kbuild: allow to use GCC toolchain not in Clang search path

commit ef8c4ed9db80261f397f0c0bf723684601ae3b52 upstream.

When using a GCC cross toolchain which is not in a compiled in
Clang search path, Clang reverts to the system assembler and
linker. This leads to assembler or linker errors, depending on
which tool is first used for a given architecture.

It seems that Clang is not searching $PATH for a matching
assembler or linker.

Make sure that Clang picks up the correct assembler or linker by
passing the cross compilers bin directory as search path.

This allows to use Clang provided by distributions with GCC
toolchains not in /usr/bin.

Change-Id: Ia5683ef2a9def6122219bb38d834ad45e2d204c2
Link: https://github.com/ClangBuiltLinux/linux/issues/78
Signed-off-by: Stefan Agner <stefan@agner.ch>
Reviewed-and-tested-by: Nick Desaulniers <ndesaulniers@google.com>
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
[nc: Adjust context]
Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

FROMLIST: arm64: Build vDSO with -ffixed-x18

The vDSO needs to be build with x18 reserved in order to accommodate
userspace platform ABIs built on top of Linux that use the register
to carry inter-procedural state, as provided for by the AAPCS.
An example of such a platform ABI is the one that will be used by an
upcoming version of Android.

Although this change is currently a no-op due to the fact that the vDSO
is currently implemented in pure assembly on arm64, it is necessary
in order to prepare for another change [1] that will add C code to
the vDSO.

[1] https://patchwork.kernel.org/patch/10044501/

Change-Id: Icaac4b1c9127d81d754d3b8688274e9afc781760
Signed-off-by: Peter Collingbourne <pcc@google.com>
Signed-off-by: Vincenzo Frascino <vincenzo.frascino@arm.com>
Cc: Mark Salyzyn <salyzyn@google.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: linux-arm-kernel@lists.infradead.org

FROMLIST: arm64: vdso32: Use full path to Clang instead of relying on PATH

Currently, in order to build the compat VDSO with Clang, this format
has to be used:

  PATH=${BIN_FOLDER}:${PATH} make CC=clang

Prior to the addition of this file, this format would also be
acceptable:

  make CC=${BIN_FOLDER}/clang

This is because the vdso32 Makefile uses cc-name instead of CC. After
this path, CC will still evaluate to clang for the first case as
expected but now the second case will use the specified Clang, rather
than the host's copy, which may not be compatible as shown below.

/usr/bin/as: unrecognized option '-mfloat-abi=soft'
clang-6.0: error: assembler command failed with exit code 1

Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>
(cherry picked from https://patchwork.kernel.org/patch/10419665)
Bug: 80184372
Change-Id: If90a5a4edbc2b5883b4c78161081ebeafbebdcde

ANDROID: clock_gettime(CLOCK_BOOTTIME,) slows down >20x

clock_gettime(CLOCK_BOOTTIME,) slows down after significant
accumulation of suspend time creating a large offset between it and
CLOCK_MONOTONIC time.  The __iter_div_u64_rem() is only for the usage
of adding a few second+nanosecond times and saving cycles on more
expensive remainder and division operations, but iterates one second
at a time which quickly goes out of scale in CLOCK_BOOTTIME's case
since it was specified as nanoseconds only.

The fix is to split off seconds from the boot time and cap the
nanoseconds so that __iter_div_u64_rem does not iterate.

Signed-off-by: Mark Salyzyn <salyzyn@google.com>
Bug: 72406285
Change-Id: Ia647ef1e76b7ba3b0c003028d4b3b955635adabb

ANDROID: arm64: vdso32: fix CC=clang build

clang build had wrong --gcc-toolchain for cross compiler.

Signed-off-by: Mark Salyzyn <salyzyn@google.com>
Bug: 63737556
Change-Id: Ib0c3bbb250943a0309da021ba5335a8320b584f4

ANDROID: CROSS_COMPILE_ARM32 must work if CONFIG_COMPAT_VDSO

Prevent surprise loss of vdso32 support.

Signed-off-by: Mark Salyzyn <salyzyn@google.com>
Bug: 63737556
Bug: 20045882
Bug: 19198045
Change-Id: I8b381f7649b95b298ea9e1a99aa3794c7bc08d09

FROMLIST: BACKPORT: [PATCH 6/6] arm64: Wire up and expose the new compat vDSO

(cherry pick from url https://patchwork.kernel.org/patch/10060447/)

Expose the new compat vDSO via the COMPAT_VDSO config option.

The option is not enabled in defconfig because we really need a 32-bit
compiler this time, and we rely on the user to provide it themselves
by setting CROSS_COMPILE_ARM32. Therefore enabling the option by
default would make little sense, since the user must explicitly set a
non-standard environment variable anyway.

CONFIG_COMPAT_VDSO is not directly used in the code, because we want
to ignore it (build as if it were not set) if the user didn't set
CROSS_COMPILE_ARM32. If the variable has been set to a valid prefix,
CONFIG_VDSO32 will be set; this is the option that the code and
Makefiles test.

For more flexibility, like CROSS_COMPILE, CROSS_COMPILE_ARM32 can also
be set via CONFIG_CROSS_COMPILE_ARM32 (the environment variable
overrides the config option, as expected).

Signed-off-by: Kevin Brodsky <kevin.brodsky@arm.com>
Also needs to set CONFIG_ARM_ARCH_TIMER_VCT_ACCESS when CONFIG_VDSO32
is selected.

Signed-off-by: Mark Salyzyn <salyzyn@android.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Dave Martin <Dave.Martin@arm.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: linux-arm-kernel@lists.infradead.org
Cc: linux-kernel@vger.kernel.org
Bug: 63737556
Bug: 20045882
Change-Id: Ie8a7d6c2b5ba3edca591a9a953ce99ec792da882

FROMLIST: BACKPORT: [PATCH 5/6] arm64: compat: 32-bit vDSO setup

(cherry pick from url https://patchwork.kernel.org/patch/10060459/)

If the compat vDSO is enabled, install it in compat processes. In this
case, the compat vDSO replaces the sigreturn page (it provides its own
sigreturn trampolines).

Signed-off-by: Kevin Brodsky <kevin.brodsky@arm.com>
Signed-off-by: Mark Salyzyn <salyzyn@android.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Dave Martin <Dave.Martin@arm.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: linux-arm-kernel@lists.infradead.org
Cc: linux-kernel@vger.kernel.org
Bug: 63737556
Bug: 20045882
Change-Id: Ia6acf4c3ffea636bc750ac00853ea762c182e5b5

FROMLIST: [PATCH 4/6] arm64: compat: Add a 32-bit vDSO

(cherry pick from url https://patchwork.kernel.org/patch/10060445/)

Provide the files necessary for building a compat (AArch32) vDSO in
kernel/vdso32.

This is mostly an adaptation of the arm vDSO. The most significant
change in vgettimeofday.c is the use of the arm64 vdso_data struct,
allowing the vDSO data page to be shared between the 32 and 64-bit
vDSOs. Additionally, a different set of barrier macros is used (see
aarch32-barrier.h), as we want to support old 32-bit compilers that
may not support ARMv8 and its new barrier arguments (*ld).

In addition to the time functions, sigreturn trampolines are also
provided, aiming at replacing those in the sigreturn page as the
latter don't provide any unwinding information (and it's easier to
have just one "user code" page). arm-specific unwinding directives are
used, based on glibc's implementation. Symbol offsets are made
available to the kernel using the same method as the 64-bit vDSO.

There is unfortunately an important caveat: we cannot get away with
hand-coding 32-bit instructions like in kernel/kuser32.S, this time we
really need a 32-bit compiler. The compat vDSO Makefile relies on
CROSS_COMPILE_ARM32 to provide a 32-bit compiler, appropriate logic
will be added to the arm64 Makefile later on to ensure that an attempt
to build the compat vDSO is made only if this variable has been set
properly.

Signed-off-by: Kevin Brodsky <kevin.brodsky@arm.com>
Take an effort to recode the arm64 vdso code from assembler to C
previously submitted by Andrew Pinski <apinski@cavium.com>, rework
it for use in both arm and arm64, overlapping any optimizations
for each architecture.

