[STABLE candidates] backport candidates from Allwinner kernels

[STABLE candidates] backport candidates from Allwinner kernels

[Arnd Bergmann]

These are from https://github.com/BPI-SINOVOIP/BPI-M2P-bsp-4.4

They contain a bunch of things from an older Android kernel that looks
like it would still
make sense for normal stable kernels as well, very little here is
allwinner specific:

0905f04eb21f sched/fair: Fix new task's load avg removed from source
CPU in wake_up_new_task()
382c55f88ffe mmc: block: Allow more than 8 partitions per card
b9b7aebb42d1 arm64: fix COMPAT_SHMLBA definition for large pages
b523e185bba3 efi: stub: define DISABLE_BRANCH_PROFILING for all architectures
6c044fecdf78 ARM: 8458/1: bL_switcher: add GIC dependency
ad84f56bf6d6 ARM: 8494/1: mm: Enable PXN when running non-LPAE kernel
on LPAE processor
2010b93e9317 net: tcp: deal with listen sockets properly in tcp_abort.
699f68556943 android: unconditionally remove callbacks in sync_fence_free()
0eb77e988032 vmstat: make vmstat_updater deferrable again and shut down on idle
8d43b49e7e00 hid-sensor-hub.c: fix wrong do_div() usage
75feee3d9d51 arm64: hide __efistub_ aliases from kallsyms
728dabd6d175 Eliminate the .eh_frame sections from the aarch64 vmlinux
and kernel modules
45c815f06b80 perf: Synchronously free aux pages in case of allocation failure
7c1306723ee9 net: diag: support v4mapped sockets in inet_diag_find_one_icsk()
a5ebb87db843 Revert "mmc: block: don't use parameter prefix if built as module"
3d65ae4634ed writeback: initialize inode members that track writeback history
5ef11c35ce86 mm: ASLR: use get_random_long()
0d9c51a6e1df block: partition: add partition specific uevent callbacks
for partition info
a9ddc71f5840 coresight: fixing lockdep error
fae54158792a coresight: coresight_unregister() function cleanup
f2dfab3568fc coresight: release reference taken by 'bus_find_device()'
ad725aee070c coresight: remove csdev's link from topology
b15f0fb657e0 coresight: removing bind/unbind options from sysfs
4a2e2b19f96a stm class: Hide STM-specific options if STM is disabled
4c127fd16e6b stm class: Fix locking in unbinding policy path
c74f7e8281ad stm class: Fix link list locking
f08b18266c71 stm class: Prevent user-controllable allocations
59be422e4ce1 stm class: Support devices with multiple instances
1810f2c44817 stm class: Fix unlocking braino in the error path
cde4ad836884 stm class: Guard output assignment against concurrency
f7c81c7176c7 stm class: Fix unbalanced module/device refcounting
b4ca34aaf78e stm class: Fix a race in unlinking
bf16e5b8cdea coresight: "DEVICE_ATTR_RO" should defined as static.
61390593f723 coresight: etm4x: Check every parameter used by dma_xx_coherent.
3694bd76781b asm-generic: Fix local variable shadow in __set_fixmap_offset
18e77054de74 staging: ashmem: Avoid deadlock with mmap/shrink
90a2f171383b staging: ashmem: Add missing include
1328d8efef17 staging: ion: Set minimum carveout heap allocation order
to PAGE_SHIFT
4532150762ce staging: goldfish: audio: fix compiliation on arm
1b9bdf5c1661 ARM: 8510/1: rework ARM_CPU_SUSPEND dependencies
b660950c60a7 arm64/kernel: fix incorrect EL0 check in inv_entry macro
b4201cc4fc6e mac80211: fix "warning: ‘target_metric’ may be used uninitialized"
dcb10a967ce8 perf/ring_buffer: Refuse to begin AUX transaction after
rb->aux_mmap_count drops
812264550dcb arm64: kernel: Include _AC definition in page.h
f6cf0545ec69 PM / Hibernate: Call flush_icache_range() on pages
restored in-place
cbe4a61d1ddc stm class: Do not leak the chrdev in error path
389b6699a2aa stm class: Fix stm device initialization order
bbb1681ee365 arm64: mm: mark fault_info table const
5014e904681d coresight: Handle build path error
dcb94b88c09c ipv6: fix endianness error in icmpv6_err
cee51c33f52e usb: gadget: configfs: add mutex lock before unregister gadget
207707d8fd48 usb: gadget: rndis: free response queue during
REMOTE_NDIS_RESET_MSG
7b996243fab4 tun: fix transmit timestamp support
01b411590665 cpu/hotplug: Handle unbalanced hotplug enable/disable
359be67862c5 dma mapping : export caller to vmallocinfo
ace6033ec5c3 video: fbdev: Set pixclock = 0 in goldfishfb

These are from https://github.com/BPI-SINOVOIP/BPI-M64-bsp-4.4:
91d372117695 s390: add DEBUG_RODATA support
99a507771fa5 arm64: kconfig: drop CONFIG_RTC_LIB dependency
38b850a73034 arm64: spinlock: order spin_{is_locked,unlock_wait}
against local locks
987aa5f80596 mmc: mmc: fix switch timeout issue caused by jiffies precision
4ef8c1c93f84 cfg80211: size various nl80211 messages correctly
c88215d7050f cfg80211: Fix documentation for connect result
5c2a625937ba arm64: support keyctl() system call in 32-bit mode

And finally, one from
https://github.com/BPI-SINOVOIP/BPI-Mainline-kernel/tree/master/linux-4.14:
a830405ee452 stmmac: copy unicast mac address to MAC registers

I can post them as patches as well if that helps.

