cascardo/linux.git
8 years agoMerge tag 'pwm/for-4.6-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/thierry...
Linus Torvalds [Thu, 24 Mar 2016 00:14:09 +0000 (17:14 -0700)]
Merge tag 'pwm/for-4.6-rc1' of git://git./linux/kernel/git/thierry.reding/linux-pwm

Pull pwm updates from Thierry Reding:
 "No new drivers this time around, but a handful of cleanups and fixes"

* tag 'pwm/for-4.6-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/thierry.reding/linux-pwm:
  pwm: omap-dmtimer: Add debug message for effective period and duty cycle
  pwm: omap-dmtimer: Round load and match values rather than truncate
  pwm: omap-dmtimer: Add sanity checking for load and match values
  pwm: omap-dmtimer: Fix inaccurate period and duty cycle calculations
  pwm: brcmstb: Fix check of devm_ioremap_resource() return code
  pwm: rcar: Depend on ARCH_RENESAS instead of ARCH_SHMOBILE
  pwm: lpc18xx-sct: Test clock rate to avoid division by 0
  pwm: img: Test clock rate to avoid division by 0

8 years agoMerge branch 'for-next-merge' of git://git.kernel.org/pub/scm/linux/kernel/git/nab...
Linus Torvalds [Wed, 23 Mar 2016 22:57:39 +0000 (15:57 -0700)]
Merge branch 'for-next-merge' of git://git./linux/kernel/git/nab/target-pending

Pull more SCSI target updates from Nicholas Bellinger:
 "This series contains cxgb4 driver prerequisites for supporting iscsi
  segmentation offload (ISO), that will be utilized for a number of
  future v4.7 developments in iscsi-target for supporting generic hw
  offloads"

* 'for-next-merge' of git://git.kernel.org/pub/scm/linux/kernel/git/nab/target-pending:
  cxgb4: update Kconfig and Makefile
  cxgb4: add iSCSI DDP page pod manager
  cxgb4, iw_cxgb4: move delayed ack macro definitions
  cxgb4: move VLAN_NONE macro definition
  cxgb4: update struct cxgb4_lld_info definition
  cxgb4: add definitions for iSCSI target ULD
  cxgb4, cxgb4i: move struct cpl_rx_data_ddp definition
  cxgb4, iw_cxgb4, cxgb4i: remove duplicate definitions
  cxgb4, iw_cxgb4: move definitions to common header file
  cxgb4: large receive offload support
  cxgb4: allocate resources for CXGB4_ULD_ISCSIT
  cxgb4: add new ULD type CXGB4_ULD_ISCSIT

8 years agopwm: omap-dmtimer: Add debug message for effective period and duty cycle
David Rivshin [Sat, 30 Jan 2016 04:26:54 +0000 (23:26 -0500)]
pwm: omap-dmtimer: Add debug message for effective period and duty cycle

After going through the math and constraints checking to compute load
and match values, it is helpful to know what the resultant period and
duty cycle are.

Signed-off-by: David Rivshin <drivshin@allworx.com>
Acked-by: Neil Armstrong <narmstrong@baylibre.com>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
8 years agopwm: omap-dmtimer: Round load and match values rather than truncate
David Rivshin [Sat, 30 Jan 2016 04:26:53 +0000 (23:26 -0500)]
pwm: omap-dmtimer: Round load and match values rather than truncate

When converting period and duty_cycle from nanoseconds to fclk cycles,
the error introduced by the integer division can be appreciable, especially
in the case of slow fclk or short period. Use DIV_ROUND_CLOSEST_ULL() so
that the error is kept to +/- 0.5 clock cycles.

Fixes: 6604c6556db9 ("pwm: Add PWM driver for OMAP using dual-mode timers")
Signed-off-by: David Rivshin <drivshin@allworx.com>
Acked-by: Neil Armstrong <narmstrong@baylibre.com>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
8 years agopwm: omap-dmtimer: Add sanity checking for load and match values
David Rivshin [Sat, 30 Jan 2016 04:26:52 +0000 (23:26 -0500)]
pwm: omap-dmtimer: Add sanity checking for load and match values

Add sanity checking to ensure that we do not program load or match values
that are out of range if a user requests period or duty_cycle values which
are not achievable. The match value cannot be less than the load value (but
can be equal), and neither can be 0xffffffff. This means that there must be
at least one fclk cycle between load and match, and another between match
and overflow.

Fixes: 6604c6556db9 ("pwm: Add PWM driver for OMAP using dual-mode timers")
Signed-off-by: David Rivshin <drivshin@allworx.com>
Acked-by: Neil Armstrong <narmstrong@baylibre.com>
[thierry.reding@gmail.com: minor coding style cleanups]
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
8 years agopwm: omap-dmtimer: Fix inaccurate period and duty cycle calculations
David Rivshin [Sat, 30 Jan 2016 04:26:51 +0000 (23:26 -0500)]
pwm: omap-dmtimer: Fix inaccurate period and duty cycle calculations

Fix the calculation of load_value and match_value. Currently they
are slightly too low, which produces a noticeably wrong PWM rate with
sufficiently short periods (i.e. when 1/period approaches clk_rate/2).

Example:
 clk_rate=32768Hz, period=122070ns, duty_cycle=61035ns (8192Hz/50% PWM)
 Correct values: load = 0xfffffffc, match = 0xfffffffd
 Current values: load = 0xfffffffa, match = 0xfffffffc
 effective PWM: period=183105ns, duty_cycle=91553ns (5461Hz/50% PWM)

Fixes: 6604c6556db9 ("pwm: Add PWM driver for OMAP using dual-mode timers")
Signed-off-by: David Rivshin <drivshin@allworx.com>
Acked-by: Neil Armstrong <narmstrong@baylibre.com>
Tested-by: Adam Ford <aford173@gmail.com>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
8 years agopwm: brcmstb: Fix check of devm_ioremap_resource() return code
Vladimir Zapolskiy [Sun, 6 Mar 2016 01:21:46 +0000 (03:21 +0200)]
pwm: brcmstb: Fix check of devm_ioremap_resource() return code

The change fixes potential oops while accessing iomem on invalid address
if devm_ioremap_resource() fails due to some reason.

The devm_ioremap_resource() function returns ERR_PTR() and never returns
NULL, which makes useless a following check for NULL.

Signed-off-by: Vladimir Zapolskiy <vz@mleia.com>
Fixes: 3a9f5957020f ("pwm: Add Broadcom BCM7038 PWM controller support")
Acked-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
8 years agopwm: rcar: Depend on ARCH_RENESAS instead of ARCH_SHMOBILE
Simon Horman [Thu, 25 Feb 2016 01:03:23 +0000 (10:03 +0900)]
pwm: rcar: Depend on ARCH_RENESAS instead of ARCH_SHMOBILE

This is part of an ongoing process to migrate from ARCH_SHMOBILE to
ARCH_RENESAS the motivation for which being that RENESAS seems to be a
more appropriate name than SHMOBILE for the majority of Renesas ARM
based SoCs.

Signed-off-by: Simon Horman <horms+renesas@verge.net.au>
Acked-by: Geert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
8 years agopwm: lpc18xx-sct: Test clock rate to avoid division by 0
Wolfram Sang [Wed, 2 Mar 2016 22:57:09 +0000 (23:57 +0100)]
pwm: lpc18xx-sct: Test clock rate to avoid division by 0

The clk API may return 0 on clk_get_rate(), so we should check the
result before using it as a divisor.

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Acked-by: Joachim Eastwood <manabian@gmail.com>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
8 years agopwm: img: Test clock rate to avoid division by 0
Wolfram Sang [Wed, 2 Mar 2016 22:33:34 +0000 (23:33 +0100)]
pwm: img: Test clock rate to avoid division by 0

The clk API may return 0 on clk_get_rate(), so we should check the
result before using it as a divisor.

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
8 years agoMerge branch 'linus' of git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6
Linus Torvalds [Wed, 23 Mar 2016 13:12:39 +0000 (06:12 -0700)]
Merge branch 'linus' of git://git./linux/kernel/git/herbert/crypto-2.6

Pull crypto fixes from Herbert Xu:
 "This fixes the following issues:

  API:
   - Fix kzalloc error path crash in ecryptfs added by skcipher
     conversion.  Note the subject of the commit is screwed up and the
     correct subject is actually in the body.

  Drivers:
   - A number of fixes to the marvell cesa hashing code.
   - Remove bogus nested irqsave that clobbers the saved flags in ccp"

* 'linus' of git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6:
  crypto: marvell/cesa - forward devm_ioremap_resource() error code
  crypto: marvell/cesa - initialize hash states
  crypto: marvell/cesa - fix memory leak
  crypto: ccp - fix lock acquisition code
  eCryptfs: Use skcipher and shash

8 years agoMerge branch 'mailbox-for-next' of git://git.linaro.org/landing-teams/working/fujitsu...
Linus Torvalds [Wed, 23 Mar 2016 13:09:15 +0000 (06:09 -0700)]
Merge branch 'mailbox-for-next' of git://git.linaro.org/landing-teams/working/fujitsu/integration

Pull more mailbox updates from Jassi Brar:
 "Device tree bindings and driver for TI's Message-Manager controller.

  Due to some last minute cosmetic changes, the driver was not included
  in the first pull request, otherwise the driver has been reviewed
  twice"

* 'mailbox-for-next' of git://git.linaro.org/landing-teams/working/fujitsu/integration:
  mailbox: Introduce TI message manager driver
  Documentation: dt: mailbox: Add TI Message Manager

8 years agoMerge tag 'clk-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/clk/linux
Linus Torvalds [Wed, 23 Mar 2016 13:06:45 +0000 (06:06 -0700)]
Merge tag 'clk-for-linus' of git://git./linux/kernel/git/clk/linux

Pull clk updates from Stephen Boyd:
 "The clk changes for this release cycle are mostly dominated by new
  device support in terms of LoC, but there has been some cleanup in the
  core as well as the usual minor clk additions to various drivers.

  Core:
   - parent tracking has been simplified
   - CLK_IS_ROOT is now a no-op flag, cleaning up drivers has started
   - of_clk_init() doesn't consider disabled DT nodes anymore
   - clk_unregister() had an error path bug squashed
   - of_clk_get_parent_count() has been fixed to only return unsigned ints
   - HAVE_MACH_CLKDEV is removed now that the last arch user (ARM) is gone

  New Drivers:
   - NXP LPC18xx creg
   - QCOM IPQ4019 GCC
   - TI dm814x ADPLL
   - i.MX6QP

  Updates:
   - Cyngus audio clks found on Broadcom iProc devices
   - Non-critical fixes for BCM2385 PLLs
   - Samsung exynos5433 updates for clk id errors, HDMI support,
     suspend/resume simplifications
   - USB, CAN, LVDS, and FCP clks on shmobile devices
   - sunxi got support for more clks on new SoCs and went through a
     minor refactoring/rewrite to use a simpler factor clk construct
   - rockchip added some more clk ids and added suport for fraction
     dividers
   - QCOM GDSCs in msm8996
   - A new devm helper to make adding custom actions simpler (acked by Greg)"

* tag 'clk-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/clk/linux: (197 commits)
  clk: bcm2835: fix check of error code returned by devm_ioremap_resource()
  clk: renesas: div6: use RENESAS for #define
  clk: renesas: Rename header file renesas.h
  clk: max77{686,802}: Remove CLK_IS_ROOT
  clk: versatile: Remove CLK_IS_ROOT
  clk: sunxi: Remove use of variable length array
  clk: fixed-rate: Remove CLK_IS_ROOT
  clk: qcom: Remove CLK_IS_ROOT
  doc: dt: add documentation for lpc1850-creg-clk driver
  clk: add lpc18xx creg clk driver
  clk: lpc32xx: fix compilation warning
  clk: xgene: Add missing parenthesis when clearing divider value
  clk: mb86s7x: Remove CLK_IS_ROOT
  clk: x86: Remove clkdev.h and clk.h includes
  clk: x86: Remove CLK_IS_ROOT
  clk: mvebu: Remove CLK_IS_ROOT
  clk: renesas: move drivers to renesas directory
  clk: si5{14,351,70}: Remove CLK_IS_ROOT
  clk: scpi: Remove CLK_IS_ROOT
  clk: s2mps11: Remove CLK_IS_ROOT
  ...

