muen/linux.git
18 months agoMerge tag 'acpi-5.0-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael...
Linus Torvalds [Fri, 11 Jan 2019 17:04:36 +0000 (09:04 -0800)]
Merge tag 'acpi-5.0-rc2' of git://git./linux/kernel/git/rafael/linux-pm

Pull ACPI fixes from Rafael Wysocki:
 "Fix a build failure introduced recently, fix the xpower PMIC ACPI
  driver, clean up the handling of duplicate entries in _PRx power
  resource lists and fix addresses in NUMA-related messages on 32-bit
  with PAE.

  Specifics:

   - Fix build failures with both CONFIG_NLS and CONFIG_PCI unset that
     can occur since ACPI can be built without PCI now (Sinan Kaya).

   - Clean up the handling of duplicate entries in power resource lists
     returned by _PRx evaluation to avoid triggering WARN_ON() on
     attempts to add duplicate symlinks in sysfs (Hans de Goede).

   - Fix issues with the TS current-source switching on systems using
     the xpower PMIC by avoiding to update unrelated bits in the TS
     pin-ctrl register and avoiding to unconditionally enable TS
     current-source on systems where it is not used (Hans de Goede).

   - Fix addresses in NUMA-related messages on 32-bit with PAE which can
     be truncated due to integer type conversions (Chao Fan)"

* tag 'acpi-5.0-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
  ACPI / PMIC: xpower: Fix TS-pin current-source handling
  ACPI: NUMA: Use correct type for printing addresses on i386-PAE
  ACPI: power: Skip duplicate power resource references in _PRx
  ACPI: Fix build failure when CONFIG_NLS is set to 'n'

18 months agoMerge tag 'pm-5.0-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm
Linus Torvalds [Fri, 11 Jan 2019 17:01:43 +0000 (09:01 -0800)]
Merge tag 'pm-5.0-rc2' of git://git./linux/kernel/git/rafael/linux-pm

Pull power management updates from Rafael Wysocki:
 "These fix fallout after starting to use hrtimers in the runtime PM
  framework, fix a few cpufreq issues, fix a recently broken reference
  to cpuidle documentation, update MAINTAINERS entries for cpufreq and
  cpuidle and make the recently added system suspend and resume support
  in devfreq actually work.

  Specifics:

   - Prevent integer overflows from occurring on 32-bit when converting
     milliseconds to nanoseconds in the runtime PM framework and update
     comments that still refer to jiffies in it (Vincent Guittot,
     Ladislav Michl).

   - Fix the SCMI cpufreq driver to always use the same frequency units
     for arch_set_freq_scale() and make the scale-invariant load
     tracking acutally work with this driver (Quentin Perret).

   - Fix freeing of dynamic OPPs in the SCPI and SCMI cpufreq drivers
     broken during the 4.20 defelopment cycle (Viresh Kumar).

   - Prevent the cpufreq core from attempting to return the current
     frequency of offline CPUs (Sudeep Holla).

   - Add devfreq suspend and resume hooks (missed previously) to the PM
     core to make the recently added system suspend and resume support
     in devfreq actually work (Lukasz Luba).

   - Update MAINTAINERS entries for cpufreq and cpuidle, mostly to add
     references to new/current documentation to them (Rafael Wysocki).

   - Fix a recently broken reference to cpuidle documentation (Otto
     Sabart)"

* tag 'pm-5.0-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
  PM-runtime: Fix autosuspend_delay on 32bits arch
  PM-runtime: Fix 'jiffies' in comments after switch to hrtimers
  cpufreq: scmi: Fix frequency invariance in slow path
  doc: trace: fix reference to cpuidle documentation file
  cpufreq: check if policy is inactive early in __cpufreq_get()
  cpufreq: scpi/scmi: Fix freeing of dynamic OPPs
  cpuidle / Documentation: Update cpuidle MAINTAINERS entry
  cpufreq / Documentation: Update cpufreq MAINTAINERS entry
  PM: sleep: call devfreq suspend/resume

18 months agoMerge tag 'drm-fixes-2019-01-11' of git://anongit.freedesktop.org/drm/drm
Linus Torvalds [Fri, 11 Jan 2019 16:58:02 +0000 (08:58 -0800)]
Merge tag 'drm-fixes-2019-01-11' of git://anongit.freedesktop.org/drm/drm

Pull drm fixes from Dave Airlie:
 "Not a huge amount for rc2, assume the usual quiet period, and rc3 will
  be most of it.

  amdgpu:
   - Powerplay fixes
   - Virtual display pinning fixes
   - Golden register updates for Vega
   - Pitch and gem size validation fixes
   - SR-IOV init error fix
   - Pagetables in system RAM disable for some Raven system
   - DP-MST resume fixes

  tc358767 bridge:
   - fix to work with displayport connector"

* tag 'drm-fixes-2019-01-11' of git://anongit.freedesktop.org/drm/drm: (26 commits)
  drm/amdgpu: disable system memory page tables for now
  drm/amdgpu: set WRITE_BURST_LENGTH to 64B to workaround SDMA1 hang
  drm/amdgpu: fix CPDMA hang in PRT mode for VEGA20
  drm/bridge: tc358767: use DP connector if no panel set
  drm/bridge: tc358767: fix output H/V syncs
  drm/bridge: tc358767: reject modes which require too much BW
  drm/bridge: tc358767: fix initial DP0/1_SRCCTRL value
  drm/bridge: tc358767: fix single lane configuration
  drm/bridge: tc358767: add defines for DP1_SRCCTRL & PHY_2LANE
  drm/bridge: tc358767: add bus flags
  drm/dp_mst: Add __must_check to drm_dp_mst_topology_mgr_resume()
  drm/amdgpu: Don't fail resume process if resuming atomic state fails
  drm/amdgpu: Don't ignore rc from drm_dp_mst_topology_mgr_resume()
  drm/amdgpu: validate user GEM object size
  drm/amdgpu: validate user pitch alignment
  drm/amd/powerplay: drop the unnecessary uclk hard min setting
  drm/amd/powerplay: avoid possible buffer overflow
  drm/amd/powerplay: create pp_od_clk_voltage device file under OD support
  drm/amd/powerplay: update OD support flag for SKU with no OD capabilities
  drm/amdgpu: make gfx9 enter into rlc safe mode when set MGCG
  ...

18 months agoMerge branches 'acpi-pci', 'acpi-power' and 'acpi-misc'
Rafael J. Wysocki [Fri, 11 Jan 2019 09:12:07 +0000 (10:12 +0100)]
Merge branches 'acpi-pci', 'acpi-power' and 'acpi-misc'

* acpi-pci:
  ACPI: Fix build failure when CONFIG_NLS is set to 'n'

* acpi-power:
  ACPI: power: Skip duplicate power resource references in _PRx

* acpi-misc:
  ACPI: NUMA: Use correct type for printing addresses on i386-PAE

18 months agoMerge branches 'pm-cpuidle', 'pm-cpufreq' and 'pm-sleep'
Rafael J. Wysocki [Fri, 11 Jan 2019 09:09:51 +0000 (10:09 +0100)]
Merge branches 'pm-cpuidle', 'pm-cpufreq' and 'pm-sleep'

* pm-cpuidle:
  doc: trace: fix reference to cpuidle documentation file
  cpuidle / Documentation: Update cpuidle MAINTAINERS entry

* pm-cpufreq:
  cpufreq: scmi: Fix frequency invariance in slow path
  cpufreq: check if policy is inactive early in __cpufreq_get()
  cpufreq: scpi/scmi: Fix freeing of dynamic OPPs
  cpufreq / Documentation: Update cpufreq MAINTAINERS entry

* pm-sleep:
  PM: sleep: call devfreq suspend/resume

18 months agoMerge tag 'pci-v5.0-fixes-1' of git://git.kernel.org/pub/scm/linux/kernel/git/helgaas/pci
Linus Torvalds [Thu, 10 Jan 2019 22:16:47 +0000 (14:16 -0800)]
Merge tag 'pci-v5.0-fixes-1' of git://git./linux/kernel/git/helgaas/pci

Pull PCI fix from Bjorn Helgaas:
 "Fix Amlogic Meson host controller driver build failure (Corentin
  Labbe)"

* tag 'pci-v5.0-fixes-1' of git://git.kernel.org/pub/scm/linux/kernel/git/helgaas/pci:
  PCI: amlogic: Fix build failure due to missing gpio header

18 months agoMerge branch 'drm-fixes-5.0' of git://people.freedesktop.org/~agd5f/linux into drm...
Dave Airlie [Thu, 10 Jan 2019 21:37:56 +0000 (07:37 +1000)]
Merge branch 'drm-fixes-5.0' of git://people.freedesktop.org/~agd5f/linux into drm-fixes

- Powerplay fixes
- Virtual display pinning fixes
- Golden register updates for vega
- Pitch and gem size validation fixes
- Fix for error case in sr-iov init
- Disable page tables in system memory on RV due to issues with IOMMU
  reported on some platforms

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Alex Deucher <alexdeucher@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190109204336.3315-1-alexander.deucher@amd.com
18 months agoMerge tag 'riscv-for-linus-4.21-rc2-v2' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Thu, 10 Jan 2019 21:36:53 +0000 (13:36 -0800)]
Merge tag 'riscv-for-linus-4.21-rc2-v2' of git://git./linux/kernel/git/palmer/riscv-linux

Pull RISC-V updates from Palmer Dabbelt:
 "This tag contains a handful of updates that slipped through the cracks
  during the merge window due to the holidays. The fixes are mostly
  independent, with the exception of one larger audit-related branch.

  Core RISC-V updates:

   - The BSS has been moved, which shrinks flat images.

   - A fix to test-bpf so it compiles on RV64I-based systems.

   - A fix to respect the kernel commandline when there is no device
     tree.

   - A fix to prevent CPUs from trying to put themselves to sleep when
     bringing down the system.

   - Support for MODULE_SECTIONS on RV32I-based systems.

   - [new in v2] The addition of an SBI earlycon driver. This is
     definately a new feature, but I'd like to include it now because I
     dropped this patch when submitting the merge window PR that removed
     our EARLY_PRINTK support.

  RISC-V audit updates:

   - The addition of NR_syscalls into unistd.h, which is necessary for
     CONFIG_FTRACE_SYSCALLS.

   - The definition of CREATE_TRACE_POINTS so __tracepoint_sys_{enter,exit}
     get defined.

   - A fix for trace_sys_exit() so we can enable HAVE_SYSCALL_TRACEPOINTS

  As usual, I've tested this by booting a Fedora-based image on a recent
  QEMU (this time just whatever I had lying around).

* tag 'riscv-for-linus-4.21-rc2-v2' of git://git.kernel.org/pub/scm/linux/kernel/git/palmer/riscv-linux:
  tty/serial: Add RISC-V SBI earlycon support
  riscv: add HAVE_SYSCALL_TRACEPOINTS to Kconfig
  riscv: fix trace_sys_exit hook
  riscv: define CREATE_TRACE_POINTS in ptrace.c
  riscv: define NR_syscalls in unistd.h
  riscv: audit: add audit hook in do_syscall_trace_enter/exit()
  riscv: add audit support
  RISC-V: Support MODULE_SECTIONS mechanism on RV32
  MAINTAINERS: SiFive drivers: add myself as a SiFive driver maintainer
  MAINTAINERS: SiFive drivers: change the git tree to a SiFive git tree
  riscv: don't stop itself in smp_send_stop
  arch: riscv: support kernel command line forcing when no DTB passed
  tools uapi: fix RISC-V 64-bit support
  RISC-V: Make BSS section as the last section in vmlinux.lds.S

18 months agoMerge tag 'drm-misc-fixes-2019-01-10' of git://anongit.freedesktop.org/drm/drm-misc...
Dave Airlie [Thu, 10 Jan 2019 20:32:37 +0000 (06:32 +1000)]
Merge tag 'drm-misc-fixes-2019-01-10' of git://anongit.freedesktop.org/drm/drm-misc into drm-fixes

Pull request for drm-misc-fixes for v5.0-rc2:
- Fixes for the tc358767 bridge to work correctly with
  tc358867 using a DP connector.
- Make resume work on amdgpu when a DP-MST display is unplugged.

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/1c47722d-c416-184d-4340-0dc6a614d685@linux.intel.com
18 months agoMerge tag 'vfio-v5.0-rc2' of git://github.com/awilliam/linux-vfio
Linus Torvalds [Thu, 10 Jan 2019 17:20:46 +0000 (09:20 -0800)]
Merge tag 'vfio-v5.0-rc2' of git://github.com/awilliam/linux-vfio

Pull VFIO fixes from Alex Williamson:

 - Fix trace header include path for in-tree builds (Masahiro Yamada)

 - Fix overflow in unmap wrap-around test (Alex Williamson)

* tag 'vfio-v5.0-rc2' of git://github.com/awilliam/linux-vfio:
  vfio/type1: Fix unmap overflow off-by-one
  vfio/pci: set TRACE_INCLUDE_PATH to fix the build error

18 months agoMerge tag 'sound-5.0-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai...
Linus Torvalds [Thu, 10 Jan 2019 17:17:48 +0000 (09:17 -0800)]
Merge tag 'sound-5.0-rc2' of git://git./linux/kernel/git/tiwai/sound

Pull sound fixes from Takashi Iwai:
 "A collection of small fixes for USB-audio, HD-audio and cs46xx.

