ubsan: include bug type in report header

When syzbot tries to figure out how to deduplicate bug reports, it prefers
seeing a hint about a specific bug type (we can do better than just
"UBSAN").  This lifts the handler reason into the UBSAN report line that
includes the file path that tripped a check.  Unfortunately, UBSAN does
not provide function names.

Suggested-by: Dmitry Vyukov <dvyukov@google.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Alexander Potapenko <glider@google.com>
Cc: Andrey Konovalov <andreyknvl@google.com>
Cc: Andrey Ryabinin <aryabinin@virtuozzo.com>
Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Dan Carpenter <dan.carpenter@oracle.com>
Cc: Elena Petrova <lenaptr@google.com>
Cc: "Gustavo A. R. Silva" <gustavo@embeddedor.com>
Link: http://lkml.kernel.org/r/20200227193516.32566-7-keescook@chromium.org
Link: https://lore.kernel.org/lkml/CACT4Y+bsLJ-wFx_TaXqax3JByUOWB3uk787LsyMVcfW6JzzGvg@mail.gmail.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

kasan: unset panic_on_warn before calling panic()

As done in the full WARN() handler, panic_on_warn needs to be cleared
before calling panic() to avoid recursive panics.

Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Dmitry Vyukov <dvyukov@google.com>
Cc: Alexander Potapenko <glider@google.com>
Cc: Andrey Konovalov <andreyknvl@google.com>
Cc: Andrey Ryabinin <aryabinin@virtuozzo.com>
Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Dan Carpenter <dan.carpenter@oracle.com>
Cc: Elena Petrova <lenaptr@google.com>
Cc: "Gustavo A. R. Silva" <gustavo@embeddedor.com>
Link: http://lkml.kernel.org/r/20200227193516.32566-6-keescook@chromium.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

ubsan: check panic_on_warn

Syzkaller expects kernel warnings to panic when the panic_on_warn sysctl
is set.  More work is needed here to have UBSan reuse the WARN
infrastructure, but for now, just check the flag manually.

Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Alexander Potapenko <glider@google.com>
Cc: Andrey Konovalov <andreyknvl@google.com>
Cc: Andrey Ryabinin <aryabinin@virtuozzo.com>
Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Dan Carpenter <dan.carpenter@oracle.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Elena Petrova <lenaptr@google.com>
Cc: "Gustavo A. R. Silva" <gustavo@embeddedor.com>
Link: https://lore.kernel.org/lkml/CACT4Y+bsLJ-wFx_TaXqax3JByUOWB3uk787LsyMVcfW6JzzGvg@mail.gmail.com
Link: http://lkml.kernel.org/r/20200227193516.32566-5-keescook@chromium.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

drivers/misc/lkdtm/bugs.c: add arithmetic overflow and array bounds checks

Adds LKDTM tests for arithmetic overflow (both signed and unsigned), as
well as array bounds checking.

Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Dmitry Vyukov <dvyukov@google.com>
Cc: Alexander Potapenko <glider@google.com>
Cc: Andrey Konovalov <andreyknvl@google.com>
Cc: Andrey Ryabinin <aryabinin@virtuozzo.com>
Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Dan Carpenter <dan.carpenter@oracle.com>
Cc: Elena Petrova <lenaptr@google.com>
Cc: "Gustavo A. R. Silva" <gustavo@embeddedor.com>
Link: http://lkml.kernel.org/r/20200227193516.32566-4-keescook@chromium.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

ubsan: split "bounds" checker from other options

In order to do kernel builds with the bounds checker individually
available, introduce CONFIG_UBSAN_BOUNDS, with the remaining options under
CONFIG_UBSAN_MISC.

For example, using this, we can start to expand the coverage syzkaller is
providing.  Right now, all of UBSan is disabled for syzbot builds because
taken as a whole, it is too noisy.  This will let us focus on one feature
at a time.

For the bounds checker specifically, this provides a mechanism to
eliminate an entire class of array overflows with close to zero
performance overhead (I cannot measure a difference).  In my (mostly)
defconfig, enabling bounds checking adds ~4200 checks to the kernel.
Performance changes are in the noise, likely due to the branch predictors
optimizing for the non-fail path.

Some notes on the bounds checker:

- it does not instrument {mem,str}*()-family functions, it only
  instruments direct indexed accesses (e.g. "foo[i]"). Dealing with
  the {mem,str}*()-family functions is a work-in-progress around
  CONFIG_FORTIFY_SOURCE[1].

- it ignores flexible array members, including the very old single
  byte (e.g. "int foo[1];") declarations. (Note that GCC's
  implementation appears to ignore _all_ trailing arrays, but Clang only
  ignores empty, 0, and 1 byte arrays[2].)

[1] https://github.com/KSPP/linux/issues/6
[2] https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92589

Suggested-by: Elena Petrova <lenaptr@google.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Andrey Ryabinin <aryabinin@virtuozzo.com>
Acked-by: Dmitry Vyukov <dvyukov@google.com>
Cc: Alexander Potapenko <glider@google.com>
Cc: Andrey Konovalov <andreyknvl@google.com>
Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Dan Carpenter <dan.carpenter@oracle.com>
Cc: "Gustavo A. R. Silva" <gustavo@embeddedor.com>
Link: http://lkml.kernel.org/r/20200227193516.32566-3-keescook@chromium.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

ubsan: add trap instrumentation option

Patch series "ubsan: Split out bounds checker", v5.

This splits out the bounds checker so it can be individually used.  This
is enabled in Android and hopefully for syzbot.  Includes LKDTM tests for
behavioral corner-cases (beyond just the bounds checker), and adjusts
ubsan and kasan slightly for correct panic handling.

This patch (of 6):

The Undefined Behavior Sanitizer can operate in two modes: warning
reporting mode via lib/ubsan.c handler calls, or trap mode, which uses
__builtin_trap() as the handler.  Using lib/ubsan.c means the kernel image
is about 5% larger (due to all the debugging text and reporting structures
to capture details about the warning conditions).  Using the trap mode,
the image size changes are much smaller, though at the loss of the
"warning only" mode.

In order to give greater flexibility to system builders that want minimal
changes to image size and are prepared to deal with kernel code being
aborted and potentially destabilizing the system, this introduces
CONFIG_UBSAN_TRAP.  The resulting image sizes comparison:

   text    data     bss       dec       hex     filename
19533663   6183037  18554956  44271656  2a38828 vmlinux.stock
19991849   7618513  18874448  46484810  2c54d4a vmlinux.ubsan
19712181   6284181  18366540  44362902  2a4ec96 vmlinux.ubsan-trap

CONFIG_UBSAN=y:      image +4.8% (text +2.3%, data +18.9%)
CONFIG_UBSAN_TRAP=y: image +0.2% (text +0.9%, data +1.6%)

Additionally adjusts the CONFIG_UBSAN Kconfig help for clarity and removes
the mention of non-existing boot param "ubsan_handle".

Suggested-by: Elena Petrova <lenaptr@google.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Dmitry Vyukov <dvyukov@google.com>
Cc: Andrey Ryabinin <aryabinin@virtuozzo.com>
Cc: Andrey Konovalov <andreyknvl@google.com>
Cc: Alexander Potapenko <glider@google.com>
Cc: Dan Carpenter <dan.carpenter@oracle.com>
Cc: "Gustavo A. R. Silva" <gustavo@embeddedor.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Link: http://lkml.kernel.org/r/20200227193516.32566-2-keescook@chromium.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

init/Kconfig: clean up ANON_INODES and old IO schedulers options

CONFIG_ANON_INODES is gone since commit 5dd50aaeb185 ("Make anon_inodes
unconditional").

CONFIG_CFQ_GROUP_IOSCHED was replaced with CONFIG_BFQ_GROUP_IOSCHED in
commit f382fb0bcef4 ("block: remove legacy IO schedulers").

Signed-off-by: Krzysztof Kozlowski <krzk@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Masahiro Yamada <yamada.masahiro@socionext.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Link: http://lkml.kernel.org/r/20200130192419.3026-1-krzk@kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

kernel/gcov/fs.c: replace zero-length array with flexible-array member

The current codebase makes use of the zero-length array language extension
to the C90 standard, but the preferred mechanism to declare
variable-length types such as these ones is a flexible array member[1][2],
introduced in C99:

struct foo {
        int stuff;
        struct boo array[];
};

By making use of the mechanism above, we will get a compiler warning in
case the flexible array does not occur last in the structure, which will
help us prevent some kind of undefined behavior bugs from being
inadvertently introduced[3] to the codebase from now on.

Also, notice that, dynamic memory allocations won't be affected by this
change:

"Flexible array members have incomplete type, and so the sizeof operator
may not be applied.  As a quirk of the original implementation of
zero-length arrays, sizeof evaluates to zero."[1]

This issue was found with the help of Coccinelle.

[1] https://gcc.gnu.org/onlinedocs/gcc/Zero-Length.html
[2] https://github.com/KSPP/linux/issues/21
[3] commit 76497732932f ("cxgb3/l2t: Fix undefined behaviour")

Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Peter Oberparleiter <oberpar@linux.ibm.com>
Link: http://lkml.kernel.org/r/20200302224851.GA26467@embeddedor
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

gcov: gcc_3_4: replace zero-length array with flexible-array member

The current codebase makes use of the zero-length array language extension
to the C90 standard, but the preferred mechanism to declare
variable-length types such as these ones is a flexible array member[1][2],
introduced in C99:

struct foo {
        int stuff;
        struct boo array[];
};

By making use of the mechanism above, we will get a compiler warning in
case the flexible array does not occur last in the structure, which will
help us prevent some kind of undefined behavior bugs from being
inadvertently introduced[3] to the codebase from now on.

Also, notice that, dynamic memory allocations won't be affected by this
change:

"Flexible array members have incomplete type, and so the sizeof operator
may not be applied.  As a quirk of the original implementation of
zero-length arrays, sizeof evaluates to zero."[1]

This issue was found with the help of Coccinelle.

[1] https://gcc.gnu.org/onlinedocs/gcc/Zero-Length.html
[2] https://github.com/KSPP/linux/issues/21
[3] commit 76497732932f ("cxgb3/l2t: Fix undefined behaviour")

Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Peter Oberparleiter <oberpar@linux.ibm.com>
Link: http://lkml.kernel.org/r/20200302224501.GA14175@embeddedor
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

gcov: gcc_4_7: replace zero-length array with flexible-array member

The current codebase makes use of the zero-length array language extension
to the C90 standard, but the preferred mechanism to declare
variable-length types such as these ones is a flexible array member[1][2],
introduced in C99:

struct foo {
        int stuff;
        struct boo array[];
};

By making use of the mechanism above, we will get a compiler warning in
case the flexible array does not occur last in the structure, which will
help us prevent some kind of undefined behavior bugs from being
inadvertently introduced[3] to the codebase from now on.

Also, notice that, dynamic memory allocations won't be affected by this
change:

"Flexible array members have incomplete type, and so the sizeof operator
may not be applied.  As a quirk of the original implementation of
zero-length arrays, sizeof evaluates to zero."[1]

This issue was found with the help of Coccinelle.