8 years agoMerge branch 'akpm' (patches from Andrew)
Linus Torvalds [Wed, 23 Mar 2016 00:09:14 +0000 (17:09 -0700)]
Merge branch 'akpm' (patches from Andrew)

Merge third patch-bomb from Andrew Morton:

 - more ocfs2 changes

 - a few hotfixes

 - Andy's compat cleanups

 - misc fixes to fatfs, ptrace, coredump, cpumask, creds, eventfd,
   panic, ipmi, kgdb, profile, kfifo, ubsan, etc.

 - many rapidio updates: fixes, new drivers.

 - kcov: kernel code coverage feature.  Like gcov, but not
   "prohibitively expensive".

 - extable code consolidation for various archs

* emailed patches from Andrew Morton <akpm@linux-foundation.org>: (81 commits)
  ia64/extable: use generic search and sort routines
  x86/extable: use generic search and sort routines
  s390/extable: use generic search and sort routines
  alpha/extable: use generic search and sort routines
  kernel/...: convert pr_warning to pr_warn
  drivers: dma-coherent: use memset_io for DMA_MEMORY_IO mappings
  drivers: dma-coherent: use MEMREMAP_WC for DMA_MEMORY_MAP
  memremap: add MEMREMAP_WC flag
  memremap: don't modify flags
  kernel/signal.c: add compile-time check for __ARCH_SI_PREAMBLE_SIZE
  mm/mprotect.c: don't imply PROT_EXEC on non-exec fs
  ipc/sem: make semctl setting sempid consistent
  ubsan: fix tree-wide -Wmaybe-uninitialized false positives
  kfifo: fix sparse complaints
  scripts/gdb: account for changes in module data structure
  scripts/gdb: add cmdline reader command
  scripts/gdb: add version command
  kernel: add kcov code coverage
  profile: hide unused functions when !CONFIG_PROC_FS
  hpwdt: use nmi_panic() when kernel panics in NMI handler
  ...

8 years agoMerge tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm
Linus Torvalds [Tue, 22 Mar 2016 23:28:22 +0000 (16:28 -0700)]
Merge tag 'for-linus' of git://git./virt/kvm/kvm

Pull more KVM updates from Paolo Bonzini:
 "Second round of KVM changes for 4.6:

   - build fixes for PPC KVM
   - miscellaneous bugfixes for ARM KVM
   - cleanup of memory barrier and removal of redundant barriers
   - x86 fixes: page tracking oops, support for old buggy KVM nested on 4.5
   - support for protection keys in guests
   - lockdep fix
   - another conversion to simple wait queues and raw spinlocks,
     backported from PREEMPT_RT"

* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm: (27 commits)
  KVM: page_track: fix access to NULL slot
  KVM: PPC: do not compile in vfio.o unconditionally
  kvm, rt: change async pagefault code locking for PREEMPT_RT
  KVM/PPC: update the comment of memory barrier in the kvmppc_prepare_to_enter()
  KVM/x86: update the comment of memory barrier in the vcpu_enter_guest()
  KVM: Replace smp_mb() with smp_load_acquire() in the kvm_flush_remote_tlbs()
  KVM/x86: Call smp_wmb() before increasing tlbs_dirty
  KVM: Replace smp_mb() with smp_mb_after_atomic() in the kvm_make_all_cpus_request()
  KVM/x86: Replace smp_mb() with smp_store_mb/release() in the walk_shadow_page_lockless_begin/end()
  KVM: Remove redundant smp_mb() in the kvm_mmu_commit_zap_page()
  KVM, pkeys: expose CPUID/CR4 to guest
  KVM, pkeys: add pkeys support for permission_fault
  KVM, pkeys: introduce pkru_mask to cache conditions
  KVM, pkeys: save/restore PKRU when guest/host switches
  x86: pkey: introduce write_pkru() for KVM
  KVM, pkeys: add pkeys support for xsave state
  KVM, pkeys: disable pkeys for guests in non-paging mode
  KVM: x86: remove magic number with enum cpuid_leafs
  KVM: MMU: return page fault error code from permission_fault
  KVM: fix spin_lock_init order on x86
  ...

8 years agoMerge tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dledford/rdma
Linus Torvalds [Tue, 22 Mar 2016 22:48:44 +0000 (15:48 -0700)]
Merge tag 'for-linus' of git://git./linux/kernel/git/dledford/rdma

Pull more rdma updates from Doug Ledford:
 "Round two of 4.6 merge window patches.

  This is a monster pull request.  I held off on the hfi1 driver updates
  (the hfi1 driver is intimately tied to the qib driver and the new
  rdmavt software library that was created to help both of them) in my
  first pull request.  The hfi1/qib/rdmavt update is probably 90% of
  this pull request.  The hfi1 driver is being left in staging so that
  it can be fixed up in regards to the API that Al and yourself didn't
  like.  Intel has agreed to do the work, but in the meantime, this
  clears out 300+ patches in the backlog queue and brings my tree and
  their tree closer to sync.

  This also includes about 10 patches to the core and a few to mlx5 to
  create an infrastructure for configuring SRIOV ports on IB devices.
  That series includes one patch to the net core that we sent to netdev@
  and Dave Miller with each of the three revisions to the series.  We
  didn't get any response to the patch, so we took that as implicit
  approval.

  Finally, this series includes Intel's new iWARP driver for their x722
  cards.  It's not nearly the beast as the hfi1 driver.  It also has a
  linux-next merge issue, but that has been resolved and it now passes
  just fine.

  Summary:

   - A few minor core fixups needed for the next patch series

   - The IB SRIOV series.  This has bounced around for several versions.
     Of note is the fact that the first patch in this series effects the
     net core.  It was directed to netdev and DaveM for each iteration
     of the series (three versions total).  Dave did not object, but did
     not respond either.  I've taken this as permission to move forward
     with the series.

   - The new Intel X722 iWARP driver

   - A huge set of updates to the Intel hfi1 driver.  Of particular
     interest here is that we have left the driver in staging since it
     still has an API that people object to.  Intel is working on a fix,
     but getting these patches in now helps keep me sane as the upstream
     and Intel's trees were over 300 patches apart"

* tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dledford/rdma: (362 commits)
  IB/ipoib: Allow mcast packets from other VFs
  IB/mlx5: Implement callbacks for manipulating VFs
  net/mlx5_core: Implement modify HCA vport command
  net/mlx5_core: Add VF param when querying vport counter
  IB/ipoib: Add ndo operations for configuring VFs
  IB/core: Add interfaces to control VF attributes
  IB/core: Support accessing SA in virtualized environment
  IB/core: Add subnet prefix to port info
  IB/mlx5: Fix decision on using MAD_IFC
  net/core: Add support for configuring VF GUIDs
  IB/{core, ulp} Support above 32 possible device capability flags
  IB/core: Replace setting the zero values in ib_uverbs_ex_query_device
  net/mlx5_core: Introduce offload arithmetic hardware capabilities
  net/mlx5_core: Refactor device capability function
  net/mlx5_core: Fix caching ATOMIC endian mode capability
  ib_srpt: fix a WARN_ON() message
  i40iw: Replace the obsolete crypto hash interface with shash
  IB/hfi1: Add SDMA cache eviction algorithm
  IB/hfi1: Switch to using the pin query function
  IB/hfi1: Specify mm when releasing pages
  ...

8 years agoia64/extable: use generic search and sort routines
Ard Biesheuvel [Tue, 22 Mar 2016 21:28:20 +0000 (14:28 -0700)]
ia64/extable: use generic search and sort routines

Replace the arch specific versions of search_extable() and
sort_extable() with calls to the generic ones, which now support
relative exception tables as well.

Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Acked-by: Tony Luck <tony.luck@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agox86/extable: use generic search and sort routines
Ard Biesheuvel [Tue, 22 Mar 2016 21:28:17 +0000 (14:28 -0700)]
x86/extable: use generic search and sort routines

Replace the arch specific versions of search_extable() and
sort_extable() with calls to the generic ones, which now support
relative exception tables as well.

Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Acked-by: H. Peter Anvin <hpa@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agos390/extable: use generic search and sort routines
Ard Biesheuvel [Tue, 22 Mar 2016 21:28:14 +0000 (14:28 -0700)]
s390/extable: use generic search and sort routines

Replace the arch specific versions of search_extable() and
sort_extable() with calls to the generic ones, which now support
relative exception tables as well.

Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Acked-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agoalpha/extable: use generic search and sort routines
Ard Biesheuvel [Tue, 22 Mar 2016 21:28:12 +0000 (14:28 -0700)]
alpha/extable: use generic search and sort routines

Replace the arch specific versions of search_extable() and
sort_extable() with calls to the generic ones, which now support
relative exception tables as well.

Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Acked-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agokernel/...: convert pr_warning to pr_warn
Joe Perches [Tue, 22 Mar 2016 21:28:09 +0000 (14:28 -0700)]
kernel/...: convert pr_warning to pr_warn

Use the more common logging method with the eventual goal of removing
pr_warning altogether.

Miscellanea:

 - Realign arguments
 - Coalesce formats
 - Add missing space between a few coalesced formats

Signed-off-by: Joe Perches <joe@perches.com>
Acked-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> [kernel/power/suspend.c]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agodrivers: dma-coherent: use memset_io for DMA_MEMORY_IO mappings
Brian Starkey [Tue, 22 Mar 2016 21:28:06 +0000 (14:28 -0700)]
drivers: dma-coherent: use memset_io for DMA_MEMORY_IO mappings

Use memset_io() for DMA_MEMORY_IO mappings which are mapped as I/O
memory, and regular memset() for DMA_MEMORY_MAP mappings.

This fixes the below alignment fault on arm64 for DMA_MEMORY_IO
mappings, where memset() uses the DC ZVA instruction which is invalid on
device memory.

