openwrt/staging/blogic.git
9 years agoarm64: compat: Remove incorrect comment in compat_siginfo
Catalin Marinas [Mon, 2 Feb 2015 16:44:39 +0000 (16:44 +0000)]
arm64: compat: Remove incorrect comment in compat_siginfo

The comment was right originally but the _pad array size was wrong. It
was fixed in the meantime but the comment not updated.

Reported-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
9 years agoarm64: Fix section mismatch on alloc_init_p[mu]d()
Catalin Marinas [Thu, 29 Jan 2015 17:33:35 +0000 (17:33 +0000)]
arm64: Fix section mismatch on alloc_init_p[mu]d()

Commit 523d6e9fae93 (arm64:mm: free the useless initial page table)
introduced a BUG_ON checking for the allocation type but it was
referring the early_alloc() function in the __init section. This patch
changes the check to slab_is_available() and also relaxes the BUG to a
WARN_ON_ONCE.

Reported-by: Will Deacon <will.deacon@arm.com>
Acked-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
9 years agoarm64: Avoid breakage caused by .altmacro in fpsimd save/restore macros
Dave P Martin [Thu, 29 Jan 2015 16:24:43 +0000 (16:24 +0000)]
arm64: Avoid breakage caused by .altmacro in fpsimd save/restore macros

Alternate macro mode is not a property of a macro definition, but a
gas runtime state that alters the way macros are expanded for ever
after (until .noaltmacro is seen).

This means that subsequent assembly code that calls other macros can
break if fpsimdmacros.h is included.

Since these instruction sequences are simple (if dull -- but in a
good way), this patch solves the problem by simply expanding the
.irp loops.  The pre-existing fpsimd_{save,restore} macros weren't
rolled with .irp anyway and the sequences affected are short, so
this change restores consistency at little cost.

Signed-off-by: Dave Martin <Dave.Martin@arm.com>
Acked-by: Marc Zyngier <marc.zyngier@arm.com>
Acked-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
9 years agoarm64: mm: use *_sect to check for section maps
Mark Rutland [Tue, 27 Jan 2015 16:36:30 +0000 (16:36 +0000)]
arm64: mm: use *_sect to check for section maps

The {pgd,pud,pmd}_bad family of macros have slightly fuzzy
cross-architecture semantics, and seem to imply a populated entry that
is not a next-level table, rather than a particular type of entry (e.g.
a section map).

In arm64 code, for those cases where we care about whether an entry is a
section mapping, we can instead use the {pud,pmd}_sect macros to
explicitly check for this case. This helps to document precisely what we
care about, making the code easier to read, and allows for future
relaxation of the *_bad macros to check for other "bad" entries.

To that end this patch updates the table dumping and initial table setup
to check for section mappings with {pud,pmd}_sect, and adds/restores
BUG_ON(*_bad((*p)) checks after we've handled the *_sect and *_none
cases so as to catch remaining "bad" cases.

In the fault handling code, show_pte is left with *_bad checks as it
only cares about whether it can walk the next level table, and this path
is used for both kernel and userspace fault handling. The former case
will be followed by a die() where we'll report the address that
triggered the fault, which can be useful context for debugging.

Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Acked-by: Steve Capper <steve.capper@linaro.org>
Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Cc: Kees Cook <keescook@chromium.org>
Cc: Laura Abbott <lauraa@codeaurora.org>
Cc: Will Deacon <will.deacon@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
9 years agoarm64: drop unnecessary cache+tlb maintenance
Mark Rutland [Tue, 27 Jan 2015 16:52:50 +0000 (16:52 +0000)]
arm64: drop unnecessary cache+tlb maintenance

In paging_init, we call flush_cache_all, but this is backed by Set/Way
operations which may not achieve anything in the presence of cache line
migration and/or system caches. If the caches are already in an
inconsistent state at this point, there is nothing we can do (short of
flushing the entire physical address space by VA) to empty architected
and system caches. As such, flush_cache_all only serves to mask other
potential bugs. Hence, this patch removes the boot-time call to
flush_cache_all.

Immediately after the cache maintenance we flush the TLBs, but this is
also unnecessary. Before enabling the MMU, the TLBs are invalidated, and
thus are initially clean. When changing the contents of active tables
(e.g. in fixup_executable() for DEBUG_RODATA) we perform the required
TLB maintenance following the update, and therefore no additional
maintenance is required to ensure the new table entries are in effect.
Since activating the MMU we will not have modified system register
fields permitted to be cached in a TLB, and therefore do not need
maintenance for any cached system register fields. Hence, the TLB flush
is unnecessary.

Shortly after the unnecessary TLB flush, we update TTBR0 to point to an
empty zero page rather than the idmap, and flush the TLBs. This
maintenance is necessary to remove the global idmap entries from the
TLBs (as they would conflict with userspace mappings), and is retained.

Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Acked-by: Marc Zyngier <marc.zyngier@arm.com>
Acked-by: Steve Capper <steve.capper@linaro.org>
Cc: Will Deacon <will.deacon@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
9 years agoarm64:mm: free the useless initial page table
zhichang.yuan [Tue, 9 Dec 2014 07:26:47 +0000 (07:26 +0000)]
arm64:mm: free the useless initial page table

For 64K page system, after mapping a PMD section, the corresponding initial
page table is not needed any more. That page can be freed.

Signed-off-by: Zhichang Yuan <zhichang.yuan@linaro.org>
[catalin.marinas@arm.com: added BUG_ON() to catch late memblock freeing]
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
9 years agoarm64: Enable CPU_IDLE in defconfig
Catalin Marinas [Tue, 27 Jan 2015 18:25:25 +0000 (18:25 +0000)]
arm64: Enable CPU_IDLE in defconfig

This patch enables CPU_IDLE and the generic arm64 cpuidle driver
(ARM64_CPUIDLE).

Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
9 years agoarm64: kernel: remove ARM64_CPU_SUSPEND config option
Lorenzo Pieralisi [Mon, 26 Jan 2015 18:33:44 +0000 (18:33 +0000)]
arm64: kernel: remove ARM64_CPU_SUSPEND config option

ARM64_CPU_SUSPEND config option was introduced to make code providing
context save/restore selectable only on platforms requiring power
management capabilities.

Currently ARM64_CPU_SUSPEND depends on the PM_SLEEP config option which
in turn is set by the SUSPEND config option.

The introduction of CPU_IDLE for arm64 requires that code configured
by ARM64_CPU_SUSPEND (context save/restore) should be compiled in
in order to enable the CPU idle driver to rely on CPU operations
carrying out context save/restore.

The ARM64_CPUIDLE config option (ARM64 generic idle driver) is therefore
forced to select ARM64_CPU_SUSPEND, even if there may be (ie PM_SLEEP)
failed dependencies, which is not a clean way of handling the kernel
configuration option.

For these reasons, this patch removes the ARM64_CPU_SUSPEND config option
and makes the context save/restore dependent on CPU_PM, which is selected
whenever either SUSPEND or CPU_IDLE are configured, cleaning up dependencies
in the process.

This way, code previously configured through ARM64_CPU_SUSPEND is
compiled in whenever a power management subsystem requires it to be
present in the kernel (SUSPEND || CPU_IDLE), which is the behaviour
expected on ARM64 kernels.

The cpu_suspend and cpu_init_idle CPU operations are added only if
CPU_IDLE is selected, since they are CPU_IDLE specific methods and
should be grouped and defined accordingly.

PSCI CPU operations are updated to reflect the introduced changes.

Signed-off-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Krzysztof Kozlowski <k.kozlowski@samsung.com>
Cc: Daniel Lezcano <daniel.lezcano@linaro.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
9 years agoarm64: make sys_call_table const
Mark Rutland [Thu, 8 Jan 2015 11:42:59 +0000 (11:42 +0000)]
arm64: make sys_call_table const

As with x86, mark the sys_call_table const such that it will be placed
in the .rodata section. This will cause attempts to modify the table
(accidental or deliberate) to fail when strict page permissions are in
place. In the absence of strict page permissions, there should be no
functional change.

Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Acked-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
9 years agoarm64: Remove asm/syscalls.h
Catalin Marinas [Tue, 6 Jan 2015 17:01:56 +0000 (17:01 +0000)]
arm64: Remove asm/syscalls.h

This patch moves the sys_rt_sigreturn_wrapper prototype to
arch/arm64/kernel/sys.c and removes the asm/syscalls.h header.

Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
9 years agoarm64: Implement the compat_sys_call_table in C
Catalin Marinas [Tue, 6 Jan 2015 16:42:32 +0000 (16:42 +0000)]
arm64: Implement the compat_sys_call_table in C

Unlike the sys_call_table[], the compat one was implemented in sys32.S
making it impossible to notice discrepancies between the number of
compat syscalls and the __NR_compat_syscalls macro, the latter having to
be defined in asm/unistd.h as including asm/unistd32.h would cause
conflicts on __NR_* definitions. With this patch, incorrect
__NR_compat_syscalls values will result in a build-time error.

Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Suggested-by: Mark Rutland <mark.rutland@arm.com>
Acked-by: Mark Rutland <mark.rutland@arm.com>
9 years agosyscalls: Declare sys_*stat64 prototypes if __ARCH_WANT_(COMPAT_)STAT64
Catalin Marinas [Tue, 6 Jan 2015 16:52:38 +0000 (16:52 +0000)]
syscalls: Declare sys_*stat64 prototypes if __ARCH_WANT_(COMPAT_)STAT64

Currently, the sys_stat64, sys_fstat64 and sys_lstat64 prototpyes are
only declared if BITS_PER_LONG == 32. Following commit 0753f70f07fb
(fs: Build sys_stat64() and friends if __ARCH_WANT_COMPAT_STAT64), the
implementation of these functions is allowed on 64-bit systems for
compat support. The patch changes the condition on the prototype
declaration from BITS_PER_LONG == 32 to defined(__ARCH_WANT_STAT64) ||
defined(__ARCH_WANT_COMPAT_STAT64).

In addition, it moves the sys_fstatat64 prototype under the same #if
block

Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Acked-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Arnd Bergmann <arnd@arndb.de>
9 years agocompat: Declare compat_sys_sigpending and compat_sys_sigprocmask prototypes
Catalin Marinas [Tue, 6 Jan 2015 16:48:36 +0000 (16:48 +0000)]
compat: Declare compat_sys_sigpending and compat_sys_sigprocmask prototypes

__ARCH_WANT_SYS_SIGPENDING or __ARCH_WANT_SYS_SIGPROGMASK may be defined
for compat support but the corresponding prototypes are missing from
linux/compat.h.

Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Acked-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Arnd Bergmann <arnd@arndb.de>
9 years agoarm64: uapi: expose our struct ucontext to the uapi headers
Will Deacon [Fri, 16 Jan 2015 13:52:14 +0000 (13:52 +0000)]
arm64: uapi: expose our struct ucontext to the uapi headers

arm64 defines its own ucontext structure which is incompatible with the
struct defined (and exposed to userspace by) the asm-generic headers.

glibc carries its own struct definition that is compatible with the
arm64 definition, but we should expose our format in the uapi headers in
case other libraries want to make use of the ucontext pushed as part of
an arm64 sigframe.

This patch moves the arm64 asm/ucontext.h to the uapi headers, along
with the necessary #include of linux/types.h.

Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Marcus Shawcroft <marcus.shawcroft@arm.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
9 years agosmp, ARM64: Kill SMP single function call interrupt
Jiang Liu [Fri, 23 Jan 2015 05:36:42 +0000 (05:36 +0000)]
smp, ARM64: Kill SMP single function call interrupt

Commit 9a46ad6d6df3b54 "smp: make smp_call_function_many() use logic
similar to smp_call_function_single()" has unified the way to handle
single and multiple cross-CPU function calls. Now only one interrupt
is needed for architecture specific code to support generic SMP function
call interfaces, so kill the redundant single function call interrupt.

