cfq]

$ uname -a
Linux s15 4.12.0-041200rc5-generic #201706112031 SMP Mon Jun 12 00:32:34 UTC 2017 x86_64 x86_64 x86_64 GNU/Linux

Then alter the kernel configuration (I also disable some debug stuff, halving the compile time and kernel size):

$ scripts/diffconfig .config-4.12.0-041200rc5-generic .config
-DEBUG_INFO_DWARF4 y
-DEBUG_INFO_REDUCED n
-DEBUG_INFO_SPLIT n
-GDB_SCRIPTS y
 DEBUG_INFO y -> n
 SCSI_MQ_DEFAULT n -> y

And get:

~$ uname -a
Linux s15 4.12.0-rc5-stockq #262 SMP Thu Jun 15 14:53:52 PDT 2017 x86_64 x86_64 x86_64 GNU/Linux

doug@s15:~$ cat /sys/block/sd?/queue/scheduler
[none]
[none]

[Rustan]

~$ uname -a
Linux home 4.12.0-2-generic #3-Ubuntu SMP Mon Jun 12 12:40:30 UTC 2017 x86_64 x86_64 x86_64 GNU/Linux
~$ cat /sys/block/sd*/queue/scheduler
[none] 
[none] 
~$ sudo modprobe bfq
~$ cat /sys/block/sd*/queue/scheduler
[none] bfq 
[none] bfq 
~$ echo bfq > /sys/block/sda/queue/scheduler
~$ cat /sys/block/sd*/queue/scheduler
[bfq] none
[none] bfq