   Unhandled fault: alignment fault (0x96000061) at 0xffffff8000380000
   Internal error: : 96000061 [#1] PREEMPT SMP
   Modules linked in: hdlcd(+) clk_scpi
   CPU: 4 PID: 1355 Comm: systemd-udevd Not tainted 4.4.0-rc1+ #5
   Hardware name: ARM Juno development board (r0) (DT)
   task: ffffffc9763eee00 ti: ffffffc9758c4000 task.ti: ffffffc9758c4000
   PC is at __efistub_memset+0x1ac/0x200
   LR is at dma_alloc_from_coherent+0xb0/0x120
   pc : [<ffffffc00030ff2c>] lr : [<ffffffc00042a918>] pstate: 400001c5
   sp : ffffffc9758c79a0
   x29: ffffffc9758c79a0 x28: ffffffc000635cd0
   x27: 0000000000000124 x26: ffffffc000119ef4
   x25: 0000000000010000 x24: 0000000000000140
   x23: ffffffc07e9ac3a8 x22: ffffffc9758c7a58
   x21: ffffffc9758c7a68 x20: 0000000000000004
   x19: ffffffc07e9ac380 x18: 0000000000000001
   x17: 0000007fae1bbba8 x16: ffffffc0001b2d1c
   x15: ffffffffffffffff x14: 0ffffffffffffffe
   x13: 0000000000000010 x12: ffffff800837ffff
   x11: ffffff800837ffff x10: 0000000040000000
   x9 : 0000000000000000 x8 : ffffff8000380000
   x7 : 0000000000000000 x6 : 000000000000003f
   x5 : 0000000000000040 x4 : 0000000000000000
   x3 : 0000000000000004 x2 : 000000000000ffc0
   x1 : 0000000000000000 x0 : ffffff8000380000

Signed-off-by: Brian Starkey <brian.starkey@arm.com>
Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agodrivers: dma-coherent: use MEMREMAP_WC for DMA_MEMORY_MAP
Brian Starkey [Tue, 22 Mar 2016 21:28:03 +0000 (14:28 -0700)]
drivers: dma-coherent: use MEMREMAP_WC for DMA_MEMORY_MAP

When the DMA_MEMORY_MAP flag is used, memory which can be accessed
directly should be returned, so use memremap(..., MEMREMAP_WC) to
provide a writecombine mapping.

Signed-off-by: Brian Starkey <brian.starkey@arm.com>
Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agomemremap: add MEMREMAP_WC flag
Brian Starkey [Tue, 22 Mar 2016 21:28:00 +0000 (14:28 -0700)]
memremap: add MEMREMAP_WC flag

Add a flag to memremap() for writecombine mappings.  Mappings satisfied
by this flag will not be cached, however writes may be delayed or
combined into more efficient bursts.  This is most suitable for buffers
written sequentially by the CPU for use by other DMA devices.

Signed-off-by: Brian Starkey <brian.starkey@arm.com>
Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agomemremap: don't modify flags
Brian Starkey [Tue, 22 Mar 2016 21:27:57 +0000 (14:27 -0700)]
memremap: don't modify flags

These patches implement a MEMREMAP_WC flag for memremap(), which can be
used to obtain writecombine mappings.  This is then used for setting up
dma_coherent_mem regions which use the DMA_MEMORY_MAP flag.

The motivation is to fix an alignment fault on arm64, and the suggestion
to implement MEMREMAP_WC for this case was made at [1].  That particular
issue is handled in patch 4, which makes sure that the appropriate
memset function is used when zeroing allocations mapped as IO memory.

This patch (of 4):

Don't modify the flags input argument to memremap(). MEMREMAP_WB is
already a special case so we can check for it directly instead of
clearing flag bits in each mapper.

Signed-off-by: Brian Starkey <brian.starkey@arm.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agokernel/signal.c: add compile-time check for __ARCH_SI_PREAMBLE_SIZE
Helge Deller [Tue, 22 Mar 2016 21:27:54 +0000 (14:27 -0700)]
kernel/signal.c: add compile-time check for __ARCH_SI_PREAMBLE_SIZE

The value of __ARCH_SI_PREAMBLE_SIZE defines the size (including
padding) of the part of the struct siginfo that is before the union, and
it is then used to calculate the needed padding (SI_PAD_SIZE) to make
the size of struct siginfo equal to 128 (SI_MAX_SIZE) bytes.

Depending on the target architecture and word width it equals to either
3 or 4 times sizeof int.

Since the very beginning we had __ARCH_SI_PREAMBLE_SIZE wrong on the
parisc architecture for the 64bit kernel build.  It's even more
frustrating, because it can easily be checked at compile time if the
value was defined correctly.

This patch adds such a check for the correctness of
__ARCH_SI_PREAMBLE_SIZE in the hope that it will prevent existing and
future architectures from running into the same problem.

I refrained from replacing __ARCH_SI_PREAMBLE_SIZE by offsetof() in
copy_siginfo() in include/asm-generic/siginfo.h, because a) it doesn't
make any difference and b) it's used in the Documentation/kmemcheck.txt
example.

I ran this patch through the 0-DAY kernel test infrastructure and only
the parisc architecture triggered as expected.  That means that this
patch should be OK for all major architectures.

Signed-off-by: Helge Deller <deller@gmx.de>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agomm/mprotect.c: don't imply PROT_EXEC on non-exec fs
Piotr Kwapulinski [Tue, 22 Mar 2016 21:27:51 +0000 (14:27 -0700)]
mm/mprotect.c: don't imply PROT_EXEC on non-exec fs

The mprotect(PROT_READ) fails when called by the READ_IMPLIES_EXEC
binary on a memory mapped file located on non-exec fs.  The mprotect
does not check whether fs is _executable_ or not.  The PROT_EXEC flag is
set automatically even if a memory mapped file is located on non-exec
fs.  Fix it by checking whether a memory mapped file is located on a
non-exec fs.  If so the PROT_EXEC is not implied by the PROT_READ.  The
implementation uses the VM_MAYEXEC flag set properly in mmap.  Now it is
consistent with mmap.

I did the isolated tests (PT_GNU_STACK X/NX, multiple VMAs, X/NX fs).  I
also patched the official 3.19.0-47-generic Ubuntu 14.04 kernel and it
seems to work.

Signed-off-by: Piotr Kwapulinski <kwapulinski.piotr@gmail.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Konstantin Khlebnikov <koct9i@gmail.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agoipc/sem: make semctl setting sempid consistent
Davidlohr Bueso [Tue, 22 Mar 2016 21:27:48 +0000 (14:27 -0700)]
ipc/sem: make semctl setting sempid consistent

As indicated by bug#112271, Linux sets the sempid value upon semctl, and
not only for semop calls.  However, within semctl we only do this for
SETVAL, leaving SETALL without updating the field, and therefore rather
inconsistent behavior when compared to other Unices.

There is really no documentation regarding this and therefore users
should not make assumptions.  With this patch, along with updating
semctl.2 manpages, this scenario should become less ambiguous As such,
set sempid on SETALL cmd.

Also update some in-code documentation, specifying where the sempid is
set.

Passes ltp and custom testcase where a child (fork) does SETALL to the
set.

Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Reported-by: Philip Semanchuk <linux_kernel.20.ick@spamgourmet.com>
Cc: Michael Kerrisk <mtk.manpages@gmail.com>
Cc: PrasannaKumar Muralidharan <prasannatsmkumar@gmail.com>
Cc: Manfred Spraul <manfred@colorfullife.com>
Cc: Herton R. Krzesinski <herton@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agoubsan: fix tree-wide -Wmaybe-uninitialized false positives
Andrey Ryabinin [Tue, 22 Mar 2016 21:27:45 +0000 (14:27 -0700)]
ubsan: fix tree-wide -Wmaybe-uninitialized false positives

-fsanitize=* options makes GCC less smart than usual and increase number
of 'maybe-uninitialized' false-positives. So this patch does two things:

 * Add -Wno-maybe-uninitialized to CFLAGS_UBSAN which will disable all
   such warnings for instrumented files.

 * Remove CONFIG_UBSAN_SANITIZE_ALL from all[yes|mod]config builds. So
   the all[yes|mod]config build goes without -fsanitize=* and still with
   -Wmaybe-uninitialized.

Signed-off-by: Andrey Ryabinin <aryabinin@virtuozzo.com>
Reported-by: Fengguang Wu <fengguang.wu@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agokfifo: fix sparse complaints
Stefani Seibold [Tue, 22 Mar 2016 21:27:42 +0000 (14:27 -0700)]
kfifo: fix sparse complaints

This patch fix complaints by the sparse tool when using kfifo_put() with
non scalar types like structures (i.e.
drivers/iio/industrialio-event.c).

Casting a pointer to the value and read this pointer instead of directly
casting the value will fix this.

The generated code is equal.

Signed-off-by: Stefani Seibold <stefani@seibold.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agoscripts/gdb: account for changes in module data structure
Jan Kiszka [Tue, 22 Mar 2016 21:27:39 +0000 (14:27 -0700)]
scripts/gdb: account for changes in module data structure

Commit 7523e4dc5057 ("module: use a structure to encapsulate layout.")
factored out the module_layout structure.  Adjust the symbol loader and
the lsmod command to this.

Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Reviewed-by: Kieran Bingham <kieran.bingham@linaro.org>
Tested-by: Kieran Bingham <kieran.bingham@linaro.org> (qemu-{ARM,x86})
Cc: Rusty Russell <rusty@rustcorp.com.au>
Cc: Jiri Kosina <jkosina@suse.cz>
Cc: Jason Wessel <jason.wessel@windriver.com>
Cc: <stable@vger.kernel.org> [4.4+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agoscripts/gdb: add cmdline reader command
Kieran Bingham [Tue, 22 Mar 2016 21:27:36 +0000 (14:27 -0700)]
scripts/gdb: add cmdline reader command

lx-cmdline Report the Linux Commandline used in the current kernel

[jan.kiszka@siemens.com: remove blank line from help output and fix pep8 warning]
Signed-off-by: Kieran Bingham <kieran.bingham@linaro.org>
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Cc: Jason Wessel <jason.wessel@windriver.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agoscripts/gdb: add version command
Kieran Bingham [Tue, 22 Mar 2016 21:27:33 +0000 (14:27 -0700)]
scripts/gdb: add version command

lx-version Report the Linux Version of the current kernel.

Add a command to identify the version specified by the banner in the
debugged kernel.

This lets the user identify the kernel of the running kernel, and will
let later scripts compare the banner of the attached kernel against the
banner in the vmlinux symbols files to verify that the files are
correct.

[jan.kiszka@siemens.com: remove blank line from help output and fix pep8 warning]
Signed-off-by: Kieran Bingham <kieran.bingham@linaro.org>
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Cc: Jason Wessel <jason.wessel@windriver.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agokernel: add kcov code coverage
Dmitry Vyukov [Tue, 22 Mar 2016 21:27:30 +0000 (14:27 -0700)]
kernel: add kcov code coverage

kcov provides code coverage collection for coverage-guided fuzzing
(randomized testing).  Coverage-guided fuzzing is a testing technique
that uses coverage feedback to determine new interesting inputs to a
system.  A notable user-space example is AFL
(http://lcamtuf.coredump.cx/afl/).  However, this technique is not
widely used for kernel testing due to missing compiler and kernel
support.

kcov does not aim to collect as much coverage as possible.  It aims to
collect more or less stable coverage that is function of syscall inputs.
To achieve this goal it does not collect coverage in soft/hard
interrupts and instrumentation of some inherently non-deterministic or
non-interesting parts of kernel is disbled (e.g.  scheduler, locking).

Currently there is a single coverage collection mode (tracing), but the
API anticipates additional collection modes.  Initially I also
implemented a second mode which exposes coverage in a fixed-size hash
table of counters (what Quentin used in his original patch).  I've
dropped the second mode for simplicity.

This patch adds the necessary support on kernel side.  The complimentary
compiler support was added in gcc revision 231296.

