intel_pstate no_turbo patch from upstream

[Shuhao]

If you're on ubuntu_saucy and enabled intel_pstate like you should via the kernel boot parameter intel_pstate=enabled, you will find that the no_turbo setting[1] sometimes does not work.

This is fixed in 3.12[2] and it should be backported to Saucy as it is a critical feature for those who experience overheating with turbo on.

[1]: echo 1 > /sys/devices/system/cpu/intel_pstate/no_turbo
[2]: http://comments.gmane.org/gmane.linux.power-management.general/39243

[Shuhao (shuhao)]

Also, intel_pstate should be enabled by default. Thermal daemon[1] is also a good addition to keep temperature in check.

[1]: https://github.com/01org/thermal_daemon

[Brad Figg (brad-figg)]

This bug is missing log files that will aid in diagnosing the problem. From a terminal window please run:

apport-collect 1254091

and then change the status of the bug to 'Confirmed'.

If, due to the nature of the issue you have encountered, you are unable to run this command, please add a comment stating that fact and change the bug status to 'Confirmed'.

This change has been made by an automated script, maintained by the Ubuntu Kernel Team.

[Joseph Salisbury (jsalisbury)]

I built a saucy test kernel that has a cherry pick of commit 1ccf7a1 from upstream. This test kernel can be downloaded from:

http://kernel.ubuntu.com/~jsalisbury/lp1254091/

Can you test this kernel and post back if it resolves this bug? If it does, I can submit an SRU request for this commit to be included in the stable kernels.

[Shuhao (shuhao)]

Sure. I'll take a look at it. My problem is that I don't actually know
how to reliably reproduce this bug yet.

Shuhao

On 11/22/2013 12:21 PM, Joseph Salisbury wrote:
> I built a saucy test kernel that has a cherry pick of commit 1ccf7a1
> from upstream. This test kernel can be downloaded from:
>
> http://kernel.ubuntu.com/~jsalisbury/lp1254091/
>
> Can you test this kernel and post back if it resolves this bug? If it
> does, I can submit an SRU request for this commit to be included in the
> stable kernels.
>

[Shuhao (shuhao)]

I'm getting:

     dpkg: dependency problems prevent configuration of
linux-tools-3.11.0-14-generic:
      linux-tools-3.11.0-14-generic depends on linux-tools-3.11.0-14;
however:
       Package linux-tools-3.11.0-14 is not installed.

     dpkg: error processing linux-tools-3.11.0-14-generic (--install):
      dependency problems - leaving unconfigured
     Errors were encountered while processing:
      linux-tools-3.11.0-14-generic

Shuhao

On 11/22/2013 12:21 PM, Joseph Salisbury wrote:
> I built a saucy test kernel that has a cherry pick of commit 1ccf7a1
> from upstream. This test kernel can be downloaded from:
>
> http://kernel.ubuntu.com/~jsalisbury/lp1254091/
>
> Can you test this kernel and post back if it resolves this bug? If it
> does, I can submit an SRU request for this commit to be included in the
> stable kernels.
>

[Joseph Salisbury (jsalisbury)]

You should only have to install the linux-image and linux-image-extra .deb packages.

[Shuhao (shuhao)]

I'm not sure if this is actually working.

However, that might be a bug upstream.

[Shuhao (shuhao)]

[~]$ cat /sys/devices/system/cpu/intel_pstate/no_turbo
1
[~]$ cat /proc/cpuinfo | grep MHz
cpu MHz : 2691.000
cpu MHz : 2691.000
cpu MHz : 2783.000
cpu MHz : 2806.000
[~]$ uname -a
Linux thinkpadt420 3.11.0-14-generic #21 SMP Fri Nov 22 16:40:18 UTC 2013 x86_64 x86_64 x86_64 GNU/Linux

[Shuhao (shuhao)]

Today this seem to be working. Compiled android for about 3 hours and didn't observe the clock freq to go above the cap (2300)

[Joseph Salisbury (jsalisbury)]

Is it now working with my test kernel from comment #3, or with a stock Ubuntu kernel?

[Shuhao (shuhao)]

Your kernel. Again I'm not exactly sure how to reproduce this reliably...
So I'll take another stab at it the next time I compile android.

Shuhao
Sent from my phone.
On Dec 9, 2013 3:50 PM, "Joseph Salisbury" <email address hidden>
wrote:

> Is it now working with my test kernel from comment #3, or with a stock
> Ubuntu kernel?
>
> ** Changed in: linux (Ubuntu)
> Status: Incomplete => In Progress
>
> --
> You received this bug notification because you are subscribed to the bug
> report.
> https://bugs.launchpad.net/bugs/1254091
>
> Title:
> intel_pstate no_turbo patch from upstream
>
> Status in “linux” package in Ubuntu:
> In Progress
>
> Bug description:
> If you're on ubuntu_saucy and enabled intel_pstate like you should via
> the kernel boot parameter intel_pstate=enabled, you will find that the
> no_turbo setting[1] sometimes does not work.
>
> This is fixed in 3.12[2] and it should be backported to Saucy as it is
> a critical feature for those who experience overheating with turbo on.
>
> [1]: echo 1 > /sys/devices/system/cpu/intel_pstate/no_turbo
> [2]:
> http://comments.gmane.org/gmane.linux.power-management.general/39243
>
> To manage notifications about this bug go to:
> https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1254091/+subscriptions
>

[Shuhao (shuhao)]

I think it is working with your kernel. Just did some android compilation and didn't see any spikes above ~2300 (which is my no turbo cap).