Signed-off-by: Jiang Liu <jiang.liu@linux.intel.com>
Acked-by: Will Deacon <will.deacon@arm.com>
Cc: linux-arm-kernel@lists.infradead.org
Cc: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
9 years agoarm64: Emulate SETEND for AArch32 tasks
Suzuki K. Poulose [Wed, 21 Jan 2015 12:43:11 +0000 (12:43 +0000)]
arm64: Emulate SETEND for AArch32 tasks

Emulate deprecated 'setend' instruction for AArch32 bit tasks.

setend [le/be] - Sets the endianness of EL0

On systems with CPUs which support mixed endian at EL0, the hardware
support for the instruction can be enabled by setting the SCTLR_EL1.SED
bit. Like the other emulated instructions it is controlled by an entry in
/proc/sys/abi/. For more information see :
Documentation/arm64/legacy_instructions.txt

The instruction is emulated by setting/clearing the SPSR_EL1.E bit, which
will be reflected in the PSTATE.E in AArch32 context.

This patch also restores the native endianness for the execution of signal
handlers, since the process could have changed the endianness.

Note: All CPUs on the system must have mixed endian support at EL0. Once the
handler is registered, hotplugging a CPU which doesn't support mixed endian,
could lead to unexpected results/behavior in applications.

Signed-off-by: Suzuki K. Poulose <suzuki.poulose@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Punit Agrawal <punit.agrawal@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
9 years agoarm64: Consolidate hotplug notifier for instruction emulation
Suzuki K. Poulose [Wed, 21 Jan 2015 12:43:10 +0000 (12:43 +0000)]
arm64: Consolidate hotplug notifier for instruction emulation

As of now each insn_emulation has a cpu hotplug notifier that
enables/disables the CPU feature bit for the functionality. This
patch re-arranges the code, such that there is only one notifier
that runs through the list of registered emulation hooks and runs
their corresponding set_hw_mode.

We do nothing when a CPU is dying as we will set the appropriate bits
as it comes back online based on the state of the hooks.

Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Suzuki K. Poulose <suzuki.poulose@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Punit Agrawal <punit.agrawal@arm.com>
[catalin.marinas@arm.com: fix pr_warn compilation error]
[catalin.marinas@arm.com: remove unnecessary "insn" check]
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
9 years agoarm64: Track system support for mixed endian EL0
Suzuki K. Poulose [Wed, 21 Jan 2015 12:43:09 +0000 (12:43 +0000)]
arm64: Track system support for mixed endian EL0

This patch keeps track of the mixed endian EL0 support across
the system and provides helper functions to export it. The status
is a boolean indicating whether all the CPUs on the system supports
mixed endian at EL0.

Signed-off-by: Suzuki K. Poulose <suzuki.poulose@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
9 years agoarm64: implement generic IOMMU configuration
Robin Murphy [Mon, 12 Jan 2015 20:48:54 +0000 (20:48 +0000)]
arm64: implement generic IOMMU configuration

Add the necessary call to of_iommu_init.

Acked-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Robin Murphy <robin.murphy@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
9 years agoarm64: Combine coherent and non-coherent swiotlb dma_ops
Catalin Marinas [Mon, 12 Jan 2015 20:48:53 +0000 (20:48 +0000)]
arm64: Combine coherent and non-coherent swiotlb dma_ops

Since dev_archdata now has a dma_coherent state, combine the two
coherent and non-coherent operations and remove their declaration,
together with set_dma_ops, from the arch dma-mapping.h file.

Acked-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
9 years agoarm64: Fix SCTLR_EL1 initialisation
Suzuki K. Poulose [Wed, 17 Dec 2014 15:50:21 +0000 (15:50 +0000)]
arm64: Fix SCTLR_EL1 initialisation

We initialise the SCTLR_EL1 value by read-modify-writeback
of the desired bits, leaving the other bits (including reserved
bits(RESx)) untouched. However, sometimes the boot monitor could
leave garbage values in the RESx bits which could have different
implications. This patch makes sure that all the bits, including
the RESx bits, are set to the proper state, except for the
'endianness' control bits, EE(25) & E0E(24)- which are set early
in the el2_setup.

Updated the state of the Bit[6] in the comment to RES0 in the
comment.

Signed-off-by: Suzuki K. Poulose <suzuki.poulose@arm.com>
Acked-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
9 years agoarm64: add ioremap physical address information
Min-Hua Chen [Fri, 26 Dec 2014 16:52:10 +0000 (16:52 +0000)]
arm64: add ioremap physical address information

In /proc/vmallocinfo, it's good to show the physical address
of each ioremap in vmallocinfo. Add physical address information
in arm64 ioremap.

0xffffc900047f2000-0xffffc900047f4000    8192 _nv013519rm+0x57/0xa0
[nvidia] phys=f8100000 ioremap
0xffffc900047f4000-0xffffc900047f6000    8192 _nv013519rm+0x57/0xa0
[nvidia] phys=f8008000 ioremap
0xffffc90004800000-0xffffc90004821000  135168 e1000_probe+0x22c/0xb95
[e1000e] phys=f4300000 ioremap
0xffffc900049c0000-0xffffc900049e1000  135168 _nv013521rm+0x4d/0xd0
[nvidia] phys=e0140000 ioremap

Signed-off-by: Min-Hua Chen <orca.chen@gmail.com>
Acked-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
9 years agoarm64: mm: dump: add missing includes
Mark Rutland [Thu, 22 Jan 2015 18:20:36 +0000 (18:20 +0000)]
arm64: mm: dump: add missing includes

The arm64 dump code is currently relying on some definitions which are
pulled in via transitive dependencies. It seems we have implicit
dependencies on the following definitions:

* MODULES_VADDR         (asm/memory.h)
* MODULES_END           (asm/memory.h)
* PAGE_OFFSET           (asm/memory.h)
* PTE_*                 (asm/pgtable-hwdef.h)
* ENOMEM                (linux/errno.h)
* device_initcall       (linux/init.h)

This patch ensures we explicitly include the relevant headers for the
above items, fixing the observed build issue and hopefully preventing
future issues as headers are refactored.

Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Reported-by: Mark Brown <broonie@kernel.org>
Acked-by: Steve Capper <steve.capper@linaro.org>
Cc: Laura Abbott <lauraa@codeaurora.org>
Cc: Will Deacon <will.deacon@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
9 years agoarm64: Fix overlapping VA allocations
Mark Rutland [Thu, 22 Jan 2015 18:20:35 +0000 (18:20 +0000)]
arm64: Fix overlapping VA allocations

PCI IO space was intended to be 16MiB, at 32MiB below MODULES_VADDR, but
commit d1e6dc91b532d3d3 ("arm64: Add architectural support for PCI")
extended this to cover the full 32MiB. The final 8KiB of this 32MiB is
also allocated for the fixmap, allowing for potential clashes between
the two.

This change was masked by assumptions in mem_init and the page table
dumping code, which assumed the I/O space to be 16MiB long through
seaparte hard-coded definitions.

This patch changes the definition of the PCI I/O space allocation to
live in asm/memory.h, along with the other VA space allocations. As the
fixmap allocation depends on the number of fixmap entries, this is moved
below the PCI I/O space allocation. Both the fixmap and PCI I/O space
are guarded with 2MB of padding. Sites assuming the I/O space was 16MiB
are moved over use new PCI_IO_{START,END} definitions, which will keep
in sync with the size of the IO space (now restored to 16MiB).

As a useful side effect, the use of the new PCI_IO_{START,END}
definitions prevents a build issue in the dumping code due to a (now
redundant) missing include of io.h for PCI_IOBASE.

Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Laura Abbott <lauraa@codeaurora.org>
Cc: Liviu Dudau <liviu.dudau@arm.com>
Cc: Steve Capper <steve.capper@linaro.org>
Cc: Will Deacon <will.deacon@arm.com>
[catalin.marinas@arm.com: reorder FIXADDR and PCI_IO address_markers_idx enum]
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
9 years agoarm64/efi: move virtmap init to early initcall
Ard Biesheuvel [Thu, 22 Jan 2015 10:01:40 +0000 (10:01 +0000)]
arm64/efi: move virtmap init to early initcall

Now that the create_mapping() code in mm/mmu.c is able to support
setting up kernel page tables at initcall time, we can move the whole
virtmap creation to arm64_enable_runtime_services() instead of having
a distinct stage during early boot. This also allows us to drop the
arm64-specific EFI_VIRTMAP flag.

Signed-off-by: Ard Biesheuvel <ard.biesheuvel-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
9 years agoarm64: add better page protections to arm64
Laura Abbott [Thu, 22 Jan 2015 01:36:06 +0000 (17:36 -0800)]
arm64: add better page protections to arm64

Add page protections for arm64 similar to those in arm.
This is for security reasons to prevent certain classes
of exploits. The current method:

- Map all memory as either RWX or RW. We round to the nearest
  section to avoid creating page tables before everything is mapped
- Once everything is mapped, if either end of the RWX section should
  not be X, we split the PMD and remap as necessary
- When initmem is to be freed, we change the permissions back to
  RW (using stop machine if necessary to flush the TLB)
- If CONFIG_DEBUG_RODATA is set, the read only sections are set
  read only.

Acked-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Tested-by: Kees Cook <keescook@chromium.org>
Tested-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Laura Abbott <lauraa@codeaurora.org>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
9 years agoarm64: use fixmap for text patching
Laura Abbott [Thu, 22 Jan 2015 01:36:05 +0000 (01:36 +0000)]
arm64: use fixmap for text patching

When kernel text is marked as read only, it cannot be modified directly.
Use a fixmap to modify the text instead in a similar manner to
x86 and arm.

Reviewed-by: Kees Cook <keescook@chromium.org>
Reviewed-by: Mark Rutland <mark.rutland@arm.com>
Tested-by: Kees Cook <keescook@chromium.org>
Tested-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Laura Abbott <lauraa@codeaurora.org>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
9 years agoarm64: respect mem= for EFI
Mark Rutland [Thu, 15 Jan 2015 16:42:14 +0000 (16:42 +0000)]
arm64: respect mem= for EFI

When booting with EFI, we acquire the EFI memory map after parsing the
early params. This unfortuantely renders the option useless as we call
memblock_enforce_memory_limit (which uses memblock_remove_range behind
the scenes) before we've added any memblocks. We end up removing
nothing, then adding all of memory later when efi_init calls
reserve_regions.

Instead, we can log the limit and apply this later when we do the rest
of the memblock work in memblock_init, which should work regardless of
the presence of EFI. At the same time we may as well move the early
parameter into arm64's mm/init.c, close to arm64_memblock_init.

Any memory which must be mapped (e.g. for use by EFI runtime services)
must be mapped explicitly reather than relying on the linear mapping,
which may be truncated as a result of a mem= option passed on the kernel
command line.

Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Acked-by: Catalin Marinas <catalin.marinas@arm.com>
Acked-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Tested-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Cc: Leif Lindholm <leif.lindholm@linaro.org>
Cc: Will Deacon <will.deacon@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
9 years agoarm64/efi: handle potential failure to remap memory map
Ard Biesheuvel [Thu, 15 Jan 2015 12:01:06 +0000 (12:01 +0000)]
arm64/efi: handle potential failure to remap memory map

When remapping the UEFI memory map using ioremap_cache(), we
have to deal with potential failure. Note that, even if the
common case is for ioremap_cache() to return the existing linear
mapping of the memory map, we cannot rely on that to be always the
case, e.g., in the presence of a mem= kernel parameter.

At the same time, remove a stale comment and move the memmap code
together.

Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Acked-by: Mark Rutland <mark.rutland@arm.com>
Acked-by: Mark Salter <msalter@redhat.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
9 years agoarm64: kill off the libgcc dependency
Kevin Hao [Thu, 15 Jan 2015 12:07:33 +0000 (12:07 +0000)]
arm64: kill off the libgcc dependency

The arm64 kernel builds fine without the libgcc. Actually it should not
be used at all in the kernel. The following are the reasons indicated
by Russell King:

  Although libgcc is part of the compiler, libgcc is built with the
  expectation that it will be running in userland - it expects to link
  to a libc.  That's why you can't build libgcc without having the glibc
  headers around.

  [...]

  Meanwhile, having the kernel build the compiler support functions that
  it needs ensures that (a) we know what compiler support functions are
  being used, (b) we know the implementation of those support functions
  are sane for use in the kernel, (c) we can build them with appropriate
  compiler flags for best performance, and (d) we remove an unnecessary
  dependency on the build toolchain.

Signed-off-by: Kevin Hao <haokexin@gmail.com>
Acked-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
9 years agoMerge branch 'arm64/common-esr-macros' of git://git.kernel.org/pub/scm/linux/kernel...
Catalin Marinas [Thu, 15 Jan 2015 15:44:44 +0000 (15:44 +0000)]
Merge branch 'arm64/common-esr-macros' of git://git./linux/kernel/git/mark/linux

ESR_ELx definitions clean-up from Mark Rutland.

* 'arm64/common-esr-macros' of git://git.kernel.org/pub/scm/linux/kernel/git/mark/linux:
  arm64: kvm: decode ESR_ELx.EC when reporting exceptions
  arm64: kvm: remove ESR_EL2_* macros
  arm64: remove ESR_EL1_* macros
  arm64: kvm: move to ESR_ELx macros
  arm64: decode ESR_ELx.EC when reporting exceptions
  arm64: move to ESR_ELx macros
  arm64: introduce common ESR_ELx_* definitions

9 years agoarm64: kvm: decode ESR_ELx.EC when reporting exceptions
Mark Rutland [Wed, 7 Jan 2015 11:26:18 +0000 (11:26 +0000)]
arm64: kvm: decode ESR_ELx.EC when reporting exceptions

To aid the developer when something triggers an unexpected exception,
decode the ESR_ELx.EC field when logging an ESR_ELx value using the
newly introduced esr_get_class_string. This doesn't tell the developer
the specifics of the exception encoded in the remaining IL and ISS bits,
but it can be helpful to distinguish between exception classes (e.g.
SError and a data abort) without having to manually decode the field,
which can be tiresome.

Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Acked-by: Christoffer Dall <christoffer.dall@linaro.org>
Reviewed-by: Christoffer Dall <christoffer.dall@linaro.org>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Marc Zyngier <marc.zyngier@arm.com>
Cc: Peter Maydell <peter.maydell@linaro.org>
Cc: Will Deacon <will.deacon@arm.com>
9 years agoarm64: kvm: remove ESR_EL2_* macros
Mark Rutland [Mon, 24 Nov 2014 14:05:44 +0000 (14:05 +0000)]
arm64: kvm: remove ESR_EL2_* macros

Now that all users have been moved over to the common ESR_ELx_* macros,
remove the redundant ESR_EL2 macros. To maintain compatibility with the
fault handling code shared with 32-bit, the FSC_{FAULT,PERM} macros are
retained as aliases for the common ESR_ELx_FSC_{FAULT,PERM} definitions.

There should be no functional change as a result of this patch.

Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Acked-by: Christoffer Dall <christoffer.dall@linaro.org>
Reviewed-by: Christoffer Dall <christoffer.dall@linaro.org>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Marc Zyngier <marc.zyngier@arm.com>
Cc: Peter Maydell <peter.maydell@linaro.org>
Cc: Will Deacon <will.deacon@arm.com>
9 years agoarm64: remove ESR_EL1_* macros
Mark Rutland [Mon, 24 Nov 2014 14:03:52 +0000 (14:03 +0000)]
arm64: remove ESR_EL1_* macros

Now that all users have been moved over to the common ESR_ELx_* macros,
remove the redundant ESR_EL1 macros.

There should be no functional change as a result of this patch.

Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Acked-by: Catalin Marinas <catalin.marinas@arm.com>
Reviewed-by: Christoffer Dall <christoffer.dall@linaro.org>
Cc: Marc Zyngier <marc.zyngier@arm.com>
Cc: Peter Maydell <peter.maydell@linaro.org>
Cc: Will Deacon <will.deacon@arm.com>
9 years agoarm64: kvm: move to ESR_ELx macros
Mark Rutland [Mon, 24 Nov 2014 13:59:30 +0000 (13:59 +0000)]
arm64: kvm: move to ESR_ELx macros

Now that we have common ESR_ELx macros, make use of them in the arm64
KVM code. The addition of <asm/esr.h> to the include path highlighted
badly ordered (i.e. not alphabetical) include lists; these are changed
to alphabetical order.

There should be no functional change as a result of this patch.

Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Reviewed-by: Christoffer Dall <christoffer.dall@linaro.org>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Marc Zyngier <marc.zyngier@arm.com>
Cc: Peter Maydell <peter.maydell@linaro.org>
Cc: Will Deacon <will.deacon@arm.com>
9 years agoarm64: decode ESR_ELx.EC when reporting exceptions
Mark Rutland [Tue, 18 Nov 2014 12:16:30 +0000 (12:16 +0000)]
arm64: decode ESR_ELx.EC when reporting exceptions

To aid the developer when something triggers an unexpected exception,
decode the ESR_ELx.EC field when logging an ESR_ELx value. This doesn't
tell the developer the specifics of the exception encoded in the
remaining IL and ISS bits, but it can be helpful to distinguish between
exception classes (e.g. SError and a data abort) without having to
manually decode the field, which can be tiresome.

Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Acked-by: Catalin Marinas <catalin.marinas@arm.com>
Reviewed-by: Christoffer Dall <christoffer.dall@linaro.org>
Cc: Marc Zyngier <marc.zyngier@arm.com>
Cc: Peter Maydell <peter.maydell@linaro.org>
Cc: Will Deacon <will.deacon@arm.com>
9 years agoarm64: move to ESR_ELx macros
Mark Rutland [Mon, 24 Nov 2014 12:31:40 +0000 (12:31 +0000)]
arm64: move to ESR_ELx macros

Now that we have common ESR_ELx_* macros, move the core arm64 code over
to them.

There should be no functional change as a result of this patch.

Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Acked-by: Catalin Marinas <catalin.marinas@arm.com>
Reviewed-by: Christoffer Dall <christoffer.dall@linaro.org>
Cc: Marc Zyngier <marc.zyngier@arm.com>
Cc: Peter Maydell <peter.maydell@linaro.org>
Cc: Will Deacon <will.deacon@arm.com>
9 years agoarm64: introduce common ESR_ELx_* definitions
Mark Rutland [Mon, 24 Nov 2014 12:03:32 +0000 (12:03 +0000)]
arm64: introduce common ESR_ELx_* definitions

Currently we have separate ESR_EL{1,2}_* macros, despite the fact that
the encodings are common. While encodings are architected to refer to
the current EL or a lower EL, the macros refer to particular ELs (e.g.
ESR_ELx_EC_DABT_EL0). Having these duplicate definitions is redundant,
and their naming is misleading.

This patch introduces common ESR_ELx_* macros that can be used in all
cases, in preparation for later patches which will migrate existing
users over. Some additional cleanups are made in the process:

* Suffixes for particular exception levelts (e.g. _EL0, _EL1) are
  replaced with more general _LOW and _CUR suffixes, matching the
  architectural intent.

* ESR_ELx_EC_WFx, rather than ESR_ELx_EC_WFI is introduced, as this
  EC encoding covers traps from both WFE and WFI. Similarly,
  ESR_ELx_WFx_ISS_WFE rather than ESR_ELx_EC_WFI_ISS_WFE is introduced.

* Multi-bit fields are given consistently named _SHIFT and _MASK macros.

* UL() is used for compatiblity with assembly files.

* Comments are added for currently unallocated ESR_ELx.EC encodings.

For fields other than ESR_ELx.EC, macros are only implemented for fields
for which there is already an ESR_EL{1,2}_* macro.

Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Acked-by: Catalin Marinas <catalin.marinas@arm.com>
Reviewed-by: Christoffer Dall <christoffer.dall@linaro.org>
Cc: Marc Zyngier <marc.zyngier@arm.com>
Cc: Peter Maydell <peter.maydell@linaro.org>
Cc: Will Deacon <will.deacon@arm.com>
9 years agoarm64: kernel: add support for cpu cache information
Sudeep Holla [Thu, 8 Jan 2015 10:42:34 +0000 (10:42 +0000)]
arm64: kernel: add support for cpu cache information

This patch adds support for cacheinfo on ARM64.

On ARMv8, the cache hierarchy can be identified through Cache Level ID
(CLIDR) register while the cache geometry is provided by Cache Size ID
(CCSIDR) register.

Since the architecture doesn't provide any way of detecting the cpus
sharing particular cache, device tree is used for the same purpose.

Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
9 years agoarm64: remove broken cachepolicy code
Mark Rutland [Thu, 8 Jan 2015 17:07:47 +0000 (17:07 +0000)]
arm64: remove broken cachepolicy code

The cachepolicy kernel parameter was intended to aid in the debugging of
coherency issues, but it is fundamentally broken for several reasons:

 * On SMP platforms, only the boot CPU's tcr_el1 is altered. Secondary
   CPUs may therefore use differ w.r.t. the attributes they apply to
   MT_NORMAL memory, resulting in a loss of coherency.

 * The cache maintenance using flush_dcache_all (based on Set/Way
   operations) is not guaranteed to empty a given CPU's cache hierarchy
   while said CPU has caches enabled, it cannot empty the caches of
   other coherent PEs, nor is it guaranteed to flush data to the PoC
   even when caches are disabled.

 * The TLBs are not invalidated around the modification of MAIR_EL1 and
   TCR_EL1, as required by the architecture (as both are permitted to be
   cached in a TLB). This may result in CPUs using attributes other than
   those expected for some memory accesses, resulting in a loss of
   coherency.

 * Exclusive accesses are not architecturally guaranteed to function as
   expected on memory marked as Write-Through or Non-Cacheable. Thus
   changing the attributes of MT_NORMAL away from the (architecurally
   safe) defaults may cause uses of these instructions (e.g. atomics) to
   behave erratically.

Given this, the cachepolicy code cannot be used for debugging purposes
as it alone is likely to cause coherency issues. This patch removes the
broken cachepolicy code.

Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Acked-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
9 years agoMerge tag 'for-3.20' of http://git.linaro.org/people/ard.biesheuvel/linux-arm into...
Catalin Marinas [Tue, 13 Jan 2015 22:42:31 +0000 (22:42 +0000)]
Merge tag 'for-3.20' of git.linaro.org/people/ard.biesheuvel/linux-arm into upstream

UEFI updates for arm64

This series consists of a reimplementation of the virtual remapping of
UEFI Runtime Services in a way that is stable across kexec, including
the required preparatory refactoring and other work to set the stage,
and some cleaning up afterwards to remove boot services memory and
identitity map handling that has now become redundant.