We've used this support to build syzkaller system call fuzzer, which has
found 90 kernel bugs in just 2 months:

  https://github.com/google/syzkaller/wiki/Found-Bugs

We've also found 30+ bugs in our internal systems with syzkaller.
Another (yet unexplored) direction where kcov coverage would greatly
help is more traditional "blob mutation".  For example, mounting a
random blob as a filesystem, or receiving a random blob over wire.

Why not gcov.  Typical fuzzing loop looks as follows: (1) reset
coverage, (2) execute a bit of code, (3) collect coverage, repeat.  A
typical coverage can be just a dozen of basic blocks (e.g.  an invalid
input).  In such context gcov becomes prohibitively expensive as
reset/collect coverage steps depend on total number of basic
blocks/edges in program (in case of kernel it is about 2M).  Cost of
kcov depends only on number of executed basic blocks/edges.  On top of
that, kernel requires per-thread coverage because there are always
background threads and unrelated processes that also produce coverage.
With inlined gcov instrumentation per-thread coverage is not possible.

kcov exposes kernel PCs and control flow to user-space which is
insecure.  But debugfs should not be mapped as user accessible.

Based on a patch by Quentin Casasnovas.

[akpm@linux-foundation.org: make task_struct.kcov_mode have type `enum kcov_mode']
[akpm@linux-foundation.org: unbreak allmodconfig]
[akpm@linux-foundation.org: follow x86 Makefile layout standards]
Signed-off-by: Dmitry Vyukov <dvyukov@google.com>
Reviewed-by: Kees Cook <keescook@chromium.org>
Cc: syzkaller <syzkaller@googlegroups.com>
Cc: Vegard Nossum <vegard.nossum@oracle.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Tavis Ormandy <taviso@google.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Quentin Casasnovas <quentin.casasnovas@oracle.com>
Cc: Kostya Serebryany <kcc@google.com>
Cc: Eric Dumazet <edumazet@google.com>
Cc: Alexander Potapenko <glider@google.com>
Cc: Kees Cook <keescook@google.com>
Cc: Bjorn Helgaas <bhelgaas@google.com>
Cc: Sasha Levin <sasha.levin@oracle.com>
Cc: David Drysdale <drysdale@google.com>
Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Cc: Andrey Ryabinin <ryabinin.a.a@gmail.com>
Cc: Kirill A. Shutemov <kirill@shutemov.name>
Cc: Jiri Slaby <jslaby@suse.cz>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agoprofile: hide unused functions when !CONFIG_PROC_FS
Arnd Bergmann [Tue, 22 Mar 2016 21:27:26 +0000 (14:27 -0700)]
profile: hide unused functions when !CONFIG_PROC_FS

A couple of functions and variables in the profile implementation are
used only on SMP systems by the procfs code, but are unused if either
procfs is disabled or in uniprocessor kernels.  gcc prints a harmless
warning about the unused symbols:

  kernel/profile.c:243:13: error: 'profile_flip_buffers' defined but not used [-Werror=unused-function]
   static void profile_flip_buffers(void)
               ^
  kernel/profile.c:266:13: error: 'profile_discard_flip_buffers' defined but not used [-Werror=unused-function]
   static void profile_discard_flip_buffers(void)
               ^
  kernel/profile.c:330:12: error: 'profile_cpu_callback' defined but not used [-Werror=unused-function]
   static int profile_cpu_callback(struct notifier_block *info,
              ^

This adds further #ifdef to the file, to annotate exactly in which cases
they are used.  I have done several thousand ARM randconfig kernels with
this patch applied and no longer get any warnings in this file.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Robin Holt <robinmholt@gmail.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Christoph Lameter <cl@linux.com>
Cc: Tejun Heo <tj@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agohpwdt: use nmi_panic() when kernel panics in NMI handler
Hidehiro Kawai [Tue, 22 Mar 2016 21:27:24 +0000 (14:27 -0700)]
hpwdt: use nmi_panic() when kernel panics in NMI handler

Commit 1717f2096b54 ("panic, x86: Fix re-entrance problem due to panic
on NMI") introduced nmi_panic() which prevents concurrent and recursive
execution of panic().  It also saves registers for the crash dump on x86
by later commit 58c5661f2144 ("panic, x86: Allow CPUs to save registers
even if looping in NMI context").

hpwdt driver can call panic() from NMI handler, so replace it with
nmi_panic().  Also, do some cleanups.

Signed-off-by: Hidehiro Kawai <hidehiro.kawai.ez@hitachi.com>
Acked-by: Guenter Roeck <linux@roeck-us.net>
Cc: Thomas Mingarelli <thomas.mingarelli@hpe.com>
Cc: Wim Van Sebroeck <wim@iguana.be>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agoipmi/watchdog: use nmi_panic() when kernel panics in NMI handler
Hidehiro Kawai [Tue, 22 Mar 2016 21:27:21 +0000 (14:27 -0700)]
ipmi/watchdog: use nmi_panic() when kernel panics in NMI handler

Commit 1717f2096b54 ("panic, x86: Fix re-entrance problem due to panic
on NMI") introduced nmi_panic() which prevents concurrent and recursive
execution of panic().  It also saves registers for the crash dump on x86
by later commit 58c5661f2144 ("panic, x86: Allow CPUs to save registers
even if looping in NMI context").

ipmi_watchdog driver can call panic() from NMI handler, so replace it
with nmi_panic().

Signed-off-by: Hidehiro Kawai <hidehiro.kawai.ez@hitachi.com>
Acked-by: Corey Minyard <cminyard@mvista.com>
Acked-by: Guenter Roeck <linux@roeck-us.net>
Reviewed-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agopanic: change nmi_panic from macro to function
Hidehiro Kawai [Tue, 22 Mar 2016 21:27:17 +0000 (14:27 -0700)]
panic: change nmi_panic from macro to function

Commit 1717f2096b54 ("panic, x86: Fix re-entrance problem due to panic
on NMI") and commit 58c5661f2144 ("panic, x86: Allow CPUs to save
registers even if looping in NMI context") introduced nmi_panic() which
prevents concurrent/recursive execution of panic().  It also saves
registers for the crash dump on x86.

However, there are some cases where NMI handlers still use panic().
This patch set partially replaces them with nmi_panic() in those cases.

Even this patchset is applied, some NMI or similar handlers (e.g.  MCE
handler) continue to use panic().  This is because I can't test them
well and actual problems won't happen.  For example, the possibility
that normal panic and panic on MCE happen simultaneously is very low.

This patch (of 3):

Convert nmi_panic() to a proper function and export it instead of
exporting internal implementation details to modules, for obvious
reasons.

Signed-off-by: Hidehiro Kawai <hidehiro.kawai.ez@hitachi.com>
Acked-by: Borislav Petkov <bp@suse.de>
Acked-by: Michal Nazarewicz <mina86@mina86.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Rasmus Villemoes <linux@rasmusvillemoes.dk>
Cc: Nicolas Iooss <nicolas.iooss_linux@m4x.org>
Cc: Javi Merino <javi.merino@arm.com>
Cc: Gobinda Charan Maji <gobinda.cemk07@gmail.com>
Cc: "Steven Rostedt (Red Hat)" <rostedt@goodmis.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Vitaly Kuznetsov <vkuznets@redhat.com>
Cc: HATAYAMA Daisuke <d.hatayama@jp.fujitsu.com>
Cc: Tejun Heo <tj@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agoeventfd: document lockless access in eventfd_poll
Paolo Bonzini [Tue, 22 Mar 2016 21:27:14 +0000 (14:27 -0700)]
eventfd: document lockless access in eventfd_poll

Since commit e22553e2a25e ("eventfd: don't take the spinlock in
eventfd_poll", 2015-02-17), eventfd is reading ctx->count outside
ctx->wqh.lock.

However, things aren't as simple as the read barrier in eventfd_poll
would suggest.  In fact, the read barrier, besides lacking a comment, is
not paired in any obvious manner with another read barrier, and it is
pointless because it is sitting between a write (deep in poll_wait) and
the read of ctx->count.  The read barrier is acting just as a compiler
barrier, for which we can use READ_ONCE instead.  This is what the code
change in this patch does.

The documentation change is just as important, however.  The question,
posed by Andrea Arcangeli, is then why the thing is safe on
architectures where spin_unlock does not imply a store-load memory
barrier.  The answer is that it's safe because writes of ctx->count use
the same lock as poll_wait, and hence an acquire barrier implicit in
poll_wait provides the necessary synchronization between eventfd_poll
and callers of wake_up_locked_poll.  This is sort of mentioned in the
commit message with respect to eventfd_ctx_read ("eventfd_read is
similar, it will do a single decrement with the lock held") but it
applies to all other callers too.  It's tricky enough that it should be
documented in the code.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Andrea Arcangeli <aarcange@redhat.com>
Cc: Chris Mason <clm@fb.com>
Cc: Davide Libenzi <davidel@xmailserver.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agocred/userns: define current_user_ns() as a function
Arnd Bergmann [Tue, 22 Mar 2016 21:27:11 +0000 (14:27 -0700)]
cred/userns: define current_user_ns() as a function

The current_user_ns() macro currently returns &init_user_ns when user
namespaces are disabled, and that causes several warnings when building
with gcc-6.0 in code that compares the result of the macro to
&init_user_ns itself:

  fs/xfs/xfs_ioctl.c: In function 'xfs_ioctl_setattr_check_projid':
  fs/xfs/xfs_ioctl.c:1249:22: error: self-comparison always evaluates to true [-Werror=tautological-compare]
    if (current_user_ns() == &init_user_ns)

This is a legitimate warning in principle, but here it isn't really
helpful, so I'm reprasing the definition in a way that shuts up the
warning.  Apparently gcc only warns when comparing identical literals,
but it can figure out that the result of an inline function can be
identical to a constant expression in order to optimize a condition yet
not warn about the fact that the condition is known at compile time.
This is exactly what we want here, and it looks reasonable because we
generally prefer inline functions over macros anyway.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: Serge Hallyn <serge.hallyn@canonical.com>
Cc: David Howells <dhowells@redhat.com>
Cc: Yaowei Bai <baiyaowei@cmss.chinamobile.com>
Cc: James Morris <james.l.morris@oracle.com>
Cc: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agorapidio: add mport char device driver
Alexandre Bounine [Tue, 22 Mar 2016 21:27:08 +0000 (14:27 -0700)]
rapidio: add mport char device driver

Add mport character device driver to provide user space interface to
basic RapidIO subsystem operations.

See included Documentation/rapidio/mport_cdev.txt for more details.

[akpm@linux-foundation.org: fix printk warning on i386]
[dan.carpenter@oracle.com: mport_cdev: fix some error codes]
Signed-off-by: Alexandre Bounine <alexandre.bounine@idt.com>
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Tested-by: Barry Wood <barry.wood@idt.com>
Cc: Matt Porter <mporter@kernel.crashing.org>
Cc: Aurelien Jacquiot <a-jacquiot@ti.com>
Cc: Andre van Herk <andre.van.herk@prodrive-technologies.com>
Cc: Barry Wood <barry.wood@idt.com>
Cc: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agorapidio/tsi721_dma: fix hardware error handling
Alexandre Bounine [Tue, 22 Mar 2016 21:27:05 +0000 (14:27 -0700)]
rapidio/tsi721_dma: fix hardware error handling

Add DMA channel re-initialization after an error to avoid termination of
all pending transfer requests.