[1] https://gcc.gnu.org/onlinedocs/gcc/Zero-Length.html
[2] https://github.com/KSPP/linux/issues/21
[3] commit 76497732932f ("cxgb3/l2t: Fix undefined behaviour")

Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Peter Oberparleiter <oberpar@linux.ibm.com>
Link: http://lkml.kernel.org/r/20200213152241.GA877@embeddedor
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

kernel/kmod.c: fix a typo "assuems" -> "assumes"

There is a typo in comment.  Fix it.  s/assuems/assumes/

Signed-off-by: Qiujun Huang <hqjagain@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Luis Chamberlain <mcgrof@kernel.org>
Link: http://lkml.kernel.org/r/1585891029-6450-1-git-send-email-hqjagain@gmail.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

reiserfs: clean up several indentation issues

There are several places where code is indented incorrectly. Fix these.

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Link: http://lkml.kernel.org/r/20200325135018.113431-1-colin.king@canonical.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

kallsyms: unexport kallsyms_lookup_name() and kallsyms_on_each_symbol()

kallsyms_lookup_name() and kallsyms_on_each_symbol() are exported to
modules despite having no in-tree users and being wide open to abuse by
out-of-tree modules that can use them as a method to invoke arbitrary
non-exported kernel functions.

Unexport kallsyms_lookup_name() and kallsyms_on_each_symbol().

Signed-off-by: Will Deacon <will@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Masami Hiramatsu <mhiramat@kernel.org>
Reviewed-by: Quentin Perret <qperret@google.com>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Frederic Weisbecker <frederic@kernel.org>
Cc: K.Prasad <prasad@linux.vnet.ibm.com>
Cc: Miroslav Benes <mbenes@suse.cz>
Cc: Petr Mladek <pmladek@suse.com>
Cc: Joe Lawrence <joe.lawrence@redhat.com>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Link: http://lkml.kernel.org/r/20200221114404.14641-4-will@kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

samples/hw_breakpoint: drop use of kallsyms_lookup_name()

The 'data_breakpoint' test code is the only modular user of
kallsyms_lookup_name(), which was exported as part of fixing the test in
f60d24d2ad04 ("hw-breakpoints: Fix broken hw-breakpoint sample module").

In preparation for un-exporting this symbol, switch the test over to using
__symbol_get(), which can be used to place breakpoints on exported
symbols.

Signed-off-by: Will Deacon <will@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Masami Hiramatsu <mhiramat@kernel.org>
Reviewed-by: Quentin Perret <qperret@google.com>
Cc: K.Prasad <prasad@linux.vnet.ibm.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Frederic Weisbecker <frederic@kernel.org>
Cc: Alexei Starovoitov <ast@kernel.org>
Cc: Miroslav Benes <mbenes@suse.cz>
Cc: Petr Mladek <pmladek@suse.com>
Cc: Joe Lawrence <joe.lawrence@redhat.com>
Link: http://lkml.kernel.org/r/20200221114404.14641-3-will@kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

samples/hw_breakpoint: drop HW_BREAKPOINT_R when reporting writes

Patch series "Unexport kallsyms_lookup_name() and kallsyms_on_each_symbol()".

Despite having just a single modular in-tree user that I could spot,
kallsyms_lookup_name() is exported to modules and provides a mechanism
for out-of-tree modules to access and invoke arbitrary, non-exported
kernel symbols when kallsyms is enabled.

This patch series fixes up that one user and unexports the symbol along
with kallsyms_on_each_symbol(), since that could also be abused in a
similar manner.

I would like to avoid out-of-tree modules being easily able to call
functions that are not exported.  kallsyms_lookup_name() makes this
trivial to the point that there is very little incentive to rework these
modules to either use upstream interfaces correctly or propose
functionality which may be otherwise missing upstream.  Both of these
latter solutions would be pre-requisites to upstreaming these modules, and
the current state of things actively discourages that approach.

The background here is that we are aiming for Android devices to be able
to use a generic binary kernel image closely following upstream, with any
vendor extensions coming in as kernel modules.  In this case, we (Google)
end up maintaining the binary module ABI within the scope of a single LTS
kernel.  Monitoring and managing the ABI surface is not feasible if it
effectively includes all data and functions via kallsyms_lookup_name().
Of course, we could just carry this patch in the Android kernel tree, but
we're aiming to carry as little as possible (ideally nothing) and I think
it's a sensible change in its own right.  I'm surprised you object to it,
in all honesty.

Now, you could turn around and say "that's not upstream's problem", but it
still seems highly undesirable to me to have an upstream bypass for
exported symbols that isn't even used by upstream modules.  It's ripe for
abuse and encourages people to work outside of the upstream tree.  The
usual rule is that we don't export symbols without a user in the tree and
that seems especially relevant in this case.

Joe Lawrence said:

: FWIW, kallsyms was historically used by the out-of-tree kpatch support
: module to resolve external symbols as well as call set_memory_r{w,o}()
: API.  All of that support code has been merged upstream, so modern kpatch
: modules* no longer leverage kallsyms by default.
:
: That said, there are still some users who still use the deprecated support
: module with newer kernels, but that is not officially supported by the
: project.

This patch (of 3):

Given the name of a kernel symbol, the 'data_breakpoint' test claims to
"report any write operations on the kernel symbol".  However, it creates
the breakpoint using both HW_BREAKPOINT_W and HW_BREAKPOINT_R, which menas
it also fires for read access.

Drop HW_BREAKPOINT_R from the breakpoint attributes.

Signed-off-by: Will Deacon <will@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Masami Hiramatsu <mhiramat@kernel.org>
Reviewed-by: Quentin Perret <qperret@google.com>
Cc: K.Prasad <prasad@linux.vnet.ibm.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Frederic Weisbecker <frederic@kernel.org>
Cc: Alexei Starovoitov <ast@kernel.org>
Cc: Miroslav Benes <mbenes@suse.cz>
Cc: Petr Mladek <pmladek@suse.com>
Cc: Joe Lawrence <joe.lawrence@redhat.com>
Link: http://lkml.kernel.org/r/20200221114404.14641-2-will@kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

fs/binfmt_elf.c: don't free interpreter's ELF pheaders on common path

Static executables don't need to free NULL pointer.

It doesn't matter really because static executable is not common scenario
but do it anyway out of pedantry.

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Link: http://lkml.kernel.org/r/20200219185330.GA4933@avx2
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

fs/binfmt_elf.c: allocate less for static executable

PT_INTERP ELF header can be spared if executable is static.

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Link: http://lkml.kernel.org/r/20200219185012.GB4871@avx2
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

fs/binfmt_elf.c: delete "loc" variable

"loc" variable became just a wrapper for PT_INTERP ELF header after main
ELF header was moved to "bprm->buf".  Delete it.

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Link: http://lkml.kernel.org/r/20200219184847.GA4871@avx2
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

fs/epoll: make nesting accounting safe for -rt kernel

Davidlohr Bueso pointed out that when CONFIG_DEBUG_LOCK_ALLOC is set
ep_poll_safewake() can take several non-raw spinlocks after disabling
interrupts.  Since a spinlock can block in the -rt kernel, we can't take a
spinlock after disabling interrupts.  So let's re-work how we determine
the nesting level such that it plays nicely with the -rt kernel.

Let's introduce a 'nests' field in struct eventpoll that records the
current nesting level during ep_poll_callback().  Then, if we nest again
we can find the previous struct eventpoll that we were called from and
increase our count by 1.  The 'nests' field is protected by
ep->poll_wait.lock.

I've also moved the visited field to reduce the size of struct eventpoll
from 184 bytes to 176 bytes on x86_64 for !CONFIG_DEBUG_LOCK_ALLOC, which
is typical for a production config.

Reported-by: Davidlohr Bueso <dbueso@suse.de>
Signed-off-by: Jason Baron <jbaron@akamai.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Davidlohr Bueso <dbueso@suse.de>
Cc: Roman Penyaev <rpenyaev@suse.de>
Cc: Eric Wong <normalperson@yhbt.net>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Link: http://lkml.kernel.org/r/1582739816-13167-1-git-send-email-jbaron@akamai.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

kselftest: introduce new epoll test case

This testcase repeats epollbug.c from the bug:

  https://bugzilla.kernel.org/show_bug.cgi?id=205933

What it tests?  It tests the race between epoll_ctl() and epoll_wait().
New event mask passed to epoll_ctl() triggers wake up, which can be missed
because of the bug described in the link.  Reproduction is 100%, so easy
to fix.  Kudos, Max, for wonderful test case.

Signed-off-by: Roman Penyaev <rpenyaev@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Max Neunhoeffer <max@arangodb.com>
Cc: Jakub Kicinski <kuba@kernel.org>
Cc: Christopher Kohlhoff <chris.kohlhoff@clearpool.io>
Cc: Davidlohr Bueso <dbueso@suse.de>
Cc: Jason Baron <jbaron@akamai.com>
Cc: Jes Sorensen <jes.sorensen@gmail.com>
Link: http://lkml.kernel.org/r/20200214170211.561524-2-rpenyaev@suse.de
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

checkpatch: avoid warning about uninitialized_var()

WARNING: function definition argument 'flags' should also have an identifier name
#26: FILE: drivers/tty/serial/sh-sci.c:1348:
+       unsigned long uninitialized_var(flags);

Special-case uninitialized_var() to prevent this.

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Tested-by: Andrew Morton <akpm@linux-foundation.org>
Link: http://lkml.kernel.org/r/7db7944761b0bd88c70eb17d4b7f40fe589e14ed.camel@perches.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

checkpatch: check proper licensing of Devicetree bindings

According to Devicetree maintainers (see Link: below), the Devicetree
binding documents are preferrably licensed (GPL-2.0-only OR BSD-2-Clause).

Let's check that.  The actual check is a bit more relaxed, to allow more
liberal but compatible licensing (e.g.  GPL-2.0-or-later OR BSD-2-Clause).

Signed-off-by: Lubomir Rintel <lkundrak@v3.sk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Joe Perches <joe@perches.com>
Acked-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Cc: Rob Herring <robh@kernel.org>
Cc: Neil Armstrong <narmstrong@baylibre.com>
Cc: Laurent Pinchart <Laurent.pinchart@ideasonboard.com>,
Cc: Jonas Karlman <jonas@kwiboo.se>,
Cc: Jernej Skrabec <jernej.skrabec@siol.net>,
Cc: Mark Rutland <mark.rutland@arm.com>,
Cc: David Airlie <airlied@linux.ie>
Cc: Daniel Vetter <daniel@ffwll.ch>,
Link: https://lore.kernel.org/lkml/20200108142132.GA4830@bogus/
Link: http://lkml.kernel.org/r/20200309215153.38824-1-lkundrak@v3.sk
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

checkpatch: improve Gerrit Change-Id: test

The Gerrit Change-Id: entry is sometimes placed after a Signed-off-by:
line.  When this occurs, the Gerrit warning is not currently emitted as
the first Signed-off-by: signature sets a flag to stop looking.

Change the test to add a test for the --- patch separator and emit the
warning before any before the --- and also before any diff file name.

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Tested-by: John Stultz <john.stultz@linaro.org>
Link: http://lkml.kernel.org/r/2f6d5f8766fe7439a116c77ea8cc721a3f2d77a2.camel@perches.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

checkpatch: add command-line option for TAB size

Linux kernel coding style requires a size of 8 characters for both TAB and
indentation, and such value is embedded as magic value allover the
checkpatch script.