[Brad Figg (brad-figg)]

This bug is awaiting verification that the kernel in -proposed solves the problem. Please test the kernel and update this bug with the results. If the problem is solved, change the tag 'verification-needed-saucy' to 'verification-done-saucy'.

If verification is not done by 5 working days from today, this fix will be dropped from the source code, and this bug will be closed.

See https://wiki.ubuntu.com/Testing/EnableProposed for documentation how to enable and use -proposed. Thank you!

[Shuhao (shuhao)]

I /think/ this is working. I have no observed any spikes with no_turbo set to true after this patch.

[Launchpad Janitor (janitor)]

This bug was fixed in the package linux - 3.11.0-17.31

---------------
linux (3.11.0-17.31) saucy; urgency=low

  [ Brad Figg ]

  * Release Tracking Bug
    - LP: #1275899
  * Revert "UBUNTU: SAUCE: Fix compat_sys_recvmsg on x32 archs"

  [ John Johansen ]

  * [Upstream] x86, x32: Correct invalid use of user timespec in the kernel
    - LP: #1274754

linux (3.11.0-17.30) saucy; urgency=low

  [ Brad Figg ]

  * Release Tracking Bug
    - LP: #1270292

  [ Brad Figg ]

  * Start new release

  [ Upstream Kernel Changes ]

  * Revert "ACPI: Add BayTrail SoC GPIO and LPSS ACPI IDs"

linux (3.11.0-17.29) saucy; urgency=low

  [ Brad Figg ]

  * Start new release

  [ Upstream Kernel Changes ]

  * GFS2: Fix ref count bug relating to atomic_open
    - LP: #1269863
  * aio: restore locking of ioctx list on removal
    - LP: #1269863
  * mmc: block: fix a bug of error handling in MMC driver
    - LP: #1269863
  * net/mlx4_en: Fixed crash when port type is changed
    - LP: #1269863
  * net: Fix "ip rule delete table 256"
    - LP: #1269863
  * ipv6: use rt6_get_dflt_router to get default router in rt6_route_rcv
    - LP: #1269863
  * ipv6: protect for_each_sk_fl_rcu in mem_check with rcu_read_lock_bh
    - LP: #1269863

linux (3.11.0-17.28) saucy; urgency=low

  [ Brad Figg ]

  * Release Tracking Bug
    - LP: #1269875

  [ Brad Figg ]

  * Start new release

  [ Upstream Kernel Changes ]

  * igb: Update link modes display in ethtool
  * Revert "mac80211: allow disable power save in mesh"
    - LP: #1269863
  * Revert "of/address: Handle #address-cells > 2 specially"
    - LP: #1269863
  * net: clamp ->msg_namelen instead of returning an error
    - LP: #1269053
  * random32: fix off-by-one in seeding requirement
    - LP: #1269863
  * bonding: don't permit to use ARP monitoring in 802.3ad mode
    - LP: #1269863
  * usbnet: fix status interrupt urb handling
    - LP: #1269863
  * 6lowpan: Uncompression of traffic class field was incorrect
    - LP: #1269863
  * tuntap: limit head length of skb allocated
    - LP: #1269863
  * macvtap: limit head length of skb allocated
    - LP: #1269863
  * tcp: tsq: restore minimal amount of queueing
    - LP: #1269863
  * bonding: fix two race conditions in bond_store_updelay/downdelay
    - LP: #1269863
  * net-tcp: fix panic in tcp_fastopen_cache_set()
    - LP: #1269863
  * isdnloop: use strlcpy() instead of strcpy()
    - LP: #1269863
  * connector: improved unaligned access error fix
    - LP: #1269863
  * ipv4: fix possible seqlock deadlock
    - LP: #1269863
  * inet: prevent leakage of uninitialized memory to user in recv syscalls
    - LP: #1269863
  * net: rework recvmsg handler msg_name and msg_namelen logic
    - LP: #1269863
  * net: add BUG_ON if kernel advertises msg_namelen > sizeof(struct
    sockaddr_storage)
    - LP: #1269863
  * inet: fix addr_len/msg->msg_namelen assignment in recv_error and rxpmtu
    functions
    - LP: #1269863
  * ipv6: fix leaking uninitialized port number of offender sockaddr
    - LP: #1269863
  * ip6_output: fragment outgoing reassembled skb properly
    - LP: #1269863
  * netfilter: push reasm skb through instead of original frag skbs
    - LP: #1269863
  * xf...