Signed-off-by: Alexandre Bounine <alexandre.bounine@idt.com>
Reported-by: Barry Wood <barry.wood@idt.com>
Tested-by: Barry Wood <barry.wood@idt.com>
Cc: Matt Porter <mporter@kernel.crashing.org>
Cc: Aurelien Jacquiot <a-jacquiot@ti.com>
Cc: Andre van Herk <andre.van.herk@prodrive-technologies.com>
Cc: Barry Wood <barry.wood@idt.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agorapidio/tsi721_dma: fix synchronization issues
Alexandre Bounine [Tue, 22 Mar 2016 21:27:02 +0000 (14:27 -0700)]
rapidio/tsi721_dma: fix synchronization issues

Fix synchronization issues found during testing using multiple DMA
transfer requests to the same channel:

 - lost MSI-X interrupt notifications
 - non-synchronized attempts to start DMA channel HW resulting in error
   message from the driver
 - cookie tracking/update race conditions resulting in incorrect DMA
   transfer status report

Signed-off-by: Alexandre Bounine <alexandre.bounine@idt.com>
Reported-by: Barry Wood <barry.wood@idt.com>
Tested-by: Barry Wood <barry.wood@idt.com>
Cc: Matt Porter <mporter@kernel.crashing.org>
Cc: Aurelien Jacquiot <a-jacquiot@ti.com>
Cc: Andre van Herk <andre.van.herk@prodrive-technologies.com>
Cc: Barry Wood <barry.wood@idt.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agorapidio/tsi721_dma: update error reporting from prep_sg callback
Alexandre Bounine [Tue, 22 Mar 2016 21:26:59 +0000 (14:26 -0700)]
rapidio/tsi721_dma: update error reporting from prep_sg callback

Switch to returning error-valued pointer instead of simple NULL pointer.
This allows to properly identify situation when request queue is full
and therefore gives to upper layer an option to retry operation later.

Signed-off-by: Alexandre Bounine <alexandre.bounine@idt.com>
Cc: Matt Porter <mporter@kernel.crashing.org>
Cc: Aurelien Jacquiot <a-jacquiot@ti.com>
Cc: Andre van Herk <andre.van.herk@prodrive-technologies.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agorapidio/tsi721: add filtered debug output
Alexandre Bounine [Tue, 22 Mar 2016 21:26:56 +0000 (14:26 -0700)]
rapidio/tsi721: add filtered debug output

Replace "all-or-nothing" debug output with controlled debug output using
functional block masks.  This allows run time control of debug messages
through 'dbg_level' module parameter.

Signed-off-by: Alexandre Bounine <alexandre.bounine@idt.com>
Cc: Matt Porter <mporter@kernel.crashing.org>
Cc: Aurelien Jacquiot <a-jacquiot@ti.com>
Cc: Andre van Herk <andre.van.herk@prodrive-technologies.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agorapidio/tsi721: add outbound windows mapping support
Alexandre Bounine [Tue, 22 Mar 2016 21:26:53 +0000 (14:26 -0700)]
rapidio/tsi721: add outbound windows mapping support

Add device-specific callback functions to support outbound windows
mapping and release.

Signed-off-by: Alexandre Bounine <alexandre.bounine@idt.com>
Cc: Matt Porter <mporter@kernel.crashing.org>
Cc: Aurelien Jacquiot <a-jacquiot@ti.com>
Cc: Andre van Herk <andre.van.herk@prodrive-technologies.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agorapidio: add outbound window support
Alexandre Bounine [Tue, 22 Mar 2016 21:26:50 +0000 (14:26 -0700)]
rapidio: add outbound window support

Add RapidIO controller (mport) outbound window configuration operations.

This patch is a part of the original patch submitted by Li Yang:

   https://lists.ozlabs.org/pipermail/linuxppc-dev/2009-April/071210.html

For some reason the original part was not applied to mainline code
tree.  The inbound window mapping part has been applied later during
tsi721 mport driver submission.  Now goes the second part with
corresponding HW support.

Signed-off-by: Alexandre Bounine <alexandre.bounine@idt.com>
Cc: Matt Porter <mporter@kernel.crashing.org>
Cc: Li Yang <leoli@freescale.com>
Cc: Aurelien Jacquiot <a-jacquiot@ti.com>
Cc: Andre van Herk <andre.van.herk@prodrive-technologies.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agorapidio/tsi721: fix locking in OB_MSG processing
Alexandre Bounine [Tue, 22 Mar 2016 21:26:47 +0000 (14:26 -0700)]
rapidio/tsi721: fix locking in OB_MSG processing

- Add spinlock protection into outbound message queuing routine.

- Change outbound message interrupt handler to avoid deadlock when
  calling registered callback routine.

- Allow infinite retries for outbound messages to avoid retry threshold
  error signaling in systems with nodes that have slow message receive
  queue processing.

Signed-off-by: Alexandre Bounine <alexandre.bounine@idt.com>
Cc: Matt Porter <mporter@kernel.crashing.org>
Cc: Aurelien Jacquiot <a-jacquiot@ti.com>
Cc: Andre van Herk <andre.van.herk@prodrive-technologies.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agorapidio: add global inbound port write interfaces
Alexandre Bounine [Tue, 22 Mar 2016 21:26:44 +0000 (14:26 -0700)]
rapidio: add global inbound port write interfaces

Add new Port Write handler registration interfaces that attach PW
handlers to local mport device objects.  This is different from old
interface that attaches PW callback to individual RapidIO device.  The
new interfaces are intended for use for common event handling (e.g.
hot-plug notifications) while the old interface is available for
individual device drivers.

This patch is based on patch proposed by Andre van Herk but preserves
existing per-device interface and adds lock protection for list
handling.

Signed-off-by: Alexandre Bounine <alexandre.bounine@idt.com>
Cc: Matt Porter <mporter@kernel.crashing.org>
Cc: Aurelien Jacquiot <a-jacquiot@ti.com>
Cc: Andre van Herk <andre.van.herk@prodrive-technologies.com>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agorapidio: move rio_pw_enable into core code
Alexandre Bounine [Tue, 22 Mar 2016 21:26:41 +0000 (14:26 -0700)]
rapidio: move rio_pw_enable into core code

Make rio_pw_enable() routine available to other RapidIO drivers.

Signed-off-by: Alexandre Bounine <alexandre.bounine@idt.com>
Cc: Matt Porter <mporter@kernel.crashing.org>
Cc: Aurelien Jacquiot <a-jacquiot@ti.com>
Cc: Andre van Herk <andre.van.herk@prodrive-technologies.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agorapidio: move rio_local_set_device_id function to the common core
Alexandre Bounine [Tue, 22 Mar 2016 21:26:38 +0000 (14:26 -0700)]
rapidio: move rio_local_set_device_id function to the common core

Make function rio_local_set_device_id() common for all components of
RapidIO subsystem.

Signed-off-by: Alexandre Bounine <alexandre.bounine@idt.com>
Cc: Matt Porter <mporter@kernel.crashing.org>
Cc: Aurelien Jacquiot <a-jacquiot@ti.com>
Cc: Andre van Herk <andre.van.herk@prodrive-technologies.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agorapidio: add lock protection for doorbell list
Alexandre Bounine [Tue, 22 Mar 2016 21:26:35 +0000 (14:26 -0700)]
rapidio: add lock protection for doorbell list

Add lock protection around doorbell list handling to prevent list
corruption on SMP platforms.

Signed-off-by: Alexandre Bounine <alexandre.bounine@idt.com>
Cc: Matt Porter <mporter@kernel.crashing.org>
Cc: Aurelien Jacquiot <a-jacquiot@ti.com>
Cc: Andre van Herk <andre.van.herk@prodrive-technologies.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agorapidio/rionet: add mport removal handling
Alexandre Bounine [Tue, 22 Mar 2016 21:26:32 +0000 (14:26 -0700)]
rapidio/rionet: add mport removal handling

Add handling of a local mport device removal.

RIONET driver registers itself as class interface that supports only
removal notification, 'add_device' callback is not provided because
RIONET network device can be initialized only after enumeration is
completed and the existing method (using remote peer addition) satisfies
this condition.

Signed-off-by: Alexandre Bounine <alexandre.bounine@idt.com>
Cc: Matt Porter <mporter@kernel.crashing.org>
Cc: Aurelien Jacquiot <a-jacquiot@ti.com>
Cc: Andre van Herk <andre.van.herk@prodrive-technologies.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agorapidio/rionet: add locking into add/remove device
Alexandre Bounine [Tue, 22 Mar 2016 21:26:29 +0000 (14:26 -0700)]
rapidio/rionet: add locking into add/remove device

Add spinlock protection when handling list of connected peers and
ability to handle new peer device addition after the RIONET device was
open.  Before his update RIONET was sending JOIN requests only when it
have been opened, peer devices added later have been missing from this
process.

Signed-off-by: Alexandre Bounine <alexandre.bounine@idt.com>
Cc: Matt Porter <mporter@kernel.crashing.org>
Cc: Aurelien Jacquiot <a-jacquiot@ti.com>
Cc: Andre van Herk <andre.van.herk@prodrive-technologies.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agopowerpc/fsl_rio: changes to mport registration
Alexandre Bounine [Tue, 22 Mar 2016 21:26:26 +0000 (14:26 -0700)]
powerpc/fsl_rio: changes to mport registration

Change mport object initialization/registration sequence to match
reworked version of rio_register_mport() in the core code.

Signed-off-by: Alexandre Bounine <alexandre.bounine@idt.com>
Cc: Matt Porter <mporter@kernel.crashing.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Aurelien Jacquiot <a-jacquiot@ti.com>
Cc: Andre van Herk <andre.van.herk@prodrive-technologies.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agorapidio/tsi721: add HW specific mport removal
Alexandre Bounine [Tue, 22 Mar 2016 21:26:23 +0000 (14:26 -0700)]
rapidio/tsi721: add HW specific mport removal

Add hardware-specific device removal support for Tsi721 PCIe-to-RapidIO
bridge.  To avoid excessive data type conversions, parameters passed to
some internal functions have been revised.  Dynamic memory allocations
of rio_mport and rio_ops have been replaced to reduce references between
data structures.

Signed-off-by: Alexandre Bounine <alexandre.bounine@idt.com>
Cc: Matt Porter <mporter@kernel.crashing.org>
Cc: Aurelien Jacquiot <a-jacquiot@ti.com>
Cc: Andre van Herk <andre.van.herk@prodrive-technologies.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agorapidio: add core mport removal support
Alexandre Bounine [Tue, 22 Mar 2016 21:26:20 +0000 (14:26 -0700)]
rapidio: add core mport removal support

Add common mport removal support functions into the RapidIO subsystem
core.

Changes to the existing mport registration process have been made to
avoid race conditions with active subsystem interfaces immediately after
mport device registration: part of initialization code from
rio_register_mport() have been moved into separate function
rio_mport_initialize() to allow to perform mport registration as the
final step of setup process.

Signed-off-by: Alexandre Bounine <alexandre.bounine@idt.com>
Cc: Matt Porter <mporter@kernel.crashing.org>
Cc: Aurelien Jacquiot <a-jacquiot@ti.com>
Cc: Andre van Herk <andre.van.herk@prodrive-technologies.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agorapidio: move net allocation into core code
Alexandre Bounine [Tue, 22 Mar 2016 21:26:17 +0000 (14:26 -0700)]
rapidio: move net allocation into core code

Make net allocation/release routines available to all components of
RapidIO subsystem by moving code from rio-scan enumerator.