Signed-off-by: Mark Salyzyn <salyzyn@android.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Dave Martin <Dave.Martin@arm.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: linux-arm-kernel@lists.infradead.org
Cc: linux-kernel@vger.kernel.org
Bug: 63737556
Bug: 20045882
Change-Id: I3fb9d21b29bd9fec1408f2274d090e6def546b0d

FROMLIST: [PATCH 3/6] arm64: Refactor vDSO init/setup

(cherry pick from url https://patchwork.kernel.org/patch/10060439/)

Move the logic for setting up mappings and pages for the vDSO into
static functions. This makes the vDSO setup code more consistent with
the compat side and will allow to reuse it for the future compat vDSO.

Signed-off-by: Kevin Brodsky <kevin.brodsky@arm.com>
Signed-off-by: Mark Salyzyn <salyzyn@android.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Dave Martin <Dave.Martin@arm.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: linux-arm-kernel@lists.infradead.org
Cc: linux-kernel@vger.kernel.org
Bug: 63737556
Bug: 20045882
Change-Id: I13e84479591091669190360f2a7f4d04462e6344

FROMLIST: [PATCH 2/6] arm64: elf: Set AT_SYSINFO_EHDR in compat processes

(cherry pick from url https://patchwork.kernel.org/patch/10060431/)

If the compat vDSO is enabled, we need to set AT_SYSINFO_EHDR in the
auxiliary vector of compat processes to the address of the vDSO code
page, so that the dynamic linker can find it (just like the regular vDSO).

Note that we cast context.vdso to Elf64_Off, instead of elf_addr_t,
because elf_addr_t is Elf32_Off in compat_binfmt_elf.c, and casting
context.vdso to u32 would trigger a pointer narrowing warning.

Signed-off-by: Kevin Brodsky <kevin.brodsky@arm.com>
Signed-off-by: Mark Salyzyn <salyzyn@android.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Dave Martin <Dave.Martin@arm.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: linux-arm-kernel@lists.infradead.org
Cc: linux-kernel@vger.kernel.org
Bug: 63737556
Bug: 20045882
Change-Id: I5d0b191d3b2f4c0b2ec31fe9faef0246253635ce

FROMLIST: [PATCH 1/6] arm64: compat: Use vDSO sigreturn trampolines if available

(cherry pick from url https://patchwork.kernel.org/patch/10060449/)

If the compat vDSO is enabled, it replaces the sigreturn page.
Therefore, we use the sigreturn trampolines the vDSO provides instead.

Signed-off-by: Kevin Brodsky <kevin.brodsky@arm.com>
Signed-off-by: Mark Salyzyn <salyzyn@android.com>
Bug: 63737556
Bug: 20045882
Change-Id: Ic0933741e321e1bf66409b7e190a776f12948024

FROMLIST: lib: vdso: add support for time

(cherry pick from url https://patchwork.kernel.org/patch/10053549/)

Add time() vdso support to match up with existing support in the x86's
vdso.  Currently benefitting arm and arm64 which uses the common
vgettimeofday.c implementation.  On arm provides about a ~14 fold
improvement in speed over the straight syscall, and about a ~5 fold
improvement in speed over an alternate library implementation that
relies on the vdso call to gettimeofday to fulfill the request.

We can provide __vdso_time even if we can not provide a speed
enhanced __vdso_gettimeofday.

Signed-off-by: Mark Salyzyn <salyzyn@android.com>
Bug: 63737556
Bug: 20045882
Change-Id: I0bb3c6bafe57f9ed69350e2dd54edaae58316e8f

FROMLIST: [PATCH] arm64: compat: Expose offset to registers in sigframes

(cherry picked from url https://patchwork.kernel.org/patch/10006025/)

This will be needed to provide unwinding information in compat
sigreturn trampolines, part of the future compat vDSO. There is no
obvious header the compat_sig* struct's should be moved to, so let's
put them in signal32.h.

Also fix minor style issues reported by checkpatch.

Signed-off-by: Kevin Brodsky <kevin.brodsky@arm.com>
Signed-off-by: Mark Salyzyn <salyzyn@android.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Andy Gross <andy.gross@linaro.org>
Cc: Dave Martin <Dave.Martin@arm.com>
Cc: Eric W. Biederman <ebiederm@xmission.com>
Cc: linux-arm-kernel@lists.infradead.org
Cc: linux-kernel@vger.kernel.org
Bug: 63737556
Bug: 20045882
Change-Id: I9c23dd6b56ca48c0953cbf78ccb7b49ded906052

FROMLIST: [PATCH v5 12/12] lib: vdso: do not expose gettimeofday, if no arch supported timer

(cherry pick from url https://patchwork.kernel.org/patch/10044539/)

Take an effort to recode the arm64 vdso code from assembler to C
previously submitted by Andrew Pinski <apinski@cavium.com>, rework
it for use in both arm and arm64, overlapping any optimizations
for each architecture. But instead of landing it in arm64, land the
result into lib/vdso and unify both implementations to simplify
future maintenance.

If ARCH_PROVIDES_TIMER is not defined, do not expose gettimeofday.
libc will default directly to syscall.  Also ifdef clock_gettime
switch cases and stubs if not supported and other unused components.

Signed-off-by: Mark Salyzyn <salyzyn@android.com>
Cc: James Morse <james.morse@arm.com>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Dmitry Safonov <dsafonov@virtuozzo.com>
Cc: John Stultz <john.stultz@linaro.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Laura Abbott <labbott@redhat.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Cc: Andy Gross <andy.gross@linaro.org>
Cc: Kevin Brodsky <kevin.brodsky@arm.com>
Cc: Andrew Pinski <apinski@cavium.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Cc: linux-arm-kernel@lists.infradead.org
Bug: 63737556
Bug: 20045882
Change-Id: I362a7114db0aac800e16eb90d14a8739e18f42e4

FROMLIST: [PATCH v5 11/12] lib: vdso: Add support for CLOCK_BOOTTIME

(cherry pick from url https://patchwork.kernel.org/patch/10044503/)

Take an effort to recode the arm64 vdso code from assembler to C
previously submitted by Andrew Pinski <apinski@cavium.com>, rework
it for use in both arm and arm64, overlapping any optimizations
for each architecture. But instead of landing it in arm64, land the
result into lib/vdso and unify both implementations to simplify
future maintenance.

Add a case for CLOCK_BOOTTIME as it is popular for measuring
relative time on systems expected to suspend() or hibernate().

Android uses CLOCK_BOOTTIME for all relative time measurements
and timeouts. Switching to vdso reduced CPU utilization and improves
accuracy. There is also a desire by some partners to switch all
logging over to CLOCK_BOOTTIME, and thus this operation alone would
contribute to a near percentile CPU load.

Signed-off-by: Mark Salyzyn <salyzyn@android.com>
Cc: James Morse <james.morse@arm.com>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Dmitry Safonov <dsafonov@virtuozzo.com>
Cc: John Stultz <john.stultz@linaro.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Laura Abbott <labbott@redhat.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Cc: Andy Gross <andy.gross@linaro.org>
Cc: Kevin Brodsky <kevin.brodsky@arm.com>
Cc: Andrew Pinski <apinski@cavium.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Cc: linux-arm-kernel@lists.infradead.org
Bug: 63737556
Bug: 20045882
Change-Id: I76c26b054baf7f1100e03c65d6b16fe649b883b1

FROMLIST: [PATCH v5 10/12] arm64: vdso: replace gettimeofday.S with global vgettimeofday.C

(cherry picked from url https://patchwork.kernel.org/patch/10044501/)

Take an effort from the previous 9 patches to recode the arm64 vdso
code from assembler to C previously submitted by
Andrew Pinski <apinski@cavium.com>, rework it for use in both arm and
arm64, overlapping any optimizations for each architecture. But
instead of landing it in arm64, land the result into lib/vdso and
unify both implementations to simplify future maintenance.

apinski@cavium.com makes the following claims in the original patch:

This allows the compiler to optimize the divide by 1000 and remove
the other divides.

On ThunderX, gettimeofday improves by 32%.  On ThunderX 2,
gettimeofday improves by 18%.

Note I noticed a bug in the old (arm64) implementation of
__kernel_clock_getres; it was checking only the lower 32bits of the
pointer; this would work for most cases but could fail in a few.