  The USB-audio fixes are for out-of-bound accesses and a regression in
  the recent cleanup, while HD-audio fixes are usual device-specific
  quirks"

* tag 'sound-5.0-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound:
  ALSA: hda/realtek - Disable headset Mic VREF for headset mode of ALC225
  ALSA: hda/realtek - Add unplug function into unplug state of Headset Mode for ALC225
  ALSA: usb-audio: fix CM6206 register definitions
  ALSA: cs46xx: Potential NULL dereference in probe
  ALSA: hda/realtek - Support Dell headset mode for New AIO platform
  ALSA: usb-audio: Fix an out-of-bound read in create_composite_quirks
  ALSA: usb-audio: Always check descriptor sizes in parser code
  ALSA: usb-audio: Check mixer unit descriptors more strictly
  ALSA: usb-audio: Avoid access before bLength check in build_audio_procunit()

18 months agoMerge tag 'mtd/fixes-for-5.0-rc2' of git://git.infradead.org/linux-mtd
Linus Torvalds [Thu, 10 Jan 2019 17:14:12 +0000 (09:14 -0800)]
Merge tag 'mtd/fixes-for-5.0-rc2' of git://git.infradead.org/linux-mtd

Pull mtd fixes from Boris Brezillon:
 "Core MTD Fixes:

   - Fix a bug introduced when exposing MTD devs as NVMEM providers and
     check for add_mtd_device() return code everywhere

  raw NAND fixes:

   - Fix a memory corruption in the QCOM driver"

* tag 'mtd/fixes-for-5.0-rc2' of git://git.infradead.org/linux-mtd:
  mtd: rawnand: qcom: fix memory corruption that causes panic
  mtd: Check add_mtd_device() ret code
  mtd: Fix the check on nvmem_register() ret code

18 months agocsky: fixup compile error with CPU 810.
Guo Ren [Thu, 10 Jan 2019 12:28:39 +0000 (20:28 +0800)]
csky: fixup compile error with CPU 810.

This bug is from commit f553aa1c13cb ("csky: fixup relocation error with
807 & 860").

I forgot to compile with 810 for that patch.

Signed-off-by: Guo Ren <ren_guo@c-sky.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
18 months agomm/mmu_notifier: mm/rmap.c: Fix a mmu_notifier range bug in try_to_unmap_one
Sean Christopherson [Thu, 10 Jan 2019 00:51:17 +0000 (16:51 -0800)]
mm/mmu_notifier: mm/rmap.c: Fix a mmu_notifier range bug in try_to_unmap_one

The conversion to use a structure for mmu_notifier_invalidate_range_*()
unintentionally changed the usage in try_to_unmap_one() to init the
'struct mmu_notifier_range' with vma->vm_start instead of @address,
i.e. it invalidates the wrong address range.  Revert to the correct
address range.

Manifests as KVM use-after-free WARNINGs and subsequent "BUG: Bad page
state in process X" errors when reclaiming from a KVM guest due to KVM
removing the wrong pages from its own mappings.

Reported-by: leozinho29_eu@hotmail.com
Reported-by: Mike Galbraith <efault@gmx.de>
Reported-and-tested-by: Adam Borowski <kilobyte@angband.pl>
Reviewed-by: Jérôme Glisse <jglisse@redhat.com>
Reviewed-by: Pankaj gupta <pagupta@redhat.com>
Cc: Christian König <christian.koenig@amd.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Matthew Wilcox <mawilcox@microsoft.com>
Cc: Ross Zwisler <zwisler@kernel.org>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Felix Kuehling <felix.kuehling@amd.com>
Cc: Ralph Campbell <rcampbell@nvidia.com>
Cc: John Hubbard <jhubbard@nvidia.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Fixes: ac46d4f3c432 ("mm/mmu_notifier: use structure for invalidate_range_start/end calls v2")
Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
18 months agoPM-runtime: Fix autosuspend_delay on 32bits arch
Vincent Guittot [Thu, 10 Jan 2019 09:00:40 +0000 (10:00 +0100)]
PM-runtime: Fix autosuspend_delay on 32bits arch

Cast autosuspend_delay to u64 to make sure that the full computation
of 'expires' or slack will be done in u64, even on 32bits arch.

Otherwise, any delay greater than 2^31 nsec can overflow if signed
32bits is used when converting delay from msec to nsec.

Fixes: 8234f6734c5d (PM-runtime: Switch autosuspend over to using hrtimers)
Reported-by: Tony Lindgren <tony@atomide.com>
Tested-by: Tony Lindgren <tony@atomide.com>
Signed-off-by: Vincent Guittot <vincent.guittot@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
18 months agoPM-runtime: Fix 'jiffies' in comments after switch to hrtimers
Ladislav Michl [Wed, 9 Jan 2019 23:19:44 +0000 (00:19 +0100)]
PM-runtime: Fix 'jiffies' in comments after switch to hrtimers

PM-runtime now uses the hrtimers infrastructure for autosuspend, however
comments still reference 'jiffies'.

Fixes: 8234f6734c5d (PM-runtime: Switch autosuspend over to using hrtimers)
Signed-off-by: Ladislav Michl <ladis@linux-mips.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
18 months agotty/serial: Add RISC-V SBI earlycon support
Anup Patel [Tue, 4 Dec 2018 13:55:05 +0000 (19:25 +0530)]
tty/serial: Add RISC-V SBI earlycon support

In RISC-V, the M-mode runtime firmware provide SBI calls for
debug prints. This patch adds earlycon support using RISC-V
SBI console calls. To enable it, just pass "earlycon=sbi" in
kernel parameters.

Signed-off-by: Anup Patel <anup@brainfault.org>
Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Reviewed-by: Palmer Dabbelt <palmer@sifive.com>
Signed-off-by: Palmer Dabbelt <palmer@sifive.com>
18 months agodrm/amdgpu: disable system memory page tables for now
Christian König [Mon, 7 Jan 2019 13:43:55 +0000 (14:43 +0100)]
drm/amdgpu: disable system memory page tables for now

We hit a problem with IOMMU with that. Disable until we have time to
debug further.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
18 months agodrm/amdgpu: set WRITE_BURST_LENGTH to 64B to workaround SDMA1 hang
Jim Qu [Mon, 17 Dec 2018 09:00:50 +0000 (17:00 +0800)]
drm/amdgpu: set WRITE_BURST_LENGTH to 64B to workaround SDMA1 hang

effect asics: VEGA10 and VEGA12

Signed-off-by: Jim Qu <Jim.Qu@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
18 months agodrm/amdgpu: fix CPDMA hang in PRT mode for VEGA20
Tao Zhou [Tue, 8 Jan 2019 07:08:44 +0000 (15:08 +0800)]
drm/amdgpu: fix CPDMA hang in PRT mode for VEGA20

Fix CPDMA hang in PRT mode for both VEGA10 and VEGA20

Signed-off-by: Tao Zhou <tao.zhou1@amd.com>
Tested-by: Yukun.Li <yukun1.li@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
18 months agoMerge tag 'csky-for-linus-5.0-rc1' of git://github.com/c-sky/csky-linux
Linus Torvalds [Wed, 9 Jan 2019 18:00:11 +0000 (10:00 -0800)]
Merge tag 'csky-for-linus-5.0-rc1' of git://github.com/c-sky/csky-linux

Pull arch/csky bug fixes from Guo Ren:
 "Here are some fixup patches for 5.0-rc1:

   - fix compile error with pte_alloc

   - fix handle_irq_perbit break irq flow

   - fix CACHEV1 store instruction fast retire

   - fix module relocation error with 807 & 860

   - add csky kernel features to documentation"

* tag 'csky-for-linus-5.0-rc1' of git://github.com/c-sky/csky-linux:
  irqchip/csky: fixup handle_irq_perbit break irq
  csky: fixup compile error with pte_alloc
  csky: fixup CACHEV1 store instruction fast retire
  csky: fixup relocation error with 807 & 860
  Documentation/features: Add csky kernel features

18 months agocpufreq: scmi: Fix frequency invariance in slow path
Quentin Perret [Wed, 9 Jan 2019 10:42:36 +0000 (10:42 +0000)]
cpufreq: scmi: Fix frequency invariance in slow path

The scmi-cpufreq driver calls the arch_set_freq_scale() callback on
frequency changes to provide scale-invariant load-tracking signals to
the scheduler. However, in the slow path, it does so while specifying
the current and max frequencies in different units, hence resulting in a
broken freq_scale factor.

Fix this by passing all frequencies in KHz, as stored in the CPUFreq
frequency table.

Fixes: 99d6bdf33877 (cpufreq: add support for CPU DVFS based on SCMI message protocol)
Signed-off-by: Quentin Perret <quentin.perret@arm.com>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
Acked-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: 4.17+ <stable@vger.kernel.org> # v4.17+
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
18 months agodoc: trace: fix reference to cpuidle documentation file
Otto Sabart [Tue, 8 Jan 2019 23:56:51 +0000 (00:56 +0100)]
doc: trace: fix reference to cpuidle documentation file

Old cpuidle/sysfs.txt file was replaced in aa5eee355b46. So, refer to
an updated file.

Fixes: aa5eee355b46 (Documentation: admin-guide: PM: Add cpuidle document)
Signed-off-by: Otto Sabart <ottosabart@seberm.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
18 months agodrm/bridge: tc358767: use DP connector if no panel set
Tomi Valkeinen [Thu, 3 Jan 2019 11:59:54 +0000 (13:59 +0200)]
drm/bridge: tc358767: use DP connector if no panel set

tc358767 driver sets the connector type always to eDP.

This patch sets the type to DP if there is no panel defined, which
implies that there's a DP connector on the board.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Reviewed-by: Andrzej Hajda <a.hajda@samsung.com>
Signed-off-by: Andrzej Hajda <a.hajda@samsung.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190103115954.12785-8-tomi.valkeinen@ti.com
18 months agodrm/bridge: tc358767: fix output H/V syncs
Tomi Valkeinen [Thu, 3 Jan 2019 11:59:53 +0000 (13:59 +0200)]
drm/bridge: tc358767: fix output H/V syncs

The H and V syncs of the DP output are always set to active high. This
patch fixes the syncs by configuring them according to the videomode.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Reviewed-by: Andrzej Hajda <a.hajda@samsung.com>
Signed-off-by: Andrzej Hajda <a.hajda@samsung.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190103115954.12785-7-tomi.valkeinen@ti.com
18 months agodrm/bridge: tc358767: reject modes which require too much BW
Tomi Valkeinen [Thu, 3 Jan 2019 11:59:52 +0000 (13:59 +0200)]
drm/bridge: tc358767: reject modes which require too much BW

The current driver accepts any videomode with pclk < 154MHz. This is not
correct, as with 1 lane and/or 1.62Mbps speed not all videomodes can be
supported.

Add code to reject modes that require more bandwidth that is available.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Reviewed-by: Andrzej Hajda <a.hajda@samsung.com>
Signed-off-by: Andrzej Hajda <a.hajda@samsung.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190103115954.12785-6-tomi.valkeinen@ti.com
18 months agodrm/bridge: tc358767: fix initial DP0/1_SRCCTRL value
Tomi Valkeinen [Thu, 3 Jan 2019 11:59:51 +0000 (13:59 +0200)]
drm/bridge: tc358767: fix initial DP0/1_SRCCTRL value

Initially DP0_SRCCTRL is set to a static value which includes
DP0_SRCCTRL_LANES_2 and DP0_SRCCTRL_BW27, even when only 1 lane of
1.62Gbps speed is used. DP1_SRCCTRL is configured to a magic number.

This patch changes the configuration as follows:

Configure DP0_SRCCTRL by using tc_srcctrl() which provides the correct
value.

DP1_SRCCTRL needs two bits to be set to the same value as DP0_SRCCTRL:
SSCG and BW27. All other bits can be zero.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Reviewed-by: Andrzej Hajda <a.hajda@samsung.com>
Signed-off-by: Andrzej Hajda <a.hajda@samsung.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190103115954.12785-5-tomi.valkeinen@ti.com
18 months agodrm/bridge: tc358767: fix single lane configuration
Tomi Valkeinen [Thu, 3 Jan 2019 11:59:50 +0000 (13:59 +0200)]
drm/bridge: tc358767: fix single lane configuration

PHY_2LANE bit is always set in DP_PHY_CTRL, breaking 1 lane use.