* tag 'for-3.20' of http://git.linaro.org/people/ard.biesheuvel/linux-arm:
  arm64/efi: remove idmap manipulations from UEFI code
  arm64/efi: remove free_boot_services() and friends
  arm64/efi: move SetVirtualAddressMap() to UEFI stub
  arm64/efi: set EFI_ALLOC_ALIGN to 64 KB
  efi: efistub: allow allocation alignment larger than EFI_PAGE_SIZE
  efi: split off remapping code from efi_config_init()
  arm64/mm: add create_pgd_mapping() to create private page tables
  arm64/mm: add explicit struct_mm argument to __create_mapping()

9 years agoarm64/efi: remove idmap manipulations from UEFI code
Ard Biesheuvel [Mon, 20 Oct 2014 14:41:38 +0000 (16:41 +0200)]
arm64/efi: remove idmap manipulations from UEFI code

Now that we have moved the call to SetVirtualAddressMap() to the stub,
UEFI has no use for the ID map, so we can drop the code that installs
ID mappings for UEFI memory regions.

Acked-by: Leif Lindholm <leif.lindholm@linaro.org>
Acked-by: Will Deacon <will.deacon@arm.com>
Tested-by: Leif Lindholm <leif.lindholm@linaro.org>
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
9 years agoarm64/efi: remove free_boot_services() and friends
Ard Biesheuvel [Mon, 20 Oct 2014 14:35:21 +0000 (16:35 +0200)]
arm64/efi: remove free_boot_services() and friends

Now that we are calling SetVirtualAddressMap() from the stub, there is no
need to reserve boot-only memory regions, which implies that there is also
no reason to free them again later.

Acked-by: Leif Lindholm <leif.lindholm@linaro.org>
Acked-by: Will Deacon <will.deacon@arm.com>
Tested-by: Leif Lindholm <leif.lindholm@linaro.org>
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
9 years agoarm64/efi: move SetVirtualAddressMap() to UEFI stub
Ard Biesheuvel [Mon, 20 Oct 2014 14:27:26 +0000 (16:27 +0200)]
arm64/efi: move SetVirtualAddressMap() to UEFI stub

In order to support kexec, the kernel needs to be able to deal with the
state of the UEFI firmware after SetVirtualAddressMap() has been called.
To avoid having separate code paths for non-kexec and kexec, let's move
the call to SetVirtualAddressMap() to the stub: this will guarantee us
that it will only be called once (since the stub is not executed during
kexec), and ensures that the UEFI state is identical between kexec and
normal boot.

This implies that the layout of the virtual mapping needs to be created
by the stub as well. All regions are rounded up to a naturally aligned
multiple of 64 KB (for compatibility with 64k pages kernels) and recorded
in the UEFI memory map. The kernel proper reads those values and installs
the mappings in a dedicated set of page tables that are swapped in during
UEFI Runtime Services calls.

Acked-by: Leif Lindholm <leif.lindholm@linaro.org>
Acked-by: Matt Fleming <matt.fleming@intel.com>
Tested-by: Leif Lindholm <leif.lindholm@linaro.org>
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
9 years agoarm64/efi: set EFI_ALLOC_ALIGN to 64 KB
Ard Biesheuvel [Mon, 17 Nov 2014 12:50:21 +0000 (13:50 +0100)]
arm64/efi: set EFI_ALLOC_ALIGN to 64 KB

Set EFI_ALLOC_ALIGN to 64 KB so that all allocations done by the stub
are naturally compatible with a 64 KB granule kernel.

Acked-by: Leif Lindholm <leif.lindholm@linaro.org>
Tested-by: Leif Lindholm <leif.lindholm@linaro.org>
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
9 years agoefi: efistub: allow allocation alignment larger than EFI_PAGE_SIZE
Ard Biesheuvel [Mon, 17 Nov 2014 12:46:44 +0000 (13:46 +0100)]
efi: efistub: allow allocation alignment larger than EFI_PAGE_SIZE

On systems with 64 KB pages, it is preferable for UEFI memory map
entries to be 64 KB aligned multiples of 64 KB, because it relieves
us of having to deal with the residues.
So, if EFI_ALLOC_ALIGN is #define'd by the platform, use it to round
up all memory allocations made.

Acked-by: Matt Fleming <matt.fleming@intel.com>
Acked-by: Borislav Petkov <bp@suse.de>
Tested-by: Leif Lindholm <leif.lindholm@linaro.org>
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
9 years agoefi: split off remapping code from efi_config_init()
Ard Biesheuvel [Sat, 18 Oct 2014 13:04:15 +0000 (15:04 +0200)]
efi: split off remapping code from efi_config_init()

Split of the remapping code from efi_config_init() so that the caller
can perform its own remapping. This is necessary to correctly handle
virtually remapped UEFI memory regions under kexec, as efi.systab will
have been updated to a virtual address.

Acked-by: Matt Fleming <matt.fleming@intel.com>
Tested-by: Leif Lindholm <leif.lindholm@linaro.org>
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
9 years agoarm64/mm: add create_pgd_mapping() to create private page tables
Ard Biesheuvel [Mon, 20 Oct 2014 13:42:07 +0000 (15:42 +0200)]
arm64/mm: add create_pgd_mapping() to create private page tables

For UEFI, we need to install the memory mappings used for Runtime Services
in a dedicated set of page tables. Add create_pgd_mapping(), which allows
us to allocate and install those page table entries early.

Reviewed-by: Will Deacon <will.deacon@arm.com>
Tested-by: Leif Lindholm <leif.lindholm@linaro.org>
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
9 years agoarm64/mm: add explicit struct_mm argument to __create_mapping()
Ard Biesheuvel [Mon, 20 Oct 2014 12:02:15 +0000 (14:02 +0200)]
arm64/mm: add explicit struct_mm argument to __create_mapping()

Currently, swapper_pg_dir and idmap_pg_dir share the init_mm mm_struct
instance. To allow the introduction of other pg_dir instances, for instance,
for UEFI's mapping of Runtime Services, make the struct_mm instance an
explicit argument that gets passed down to the pmd and pte instantiation
functions. Note that the consumers (pmd_populate/pgd_populate) of the
mm_struct argument don't actually inspect it, but let's fix it for
correctness' sake.

Acked-by: Steve Capper <steve.capper@linaro.org>
Tested-by: Leif Lindholm <leif.lindholm@linaro.org>
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
9 years agolinux 3.19-rc4
Linus Torvalds [Sun, 11 Jan 2015 20:44:53 +0000 (12:44 -0800)]
linux 3.19-rc4

9 years agoMerge branch 'fixes' of git://ftp.arm.linux.org.uk/~rmk/linux-arm
Linus Torvalds [Sun, 11 Jan 2015 20:44:10 +0000 (12:44 -0800)]
Merge branch 'fixes' of git://ftp.arm.linux.org.uk/~rmk/linux-arm

Pull ARM fixes from Russell King:
 "Three small fixes from over the Christmas period, and wiring up the
  new execveat syscall for ARM"

* 'fixes' of git://ftp.arm.linux.org.uk/~rmk/linux-arm:
  ARM: 8275/1: mm: fix PMD_SECT_RDONLY undeclared compile error
  ARM: 8253/1: mm: use phys_addr_t type in map_lowmem() for kernel mem region
  ARM: 8249/1: mm: dump: don't skip regions
  ARM: wire up execveat syscall

9 years agoMerge branch 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 11 Jan 2015 19:53:46 +0000 (11:53 -0800)]
Merge branch 'x86-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull x86 fixes from Ingo Molnar:
 "Misc fixes: two vdso fixes, two kbuild fixes and a boot failure fix
  with certain odd memory mappings"

* 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86, vdso: Use asm volatile in __getcpu
  x86/build: Clean auto-generated processor feature files
  x86: Fix mkcapflags.sh bash-ism
  x86: Fix step size adjustment during initial memory mapping
  x86_64, vdso: Fix the vdso address randomization algorithm

9 years agoMerge branch 'sched-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 11 Jan 2015 19:51:49 +0000 (11:51 -0800)]
Merge branch 'sched-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull scheduler fixes from Ingo Molnar:
 "Misc fixes: group scheduling corner case fix, two deadline scheduler
  fixes, effective_load() overflow fix, nested sleep fix, 6144 CPUs
  system fix"

* 'sched-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  sched/fair: Fix RCU stall upon -ENOMEM in sched_create_group()
  sched/deadline: Avoid double-accounting in case of missed deadlines
  sched/deadline: Fix migration of SCHED_DEADLINE tasks
  sched: Fix odd values in effective_load() calculations
  sched, fanotify: Deal with nested sleeps
  sched: Fix KMALLOC_MAX_SIZE overflow during cpumask allocation

9 years agoMerge branch 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 11 Jan 2015 19:47:45 +0000 (11:47 -0800)]
Merge branch 'perf-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull perf fixes from Ingo Molnar:
 "Mostly tooling fixes, but also some kernel side fixes: uncore PMU
  driver fix, user regs sampling fix and an instruction decoder fix that
  unbreaks PEBS precise sampling"

* 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  perf/x86/uncore/hsw-ep: Handle systems with only two SBOXes
  perf/x86_64: Improve user regs sampling
  perf: Move task_pt_regs sampling into arch code
  x86: Fix off-by-one in instruction decoder
  perf hists browser: Fix segfault when showing callchain
  perf callchain: Free callchains when hist entries are deleted
  perf hists: Fix children sort key behavior
  perf diff: Fix to sort by baseline field by default
  perf list: Fix --raw-dump option
  perf probe: Fix crash in dwarf_getcfi_elf
  perf probe: Fix to fall back to find probe point in symbols
  perf callchain: Append callchains only when requested
  perf ui/tui: Print backtrace symbols when segfault occurs
  perf report: Show progress bar for output resorting

9 years agoMerge branch 'locking-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 11 Jan 2015 19:46:31 +0000 (11:46 -0800)]
Merge branch 'locking-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull locking fixes from Ingo Molnar:
 "A liblockdep fix and a mutex_unlock() mutex-debugging fix"

* 'locking-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  mutex: Always clear owner field upon mutex_unlock()
  tools/liblockdep: Fix debug_check thinko in mutex destroy

9 years agomm: fix corner case in anon_vma endless growing prevention
Konstantin Khlebnikov [Sun, 11 Jan 2015 13:54:06 +0000 (16:54 +0300)]
mm: fix corner case in anon_vma endless growing prevention

Fix for BUG_ON(anon_vma->degree) splashes in unlink_anon_vmas() ("kernel
BUG at mm/rmap.c:399!") caused by commit 7a3ef208e662 ("mm: prevent
endless growth of anon_vma hierarchy")

Anon_vma_clone() is usually called for a copy of source vma in
destination argument.  If source vma has anon_vma it should be already
in dst->anon_vma.  NULL in dst->anon_vma is used as a sign that it's
called from anon_vma_fork().  In this case anon_vma_clone() finds
anon_vma for reusing.

Vma_adjust() calls it differently and this breaks anon_vma reusing
logic: anon_vma_clone() links vma to old anon_vma and updates degree
counters but vma_adjust() overrides vma->anon_vma right after that.  As
a result final unlink_anon_vmas() decrements degree for wrong anon_vma.

This patch assigns ->anon_vma before calling anon_vma_clone().

Signed-off-by: Konstantin Khlebnikov <koct9i@gmail.com>
Reported-and-tested-by: Chris Clayton <chris2553@googlemail.com>
Reported-and-tested-by: Oded Gabbay <oded.gabbay@amd.com>
Reported-and-tested-by: Chih-Wei Huang <cwhuang@android-x86.org>
Acked-by: Rik van Riel <riel@redhat.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Daniel Forrest <dan.forrest@ssec.wisc.edu>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: stable@vger.kernel.org # to match back-porting of 7a3ef208e662
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agomm: Don't count the stack guard page towards RLIMIT_STACK
Linus Torvalds [Sun, 11 Jan 2015 19:33:57 +0000 (11:33 -0800)]
mm: Don't count the stack guard page towards RLIMIT_STACK

Commit fee7e49d4514 ("mm: propagate error from stack expansion even for
guard page") made sure that we return the error properly for stack
growth conditions.  It also theorized that counting the guard page
towards the stack limit might break something, but also said "Let's see
if anybody notices".