This makes hard to reuse the script by other projects with different
requirements in their coding style (e.g.  OpenOCD [1] requires TAB size of
4 characters [2]).

Replace the magic value 8 with a variable.

Add a command-line option "--tab-size" to let the user select a
TAB size value other than 8.

[1] http://openocd.org/
[2] http://openocd.org/doc/doxygen/html/stylec.html#styleformat

Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Signed-off-by: Erik Ahlén <erik.ahlen@avalonenterprise.com>
Signed-off-by: Spencer Oliver <spen@spen-soft.co.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Joe Perches <joe@perches.com>
Link: http://lkml.kernel.org/r/20200122163852.124417-3-borneo.antonio@gmail.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

checkpatch: fix multiple const * types

Commit 1574a29f8e76 ("checkpatch: allow multiple const * types") claims to
support repetition of pattern "const *", but it actually allows only one
extra instance.

Check the following lines
int a(char const * const x[]);
int b(char const * const *x);
int c(char const * const * const x[]);
int d(char const * const * const *x);

with command

./scripts/checkpatch.pl --show-types -f filename

to find that only the first line passes the test, while a warning
is triggered by the other 3 lines:

WARNING:FUNCTION_ARGUMENTS: function definition argument
'char const * const' should also have an identifier name

The reason is that the pattern match halts at the second asterisk in the
line, thus the remaining text starting with asterisk fails to match a
valid name for a variable.

Fixed by replacing "?" (Match 1 or 0 times) with "{0,4}" (Match no more
than 4 times) in the regular expression.  Fix also the similar test for
types in unusual order.

Fixes: 1574a29f8e76 ("checkpatch: allow multiple const * types")
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Joe Perches <joe@perches.com>
Link: http://lkml.kernel.org/r/20200122163852.124417-1-borneo.antonio@gmail.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

checkpatch: fix minor typo and mixed space+tab in indentation

Fix spelling of "concatenation".
Don't use tab after space in indentation.

Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Joe Perches <joe@perches.com>
Link: http://lkml.kernel.org/r/20200122163852.124417-2-borneo.antonio@gmail.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

checkpatch: prefer fallthrough; over fallthrough comments

commit 294f69e662d1 ("compiler_attributes.h: Add 'fallthrough' pseudo
keyword for switch/case use") added the pseudo keyword so add a test for
it to checkpatch.

Warn on a patch or use --strict for files.

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Link: http://lkml.kernel.org/r/8b6c1b9031ab9f3cdebada06b8d46467f1492d68.camel@perches.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

checkpatch: support "base-commit:" format

In order to support the get-lore-mbox.py tool described in [1], I ran:

    git format-patch --base=<commit> --cover-letter <revrange>

...  which generated a "base-commit: <commit-hash>" tag at the end of the
cover letter.  However, checkpatch.pl generated an error upon encounting
"base-commit:" in the cover letter:

    "ERROR: Please use git commit description style..."

...  because it found the "commit" keyword, and failed to recognize that
it was part of the "base-commit" phrase, and as such, should not be
subjected to the same commit description style rules.

Update checkpatch.pl to include a special case for "base-commit:" (at the
start of the line, possibly with some leading whitespace) so that that tag
no longer generates a checkpatch error.

[1] https://lwn.net/Articles/811528/ "Better tools for kernel
    developers"

Suggested-by: Joe Perches <joe@perches.com>
Signed-off-by: John Hubbard <jhubbard@nvidia.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Joe Perches <joe@perches.com>
Cc: Andy Whitcroft <apw@canonical.com>
Cc: Konstantin Ryabitsev <konstantin@linuxfoundation.org>
Cc: Jonathan Corbet <corbet@lwn.net>
Link: http://lkml.kernel.org/r/20200213055004.69235-2-jhubbard@nvidia.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

checkpatch: check SPDX tags in YAML files

This adds a warning when a YAML file is lacking a SPDX header on first
line, or it uses incorrect commenting style.

Currently the only YAML files in the tree are Devicetree binding
documents.

Signed-off-by: Lubomir Rintel <lkundrak@v3.sk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Joe Perches <joe@perches.com>
Cc: Rob Herring <robh@kernel.org>
Link: http://lkml.kernel.org/r/20200129123356.388669-1-lkundrak@v3.sk
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

checkpatch: remove email address comment from email address comparisons

About 2% of the last 100K commits have email addresses that include an
RFC2822 compliant comment like:

Peter Zijlstra (Intel) <peterz@infradead.org>

checkpatch currently does a comparison of the complete name and address to
the submitted author to determine if the author has signed-off and emits a
warning if the exact email names and addresses do not match.

Unfortunately, the author email address can be written without the comment
like:

Peter Zijlstra <peterz@infradead.org>

Add logic to compare the comment stripped email addresses to avoid this
warning.

Reported-by: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: http://lkml.kernel.org/r/ebaa2f7c8f94e25520981945cddcc1982e70e072.camel@perches.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

lib/dynamic_debug.c: use address-of operator on section symbols

Clang warns:

../lib/dynamic_debug.c:1034:24: warning: array comparison always
evaluates to false [-Wtautological-compare]
        if (__start___verbose == __stop___verbose) {
                              ^
1 warning generated.

These are not true arrays, they are linker defined symbols, which are just
addresses.  Using the address of operator silences the warning and does
not change the resulting assembly with either clang/ld.lld or gcc/ld
(tested with diff + objdump -Dr).

Suggested-by: Nick Desaulniers <ndesaulniers@google.com>
Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Jason Baron <jbaron@akamai.com>
Link: https://github.com/ClangBuiltLinux/linux/issues/894
Link: http://lkml.kernel.org/r/20200220051320.10739-1-natechancellor@gmail.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

linux/bits.h: add compile time sanity check of GENMASK inputs

GENMASK() and GENMASK_ULL() are supposed to be called with the high bit as
the first argument and the low bit as the second argument.  Mixing them
will return a mask with zero bits set.

Recent commits show getting this wrong is not uncommon, see e.g.  commit
aa4c0c9091b0 ("net: stmmac: Fix misuses of GENMASK macro") and commit
9bdd7bb3a844 ("clocksource/drivers/npcm: Fix misuse of GENMASK macro").

To prevent such mistakes from appearing again, add compile time sanity
checking to the arguments of GENMASK() and GENMASK_ULL().  If both
arguments are known at compile time, and the low bit is higher than the
high bit, break the build to detect the mistake immediately.

Since GENMASK() is used in declarations, BUILD_BUG_ON_ZERO() must be used
instead of BUILD_BUG_ON().

__builtin_constant_p does not evaluate is argument, it only checks if it
is a constant or not at compile time, and __builtin_choose_expr does not
evaluate the expression that is not chosen.  Therefore, GENMASK(x++, 0)
does only evaluate x++ once.

Commit 95b980d62d52 ("linux/bits.h: make BIT(), GENMASK(), and friends
available in assembly") made the macros in linux/bits.h available in
assembly.  Since BUILD_BUG_OR_ZERO() is not asm compatible, disable the
checks if the file is included in an asm file.

Due to bugs in GCC versions before 4.9 [0], disable the check if building
with a too old GCC compiler.

[0]: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=19449

Signed-off-by: Rikard Falkeborn <rikard.falkeborn@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Kees Cook <keescook@chromium.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Haren Myneni <haren@us.ibm.com>
Cc: Joe Perches <joe@perches.com>
Cc: Johannes Berg <johannes@sipsolutions.net>
Cc: lkml <linux-kernel@vger.kernel.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/20200308193954.2372399-1-rikard.falkeborn@gmail.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

lib/test_kmod.c: remove a NULL test

The "info" pointer has already been dereferenced so checking here is too
late.  Fortunately, we never pass NULL pointers to the
test_kmod_put_module() function so the test can simply be removed.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Luis Chamberlain <mcgrof@kernel.org>
Link: http://lkml.kernel.org/r/20200228092452.vwkhthsn77nrxdy6@kili.mountain
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

lib/rbtree: fix coding style of assignments

Leave blank space between the right-hand and left-hand side of the
assignment to meet the kernel coding style better.

Signed-off-by: chenqiwu <chenqiwu@xiaomi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Michel Lespinasse <walken@google.com>
Link: http://lkml.kernel.org/r/1582621140-25850-1-git-send-email-qiwuchen55@gmail.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

lib/test_bitmap.c: make use of EXP2_IN_BITS

Commit 30544ed5de43 ("lib/bitmap: introduce bitmap_replace() helper")
introduced some new test cases to the test_bitmap.c module.  Among these
it also introduced an (unused) definition.  Let's make use of
EXP2_IN_BITS.

Reported-by: Alex Shi <alex.shi@linux.alibaba.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Alex Shi <alex.shi@linux.alibaba.com>
Link: http://lkml.kernel.org/r/20200121151847.75223-1-andriy.shevchenko@linux.intel.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

percpu_counter: fix a data race at vm_committed_as

"vm_committed_as.count" could be accessed concurrently as reported by
KCSAN,

 BUG: KCSAN: data-race in __vm_enough_memory / percpu_counter_add_batch

 write to 0xffffffff9451c538 of 8 bytes by task 65879 on cpu 35:
  percpu_counter_add_batch+0x83/0xd0
  percpu_counter_add_batch at lib/percpu_counter.c:91
  __vm_enough_memory+0xb9/0x260
  dup_mm+0x3a4/0x8f0
  copy_process+0x2458/0x3240
  _do_fork+0xaa/0x9f0
  __do_sys_clone+0x125/0x160
  __x64_sys_clone+0x70/0x90
  do_syscall_64+0x91/0xb05
  entry_SYSCALL_64_after_hwframe+0x49/0xbe

 read to 0xffffffff9451c538 of 8 bytes by task 66773 on cpu 19:
  __vm_enough_memory+0x199/0x260
  percpu_counter_read_positive at include/linux/percpu_counter.h:81
  (inlined by) __vm_enough_memory at mm/util.c:839
  mmap_region+0x1b2/0xa10
  do_mmap+0x45c/0x700
  vm_mmap_pgoff+0xc0/0x130
  ksys_mmap_pgoff+0x6e/0x300
  __x64_sys_mmap+0x33/0x40
  do_syscall_64+0x91/0xb05
  entry_SYSCALL_64_after_hwframe+0x49/0xbe

The read is outside percpu_counter::lock critical section which results in
a data race.  Fix it by adding a READ_ONCE() in
percpu_counter_read_positive() which could also service as the existing
compiler memory barrier.

Signed-off-by: Qian Cai <cai@lca.pw>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Marco Elver <elver@google.com>
Link: http://lkml.kernel.org/r/1582302724-2804-1-git-send-email-cai@lca.pw
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

kasan: stackdepot: move filter_irq_stacks() to stackdepot.c

filter_irq_stacks() can be used by other tools (e.g.  KMSAN), so it needs
to be moved to a common location.  lib/stackdepot.c seems a good place, as
filter_irq_stacks() is usually applied to the output of
stack_trace_save().

This patch has been previously mailed as part of KMSAN RFC patch series.