<end of claim>

Signed-off-by: Mark Salyzyn <salyzyn@android.com>
Cc: James Morse <james.morse@arm.com>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Dmitry Safonov <dsafonov@virtuozzo.com>
Cc: John Stultz <john.stultz@linaro.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Laura Abbott <labbott@redhat.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Cc: Andy Gross <andy.gross@linaro.org>
Cc: Kevin Brodsky <kevin.brodsky@arm.com>
Cc: Andrew Pinski <apinski@cavium.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Cc: linux-arm-kernel@lists.infradead.org
Bug: 63737556
Bug: 20045882
Change-Id: I71ff27ff5bfa323354fda6867b01ec908d8d6cbd

FROMLIST: [PATCH v5 09/12] arm: vdso: move vgettimeofday.c to lib/vdso/

(cherry pick from url https://patchwork.kernel.org/patch/10044497/)

Take an effort to recode the arm64 vdso code from assembler to C
previously submitted by Andrew Pinski <apinski@cavium.com>, rework
it for use in both arm and arm64, overlapping any optimizations
for each architecture. But instead of landing it in arm64, land the
result into lib/vdso and unify both implementations to simplify
future maintenance.

Declare arch/arm/vdso/vgettimeofday.c to be a candidate for a global
implementation of the vdso timer calls.  The hope is that new
architectures can take advantage of the current unification of
arm and arm64 implementations.

We urge future efforts to merge their implementations into the
global vgettimeofday.c file and thus provide functional parity.

Signed-off-by: Mark Salyzyn <salyzyn@android.com>
Cc: James Morse <james.morse@arm.com>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Dmitry Safonov <dsafonov@virtuozzo.com>
Cc: John Stultz <john.stultz@linaro.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Laura Abbott <labbott@redhat.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Cc: Andy Gross <andy.gross@linaro.org>
Cc: Kevin Brodsky <kevin.brodsky@arm.com>
Cc: Andrew Pinski <apinski@cavium.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Cc: linux-arm-kernel@lists.infradead.org
Bug: 63737556
Bug: 20045882
Change-Id: If7da1d8144684d52ed9520a581e6023c623df931

FROMLIST: [PATCH v5 08/12] arm: vdso: Add ARCH_CLOCK_FIXED_MASK

(cherry picked from url https://patchwork.kernel.org/patch/10044543/)

Take an effort to recode the arm64 vdso code from assembler to C
previously submitted by Andrew Pinski <apinski@cavium.com>, rework
it for use in both arm and arm64, overlapping any optimizations
for each architecture. But instead of landing it in arm64, land the
result into lib/vdso and unify both implementations to simplify
future maintenance.

Add ARCH_CLOCK_FIXED_MASK as an efficiency since arm64 has no
purpose for cs_mask vdso_data variable.

Signed-off-by: Mark Salyzyn <salyzyn@android.com>
Cc: James Morse <james.morse@arm.com>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Dmitry Safonov <dsafonov@virtuozzo.com>
Cc: John Stultz <john.stultz@linaro.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Laura Abbott <labbott@redhat.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Cc: Andy Gross <andy.gross@linaro.org>
Cc: Kevin Brodsky <kevin.brodsky@arm.com>
Cc: Andrew Pinski <apinski@cavium.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Cc: linux-arm-kernel@lists.infradead.org
Bug: 63737556
Bug: 20057882
Change-Id: Iadf94bed6166d2ee43bb46bdf54636618e4b8854

FROMLIST: [PATCH v5 07/12] arm: vdso: disable profiling

(cherry pick from url https://patchwork.kernel.org/patch/10044491/)

Take an effort to recode the arm64 vdso code from assembler to C
previously submitted by Andrew Pinski <apinski@cavium.com>, rework
it for use in both arm and arm64, overlapping any optimizations
for each architecture. But instead of landing it in arm64, land the
result into lib/vdso and unify both implementations to simplify
future maintenance.

Make sure kasan and ubsan profiling, and kcov instrumentation,
is turned off for VDSO code.

Signed-off-by: Mark Salyzyn <salyzyn@android.com>
Cc: James Morse <james.morse@arm.com>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Dmitry Safonov <dsafonov@virtuozzo.com>
Cc: John Stultz <john.stultz@linaro.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Laura Abbott <labbott@redhat.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Cc: Andy Gross <andy.gross@linaro.org>
Cc: Kevin Brodsky <kevin.brodsky@arm.com>
Cc: Andrew Pinski <apinski@cavium.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Cc: linux-arm-kernel@lists.infradead.org
Bug: 63737556
Bug: 20045882
Change-Id: I2b44c1edd81665b8bb235a65ba642767c35f1e61

FROMLIST: [PATCH v5 06/12] arm: vdso: add support for clock_getres

(cherry picked from url https://patchwork.kernel.org/patch/10044545/)

Take an effort to recode the arm64 vdso code from assembler to C
previously submitted by Andrew Pinski <apinski@cavium.com>, rework
it for use in both arm and arm64, overlapping any optimizations
for each architecture. But instead of landing it in arm64, land the
result into lib/vdso and unify both implementations to simplify
future maintenance.

Add clock_getres vdso support to match up with existing support in
the arm64's vdso.

Signed-off-by: Mark Salyzyn <salyzyn@android.com>
Cc: James Morse <james.morse@arm.com>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Dmitry Safonov <dsafonov@virtuozzo.com>
Cc: John Stultz <john.stultz@linaro.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Laura Abbott <labbott@redhat.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Cc: Andy Gross <andy.gross@linaro.org>
Cc: Kevin Brodsky <kevin.brodsky@arm.com>
Cc: Andrew Pinski <apinski@cavium.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Cc: linux-arm-kernel@lists.infradead.org
Bug: 63737556
Bug: 20045882
Change-Id: Ie37bf76d2992027f06a2cdd001d8654a860d2aac

FROMLIST: [PATCH v6 05/12] arm: vdso: Add support for CLOCK_MONOTONIC_RAW

(cherry pick from url https://patchwork.kernel.org/patch/10052099/)

Take an effort to recode the arm64 vdso code from assembler to C
previously submitted by Andrew Pinski <apinski@cavium.com>, rework
it for use in both arm and arm64, overlapping any optimizations
for each architecture. But instead of landing it in arm64, land the
result into lib/vdso and unify both implementations to simplify
future maintenance.

Add a case for CLOCK_MONOTONIC_RAW to match up with support that
is available in arm64's vdso.

Signed-off-by: Mark Salyzyn <salyzyn@android.com>
Cc: James Morse <james.morse@arm.com>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Dmitry Safonov <dsafonov@virtuozzo.com>
Cc: John Stultz <john.stultz@linaro.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Laura Abbott <labbott@redhat.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Cc: Andy Gross <andy.gross@linaro.org>
Cc: Kevin Brodsky <kevin.brodsky@arm.com>
Cc: Andrew Pinski <apinski@cavium.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Cc: linux-arm-kernel@lists.infradead.org
Bug: 63737556
Bug: 20045882
Change-Id: If9c09d131e236ba4a483dbc122e6b876f471df72

FROMLIST: [PATCH v5 04/12] arm: vdso: do calculations outside reader loops

(cherry picked from url https://patchwork.kernel.org/patch/10044477/)

Take an effort to recode the arm64 vdso code from assembler to C
previously submitted by Andrew Pinski <apinski@cavium.com>, rework
it for use in both arm and arm64, overlapping any optimizations
for each architecture. But instead of landing it in arm64, land the
result into lib/vdso and unify both implementations to simplify
future maintenance.

In variable timer reading loops, pick up just the values until all
are synchronized, then outside of loop pick up cntvct and perform
calculations to determine final offset, shifted and multiplied
output value.

This replaces get_ns with get_clock_shifted_nsec as cntvct reader.

Signed-off-by: Mark Salyzyn <salyzyn@android.com>
Cc: James Morse <james.morse@arm.com>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Dmitry Safonov <dsafonov@virtuozzo.com>
Cc: John Stultz <john.stultz@linaro.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Laura Abbott <labbott@redhat.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Cc: Andy Gross <andy.gross@linaro.org>
Cc: Kevin Brodsky <kevin.brodsky@arm.com>
Cc: Andrew Pinski <apinski@cavium.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Cc: linux-arm-kernel@lists.infradead.org
Bug: 63737556
Bug: 20045882
Change-Id: I8008197f08485ef89b267128e41624ff69c33f6b

FROMLIST: [PATCH v5 03/12] arm: vdso: inline assembler operations to compiler.h

(cherry picked from commit https://patchwork.kernel.org/patch/10044507/)

Take an effort to recode the arm64 vdso code from assembler to C
previously submitted by Andrew Pinski <apinski@cavium.com>, rework
it for use in both arm and arm64, overlapping any optimizations
for each architecture. But instead of landing it in arm64, land the
result into lib/vdso and unify both implementations to simplify
future maintenance.