Set PHY_2LANE only when 2 lanes are used.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Reviewed-by: Andrzej Hajda <a.hajda@samsung.com>
Signed-off-by: Andrzej Hajda <a.hajda@samsung.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190103115954.12785-4-tomi.valkeinen@ti.com
18 months agodrm/bridge: tc358767: add defines for DP1_SRCCTRL & PHY_2LANE
Tomi Valkeinen [Thu, 3 Jan 2019 11:59:49 +0000 (13:59 +0200)]
drm/bridge: tc358767: add defines for DP1_SRCCTRL & PHY_2LANE

DP1_SRCCTRL register and PHY_2LANE field did not have matching defines.
Add these.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Reviewed-by: Andrzej Hajda <a.hajda@samsung.com>
Signed-off-by: Andrzej Hajda <a.hajda@samsung.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190103115954.12785-3-tomi.valkeinen@ti.com
18 months agodrm/bridge: tc358767: add bus flags
Tomi Valkeinen [Thu, 3 Jan 2019 11:59:48 +0000 (13:59 +0200)]
drm/bridge: tc358767: add bus flags

tc358767 driver does not set DRM bus_flags, even if it does configures
the polarity settings into its registers. This means that the DPI source
can't configure the polarities correctly.

Add sync flags accordingly.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Reviewed-by: Andrzej Hajda <a.hajda@samsung.com>
Signed-off-by: Andrzej Hajda <a.hajda@samsung.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190103115954.12785-2-tomi.valkeinen@ti.com
18 months agoALSA: hda/realtek - Disable headset Mic VREF for headset mode of ALC225
Kailang Yang [Wed, 9 Jan 2019 09:05:24 +0000 (17:05 +0800)]
ALSA: hda/realtek - Disable headset Mic VREF for headset mode of ALC225

Disable Headset Mic VREF for headset mode of ALC225.
This will be controlled by coef bits of headset mode functions.

[ Fixed a compile warning and code simplification -- tiwai ]

Signed-off-by: Kailang Yang <kailang@realtek.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
18 months agoALSA: hda/realtek - Add unplug function into unplug state of Headset Mode for ALC225
Kailang Yang [Wed, 9 Jan 2019 08:23:37 +0000 (16:23 +0800)]
ALSA: hda/realtek - Add unplug function into unplug state of Headset Mode for ALC225

Forgot to add unplug function to unplug state of headset mode
for ALC225.

Signed-off-by: Kailang Yang <kailang@realtek.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
18 months agoMerge branch 'akpm' (patches from Andrew)
Linus Torvalds [Wed, 9 Jan 2019 02:58:29 +0000 (18:58 -0800)]
Merge branch 'akpm' (patches from Andrew)

Merge misc fixes from Andrew Morton:
 "14 fixes"

* emailed patches from Andrew Morton <akpm@linux-foundation.org>:
  mm, page_alloc: do not wake kswapd with zone lock held
  hugetlbfs: revert "use i_mmap_rwsem for more pmd sharing synchronization"
  hugetlbfs: revert "Use i_mmap_rwsem to fix page fault/truncate race"
  mm: page_mapped: don't assume compound page is huge or THP
  mm/memory.c: initialise mmu_notifier_range correctly
  tools/vm/page_owner: use page_owner_sort in the use example
  kasan: fix krealloc handling for tag-based mode
  kasan: make tag based mode work with CONFIG_HARDENED_USERCOPY
  kasan, arm64: use ARCH_SLAB_MINALIGN instead of manual aligning
  mm, memcg: fix reclaim deadlock with writeback
  mm/usercopy.c: no check page span for stack objects
  slab: alien caches must not be initialized if the allocation of the alien cache failed
  fork, memcg: fix cached_stacks case
  zram: idle writeback fixes and cleanup

18 months agoarch/openrisc: Fix issues with access_ok()
Stafford Horne [Tue, 8 Jan 2019 13:15:15 +0000 (22:15 +0900)]
arch/openrisc: Fix issues with access_ok()

The commit 594cc251fdd0 ("make 'user_access_begin()' do 'access_ok()'")
exposed incorrect implementations of access_ok() macro in several
architectures.  This change fixes 2 issues found in OpenRISC.

OpenRISC was not properly using parenthesis for arguments and also using
arguments twice.  This patch fixes those 2 issues.

I test booted this patch with v5.0-rc1 on qemu and it's working fine.

Cc: Guenter Roeck <linux@roeck-us.net>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Reported-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Stafford Horne <shorne@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
18 months agomm, page_alloc: do not wake kswapd with zone lock held
Mel Gorman [Tue, 8 Jan 2019 23:23:39 +0000 (15:23 -0800)]
mm, page_alloc: do not wake kswapd with zone lock held

syzbot reported the following regression in the latest merge window and
it was confirmed by Qian Cai that a similar bug was visible from a
different context.

  ======================================================
  WARNING: possible circular locking dependency detected
  4.20.0+ #297 Not tainted
  ------------------------------------------------------
  syz-executor0/8529 is trying to acquire lock:
  000000005e7fb829 (&pgdat->kswapd_wait){....}, at:
  __wake_up_common_lock+0x19e/0x330 kernel/sched/wait.c:120

  but task is already holding lock:
  000000009bb7bae0 (&(&zone->lock)->rlock){-.-.}, at: spin_lock
  include/linux/spinlock.h:329 [inline]
  000000009bb7bae0 (&(&zone->lock)->rlock){-.-.}, at: rmqueue_bulk
  mm/page_alloc.c:2548 [inline]
  000000009bb7bae0 (&(&zone->lock)->rlock){-.-.}, at: __rmqueue_pcplist
  mm/page_alloc.c:3021 [inline]
  000000009bb7bae0 (&(&zone->lock)->rlock){-.-.}, at: rmqueue_pcplist
  mm/page_alloc.c:3050 [inline]
  000000009bb7bae0 (&(&zone->lock)->rlock){-.-.}, at: rmqueue
  mm/page_alloc.c:3072 [inline]
  000000009bb7bae0 (&(&zone->lock)->rlock){-.-.}, at:
  get_page_from_freelist+0x1bae/0x52a0 mm/page_alloc.c:3491

It appears to be a false positive in that the only way the lock ordering
should be inverted is if kswapd is waking itself and the wakeup
allocates debugging objects which should already be allocated if it's
kswapd doing the waking.  Nevertheless, the possibility exists and so
it's best to avoid the problem.

This patch flags a zone as needing a kswapd using the, surprisingly,
unused zone flag field.  The flag is read without the lock held to do
the wakeup.  It's possible that the flag setting context is not the same
as the flag clearing context or for small races to occur.  However, each
race possibility is harmless and there is no visible degredation in
fragmentation treatment.

While zone->flag could have continued to be unused, there is potential
for moving some existing fields into the flags field instead.
Particularly read-mostly ones like zone->initialized and
zone->contiguous.

Link: http://lkml.kernel.org/r/20190103225712.GJ31517@techsingularity.net
Fixes: 1c30844d2dfe ("mm: reclaim small amounts of memory when an external fragmentation event occurs")
Reported-by: syzbot+93d94a001cfbce9e60e1@syzkaller.appspotmail.com
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Tested-by: Qian Cai <cai@lca.pw>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Michal Hocko <mhocko@suse.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
18 months agohugetlbfs: revert "use i_mmap_rwsem for more pmd sharing synchronization"
Mike Kravetz [Tue, 8 Jan 2019 23:23:36 +0000 (15:23 -0800)]
hugetlbfs: revert "use i_mmap_rwsem for more pmd sharing synchronization"

This reverts b43a9990055958e70347c56f90ea2ae32c67334c

The reverted commit caused issues with migration and poisoning of anon
huge pages.  The LTP move_pages12 test will cause an "unable to handle
kernel NULL pointer" BUG would occur with stack similar to:

  RIP: 0010:down_write+0x1b/0x40
  Call Trace:
    migrate_pages+0x81f/0xb90
    __ia32_compat_sys_migrate_pages+0x190/0x190
    do_move_pages_to_node.isra.53.part.54+0x2a/0x50
    kernel_move_pages+0x566/0x7b0
    __x64_sys_move_pages+0x24/0x30
    do_syscall_64+0x5b/0x180
    entry_SYSCALL_64_after_hwframe+0x44/0xa9

The purpose of the reverted patch was to fix some long existing races
with huge pmd sharing.  It used i_mmap_rwsem for this purpose with the
idea that this could also be used to address truncate/page fault races
with another patch.  Further analysis has determined that i_mmap_rwsem
can not be used to address all these hugetlbfs synchronization issues.
Therefore, revert this patch while working an another approach to the
underlying issues.

Link: http://lkml.kernel.org/r/20190103235452.29335-2-mike.kravetz@oracle.com
Signed-off-by: Mike Kravetz <mike.kravetz@oracle.com>
Reported-by: Jan Stancek <jstancek@redhat.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Cc: "Aneesh Kumar K . V" <aneesh.kumar@linux.vnet.ibm.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: "Kirill A . Shutemov" <kirill.shutemov@linux.intel.com>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Prakash Sangappa <prakash.sangappa@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
18 months agohugetlbfs: revert "Use i_mmap_rwsem to fix page fault/truncate race"
Mike Kravetz [Tue, 8 Jan 2019 23:23:32 +0000 (15:23 -0800)]
hugetlbfs: revert "Use i_mmap_rwsem to fix page fault/truncate race"

This reverts c86aa7bbfd5568ba8a82d3635d8f7b8a8e06fe54

The reverted commit caused ABBA deadlocks when file migration raced with
file eviction for specific hugetlbfs files.  This was discovered with a
modified version of the LTP move_pages12 test.

The purpose of the reverted patch was to close a long existing race
between hugetlbfs file truncation and page faults.  After more analysis
of the patch and impacted code, it was determined that i_mmap_rwsem can
not be used for all required synchronization.  Therefore, revert this
patch while working an another approach to the underlying issue.

Link: http://lkml.kernel.org/r/20190103235452.29335-1-mike.kravetz@oracle.com
Signed-off-by: Mike Kravetz <mike.kravetz@oracle.com>
Reported-by: Jan Stancek <jstancek@redhat.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Cc: "Aneesh Kumar K . V" <aneesh.kumar@linux.vnet.ibm.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: "Kirill A . Shutemov" <kirill.shutemov@linux.intel.com>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Prakash Sangappa <prakash.sangappa@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
18 months agomm: page_mapped: don't assume compound page is huge or THP
Jan Stancek [Tue, 8 Jan 2019 23:23:28 +0000 (15:23 -0800)]
mm: page_mapped: don't assume compound page is huge or THP

LTP proc01 testcase has been observed to rarely trigger crashes
on arm64:
    page_mapped+0x78/0xb4
    stable_page_flags+0x27c/0x338
    kpageflags_read+0xfc/0x164
    proc_reg_read+0x7c/0xb8
    __vfs_read+0x58/0x178
    vfs_read+0x90/0x14c
    SyS_read+0x60/0xc0

The issue is that page_mapped() assumes that if compound page is not
huge, then it must be THP.  But if this is 'normal' compound page
(COMPOUND_PAGE_DTOR), then following loop can keep running (for
HPAGE_PMD_NR iterations) until it tries to read from memory that isn't
mapped and triggers a panic:

        for (i = 0; i < hpage_nr_pages(page); i++) {
                if (atomic_read(&page[i]._mapcount) >= 0)
                        return true;
}

I could replicate this on x86 (v4.20-rc4-98-g60b548237fed) only
with a custom kernel module [1] which:
 - allocates compound page (PAGEC) of order 1
 - allocates 2 normal pages (COPY), which are initialized to 0xff (to
   satisfy _mapcount >= 0)
 - 2 PAGEC page structs are copied to address of first COPY page
 - second page of COPY is marked as not present
 - call to page_mapped(COPY) now triggers fault on access to 2nd COPY
   page at offset 0x30 (_mapcount)

[1] https://github.com/jstancek/reproducers/blob/master/kernel/page_mapped_crash/repro.c

Fix the loop to iterate for "1 << compound_order" pages.

Kirrill said "IIRC, sound subsystem can producuce custom mapped compound
pages".

Link: http://lkml.kernel.org/r/c440d69879e34209feba21e12d236d06bc0a25db.1543577156.git.jstancek@redhat.com
Fixes: e1534ae95004 ("mm: differentiate page_mapped() from page_mapcount() for compound pages")
Signed-off-by: Jan Stancek <jstancek@redhat.com>
Debugged-by: Laszlo Ersek <lersek@redhat.com>
Suggested-by: "Kirill A. Shutemov" <kirill@shutemov.name>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Andrea Arcangeli <aarcange@redhat.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
18 months agomm/memory.c: initialise mmu_notifier_range correctly
Matthew Wilcox [Tue, 8 Jan 2019 23:23:25 +0000 (15:23 -0800)]
mm/memory.c: initialise mmu_notifier_range correctly

One of the paths in follow_pte_pmd() initialised the mmu_notifier_range
incorrectly.

Link: http://lkml.kernel.org/r/20190103002126.GM6310@bombadil.infradead.org
Fixes: ac46d4f3c432 ("mm/mmu_notifier: use structure for invalidate_range_start/end calls v2")
Signed-off-by: Matthew Wilcox <willy@infradead.org>
Tested-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Jérôme Glisse <jglisse@redhat.com>
Cc: John Hubbard <jhubbard@nvidia.com>
Cc: Jan Kara <jack@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
18 months agotools/vm/page_owner: use page_owner_sort in the use example
Miles Chen [Tue, 8 Jan 2019 23:23:22 +0000 (15:23 -0800)]
tools/vm/page_owner: use page_owner_sort in the use example

The example in comment does not useable because the output binary is
named "page_owner_sort", not "sort".