Somebody did notice.  Apparently android-x86 sets the stack limit very
close to the limit indeed, and including the guard page in the rlimit
check causes the android 'zygote' process problems.

So this adds the (fairly trivial) code to make the stack rlimit check be
against the actual real stack size, rather than the size of the vma that
includes the guard page.

Reported-and-tested-by: Chih-Wei Huang <cwhuang@android-x86.org>
Cc: Jay Foad <jay.foad@gmail.com>
Cc: stable@kernel.org # to match back-porting of fee7e49d4514
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoMerge branch 'core/urgent' into locking/urgent, to collect all pending locking fixes
Ingo Molnar [Sun, 11 Jan 2015 08:18:05 +0000 (09:18 +0100)]
Merge branch 'core/urgent' into locking/urgent, to collect all pending locking fixes

Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agoMerge tag 'vfio-v3.19-rc4' of git://github.com/awilliam/linux-vfio
Linus Torvalds [Sat, 10 Jan 2015 20:23:03 +0000 (12:23 -0800)]
Merge tag 'vfio-v3.19-rc4' of git://github.com/awilliam/linux-vfio

Pull VFIO fix from Alex Williamson:
 "Fix PCI header check in vfio_pci_probe() (Wei Yang)"

* tag 'vfio-v3.19-rc4' of git://github.com/awilliam/linux-vfio:
  vfio-pci: Fix the check on pci device type in vfio_pci_probe()

9 years agoMerge tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi
Linus Torvalds [Sat, 10 Jan 2015 19:59:25 +0000 (11:59 -0800)]
Merge tag 'scsi-fixes' of git://git./linux/kernel/git/jejb/scsi

Pull SCSI fix from James Bottomley:
 "Just one fix: a qlogic busy wait regression"

* tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi:
  qla2xxx: fix busy wait regression

9 years agoMerge tag 'sound-3.19-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai...
Linus Torvalds [Sat, 10 Jan 2015 05:23:27 +0000 (21:23 -0800)]
Merge tag 'sound-3.19-rc4' of git://git./linux/kernel/git/tiwai/sound

Pull sound fixes from Takashi Iwai:
 "All a few small regression or stable fixes: a Nvidia HDMI ID addition,
  a regression fix for CAIAQ stream count, a typo fix for GPIO setup
  with STAC/IDT HD-audio codecs, and a Fireworks big-endian fix"

* tag 'sound-3.19-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound:
  ALSA: fireworks: fix an endianness bug for transaction length
  ALSA: hda - Add new GPU codec ID 0x10de0072 to snd-hda
  ALSA: hda - Fix wrong gpio_dir & gpio_mask hint setups for IDT/STAC codecs
  ALSA: snd-usb-caiaq: fix stream count check

9 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jikos/hid
Linus Torvalds [Sat, 10 Jan 2015 05:18:48 +0000 (21:18 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/jikos/hid

Pull HID updates from Jiri Kosina:

 - bounds checking fixes in logitech and roccat drivers, from Peter Wu
   and Dan Carpenter

 - double-kfree fix in i2c-hid driver on bus shutdown, from Mika
   Westerberg

 - a couple of various small driver fixes

 - a few device id additions

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jikos/hid:
  HID: roccat: potential out of bounds in pyra_sysfs_write_settings()
  HID: Add a new id 0x501a for Genius MousePen i608X
  HID: logitech-hidpp: prefix the name with "Logitech"
  HID: logitech-hidpp: avoid unintended fall-through
  HID: Allow HID_BATTERY_STRENGTH to be enabled
  HID: i2c-hid: Do not free buffers in i2c_hid_stop()
  HID: add battery quirk for USB_DEVICE_ID_APPLE_ALU_WIRELESS_2011_ISO keyboard
  HID: logitech-hidpp: check WTP report length
  HID: logitech-dj: check report length

9 years agoMerge branch 'drm-fixes' of git://people.freedesktop.org/~airlied/linux
Linus Torvalds [Sat, 10 Jan 2015 05:13:34 +0000 (21:13 -0800)]
Merge branch 'drm-fixes' of git://people.freedesktop.org/~airlied/linux

Pull drm fixes from Dave Airlie:
 "I'm briefly working between holidays and LCA, so this is close to a
  couple of weeks of fixes,

  Two sets of amdkfd fixes, this is a new feature this kernel, and this
  pull fixes a few issues since it got merged, ordering when built-in to
  kernel and also the iommu vs gpu ordering patch, it also reworks the
  ioctl before the initial release.

  Otherwise:
   - radeon: some misc fixes all over, hdmi, 4k, dpm
   - nouveau: mcp77 init fixes, oops fix, bug on fix, msi fix
   - i915: power fixes, revert VGACNTR patch

  Probably be quiteer next week since I'll be at LCA anyways"

* 'drm-fixes' of git://people.freedesktop.org/~airlied/linux: (33 commits)
  drm/amdkfd: rewrite kfd_ioctl() according to drm_ioctl()
  drm/amdkfd: reformat IOCTL definitions to drm-style
  drm/amdkfd: Do copy_to/from_user in general kfd_ioctl()
  drm/radeon: integer underflow in radeon_cp_dispatch_texture()
  drm/radeon: adjust default bapm settings for KV
  drm/radeon: properly filter DP1.2 4k modes on non-DP1.2 hw
  drm/radeon: fix sad_count check for dce3
  drm/radeon: KV has three PPLLs (v2)
  drm/amdkfd: unmap VMID<-->PASID when relesing VMID (non-HWS)
  drm/radeon: Init amdkfd only if it was compiled
  amdkfd: actually allocate longs for the pasid bitmask
  drm/nouveau/nouveau: Do not BUG_ON(!spin_is_locked()) on UP
  drm/nv4c/mc: disable msi
  drm/nouveau/fb/ram/mcp77: enable NISO poller
  drm/nouveau/fb/ram/mcp77: use carveout reg to determine size
  drm/nouveau/fb/ram/mcp77: subclass nouveau_ram
  drm/nouveau: wake up the card if necessary during gem callbacks
  drm/nouveau/device: Add support for GK208B, resolves bug 86935
  drm/nouveau: fix missing return statement in nouveau_ttm_tt_unpopulate
  drm/nouveau/bios: fix oops on pre-nv50 chipsets
  ...

9 years agoMerge tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux
Linus Torvalds [Sat, 10 Jan 2015 04:57:58 +0000 (20:57 -0800)]
Merge tag 'arm64-fixes' of git://git./linux/kernel/git/arm64/linux

Pull arm64 fixes from Will Deacon:
 "Here is a handful of minor arm64 fixes discovered and fixed over the
  Christmas break.  The main part is adding some missing #includes that
  we seem to be getting transitively but have started causing problems
  in -next.

   - Fix early mapping fixmap corruption by EFI runtime services
   - Fix __NR_compat_syscalls off-by-one
   - Add missing sanity checks for some 32-bit registers
   - Add some missing #includes which we get transitively
   - Remove unused prepare_to_copy() macro"

* tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux:
  arm64/efi: add missing call to early_ioremap_reset()
  arm64: fix missing asm/io.h include in kernel/smp_spin_table.c
  arm64: fix missing asm/alternative.h include in kernel/module.c
  arm64: fix missing linux/bug.h include in asm/arch_timer.h
  arm64: fix missing asm/pgtable-hwdef.h include in asm/processor.h
  arm64: sanity checks: add missing AArch32 registers
  arm64: Remove unused prepare_to_copy()
  arm64: Correct __NR_compat_syscalls for bpf

9 years agoMerge tag 'for_linus-3.19-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/jwess...
Linus Torvalds [Sat, 10 Jan 2015 04:51:10 +0000 (20:51 -0800)]
Merge tag 'for_linus-3.19-rc4' of git://git./linux/kernel/git/jwessel/kgdb

Pull kgdb/kdb fixes from Jason Wessel:
 "These have been around since 3.17 and in kgdb-next for the last 9
  weeks and some will go back to -stable.

  Summary of changes:

  Cleanups
   - kdb: Remove unused command flags, repeat flags and KDB_REPEAT_NONE

  Fixes
   - kgdb/kdb: Allow access on a single core, if a CPU round up is
     deemed impossible, which will allow inspection of the now "trashed"
     kernel
   - kdb: Add enable mask for the command groups
   - kdb: access controls to restrict sensitive commands"

* tag 'for_linus-3.19-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/jwessel/kgdb:
  kernel/debug/debug_core.c: Logging clean-up
  kgdb: timeout if secondary CPUs ignore the roundup
  kdb: Allow access to sensitive commands to be restricted by default
  kdb: Add enable mask for groups of commands
  kdb: Categorize kdb commands (similar to SysRq categorization)
  kdb: Remove KDB_REPEAT_NONE flag
  kdb: Use KDB_REPEAT_* values as flags
  kdb: Rename kdb_register_repeat() to kdb_register_flags()
  kdb: Rename kdb_repeat_t to kdb_cmdflags_t, cmd_repeat to cmd_flags
  kdb: Remove currently unused kdbtab_t->cmd_flags

9 years agoMerge branch 'for-3.19' of git://linux-nfs.org/~bfields/linux
Linus Torvalds [Sat, 10 Jan 2015 02:10:48 +0000 (18:10 -0800)]
Merge branch 'for-3.19' of git://linux-nfs.org/~bfields/linux

Pull two nfsd bugfixes from Bruce Fields.

* 'for-3.19' of git://linux-nfs.org/~bfields/linux:
  rpc: fix xdr_truncate_encode to handle buffer ending on page boundary
  nfsd: fix fi_delegees leak when fi_had_conflict returns true

9 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/sage/ceph...
Linus Torvalds [Sat, 10 Jan 2015 01:55:00 +0000 (17:55 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/sage/ceph-client

Pull two Ceph fixes from Sage Weil:
 "These are both pretty trivial: a sparse warning fix and size_t printk
  thing"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/sage/ceph-client:
  libceph: fix sparse endianness warnings
  ceph: use %zu for len in ceph_fill_inline_data()

9 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux...
Linus Torvalds [Sat, 10 Jan 2015 01:46:07 +0000 (17:46 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/mason/linux-btrfs

Pull btrfs fixes from Chris Mason:
 "None of these are huge, but my commit does fix a regression from 3.18
  that could cause lost files during log replay.

  This also adds Dave Sterba to the list of Btrfs maintainers.  It
  doesn't mean we're doing things differently, but Dave has really been
  helping with the maintainer workload for years"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs:
  Btrfs: don't delay inode ref updates during log replay
  Btrfs: correctly get tree level in tree_backref_for_extent
  Btrfs: call inode_dec_link_count() on mkdir error path
  Btrfs: abort transaction if we don't find the block group
  Btrfs, scrub: uninitialized variable in scrub_extent_for_parity()
  Btrfs: add more maintainers

9 years agoMerge branch 'akpm' (patches from Andrew)
Linus Torvalds [Fri, 9 Jan 2015 23:10:59 +0000 (15:10 -0800)]
Merge branch 'akpm' (patches from Andrew)

Merge misc fixes from Andrew Morton:
 "12 fixes"

* emailed patches from Andrew Morton <akpm@linux-foundation.org>:
  mm, vmscan: prevent kswapd livelock due to pfmemalloc-throttled process being killed
  memcg: fix destination cgroup leak on task charges migration
  mm: memcontrol: switch soft limit default back to infinity
  mm/debug_pagealloc: remove obsolete Kconfig options
  vfs: renumber FMODE_NONOTIFY and add to uniqueness check
  arch/blackfin/mach-bf533/boards/stamp.c: add linux/delay.h
  ocfs2: fix the wrong directory passed to ocfs2_lookup_ino_from_name() when link file
  MAINTAINERS: update rydberg's addresses
  mm: protect set_page_dirty() from ongoing truncation
  mm: prevent endless growth of anon_vma hierarchy
  exit: fix race between wait_consider_task() and wait_task_zombie()
  ocfs2: remove bogus check in dlm_process_recovery_data

9 years agoARM: 8275/1: mm: fix PMD_SECT_RDONLY undeclared compile error
Victor Kamensky [Fri, 9 Jan 2015 17:55:45 +0000 (18:55 +0100)]
ARM: 8275/1: mm: fix PMD_SECT_RDONLY undeclared compile error

In v3.19-rc3 tree when CONFIG_ARM_LPAE and CONFIG_DEBUG_RODATA are enabled
image failed to compile with the following error:

arch/arm/mm/init.c:661:14: error: â€˜PMD_SECT_RDONLY’ undeclared here (not in a function)

It seems that '80d6b0c ARM: mm: allow text and rodata sections to be read-only'
and 'ded9477 ARM: 8109/1: mm: Modify pte_write and pmd_write logic for LPAE'
commits crossed. 80d6b0c uses PMD_SECT_RDONLY macro but ded9477 renames it
and uses software bits L_PMD_SECT_RDONLY instead.