Move compiler-specific code to a local compiler.h file:

- CONFIG_AEABI dependency check.
- System call fallback functions standardized into a
  DEFINE_FALLBACK macro.
- Replace arch_counter_get_cntvct() with arch_vdso_read_counter.
- Deal with architecture specific unresolved references emitted
  by GCC.
- Optimize handling of fallback calls in callers.
- For time functions that always return success, do not waste time
  checking return value for switch to fallback.
- Optimize unlikely nullptr checking in __vdso_gettimeofday,
  if tv null no need to proceed to fallback, as vdso is still
  capable of filling in the tv values.

Signed-off-by: Mark Salyzyn <salyzyn@android.com>
Cc: James Morse <james.morse@arm.com>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Dmitry Safonov <dsafonov@virtuozzo.com>
Cc: John Stultz <john.stultz@linaro.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Laura Abbott <labbott@redhat.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Cc: Andy Gross <andy.gross@linaro.org>
Cc: Kevin Brodsky <kevin.brodsky@arm.com>
Cc: Andrew Pinski <apinski@cavium.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Cc: linux-arm-kernel@lists.infradead.org
Bug: 63737556
Bug: 20045882
Change-Id: I468e4c32b5136d199982bf25df8967321e384d90

FROMLIST: [PATCH v5 02/12] arm: vdso: add include file defining __get_datapage()

(cherry picked from url https://patchwork.kernel.org/patch/10044481/)

Take an effort to recode the arm64 vdso code from assembler to C
previously submitted by Andrew Pinski <apinski@cavium.com>, rework
it for use in both arm and arm64, overlapping any optimizations
for each architecture. But instead of landing it in arm64, land the
result into lib/vdso and unify both implementations to simplify
future maintenance.

Define the prototype for __get_datapage() in local datapage.h header.
Rename all vdata variable that point to the datapage shortened to vd
to relect a consistent and concise style. Make sure that all
references to the datapage in vdso operations are readonly (const).
Make sure datapage is first parameter to all subroutines to also
be consistent.

Signed-off-by: Mark Salyzyn <salyzyn@android.com>
Cc: James Morse <james.morse@arm.com>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Dmitry Safonov <dsafonov@virtuozzo.com>
Cc: John Stultz <john.stultz@linaro.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Laura Abbott <labbott@redhat.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Cc: Andy Gross <andy.gross@linaro.org>
Cc: Kevin Brodsky <kevin.brodsky@arm.com>
Cc: Andrew Pinski <apinski@cavium.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Cc: linux-arm-kernel@lists.infradead.org
Bug: 63737556
Bug: 20045882
Change-Id: I9512b49d36d53ca1b71d3ff82219a7c64e0fc613

FROMLIST: [PATCH v5 01/12] arm: vdso: rename vdso_datapage variables

(cherry picked from url https://patchwork.kernel.org/patch/10044505/)

Take an effort to recode the arm64 vdso code from assembler to C
previously submitted by Andrew Pinski <apinski@cavium.com>, rework
it for use in both arm and arm64, overlapping any optimizations
for each architecture. But instead of landing it in arm64, land the
result into lib/vdso and unify both implementations to simplify
future maintenance.

Rename seq_count to tb_seq_count. Rename tk_is_cntvct to use_syscall.
Rename cs_mult to cs_mono_mult. All to align with the variables in the
arm64 vdso datapage. Rework vdso_read_begin() and vdso_read_retry()
functions to reflect modern access patterns for tb_seq_count field.

Update copyright message to reflect the start of the contributions in
this series.

Signed-off-by: Mark Salyzyn <salyzyn@android.com>
Cc: James Morse <james.morse@arm.com>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Dmitry Safonov <dsafonov@virtuozzo.com>
Cc: John Stultz <john.stultz@linaro.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Laura Abbott <labbott@redhat.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Cc: Andy Gross <andy.gross@linaro.org>
Cc: Kevin Brodsky <kevin.brodsky@arm.com>
Cc: Andrew Pinski <apinski@cavium.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Cc: linux-arm-kernel@lists.infradead.org
Bug: 63737556
Bug: 20045882
Change-Id: I13f16e71b1ecba3d72b999caafef72e3c7f48dfe

FROMLIST: [PATCH v2 3/3] arm64: compat: Add CONFIG_KUSER_HELPERS

(cherry picked from url http://lkml.iu.edu/hypermail/linux/kernel/1709.1/01903.html)

Make it possible to disable the kuser helpers by adding a KUSER_HELPERS
config option (enabled by default). When disabled, all kuser
helpers-related code is removed from the kernel and no mapping is done
at the fixed high address (0xffff0000); any attempt to use a kuser
helper from a 32-bit process will result in a segfault.

Signed-off-by: Kevin Brodsky <kevin.brodsky@arm.com>
Signed-off-by: Mark Salyzyn <salyzyn@android.com>
Bug: 9674955
Bug: 63737556
Bug: 20045882
Change-Id: Ie8c543301d39bfe88ef71fb6a669e571914b117b

FROMLIST: [PATCH v2 2/3] arm64: compat: Split the sigreturn trampolines and kuser helpers (assembler sources)

(cherry picked from url http://lkml.iu.edu/hypermail/linux/kernel/1709.1/01902.html)

AArch32 processes are currently installed a special [vectors] page that
contains the sigreturn trampolines and the kuser helpers, at the fixed
address mandated by the kuser helpers ABI.

Having both functionalities in the same page has become problematic,
because:

* It makes it impossible to disable the kuser helpers (the sigreturn
  trampolines cannot be removed), which is possible on arm.

* A future 32-bit vDSO would provide the sigreturn trampolines itself,
  making those in [vectors] redundant.

This patch addresses the problem by moving the sigreturn trampolines
sources to its own file.  Wrapped the comments to reduce the wrath of
checkpatch.pl.

Signed-off-by: Kevin Brodsky <kevin.brodsky@arm.com>
Signed-off-by: Mark Salyzyn <salyzyn@android.com>
Bug: 9674955
Bug: 63737556
Bug: 20045882
Change-Id: I1d7b96e7cfbe979ecf4cb4996befd1f3ae0e64fd

FROMLIST: [PATCH v2 1/3] arm64: compat: Split the sigreturn trampolines and kuser helpers (C sources)

(cherry picked from url http://lkml.iu.edu/hypermail/linux/kernel/1709.1/01901.html)

AArch32 processes are currently installed a special [vectors] page that
contains the sigreturn trampolines and the kuser helpers, at the fixed
address mandated by the kuser helpers ABI.

Having both functionalities in the same page has become problematic,
because:

* It makes it impossible to disable the kuser helpers (the sigreturn
  trampolines cannot be removed), which is possible on arm.

* A future 32-bit vDSO would provide the sigreturn trampolines itself,
  making those in [vectors] redundant.

This patch addresses the problem by moving the sigreturn trampolines to
a separate [sigpage] page, mirroring [sigpage] on arm.

Even though [vectors] has always been a misnomer on arm64/compat, as
there is no AArch32 vector there (and now only the kuser helpers),
its name has been left unchanged, for compatibility with arm (there
are reports of software relying on [vectors] being there as the last
mapping in /proc/maps).

mm->context.vdso used to point to the [vectors] page, which is
unnecessary (as its address is fixed). It now points to the [sigpage]
page (whose address is randomized like a vDSO).

Signed-off-by: Kevin Brodsky <kevin.brodsky@arm.com>
Signed-off-by: Mark Salyzyn <salyzyn@android.com>
Bug: 9674955
Bug: 63737556
Bug: 20045882
Change-Id: I52a56ea71d7326df8c784f90eb73b5c324fe9d20

Revert "kbuild: use -Oz instead of -Os when using clang"

commit a75bb4eb9e565b9f5115e2e8c07377ce32cbe69a upstream.

The clang option -Oz enables *aggressive* optimization for size,
which doesn't necessarily result in smaller images, but can have
negative impact on performance. Switch back to the less aggressive
-Os.

This reverts commit 6748cb3c299de1ffbe56733647b01dbcc398c419.

Suggested-by: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Matthias Kaehlcke <mka@chromium.org>
Reviewed-by: Nick Desaulniers <ndesaulniers@google.com>
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

defconfig: Disable stack frame size warning

Signed-off-by: Michael Benedict <michaelbt@live.com>

bcmdhd4361: fake platform version

Signed-off-by: Michael Benedict <michaelbt@live.com>

decon_reg: silence literal conversion warning

Signed-off-by: Michael Benedict <michaelbt@live.com>

selinux: silence unused variable warning

Signed-off-by: Michael Benedict <michaelbt@live.com>

input: silence strlcpy-strlcat-size warning

Signed-off-by: Michael Benedict <michaelbt@live.com>

wireless: fix all kind of warnings

Signed-off-by: Michael Benedict <michaelbt@live.com>

mfc: silence enum conversion warning

Signed-off-by: Michael Benedict <michaelbt@live.com>

PCI: Skip MPS logic for Virtual Functions (VFs)

commit 3dbe97efe8bf450b183d6dee2305cbc032e6b8a4 upstream.