Overall, my script found 380, 80 and 55 patches backported from mainline here,
respectively, but most of them are clearly not stable candidates.

       Arnd

Re: [STABLE candidates] backport candidates from Allwinner kernels

[Sasha Levin]

On Fri, Mar 22, 2019 at 09:46:18PM +0100, Arnd Bergmann wrote:
>These are from https://github.com/BPI-SINOVOIP/BPI-M2P-bsp-4.4
>
>They contain a bunch of things from an older Android kernel that looks
>like it would still
>make sense for normal stable kernels as well, very little here is
>allwinner specific:
>
>0905f04eb21f sched/fair: Fix new task's load avg removed from source
>CPU in wake_up_new_task()
>382c55f88ffe mmc: block: Allow more than 8 partitions per card
>b9b7aebb42d1 arm64: fix COMPAT_SHMLBA definition for large pages
>b523e185bba3 efi: stub: define DISABLE_BRANCH_PROFILING for all architectures
>6c044fecdf78 ARM: 8458/1: bL_switcher: add GIC dependency
>ad84f56bf6d6 ARM: 8494/1: mm: Enable PXN when running non-LPAE kernel
>on LPAE processor
>2010b93e9317 net: tcp: deal with listen sockets properly in tcp_abort.
>699f68556943 android: unconditionally remove callbacks in sync_fence_free()
>0eb77e988032 vmstat: make vmstat_updater deferrable again and shut down on idle
>8d43b49e7e00 hid-sensor-hub.c: fix wrong do_div() usage
>75feee3d9d51 arm64: hide __efistub_ aliases from kallsyms
>728dabd6d175 Eliminate the .eh_frame sections from the aarch64 vmlinux
>and kernel modules
>45c815f06b80 perf: Synchronously free aux pages in case of allocation failure
>7c1306723ee9 net: diag: support v4mapped sockets in inet_diag_find_one_icsk()
>a5ebb87db843 Revert "mmc: block: don't use parameter prefix if built as module"
>3d65ae4634ed writeback: initialize inode members that track writeback history
>5ef11c35ce86 mm: ASLR: use get_random_long()
>0d9c51a6e1df block: partition: add partition specific uevent callbacks
>for partition info
>a9ddc71f5840 coresight: fixing lockdep error
>fae54158792a coresight: coresight_unregister() function cleanup
>f2dfab3568fc coresight: release reference taken by 'bus_find_device()'
>ad725aee070c coresight: remove csdev's link from topology
>b15f0fb657e0 coresight: removing bind/unbind options from sysfs
>4a2e2b19f96a stm class: Hide STM-specific options if STM is disabled
>4c127fd16e6b stm class: Fix locking in unbinding policy path
>c74f7e8281ad stm class: Fix link list locking
>f08b18266c71 stm class: Prevent user-controllable allocations
>59be422e4ce1 stm class: Support devices with multiple instances
>1810f2c44817 stm class: Fix unlocking braino in the error path
>cde4ad836884 stm class: Guard output assignment against concurrency
>f7c81c7176c7 stm class: Fix unbalanced module/device refcounting
>b4ca34aaf78e stm class: Fix a race in unlinking
>bf16e5b8cdea coresight: "DEVICE_ATTR_RO" should defined as static.
>61390593f723 coresight: etm4x: Check every parameter used by dma_xx_coherent.
>3694bd76781b asm-generic: Fix local variable shadow in __set_fixmap_offset
>18e77054de74 staging: ashmem: Avoid deadlock with mmap/shrink
>90a2f171383b staging: ashmem: Add missing include
>1328d8efef17 staging: ion: Set minimum carveout heap allocation order
>to PAGE_SHIFT
>4532150762ce staging: goldfish: audio: fix compiliation on arm
>1b9bdf5c1661 ARM: 8510/1: rework ARM_CPU_SUSPEND dependencies
>b660950c60a7 arm64/kernel: fix incorrect EL0 check in inv_entry macro
>b4201cc4fc6e mac80211: fix "warning: ‘target_metric’ may be used uninitialized"
>dcb10a967ce8 perf/ring_buffer: Refuse to begin AUX transaction after
>rb->aux_mmap_count drops
>812264550dcb arm64: kernel: Include _AC definition in page.h
>f6cf0545ec69 PM / Hibernate: Call flush_icache_range() on pages
>restored in-place
>cbe4a61d1ddc stm class: Do not leak the chrdev in error path
>389b6699a2aa stm class: Fix stm device initialization order
>bbb1681ee365 arm64: mm: mark fault_info table const
>5014e904681d coresight: Handle build path error
>dcb94b88c09c ipv6: fix endianness error in icmpv6_err
>cee51c33f52e usb: gadget: configfs: add mutex lock before unregister gadget
>207707d8fd48 usb: gadget: rndis: free response queue during
>REMOTE_NDIS_RESET_MSG
>7b996243fab4 tun: fix transmit timestamp support
>01b411590665 cpu/hotplug: Handle unbalanced hotplug enable/disable
>359be67862c5 dma mapping : export caller to vmallocinfo
>ace6033ec5c3 video: fbdev: Set pixclock = 0 in goldfishfb
>
>These are from https://github.com/BPI-SINOVOIP/BPI-M64-bsp-4.4:
>91d372117695 s390: add DEBUG_RODATA support
>99a507771fa5 arm64: kconfig: drop CONFIG_RTC_LIB dependency
>38b850a73034 arm64: spinlock: order spin_{is_locked,unlock_wait}
>against local locks
>987aa5f80596 mmc: mmc: fix switch timeout issue caused by jiffies precision
>4ef8c1c93f84 cfg80211: size various nl80211 messages correctly
>c88215d7050f cfg80211: Fix documentation for connect result
>5c2a625937ba arm64: support keyctl() system call in 32-bit mode
>
>And finally, one from
>https://github.com/BPI-SINOVOIP/BPI-Mainline-kernel/tree/master/linux-4.14:
>a830405ee452 stmmac: copy unicast mac address to MAC registers
>
>I can post them as patches as well if that helps.
>
>Overall, my script found 380, 80 and 55 patches backported from mainline here,
>respectively, but most of them are clearly not stable candidates.