Make destination ID allocation method private to existing enumerator
because other enumeration methods can use their own algorithm.

Setup net device object as a parent of all RapidIO devices residing in
it and register net as a child of active mport device.

Signed-off-by: Alexandre Bounine <alexandre.bounine@idt.com>
Cc: Matt Porter <mporter@kernel.crashing.org>
Cc: Aurelien Jacquiot <a-jacquiot@ti.com>
Cc: Andre van Herk <andre.van.herk@prodrive-technologies.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agorapidio: rework common RIO device add/delete routines
Alexandre Bounine [Tue, 22 Mar 2016 21:26:14 +0000 (14:26 -0700)]
rapidio: rework common RIO device add/delete routines

This patch moves per-net device list handling from rio-scan to common
RapidIO core and adds a matching device deletion routine.  This makes
device object creation/removal available to other implementations of
enumeration/discovery process.

Signed-off-by: Alexandre Bounine <alexandre.bounine@idt.com>
Cc: Matt Porter <mporter@kernel.crashing.org>
Cc: Aurelien Jacquiot <a-jacquiot@ti.com>
Cc: Andre van Herk <andre.van.herk@prodrive-technologies.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agorapidio/rionet: add shutdown event handling
Alexandre Bounine [Tue, 22 Mar 2016 21:26:11 +0000 (14:26 -0700)]
rapidio/rionet: add shutdown event handling

Add shutdown notification handler which terminates active connections
with remote RapidIO nodes.  This prevents remote nodes from sending
packets to the powered off node and eliminates hardware error events on
remote nodes.

Signed-off-by: Alexandre Bounine <alexandre.bounine@idt.com>
Cc: Matt Porter <mporter@kernel.crashing.org>
Cc: Aurelien Jacquiot <a-jacquiot@ti.com>
Cc: Andre van Herk <andre.van.herk@prodrive-technologies.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agorapidio/tsi721: add shutdown notification callback
Alexandre Bounine [Tue, 22 Mar 2016 21:26:08 +0000 (14:26 -0700)]
rapidio/tsi721: add shutdown notification callback

Add device driver specific shutdown notification callback.

Signed-off-by: Alexandre Bounine <alexandre.bounine@idt.com>
Cc: Matt Porter <mporter@kernel.crashing.org>
Cc: Aurelien Jacquiot <a-jacquiot@ti.com>
Cc: Andre van Herk <andre.van.herk@prodrive-technologies.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agorapidio: add shutdown notification for RapidIO devices
Alexandre Bounine [Tue, 22 Mar 2016 21:26:05 +0000 (14:26 -0700)]
rapidio: add shutdown notification for RapidIO devices

Add bus-specific callback to stop RapidIO devices during a system
shutdown.

Signed-off-by: Alexandre Bounine <alexandre.bounine@idt.com>
Cc: Matt Porter <mporter@kernel.crashing.org>
Cc: Aurelien Jacquiot <a-jacquiot@ti.com>
Cc: Andre van Herk <andre.van.herk@prodrive-technologies.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agorapidio/tsi721: add query_mport callback
Alexandre Bounine [Tue, 22 Mar 2016 21:26:02 +0000 (14:26 -0700)]
rapidio/tsi721: add query_mport callback

Add device-specific implementation of query_mport callback function.

Signed-off-by: Alexandre Bounine <alexandre.bounine@idt.com>
Cc: Matt Porter <mporter@kernel.crashing.org>
Cc: Aurelien Jacquiot <a-jacquiot@ti.com>
Cc: Andre van Herk <andre.van.herk@prodrive-technologies.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agorapidio: add query_mport operation
Alexandre Bounine [Tue, 22 Mar 2016 21:26:00 +0000 (14:26 -0700)]
rapidio: add query_mport operation

Add mport query operation to report master port RapidIO capabilities and
run time configuration to upper level drivers.

Signed-off-by: Alexandre Bounine <alexandre.bounine@idt.com>
Cc: Matt Porter <mporter@kernel.crashing.org>
Cc: Aurelien Jacquiot <a-jacquiot@ti.com>
Cc: Andre van Herk <andre.van.herk@prodrive-technologies.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agorapidio/tsi721_dma: fix pending transaction queue handling
Alexandre Bounine [Tue, 22 Mar 2016 21:25:57 +0000 (14:25 -0700)]
rapidio/tsi721_dma: fix pending transaction queue handling

Fix pending DMA request queue handling to avoid broken ordering during
concurrent request submissions.

Signed-off-by: Alexandre Bounine <alexandre.bounine@idt.com>
Cc: Matt Porter <mporter@kernel.crashing.org>
Cc: Aurelien Jacquiot <a-jacquiot@ti.com>
Cc: Andre van Herk <andre.van.herk@prodrive-technologies.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agorapidio/tsi721: add option to configure direct mapping of IB window
Alexandre Bounine [Tue, 22 Mar 2016 21:25:54 +0000 (14:25 -0700)]
rapidio/tsi721: add option to configure direct mapping of IB window

Add an option to configure mapping of Inbound Window without RIO-to-PCIe
address translation.

If a local memory buffer is not properly aligned to meet HW requirements
for RapidIO address mapping with address translation, caller can request
an inbound window with matching RapidIO address assigned to it.  This
implementation selects RapidIO base address and size for inbound window
that are capable to accommodate the local memory buffer.

Signed-off-by: Alexandre Bounine <alexandre.bounine@idt.com>
Cc: Matt Porter <mporter@kernel.crashing.org>
Cc: Aurelien Jacquiot <a-jacquiot@ti.com>
Cc: Andre van Herk <andre.van.herk@prodrive-technologies.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agorapidio/tsi721: add check for overlapped IB window mappings
Alexandre Bounine [Tue, 22 Mar 2016 21:25:51 +0000 (14:25 -0700)]
rapidio/tsi721: add check for overlapped IB window mappings

Add check for attempts to request mapping of inbound RapidIO address
space that overlaps with existing active mapping windows.

Tsi721 device does not support overlapped inbound windows and SRIO
address decoding behavior is not defined in such cases.

This patch is applicable to kernel versions starting from v3.7.

Signed-off-by: Alexandre Bounine <alexandre.bounine@idt.com>
Cc: Matt Porter <mporter@kernel.crashing.org>
Cc: Aurelien Jacquiot <a-jacquiot@ti.com>
Cc: Andre van Herk <andre.van.herk@prodrive-technologies.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agorapidio/tsi721: fix hardcoded MRRS setting
Alexandre Bounine [Tue, 22 Mar 2016 21:25:48 +0000 (14:25 -0700)]
rapidio/tsi721: fix hardcoded MRRS setting

Remove use of hardcoded setting for Maximum Read Request Size (MRRS)
value and use one set by PCIe bus driver.

Using hardcoded value can cause PCIe bus errors on platforms that have
tsi721 device on PCIe path that allows only smaller read request sizes.

This fix is applicable to kernel versions starting from v3.2.

Signed-off-by: Alexandre Bounine <alexandre.bounine@idt.com>
Cc: Matt Porter <mporter@kernel.crashing.org>
Cc: Aurelien Jacquiot <a-jacquiot@ti.com>
Cc: Andre van Herk <andre.van.herk@prodrive-technologies.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agorapidio/rionet: add capability to change MTU
Aurelien Jacquiot [Tue, 22 Mar 2016 21:25:45 +0000 (14:25 -0700)]
rapidio/rionet: add capability to change MTU

These patches are the result of extensive collaboration within the
RapidIO.org Software Task Group between Texas Instruments, Freescale,
Prodrive Technologies, Nokia Networks, BAE and IDT.  Additional input
was received from other members of RapidIO.org.  The objective was to
create a character mode driver interface which exposes the capabilities
of RapidIO devices directly to applications, in a manner that allows the
numerous and varied RapidIO implementations to interoperate.

The Software Task Group has also developed fabric management, Remote
Memory Access, and sockets applications which make use of these
interfaces in user space.  Intensive testing with these applications
prompted the RapidIO subsystem updates provided within this set of
patches.

This patch (of 29):

Replace default Ethernet-specific routine by the custom one to allow
setting of larger MTU supported by RapidIO messaging (max RIO packet
size is 4096 bytes).

Signed-off-by: Aurelien Jacquiot <a-jacquiot@ti.com>
Signed-off-by: Alexandre Bounine <alexandre.bounine@idt.com>
Cc: Matt Porter <mporter@kernel.crashing.org>
Cc: Andre van Herk <andre.van.herk@prodrive-technologies.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agorapidio/rionet: fix deadlock on SMP
Aurelien Jacquiot [Tue, 22 Mar 2016 21:25:42 +0000 (14:25 -0700)]
rapidio/rionet: fix deadlock on SMP

Fix deadlocking during concurrent receive and transmit operations on SMP
platforms caused by the use of incorrect lock: on transmit 'tx_lock'
spinlock should be used instead of 'lock' which is used for receive
operation.

This fix is applicable to kernel versions starting from v2.15.

Signed-off-by: Aurelien Jacquiot <a-jacquiot@ti.com>
Signed-off-by: Alexandre Bounine <alexandre.bounine@idt.com>
Cc: Matt Porter <mporter@kernel.crashing.org>
Cc: Andre van Herk <andre.van.herk@prodrive-technologies.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agocpumask: remove incorrect information from comment
Eric Biggers [Tue, 22 Mar 2016 21:25:39 +0000 (14:25 -0700)]
cpumask: remove incorrect information from comment

Since commit cdfdef75e795 ("cpumask: only allocate nr_cpumask_bits."),
this comment above cpumask_size() is no longer relevant.

Signed-off-by: Eric Biggers <ebiggers3@gmail.com>
Cc: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agofs/coredump: prevent fsuid=0 dumps into user-controlled directories
Jann Horn [Tue, 22 Mar 2016 21:25:36 +0000 (14:25 -0700)]
fs/coredump: prevent fsuid=0 dumps into user-controlled directories

This commit fixes the following security hole affecting systems where
all of the following conditions are fulfilled:

 - The fs.suid_dumpable sysctl is set to 2.
 - The kernel.core_pattern sysctl's value starts with "/". (Systems
   where kernel.core_pattern starts with "|/" are not affected.)
 - Unprivileged user namespace creation is permitted. (This is
   true on Linux >=3.8, but some distributions disallow it by
   default using a distro patch.)

Under these conditions, if a program executes under secure exec rules,
causing it to run with the SUID_DUMP_ROOT flag, then unshares its user
namespace, changes its root directory and crashes, the coredump will be
written using fsuid=0 and a path derived from kernel.core_pattern - but
this path is interpreted relative to the root directory of the process,
allowing the attacker to control where a coredump will be written with
root privileges.

To fix the security issue, always interpret core_pattern for dumps that
are written under SUID_DUMP_ROOT relative to the root directory of init.