[glider@google.co: nds32: linker script: add SOFTIRQENTRY_TEXT\
Link: http://lkml.kernel.org/r/20200311121002.241430-1-glider@google.com
[glider@google.com: add IRQENTRY_TEXT and SOFTIRQENTRY_TEXT to linker script]
Link: http://lkml.kernel.org/r/20200311121124.243352-1-glider@google.com
Signed-off-by: Alexander Potapenko <glider@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Vegard Nossum <vegard.nossum@oracle.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Marco Elver <elver@google.com>
Cc: Andrey Konovalov <andreyknvl@google.com>
Cc: Andrey Ryabinin <aryabinin@virtuozzo.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
Link: http://lkml.kernel.org/r/20200220141916.55455-3-glider@google.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

lib/stackdepot.c: build with -fno-builtin

Clang may replace stackdepot_memcmp() with a call to instrumented bcmp(),
which is exactly what we wanted to avoid creating stackdepot_memcmp().
Building the file with -fno-builtin prevents such optimizations.

This patch has been previously mailed as part of KMSAN RFC patch series.

Signed-off-by: Alexander Potapenko <glider@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Vegard Nossum <vegard.nossum@oracle.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Marco Elver <elver@google.com>
Cc: Andrey Konovalov <andreyknvl@google.com>
Cc: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Andrey Ryabinin <aryabinin@virtuozzo.com>
Link: http://lkml.kernel.org/r/20200220141916.55455-2-glider@google.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

lib/stackdepot.c: check depot_index before accessing the stack slab

Avoid crashes on corrupted stack ids.  Despite stack ID corruption may
indicate other bugs in the program, we'd better fail gracefully on such
IDs instead of crashing the kernel.

This patch has been previously mailed as part of KMSAN RFC patch series.

Link: http://lkml.kernel.org/r/20200220141916.55455-1-glider@google.com
Signed-off-by: Alexander Potapenko <glider@google.com>
Cc: Vegard Nossum <vegard.nossum@oracle.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Marco Elver <elver@google.com>
Cc: Andrey Konovalov <andreyknvl@google.com>
Cc: Andrey Ryabinin <aryabinin@virtuozzo.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
From: Dan Carpenter <dan.carpenter@oracle.com>
Subject: lib/stackdepot.c: fix a condition in stack_depot_fetch()

We should check for a NULL pointer first before adding the offset.
Otherwise if the pointer is NULL and the offset is non-zero, it will lead
to an Oops.

Fixes: d45048e65a59 ("lib/stackdepot.c: check depot_index before accessing the stack slab")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Alexander Potapenko <glider@google.com>
Link: http://lkml.kernel.org/r/20200312113006.GA20562@mwanda
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

lib: test_stackinit.c: XFAIL switch variable init tests

The tests for initializing a variable defined between a switch statement's
test and its first "case" statement are currently not initialized in
Clang[1] nor the proposed auto-initialization feature in GCC.

We should retain the test (so that we can evaluate compiler fixes), but
mark it as an "expected fail".  The rest of the kernel source will be
adjusted to avoid this corner case.

Also disable -Wswitch-unreachable for the test so that the intentionally
broken code won't trigger warnings for GCC (nor future Clang) when
initialization happens this unhandled place.

[1] https://bugs.llvm.org/show_bug.cgi?id=44916

Suggested-by: Alexander Potapenko <glider@google.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Jann Horn <jannh@google.com>
Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Link: http://lkml.kernel.org/r/202002191358.2897A07C6@keescook
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

lib/scatterlist: fix sg_copy_buffer() kerneldoc

Add the missing closing parenthesis to the description for the to_buffer
parameter of sg_copy_buffer().

Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Akinobu Mita <akinobu.mita@gmail.com
Link: http://lkml.kernel.org/r/20200212084241.8778-1-geert+renesas@glider.be
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

lib/ts_kmp.c: replace zero-length array with flexible-array member

The current codebase makes use of the zero-length array language extension
to the C90 standard, but the preferred mechanism to declare
variable-length types such as these ones is a flexible array member[1][2],
introduced in C99:

struct foo {
        int stuff;
        struct boo array[];
};

By making use of the mechanism above, we will get a compiler warning in
case the flexible array does not occur last in the structure, which will
help us prevent some kind of undefined behavior bugs from being
inadvertenly introduced[3] to the codebase from now on.

This issue was found with the help of Coccinelle.

[1] https://gcc.gnu.org/onlinedocs/gcc/Zero-Length.html
[2] https://github.com/KSPP/linux/issues/21
[3] commit 76497732932f ("cxgb3/l2t: Fix undefined behaviour")

Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Link: http://lkml.kernel.org/r/20200211205948.GA26459@embeddedor
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

lib/ts_fsm.c: replace zero-length array with flexible-array member

The current codebase makes use of the zero-length array language extension
to the C90 standard, but the preferred mechanism to declare
variable-length types such as these ones is a flexible array member[1][2],
introduced in C99:

struct foo {
        int stuff;
        struct boo array[];
};

By making use of the mechanism above, we will get a compiler warning in
case the flexible array does not occur last in the structure, which will
help us prevent some kind of undefined behavior bugs from being
inadvertenly introduced[3] to the codebase from now on.

This issue was found with the help of Coccinelle.

[1] https://gcc.gnu.org/onlinedocs/gcc/Zero-Length.html
[2] https://github.com/KSPP/linux/issues/21
[3] commit 76497732932f ("cxgb3/l2t: Fix undefined behaviour")

Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Link: http://lkml.kernel.org/r/20200211205813.GA25602@embeddedor
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

lib/ts_bm.c: replace zero-length array with flexible-array member

The current codebase makes use of the zero-length array language extension
to the C90 standard, but the preferred mechanism to declare
variable-length types such as these ones is a flexible array member[1][2],
introduced in C99:

struct foo {
        int stuff;
        struct boo array[];
};

By making use of the mechanism above, we will get a compiler warning in
case the flexible array does not occur last in the structure, which will
help us prevent some kind of undefined behavior bugs from being
inadvertenly introduced[3] to the codebase from now on.

This issue was found with the help of Coccinelle.

[1] https://gcc.gnu.org/onlinedocs/gcc/Zero-Length.html
[2] https://github.com/KSPP/linux/issues/21
[3] commit 76497732932f ("cxgb3/l2t: Fix undefined behaviour")

Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Link: http://lkml.kernel.org/r/20200211205620.GA24694@embeddedor
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

lib/bch.c: replace zero-length array with flexible-array member

The current codebase makes use of the zero-length array language extension
to the C90 standard, but the preferred mechanism to declare
variable-length types such as these ones is a flexible array member[1][2],
introduced in C99:

struct foo {
        int stuff;
        struct boo array[];
};

By making use of the mechanism above, we will get a compiler warning in
case the flexible array does not occur last in the structure, which will
help us prevent some kind of undefined behavior bugs from being
inadvertenly introduced[3] to the codebase from now on.

This issue was found with the help of Coccinelle.

[1] https://gcc.gnu.org/onlinedocs/gcc/Zero-Length.html
[2] https://github.com/KSPP/linux/issues/21
[3] commit 76497732932f ("cxgb3/l2t: Fix undefined behaviour")

Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Link: http://lkml.kernel.org/r/20200211205119.GA21234@embeddedor
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

lib/test_lockup.c: add parameters for locking generic vfs locks

file_path=<path> defines file or directory to open
lock_inode=Y set lock_rwsem_ptr to inode->i_rwsem
lock_mapping=Y set lock_rwsem_ptr to mapping->i_mmap_rwsem
lock_sb_umount=Y set lock_rwsem_ptr to sb->s_umount

This gives safe and simple way to see how system reacts to contention of
common vfs locks and how syscalls depend on them directly or indirectly.

For example to block s_umount for 60 seconds:
# modprobe test_lockup file_path=. lock_sb_umount time_secs=60 state=S

This is useful for checking/testing scalability issues like this:
https://lore.kernel.org/lkml/158497590858.7371.9311902565121473436.stgit@buzz/

Signed-off-by: Konstantin Khlebnikov <khlebnikov@yandex-team.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Colin Ian King <colin.king@canonical.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Guenter Roeck <linux@roeck-us.net>
Cc: Kees Cook <keescook@chromium.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Petr Mladek <pmladek@suse.com>
Cc: Sasha Levin <sashal@kernel.org>
Cc: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Link: http://lkml.kernel.org/r/158498153964.5621.83061779039255681.stgit@buzz
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

lib/test_lockup.c: fix spelling mistake "iteraions" -> "iterations"

There is a spelling mistake in a pr_notice message.  Fix it.

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Guenter Roeck <linux@roeck-us.net>
Cc: Kees Cook <keescook@chromium.org>
Cc: Konstantin Khlebnikov <khlebnikov@yandex-team.ru>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Petr Mladek <pmladek@suse.com>
Cc: Sasha Levin <sashal@kernel.org>
Cc: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Link: http://lkml.kernel.org/r/20200221155145.79522-1-colin.king@canonical.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

lib/test_lockup: test module to generate lockups

CONFIG_TEST_LOCKUP=m adds module "test_lockup" that helps to make sure
that watchdogs and lockup detectors are working properly.

Depending on module parameters test_lockup could emulate soft or hard
lockup, "hung task", hold arbitrary lock, allocate bunch of pages.

Also it could generate series of lockups with cooling-down periods, in
this way it could be used as "ping" for locks or page allocator.  Loop
checks signals between iteration thus could be stopped by ^C.

# modinfo test_lockup
...
parm:           time_secs:lockup time in seconds, default 0 (uint)
parm:           time_nsecs:nanoseconds part of lockup time, default 0 (uint)
parm:           cooldown_secs:cooldown time between iterations in seconds, default 0 (uint)
parm:           cooldown_nsecs:nanoseconds part of cooldown, default 0 (uint)
parm:           iterations:lockup iterations, default 1 (uint)
parm:           all_cpus:trigger lockup at all cpus at once (bool)
parm:           state:wait in 'R' running (default), 'D' uninterruptible, 'K' killable, 'S' interruptible state (charp)
parm:           use_hrtimer:use high-resolution timer for sleeping (bool)
parm:           iowait:account sleep time as iowait (bool)
parm:           lock_read:lock read-write locks for read (bool)
parm:           lock_single:acquire locks only at one cpu (bool)
parm:           reacquire_locks:release and reacquire locks/irq/preempt between iterations (bool)
parm:           touch_softlockup:touch soft-lockup watchdog between iterations (bool)
parm:           touch_hardlockup:touch hard-lockup watchdog between iterations (bool)
parm:           call_cond_resched:call cond_resched() between iterations (bool)
parm:           measure_lock_wait:measure lock wait time (bool)
parm:           lock_wait_threshold:print lock wait time longer than this in nanoseconds, default off (ulong)
parm:           disable_irq:disable interrupts: generate hard-lockups (bool)
parm:           disable_softirq:disable bottom-half irq handlers (bool)
parm:           disable_preempt:disable preemption: generate soft-lockups (bool)
parm:           lock_rcu:grab rcu_read_lock: generate rcu stalls (bool)
parm:           lock_mmap_sem:lock mm->mmap_sem: block procfs interfaces (bool)
parm:           lock_rwsem_ptr:lock rw_semaphore at address (ulong)
parm:           lock_mutex_ptr:lock mutex at address (ulong)
parm:           lock_spinlock_ptr:lock spinlock at address (ulong)
parm:           lock_rwlock_ptr:lock rwlock at address (ulong)
parm:           alloc_pages_nr:allocate and free pages under locks (uint)
parm:           alloc_pages_order:page order to allocate (uint)
parm:           alloc_pages_gfp:allocate pages with this gfp_mask, default GFP_KERNEL (uint)
parm:           alloc_pages_atomic:allocate pages with GFP_ATOMIC (bool)
parm:           reallocate_pages:free and allocate pages between iterations (bool)

Parameters for locking by address are unsafe and taints kernel. With
CONFIG_DEBUG_SPINLOCK=y they at least check magics for embedded spinlocks.