PCIe r4.0, sec 9.3.5.4, "Device Control Register", shows both
Max_Payload_Size (MPS) and Max_Read_request_Size (MRRS) to be 'RsvdP' for
VFs.  Just prior to the table it states:

  "PF and VF functionality is defined in Section 7.5.3.4 except where
   noted in Table 9-16.  For VF fields marked 'RsvdP', the PF setting
   applies to the VF."

All of which implies that with respect to Max_Payload_Size Supported
(MPSS), MPS, and MRRS values, we should not be paying any attention to the
VF's fields, but rather only to the PF's.  Only looking at the PF's fields
also logically makes sense as it's the sole physical interface to the PCIe
bus.

Link: https://bugzilla.kernel.org/show_bug.cgi?id=200527
Fixes: 27d868b5e6cf ("PCI: Set MPS to match upstream bridge")
Signed-off-by: Myron Stowe <myron.stowe@redhat.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Cc: stable@vger.kernel.org # 4.3+
Cc: Keith Busch <keith.busch@intel.com>
Cc: Sinan Kaya <okaya@kernel.org>
Cc: Dongdong Liu <liudongdong3@huawei.com>
Cc: Jon Mason <jdmason@kudzu.us>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

PCI: Make PCI_ROM_ADDRESS_MASK a 32-bit constant

commit 76dc52684d0f72971d9f6cc7d5ae198061b715bd upstream.

A 64-bit value is not needed since a PCI ROM address consists in 32 bits.
This fixes a clang warning about "implicit conversion from 'unsigned long'
to 'u32'".

Also remove now unnecessary casts to u32 from __pci_read_base() and
pci_std_update_resource().

Signed-off-by: Matthias Kaehlcke <mka@chromium.org>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Cc: Nathan Chancellor <natechancellor@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ANDROID: xt_qtaguid: Remove unnecessary null checks to device's name

'name' will never be NULL since it isn't a plain pointer but an array
of char values.

../net/netfilter/xt_qtaguid.c:1195:27: warning: address of array
'(*el_dev)->name' will always evaluate to 'true'
[-Wpointer-bool-conversion]
        if (unlikely(!(*el_dev)->name)) {
                     ~~~~~~~~~~~~^~~~

Change-Id: If3b25f17829b43e8a639193fb9cd04ae45947200
Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>

ANDROID: xt_qtaguid: Remove unnecessary null checks to ifa_label

'ifa_label' will never be NULL since it isn't a plain pointer but an
array of char values.

../net/netfilter/xt_qtaguid.c:971:11: warning: address of array
'ifa->ifa_label' will always evaluate to 'true'
[-Wpointer-bool-conversion]
                        ifa->ifa_label ? ifa->ifa_label : "(null)");
                        ~~~~~^~~~~~~~~ ~
../net/netfilter/xt_qtaguid.c:972:13: warning: address of array
'ifa->ifa_label' will always evaluate to 'true'
[-Wpointer-bool-conversion]
                if (ifa->ifa_label && !strcmp(ifname, ifa->ifa_label))
                    ~~~~~^~~~~~~~~ ~~

Change-Id: I3c87a5d4b830aaa21a59e9c39cfe0a1d60d7f830
Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>

dm ioctl: remove double parentheses

commit e36215d87f301f9567c8c99fd34e6c3521a94ddf upstream.

The extra pair of parantheses is not needed and causes clang to generate
warnings about the DM_DEV_CREATE_CMD comparison in validate_params().

Also remove another double parentheses that doesn't cause a warning.

Signed-off-by: Matthias Kaehlcke <mka@chromium.org>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Cc: Nathan Chancellor <natechancellor@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Input: mousedev - fix implicit conversion warning

commit dae1a432ab1fe79ae53129ededeaece35a2dc14d upstream.

Clang warns:

drivers/input/mousedev.c:653:63: error: implicit conversion from 'int'
to 'signed char' changes value from 200 to -56
[-Wconstant-conversion]
  client->ps2[1] = 0x60; client->ps2[2] = 3; client->ps2[3] = 200;
                                                            ~ ^~~
As the PS2 data is really a stream of bytes, let's switch to using u8 type
for it, which silences this warning.

Signed-off-by: Nick Desaulniers <nick.desaulniers@gmail.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Cc: Nathan Chancellor <natechancellor@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

netfilter: nf_nat_h323: fix logical-not-parentheses warning

commit eee6ebbac18a189ef33d25ea9b8bcae176515e49 upstream.

Clang produces the following warning:

net/ipv4/netfilter/nf_nat_h323.c:553:6: error:
logical not is only applied to the left hand side of this comparison
  [-Werror,-Wlogical-not-parentheses]
if (!set_h225_addr(skb, protoff, data, dataoff, taddr,
    ^
add parentheses after the '!' to evaluate the comparison first
add parentheses around left hand side expression to silence this warning

There's not necessarily a bug here, but it's cleaner to return early,
ex:

if (x)
  return
...

rather than:

if (x == 0)
  ...
else
  return

Also added a return code check that seemed to be missing in one
instance.

Signed-off-by: Nick Desaulniers <ndesaulniers@google.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Cc: Nathan Chancellor <natechancellor@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

phonet: fix building with clang

[ Upstream commit 6321aa197547da397753757bd84c6ce64b3e3d89 ]

clang warns about overflowing the data[] member in the struct pnpipehdr:

net/phonet/pep.c:295:8: warning: array index 4 is past the end of the array (which contains 1 element) [-Warray-bounds]
                        if (hdr->data[4] == PEP_IND_READY)
                            ^         ~
include/net/phonet/pep.h:66:3: note: array 'data' declared here
                u8              data[1];

Using a flexible array member at the end of the struct avoids the
warning, but since we cannot have a flexible array member inside
of the union, each index now has to be moved back by one, which
makes it a little uglier.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: Rémi Denis-Courmont <remi@remlab.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

fs: compat: Remove warning from COMPATIBLE_IOCTL

commit 9280cdd6fe5b8287a726d24cc1d558b96c8491d7 upstream.

cmd in COMPATIBLE_IOCTL is always a u32, so cast it so there isn't a
warning about an overflow in XFORM.

From: Mark Charlebois <charlebm@gmail.com>
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
Signed-off-by: Behan Webster <behanw@converseincode.com>
Signed-off-by: Matthias Kaehlcke <mka@chromium.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Cc: Nathan Chancellor <natechancellor@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

netfilter: ctnetlink: fix incorrect nf_ct_put during hash resize

[ Upstream commit fefa92679dbe0c613e62b6c27235dcfbe9640ad1 ]

If nf_conntrack_htable_size was adjusted by the user during the ct
dump operation, we may invoke nf_ct_put twice for the same ct, i.e.
the "last" ct. This will cause the ct will be freed but still linked
in hash buckets.

It's very easy to reproduce the problem by the following commands:
  # while : ; do
  echo $RANDOM > /proc/sys/net/netfilter/nf_conntrack_buckets
  done
  # while : ; do
  conntrack -L
  done
  # iperf -s 127.0.0.1 &
  # iperf -c 127.0.0.1 -P 60 -t 36000

After a while, the system will hang like this:
  NMI watchdog: BUG: soft lockup - CPU#1 stuck for 22s! [bash:20184]
  NMI watchdog: BUG: soft lockup - CPU#0 stuck for 22s! [iperf:20382]
  ...

So at last if we find cb->args[1] is equal to "last", this means hash
resize happened, then we can set cb->args[1] to 0 to fix the above
issue.

Fixes: d205dc40798d ("[NETFILTER]: ctnetlink: fix deadlock in table dumping")
Signed-off-by: Liping Zhang <zlpnobody@gmail.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

netfilter: ctnetlink: Make some parameters integer to avoid enum mismatch

commit a2b7cbdd2559aff06cebc28a7150f81c307a90d3 upstream.

Not all parameters passed to ctnetlink_parse_tuple() and
ctnetlink_exp_dump_tuple() match the enum type in the signatures of these
functions. Since this is intended change the argument type of to be an
unsigned integer value.