Also add a reference to Documentation/vm/page_owner.rst

Link: http://lkml.kernel.org/r/1546515361-8317-1-git-send-email-miles.chen@mediatek.com
Signed-off-by: Miles Chen <miles.chen@mediatek.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
18 months agokasan: fix krealloc handling for tag-based mode
Andrey Konovalov [Tue, 8 Jan 2019 23:23:18 +0000 (15:23 -0800)]
kasan: fix krealloc handling for tag-based mode

Right now tag-based KASAN can retag the memory that is reallocated via
krealloc and return a differently tagged pointer even if the same slab
object gets used and no reallocated technically happens.

There are a few issues with this approach.  One is that krealloc callers
can't rely on comparing the return value with the passed argument to
check whether reallocation happened.  Another is that if a caller knows
that no reallocation happened, that it can access object memory through
the old pointer, which leads to false positives.  Look at
nf_ct_ext_add() to see an example.

Fix this by keeping the same tag if the memory don't actually gets
reallocated during krealloc.

Link: http://lkml.kernel.org/r/bb2a71d17ed072bcc528cbee46fcbd71a6da3be4.1546540962.git.andreyknvl@google.com
Signed-off-by: Andrey Konovalov <andreyknvl@google.com>
Cc: Andrey Ryabinin <aryabinin@virtuozzo.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Vincenzo Frascino <vincenzo.frascino@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
18 months agokasan: make tag based mode work with CONFIG_HARDENED_USERCOPY
Andrey Konovalov [Tue, 8 Jan 2019 23:23:15 +0000 (15:23 -0800)]
kasan: make tag based mode work with CONFIG_HARDENED_USERCOPY

With CONFIG_HARDENED_USERCOPY enabled __check_heap_object() compares and
then subtracts a potentially tagged pointer with a non-tagged address of
the page that this pointer belongs to, which leads to unexpected
behavior.

Untag the pointer in __check_heap_object() before doing any of these
operations.

Link: http://lkml.kernel.org/r/7e756a298d514c4482f52aea6151db34818d395d.1546540962.git.andreyknvl@google.com
Signed-off-by: Andrey Konovalov <andreyknvl@google.com>
Cc: Andrey Ryabinin <aryabinin@virtuozzo.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Vincenzo Frascino <vincenzo.frascino@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
18 months agokasan, arm64: use ARCH_SLAB_MINALIGN instead of manual aligning
Andrey Konovalov [Tue, 8 Jan 2019 23:23:11 +0000 (15:23 -0800)]
kasan, arm64: use ARCH_SLAB_MINALIGN instead of manual aligning

Instead of changing cache->align to be aligned to KASAN_SHADOW_SCALE_SIZE
in kasan_cache_create() we can reuse the ARCH_SLAB_MINALIGN macro.

Link: http://lkml.kernel.org/r/52ddd881916bcc153a9924c154daacde78522227.1546540962.git.andreyknvl@google.com
Signed-off-by: Andrey Konovalov <andreyknvl@google.com>
Suggested-by: Vincenzo Frascino <vincenzo.frascino@arm.com>
Cc: Andrey Ryabinin <aryabinin@virtuozzo.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Vincenzo Frascino <vincenzo.frascino@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
18 months agomm, memcg: fix reclaim deadlock with writeback
Michal Hocko [Tue, 8 Jan 2019 23:23:07 +0000 (15:23 -0800)]
mm, memcg: fix reclaim deadlock with writeback

Liu Bo has experienced a deadlock between memcg (legacy) reclaim and the
ext4 writeback

  task1:
    wait_on_page_bit+0x82/0xa0
    shrink_page_list+0x907/0x960
    shrink_inactive_list+0x2c7/0x680
    shrink_node_memcg+0x404/0x830
    shrink_node+0xd8/0x300
    do_try_to_free_pages+0x10d/0x330
    try_to_free_mem_cgroup_pages+0xd5/0x1b0
    try_charge+0x14d/0x720
    memcg_kmem_charge_memcg+0x3c/0xa0
    memcg_kmem_charge+0x7e/0xd0
    __alloc_pages_nodemask+0x178/0x260
    alloc_pages_current+0x95/0x140
    pte_alloc_one+0x17/0x40
    __pte_alloc+0x1e/0x110
    alloc_set_pte+0x5fe/0xc20
    do_fault+0x103/0x970
    handle_mm_fault+0x61e/0xd10
    __do_page_fault+0x252/0x4d0
    do_page_fault+0x30/0x80
    page_fault+0x28/0x30

  task2:
    __lock_page+0x86/0xa0
    mpage_prepare_extent_to_map+0x2e7/0x310 [ext4]
    ext4_writepages+0x479/0xd60
    do_writepages+0x1e/0x30
    __writeback_single_inode+0x45/0x320
    writeback_sb_inodes+0x272/0x600
    __writeback_inodes_wb+0x92/0xc0
    wb_writeback+0x268/0x300
    wb_workfn+0xb4/0x390
    process_one_work+0x189/0x420
    worker_thread+0x4e/0x4b0
    kthread+0xe6/0x100
    ret_from_fork+0x41/0x50

He adds
 "task1 is waiting for the PageWriteback bit of the page that task2 has
  collected in mpd->io_submit->io_bio, and tasks2 is waiting for the
  LOCKED bit the page which tasks1 has locked"

More precisely task1 is handling a page fault and it has a page locked
while it charges a new page table to a memcg.  That in turn hits a
memory limit reclaim and the memcg reclaim for legacy controller is
waiting on the writeback but that is never going to finish because the
writeback itself is waiting for the page locked in the #PF path.  So
this is essentially ABBA deadlock:

                                        lock_page(A)
                                        SetPageWriteback(A)
                                        unlock_page(A)
  lock_page(B)
                                        lock_page(B)
  pte_alloc_pne
    shrink_page_list
      wait_on_page_writeback(A)
                                        SetPageWriteback(B)
                                        unlock_page(B)

                                        # flush A, B to clear the writeback

This accumulating of more pages to flush is used by several filesystems
to generate a more optimal IO patterns.

Waiting for the writeback in legacy memcg controller is a workaround for
pre-mature OOM killer invocations because there is no dirty IO
throttling available for the controller.  There is no easy way around
that unfortunately.  Therefore fix this specific issue by pre-allocating
the page table outside of the page lock.  We have that handy
infrastructure for that already so simply reuse the fault-around pattern
which already does this.

There are probably other hidden __GFP_ACCOUNT | GFP_KERNEL allocations
from under a fs page locked but they should be really rare.  I am not
aware of a better solution unfortunately.

[akpm@linux-foundation.org: fix mm/memory.c:__do_fault()]
[akpm@linux-foundation.org: coding-style fixes]
[mhocko@kernel.org: enhance comment, per Johannes]
Link: http://lkml.kernel.org/r/20181214084948.GA5624@dhcp22.suse.cz
Link: http://lkml.kernel.org/r/20181213092221.27270-1-mhocko@kernel.org
Fixes: c3b94f44fcb0 ("memcg: further prevent OOM with too many dirty pages")
Signed-off-by: Michal Hocko <mhocko@suse.com>
Reported-by: Liu Bo <bo.liu@linux.alibaba.com>
Debugged-by: Liu Bo <bo.liu@linux.alibaba.com>
Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Liu Bo <bo.liu@linux.alibaba.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Theodore Ts'o <tytso@mit.edu>
Cc: Vladimir Davydov <vdavydov.dev@gmail.com>
Cc: Shakeel Butt <shakeelb@google.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
18 months agomm/usercopy.c: no check page span for stack objects
Qian Cai [Tue, 8 Jan 2019 23:23:04 +0000 (15:23 -0800)]
mm/usercopy.c: no check page span for stack objects

It is easy to trigger this with CONFIG_HARDENED_USERCOPY_PAGESPAN=y,

  usercopy: Kernel memory overwrite attempt detected to spans multiple pages (offset 0, size 23)!
  kernel BUG at mm/usercopy.c:102!

For example,

print_worker_info
char name[WQ_NAME_LEN] = { };
char desc[WORKER_DESC_LEN] = { };
  probe_kernel_read(name, wq->name, sizeof(name) - 1);
  probe_kernel_read(desc, worker->desc, sizeof(desc) - 1);
    __copy_from_user_inatomic
      check_object_size
        check_heap_object
          check_page_span

This is because on-stack variables could cross PAGE_SIZE boundary, and
failed this check,

if (likely(((unsigned long)ptr & (unsigned long)PAGE_MASK) ==
   ((unsigned long)end & (unsigned long)PAGE_MASK)))

ptr = FFFF889007D7EFF8
end = FFFF889007D7F00E

Hence, fix it by checking if it is a stack object first.

[keescook@chromium.org: improve comments after reorder]
Link: http://lkml.kernel.org/r/20190103165151.GA32845@beast
Link: http://lkml.kernel.org/r/20181231030254.99441-1-cai@lca.pw
Signed-off-by: Qian Cai <cai@lca.pw>
Signed-off-by: Kees Cook <keescook@chromium.org>
Acked-by: Kees Cook <keescook@chromium.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
18 months agoslab: alien caches must not be initialized if the allocation of the alien cache failed
Christoph Lameter [Tue, 8 Jan 2019 23:23:00 +0000 (15:23 -0800)]
slab: alien caches must not be initialized if the allocation of the alien cache failed

Callers of __alloc_alien() check for NULL.  We must do the same check in
__alloc_alien_cache to avoid NULL pointer dereferences on allocation
failures.

Link: http://lkml.kernel.org/r/010001680f42f192-82b4e12e-1565-4ee0-ae1f-1e98974906aa-000000@email.amazonses.com
Fixes: 49dfc304ba241 ("slab: use the lock on alien_cache, instead of the lock on array_cache")
Fixes: c8522a3a5832b ("Slab: introduce alloc_alien")
Signed-off-by: Christoph Lameter <cl@linux.com>
Reported-by: syzbot+d6ed4ec679652b4fd4e4@syzkaller.appspotmail.com
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Pekka Enberg <penberg@kernel.org>
Cc: David Rientjes <rientjes@google.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
18 months agofork, memcg: fix cached_stacks case
Shakeel Butt [Tue, 8 Jan 2019 23:22:57 +0000 (15:22 -0800)]
fork, memcg: fix cached_stacks case

Commit 5eed6f1dff87 ("fork,memcg: fix crash in free_thread_stack on
memcg charge fail") fixes a crash caused due to failed memcg charge of
the kernel stack.  However the fix misses the cached_stacks case which
this patch fixes.  So, the same crash can happen if the memcg charge of
a cached stack is failed.

Link: http://lkml.kernel.org/r/20190102180145.57406-1-shakeelb@google.com
Fixes: 5eed6f1dff87 ("fork,memcg: fix crash in free_thread_stack on memcg charge fail")
Signed-off-by: Shakeel Butt <shakeelb@google.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Rik van Riel <riel@surriel.com>
Cc: Rik van Riel <riel@surriel.com>
Cc: Roman Gushchin <guro@fb.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
18 months agozram: idle writeback fixes and cleanup
Minchan Kim [Tue, 8 Jan 2019 23:22:53 +0000 (15:22 -0800)]
zram: idle writeback fixes and cleanup

This patch includes some fixes and cleanup for idle-page writeback.

1. writeback_limit interface

Now writeback_limit interface is rather conusing.  For example, once
writeback limit budget is exausted, admin can see 0 from
/sys/block/zramX/writeback_limit which is same semantic with disable
writeback_limit at this moment.  IOW, admin cannot tell that zero came
from disable writeback limit or exausted writeback limit.

To make the interface clear, let's sepatate enable of writeback limit to
another knob - /sys/block/zram0/writeback_limit_enable

* before:
  while true :
    # to re-enable writeback limit once previous one is used up
    echo 0 > /sys/block/zram0/writeback_limit
    echo $((200<<20)) > /sys/block/zram0/writeback_limit
    ..
    .. # used up the writeback limit budget

* new
  # To enable writeback limit, from the beginning, admin should
  # enable it.
  echo $((200<<20)) > /sys/block/zram0/writeback_limit
  echo 1 > /sys/block/zram/0/writeback_limit_enable
  while true :
    echo $((200<<20)) > /sys/block/zram0/writeback_limit
    ..
    .. # used up the writeback limit budget

It's much strightforward.

2. fix condition check idle/huge writeback mode check

The mode in writeback_store is not bit opeartion any more so no need to
use bit operations.  Furthermore, current condition check is broken in
that it does writeback every pages regardless of huge/idle.

3. clean up idle_store

No need to use goto.

[minchan@kernel.org: missed spin_lock_init]
Link: http://lkml.kernel.org/r/20190103001601.GA255139@google.com
Link: http://lkml.kernel.org/r/20181224033529.19450-1-minchan@kernel.org
Signed-off-by: Minchan Kim <minchan@kernel.org>
Suggested-by: John Dias <joaodias@google.com>
Cc: Sergey Senozhatsky <sergey.senozhatsky.work@gmail.com>
Cc: John Dias <joaodias@google.com>
Cc: Srinivas Paladugu <srnvs@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
18 months agodrm/dp_mst: Add __must_check to drm_dp_mst_topology_mgr_resume()
Lyude Paul [Tue, 8 Jan 2019 21:11:29 +0000 (16:11 -0500)]
drm/dp_mst: Add __must_check to drm_dp_mst_topology_mgr_resume()

Since I've had to fix two cases of drivers not checking the return code
from this function, let's make the compiler complain so this doesn't
come up again in the future.