Examples:

task hang in D-state:
modprobe test_lockup time_secs=1 iterations=60 state=D

task hang in io-wait D-state:
modprobe test_lockup time_secs=1 iterations=60 state=D iowait

softlockup:
modprobe test_lockup time_secs=1 iterations=60 state=R

hardlockup:
modprobe test_lockup time_secs=1 iterations=60 state=R disable_irq

system-wide hardlockup:
modprobe test_lockup time_secs=1 iterations=60 state=R \
 disable_irq all_cpus

rcu stall:
modprobe test_lockup time_secs=1 iterations=60 state=R \
 lock_rcu touch_softlockup

lock mmap_sem / block procfs interfaces:
modprobe test_lockup time_secs=1 iterations=60 state=S lock_mmap_sem

lock tasklist_lock for read / block forks:
TASKLIST_LOCK=$(awk '$3 == "tasklist_lock" {print "0x"$1}' /proc/kallsyms)
modprobe test_lockup time_secs=1 iterations=60 state=R \
 disable_irq lock_read lock_rwlock_ptr=$TASKLIST_LOCK

lock namespace_sem / block vfs mount operations:
NAMESPACE_SEM=$(awk '$3 == "namespace_sem" {print "0x"$1}' /proc/kallsyms)
modprobe test_lockup time_secs=1 iterations=60 state=S \
 lock_rwsem_ptr=$NAMESPACE_SEM

lock cgroup mutex / block cgroup operations:
CGROUP_MUTEX=$(awk '$3 == "cgroup_mutex" {print "0x"$1}' /proc/kallsyms)
modprobe test_lockup time_secs=1 iterations=60 state=S \
 lock_mutex_ptr=$CGROUP_MUTEX

ping cgroup_mutex every second and measure maximum lock wait time:
modprobe test_lockup cooldown_secs=1 iterations=60 state=S \
 lock_mutex_ptr=$CGROUP_MUTEX reacquire_locks measure_lock_wait

[linux@roeck-us.net: rename disable_irq to fix build error]
Link: http://lkml.kernel.org/r/20200317133614.23152-1-linux@roeck-us.net
Signed-off-by: Konstantin Khlebnikov <khlebnikov@yandex-team.ru>
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Sasha Levin <sashal@kernel.org>
Cc: Petr Mladek <pmladek@suse.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
Cc: Dmitry Monakhov <dmtrmonakhov@yandex-team.ru
Cc: Colin Ian King <colin.king@canonical.com>
Cc: Guenter Roeck <linux@roeck-us.net>
Link: http://lkml.kernel.org/r/158132859146.2797.525923171323227836.stgit@buzz
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

bitops: always inline sign extension helpers

With CONFIG_CC_OPTIMIZE_FOR_SIZE, objtool reports:

  drivers/gpu/drm/i915/gem/i915_gem_execbuffer.o: warning: objtool: i915_gem_execbuffer2_ioctl()+0x5b7: call to gen8_canonical_addr() with UACCESS enabled

This means i915_gem_execbuffer2_ioctl() is calling gen8_canonical_addr()
from the user_access_begin/end critical region (i.e, with SMAP disabled).

While it's probably harmless in this case, in general we like to avoid
extra function calls in SMAP-disabled regions because it can open up
inadvertent security holes.

Fix the warning by changing the sign extension helpers to __always_inline.
This convinces GCC to inline gen8_canonical_addr().

The sign extension functions are trivial anyway, so it makes sense to
always inline them.  With my test optimize-for-size-based config, this
actually shrinks the text size of i915_gem_execbuffer.o by 45 bytes -- and
no change for vmlinux.

Reported-by: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Link: http://lkml.kernel.org/r/740179324b2b18b750b16295c48357f00b5fa9ed.1582982020.git.jpoimboe@redhat.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

MAINTAINERS: list the section entries in the preferred order

The MAINTAINERS file header has never shown a preferred order for the
section entries but scripts/parse-maintainers.pl added a preferred order
with commit 61f741645a35 ("parse-maintainers: Add section pattern
sorting")

Commit 5cdbec108fd2 ("parse-maintainers: Do not sort section content by
default") changed the preferred order to be a bit more sensible.

Update the MAINTAINERS section description block to use this preferred
section entry ordering.

Add a slightly better description for the N: entry too.

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Link: http://lkml.kernel.org/r/5aa5aad6fb1678230c260337dc066cd449a2bf32.camel@perches.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

compiler.h: fix error in BUILD_BUG_ON() reporting

compiletime_assert() uses __LINE__ to create a unique function name.  This
means that if you have more than one BUILD_BUG_ON() in the same source
line (which can happen if they appear e.g.  in a macro), then the error
message from the compiler might output the wrong condition.

For this source file:

#include <linux/build_bug.h>

#define macro() \
BUILD_BUG_ON(1); \
BUILD_BUG_ON(0);

void foo()
{
macro();
}

gcc would output:

./include/linux/compiler.h:350:38: error: call to `__compiletime_assert_9' declared with attribute error: BUILD_BUG_ON failed: 0
  _compiletime_assert(condition, msg, __compiletime_assert_, __LINE__)

However, it was not the BUILD_BUG_ON(0) that failed, so it should say 1
instead of 0. With this patch, we use __COUNTER__ instead of __LINE__, so
each BUILD_BUG_ON() gets a different function name and the correct
condition is printed:

./include/linux/compiler.h:350:38: error: call to `__compiletime_assert_0' declared with attribute error: BUILD_BUG_ON failed: 1
  _compiletime_assert(condition, msg, __compiletime_assert_, __COUNTER__)

Signed-off-by: Vegard Nossum <vegard.nossum@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Daniel Santos <daniel.santos@pobox.com>
Cc: Rasmus Villemoes <linux@rasmusvillemoes.dk>
Cc: Ian Abbott <abbotti@mev.co.uk>
Cc: Joe Perches <joe@perches.com>
Link: http://lkml.kernel.org/r/20200331112637.25047-1-vegard.nossum@oracle.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

compiler: remove CONFIG_OPTIMIZE_INLINING entirely

Commit ac7c3e4ff401 ("compiler: enable CONFIG_OPTIMIZE_INLINING
forcibly") made this always-on option. We released v5.4 and v5.5
including that commit.

Remove the CONFIG option and clean up the code now.

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Miguel Ojeda <miguel.ojeda.sandonis@gmail.com>
Reviewed-by: Nathan Chancellor <natechancellor@gmail.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Borislav Petkov <bp@alien8.de>
Cc: David Miller <davem@davemloft.net>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/20200220110807.32534-2-masahiroy@kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

sparc,x86: vdso: remove meaningless undefining CONFIG_OPTIMIZE_INLINING

The code, #undef CONFIG_OPTIMIZE_INLINING, is not working as expected
because <linux/compiler_types.h> is parsed before vclock_gettime.c since
28128c61e08e ("kconfig.h: Include compiler types to avoid missed struct
attributes").

Since then, <linux/compiler_types.h> is included really early by using the
'-include' option.  So, you cannot negate the decision of
<linux/compiler_types.h> in this way.

You can confirm it by checking the pre-processed code, like this:

  $ make arch/x86/entry/vdso/vdso32/vclock_gettime.i

There is no difference with/without CONFIG_CC_OPTIMIZE_FOR_SIZE.

It is about two years since 28128c61e08e.  Nobody has reported a problem
(or, nobody has even noticed the fact that this code is not working).

It is ugly and unreliable to attempt to undefine a CONFIG option from C
files, and anyway the inlining heuristic is up to the compiler.

Just remove the broken code.

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Nathan Chancellor <natechancellor@gmail.com>
Acked-by: Miguel Ojeda <miguel.ojeda.sandonis@gmail.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Masahiro Yamada <masahiroy@kernel.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: David Miller <davem@davemloft.net>
Link: http://lkml.kernel.org/r/20200220110807.32534-1-masahiroy@kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

kernel/extable.c: use address-of operator on section symbols

Clang warns:

../kernel/extable.c:37:52: warning: array comparison always evaluates to
a constant [-Wtautological-compare]
        if (main_extable_sort_needed && __stop___ex_table > __start___ex_table) {
                                                          ^
1 warning generated.

These are not true arrays, they are linker defined symbols, which are just
addresses.  Using the address of operator silences the warning and does
not change the resulting assembly with either clang/ld.lld or gcc/ld
(tested with diff + objdump -Dr).

Suggested-by: Nick Desaulniers <ndesaulniers@google.com>
Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Link: https://github.com/ClangBuiltLinux/linux/issues/892
Link: http://lkml.kernel.org/r/20200219202036.45702-1-natechancellor@gmail.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

asm-generic: fix unistd_32.h generation format

Generated files are also checked by sparse that's why add newline to
remove sparse (C=1) warning.

The issue was found on Microblaze and reported like this:
./arch/microblaze/include/generated/uapi/asm/unistd_32.h:438:45: warning:
no newline at end of file

Mips and PowerPC have it already but let's align with style used by m68k.

Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Stefan Asserhall <stefan.asserhall@xilinx.com>
Acked-by: Max Filippov <jcmvbkbc@gmail.com> (xtensa)
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Max Filippov <jcmvbkbc@gmail.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Chris Zankel <chris@zankel.net>
Cc: David S. Miller <davem@davemloft.net>
Cc: Fenghua Yu <fenghua.yu@intel.com>
Cc: Helge Deller <deller@gmx.de>
Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru>
Cc: James Bottomley <James.Bottomley@HansenPartnership.com>
Cc: Matt Turner <mattst88@gmail.com>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Paul Burton <paulburton@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Rich Felker <dalias@libc.org>
Cc: Richard Henderson <rth@twiddle.net>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Link: http://lkml.kernel.org/r/4d32ab4e1fb2edb691d2e1687e8fb303c09fd023.1581504803.git.michal.simek@xilinx.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

proc: inline m_next_vma into m_next

It's clearer to just put this inline.

Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Link: http://lkml.kernel.org/r/20200317193201.9924-5-adobriyan@gmail.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

seq_file: remove m->version

The process maps file was the only user of version (introduced back in
2005).  Now that it uses ppos instead, we can remove it.

Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Link: http://lkml.kernel.org/r/20200317193201.9924-4-adobriyan@gmail.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

proc: use ppos instead of m->version

The ppos is a private cursor, just like m->version.  Use the canonical
cursor, not a special one.

Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Link: http://lkml.kernel.org/r/20200317193201.9924-3-adobriyan@gmail.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

proc: remove m_cache_vma

Instead of setting m->version in the show method, set it in m_next(),
where it should be.  Also remove the fallback code for failing to find a
vma, or version being zero.

Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Link: http://lkml.kernel.org/r/20200317193201.9924-2-adobriyan@gmail.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

proc: inline vma_stop into m_stop

Instead of calling vma_stop() from m_start() and m_next(), do its work
in m_stop().

Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Link: http://lkml.kernel.org/r/20200317193201.9924-1-adobriyan@gmail.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

proc: speed up /proc/*/statm

top(1) reads all /proc/*/statm files but kernel threads will always have
zeros.  Print those zeroes directly without going through
seq_put_decimal_ull().

Speed up reading /proc/2/statm (which is kthreadd) is like 3%.

My system has more kernel threads than normal processes after booting KDE.

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Link: http://lkml.kernel.org/r/20200307154435.GA2788@avx2
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

proc: faster open/read/close with "permanent" files

Now that "struct proc_ops" exist we can start putting there stuff which
could not fly with VFS "struct file_operations"...

Most of fs/proc/inode.c file is dedicated to make open/read/.../close
reliable in the event of disappearing /proc entries which usually happens
if module is getting removed.  Files like /proc/cpuinfo which never
disappear simply do not need such protection.

Save 2 atomic ops, 1 allocation, 1 free per open/read/close sequence for such
"permanent" files.

Enable "permanent" flag for

/proc/cpuinfo
/proc/kmsg
/proc/modules
/proc/slabinfo
/proc/stat
/proc/sysvipc/*
/proc/swaps

More will come once I figure out foolproof way to prevent out module
authors from marking their stuff "permanent" for performance reasons
when it is not.

This should help with scalability: benchmark is "read /proc/cpuinfo R times
by N threads scattered over the system".

N R t, s (before) t, s (after)
-----------------------------------------------------
64 4096 1.582458 1.530502 -3.2%
256 4096 6.371926 6.125168 -3.9%
1024 4096 25.64888 24.47528 -4.6%

Benchmark source:

#include <chrono>
#include <iostream>
#include <thread>
#include <vector>

#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <unistd.h>

const int NR_CPUS = sysconf(_SC_NPROCESSORS_ONLN);
int N;
const char *filename;
int R;

int xxx = 0;

int glue(int n)
{
cpu_set_t m;
CPU_ZERO(&m);
CPU_SET(n, &m);
return sched_setaffinity(0, sizeof(cpu_set_t), &m);
}

void f(int n)
{
glue(n % NR_CPUS);

while (*(volatile int *)&xxx == 0) {
}

for (int i = 0; i < R; i++) {
int fd = open(filename, O_RDONLY);
char buf[4096];
ssize_t rv = read(fd, buf, sizeof(buf));
asm volatile ("" :: "g" (rv));
close(fd);
}
}

int main(int argc, char *argv[])
{
if (argc < 4) {
std::cerr << "usage: " << argv[0] << ' ' << "N /proc/filename R
";
return 1;
}

N = atoi(argv[1]);
filename = argv[2];
R = atoi(argv[3]);

for (int i = 0; i < NR_CPUS; i++) {
if (glue(i) == 0)
break;
}

std::vector<std::thread> T;
T.reserve(N);
for (int i = 0; i < N; i++) {
T.emplace_back(f, i);
}

auto t0 = std::chrono::system_clock::now();
{
*(volatile int *)&xxx = 1;
for (auto& t: T) {
t.join();
}
}
auto t1 = std::chrono::system_clock::now();
std::chrono::duration<double> dt = t1 - t0;
std::cout << dt.count() << '
';

return 0;
}

P.S.:
Explicit randomization marker is added because adding non-function pointer
will silently disable structure layout randomization.

[akpm@linux-foundation.org: coding style fixes]
Reported-by: kbuild test robot <lkp@intel.com>
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Joe Perches <joe@perches.com>
Link: http://lkml.kernel.org/r/20200222201539.GA22576@avx2
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

fs/proc/inode.c: annotate close_pdeo() for sparse

Fix sparse locking imbalance warning:

warning: context imbalance in close_pdeo() - unexpected unlock

Signed-off-by: Jules Irenge <jbi.octave@gmail.com>
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Link: http://lkml.kernel.org/r/20200227201538.GA30462@avx2
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm: remove dummy struct bootmem_data/bootmem_data_t

Both bootmem_data and bootmem_data_t structures are no longer defined.
Remove the dummy forward declarations.

Signed-off-by: Waiman Long <longman@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Baoquan He <bhe@redhat.com>
Acked-by: Mike Rapoport <rppt@linux.ibm.com>
Link: http://lkml.kernel.org/r/20200326022617.26208-1-longman@redhat.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/dmapool.c: micro-optimisation remove unnecessary branch

Previously there was a check if 'size' is aligned to 'align' and if not
then it was aligned.  This check was expensive as both branch and division
are expensive instructions in most architectures.  'ALIGN' function on
already aligned value will not change it, and as it is cheaper than branch
+ division it can be executed all the time and branch can be removed.

Signed-off-by: Mateusz Nosek <mateusznosek0@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Link: http://lkml.kernel.org/r/20200320173317.26408-1-mateusznosek0@gmail.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

include/linux/memremap.h: remove stale comments

Fixes: 80a72d0af05a ("memremap: remove the data field in struct dev_pagemap")
Fixes: fdc029b19dfd ("memremap: remove the dev field in struct dev_pagemap")
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Cc: Jason Gunthorpe <jgg@ziepe.ca>
Cc: Dan Williams <dan.j.williams@intel.com>
Link: http://lkml.kernel.org/r/20200316213205.145333-1-ira.weiny@intel.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

include/linux/swapops.h: correct guards for non_swap_entry()

If CONFIG_DEVICE_PRIVATE is defined, but neither CONFIG_MEMORY_FAILURE nor
CONFIG_MIGRATION, then non_swap_entry() will return 0, meaning that the
condition (non_swap_entry(entry) && is_device_private_entry(entry)) in
zap_pte_range() will never be true even if the entry is a device private
one.

Equally any other code depending on non_swap_entry() will not function as
expected.

I originally spotted this just by looking at the code, I haven't actually
observed any problems.

Looking a bit more closely it appears that actually this situation
(currently at least) cannot occur:

DEVICE_PRIVATE depends on ZONE_DEVICE
ZONE_DEVICE depends on MEMORY_HOTREMOVE
MEMORY_HOTREMOVE depends on MIGRATION

Fixes: 5042db43cc26 ("mm/ZONE_DEVICE: new type of ZONE_DEVICE for unaddressable memory")
Signed-off-by: Steven Price <steven.price@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Jérôme Glisse <jglisse@redhat.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: John Hubbard <jhubbard@nvidia.com>
Link: http://lkml.kernel.org/r/20200305130550.22693-1-steven.price@arm.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm: use fallthrough;

Convert the various /* fallthrough */ comments to the pseudo-keyword
fallthrough;

Done via script:
https://lore.kernel.org/lkml/b56602fcf79f849e733e7b521bb0e17895d390fa.1582230379.git.joe@perches.com/

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Link: http://lkml.kernel.org/r/f62fea5d10eb0ccfc05d87c242a620c261219b66.camel@perches.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/mm_init.c: clean code. Use BUILD_BUG_ON when comparing compile time constant

MAX_ZONELISTS is a compile time constant, so it should be compared using
BUILD_BUG_ON not BUG_ON.

Signed-off-by: Mateusz Nosek <mateusznosek0@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Wei Yang <richard.weiyang@gmail.com>
Link: http://lkml.kernel.org/r/20200228224617.11343-1-mateusznosek0@gmail.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm: fix ambiguous comments for better code readability

The parameter of remap_pfn_range() @pfn passed from the caller is actually
a page-frame number converted by corresponding physical address of kernel
memory, the original comment is ambiguous that may mislead the users.

Meanwhile, there is an ambiguous typo "VMM" in the comment of
vm_area_struct.  So fixing them will make the code more readable.

Signed-off-by: chenqiwu <chenqiwu@xiaomi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Link: http://lkml.kernel.org/r/1583026921-15279-1-git-send-email-qiwuchen55@gmail.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/zsmalloc: add missing annotation for unpin_tag()

Sparse reports a warning at unpin_tag()()

warning: context imbalance in unpin_tag() - unexpected unlock

The root cause is the missing annotation at unpin_tag()
Add the missing __releases(bitlock) annotation

Signed-off-by: Jules Irenge <jbi.octave@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Minchan Kim <minchan@kernel.org>
Link: http://lkml.kernel.org/r/20200214204741.94112-14-jbi.octave@gmail.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/zsmalloc: add missing annotation for pin_tag()

Sparse reports a warning at pin_tag()()

warning: context imbalance in pin_tag() - wrong count at exit

The root cause is the missing annotation at pin_tag()
Add the missing __acquires(bitlock) annotation

Signed-off-by: Jules Irenge <jbi.octave@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Minchan Kim <minchan@kernel.org>
Link: http://lkml.kernel.org/r/20200214204741.94112-13-jbi.octave@gmail.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/zsmalloc: add missing annotation for migrate_read_unlock()

Sparse reports a warning at migrate_read_unlock()()

 warning: context imbalance in migrate_read_unlock() - unexpected unlock

The root cause is the missing annotation at migrate_read_unlock()
Add the missing __releases(&zspage->lock) annotation

Signed-off-by: Jules Irenge <jbi.octave@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Minchan Kim <minchan@kernel.org>
Link: http://lkml.kernel.org/r/20200214204741.94112-12-jbi.octave@gmail.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/zsmalloc: add missing annotation for migrate_read_lock()

Sparse reports a warning at migrate_read_lock()()

 warning: context imbalance in migrate_read_lock() - wrong count at exit

The root cause is the missing annotation at migrate_read_lock()
Add the missing __acquires(&zspage->lock) annotation

Signed-off-by: Jules Irenge <jbi.octave@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Minchan Kim <minchan@kernel.org>
Link: http://lkml.kernel.org/r/20200214204741.94112-11-jbi.octave@gmail.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/slub: add missing annotation for put_map()

Sparse reports a warning at put_map()()

 warning: context imbalance in put_map() - unexpected unlock

The root cause is the missing annotation at put_map()
Add the missing __releases(&object_map_lock) annotation

Signed-off-by: Jules Irenge <jbi.octave@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Link: http://lkml.kernel.org/r/20200214204741.94112-10-jbi.octave@gmail.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/slub: add missing annotation for get_map()

Sparse reports a warning at get_map()()

 warning: context imbalance in get_map() - wrong count at exit

The root cause is the missing annotation at get_map()
Add the missing __acquires(&object_map_lock) annotation

Signed-off-by: Jules Irenge <jbi.octave@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Link: http://lkml.kernel.org/r/20200214204741.94112-9-jbi.octave@gmail.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/mempolicy: add missing annotation for queue_pages_pmd()

Sparse reports a warning at queue_pages_pmd()

context imbalance in queue_pages_pmd() - unexpected unlock

The root cause is the missing annotation at queue_pages_pmd()
Add the missing __releases(ptl)

Signed-off-by: Jules Irenge <jbi.octave@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Link: http://lkml.kernel.org/r/20200214204741.94112-8-jbi.octave@gmail.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/hugetlb: add missing annotation for gather_surplus_pages()

Sparse reports a warning at gather_surplus_pages()

warning: context imbalance in hugetlb_cow() - unexpected unlock

The root cause is the missing annotation at gather_surplus_pages()
Add the missing __must_hold(&hugetlb_lock)

Signed-off-by: Jules Irenge <jbi.octave@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com>
Link: http://lkml.kernel.org/r/20200214204741.94112-7-jbi.octave@gmail.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/compaction: add missing annotation for compact_lock_irqsave

Sparse reports a warning at compact_lock_irqsave()

warning: context imbalance in compact_lock_irqsave() - wrong count at exit

The root cause is the missing annotation at compact_lock_irqsave()
Add the missing __acquires(lock) annotation.