Signed-off-by: Matthias Kaehlcke <mka@chromium.org>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
[natechancellor: ctnetlink_exp_dump_tuple is still inline]
Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

audit: return on memory error to avoid null pointer dereference

[ Upstream commit 23138ead270045f1b3e912e667967b6094244999 ]

If there is a memory allocation error when trying to change an audit
kernel feature value, the ignored allocation error will trigger a NULL
pointer dereference oops on subsequent use of that pointer.  Return
instead.

Passes audit-testsuite.
See: https://github.com/linux-audit/audit-kernel/issues/76

Signed-off-by: Richard Guy Briggs <rgb@redhat.com>
[PM: not necessary (other funcs check for NULL), but a good practice]
Signed-off-by: Paul Moore <paul@paul-moore.com>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

audit: add tty field to LOGIN event

commit db0a6fb5d97afe01fd9c47d37c6daa82d4d4001d upstream.

The tty field was missing from AUDIT_LOGIN events.

Refactor code to create a new function audit_get_tty(), using it to
replace the call in audit_log_task_info() and to add it to
audit_log_set_loginuid().  Lock and bump the kref to protect it, adding
audit_put_tty() alias to decrement it.

Signed-off-by: Richard Guy Briggs <rgb@redhat.com>
Signed-off-by: Paul Moore <paul@paul-moore.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

netfilter: nf_ct_expect: remove the redundant slash when policy name is empty

commit b173a28f62cf929324a8a6adcc45adadce311d16 upstream.

The 'name' filed in struct nf_conntrack_expect_policy{} is not a
pointer, so check it is NULL or not will always return true. Even if the
name is empty, slash will always be displayed like follows:
  # cat /proc/net/nf_conntrack_expect
  297 l3proto = 2 proto=6 src=1.1.1.1 dst=2.2.2.2 sport=1 dport=1025 ftp/
                                                                        ^

Fixes: 3a8fc53a45c4 ("netfilter: nf_ct_helper: allocate 16 bytes for the helper and policy names")
Signed-off-by: Liping Zhang <liping.zhang@spreadtrum.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Acked-by: Michal Kubecek <mkubecek@suse.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

UPSTREAM: tracing: Use cpumask_available() to check if cpumask variable may be used

This fixes the following clang warning:

kernel/trace/trace.c:3231:12: warning: address of array 'iter->started'
  will always evaluate to 'true' [-Wpointer-bool-conversion]
        if (iter->started)

Link: http://lkml.kernel.org/r/20170421234110.117075-1-mka@chromium.org
Signed-off-by: Matthias Kaehlcke <mka@chromium.org>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
(cherry picked from commit 4dbbe2d8e95c351157f292ece067f985c30c7b53)
Bug: 78886293
Change-Id: Ib17a68ce55ca80b04bdea2d232f8ca9f88b1b8a3
Signed-off-by: Alistair Strachan <astrachan@google.com>

battery_v2: silence enum conversion warning

Signed-off-by: Michael Benedict <michaelbt@live.com>

frv: declare jiffies to be located in the .data section

commit 60b0a8c3d2480f3b57282b47b7cae7ee71c48635 upstream.

Commit 7c30f352c852 ("jiffies.h: declare jiffies and jiffies_64 with
____cacheline_aligned_in_smp") removed a section specification from the
jiffies declaration that caused conflicts on some platforms.

Unfortunately this change broke the build for frv:

  kernel/built-in.o: In function `__do_softirq': (.text+0x6460): relocation truncated to fit: R_FRV_GPREL12 against symbol
      `jiffies' defined in *ABS* section in .tmp_vmlinux1
  kernel/built-in.o: In function `__do_softirq': (.text+0x6574): relocation truncated to fit: R_FRV_GPREL12 against symbol
      `jiffies' defined in *ABS* section in .tmp_vmlinux1
  kernel/built-in.o: In function `pwq_activate_delayed_work': workqueue.c:(.text+0x15b9c): relocation truncated to fit: R_FRV_GPREL12 against
      symbol `jiffies' defined in *ABS* section in .tmp_vmlinux1
  ...

Add __jiffy_arch_data to the declaration of jiffies and use it on frv to
include the section specification.  For all other platforms
__jiffy_arch_data (currently) has no effect.

Fixes: 7c30f352c852 ("jiffies.h: declare jiffies and jiffies_64 with ____cacheline_aligned_in_smp")
Link: http://lkml.kernel.org/r/20170516221333.177280-1-mka@chromium.org
Signed-off-by: Matthias Kaehlcke <mka@chromium.org>
Reported-by: Guenter Roeck <linux@roeck-us.net>
Tested-by: Guenter Roeck <linux@roeck-us.net>
Reviewed-by: David Howells <dhowells@redhat.com>
Cc: Sudip Mukherjee <sudipm.mukherjee@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

jiffies.h: declare jiffies and jiffies_64 with ____cacheline_aligned_in_smp

commit 7c30f352c852bae2715ad65ac4a38ca9af7d7696 upstream.

jiffies_64 is defined in kernel/time/timer.c with
____cacheline_aligned_in_smp, however this macro is not part of the
declaration of jiffies and jiffies_64 in jiffies.h.

As a result clang generates the following warning:

  kernel/time/timer.c:57:26: error: section does not match previous declaration [-Werror,-Wsection]
  __visible u64 jiffies_64 __cacheline_aligned_in_smp = INITIAL_JIFFIES;
                           ^
  include/linux/cache.h:39:36: note: expanded from macro '__cacheline_aligned_in_smp'
                                     ^
  include/linux/cache.h:34:4: note: expanded from macro '__cacheline_aligned'
                   __section__(".data..cacheline_aligned")))
                   ^
  include/linux/jiffies.h:77:12: note: previous attribute is here
  extern u64 __jiffy_data jiffies_64;
             ^
  include/linux/jiffies.h:70:38: note: expanded from macro '__jiffy_data'

Link: http://lkml.kernel.org/r/20170403190200.70273-1-mka@chromium.org
Signed-off-by: Matthias Kaehlcke <mka@chromium.org>
Cc: "Jason A . Donenfeld" <Jason@zx2c4.com>
Cc: Grant Grundler <grundler@chromium.org>
Cc: Michael Davidson <md@google.com>
Cc: Greg Hackmann <ghackmann@google.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Nathan Chancellor <natechancellor@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

cpumask: Add helper cpumask_available()

commit f7e30f01a9e221067bb4b579e3cfc25cd2617467 upstream.

With CONFIG_CPUMASK_OFFSTACK=y cpumask_var_t is a struct cpumask
pointer, otherwise a struct cpumask array with a single element.

Some code dealing with cpumasks needs to validate that a cpumask_var_t
is not a NULL pointer when CONFIG_CPUMASK_OFFSTACK=y. This is typically
done by performing the check always, regardless of the underlying type
of cpumask_var_t. This works in both cases, however clang raises a
warning like this when CONFIG_CPUMASK_OFFSTACK=n:

kernel/irq/manage.c:839:28: error: address of array
'desc->irq_common_data.affinity' will always evaluate to 'true'
[-Werror,-Wpointer-bool-conversion]

Add the inline helper cpumask_available() which only performs the
pointer check if CONFIG_CPUMASK_OFFSTACK=y.

Signed-off-by: Matthias Kaehlcke <mka@chromium.org>
Cc: Grant Grundler <grundler@chromium.org>
Cc: Rusty Russell <rusty@rustcorp.com.au>
Cc: Greg Hackmann <ghackmann@google.com>
Cc: Michael Davidson <md@google.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Link: http://lkml.kernel.org/r/20170412182030.83657-1-mka@chromium.org
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Nathan Chancellor <natechancellor@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

genirq: Use cpumask_available() for check of cpumask variable

commit d170fe7dd992b313d4851ae5ab77ee7a51ed8c72 upstream.

This fixes the following clang warning when CONFIG_CPUMASK_OFFSTACK=n:

kernel/irq/manage.c:839:28: error: address of array
'desc->irq_common_data.affinity' will always evaluate to 'true'
[-Werror,-Wpointer-bool-conversion]

Signed-off-by: Matthias Kaehlcke <mka@chromium.org>
Cc: Grant Grundler <grundler@chromium.org>
Cc: Rusty Russell <rusty@rustcorp.com.au>
Cc: Greg Hackmann <ghackmann@google.com>
Cc: Michael Davidson <md@google.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Link: http://lkml.kernel.org/r/20170412182030.83657-2-mka@chromium.org
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Nathan Chancellor <natechancellor@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

selinux: Remove unnecessary check of array base in selinux_set_mapping()

commit 342e91578eb6909529bc7095964cd44b9c057c4e upstream.