Fix is to use L_PMD_SECT_RDONLY instead PMD_SECT_RDONLY as ded9477 does in
another places.

Signed-off-by: Victor Kamensky <victor.kamensky@linaro.org>
Acked-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
9 years agoHID: roccat: potential out of bounds in pyra_sysfs_write_settings()
Dan Carpenter [Fri, 9 Jan 2015 12:32:31 +0000 (15:32 +0300)]
HID: roccat: potential out of bounds in pyra_sysfs_write_settings()

This is a static checker fix.  We write some binary settings to the
sysfs file.  One of the settings is the "->startup_profile".  There
isn't any checking to make sure it fits into the
pyra->profile_settings[] array in the profile_activated() function.

I added a check to pyra_sysfs_write_settings() in both places because
I wasn't positive that the other callers were correct.

Cc: <stable@vger.kernel.org>
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
9 years agomutex: Always clear owner field upon mutex_unlock()
Chris Wilson [Tue, 6 Jan 2015 10:29:35 +0000 (10:29 +0000)]
mutex: Always clear owner field upon mutex_unlock()

Currently if DEBUG_MUTEXES is enabled, the mutex->owner field is only
cleared iff debug_locks is active. This exposes a race to other users of
the field where the mutex->owner may be still set to a stale value,
potentially upsetting mutex_spin_on_owner() among others.

References: https://bugs.freedesktop.org/show_bug.cgi?id=87955
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Acked-by: Davidlohr Bueso <dave@stgolabs.net>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Link: http://lkml.kernel.org/r/1420540175-30204-1-git-send-email-chris@chris-wilson.co.uk
Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agosched/fair: Fix RCU stall upon -ENOMEM in sched_create_group()
Tetsuo Handa [Thu, 25 Dec 2014 06:51:21 +0000 (15:51 +0900)]
sched/fair: Fix RCU stall upon -ENOMEM in sched_create_group()

When alloc_fair_sched_group() in sched_create_group() fails,
free_sched_group() is called, and free_fair_sched_group() is called by
free_sched_group(). Since destroy_cfs_bandwidth() is called by
free_fair_sched_group() without calling init_cfs_bandwidth(),
RCU stall occurs at hrtimer_cancel():

  INFO: rcu_sched self-detected stall on CPU { 1}  (t=60000 jiffies g=13074 c=13073 q=0)
  Task dump for CPU 1:
  (fprintd)       R  running task        0  6249      1 0x00000088
  ...
  Call Trace:
   <IRQ>  [<ffffffff81094988>] sched_show_task+0xa8/0x110
   [<ffffffff81097acd>] dump_cpu_task+0x3d/0x50
   [<ffffffff810c3a80>] rcu_dump_cpu_stacks+0x90/0xd0
   [<ffffffff810c7751>] rcu_check_callbacks+0x491/0x700
   [<ffffffff810cbf2b>] update_process_times+0x4b/0x80
   [<ffffffff810db046>] tick_sched_handle.isra.20+0x36/0x50
   [<ffffffff810db0a2>] tick_sched_timer+0x42/0x70
   [<ffffffff810ccb19>] __run_hrtimer+0x69/0x1a0
   [<ffffffff810db060>] ? tick_sched_handle.isra.20+0x50/0x50
   [<ffffffff810ccedf>] hrtimer_interrupt+0xef/0x230
   [<ffffffff810452cb>] local_apic_timer_interrupt+0x3b/0x70
   [<ffffffff8164a465>] smp_apic_timer_interrupt+0x45/0x60
   [<ffffffff816485bd>] apic_timer_interrupt+0x6d/0x80
   <EOI>  [<ffffffff810cc588>] ? lock_hrtimer_base.isra.23+0x18/0x50
   [<ffffffff81193cf1>] ? __kmalloc+0x211/0x230
   [<ffffffff810cc9d2>] hrtimer_try_to_cancel+0x22/0xd0
   [<ffffffff81193cf1>] ? __kmalloc+0x211/0x230
   [<ffffffff810ccaa2>] hrtimer_cancel+0x22/0x30
   [<ffffffff810a3cb5>] free_fair_sched_group+0x25/0xd0
   [<ffffffff8108df46>] free_sched_group+0x16/0x40
   [<ffffffff810971bb>] sched_create_group+0x4b/0x80
   [<ffffffff810aa383>] sched_autogroup_create_attach+0x43/0x1c0
   [<ffffffff8107dc9c>] sys_setsid+0x7c/0x110
   [<ffffffff81647729>] system_call_fastpath+0x12/0x17

Check whether init_cfs_bandwidth() was called before calling
destroy_cfs_bandwidth().

Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
[ Move the check into destroy_cfs_bandwidth() to aid compilability. ]
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Paul Turner <pjt@google.com>
Cc: Ben Segall <bsegall@google.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Link: http://lkml.kernel.org/r/201412252210.GCC30204.SOMVFFOtQJFLOH@I-love.SAKURA.ne.jp
Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agosched/deadline: Avoid double-accounting in case of missed deadlines
Luca Abeni [Wed, 17 Dec 2014 10:50:32 +0000 (11:50 +0100)]
sched/deadline: Avoid double-accounting in case of missed deadlines

The dl_runtime_exceeded() function is supposed to ckeck if
a SCHED_DEADLINE task must be throttled, by checking if its
current runtime is <= 0. However, it also checks if the
scheduling deadline has been missed (the current time is
larger than the current scheduling deadline), further
decreasing the runtime if this happens.
This "double accounting" is wrong:

- In case of partitioned scheduling (or single CPU), this
  happens if task_tick_dl() has been called later than expected
  (due to small HZ values). In this case, the current runtime is
  also negative, and replenish_dl_entity() can take care of the
  deadline miss by recharging the current runtime to a value smaller
  than dl_runtime

- In case of global scheduling on multiple CPUs, scheduling
  deadlines can be missed even if the task did not consume more
  runtime than expected, hence penalizing the task is wrong

This patch fix this problem by throttling a SCHED_DEADLINE task
only when its runtime becomes negative, and not modifying the runtime

Signed-off-by: Luca Abeni <luca.abeni@unitn.it>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Acked-by: Juri Lelli <juri.lelli@gmail.com>
Cc: <stable@vger.kernel.org>
Cc: Dario Faggioli <raistlin@linux.it>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Link: http://lkml.kernel.org/r/1418813432-20797-3-git-send-email-luca.abeni@unitn.it
Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agosched/deadline: Fix migration of SCHED_DEADLINE tasks
Luca Abeni [Wed, 17 Dec 2014 10:50:31 +0000 (11:50 +0100)]
sched/deadline: Fix migration of SCHED_DEADLINE tasks

According to global EDF, tasks should be migrated between runqueues
without checking if their scheduling deadlines and runtimes are valid.
However, SCHED_DEADLINE currently performs such a check:
a migration happens doing:

deactivate_task(rq, next_task, 0);
set_task_cpu(next_task, later_rq->cpu);
activate_task(later_rq, next_task, 0);

which ends up calling dequeue_task_dl(), setting the new CPU, and then
calling enqueue_task_dl().

enqueue_task_dl() then calls enqueue_dl_entity(), which calls
update_dl_entity(), which can modify scheduling deadline and runtime,
breaking global EDF scheduling.

As a result, some of the properties of global EDF are not respected:
for example, a taskset {(30, 80), (40, 80), (120, 170)} scheduled on
two cores can have unbounded response times for the third task even
if 30/80+40/80+120/170 = 1.5809 < 2

This can be fixed by invoking update_dl_entity() only in case of
wakeup, or if this is a new SCHED_DEADLINE task.

Signed-off-by: Luca Abeni <luca.abeni@unitn.it>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Acked-by: Juri Lelli <juri.lelli@gmail.com>
Cc: <stable@vger.kernel.org>
Cc: Dario Faggioli <raistlin@linux.it>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Link: http://lkml.kernel.org/r/1418813432-20797-2-git-send-email-luca.abeni@unitn.it
Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agosched: Fix odd values in effective_load() calculations
Yuyang Du [Fri, 19 Dec 2014 00:29:56 +0000 (08:29 +0800)]
sched: Fix odd values in effective_load() calculations

In effective_load, we have (long w * unsigned long tg->shares) / long W,
when w is negative, it is cast to unsigned long and hence the product is
insanely large. Fix this by casting tg->shares to long.

Reported-by: Sasha Levin <sasha.levin@oracle.com>
Signed-off-by: Yuyang Du <yuyang.du@intel.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Dave Jones <davej@redhat.com>
Cc: Andrey Ryabinin <a.ryabinin@samsung.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Link: http://lkml.kernel.org/r/20141219002956.GA25405@intel.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agosched, fanotify: Deal with nested sleeps
Peter Zijlstra [Tue, 16 Dec 2014 15:28:38 +0000 (16:28 +0100)]
sched, fanotify: Deal with nested sleeps

As per e23738a7300a ("sched, inotify: Deal with nested sleeps").

fanotify_read is a wait loop with sleeps in. Wait loops rely on
task_struct::state and sleeps do too, since that's the only means of
actually sleeping. Therefore the nested sleeps destroy the wait loop
state and the wait loop breaks the sleep functions that assume
TASK_RUNNING (mutex_lock).

Fix this by using the new woken_wake_function and wait_woken() stuff,
which registers wakeups in wait and thereby allows shrinking the
task_state::state changes to the actual sleep part.

Reported-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Reported-by: Sedat Dilek <sedat.dilek@gmail.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Takashi Iwai <tiwai@suse.de>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Eric Paris <eparis@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Eric Paris <eparis@redhat.com>
Link: http://lkml.kernel.org/r/20141216152838.GZ3337@twins.programming.kicks-ass.net
Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agoperf/x86/uncore/hsw-ep: Handle systems with only two SBOXes
Andi Kleen [Tue, 6 Jan 2015 22:34:35 +0000 (14:34 -0800)]
perf/x86/uncore/hsw-ep: Handle systems with only two SBOXes

There was another report of a boot failure with a #GP fault in the
uncore SBOX initialization. The earlier work around was not enough
for this system.

The boot was failing while trying to initialize the third SBOX.

This patch detects parts with only two SBOXes and limits the number
of SBOX units to two there.

Stable material, as it affects boot problems on 3.18.