Signed-off-by: Jules Irenge <jbi.octave@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Link: http://lkml.kernel.org/r/20200214204741.94112-6-jbi.octave@gmail.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/zswap: allow setting default status, compressor and allocator in Kconfig

The compressed cache for swap pages (zswap) currently needs from 1 to 3
extra kernel command line parameters in order to make it work: it has to
be enabled by adding a "zswap.enabled=1" command line parameter and if one
wants a different compressor or pool allocator than the default lzo / zbud
combination then these choices also need to be specified on the kernel
command line in additional parameters.

Using a different compressor and allocator for zswap is actually pretty
common as guides often recommend using the lz4 / z3fold pair instead of
the default one.  In such case it is also necessary to remember to enable
the appropriate compression algorithm and pool allocator in the kernel
config manually.

Let's avoid the need for adding these kernel command line parameters and
automatically pull in the dependencies for the selected compressor
algorithm and pool allocator by adding an appropriate default switches to
Kconfig.

The default values for these options match what the code was using
previously as its defaults.

Signed-off-by: Maciej S. Szmigiero <mail@maciej.szmigiero.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Vitaly Wool <vitaly.wool@konsulko.com>
Link: http://lkml.kernel.org/r/20200202000112.456103-1-mail@maciej.szmigiero.name
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm: prevent a warning when casting void* -> enum

I recently build the RISC-V port with LLVM trunk, which has introduced a
new warning when casting from a pointer to an enum of a smaller size.
This patch simply casts to a long in the middle to stop the warning.  I'd
be surprised this is the only one in the kernel, but it's the only one I
saw.

Signed-off-by: Palmer Dabbelt <palmerdabbelt@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Link: http://lkml.kernel.org/r/20200227211741.83165-1-palmer@dabbelt.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm: huge tmpfs: try to split_huge_page() when punching hole

Yang Shi writes:

Currently, when truncating a shmem file, if the range is partly in a THP
(start or end is in the middle of THP), the pages actually will just get
cleared rather than being freed, unless the range covers the whole THP.
Even though all the subpages are truncated (randomly or sequentially), the
THP may still be kept in page cache.

This might be fine for some usecases which prefer preserving THP, but
balloon inflation is handled in base page size.  So when using shmem THP
as memory backend, QEMU inflation actually doesn't work as expected since
it doesn't free memory.  But the inflation usecase really needs to get the
memory freed.  (Anonymous THP will also not get freed right away, but will
be freed eventually when all subpages are unmapped: whereas shmem THP
still stays in page cache.)

Split THP right away when doing partial hole punch, and if split fails
just clear the page so that read of the punched area will return zeroes.

Hugh Dickins adds:

Our earlier "team of pages" huge tmpfs implementation worked in the way
that Yang Shi proposes; and we have been using this patch to continue to
split the huge page when hole-punched or truncated, since converting over
to the compound page implementation.  Although huge tmpfs gives out huge
pages when available, if the user specifically asks to truncate or punch a
hole (perhaps to free memory, perhaps to reduce the memcg charge), then
the filesystem should do so as best it can, splitting the huge page.

That is not always possible: any additional reference to the huge page
prevents split_huge_page() from succeeding, so the result can be flaky.
But in practice it works successfully enough that we've not seen any
problem from that.

Add shmem_punch_compound() to encapsulate the decision of when a split is
needed, and doing the split if so.  Using this simplifies the flow in
shmem_undo_range(); and the first (trylock) pass does not need to do any
page clearing on failure, because the second pass will either succeed or
do that clearing.  Following the example of zero_user_segment() when
clearing a partial page, add flush_dcache_page() and set_page_dirty() when
clearing a hole - though I'm not certain that either is needed.

But: split_huge_page() would be sure to fail if shmem_undo_range()'s
pagevec holds further references to the huge page.  The easiest way to fix
that is for find_get_entries() to return early, as soon as it has put one
compound head or tail into the pagevec.  At first this felt like a hack;
but on examination, this convention better suits all its callers - or will
do, if the slight one-page-per-pagevec slowdown in shmem_unlock_mapping()
and shmem_seek_hole_data() is transformed into a 512-page-per-pagevec
speedup by checking for compound pages there.

Signed-off-by: Hugh Dickins <hughd@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Yang Shi <yang.shi@linux.alibaba.com>
Cc: Alexander Duyck <alexander.duyck@gmail.com>
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Link: http://lkml.kernel.org/r/alpine.LSU.2.11.2002261959020.10801@eggly.anvils
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/shmem.c: clean code by removing unnecessary assignment

Previously 0 was assigned to variable 'error' but the variable was never
read before reassignemnt later.  So the assignment can be removed.

Signed-off-by: Mateusz Nosek <mateusznosek0@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Acked-by: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Cc: Hugh Dickins <hughd@google.com>
Link: http://lkml.kernel.org/r/20200301152832.24595-1-mateusznosek0@gmail.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/shmem.c: distribute switch variables for initialization

Variables declared in a switch statement before any case statements cannot
be automatically initialized with compiler instrumentation (as they are
not part of any execution flow).  With GCC's proposed automatic stack
variable initialization feature, this triggers a warning (and they don't
get initialized).  Clang's automatic stack variable initialization (via
CONFIG_INIT_STACK_ALL=y) doesn't throw a warning, but it also doesn't
initialize such variables[1].  Note that these warnings (or silent
skipping) happen before the dead-store elimination optimization phase, so
even when the automatic initializations are later elided in favor of
direct initializations, the warnings remain.

To avoid these problems, move such variables into the "case" where they're
used or lift them up into the main function body.

mm/shmem.c: In function `shmem_getpage_gfp':
mm/shmem.c:1816:10: warning: statement will never be executed [-Wswitch-unreachable]
 1816 |   loff_t i_size;
      |          ^~~~~~

[1] https://bugs.llvm.org/show_bug.cgi?id=44916

Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Alexander Potapenko <glider@google.com>
Link: http://lkml.kernel.org/r/20200220062312.69165-1-keescook@chromium.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/memory_hotplug.c: use __pfn_to_section() instead of open-coding

Use __pfn_to_section() API instead of open-coding for better code
readability.

Signed-off-by: chenqiwu <chenqiwu@xiaomi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: David Hildenbrand <david@redhat.com>
Link: http://lkml.kernel.org/r/1584345134-16671-1-git-send-email-qiwuchen55@gmail.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/memory_hotplug: allow to specify a default online_type

For now, distributions implement advanced udev rules to essentially
- Don't online any hotplugged memory (s390x)
- Online all memory to ZONE_NORMAL (e.g., most virt environments like
  hyperv)
- Online all memory to ZONE_MOVABLE in case the zone imbalance is taken
  care of (e.g., bare metal, special virt environments)

In summary: All memory is usually onlined the same way, however, the
kernel always has to ask user space to come up with the same answer.
E.g., Hyper-V always waits for a memory block to get onlined before
continuing, otherwise it might end up adding memory faster than
onlining it, which can result in strange OOM situations.  This waiting
slows down adding of a bigger amount of memory.

Let's allow to specify a default online_type, not just "online" and
"offline".  This allows distributions to configure the default online_type
when booting up and be done with it.

We can now specify "offline", "online", "online_movable" and
"online_kernel" via
- "memhp_default_state=" on the kernel cmdline
- /sys/devices/system/memory/auto_online_blocks
just like we are able to specify for a single memory block via
/sys/devices/system/memory/memoryX/state

Signed-off-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Wei Yang <richard.weiyang@gmail.com>
Reviewed-by: Baoquan He <bhe@redhat.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Oscar Salvador <osalvador@suse.de>
Cc: "Rafael J. Wysocki" <rafael@kernel.org>
Cc: Wei Yang <richard.weiyang@gmail.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Eduardo Habkost <ehabkost@redhat.com>
Cc: Haiyang Zhang <haiyangz@microsoft.com>
Cc: Igor Mammedov <imammedo@redhat.com>
Cc: "K. Y. Srinivasan" <kys@microsoft.com>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Stephen Hemminger <sthemmin@microsoft.com>
Cc: Vitaly Kuznetsov <vkuznets@redhat.com>
Cc: Wei Liu <wei.liu@kernel.org>
Cc: Yumei Huang <yuhuang@redhat.com>
Link: http://lkml.kernel.org/r/20200317104942.11178-9-david@redhat.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/memory_hotplug: convert memhp_auto_online to store an online_type

...  and rename it to memhp_default_online_type.  This is a preparation
for more detailed default online behavior.

Signed-off-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Wei Yang <richard.weiyang@gmail.com>
Reviewed-by: Baoquan He <bhe@redhat.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Oscar Salvador <osalvador@suse.de>
Cc: "Rafael J. Wysocki" <rafael@kernel.org>
Cc: Wei Yang <richard.weiyang@gmail.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Eduardo Habkost <ehabkost@redhat.com>
Cc: Haiyang Zhang <haiyangz@microsoft.com>
Cc: Igor Mammedov <imammedo@redhat.com>
Cc: "K. Y. Srinivasan" <kys@microsoft.com>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Stephen Hemminger <sthemmin@microsoft.com>
Cc: Vitaly Kuznetsov <vkuznets@redhat.com>
Cc: Wei Liu <wei.liu@kernel.org>
Cc: Yumei Huang <yuhuang@redhat.com>
Link: http://lkml.kernel.org/r/20200317104942.11178-8-david@redhat.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/memory_hotplug: unexport memhp_auto_online

All in-tree users except the mm-core are gone. Let's drop the export.

Signed-off-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Wei Yang <richard.weiyang@gmail.com>
Reviewed-by: Baoquan He <bhe@redhat.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Cc: Oscar Salvador <osalvador@suse.de>
Cc: "Rafael J. Wysocki" <rafael@kernel.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Eduardo Habkost <ehabkost@redhat.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Haiyang Zhang <haiyangz@microsoft.com>
Cc: Igor Mammedov <imammedo@redhat.com>
Cc: "K. Y. Srinivasan" <kys@microsoft.com>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Stephen Hemminger <sthemmin@microsoft.com>
Cc: Vitaly Kuznetsov <vkuznets@redhat.com>
Cc: Wei Liu <wei.liu@kernel.org>
Cc: Yumei Huang <yuhuang@redhat.com>
Link: http://lkml.kernel.org/r/20200317104942.11178-7-david@redhat.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

hv_balloon: don't check for memhp_auto_online manually

We get the MEM_ONLINE notifier call if memory is added right from the
kernel via add_memory() or later from user space.

Let's get rid of the "ha_waiting" flag - the wait event has an inbuilt
mechanism (->done) for that.  Initialize the wait event only once and
reinitialize before adding memory.  Unconditionally call complete() and
wait_for_completion_timeout().

If there are no waiters, complete() will only increment ->done - which
will be reset by reinit_completion().  If complete() has already been
called, wait_for_completion_timeout() will not wait.

There is still the chance for a small race between concurrent
reinit_completion() and complete().  If complete() wins, we would not wait
- which is tolerable (and the race exists in current code as well).

Note: We only wait for "some" memory to get onlined, which seems to be
      good enough for now.