Changes since v1:
* Remove unneeded __must_check in function declaration - danvet

Signed-off-by: Lyude Paul <lyude@redhat.com>
Cc: Jerry Zuo <Jerry.Zuo@amd.com>
Reviewed-by: Daniel Vetter <daniel@ffwll.ch>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190108211133.32564-4-lyude@redhat.com
18 months agodrm/amdgpu: Don't fail resume process if resuming atomic state fails
Lyude Paul [Tue, 8 Jan 2019 21:11:28 +0000 (16:11 -0500)]
drm/amdgpu: Don't fail resume process if resuming atomic state fails

This is an ugly one unfortunately. Currently, all DRM drivers supporting
atomic modesetting will save the state that userspace had set before
suspending, then attempt to restore that state on resume. This probably
worked very well at one point, like many other things, until DP MST came
into the picture. While it's easy to restore state on normal display
connectors that were disconnected during suspend regardless of their
state post-resume, this can't really be done with MST because of the
fact that setting up a downstream sink requires performing sideband
transactions between the source and the MST hub, sending out the ACT
packets, etc.

Because of this, there isn't really a guarantee that we can restore the
atomic state we had before suspend once we've resumed. This sucks pretty
bad, but so far I haven't run into any compositors that this actually
causes serious issues with. Most compositors will notice the hotplug we
send afterwards, and then reprobe state.

Since nouveau and i915 also don't fail the suspend/resume process due to
failing to restore the atomic state, let's make amdgpu match this
behavior. Better to resume the GPU properly, then to stop the process
half way because of a potentially unavoidable atomic commit failure.

Eventually, we'll have a real fix for this problem on the DRM level. But
we've got some more important low-hanging fruit to deal with first.

Signed-off-by: Lyude Paul <lyude@redhat.com>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Cc: Jerry Zuo <Jerry.Zuo@amd.com>
Cc: <stable@vger.kernel.org> # v4.15+
Link: https://patchwork.freedesktop.org/patch/msgid/20190108211133.32564-3-lyude@redhat.com
18 months agodrm/amdgpu: Don't ignore rc from drm_dp_mst_topology_mgr_resume()
Lyude Paul [Tue, 8 Jan 2019 21:11:27 +0000 (16:11 -0500)]
drm/amdgpu: Don't ignore rc from drm_dp_mst_topology_mgr_resume()

drm_dp_mst_topology_mgr_resume() returns whether or not it managed to
find the topology in question after a suspend resume cycle, and the
driver is supposed to check this value and disable MST accordingly if
it's gone-in addition to sending a hotplug in order to notify userspace
that something changed during suspend.

Currently, amdgpu just makes the mistake of ignoring the return code
from drm_dp_mst_topology_mgr_resume() which means that if a topology was
removed in suspend, amdgpu never notices and assumes it's still
connected which leads to all sorts of problems.

So, fix this by actually checking the rc from
drm_dp_mst_topology_mgr_resume(). Also, reformat the rest of the
function while we're at it to fix the over-indenting.

Signed-off-by: Lyude Paul <lyude@redhat.com>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Cc: Jerry Zuo <Jerry.Zuo@amd.com>
Cc: <stable@vger.kernel.org> # v4.15+
Link: https://patchwork.freedesktop.org/patch/msgid/20190108211133.32564-2-lyude@redhat.com
18 months agoALSA: usb-audio: fix CM6206 register definitions
Amadeusz Sławiński [Tue, 8 Jan 2019 20:03:11 +0000 (21:03 +0100)]
ALSA: usb-audio: fix CM6206 register definitions

fix typo after a recent commit causing headphones to have no sound

Fixes: ad43d528a7ac (ALSA: usb-audio: Define registers for CM6206)
Signed-off-by: Amadeusz Sławiński <amade@asmblr.net>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
18 months agodrm/amdgpu: validate user GEM object size
Yu Zhao [Mon, 7 Jan 2019 22:51:15 +0000 (15:51 -0700)]
drm/amdgpu: validate user GEM object size

When creating frame buffer, userspace may request to attach to a
previously allocated GEM object that is smaller than what GPU
requires. Validation must be done to prevent out-of-bound DMA,
otherwise it could be exploited to reveal sensitive data.

This fix is not done in a common code path because individual
driver might have different requirement.

Cc: stable@vger.kernel.org # v4.2+
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Signed-off-by: Yu Zhao <yuzhao@google.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
18 months agodrm/amdgpu: validate user pitch alignment
Yu Zhao [Mon, 7 Jan 2019 22:51:14 +0000 (15:51 -0700)]
drm/amdgpu: validate user pitch alignment

Userspace may request pitch alignment that is not supported by GPU.
Some requests 32, but GPU ignores it and uses default 64 when cpp is
4. If GEM object is allocated based on the smaller alignment, GPU
DMA will go out of bound.

Cc: stable@vger.kernel.org # v4.2+
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Signed-off-by: Yu Zhao <yuzhao@google.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
18 months agodrm/amd/powerplay: drop the unnecessary uclk hard min setting
Evan Quan [Tue, 8 Jan 2019 04:22:24 +0000 (12:22 +0800)]
drm/amd/powerplay: drop the unnecessary uclk hard min setting

Since soft min setting is enough. Hard min setting is redundant.

Reported-by: Likun Gao <Likun.Gao@amd.com>
Signed-off-by: Evan Quan <evan.quan@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Likun Gao <Likun.Gao@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
18 months agodrm/amd/powerplay: avoid possible buffer overflow
Evan Quan [Tue, 8 Jan 2019 02:33:35 +0000 (10:33 +0800)]
drm/amd/powerplay: avoid possible buffer overflow

Make sure the clock level enforced is within the allowed
ranges.

Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Likun Gao <Likun.Gao@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
18 months agodrm/amd/powerplay: create pp_od_clk_voltage device file under OD support
Evan Quan [Mon, 7 Jan 2019 10:56:14 +0000 (18:56 +0800)]
drm/amd/powerplay: create pp_od_clk_voltage device file under OD support

Since pp_od_clk_voltage device file is for OD related sysfs operations.

Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
18 months agodrm/amd/powerplay: update OD support flag for SKU with no OD capabilities
Evan Quan [Mon, 7 Jan 2019 10:38:30 +0000 (18:38 +0800)]
drm/amd/powerplay: update OD support flag for SKU with no OD capabilities

For those ASICs with no overdrive capabilities, the OD support flag
will be reset.

Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
18 months agofork: record start_time late
David Herrmann [Tue, 8 Jan 2019 12:58:52 +0000 (13:58 +0100)]
fork: record start_time late

This changes the fork(2) syscall to record the process start_time after
initializing the basic task structure but still before making the new
process visible to user-space.

Technically, we could record the start_time anytime during fork(2).  But
this might lead to scenarios where a start_time is recorded long before
a process becomes visible to user-space.  For instance, with
userfaultfd(2) and TLS, user-space can delay the execution of fork(2)
for an indefinite amount of time (and will, if this causes network
access, or similar).

By recording the start_time late, it much closer reflects the point in
time where the process becomes live and can be observed by other
processes.

Lastly, this makes it much harder for user-space to predict and control
the start_time they get assigned.  Previously, user-space could fork a
process and stall it in copy_thread_tls() before its pid is allocated,
but after its start_time is recorded.  This can be misused to later-on
cycle through PIDs and resume the stalled fork(2) yielding a process
that has the same pid and start_time as a process that existed before.
This can be used to circumvent security systems that identify processes
by their pid+start_time combination.

Even though user-space was always aware that start_time recording is
flaky (but several projects are known to still rely on start_time-based
identification), changing the start_time to be recorded late will help
mitigate existing attacks and make it much harder for user-space to
control the start_time a process gets assigned.

Reported-by: Jann Horn <jannh@google.com>
Signed-off-by: Tom Gundersen <teg@jklm.no>
Signed-off-by: David Herrmann <dh.herrmann@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
18 months agovfio/type1: Fix unmap overflow off-by-one
Alex Williamson [Tue, 8 Jan 2019 05:13:22 +0000 (22:13 -0700)]
vfio/type1: Fix unmap overflow off-by-one

The below referenced commit adds a test for integer overflow, but in
doing so prevents the unmap ioctl from ever including the last page of
the address space.  Subtract one to compare to the last address of the
unmap to avoid the overflow and wrap-around.

Fixes: 71a7d3d78e3c ("vfio/type1: silence integer overflow warning")
Link: https://bugzilla.redhat.com/show_bug.cgi?id=1662291
Cc: stable@vger.kernel.org # v4.15+
Reported-by: Pei Zhang <pezhang@redhat.com>
Debugged-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Tested-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
18 months agoirqchip/csky: fixup handle_irq_perbit break irq
Guo Ren [Tue, 8 Jan 2019 12:49:24 +0000 (20:49 +0800)]
irqchip/csky: fixup handle_irq_perbit break irq

The handle_irq_perbit function loop every bit in hwirq local variable.

handle_irq_perbit(hwirq) {
  for_everyt_bit_in(hwirq) {
handle_domain_irq()
->irq_exit()
->invoke_softirq()
->__do_softirq()
->local_irq_enable() // Here will cause new interrupt.
  }
}

When new interrupt coming at local_irq_enable, it will finish another
interrupt handler and pull down the interrupt source. But hwirq is the
local variable for handle_irq_perbit(), it can't get new interrupt
controller pending reg status. So we need update hwirq with pending reg
in every loop.

Also change write_relax to writel could prevent stw from fast retire.
When local_irq is enabled, intc regs is really set-in.

Signed-off-by: Guo Ren <ren_guo@c-sky.com>
Cc: Lu Baoquan <lu.baoquan@intellif.com>
18 months agocsky: fixup compile error with pte_alloc
Guo Ren [Tue, 8 Jan 2019 12:31:43 +0000 (20:31 +0800)]
csky: fixup compile error with pte_alloc

Commit: 4cf58924951e remove the address argument of pte_alloc without
modify csky related code. linux-5.0-rc1 compile failed with csky.

Remove the unnecessary address testing in pte_alloc().

Signed-off-by: Guo Ren <ren_guo@c-sky.com>
Cc: Joel Fernandes (Google) <joel@joelfernandes.org>
Cc: Guenter Roeck <linux@roeck-us.net>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
18 months agovfio/pci: set TRACE_INCLUDE_PATH to fix the build error
Masahiro Yamada [Tue, 8 Jan 2019 03:08:03 +0000 (12:08 +0900)]
vfio/pci: set TRACE_INCLUDE_PATH to fix the build error

drivers/vfio/pci/vfio_pci_nvlink2.c cannot be compiled for in-tree
building.

    CC      drivers/vfio/pci/vfio_pci_nvlink2.o
  In file included from drivers/vfio/pci/trace.h:102,
                   from drivers/vfio/pci/vfio_pci_nvlink2.c:29:
  ./include/trace/define_trace.h:89:42: fatal error: ./trace.h: No such file or directory
   #include TRACE_INCLUDE(TRACE_INCLUDE_FILE)
                                          ^
  compilation terminated.
  make[1]: *** [scripts/Makefile.build;277: drivers/vfio/pci/vfio_pci_nvlink2.o] Error 1

To fix the build error, let's tell include/trace/define_trace.h the
location of drivers/vfio/pci/trace.h

Fixes: 7f92891778df ("vfio_pci: Add NVIDIA GV100GL [Tesla V100 SXM2] subdriver")
Reported-by: Laura Abbott <labbott@redhat.com>
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
18 months agocsky: fixup CACHEV1 store instruction fast retire
Guo Ren [Tue, 8 Jan 2019 12:17:49 +0000 (20:17 +0800)]
csky: fixup CACHEV1 store instruction fast retire

For I/O access, 810/807 store instruction fast retire will cause wrong
primitive. For example:

stw (clear interrupt source)
stw (unmask interrupt controller)
enable interrupt

stw is fast retire instruction. When PC is run at enable interrupt
stage, the clear interrupt source hasn't finished. It will cause another
wrong irq-enter.

So use mb() to prevent above.