Tested-by: Andreas Oehler <andreas@oehler-net.de>
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: <stable@vger.kernel.org>
Cc: Arnaldo Carvalho de Melo <acme@kernel.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Yan, Zheng <zheng.z.yan@intel.com>
Link: http://lkml.kernel.org/r/1420583675-9163-1-git-send-email-andi@firstfloor.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agoperf/x86_64: Improve user regs sampling
Andy Lutomirski [Sun, 4 Jan 2015 18:36:20 +0000 (10:36 -0800)]
perf/x86_64: Improve user regs sampling

Perf reports user regs for kernel-mode samples so that samples can
be backtraced through user code.  The old code was very broken in
syscall context, resulting in useless backtraces.

The new code, in contrast, is still dangerously racy, but it should
at least work most of the time.

Tested-by: Jiri Olsa <jolsa@kernel.org>
Signed-off-by: Andy Lutomirski <luto@amacapital.net>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: chenggang.qcg@taobao.com
Cc: Wu Fengguang <fengguang.wu@intel.com>
Cc: Namhyung Kim <namhyung@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Arjan van de Ven <arjan@linux.intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Link: http://lkml.kernel.org/r/243560c26ff0f739978e2459e203f6515367634d.1420396372.git.luto@amacapital.net
Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agoperf: Move task_pt_regs sampling into arch code
Andy Lutomirski [Sun, 4 Jan 2015 18:36:19 +0000 (10:36 -0800)]
perf: Move task_pt_regs sampling into arch code

On x86_64, at least, task_pt_regs may be only partially initialized
in many contexts, so x86_64 should not use it without extra care
from interrupt context, let alone NMI context.

This will allow x86_64 to override the logic and will supply some
scratch space to use to make a cleaner copy of user regs.

Tested-by: Jiri Olsa <jolsa@kernel.org>
Signed-off-by: Andy Lutomirski <luto@amacapital.net>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: chenggang.qcg@taobao.com
Cc: Wu Fengguang <fengguang.wu@intel.com>
Cc: Namhyung Kim <namhyung@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Arjan van de Ven <arjan@linux.intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Arnaldo Carvalho de Melo <acme@kernel.org>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Jean Pihet <jean.pihet@linaro.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mark Salter <msalter@redhat.com>
Cc: Russell King <linux@arm.linux.org.uk>
Cc: Will Deacon <will.deacon@arm.com>
Cc: linux-arm-kernel@lists.infradead.org
Link: http://lkml.kernel.org/r/e431cd4c18c2e1c44c774f10758527fb2d1025c4.1420396372.git.luto@amacapital.net
Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agox86: Fix off-by-one in instruction decoder
Peter Zijlstra [Tue, 16 Dec 2014 10:46:14 +0000 (11:46 +0100)]
x86: Fix off-by-one in instruction decoder

Stephane reported that the PEBS fixup was broken by the recent commit to
the instruction decoder. The thing had an off-by-one which resulted in
not being able to decode the last instruction and always bail.

Reported-by: Stephane Eranian <eranian@google.com>
Fixes: 6ba48ff46f76 ("x86: Remove arbitrary instruction size limit in instruction decoder")
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: stable@vger.kernel.org # 3.18
Cc: <ak@linux.intel.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Liang Kan <kan.liang@intel.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Jim Keniston <jkenisto@us.ibm.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com>
Link: http://lkml.kernel.org/r/20141216104614.GV3337@twins.programming.kicks-ass.net
Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agoMerge tag 'perf-urgent-for-mingo' of git://git.kernel.org/pub/scm/linux/kernel/git...
Ingo Molnar [Fri, 9 Jan 2015 10:04:13 +0000 (11:04 +0100)]
Merge tag 'perf-urgent-for-mingo' of git://git./linux/kernel/git/acme/linux into perf/urgent

Pull perf/urgent fixes from Arnaldo Carvalho de Melo:

 - Free callchains when hist entries are deleted, plugging a massive leak in
   'top -g', where hist_entries (and its callchains) are decayed over time. (Namhyung Kim)

 - Fix segfault when showing callchain in the hists browser (report & top) (Namhyung Kim)

 - Fix children sort key behavior, and also the 'perf test 32' test that
   was failing due to reliance on undefined behaviour (Namhyung Kim)

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agomm, vmscan: prevent kswapd livelock due to pfmemalloc-throttled process being killed
Vlastimil Babka [Thu, 8 Jan 2015 22:32:40 +0000 (14:32 -0800)]
mm, vmscan: prevent kswapd livelock due to pfmemalloc-throttled process being killed

Charles Shirron and Paul Cassella from Cray Inc have reported kswapd
stuck in a busy loop with nothing left to balance, but
kswapd_try_to_sleep() failing to sleep.  Their analysis found the cause
to be a combination of several factors:

1. A process is waiting in throttle_direct_reclaim() on pgdat->pfmemalloc_wait

2. The process has been killed (by OOM in this case), but has not yet been
   scheduled to remove itself from the waitqueue and die.

3. kswapd checks for throttled processes in prepare_kswapd_sleep():

        if (waitqueue_active(&pgdat->pfmemalloc_wait)) {
                wake_up(&pgdat->pfmemalloc_wait);
return false; // kswapd will not go to sleep
}

   However, for a process that was already killed, wake_up() does not remove
   the process from the waitqueue, since try_to_wake_up() checks its state
   first and returns false when the process is no longer waiting.

4. kswapd is running on the same CPU as the only CPU that the process is
   allowed to run on (through cpus_allowed, or possibly single-cpu system).

5. CONFIG_PREEMPT_NONE=y kernel is used. If there's nothing to balance, kswapd
   encounters no voluntary preemption points and repeatedly fails
   prepare_kswapd_sleep(), blocking the process from running and removing
   itself from the waitqueue, which would let kswapd sleep.

So, the source of the problem is that we prevent kswapd from going to
sleep until there are processes waiting on the pfmemalloc_wait queue,
and a process waiting on a queue is guaranteed to be removed from the
queue only when it gets scheduled.  This was done to make sure that no
process is left sleeping on pfmemalloc_wait when kswapd itself goes to
sleep.

However, it isn't necessary to postpone kswapd sleep until the
pfmemalloc_wait queue actually empties.  To prevent processes from being
left sleeping, it's actually enough to guarantee that all processes
waiting on pfmemalloc_wait queue have been woken up by the time we put
kswapd to sleep.

This patch therefore fixes this issue by substituting 'wake_up' with
'wake_up_all' and removing 'return false' in the code snippet from
prepare_kswapd_sleep() above.  Note that if any process puts itself in
the queue after this waitqueue_active() check, or after the wake up
itself, it means that the process will also wake up kswapd - and since
we are under prepare_to_wait(), the wake up won't be missed.  Also we
update the comment prepare_kswapd_sleep() to hopefully more clearly
describe the races it is preventing.

Fixes: 5515061d22f0 ("mm: throttle direct reclaimers if PF_MEMALLOC reserves are low and swap is backed by network storage")
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Vladimir Davydov <vdavydov@parallels.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Michal Hocko <mhocko@suse.cz>
Acked-by: Rik van Riel <riel@redhat.com>
Cc: <stable@vger.kernel.org> [3.6+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agomemcg: fix destination cgroup leak on task charges migration
Vladimir Davydov [Thu, 8 Jan 2015 22:32:37 +0000 (14:32 -0800)]
memcg: fix destination cgroup leak on task charges migration

We are supposed to take one css reference per each memory page and per
each swap entry accounted to a memory cgroup.  However, during task
charges migration we take a reference to the destination cgroup twice
per each swap entry: first in mem_cgroup_do_precharge()->try_charge()
and then in mem_cgroup_move_swap_account(), permanently leaking the
destination cgroup.

The hunk taking the second reference seems to be a leftover from the
pre-00501b531c472 ("mm: memcontrol: rewrite charge API") era.  Remove it
to fix the leak.

Fixes: e8ea14cc6ead (mm: memcontrol: take a css reference for each charged page)
Signed-off-by: Vladimir Davydov <vdavydov@parallels.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Michal Hocko <mhocko@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agomm: memcontrol: switch soft limit default back to infinity
Johannes Weiner [Thu, 8 Jan 2015 22:32:35 +0000 (14:32 -0800)]
mm: memcontrol: switch soft limit default back to infinity

Commit 3e32cb2e0a12 ("mm: memcontrol: lockless page counters")
accidentally switched the soft limit default from infinity to zero,
which turns all memcgs with even a single page into soft limit excessors
and engages soft limit reclaim on all of them during global memory
pressure.  This makes global reclaim generally more aggressive, but also
inverts the meaning of existing soft limit configurations where unset
soft limits are usually more generous than set ones.

Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Michal Hocko <mhocko@suse.cz>
Acked-by: Vladimir Davydov <vdavydov@parallels.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agomm/debug_pagealloc: remove obsolete Kconfig options
Joonsoo Kim [Thu, 8 Jan 2015 22:32:32 +0000 (14:32 -0800)]
mm/debug_pagealloc: remove obsolete Kconfig options

These are obsolete since commit e30825f1869a ("mm/debug-pagealloc:
prepare boottime configurable") was merged.  So remove them.

[pebolle@tiscali.nl: find obsolete Kconfig options]
Signed-off-by: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Paul Bolle <pebolle@tiscali.nl>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Dave Hansen <dave@sr71.net>
Cc: Michal Nazarewicz <mina86@mina86.com>
Cc: Jungsoo Son <jungsoo.son@lge.com>
Acked-by: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agovfs: renumber FMODE_NONOTIFY and add to uniqueness check
David Drysdale [Thu, 8 Jan 2015 22:32:29 +0000 (14:32 -0800)]
vfs: renumber FMODE_NONOTIFY and add to uniqueness check

Fix clashing values for O_PATH and FMODE_NONOTIFY on sparc.  The
clashing O_PATH value was added in commit 5229645bdc35 ("vfs: add
nonconflicting values for O_PATH") but this can't be changed as it is
user-visible.

FMODE_NONOTIFY is only used internally in the kernel, but it is in the
same numbering space as the other O_* flags, as indicated by the comment
at the top of include/uapi/asm-generic/fcntl.h (and its use in
fs/notify/fanotify/fanotify_user.c).  So renumber it to avoid the clash.

All of this has happened before (commit 12ed2e36c98a: "fanotify:
FMODE_NONOTIFY and __O_SYNC in sparc conflict"), and all of this will
happen again -- so update the uniqueness check in fcntl_init() to
include __FMODE_NONOTIFY.

Signed-off-by: David Drysdale <drysdale@google.com>
Acked-by: David S. Miller <davem@davemloft.net>
Acked-by: Jan Kara <jack@suse.cz>
Cc: Heinrich Schuchardt <xypron.glpk@gmx.de>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Cc: Eric Paris <eparis@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoarch/blackfin/mach-bf533/boards/stamp.c: add linux/delay.h
Oleg Nesterov [Thu, 8 Jan 2015 22:32:26 +0000 (14:32 -0800)]
arch/blackfin/mach-bf533/boards/stamp.c: add linux/delay.h

build error

  arch/blackfin/mach-bf533/boards/stamp.c:834:2: error: implicit declaration of function 'mdelay'

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Reported-by: Wu Fengguang <fengguang.wu@intel.com>
Acked-by: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoocfs2: fix the wrong directory passed to ocfs2_lookup_ino_from_name() when link file
Xue jiufei [Thu, 8 Jan 2015 22:32:23 +0000 (14:32 -0800)]
ocfs2: fix the wrong directory passed to ocfs2_lookup_ino_from_name() when link file

In ocfs2_link(), the parent directory inode passed to function
ocfs2_lookup_ino_from_name() is wrong.  Parameter dir is the parent of
new_dentry not old_dentry.  We should get old_dir from old_dentry and
lookup old_dentry in old_dir in case another node remove the old dentry.