Signed-off-by: Jann Horn <jann@thejh.net>
Acked-by: Kees Cook <keescook@chromium.org>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Oleg Nesterov <oleg@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>
8 years agoptrace: change __ptrace_unlink() to clear ->ptrace under ->siglock
Oleg Nesterov [Tue, 22 Mar 2016 21:25:33 +0000 (14:25 -0700)]
ptrace: change __ptrace_unlink() to clear ->ptrace under ->siglock

This test-case (simplified version of generated by syzkaller)

#include <unistd.h>
#include <sys/ptrace.h>
#include <sys/wait.h>

void test(void)
{
for (;;) {
if (fork()) {
wait(NULL);
continue;
}

ptrace(PTRACE_SEIZE, getppid(), 0, 0);
ptrace(PTRACE_INTERRUPT, getppid(), 0, 0);
_exit(0);
}
}

int main(void)
{
int np;

for (np = 0; np < 8; ++np)
if (!fork())
test();

while (wait(NULL) > 0)
;
return 0;
}

triggers the 2nd WARN_ON_ONCE(!signr) warning in do_jobctl_trap().  The
problem is that __ptrace_unlink() clears task->jobctl under siglock but
task->ptrace is cleared without this lock held; this fools the "else"
branch which assumes that !PT_SEIZED means PT_PTRACED.

Note also that most of other PTRACE_SEIZE checks can race with detach
from the exiting tracer too.  Say, the callers of ptrace_trap_notify()
assume that SEIZED can't go away after it was checked.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Reported-by: Dmitry Vyukov <dvyukov@google.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: syzkaller <syzkaller@googlegroups.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agofat: add config option to set UTF-8 mount option by default
Maciej S. Szmigiero [Tue, 22 Mar 2016 21:25:30 +0000 (14:25 -0700)]
fat: add config option to set UTF-8 mount option by default

FAT has long supported its own default file name encoding config
setting, separate from CONFIG_NLS_DEFAULT.

However, if UTF-8 encoded file names are desired FAT character set
should not be set to utf8 since this would make file names case
sensitive even if case insensitive matching is requested.  Instead,
"utf8" mount options should be provided to enable UTF-8 file names in
FAT file system.

Unfortunately, there was no possibility to set the default value of this
option so on UTF-8 system "utf8" mount option had to be added manually
to most FAT mounts.

This patch adds config option to set such default value.

Signed-off-by: Maciej S. Szmigiero <mail@maciej.szmigiero.name>
Acked-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agox86/compat: remove is_compat_task()
Andy Lutomirski [Tue, 22 Mar 2016 21:25:27 +0000 (14:25 -0700)]
x86/compat: remove is_compat_task()

x86's is_compat_task always checked the current syscall type, not the
task type.  It has no non-arch users any more, so just remove it to
avoid confusion.

On x86, nothing should really be checking the task ABI.  There are
legitimate users for the syscall ABI and for the mm ABI.

Signed-off-by: Andy Lutomirski <luto@kernel.org>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agodrivers/hid/uhid.c: check write() bitness using in_compat_syscall
Andy Lutomirski [Tue, 22 Mar 2016 21:25:24 +0000 (14:25 -0700)]
drivers/hid/uhid.c: check write() bitness using in_compat_syscall

uhid changes the format expected in write() depending on bitness.  It
should check the syscall bitness directly.

Signed-off-by: Andy Lutomirski <luto@kernel.org>
Cc: David Herrmann <dh.herrmann@googlemail.com>
Acked-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agoinput: redefine INPUT_COMPAT_TEST as in_compat_syscall()
Andy Lutomirski [Tue, 22 Mar 2016 21:25:21 +0000 (14:25 -0700)]
input: redefine INPUT_COMPAT_TEST as in_compat_syscall()

The input compat code should work like all other compat code: for 32-bit
syscalls, use the 32-bit ABI and for 64-bit syscalls, use the 64-bit
ABI.  We have a helper for that (in_compat_syscall()): just use it.

Signed-off-by: Andy Lutomirski <luto@kernel.org>
Cc: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agodrivers/gpu/drm/amd/amdkfd: use in_compat_syscall to check open() caller type
Andy Lutomirski [Tue, 22 Mar 2016 21:25:19 +0000 (14:25 -0700)]
drivers/gpu/drm/amd/amdkfd: use in_compat_syscall to check open() caller type

amdkfd wants to know syscall type, not task type.  Check directly.

Unfortunately, amdkfd is making nasty assumptions that a process'
bitness is a well-defined constant thing.  This isn't the case on x86.
I don't know how much this matters, but this patch has no effect on
generated code on x86, so amdkfd is equally broken with and without this
patch.

Signed-off-by: Andy Lutomirski <luto@kernel.org>
Cc: Oded Gabbay <oded.gabbay@gmail.com>
Cc: David Airlie <airlied@linux.ie>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agodrivers/firmware/efi/efivars.c: use in_compat_syscall() to check for compat callers
Andy Lutomirski [Tue, 22 Mar 2016 21:25:16 +0000 (14:25 -0700)]
drivers/firmware/efi/efivars.c: use in_compat_syscall() to check for compat callers

This should make no difference on any architecture, as x86's historical
is_compat_task behavior really did check whether the calling syscall was
a compat syscall.  x86's is_compat_task is going away, though.

Signed-off-by: Andy Lutomirski <luto@kernel.org>
Reviewed-by: Matt Fleming <matt@codeblueprint.co.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agofirewire: use in_compat_syscall to check ioctl compatness
Andy Lutomirski [Tue, 22 Mar 2016 21:25:13 +0000 (14:25 -0700)]
firewire: use in_compat_syscall to check ioctl compatness

Firewire was using is_compat_task to check whether it was in a compat
ioctl or a non-compat ioctl.  Use is_compat_syscall instead so it works
properly on all architectures.

Signed-off-by: Andy Lutomirski <luto@kernel.org>
Cc: Clemens Ladisch <clemens@ladisch.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agonet/xfrm_user: use in_compat_syscall to deny compat syscalls
Andy Lutomirski [Tue, 22 Mar 2016 21:25:10 +0000 (14:25 -0700)]
net/xfrm_user: use in_compat_syscall to deny compat syscalls

The code wants to prevent compat code from receiving messages.  Use
in_compat_syscall for this.

Signed-off-by: Andy Lutomirski <luto@kernel.org>
Cc: Steffen Klassert <steffen.klassert@secunet.com>
Cc: Herbert Xu <herbert@gondor.apana.org.au>
Cc: "David S. Miller" <davem@davemloft.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agonet/sctp: use in_compat_syscall for sctp_getsockopt_connectx3
Andy Lutomirski [Tue, 22 Mar 2016 21:25:07 +0000 (14:25 -0700)]
net/sctp: use in_compat_syscall for sctp_getsockopt_connectx3

SCTP unfortunately has a different ABI for SCTP_SOCKOPT_CONNECTX3 for
32-bit and 64-bit callers.  Use in_compat_syscall to correctly
distinguish them on all architectures.

Signed-off-by: Andy Lutomirski <luto@kernel.org>
Cc: Vlad Yasevich <vyasevich@gmail.com>
Cc: Neil Horman <nhorman@tuxdriver.com>
Cc: David Miller <davem@davemloft.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agoext4: in ext4_dir_llseek, check syscall bitness directly
Andy Lutomirski [Tue, 22 Mar 2016 21:25:04 +0000 (14:25 -0700)]
ext4: in ext4_dir_llseek, check syscall bitness directly

ext4 treats directory offsets differently for 32-bit and 64-bit callers.
Check the caller type using in_compat_syscall, not is_compat_task.  This
changes behavior on SPARC slightly.

Signed-off-by: Andy Lutomirski <luto@kernel.org>
Cc: "Theodore Ts'o" <tytso@mit.edu>
Cc: Andreas Dilger <adilger.kernel@dilger.ca>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agostaging/lustre: switch from is_compat_task to in_compat_syscall
Andy Lutomirski [Tue, 22 Mar 2016 21:25:01 +0000 (14:25 -0700)]
staging/lustre: switch from is_compat_task to in_compat_syscall

AFAICT, lustre is trying to determine syscall bitness.  Use the new
accessor.

Signed-off-by: Andy Lutomirski <luto@kernel.org>
Cc: Oleg Drokin <oleg.drokin@intel.com>
Cc: Andreas Dilger <adilger.kernel@dilger.ca>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agoauditsc: for seccomp events, log syscall compat state using in_compat_syscall
Andy Lutomirski [Tue, 22 Mar 2016 21:24:58 +0000 (14:24 -0700)]
auditsc: for seccomp events, log syscall compat state using in_compat_syscall

Except on SPARC, this is what the code always did.  SPARC compat seccomp
was buggy, although the impact of the bug was limited because SPARC
32-bit and 64-bit syscall numbers are the same.

Signed-off-by: Andy Lutomirski <luto@kernel.org>
Cc: Paul Moore <paul@paul-moore.com>
Cc: Eric Paris <eparis@redhat.com>
Cc: David Miller <davem@davemloft.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agoptrace: in PEEK_SIGINFO, check syscall bitness, not task bitness
Andy Lutomirski [Tue, 22 Mar 2016 21:24:55 +0000 (14:24 -0700)]
ptrace: in PEEK_SIGINFO, check syscall bitness, not task bitness

Users of the 32-bit ptrace() ABI expect the full 32-bit ABI.  siginfo
translation should check ptrace() ABI, not caller task ABI.

This is an ABI change on SPARC.  Let's hope that no one relied on the
old buggy ABI.

Signed-off-by: Andy Lutomirski <luto@kernel.org>
Cc: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agoseccomp: check in_compat_syscall, not is_compat_task, in strict mode
Andy Lutomirski [Tue, 22 Mar 2016 21:24:52 +0000 (14:24 -0700)]
seccomp: check in_compat_syscall, not is_compat_task, in strict mode

Seccomp wants to know the syscall bitness, not the caller task bitness,
when it selects the syscall whitelist.

As far as I know, this makes no difference on any architecture, so it's
not a security problem.  (It generates identical code everywhere except
sparc, and, on sparc, the syscall numbering is the same for both ABIs.)

Signed-off-by: Andy Lutomirski <luto@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agosparc/syscall: fix syscall_get_arch
Andy Lutomirski [Tue, 22 Mar 2016 21:24:49 +0000 (14:24 -0700)]
sparc/syscall: fix syscall_get_arch

Sparc's syscall_get_arch was buggy: it returned the task arch, not the
syscall arch.  This could confuse seccomp and audit.

I don't think this is as bad for seccomp as it looks: sparc's 32-bit and
64-bit syscalls are numbered the same.

Signed-off-by: Andy Lutomirski <luto@kernel.org>
Cc: David S. Miller <davem@davemloft.net>
Cc: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agosparc/compat: provide an accurate in_compat_syscall implementation
Andy Lutomirski [Tue, 22 Mar 2016 21:24:46 +0000 (14:24 -0700)]
sparc/compat: provide an accurate in_compat_syscall implementation

On sparc64 compat-enabled kernels, any task can make 32-bit and 64-bit
syscalls.  is_compat_task returns true in 32-bit tasks, which does not
necessarily imply that the current syscall is 32-bit.

Provide an in_compat_syscall implementation that checks whether the
current syscall is compat.

As far as I know, sparc is the only architecture on which is_compat_task
checks the compat status of the task and on which the compat status of a
syscall can differ from the compat status of the task.  On x86,
is_compat_task checks the syscall type, not the task type.