'perms' will never be NULL since it isn't a plain pointer but an array
of u32 values.

This fixes the following warning when building with clang:

security/selinux/ss/services.c:158:16: error: address of array
'p_in->perms' will always evaluate to 'true'
[-Werror,-Wpointer-bool-conversion]
                while (p_in->perms && p_in->perms[k]) {

Signed-off-by: Matthias Kaehlcke <mka@chromium.org>
Signed-off-by: Paul Moore <paul@paul-moore.com>
Cc: Nathan Chancellor <natechancellor@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

arm64: avoid overflow in VA_START and PAGE_OFFSET

commit 82cd588052815eb4146f9f7c5347ca5e32c56360 upstream.

The bitmask used to define these values produces overflow, as seen by
this compiler warning:

arch/arm64/kernel/head.S:47:8: warning:
      integer overflow in preprocessor expression
  #elif (PAGE_OFFSET & 0x1fffff) != 0
         ^~~~~~~~~~~
arch/arm64/include/asm/memory.h:52:46: note:
      expanded from macro 'PAGE_OFFSET'
  #define PAGE_OFFSET             (UL(0xffffffffffffffff) << (VA_BITS -
1))
                                      ~~~~~~~~~~~~~~~~~~  ^

It would be preferrable to use GENMASK_ULL() instead, but it's not set
up to be used from assembly (the UL() macro token pastes UL suffixes
when not included in assembly sources).

Suggested-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Suggested-by: Yury Norov <ynorov@caviumnetworks.com>
Suggested-by: Matthias Kaehlcke <mka@chromium.org>
Signed-off-by: Nick Desaulniers <ndesaulniers@google.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
[natechancellor: KIMAGE_VADDR doesn't exist]
Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

drivers: acpi: silence pointer bool conversion warning

Signed-off-by: Michael Benedict <michaelbt@live.com>

mm: silence enum conversion warnings

Signed-off-by: Michael Benedict <michaelbt@live.com>

soc: samsung: cal-if: add volatile keyword to prevent optimization by clang

Signed-off-by: Michael Benedict <michaelbt@live.com>

ion: silence logical-not-parentheses warning

Signed-off-by: Michael Benedict <michaelbt@live.com>

dpu: use usleep_range instead of udelay

Signed-off-by: Michael Benedict <michaelbt@live.com>

mptcp: disable incompatible-pointer-types warnings

Signed-off-by: Michael Benedict <michaelbt@live.com>

cal-if: disable sizeof-pointer-div warnings

Signed-off-by: Michael Benedict <michaelbt@live.com>

sec_debug: silence null-dereference warnings

Signed-off-by: Michael Benedict <michaelbt@live.com>

decon: merge all into decon_data_path

Signed-off-by: Michael Benedict <michaelbt@live.com>

drivers: usb: silence enum-conversion warnings

Signed-off-by: Michael Benedict <michaelbt@live.com>

drivers: thermal: silence enum-conversion warnings

Signed-off-by: Michael Benedict <michaelbt@live.com>

drivers: fimc-is2: silence enum-conversion warnings

Signed-off-by: Michael Benedict <michaelbt@live.com>

drivers: phy: silence enum-conversion warning

Signed-off-by: Michael Benedict <michaelbt@live.com>

s5p_mfc_enc_internal: replace min/max value to fix shift-overflow warning

Signed-off-by: Michael Benedict <michaelbt@live.com>

drivers: modem_v1: silence logical-not-parentheses warning

Signed-off-by: Michael Benedict <michaelbt@live.com>

drivers: modem_v1: silence enum-conversion warnings

Signed-off-by: Michael Benedict <michaelbt@live.com>

drivers: touchscreen: fix strncat-size warnings

S10 sec_ts as reference

Signed-off-by: Michael Benedict <michaelbt@live.com>

drivers: vfs8xxx: silence enum-conversion warning

Signed-off-by: Michael Benedict <michaelbt@live.com>

battery_v2: silence enum-conversion warnings

Signed-off-by: Michael Benedict <michaelbt@live.com>

UPSTREAM: mm/zsmalloc.c: fix -Wunneeded-internal-declaration warning

is_first_page() is only called from the macro VM_BUG_ON_PAGE() which is
only compiled in as a runtime check when CONFIG_DEBUG_VM is set,
otherwise is checked at compile time and not actually compiled in.

Fixes the following warning, found with Clang:

  mm/zsmalloc.c:472:12: warning: function 'is_first_page' is not needed and will not be emitted [-Wunneeded-internal-declaration]
  static int is_first_page(struct page *page)
           ^

Link: http://lkml.kernel.org/r/20170524053859.29059-1-nick.desaulniers@gmail.com
Signed-off-by: Nick Desaulniers <nick.desaulniers@gmail.com>
Reviewed-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
Acked-by: Minchan Kim <minchan@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
(cherry picked from commit 3457f4147675108aa83f9f33c136f06bb9f8518f)
Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>

UPSTREAM: mm/zsmalloc.c: change stat type parameter to int

zs_stat_inc/dec/get() uses enum zs_stat_type for the stat type, however
some callers pass an enum fullness_group value.  Change the type to int to
reflect the actual use of the functions and get rid of 'enum-conversion'
warnings

Link: http://lkml.kernel.org/r/20170731175000.56538-1-mka@chromium.org
Signed-off-by: Matthias Kaehlcke <mka@chromium.org>
Reviewed-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
Acked-by: Minchan Kim <minchan@kernel.org>
Cc: Doug Anderson <dianders@chromium.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
(cherry picked from commit 3eb95feac113d8ebad5b7b5189a65efcbd95a749)
Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>

BACKPORT: sched/sysctl: Fix attributes of some extern declarations

The definition of sysctl_sched_migration_cost, sysctl_sched_nr_migrate
and sysctl_sched_time_avg includes the attribute const_debug. This
attribute is not part of the extern declaration of these variables in
include/linux/sched/sysctl.h, while it is in kernel/sched/sched.h,
and as a result Clang generates warnings like this:

  kernel/sched/sched.h:1618:33: warning: section attribute is specified on redeclared variable [-Wsection]
  extern const_debug unsigned int sysctl_sched_time_avg;
                                ^
  ./include/linux/sched/sysctl.h:42:21: note: previous declaration is here
  extern unsigned int sysctl_sched_time_avg;

The header only declares the variables when CONFIG_SCHED_DEBUG is defined,
therefore it is not necessary to duplicate the definition of const_debug.
Instead we can use the attribute __read_mostly, which is the expansion of
const_debug when CONFIG_SCHED_DEBUG=y is set.

Signed-off-by: Matthias Kaehlcke <mka@chromium.org>
Reviewed-by: Nick Desaulniers <nick.desaulniers@gmail.com>
Cc: Douglas Anderson <dianders@chromium.org>
Cc: Guenter Roeck <groeck@chromium.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Shile Zhang <shile.zhang@nokia.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/20171030180816.170850-1-mka@chromium.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
(cherry picked from commit a9903f04e0a4ea522d959c2f287cdf0ab029e324)
Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>
Signed-off-by: kdrag0n <dragon@khronodragon.com>

decon_core: fix parentheses-equality warning

Signed-off-by: Yaroslav Furman <yaro330@gmail.com>

ufs-exynos: don't use boolean statement in switch-case

Convert it to int because it's not really a boolean.
Signed-off-by: Yaroslav Furman <yaro330@gmail.com>

vs4l: remove __u[32/16/8] and __s32 declarations

They are already defined. Fixes redefinition warnings.
Signed-off-by: Yaroslav Furman <yaro330@gmail.com>

s5p_mfc_[dec/enc]_ops: fix parentheses-equality warnings

Signed-off-by: Yaroslav Furman <yaro330@gmail.com>

ufshcd: fix parentheses-equality warning

Signed-off-by: Yaroslav Furman <yaro330@gmail.com>

arm64: include: asm: initialize a few ret(-s)

Fixes -Wsometimes-uninitialized that just
fill up build log.
Signed-off-by: Yaroslav Furman <yaro330@gmail.com>

drv2624: fix shift overflow warning

Signed-off-by: Yaroslav Furman <yaro330@gmail.com>

[COMMON] fimc-is2: fix a typo (SYNNY -> SUNNY)

Change-Id: Ieeb22b673ae6d04e41fdb66d042fff5c70491dac
Signed-off-by: Jeongtae Park <jtp.park@samsung.com>

sec_ts_fn: fix logical-not-parentheses warning

Signed-off-by: Yaroslav Furman <yaro330@gmail.com>

[cpu/gpu]_cooling: move get_static_t declaration in thermal.h

To avoid redefenition warning.
Signed-off-by: Yaroslav Furman <yaro330@gmail.com>

fs: ecryptfs: fix parentheses-equality warning

Signed-off-by: Yaroslav Furman <yaro330@gmail.com>

kbuild: clang: disable unused variable warnings only when constant

commit 0a5f41767444cc3b4fc5573921ab914b4f78baaa upstream.