Signed-off-by: Guo Ren <ren_guo@c-sky.com>
Cc: Lu Baoquan <lu.baoquan@intellif.com>
18 months agoPCI: amlogic: Fix build failure due to missing gpio header
Corentin Labbe [Mon, 7 Jan 2019 17:22:23 +0000 (17:22 +0000)]
PCI: amlogic: Fix build failure due to missing gpio header

Building the driver when GPIOLIB=n is not selected is causing the following
compilation failure:

  drivers/pci/controller/dwc/pci-meson.c: In function 'meson_pcie_assert_reset':
  drivers/pci/controller/dwc/pci-meson.c:290:2: error: implicit declaration of function 'gpiod_set_value_cansleep'; did you mean 'gpio_set_value_cansleep'? [-Werror=implicit-function-declaration]
    gpiod_set_value_cansleep(mp->reset_gpio, 0);
    ^~~~~~~~~~~~~~~~~~~~~~~~
    gpio_set_value_cansleep
  drivers/pci/controller/dwc/pci-meson.c: In function 'meson_pcie_probe':
  drivers/pci/controller/dwc/pci-meson.c:540:19: error: implicit declaration of function 'devm_gpiod_get'; did you mean 'devm_gpio_free'? [-Werror=implicit-function-declaration]
    mp->reset_gpio = devm_gpiod_get(dev, "reset", GPIOD_OUT_LOW);
     ^~~~~~~~~~~~~~
     devm_gpio_free
  drivers/pci/controller/dwc/pci-meson.c:540:48: error: 'GPIOD_OUT_LOW' undeclared (first use in this function); did you mean 'GPIOF_INIT_LOW'?
    mp->reset_gpio = devm_gpiod_get(dev, "reset", GPIOD_OUT_LOW);
  ^~~~~~~~~~~~~
  GPIOF_INIT_LOW

Add the missing linux/gpio/consumer.h header to fix it.

Fixes: 9c0ef6d34fdb ("PCI: amlogic: Add the Amlogic Meson PCIe controller driver")
Signed-off-by: Corentin Labbe <clabbe@baylibre.com>
[lorenzo.pieralisi@arm.com: commit log]
Signed-off-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
18 months agocsky: fixup relocation error with 807 & 860
Guo Ren [Tue, 8 Jan 2019 11:52:22 +0000 (19:52 +0800)]
csky: fixup relocation error with 807 & 860

810 doesn't support jsri instruction and csky-as will leave
jsri + nop for relocation. Module-probe need replace them with
lrw + jsr.

Signed-off-by: Guo Ren <ren_guo@c-sky.com>
Cc: Hui Kai <huikai@acoinfo.com>
18 months agomtd: rawnand: qcom: fix memory corruption that causes panic
Christian Lamparter [Sun, 23 Dec 2018 00:31:26 +0000 (01:31 +0100)]
mtd: rawnand: qcom: fix memory corruption that causes panic

This patch fixes a memory corruption that occurred in the
qcom-nandc driver since it was converted to nand_scan().

On boot, an affected device will panic from a NPE at a weird place:
| Unable to handle kernel NULL pointer dereference at virtual address 0
| pgd = (ptrval)
| [00000000] *pgd=00000000
| Internal error: Oops: 80000005 [#1] SMP ARM
| CPU: 0 PID: 1 Comm: swapper/0 Not tainted 4.19.9 #0
| Hardware name: Generic DT based system
| PC is at   (null)
| LR is at nand_block_isbad+0x90/0xa4
| pc : [<00000000>]    lr : [<c0592240>]    psr: 80000013
| sp : cf839d40  ip : 00000000  fp : cfae9e20
| r10: cf815810  r9 : 00000000  r8 : 00000000
| r7 : 00000000  r6 : 00000000  r5 : 00000001  r4 : cf815810
| r3 : 00000000  r2 : cfae9810  r1 : ffffffff  r0 : cf815810
| Flags: Nzcv  IRQs on  FIQs on  Mode SVC_32  ISA ARM  Segment none
| Control: 10c5387d  Table: 8020406a  DAC: 00000051
| Process swapper/0 (pid: 1, stack limit = 0x(ptrval))
| [<c0592240>] (nand_block_isbad) from [<c0580a94>]
| [<c0580a94>] (allocate_partition) from [<c05811e4>]
| [<c05811e4>] (add_mtd_partitions) from [<c0581164>]
| [<c0581164>] (parse_mtd_partitions) from [<c057def4>]
| [<c057def4>] (mtd_device_parse_register) from [<c059d274>]
| [<c059d274>] (qcom_nandc_probe) from [<c0567f00>]

The problem is that the nand_scan()'s qcom_nand_attach_chip callback
is updating the nandc->max_cwperpage from 1 to 4. This causes the
sg_init_table of clear_bam_transaction() in the driver's
qcom_nandc_block_bad() to memset much more than what was initially
allocated by alloc_bam_transaction().

This patch restores the old behavior by reallocating the shared bam
transaction alloc_bam_transaction() after the chip was identified,
but before mtd_device_parse_register() (which is an alias for
mtd_device_register() - see panic) gets called. This fixes the
corruption and the driver is working again.

Cc: stable@vger.kernel.org
Fixes: 6a3cec64f18c ("mtd: rawnand: qcom: convert driver to nand_scan()")
Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
Acked-by: Miquel Raynal <miquel.raynal@bootlin.com>
Signed-off-by: Boris Brezillon <bbrezillon@kernel.org>
18 months agocpufreq: check if policy is inactive early in __cpufreq_get()
Sudeep Holla [Mon, 7 Jan 2019 18:51:53 +0000 (18:51 +0000)]
cpufreq: check if policy is inactive early in __cpufreq_get()

cpuinfo_cur_freq gets current CPU frequency as detected by hardware
while scaling_cur_freq last known CPU frequency. Some platforms may not
allow checking the CPU frequency of an offline CPU or the associated
resources may have been released via cpufreq_exit when the CPU gets
offlined, in which case the policy would have been invalidated already.
If we attempt to get current frequency from the hardware, it may result
in hang or crash.

For example on Juno, I see:

Unable to handle kernel NULL pointer dereference at virtual address 0000000000000188
[0000000000000188] pgd=0000000000000000
Internal error: Oops: 96000004 [#1] PREEMPT SMP
Modules linked in:
CPU: 5 PID: 4202 Comm: cat Not tainted 4.20.0-08251-ga0f2c0318a15-dirty #87
Hardware name: ARM LTD ARM Juno Development Platform/ARM Juno Development Platform
pstate: 40000005 (nZcv daif -PAN -UAO)
pc : scmi_cpufreq_get_rate+0x34/0xb0
lr : scmi_cpufreq_get_rate+0x34/0xb0
Call trace:
 scmi_cpufreq_get_rate+0x34/0xb0
 __cpufreq_get+0x34/0xc0
 show_cpuinfo_cur_freq+0x24/0x78
 show+0x40/0x60
 sysfs_kf_seq_show+0xc0/0x148
 kernfs_seq_show+0x44/0x50
 seq_read+0xd4/0x480
 kernfs_fop_read+0x15c/0x208
 __vfs_read+0x60/0x188
 vfs_read+0x94/0x150
 ksys_read+0x6c/0xd8
 __arm64_sys_read+0x24/0x30
 el0_svc_common+0x78/0x100
 el0_svc_handler+0x38/0x78
 el0_svc+0x8/0xc
---[ end trace 3d1024e58f77f6b2 ]---

So fix the issue by checking if the policy is invalid early in
__cpufreq_get before attempting to get the current frequency.

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
18 months agoALSA: cs46xx: Potential NULL dereference in probe
Dan Carpenter [Tue, 8 Jan 2019 07:43:30 +0000 (10:43 +0300)]
ALSA: cs46xx: Potential NULL dereference in probe

The "chip->dsp_spos_instance" can be NULL on some of the ealier error
paths in snd_cs46xx_create().

Reported-by: "Yavuz, Tuba" <tuba@ece.ufl.edu>
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
19 months agoFix a handful of audit-related issue
Palmer Dabbelt [Mon, 7 Jan 2019 16:45:47 +0000 (08:45 -0800)]
Fix a handful of audit-related issue

This is sort of a mix between a new feature and a bug fix.  I've managed
to screw up merging this patch set a handful of times but I think it's
OK this time around.  The main new feature here is audit support for
RISC-V, with some fixes to audit-related bugs that cropped up along the
way:

* The addition of NR_syscalls into unistd.h, which is necessary for
  CONFIG_FTRACE_SYSCALLS.
* The definition of CREATE_TRACE_POINTS so
  __tracepoint_sys_{enter,exit} get defined.
* A fix for trace_sys_exit() so we can enable
  CONFIG_HAVE_SYSCALL_TRACEPOINTS.

19 months agoriscv: add HAVE_SYSCALL_TRACEPOINTS to Kconfig
David Abdurachmanov [Thu, 6 Dec 2018 15:26:35 +0000 (16:26 +0100)]
riscv: add HAVE_SYSCALL_TRACEPOINTS to Kconfig

I looked into Documentation/trace/ftrace-design.rst and, I think,
we check all the boxes needed for HAVE_SYSCALL_TRACEPOINTS.

Signed-off-by: David Abdurachmanov <david.abdurachmanov@gmail.com>
Signed-off-by: Palmer Dabbelt <palmer@sifive.com>
19 months agoriscv: fix trace_sys_exit hook
David Abdurachmanov [Thu, 6 Dec 2018 15:26:34 +0000 (16:26 +0100)]
riscv: fix trace_sys_exit hook

Fix compilation error.

Signed-off-by: David Abdurachmanov <david.abdurachmanov@gmail.com>
Signed-off-by: Palmer Dabbelt <palmer@sifive.com>
19 months agoriscv: define CREATE_TRACE_POINTS in ptrace.c
David Abdurachmanov [Mon, 10 Dec 2018 20:43:55 +0000 (21:43 +0100)]
riscv: define CREATE_TRACE_POINTS in ptrace.c

Define CREATE_TRACE_POINTS in order to create functions and structures
for the trace events. This is needed if HAVE_SYSCALL_TRACEPOINTS and
CONFIG_FTRACE_SYSCALLS are enabled, otherwise we get linking errors:

[..]
  MODPOST vmlinux.o
kernel/trace/trace_syscalls.o: In function `.L0 ':
trace_syscalls.c:(.text+0x1152): undefined reference to `__tracepoint_sys_enter'
trace_syscalls.c:(.text+0x126c): undefined reference to `__tracepoint_sys_enter'
trace_syscalls.c:(.text+0x1328): undefined reference to `__tracepoint_sys_enter'
trace_syscalls.c:(.text+0x14aa): undefined reference to `__tracepoint_sys_enter'
trace_syscalls.c:(.text+0x1684): undefined reference to `__tracepoint_sys_exit'
trace_syscalls.c:(.text+0x17a0): undefined reference to `__tracepoint_sys_exit'
trace_syscalls.c:(.text+0x185c): undefined reference to `__tracepoint_sys_exit'
trace_syscalls.c:(.text+0x19de): undefined reference to `__tracepoint_sys_exit'
arch/riscv/kernel/ptrace.o: In function `.L0 ':
ptrace.c:(.text+0x4dc): undefined reference to `__tracepoint_sys_enter'
ptrace.c:(.text+0x632): undefined reference to `__tracepoint_sys_exit'
make: *** [Makefile:1036: vmlinux] Error 1

Signed-off-by: David Abdurachmanov <david.abdurachmanov@gmail.com>
Fixes: b78002b395b4 ("riscv: add HAVE_SYSCALL_TRACEPOINTS to Kconfig")
Signed-off-by: Palmer Dabbelt <palmer@sifive.com>
19 months agoriscv: define NR_syscalls in unistd.h
David Abdurachmanov [Mon, 10 Dec 2018 20:43:54 +0000 (21:43 +0100)]
riscv: define NR_syscalls in unistd.h

This macro is used by kernel/trace/{trace.h,trace_syscalls.c} if we
have CONFIG_FTRACE_SYSCALLS enabled.

Signed-off-by: David Abdurachmanov <david.abdurachmanov@gmail.com>
Fixes: b78002b395b4 ("riscv: add HAVE_SYSCALL_TRACEPOINTS to Kconfig")
Signed-off-by: Palmer Dabbelt <palmer@sifive.com>
19 months agoriscv: audit: add audit hook in do_syscall_trace_enter/exit()
David Abdurachmanov [Mon, 29 Oct 2018 10:48:54 +0000 (11:48 +0100)]
riscv: audit: add audit hook in do_syscall_trace_enter/exit()

This patch adds auditing functions on entry to and exit from every system
call invocation.

Signed-off-by: David Abdurachmanov <david.abdurachmanov@gmail.com>
Signed-off-by: Palmer Dabbelt <palmer@sifive.com>
19 months agoriscv: add audit support
David Abdurachmanov [Mon, 29 Oct 2018 10:48:53 +0000 (11:48 +0100)]
riscv: add audit support

On RISC-V (riscv) audit is supported through generic lib/audit.c.
The patch adds required arch specific definitions.

Signed-off-by: David Abdurachmanov <david.abdurachmanov@gmail.com>
Signed-off-by: Palmer Dabbelt <palmer@sifive.com>
19 months agoRISC-V: Support MODULE_SECTIONS mechanism on RV32
Zong Li [Fri, 7 Dec 2018 09:02:16 +0000 (17:02 +0800)]
RISC-V: Support MODULE_SECTIONS mechanism on RV32

This patch supports dynamic generate got and plt sections mechanism on
rv32. It contains the modification as follows:
 - Always enable MODULE_SECTIONS (both rv64 and rv32)
 - Change the fixed size type.