[akpm@linux-foundation.org: add comment, per Sam]
[akpm@linux-foundation.org: update comment, per Andy]
Signed-off-by: Andy Lutomirski <luto@kernel.org>
Acked-by: David S. Miller <davem@davemloft.net>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: Andy Lutomirski <luto@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agocompat: add in_compat_syscall to ask whether we're in a compat syscall
Andy Lutomirski [Tue, 22 Mar 2016 21:24:43 +0000 (14:24 -0700)]
compat: add in_compat_syscall to ask whether we're in a compat syscall

A lot of code currently abuses is_compat_task to determine this.

Signed-off-by: Andy Lutomirski <luto@kernel.org>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: "Theodore Ts'o" <tytso@mit.edu>
Cc: Andreas Dilger <adilger.kernel@dilger.ca>
Cc: Clemens Ladisch <clemens@ladisch.de>
Cc: David Airlie <airlied@linux.ie>
Cc: David Herrmann <dh.herrmann@googlemail.com>
Cc: David Miller <davem@davemloft.net>
Cc: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Cc: Eric Paris <eparis@redhat.com>
Cc: Herbert Xu <herbert@gondor.apana.org.au>
Cc: Ingo Molnar <mingo@elte.hu>
Acked-by: Jiri Kosina <jkosina@suse.cz>
Cc: Matt Fleming <matt@codeblueprint.co.uk>
Cc: Neil Horman <nhorman@tuxdriver.com>
Cc: Oded Gabbay <oded.gabbay@gmail.com>
Cc: Oleg Drokin <oleg.drokin@intel.com>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Paul Moore <paul@paul-moore.com>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: Steffen Klassert <steffen.klassert@secunet.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Vlad Yasevich <vyasevich@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agokernel/hung_task.c: use timeout diff when timeout is updated
Tetsuo Handa [Tue, 22 Mar 2016 21:24:39 +0000 (14:24 -0700)]
kernel/hung_task.c: use timeout diff when timeout is updated

When new timeout is written to /proc/sys/kernel/hung_task_timeout_secs,
khungtaskd is interrupted and again sleeps for full timeout duration.

This means that hang task will not be checked if new timeout is written
periodically within old timeout duration and/or checking of hang task
will be delayed for up to previous timeout duration.  Fix this by
remembering last time khungtaskd checked hang task.

This change will allow other watchdog tasks (if any) to share khungtaskd
by sleeping for minimal timeout diff of all watchdog tasks.  Doing more
watchdog tasks from khungtaskd will reduce the possibility of printk()
collisions by multiple watchdog threads.

Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Aaron Tomlin <atomlin@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agozram: revive swap_slot_free_notify
Minchan Kim [Tue, 22 Mar 2016 21:24:36 +0000 (14:24 -0700)]
zram: revive swap_slot_free_notify

Commit b430e9d1c6d4 ("remove compressed copy from zram in-memory")
applied swap_slot_free_notify call in *end_swap_bio_read* to remove
duplicated memory between zram and memory.

However, with the introduction of rw_page in zram: 8c7f01025f7b ("zram:
implement rw_page operation of zram"), it became void because rw_page
doesn't need bio.

Memory footprint is really important in embedded platforms which have
small memory, for example, 512M) recently because it could start to kill
processes if memory footprint exceeds some threshold by LMK or some
similar memory management modules.

This patch restores the function for rw_page, thereby eliminating this
duplication.

Signed-off-by: Minchan Kim <minchan@kernel.org>
Cc: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
Cc: karam.lee <karam.lee@lge.com>
Cc: <sangseok.lee@lge.com>
Cc: Chan Jeong <chan.jeong@lge.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agoocfs2: add feature document for online file check
Gang He [Tue, 22 Mar 2016 21:24:33 +0000 (14:24 -0700)]
ocfs2: add feature document for online file check

This document will describe OCFS2 online file check feature.  OCFS2 is
often used in high-availaibility systems.  However, OCFS2 usually
converts the filesystem to read-only when encounters an error.  This may
not be necessary, since turning the filesystem read-only would affect
other running processes as well, decreasing availability.

Then, a mount option (errors=continue) is introduced, which would return
the -EIO errno to the calling process and terminate furhter processing
so that the filesystem is not corrupted further.  The filesystem is not
converted to read-only, and the problematic file's inode number is
reported in the kernel log.  The user can try to check/fix this file via
online filecheck feature.

Signed-off-by: Gang He <ghe@suse.com>
Cc: Mark Fasheh <mfasheh@suse.de>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Junxiao Bi <junxiao.bi@oracle.com>
Cc: Joseph Qi <joseph.qi@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agoocfs2: check/fix inode block for online file check
Gang He [Tue, 22 Mar 2016 21:24:30 +0000 (14:24 -0700)]
ocfs2: check/fix inode block for online file check

Implement online check or fix inode block during reading a inode block
to memory.

Signed-off-by: Gang He <ghe@suse.com>
Reviewed-by: Mark Fasheh <mfasheh@suse.de>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Junxiao Bi <junxiao.bi@oracle.com>
Cc: Joseph Qi <joseph.qi@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agoocfs2: create/remove sysfile for online file check
Gang He [Tue, 22 Mar 2016 21:24:27 +0000 (14:24 -0700)]
ocfs2: create/remove sysfile for online file check

Create online file check sysfile when ocfs2 mount, remove the related
sysfile when ocfs2 umount.

Signed-off-by: Gang He <ghe@suse.com>
Reviewed-by: Mark Fasheh <mfasheh@suse.de>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Junxiao Bi <junxiao.bi@oracle.com>
Cc: Joseph Qi <joseph.qi@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agoocfs2: sysfile interfaces for online file check
Gang He [Tue, 22 Mar 2016 21:24:24 +0000 (14:24 -0700)]
ocfs2: sysfile interfaces for online file check

Implement online file check sysfile interfaces, e.g. how to create the
related sysfile according to device name, how to display/handle file
check request from the sysfile.

Signed-off-by: Gang He <ghe@suse.com>
Reviewed-by: Mark Fasheh <mfasheh@suse.de>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Junxiao Bi <junxiao.bi@oracle.com>
Cc: Joseph Qi <joseph.qi@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agoocfs2: export ocfs2_kset for online file check
Gang He [Tue, 22 Mar 2016 21:24:20 +0000 (14:24 -0700)]
ocfs2: export ocfs2_kset for online file check

When there are errors in the ocfs2 filesystem, they are usually
accompanied by the inode number which caused the error.  This inode
number would be the input to fixing the file.  One of these options
could be considered:

A file in the sys filesytem which would accept inode numbers.  This
could be used to communication back what has to be fixed or is fixed.
You could write:

  $# echo "<inode>" > /sys/fs/ocfs2/devname/filecheck/check

or

  $# echo "<inode>" > /sys/fs/ocfs2/devname/filecheck/fix

Compare with second version, I re-design filecheck sysfs interfaces,
there are three sysfs files (check, fix and set) under filecheck
directory (see above), sysfs will accept only one argument <inode>.
Second, I adjust some code in ocfs2_filecheck_repair_inode_block()
function according to upstream feedback, we cannot just add VALID_FL
flag back as a inode block fix, then we will not fix this field
corruption currently until having a complete solution.  Compare with
first version, I use strncasecmp instead of double strncmp functions.
Second, update the source file contribution vendor.

This patch (of 4):

Export ocfs2_kset object from ocfs2_stackglue kernel module, then online
file check code will create the related sysfiles under ocfs2_kset
object.  We're exporting this because it's built in ocfs2_stackglue.ko.

Signed-off-by: Gang He <ghe@suse.com>
Reviewed-by: Mark Fasheh <mfasheh@suse.de>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Junxiao Bi <junxiao.bi@oracle.com>
Cc: Joseph Qi <joseph.qi@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agoMerge tag 'nfs-for-4.6-1' of git://git.linux-nfs.org/projects/trondmy/linux-nfs
Linus Torvalds [Tue, 22 Mar 2016 20:16:21 +0000 (13:16 -0700)]
Merge tag 'nfs-for-4.6-1' of git://git.linux-nfs.org/projects/trondmy/linux-nfs

Pull NFS client updates from Trond Myklebust:
 "Highlights include:

  Features:
   - Add support for multiple NFSv4.1 callbacks in flight
   - Initial patchset for RPC multipath support
   - Adapt RPC/RDMA to use the new completion queue API

  Bugfixes and cleanups:
   - nfs4: nfs4_ff_layout_prepare_ds should return NULL if connection failed
   - Cleanups to remove nfs_inode_dio_wait and nfs4_file_fsync
   - Fix RPC/RDMA credit accounting
   - Properly handle RDMA_ERROR replies
   - xprtrdma: Do not wait if ib_post_send() fails
   - xprtrdma: Segment head and tail XDR buffers on page boundaries
   - xprtrdma cleanups for dprintk, physical_op_map and unused macros"

* tag 'nfs-for-4.6-1' of git://git.linux-nfs.org/projects/trondmy/linux-nfs: (35 commits)
  nfs/blocklayout: make sure making a aligned read request
  nfs4: nfs4_ff_layout_prepare_ds should return NULL if connection failed
  nfs: remove nfs_inode_dio_wait
  nfs: remove nfs4_file_fsync
  xprtrdma: Use new CQ API for RPC-over-RDMA client send CQs
  xprtrdma: Use an anonymous union in struct rpcrdma_mw
  xprtrdma: Use new CQ API for RPC-over-RDMA client receive CQs
  xprtrdma: Serialize credit accounting again
  xprtrdma: Properly handle RDMA_ERROR replies
  rpcrdma: Add RPCRDMA_HDRLEN_ERR
  xprtrdma: Do not wait if ib_post_send() fails
  xprtrdma: Segment head and tail XDR buffers on page boundaries
  xprtrdma: Clean up dprintk format string containing a newline
  xprtrdma: Clean up physical_op_map()
  xprtrdma: Clean up unused RPCRDMA_INLINE_PAD_THRESH macro
  NFS add callback_ops to nfs4_proc_bind_conn_to_session_callback
  pnfs/NFSv4.1: Add multipath capabilities to pNFS flexfiles servers over NFSv3
  SUNRPC: Allow addition of new transports to a struct rpc_clnt
  NFSv4.1: nfs4_proc_bind_conn_to_session must iterate over all connections
  SUNRPC: Make NFS swap work with multipath
  ...

8 years agoMerge branch 'overlayfs-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mszer...
Linus Torvalds [Tue, 22 Mar 2016 20:11:15 +0000 (13:11 -0700)]
Merge branch 'overlayfs-linus' of git://git./linux/kernel/git/mszeredi/vfs

Pull overlayfs updates from Miklos Szeredi:
 "Various fixes and tweaks"

* 'overlayfs-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mszeredi/vfs:
  ovl: cleanup unused var in rename2
  ovl: rename is_merge to is_lowest
  ovl: fixed coding style warning
  ovl: Ensure upper filesystem supports d_type
  ovl: Warn on copy up if a process has a R/O fd open to the lower file
  ovl: honor flag MS_SILENT at mount
  ovl: verify upper dentry before unlink and rename

8 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mszeredi...
Linus Torvalds [Tue, 22 Mar 2016 20:05:34 +0000 (13:05 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/mszeredi/fuse

Pull fuse update from Miklos Szeredi:
 "This contains direct I/O fixes"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mszeredi/fuse:
  fuse: return patrial success from fuse_direct_io()
  fuse: Add reference counting for fuse_io_priv
  fuse: do not use iocb after it may have been freed