[Xenial/snapdragon] [pull req] Ubuntu-4.4.0-1011.11

[Tim Gardner]

On 04/06/2016 04:13 PM, Paolo Pisati wrote:
> This pull contains a fix for a regression that was introduced in 4.4.0-1010.10,
> config changes to accomodate zfs compilation, and a pick up of the latest
> stability fixes from the qcom-lt 4.4 tree, including a fix for cpufreq-dt.
> 
> Patch 0001 "UBUNTU: [Config] PREEMPT_VOLUNTARY=y" align the config with generic
> arm64 and avoid hitting a regression that was introduced in generic master via
> this commit (and that we picked with the 4.4.0-1010.10):
> 
> 444209c printk: set may_schedule for some of console_trylock() callers
> 
> If this commit is present, and the kernel is compiled with the PREEMPT_COUNT
> option (as the snapdragon kernel was before this change), upon boot we are
> presented with lots of:
> 
> ...
> [   12.261029] BUG: using smp_processor_id() in preemptible [00000000] code: systemd/1
> [   12.267522] caller is debug_smp_processor_id+0x1c/0x28
> [   12.275059] CPU: 1 PID: 1 Comm: systemd Not tainted 4.4.0-1010-snapdragon #10
> [   12.280241] Hardware name: Qualcomm Technologies, Inc. APQ 8016 SBC (DT)
> [   12.287792] Call trace:
> [   12.294564] [<ffffffc00008ada0>] dump_backtrace+0x0/0x1b0
> [   12.296734] [<ffffffc00008af74>] show_stack+0x24/0x30
> [   12.302290] [<ffffffc0005763f8>] dump_stack+0xb0/0xe8
> [   12.307325] [<ffffffc0005967d4>] check_preemption_disabled+0x114/0x118
> [   12.312360] [<ffffffc0005967f4>] debug_smp_processor_id+0x1c/0x28
> [   12.318784] [<ffffffc00013b168>] vprintk_emit+0x338/0x6d8
> [   12.324948] [<ffffffc0002028c4>] printk_emit+0x98/0xc0
> [   12.330326] [<ffffffc00013b650>] devkmsg_write+0xf0/0x140
> [   12.335364] [<ffffffc00029a784>] do_iter_readv_writev+0x7c/0xc8
> [   12.340833] [<ffffffc00029b00c>] do_readv_writev+0x19c/0x278
> [   12.346560] [<ffffffc00029b1a4>] vfs_writev+0x4c/0x70
> [   12.352464] [<ffffffc00029bfcc>] SyS_writev+0x6c/0x110
> ...
> 
> and an wrecked system.
> 
> Arm64 generic was not affected because it didn't have that option turned on, but
> the regression is there too - we probably missed some patches during the cherry
> pick of the above patch - see also http://bugs.launchpad.net/bugs/1534216.
> 
> Patch 0002-005 "UBUNTU: [Config] disable ..." disable debugging options that
> were only useful during development and that prevented zfs to compile against
> this kernel.
> 
> Patch 0006-0010 are the latest stable fixes from the qcom-lt 4.4 tree, in
> particular the last two commits ("regulator: qcom_spmi: ...") fix an instability
> (that could be triggered with stress-ng) in cpufreq-dt - see also
> https://bugs.96boards.org/show_bug.cgi?id=305.
> 
> Patch 0011 "UBUNTU: [Config] INLINE_*_UNLOCK=y" is just an updateconfigs after
> all the config changes.
> 
> Tested on my dragonboard, went through continuous stress test and multiple
> stress-ng runs.
> 
> The following changes since commit 26bf76b56226026d6100180a5118ec4a207806af:
> 
>   UBUNTU: Ubuntu-4.4.0-1010.10 (2016-03-30 09:05:21 -0600)
> 
> are available in the git repository at:
> 
>   git://git.launchpad.net/~p-pisati/ubuntu/+source/linux 
> 
> for you to fetch changes up to 4781c1de851678021efce07bd91530defce5380d:
> 
>   UBUNTU: Ubuntu-4.4.0-1011.11 (2016-04-06 16:33:52 +0200)
> 
> ----------------------------------------------------------------
> Nicolas Dechesne (2):
>       arm64: clean up defconfig with savedefconfig
>       configs: add SPI_SPIDEV as module in distro.config
> 
> Paolo Pisati (8):
>       UBUNTU: [Config] PREEMPT_VOLUNTARY=y
>       UBUNTU: [Config] disable PROVE_LOCKING
>       UBUNTU: [Config] disable DEBUG_LOCK_ALLOC
>       UBUNTU: [Config] disable DEBUG_SPINLOCK
>       UBUNTU: [Config] disable DEBUG_MUTEXES
>       UBUNTU: [Config] INLINE_*_UNLOCK=y
>       UBUNTU: Start new release
>       UBUNTU: Ubuntu-4.4.0-1011.11
> 
> Srinivas Kandagatla (1):
>       arm64: dts: apq8016-sbc: Add aliases to spi device.
> 
> Stephen Boyd (2):
>       regulator: qcom_spmi: Add slewing delays for all SMPS types
>       regulator: qcom_spmi: Only use selector based regulator ops
> 
>  arch/arm64/boot/dts/qcom/apq8016-sbc.dtsi          |   2 +
>  arch/arm64/configs/defconfig                       |  36 +---
>  debian.snapdragon/abi/4.4.0-1009.9/abiname         |   1 -
>  debian.snapdragon/abi/4.4.0-1010.10/abiname        |   1 +
>  .../{4.4.0-1009.9 => 4.4.0-1010.10}/arm64/ignore   |   0
>  .../arm64/ignore.modules                           |   0
>  debian.snapdragon/changelog                        |  19 ++
>  debian.snapdragon/config/config.common.ubuntu      |  30 ++-
>  drivers/regulator/qcom_spmi-regulator.c            | 218 +++++++++++++--------
>  kernel/configs/distro.config                       |   1 +
>  10 files changed, 181 insertions(+), 127 deletions(-)
>  delete mode 100644 debian.snapdragon/abi/4.4.0-1009.9/abiname
>  create mode 100644 debian.snapdragon/abi/4.4.0-1010.10/abiname
>  rename debian.snapdragon/abi/{4.4.0-1009.9 => 4.4.0-1010.10}/arm64/ignore (100%)
>  rename debian.snapdragon/abi/{4.4.0-1009.9 => 4.4.0-1010.10}/arm64/ignore.modules (100%)
> 

Why is there a bunch of Wily packaging commits on top of
Ubuntu-4.4.0-17.33 ? Perhaps you could squash them out of existence.

rtg
-- 
Tim Gardner tim.gardner at canonical.com