Hi Arnd,

I've queued all but the following, which didn't look like fixes to me:

block: partition: add partition specific uevent callbacks for partition info
coresight: removing bind/unbind options from sysfs
stm class: Hide STM-specific options if STM is disabled
arm64: mm: mark fault_info table const
dma mapping : export caller to vmallocinfo
s390: add DEBUG_RODATA support
arm64: support keyctl() system call in 32-bit mode

Any reason they should be included?

--
Thanks,
Sasha

Re: [STABLE candidates] backport candidates from Allwinner kernels

[Arnd Bergmann]

On Wed, Mar 27, 2019 at 11:59 PM Sasha Levin <sashal@kernel.org> wrote:
>
> On Fri, Mar 22, 2019 at 09:46:18PM +0100, Arnd Bergmann wrote:

> Hi Arnd,
>
> I've queued all but the following, which didn't look like fixes to me:

Ok, thanks!

> block: partition: add partition specific uevent callbacks for partition info
> coresight: removing bind/unbind options from sysfs
> stm class: Hide STM-specific options if STM is disabled
> arm64: mm: mark fault_info table const
> dma mapping : export caller to vmallocinfo
> s390: add DEBUG_RODATA support
> arm64: support keyctl() system call in 32-bit mode
>
> Any reason they should be included?

The keyctl patch is probably the only one that makes
a difference here, please include that to fix a bug in the
compat layer of lacking keyctl access.

The coresight unbind fix is a bugfix that avoids a user
triggered kernel oops, but since unbind requires root
permissions, it's not a critical fix.

I assumed the STM Kconfig avoided a randconfig
build issue, but I may be misremembering that,
and I did not try it without that patch.

       Arnd

Re: [STABLE candidates] backport candidates from Allwinner kernels

[Sasha Levin]

On Thu, Mar 28, 2019 at 11:02:16AM +0800, Arnd Bergmann wrote:
>On Wed, Mar 27, 2019 at 11:59 PM Sasha Levin <sashal@kernel.org> wrote:
>>
>> On Fri, Mar 22, 2019 at 09:46:18PM +0100, Arnd Bergmann wrote:
>
>> Hi Arnd,
>>
>> I've queued all but the following, which didn't look like fixes to me:
>
>Ok, thanks!
>
>> block: partition: add partition specific uevent callbacks for partition info
>> coresight: removing bind/unbind options from sysfs
>> stm class: Hide STM-specific options if STM is disabled
>> arm64: mm: mark fault_info table const
>> dma mapping : export caller to vmallocinfo
>> s390: add DEBUG_RODATA support
>> arm64: support keyctl() system call in 32-bit mode
>>
>> Any reason they should be included?
>
>The keyctl patch is probably the only one that makes
>a difference here, please include that to fix a bug in the
>compat layer of lacking keyctl access.
>
>The coresight unbind fix is a bugfix that avoids a user
>triggered kernel oops, but since unbind requires root
>permissions, it's not a critical fix.
>
>I assumed the STM Kconfig avoided a randconfig
>build issue, but I may be misremembering that,
>and I did not try it without that patch.

I'll queue them up, thanks for the explanation.

--
Thanks,
Sasha