Currently, GCC disables -Wunused-const-variable, but not
-Wunused-variable, so warns unused variables if they are
non-constant.

While, Clang does not warn unused variables at all regardless of
the const qualifier because -Wno-unused-const-variable is implied
by the stronger option -Wno-unused-variable.

Disable -Wunused-const-variable instead of -Wunused-variable so that
GCC and Clang work in the same way.

Signed-off-by: Prasad Sodagudi <psodagud@codeaurora.org>
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

kbuild: clang: remove crufty HOSTCFLAGS

commit df16aaac26e92e97ab7234d3f93c953466adc4b5 upstream.

When compiling with `make CC=clang HOSTCC=clang`, I was seeing warnings
that clang did not recognize -fno-delete-null-pointer-checks for HOSTCC
targets.  These were added in commit 61163efae020 ("kbuild: LLVMLinux:
Add Kbuild support for building kernel with Clang").

Clang does not support -fno-delete-null-pointer-checks, so adding it to
HOSTCFLAGS if HOSTCC is clang does not make sense.

It's not clear why the other warnings were disabled, and just for
HOSTCFLAGS, but I can remove them, add -Werror to HOSTCFLAGS and compile
with clang just fine.

Suggested-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Signed-off-by: Nick Desaulniers <nick.desaulniers@gmail.com>
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
[nc: Adjust context]
Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

source: N950F DSH1

Signed-off-by: Michael Benedict <michaelbt@live.com>

source: G95xF DSH5

Signed-off-by: Michael Benedict <michaelbt@live.com>

BACKPORT: zsmalloc: introduce zs_huge_class_size()

Patch series "zsmalloc/zram: drop zram's max_zpage_size", v3.

ZRAM's max_zpage_size is a bad thing.  It forces zsmalloc to store
normal objects as huge ones, which results in bigger zsmalloc memory
usage.  Drop it and use actual zsmalloc huge-class value when decide if
the object is huge or not.

This patch (of 2):

Not every object can be share its zspage with other objects, e.g.  when
the object is as big as zspage or nearly as big a zspage.  For such
objects zsmalloc has a so called huge class - every object which belongs
to huge class consumes the entire zspage (which consists of a physical
page).  On x86_64, PAGE_SHIFT 12 box, the first non-huge class size is
3264, so starting down from size 3264, objects can share page(-s) and
thus minimize memory wastage.

ZRAM, however, has its own statically defined watermark for huge
objects, namely "3 * PAGE_SIZE / 4 = 3072", and forcibly stores every
object larger than this watermark (3072) as a PAGE_SIZE object, in other
words, to a huge class, while zsmalloc can keep some of those objects in
non-huge classes.  This results in increased memory consumption.

zsmalloc knows better if the object is huge or not.  Introduce
zs_huge_class_size() function which tells if the given object can be
stored in one of non-huge classes or not.  This will let us to drop
ZRAM's huge object watermark and fully rely on zsmalloc when we decide
if the object is huge.

[sergey.senozhatsky.work@gmail.com: add pool param to zs_huge_class_size()]
Link: http://lkml.kernel.org/r/20180314081833.1096-2-sergey.senozhatsky@gmail.com
Link: http://lkml.kernel.org/r/20180306070639.7389-2-sergey.senozhatsky@gmail.com
Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
Acked-by: Minchan Kim <minchan@kernel.org>
Cc: Mike Rapoport <rppt@linux.vnet.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
(cherry picked from commit 010b495e2fa32353d0ef6aa70a8169e5ef617a15)
Signed-off-by: Peter Kalauskas <peskal@google.com>
Bug: 113183619
Change-Id: Ic35f8c1ec75f0b78bf2d83729b6aedd2999f25c8

zram: close udev startup race condition as default groups

commit fef912bf860e upstream.
commit 98af4d4df889 upstream.

I got a report from Howard Chen that he saw zram and sysfs race(ie,
zram block device file is created but sysfs for it isn't yet)
when he tried to create new zram devices via hotadd knob.

v4.20 kernel fixes it by [1, 2] but it's too large size to merge
into -stable so this patch fixes the problem by registering defualt
group by Greg KH's approach[3].

This patch should be applied to every stable tree [3.16+] currently
existing from kernel.org because the problem was introduced at 2.6.37
by [4].

[1] fef912bf860e, block: genhd: add 'groups' argument to device_add_disk
[2] 98af4d4df889, zram: register default groups with device_add_disk()
[3] http://kroah.com/log/blog/2013/06/26/how-to-create-a-sysfs-file-correctly/
[4] 33863c21e69e9, Staging: zram: Replace ioctls with sysfs interface

Cc: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
Cc: Hannes Reinecke <hare@suse.com>
Tested-by: Howard Chen <howardsoc@google.com>
Signed-off-by: Minchan Kim <minchan@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ANDROID: zram: set comp_len to PAGE_SIZE when page is huge

This bug was introduced when two patches were applied out of order.

* zram: drop max_zpage_size and use zs_huge_class_size()
* zram: mark incompressible page as ZRAM_HUGE

Signed-off-by: Peter Kalauskas <peskal@google.com>
Bug: 119260394
Change-Id: I437d35c8d23c15237ad9c2d5bd7f99d7bff42872

BACKPORT: zram: drop max_zpage_size and use zs_huge_class_size()

Remove ZRAM's enforced "huge object" value and use zsmalloc huge-class
watermark instead, which makes more sense.

TEST
- I used a 1G zram device, LZO compression back-end, original
  data set size was 444MB. Looking at zsmalloc classes stats the
  test ended up to be pretty fair.

BASE ZRAM/ZSMALLOC
=====================
zram mm_stat

498978816 191482495 199831552        0 199831552    15634        0

zsmalloc classes

 class  size almost_full almost_empty obj_allocated   obj_used pages_used pages_per_zspage freeable
...
   151  2448           0            0          1240       1240        744                3        0
   168  2720           0            0          4200       4200       2800                2        0
   190  3072           0            0         10100      10100       7575                3        0
   202  3264           0            0           380        380        304                4        0
   254  4096           0            0         10620      10620      10620                1        0

 Total                 7           46        106982     106187      48787                         0

PATCHED ZRAM/ZSMALLOC
=====================

zram mm_stat

498978816 182579184 194248704        0 194248704    15628        0

zsmalloc classes

 class  size almost_full almost_empty obj_allocated   obj_used pages_used pages_per_zspage freeable
...
   151  2448           0            0          1240       1240        744                3        0
   168  2720           0            0          4200       4200       2800                2        0
   190  3072           0            0         10100      10100       7575                3        0
   202  3264           0            0          7180       7180       5744                4        0
   254  4096           0            0          3820       3820       3820                1        0

 Total                 8           45        106959     106193      47424                         0

As we can see, we reduced the number of objects stored in class-4096,
because a huge number of objects which we previously forcibly stored in
class-4096 now stored in non-huge class-3264.  This results in lower
memory consumption:

- zsmalloc now uses 47424 physical pages, which is less than 48787 pages
  zsmalloc used before.

- objects that we store in class-3264 share zspages.  That's why overall
  the number of pages that both class-4096 and class-3264 consumed went
  down from 10924 to 9564.

[sergey.senozhatsky.work@gmail.com: add pool param to zs_huge_class_size()]
Link: http://lkml.kernel.org/r/20180314081833.1096-3-sergey.senozhatsky@gmail.com
Link: http://lkml.kernel.org/r/20180306070639.7389-3-sergey.senozhatsky@gmail.com
Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
Acked-by: Minchan Kim <minchan@kernel.org>
Cc: Mike Rapoport <rppt@linux.vnet.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
(cherry picked from commit 60f5921a9a4f126e081318bd6bb2bc2798b7bba8)
Signed-off-by: Peter Kalauskas <peskal@google.com>
Bug: 113183619
Change-Id: I1d3ede25543e99a24802ad03f68995f33aaf79b5