With this change, hard linking works again, when paths are relative with
at least one subdirectory.  This is how the problem was reproducable:

  # mkdir a
  # mkdir b
  # touch a/test
  # ln a/test b/test
  ln: failed to create hard link `b/test' => `a/test': No such file or  directory

However when creating links in the same dir, it worked well.

Now the link gets created.

Fixes: 0e048316ff57 ("ocfs2: check existence of old dentry in ocfs2_link()")
Signed-off-by: joyce.xue <xuejiufei@huawei.com>
Reported-by: Szabo Aron - UBIT <aron@ubit.hu>
Cc: Mark Fasheh <mfasheh@suse.com>
Cc: Joel Becker <jlbec@evilplan.org>
Tested-by: Aron Szabo <aron@ubit.hu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoMAINTAINERS: update rydberg's addresses
Henrik Rydberg [Thu, 8 Jan 2015 22:32:21 +0000 (14:32 -0800)]
MAINTAINERS: update rydberg's addresses

My ISP finally gave up on the old mail address, so I am moving things
over to bitmath.org instead.  Also change the status fields to better
reflect reality.

Signed-off-by: Henrik Rydberg <rydberg@bitmath.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agomm: protect set_page_dirty() from ongoing truncation
Johannes Weiner [Thu, 8 Jan 2015 22:32:18 +0000 (14:32 -0800)]
mm: protect set_page_dirty() from ongoing truncation

Tejun, while reviewing the code, spotted the following race condition
between the dirtying and truncation of a page:

__set_page_dirty_nobuffers()       __delete_from_page_cache()
  if (TestSetPageDirty(page))
                                     page->mapping = NULL
     if (PageDirty())
       dec_zone_page_state(page, NR_FILE_DIRTY);
       dec_bdi_stat(mapping->backing_dev_info, BDI_RECLAIMABLE);
    if (page->mapping)
      account_page_dirtied(page)
        __inc_zone_page_state(page, NR_FILE_DIRTY);
__inc_bdi_stat(mapping->backing_dev_info, BDI_RECLAIMABLE);

which results in an imbalance of NR_FILE_DIRTY and BDI_RECLAIMABLE.

Dirtiers usually lock out truncation, either by holding the page lock
directly, or in case of zap_pte_range(), by pinning the mapcount with
the page table lock held.  The notable exception to this rule, though,
is do_wp_page(), for which this race exists.  However, do_wp_page()
already waits for a locked page to unlock before setting the dirty bit,
in order to prevent a race where clear_page_dirty() misses the page bit
in the presence of dirty ptes.  Upgrade that wait to a fully locked
set_page_dirty() to also cover the situation explained above.

Afterwards, the code in set_page_dirty() dealing with a truncation race
is no longer needed.  Remove it.

Reported-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agomm: prevent endless growth of anon_vma hierarchy
Konstantin Khlebnikov [Thu, 8 Jan 2015 22:32:15 +0000 (14:32 -0800)]
mm: prevent endless growth of anon_vma hierarchy

Constantly forking task causes unlimited grow of anon_vma chain.  Each
next child allocates new level of anon_vmas and links vma to all
previous levels because pages might be inherited from any level.

This patch adds heuristic which decides to reuse existing anon_vma
instead of forking new one.  It adds counter anon_vma->degree which
counts linked vmas and directly descending anon_vmas and reuses anon_vma
if counter is lower than two.  As a result each anon_vma has either vma
or at least two descending anon_vmas.  In such trees half of nodes are
leafs with alive vmas, thus count of anon_vmas is no more than two times
bigger than count of vmas.

This heuristic reuses anon_vmas as few as possible because each reuse
adds false aliasing among vmas and rmap walker ought to scan more ptes
when it searches where page is might be mapped.

Link: http://lkml.kernel.org/r/20120816024610.GA5350@evergreen.ssec.wisc.edu
Fixes: 5beb49305251 ("mm: change anon_vma linking to fix multi-process server scalability issue")
[akpm@linux-foundation.org: fix typo, per Rik]
Signed-off-by: Konstantin Khlebnikov <koct9i@gmail.com>
Reported-by: Daniel Forrest <dan.forrest@ssec.wisc.edu>
Tested-by: Michal Hocko <mhocko@suse.cz>
Tested-by: Jerome Marchand <jmarchan@redhat.com>
Reviewed-by: Michal Hocko <mhocko@suse.cz>
Reviewed-by: Rik van Riel <riel@redhat.com>
Cc: <stable@vger.kernel.org> [2.6.34+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoexit: fix race between wait_consider_task() and wait_task_zombie()
Oleg Nesterov [Thu, 8 Jan 2015 22:32:12 +0000 (14:32 -0800)]
exit: fix race between wait_consider_task() and wait_task_zombie()

wait_consider_task() checks EXIT_ZOMBIE after EXIT_DEAD/EXIT_TRACE and
both checks can fail if we race with EXIT_ZOMBIE -> EXIT_DEAD/EXIT_TRACE
change in between, gcc needs to reload p->exit_state after
security_task_wait().  In this case ->notask_error will be wrongly
cleared and do_wait() can hang forever if it was the last eligible
child.

Many thanks to Arne who carefully investigated the problem.

Note: this bug is very old but it was pure theoretical until commit
b3ab03160dfa ("wait: completely ignore the EXIT_DEAD tasks").  Before
this commit "-O2" was probably enough to guarantee that compiler won't
read ->exit_state twice.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Reported-by: Arne Goedeke <el@laramies.com>
Tested-by: Arne Goedeke <el@laramies.com>
Cc: <stable@vger.kernel.org> [3.15+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoocfs2: remove bogus check in dlm_process_recovery_data
Joseph Qi [Thu, 8 Jan 2015 22:32:09 +0000 (14:32 -0800)]
ocfs2: remove bogus check in dlm_process_recovery_data

In dlm_process_recovery_data, only when dlm_new_lock failed the ret will
be set to -ENOMEM.  And in this case, newlock is definitely NULL.  So
test newlock is meaningless, remove it.

Signed-off-by: Joseph Qi <joseph.qi@huawei.com>
Reviewed-by: Alex Chen <alex.chen@huawei.com>
Reviewed-by: Mark Fasheh <mfasheh@suse.de>
Cc: Joel Becker <jlbec@evilplan.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoMerge branch 'rc-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/mmarek/kbuild
Linus Torvalds [Thu, 8 Jan 2015 22:35:00 +0000 (14:35 -0800)]
Merge branch 'rc-fixes' of git://git./linux/kernel/git/mmarek/kbuild

Pull kbuild fix from Michal Marek:
 "make mrproper / distclean stopped removing the generated debian/
  directory in v3.16.  This fixes it"

* 'rc-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/mmarek/kbuild:
  kbuild: Fix removal of the debian/ directory

9 years agoMakefile: include arch/*/include/generated/uapi before .../generated
Michal Marek [Thu, 8 Jan 2015 15:16:39 +0000 (16:16 +0100)]
Makefile: include arch/*/include/generated/uapi before .../generated

The introduction of the uapi directories in v3.7-rc1 moved some of the
generated headers from arch/*/include/generated to the uapi directory,
keeping the #include directives intact.

This creates a problem when bisecting, because the unversioned files are
not cleaned automatically by git and the compiler might include stale
headers as a result.  Instead of cleaning them in the Makefiles, promote
arch/*/include/generated/uapi in the search path.  Under normal
circumstances, there is no overlap between this uapi subdirectory and
its parent, so the include choices remain the same.  We keep
arch/*/include/generated/uapi in the USERINCLUDE variable so that it is
usable standalone.

Note that we cannot completely swap the order of the uapi and
kernel-only directories, since the headers in include/uapi/asm-generic
are meant to be wrapped by their include/asm-generic counterparts when
building kernel code.

Reported-by: "Nicholas A. Bellinger" <nab@linux-iscsi.org>
Reported-by: David Drysdale <dmd@lurklurk.org>
Signed-off-by: Michal Marek <mmarek@suse.cz>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoMerge tag 'pinctrl-v3.19-2' of git://git.kernel.org/pub/scm/linux/kernel/git/linusw...
Linus Torvalds [Thu, 8 Jan 2015 22:19:44 +0000 (14:19 -0800)]
Merge tag 'pinctrl-v3.19-2' of git://git./linux/kernel/git/linusw/linux-pinctrl

Pull pinctrl fixes from Linus Walleij:
 "Allright allright I've been lazy over christmas and New Years.  Here
  are a few collected pin control fixes eventually.  Details:

  A set of assorted pin control fixes for the Rockchip and STi drivers"

* tag 'pinctrl-v3.19-2' of git://git.kernel.org/pub/scm/linux/kernel/git/linusw/linux-pinctrl:
  pinctrl: st: Add irq_disable hook to st_gpio_irqchip
  pinctrl: st: avoid multiple mutex lock
  pinctrl: rockchip: Fix enable/disable/mask/unmask
  pinctrl: rockchip: Handle wakeup pins

9 years agoMerge tag 'pm+acpi-3.19-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael...
Linus Torvalds [Thu, 8 Jan 2015 22:11:03 +0000 (14:11 -0800)]
Merge tag 'pm+acpi-3.19-rc4' of git://git./linux/kernel/git/rafael/linux-pm

Pull power management and ACPI fixes from Rafael Wysocki:
 "These are an ACPI device power management initialization fix (-stable
  material), two commits renaming stuff in the ACPI processor driver to
  make it more suitable for ARM64 processors and a new ACPI backlight
  blacklist entry.

  Specifics:

   - Fix ACPI power management intialization for device objects
     corresponding to devices that are not present at the init time (the
     _STA control method returns 0 for them) and therefore should not be
     regarded as power manageable (Rafael J Wysocki).

   - Rename a structure field and two functions used by the ACPI
     processor driver to make them less tied to architectures that use
     APICs (both x86 and ia64) and more suitable for ARM64 processors
     (Hanjun Guo).

   - Add a disable_native_backlight quirk for Dell XPS15 L521X designed
     in an unusual way preventing native backlight from working on that
     machine (Hans de Goede)"

* tag 'pm+acpi-3.19-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
  ACPI / video: Add disable_native_backlight quirk for Dell XPS15 L521X
  ACPI / processor: Rename acpi_(un)map_lsapic() to acpi_(un)map_cpu()
  ACPI / processor: Convert apic_id to phys_id to make it arch agnostic
  ACPI / PM: Fix PM initialization for devices that are not present

9 years agoMerge tag 'keys-fixes-20150107' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Thu, 8 Jan 2015 21:52:16 +0000 (13:52 -0800)]
Merge tag 'keys-fixes-20150107' of git://git./linux/kernel/git/dhowells/linux-fs

Pull keyrings fixes from David Howells:
 "Two fixes:

   - Fix for the order in which things are done during key garbage
     collection to prevent named keyrings causing a crash
     [CVE-2014-9529].

   - Fix assoc_array to explicitly #include rcupdate.h to prevent
     compilation errors under certain circumstances"

* tag 'keys-fixes-20150107' of git://git.kernel.org/pub/scm/linux/kernel/git/dhowells/linux-fs:
  assoc_array: Include rcupdate.h for call_rcu() definition
  KEYS: close race between key lookup and freeing

9 years agoMerge tag 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mst/vhost
Linus Torvalds [Thu, 8 Jan 2015 21:05:56 +0000 (13:05 -0800)]
Merge tag 'for_linus' of git://git./linux/kernel/git/mst/vhost

pull virtio/vhost fixes from Michael Tsirkin:
 "This fixes a couple of bugs triggered by hot-unplug of virtio devices,
  as well as a regression in vhost-net"

* tag 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mst/vhost:
  vhost/net: length miscalculation
  virtio_pci: document why we defer kfree
  virtio_pci: defer kfree until release callback
  virtio_pci: device-specific release callback
  virtio: make del_vqs idempotent