This patch had been tested by following modules:

btrfs 6795991 0 - Live 0xa544b000
test_static_keys 17304 0 - Live 0xa28be000
zstd_compress 1198986 1 btrfs, Live 0xa2a25000
zstd_decompress 608112 1 btrfs, Live 0xa24e7000
lzo 8787 0 - Live 0xa2049000
xor 27461 1 btrfs, Live 0xa2041000
zram 78849 0 - Live 0xa2276000
netdevsim 55909 0 - Live 0xa202d000
tun 211534 0 - Live 0xa21b5000
fuse 566049 0 - Live 0xa25fb000
nfs_layout_flexfiles 192597 0 - Live 0xa229b000
ramoops 74895 0 - Live 0xa2019000
xfs 3973221 0 - Live 0xa507f000
libcrc32c 3053 2 btrfs,xfs, Live 0xa34af000
lzo_compress 17302 2 btrfs,lzo, Live 0xa347d000
lzo_decompress 7178 2 btrfs,lzo, Live 0xa3451000
raid6_pq 142086 1 btrfs, Live 0xa33a4000
reed_solomon 31022 1 ramoops, Live 0xa31eb000
test_bitmap 3734 0 - Live 0xa31af000
test_bpf 1588736 0 - Live 0xa2c11000
test_kmod 41161 0 - Live 0xa29f8000
test_module 1356 0 - Live 0xa299e000
test_printf 6024 0 [permanent], Live 0xa2971000
test_static_key_base 5797 1 test_static_keys, Live 0xa2931000
test_user_copy 4382 0 - Live 0xa28c9000
xxhash 70501 2 zstd_compress,zstd_decompress, Live 0xa2055000

Signed-off-by: Zong Li <zong@andestech.com>
Signed-off-by: Palmer Dabbelt <palmer@sifive.com>
19 months agoMAINTAINERS: SiFive drivers: add myself as a SiFive driver maintainer
Paul Walmsley [Tue, 18 Dec 2018 00:09:59 +0000 (16:09 -0800)]
MAINTAINERS: SiFive drivers: add myself as a SiFive driver maintainer

I'll be helping Palmer review drivers for SiFive-specific IP blocks,
so add myself to the MAINTAINERS file.

Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Palmer Dabbelt <palmer@sifive.com>
Cc: linux-kernel@vger.kernel.org
Cc: linux-riscv@lists.infradead.org
Signed-off-by: Paul Walmsley <paul.walmsley@sifive.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Signed-off-by: Palmer Dabbelt <palmer@sifive.com>
19 months agoMAINTAINERS: SiFive drivers: change the git tree to a SiFive git tree
Paul Walmsley [Tue, 18 Dec 2018 00:09:58 +0000 (16:09 -0800)]
MAINTAINERS: SiFive drivers: change the git tree to a SiFive git tree

Update the git tree URL for drivers for SiFive-related IP blocks to
point to a SiFive-managed URL.

Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Palmer Dabbelt <palmer@sifive.com>
Cc: linux-kernel@vger.kernel.org
Cc: linux-riscv@lists.infradead.org
Signed-off-by: Paul Walmsley <paul.walmsley@sifive.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Signed-off-by: Palmer Dabbelt <palmer@sifive.com>
19 months agoriscv: don't stop itself in smp_send_stop
Andreas Schwab [Tue, 11 Dec 2018 10:20:40 +0000 (11:20 +0100)]
riscv: don't stop itself in smp_send_stop

Add IPI_CPU_STOP message and use it in smp_send_stop to stop other cpus,
but not itself.  Mark cpu offline on reception of IPI_CPU_STOP.

Signed-off-by: Andreas Schwab <schwab@suse.de>
Reviewed-by: Atish Patra <atish.patra@wdc.com>
Signed-off-by: Palmer Dabbelt <palmer@sifive.com>
19 months agoarch: riscv: support kernel command line forcing when no DTB passed
Paul Walmsley [Tue, 18 Dec 2018 03:15:12 +0000 (19:15 -0800)]
arch: riscv: support kernel command line forcing when no DTB passed

CONFIG_CMDLINE_FORCE doesn't work on RISC-V when no DTB is passed into
the kernel.  This is because the code that forces the kernel command
line only runs if a valid DTB is present at boot.  During debugging,
it's useful to have the ability to force kernel command lines even
when no DTB is present.  This patch adds support for doing so.

Cc: Palmer Dabbelt <palmer@sifive.com>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: linux-riscv@lists.infradead.org
Cc: linux-kernel@vger.kernel.org (open list)
Signed-off-by: Paul Walmsley <paul.walmsley@sifive.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Signed-off-by: Palmer Dabbelt <palmer@sifive.com>
19 months agotools uapi: fix RISC-V 64-bit support
Aurelien Jarno [Tue, 25 Dec 2018 14:46:24 +0000 (15:46 +0100)]
tools uapi: fix RISC-V 64-bit support

The BPF library is not built on 64-bit RISC-V, as the BPF feature is
not detected. Looking more in details, feature/test-bpf.c fails to build
with the following error:

| In file included from /tmp/linux-4.19.12/tools/include/uapi/asm/bitsperlong.h:17,
|                  from /tmp/linux-4.19.12/tools/include/uapi/asm-generic/unistd.h:2,
|                  from /usr/include/riscv64-linux-gnu/asm/unistd.h:1,
|                  from test-bpf.c:2:
| /tmp/linux-4.19.12/tools/include/asm-generic/bitsperlong.h:14:2: error: #error Inconsistent word size. Check asm/bitsperlong.h
|  #error Inconsistent word size. Check asm/bitsperlong.h
|   ^~~~~

The UAPI from the tools directory is missing RISC-V support, therefore
bitsperlong.h from asm-generic is used, defaulting to 32 bits.

Fix that by adding tools/arch/riscv/include/uapi/asm/bitsperlong.h as
a copy of arch/riscv/include/uapi/asm/bitsperlong.h and by updating
tools/include/uapi/asm/bitsperlong.h.

Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Signed-off-by: Palmer Dabbelt <palmer@sifive.com>
19 months agoRISC-V: Make BSS section as the last section in vmlinux.lds.S
Anup Patel [Wed, 19 Dec 2018 12:36:57 +0000 (18:06 +0530)]
RISC-V: Make BSS section as the last section in vmlinux.lds.S

The objcopy only emits loadable sections when creating flat kernel
Image. To have minimal possible size of flat kernel Image, we should
have all non-loadable sections after loadable sections.

Currently, execption table section (loadable section) is after BSS
section (non-loadable section) in the RISC-V vmlinux.lds.S. This
is not optimal for having minimal flat kernel Image size hence this
patch makes BSS section as the last section in RISC-V vmlinux.lds.S.

In addition, we make BSS section aligned to 16byte instead of PAGE
aligned which further reduces flat kernel Image size by few KBs.

The flat kernel Image size of Linux-4.20-rc4 using GCC 8.2.0 is
8819980 bytes with current RISC-V vmlinux.lds.S and it reduces to
7991740 bytes with this patch applied. In summary, this patch reduces
Linux-4.20-rc4 flat kernel Image size by 809 KB.

Signed-off-by: Anup Patel <anup@brainfault.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Signed-off-by: Palmer Dabbelt <palmer@sifive.com>
19 months agoDocumentation/features: Add csky kernel features
Guo Ren [Fri, 4 Jan 2019 03:17:33 +0000 (11:17 +0800)]
Documentation/features: Add csky kernel features

      core/ cBPF-JIT             : TODO |
      core/ eBPF-JIT             : TODO |
      core/ generic-idle-thread  :  ok  |
      core/ jump-labels          : TODO |
      core/ tracehook            :  ok  |
     debug/ KASAN                : TODO |
     debug/ gcov-profile-all     : TODO |
     debug/ kgdb                 : TODO |
     debug/ kprobes-on-ftrace    : TODO |
     debug/ kprobes              : TODO |
     debug/ kretprobes           : TODO |
     debug/ optprobes            : TODO |
     debug/ stackprotector       : TODO |
     debug/ uprobes              : TODO |
     debug/ user-ret-profiler    : TODO |
        io/ dma-contiguous       :  ok  |
   locking/ cmpxchg-local        : TODO |
   locking/ lockdep              : TODO |
   locking/ queued-rwlocks       :  ok  |
   locking/ queued-spinlocks     : TODO |
   locking/ rwsem-optimized      : TODO |
      perf/ kprobes-event        : TODO |
      perf/ perf-regs            : TODO |
      perf/ perf-stackdump       : TODO |
     sched/ membarrier-sync-core : TODO |
     sched/ numa-balancing       :  ..  |
   seccomp/ seccomp-filter       : TODO |
      time/ arch-tick-broadcast  : TODO |
      time/ clockevents          :  ok  |
      time/ context-tracking     : TODO |
      time/ irq-time-acct        : TODO |
      time/ modern-timekeeping   :  ok  |
      time/ virt-cpuacct         : TODO |
        vm/ ELF-ASLR             : TODO |
        vm/ PG_uncached          : TODO |
        vm/ THP                  :  ..  |
        vm/ batch-unmap-tlb-flush: TODO |
        vm/ huge-vmap            : TODO |
        vm/ ioremap_prot         : TODO |
        vm/ numa-memblock        :  ..  |
        vm/ pte_special          : TODO |

Signed-off-by: Guo Ren <ren_guo@c-sky.com>
Cc: Arnd Bergmann <arnd@arndb.de>
19 months agomtd: Check add_mtd_device() ret code
Boris Brezillon [Wed, 2 Jan 2019 14:36:54 +0000 (15:36 +0100)]
mtd: Check add_mtd_device() ret code

add_mtd_device() can fail. We should always check its return value
and gracefully handle the failure case. Fix the call sites where this
not done (in mtdpart.c) and add a __must_check attribute to the
prototype to avoid this kind of mistakes.

Signed-off-by: Boris Brezillon <bbrezillon@kernel.org>
19 months agomtd: Fix the check on nvmem_register() ret code
Boris Brezillon [Wed, 2 Jan 2019 14:36:53 +0000 (15:36 +0100)]
mtd: Fix the check on nvmem_register() ret code

Commit 20167b70c894 ("nvmem: use EOPNOTSUPP instead of ENOSYS") changed
the nvmem_register() ret code from ENOSYS to EOPNOTSUPP when
CONFIG_NVMEM is not enabled, but the check in mtd_nvmem_add() was not
adjusted accordingly.

Cc: Bartosz Golaszewski <brgl@bgdev.pl>
Cc: Alban Bedel <albeu@free.fr>
Fixes: c4dfa25ab307 ("mtd: add support for reading MTD devices via the nvmem API")
Reported-by: kernel test robot <rong.a.chen@intel.com>
Signed-off-by: Boris Brezillon <bbrezillon@kernel.org>
Reviewed-by: Bartosz Golaszewski <bgolaszewski@baylibre.com>
Signed-off-by: Boris Brezillon <bbrezillon@kernel.org>
19 months agoACPI / PMIC: xpower: Fix TS-pin current-source handling
Hans de Goede [Fri, 4 Jan 2019 22:10:54 +0000 (23:10 +0100)]
ACPI / PMIC: xpower: Fix TS-pin current-source handling

The current-source used for the battery temp-sensor (TS) is shared with the
GPADC. For proper fuel-gauge and charger operation the TS current-source
needs to be permanently on. But to read the GPADC we need to temporary
switch the TS current-source to ondemand, so that the GPADC can use it,
otherwise we will always read an all 0 value.

The switching from on to on-ondemand is not necessary when the TS
current-source is off (this happens on devices which do not have a TS).

Prior to this commit there were 2 issues with our handling of the TS
current-source switching:

 1) We were writing hardcoded values to the ADC TS pin-ctrl register,
 overwriting various other unrelated bits. Specifically we were overwriting
 the current-source setting for the TS and GPIO0 pins, forcing it to 80ųA
 independent of its original setting. On a Chuwi Vi10 tablet this was
 causing us to get a too high adc value (due to a too high current-source)
 resulting in acpi_lpat_raw_to_temp() returning -ENOENT, resulting in:

ACPI Error: AE_ERROR, Returned by Handler for [UserDefinedRegion]
ACPI Error: Method parse/execution failed \_SB.SXP1._TMP, AE_ERROR

This commit fixes this by using regmap_update_bits to change only the
relevant bits.

 2) At the end of intel_xpower_pmic_get_raw_temp() we were unconditionally
 enabling the TS current-source even on devices where the TS-pin is not used
 and the current-source thus was off on entry of the function.

This commit fixes this by checking if the TS current-source is off when
entering intel_xpower_pmic_get_raw_temp() and if so it is left as is.

Fixes: 58eefe2f3f53 (ACPI / PMIC: xpower: Do pinswitch ... reading GPADC)
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Cc: 4.14+ <stable@vger.kernel.org> # 4.14+
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
19 months agoALSA: hda/realtek - Support Dell headset mode for New AIO platform
Kailang Yang [Thu, 3 Jan 2019 07:53:39 +0000 (15:53 +0800)]
ALSA: hda/realtek - Support Dell headset mode for New AIO platform

Dell has new platform for ALC274.
This will support to enable headset mode.

Signed-off-by: Kailang Yang <kailang@realtek.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
19 months agoALSA: usb-audio: Fix an out-of-bound read in create_composite_quirks
Hui Peng [Tue, 25 Dec 2018 23:11:52 +0000 (18:11 -0500)]
ALSA: usb-audio: Fix an out-of-bound read in create_composite_quirks

In `create_composite_quirk`, the terminating condition of for loops is
`quirk->ifnum < 0`. So any composite quirks should end with `struct
snd_usb_audio_quirk` object with ifnum < 0.

    for (quirk = quirk_comp->data; quirk->ifnum >= 0; ++quirk) {

     .....
    }

the data field of Bower's & Wilkins PX headphones usb device device quirks
do not end with {.ifnum = -1}, wihch may result in out-of-bound read.

This Patch fix the bug by adding an ending quirk object.

Fixes: 240a8af929c7 ("ALSA: usb-audio: Add a quirck for B&W PX headphones")
Signed-off-by: Hui Peng <benquike@163.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
19 months agoALSA: usb-audio: Always check descriptor sizes in parser code
Takashi Iwai [Wed, 2 Jan 2019 16:12:21 +0000 (17:12 +0100)]
ALSA: usb-audio: Always check descriptor sizes in parser code

There are a few places where we access the data without checking the
actual object size from the USB audio descriptor.  This may result in
OOB access, as recently reported.

This patch addresses these missing checks.  Most of added codes are
simple bLength checks in the caller side.  For the input and output
terminal parsers, we put the length check in the parser functions.
For the input terminal, a new argument is added to distinguish between
UAC1 and the rest, as they treat different objects.

Reported-by: Mathias Payer <mathias.payer@nebelwelt.net>
Reported-by: Hui Peng <benquike@163.com>
Tested-by: Hui Peng <benquike@163.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
19 months agoALSA: usb-audio: Check mixer unit descriptors more strictly
Takashi Iwai [Wed, 19 Dec 2018 13:04:47 +0000 (14:04 +0100)]
ALSA: usb-audio: Check mixer unit descriptors more strictly

We've had some sanity checks of the mixer unit descriptors but they
are too loose and some corner cases are overlooked.  Add more strict
checks in uac_mixer_unit_get_channels() for avoiding possible OOB
accesses by malformed descriptors.

This also changes the semantics of uac_mixer_unit_get_channels()
slightly.  Now it returns zero for the cases where the descriptor
lacks of bmControls instead of -EINVAL.  Then the caller side skips
the mixer creation for such unit while it keeps parsing it.
This corresponds to the case like Maya44.

Cc: <stable@vger.kernel.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
19 months agoALSA: usb-audio: Avoid access before bLength check in build_audio_procunit()
Takashi Iwai [Wed, 19 Dec 2018 11:36:27 +0000 (12:36 +0100)]
ALSA: usb-audio: Avoid access before bLength check in build_audio_procunit()

The parser for the processing unit reads bNrInPins field before the
bLength sanity check, which may lead to an out-of-bound access when a
malformed descriptor is given.  Fix it by assignment after the bLength
check.

Cc: <stable@vger.kernel.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
19 months agoarch: restore generic-y += shmparam.h for some architectures
Masahiro Yamada [Mon, 7 Jan 2019 02:08:08 +0000 (11:08 +0900)]
arch: restore generic-y += shmparam.h for some architectures

For some reasons, I accidentally got rid of "generic-y += shmparam.h"
from some architectures.

Restore them to fix building c6x, h8300, hexagon, m68k, microblaze,
openrisc, and unicore32.

Fixes: d6e4b3e326d8 ("arch: remove redundant UAPI generic-y defines")
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
19 months agoLinux 5.0-rc1 v5.0-rc1
Linus Torvalds [Mon, 7 Jan 2019 01:08:20 +0000 (17:08 -0800)]
Linux 5.0-rc1

19 months agoMerge tag 'kbuild-v4.21-3' of git://git.kernel.org/pub/scm/linux/kernel/git/masahiroy...
Linus Torvalds [Mon, 7 Jan 2019 00:33:10 +0000 (16:33 -0800)]
Merge tag 'kbuild-v4.21-3' of git://git./linux/kernel/git/masahiroy/linux-kbuild

Pull more Kbuild updates from Masahiro Yamada:

 - improve boolinit.cocci and use_after_iter.cocci semantic patches

 - fix alignment for kallsyms

 - move 'asm goto' compiler test to Kconfig and clean up jump_label
   CONFIG option

 - generate asm-generic wrappers automatically if arch does not
   implement mandatory UAPI headers

 - remove redundant generic-y defines

 - misc cleanups

* tag 'kbuild-v4.21-3' of git://git.kernel.org/pub/scm/linux/kernel/git/masahiroy/linux-kbuild:
  kconfig: rename generated .*conf-cfg to *conf-cfg
  kbuild: remove unnecessary stubs for archheader and archscripts
  kbuild: use assignment instead of define ... endef for filechk_* rules
  arch: remove redundant UAPI generic-y defines
  kbuild: generate asm-generic wrappers if mandatory headers are missing
  arch: remove stale comments "UAPI Header export list"
  riscv: remove redundant kernel-space generic-y
  kbuild: change filechk to surround the given command with { }
  kbuild: remove redundant target cleaning on failure
  kbuild: clean up rule_dtc_dt_yaml
  kbuild: remove UIMAGE_IN and UIMAGE_OUT
  jump_label: move 'asm goto' support test to Kconfig
  kallsyms: lower alignment on ARM
  scripts: coccinelle: boolinit: drop warnings on named constants
  scripts: coccinelle: check for redeclaration
  kconfig: remove unused "file" field of yylval union
  nds32: remove redundant kernel-space generic-y
  nios2: remove unneeded HAS_DMA define

19 months agoMerge branch 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Mon, 7 Jan 2019 00:30:14 +0000 (16:30 -0800)]
Merge branch 'perf-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull perf tooling updates form Ingo Molnar:
 "A final batch of perf tooling changes: mostly fixes and small
  improvements"

* 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (29 commits)
  perf session: Add comment for perf_session__register_idle_thread()
  perf thread-stack: Fix thread stack processing for the idle task
  perf thread-stack: Allocate an array of thread stacks
  perf thread-stack: Factor out thread_stack__init()
  perf thread-stack: Allow for a thread stack array
  perf thread-stack: Avoid direct reference to the thread's stack
  perf thread-stack: Tidy thread_stack__bottom() usage
  perf thread-stack: Simplify some code in thread_stack__process()
  tools gpio: Allow overriding CFLAGS
  tools power turbostat: Override CFLAGS assignments and add LDFLAGS to build command
  tools thermal tmon: Allow overriding CFLAGS assignments
  tools power x86_energy_perf_policy: Override CFLAGS assignments and add LDFLAGS to build command
  perf c2c: Increase the HITM ratio limit for displayed cachelines
  perf c2c: Change the default coalesce setup
  perf trace beauty ioctl: Beautify USBDEVFS_ commands
  perf trace beauty: Export function to get the files for a thread
  perf trace: Wire up ioctl's USBDEBFS_ cmd table generator
  perf beauty ioctl: Add generator for USBDEVFS_ ioctl commands
  tools headers uapi: Grab a copy of usbdevice_fs.h
  perf trace: Store the major number for a file when storing its pathname
  ...

19 months agoChange mincore() to count "mapped" pages rather than "cached" pages
Linus Torvalds [Sun, 6 Jan 2019 01:50:59 +0000 (17:50 -0800)]
Change mincore() to count "mapped" pages rather than "cached" pages

The semantics of what "in core" means for the mincore() system call are
somewhat unclear, but Linux has always (since 2.3.52, which is when
mincore() was initially done) treated it as "page is available in page
cache" rather than "page is mapped in the mapping".

The problem with that traditional semantic is that it exposes a lot of
system cache state that it really probably shouldn't, and that users
shouldn't really even care about.

So let's try to avoid that information leak by simply changing the
semantics to be that mincore() counts actual mapped pages, not pages
that might be cheaply mapped if they were faulted (note the "might be"
part of the old semantics: being in the cache doesn't actually guarantee
that you can access them without IO anyway, since things like network
filesystems may have to revalidate the cache before use).

In many ways the old semantics were somewhat insane even aside from the
information leak issue.  From the very beginning (and that beginning is
a long time ago: 2.3.52 was released in March 2000, I think), the code
had a comment saying

  Later we can get more picky about what "in core" means precisely.

and this is that "later".  Admittedly it is much later than is really
comfortable.

NOTE! This is a real semantic change, and it is for example known to
change the output of "fincore", since that program literally does a
mmmap without populating it, and then doing "mincore()" on that mapping
that doesn't actually have any pages in it.

I'm hoping that nobody actually has any workflow that cares, and the
info leak is real.

We may have to do something different if it turns out that people have
valid reasons to want the old semantics, and if we can limit the
information leak sanely.

Cc: Kevin Easton <kevin@guarana.org>
Cc: Jiri Kosina <jikos@kernel.org>
Cc: Masatake YAMATO <yamato@redhat.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Greg KH <gregkh@linuxfoundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
19 months agoFix 'acccess_ok()' on alpha and SH
Linus Torvalds [Sun, 6 Jan 2019 19:15:04 +0000 (11:15 -0800)]
Fix 'acccess_ok()' on alpha and SH

Commit 594cc251fdd0 ("make 'user_access_begin()' do 'access_ok()'")
broke both alpha and SH booting in qemu, as noticed by Guenter Roeck.

It turns out that the bug wasn't actually in that commit itself (which
would have been surprising: it was mostly a no-op), but in how the
addition of access_ok() to the strncpy_from_user() and strnlen_user()
functions now triggered the case where those functions would test the
access of the very last byte of the user address space.

The string functions actually did that user range test before too, but
they did it manually by just comparing against user_addr_max().  But
with user_access_begin() doing the check (using "access_ok()"), it now
exposed problems in the architecture implementations of that function.

For example, on alpha, the access_ok() helper macro looked like this:

  #define __access_ok(addr, size) \
        ((get_fs().seg & (addr | size | (addr+size))) == 0)

and what it basically tests is of any of the high bits get set (the
USER_DS masking value is 0xfffffc0000000000).

And that's completely wrong for the "addr+size" check.  Because it's
off-by-one for the case where we check to the very end of the user
address space, which is exactly what the strn*_user() functions do.

Why? Because "addr+size" will be exactly the size of the address space,
so trying to access the last byte of the user address space will fail
the __access_ok() check, even though it shouldn't.  As a result, the
user string accessor functions failed consistently - because they
literally don't know how long the string is going to be, and the max
access is going to be that last byte of the user address space.

Side note: that alpha macro is buggy for another reason too - it re-uses
the arguments twice.

And SH has another version of almost the exact same bug:

  #define __addr_ok(addr) \
        ((unsigned long __force)(addr) < current_thread_info()->addr_limit.seg)

so far so good: yes, a user address must be below the limit.  But then:

  #define __access_ok(addr, size)         \
        (__addr_ok((addr) + (size)))

is wrong with the exact same off-by-one case: the case when "addr+size"
is exactly _equal_ to the limit is actually perfectly fine (think "one
byte access at the last address of the user address space")

The SH version is actually seriously buggy in another way: it doesn't
actually check for overflow, even though it did copy the _comment_ that
talks about overflow.

So it turns out that both SH and alpha actually have completely buggy
implementations of access_ok(), but they happened to work in practice
(although the SH overflow one is a serious serious security bug, not
that anybody likely cares about SH security).

This fixes the problems by using a similar macro on both alpha and SH.
It isn't trying to be clever, the end address is based on this logic:

        unsigned long __ao_end = __ao_a + __ao_b - !!__ao_b;

which basically says "add start and length, and then subtract one unless
the length was zero".  We can't subtract one for a zero length, or we'd
just hit an underflow instead.

For a lot of access_ok() users the length is a constant, so this isn't
actually as expensive as it initially looks.

Reported-and-tested-by: Guenter Roeck <linux@roeck-us.net>
Cc: Matt Turner <mattst88@gmail.com>
Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
19 months agoMerge tag 'fscrypt_for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso...
Linus Torvalds [Sun, 6 Jan 2019 20:21:11 +0000 (12:21 -0800)]
Merge tag 'fscrypt_for_linus' of git://git./linux/kernel/git/tytso/fscrypt

Pull fscrypt updates from Ted Ts'o:
 "Add Adiantum support for fscrypt"

* tag 'fscrypt_for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/fscrypt:
  fscrypt: add Adiantum support

19 months agoMerge tag 'ext4_for_linus_stable' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sun, 6 Jan 2019 20:19:23 +0000 (12:19 -0800)]
Merge tag 'ext4_for_linus_stable' of git://git./linux/kernel/git/tytso/ext4

Pull ext4 bug fixes from Ted Ts'o:
 "Fix a number of ext4 bugs"

* tag 'ext4_for_linus_stable' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4:
  ext4: fix special inode number checks in __ext4_iget()
  ext4: track writeback errors using the generic tracking infrastructure
  ext4: use ext4_write_inode() when fsyncing w/o a journal
  ext4: avoid kernel warning when writing the superblock to a dead device
  ext4: fix a potential fiemap/page fault deadlock w/ inline_data
  ext4: make sure enough credits are reserved for dioread_nolock writes