[akpm@linux-foundation.org: register_memory_notifier() after init_completion(), per David]
Signed-off-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Reviewed-by: Baoquan He <bhe@redhat.com>
Cc: "K. Y. Srinivasan" <kys@microsoft.com>
Cc: Haiyang Zhang <haiyangz@microsoft.com>
Cc: Stephen Hemminger <sthemmin@microsoft.com>
Cc: Wei Liu <wei.liu@kernel.org>
Cc: Oscar Salvador <osalvador@suse.de>
Cc: "Rafael J. Wysocki" <rafael@kernel.org>
Cc: Wei Yang <richard.weiyang@gmail.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Eduardo Habkost <ehabkost@redhat.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Igor Mammedov <imammedo@redhat.com>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Yumei Huang <yuhuang@redhat.com>
Link: http://lkml.kernel.org/r/20200317104942.11178-6-david@redhat.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

powernv/memtrace: always online added memory blocks

Let's always try to online the re-added memory blocks.  In case
add_memory() already onlined the added memory blocks, the first
device_online() call will fail and stop processing the remaining memory
blocks.

This avoids manually having to check memhp_auto_online.

Note: PPC always onlines all hotplugged memory directly from the kernel as
well - something that is handled by user space on other architectures.

Signed-off-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Wei Yang <richard.weiyang@gmail.com>
Reviewed-by: Baoquan He <bhe@redhat.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Oscar Salvador <osalvador@suse.de>
Cc: "Rafael J. Wysocki" <rafael@kernel.org>
Cc: Eduardo Habkost <ehabkost@redhat.com>
Cc: Haiyang Zhang <haiyangz@microsoft.com>
Cc: Igor Mammedov <imammedo@redhat.com>
Cc: "K. Y. Srinivasan" <kys@microsoft.com>
Cc: Stephen Hemminger <sthemmin@microsoft.com>
Cc: Vitaly Kuznetsov <vkuznets@redhat.com>
Cc: Wei Liu <wei.liu@kernel.org>
Cc: Yumei Huang <yuhuang@redhat.com>
Link: http://lkml.kernel.org/r/20200317104942.11178-5-david@redhat.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

drivers/base/memory: store mapping between MMOP_* and string in an array

Let's use a simple array which we can reuse soon.  While at it, move the
string->mmop conversion out of the device hotplug lock.

Signed-off-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Wei Yang <richard.weiyang@gmail.com>
Reviewed-by: Baoquan He <bhe@redhat.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Oscar Salvador <osalvador@suse.de>
Cc: "Rafael J. Wysocki" <rafael@kernel.org>
Cc: Wei Yang <richard.weiyang@gmail.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Eduardo Habkost <ehabkost@redhat.com>
Cc: Haiyang Zhang <haiyangz@microsoft.com>
Cc: Igor Mammedov <imammedo@redhat.com>
Cc: "K. Y. Srinivasan" <kys@microsoft.com>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Stephen Hemminger <sthemmin@microsoft.com>
Cc: Vitaly Kuznetsov <vkuznets@redhat.com>
Cc: Wei Liu <wei.liu@kernel.org>
Cc: Yumei Huang <yuhuang@redhat.com>
Link: http://lkml.kernel.org/r/20200317104942.11178-4-david@redhat.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

drivers/base/memory: map MMOP_OFFLINE to 0

Historically, we used the value -1.  Just treat 0 as the special case now.
Clarify a comment (which was wrong, when we come via device_online() the
first time, the online_type would have been 0 / MEM_ONLINE).  The default
is now always MMOP_OFFLINE.  This removes the last user of the manual
"-1", which didn't use the enum value.

This is a preparation to use the online_type as an array index.

Signed-off-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Wei Yang <richard.weiyang@gmail.com>
Reviewed-by: Baoquan He <bhe@redhat.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Oscar Salvador <osalvador@suse.de>
Cc: "Rafael J. Wysocki" <rafael@kernel.org>
Cc: Wei Yang <richard.weiyang@gmail.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Eduardo Habkost <ehabkost@redhat.com>
Cc: Haiyang Zhang <haiyangz@microsoft.com>
Cc: Igor Mammedov <imammedo@redhat.com>
Cc: "K. Y. Srinivasan" <kys@microsoft.com>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Stephen Hemminger <sthemmin@microsoft.com>
Cc: Vitaly Kuznetsov <vkuznets@redhat.com>
Cc: Wei Liu <wei.liu@kernel.org>
Cc: Yumei Huang <yuhuang@redhat.com>
Link: http://lkml.kernel.org/r/20200317104942.11178-3-david@redhat.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

drivers/base/memory: rename MMOP_ONLINE_KEEP to MMOP_ONLINE

Patch series "mm/memory_hotplug: allow to specify a default online_type", v3.

Distributions nowadays use udev rules ([1] [2]) to specify if and how to
online hotplugged memory.  The rules seem to get more complex with many
special cases.  Due to the various special cases,
CONFIG_MEMORY_HOTPLUG_DEFAULT_ONLINE cannot be used.  All memory hotplug
is handled via udev rules.

Every time we hotplug memory, the udev rule will come to the same
conclusion.  Especially Hyper-V (but also soon virtio-mem) add a lot of
memory in separate memory blocks and wait for memory to get onlined by
user space before continuing to add more memory blocks (to not add memory
faster than it is getting onlined).  This of course slows down the whole
memory hotplug process.

To make the job of distributions easier and to avoid udev rules that get
more and more complicated, let's extend the mechanism provided by
- /sys/devices/system/memory/auto_online_blocks
- "memhp_default_state=" on the kernel cmdline
to be able to specify also "online_movable" as well as "online_kernel"

=== Example /usr/libexec/config-memhotplug ===

#!/bin/bash

VIRT=`systemd-detect-virt --vm`
ARCH=`uname -p`

sense_virtio_mem() {
  if [ -d "/sys/bus/virtio/drivers/virtio_mem/" ]; then
    DEVICES=`find /sys/bus/virtio/drivers/virtio_mem/ -maxdepth 1 -type l | wc -l`
    if [ $DEVICES != "0" ]; then
        return 0
    fi
  fi
  return 1
}

if [ ! -e "/sys/devices/system/memory/auto_online_blocks" ]; then
  echo "Memory hotplug configuration support missing in the kernel"
  exit 1
fi

if grep "memhp_default_state=" /proc/cmdline > /dev/null; then
  echo "Memory hotplug configuration overridden in kernel cmdline (memhp_default_state=)"
  exit 1
fi

if [ $VIRT == "microsoft" ]; then
  echo "Detected Hyper-V on $ARCH"
  # Hyper-V wants all memory in ZONE_NORMAL
  ONLINE_TYPE="online_kernel"
elif sense_virtio_mem; then
  echo "Detected virtio-mem on $ARCH"
  # virtio-mem wants all memory in ZONE_NORMAL
  ONLINE_TYPE="online_kernel"
elif [ $ARCH == "s390x" ] || [ $ARCH == "s390" ]; then
  echo "Detected $ARCH"
  # standby memory should not be onlined automatically
  ONLINE_TYPE="offline"
elif [ $ARCH == "ppc64" ] || [ $ARCH == "ppc64le" ]; then
  echo "Detected" $ARCH
  # PPC64 onlines all hotplugged memory right from the kernel
  ONLINE_TYPE="offline"
elif [ $VIRT == "none" ]; then
  echo "Detected bare-metal on $ARCH"
  # Bare metal users expect hotplugged memory to be unpluggable. We assume
  # that ZONE imbalances on such enterpise servers cannot happen and is
  # properly documented
  ONLINE_TYPE="online_movable"
else
  # TODO: Hypervisors that want to unplug DIMMs and can guarantee that ZONE
  # imbalances won't happen
  echo "Detected $VIRT on $ARCH"
  # Usually, ballooning is used in virtual environments, so memory should go to
  # ZONE_NORMAL. However, sometimes "movable_node" is relevant.
  ONLINE_TYPE="online"
fi

echo "Selected online_type:" $ONLINE_TYPE

# Configure what to do with memory that will be hotplugged in the future
echo $ONLINE_TYPE 2>/dev/null > /sys/devices/system/memory/auto_online_blocks
if [ $? != "0" ]; then
  echo "Memory hotplug cannot be configured (e.g., old kernel or missing permissions)"
  # A backup udev rule should handle old kernels if necessary
  exit 1
fi

# Process all already pluggedd blocks (e.g., DIMMs, but also Hyper-V or virtio-mem)
if [ $ONLINE_TYPE != "offline" ]; then
  for MEMORY in /sys/devices/system/memory/memory*; do
    STATE=`cat $MEMORY/state`
    if [ $STATE == "offline" ]; then
        echo $ONLINE_TYPE > $MEMORY/state
    fi
  done
fi

=== Example /usr/lib/systemd/system/config-memhotplug.service ===

[Unit]
Description=Configure memory hotplug behavior
DefaultDependencies=no
Conflicts=shutdown.target
Before=sysinit.target shutdown.target
After=systemd-modules-load.service
ConditionPathExists=|/sys/devices/system/memory/auto_online_blocks

[Service]
ExecStart=/usr/libexec/config-memhotplug
Type=oneshot
TimeoutSec=0
RemainAfterExit=yes

[Install]
WantedBy=sysinit.target

=== Example modification to the 40-redhat.rules [2] ===

: diff --git a/40-redhat.rules b/40-redhat.rules-new
: index 2c690e5..168fd03 100644
: --- a/40-redhat.rules
: +++ b/40-redhat.rules-new
: @@ -6,6 +6,9 @@ SUBSYSTEM=="cpu", ACTION=="add", TEST=="online", ATTR{online}=="0", ATTR{online}
:  # Memory hotadd request
:  SUBSYSTEM!="memory", GOTO="memory_hotplug_end"
:  ACTION!="add", GOTO="memory_hotplug_end"
: +# memory hotplug behavior configured
: +PROGRAM=="grep online /sys/devices/system/memory/auto_online_blocks", GOTO="memory_hotplug_end"
: +
:  PROGRAM="/bin/uname -p", RESULT=="s390*", GOTO="memory_hotplug_end"
:
:  ENV{.state}="online"

===

[1] https://github.com/lnykryn/systemd-rhel/pull/281
[2] https://github.com/lnykryn/systemd-rhel/blob/staging/rules/40-redhat.rules

This patch (of 8):

The name is misleading and it's not really clear what is "kept".  Let's
just name it like the online_type name we expose to user space ("online").

Add some documentation to the types.

Signed-off-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Wei Yang <richard.weiyang@gmail.com>
Reviewed-by: Baoquan He <bhe@redhat.com>
Acked-by: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Oscar Salvador <osalvador@suse.de>
Cc: "Rafael J. Wysocki" <rafael@kernel.org>
Cc: Wei Yang <richard.weiyang@gmail.com>
Cc: Vitaly Kuznetsov <vkuznets@redhat.com>
Cc: Yumei Huang <yuhuang@redhat.com>
Cc: Igor Mammedov <imammedo@redhat.com>
Cc: Eduardo Habkost <ehabkost@redhat.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Haiyang Zhang <haiyangz@microsoft.com>
Cc: K. Y. Srinivasan <kys@microsoft.com>
Cc: Michael Ellerman <mpe@ellerman.id.au> (powerpc)
Cc: Paul Mackerras <paulus@samba.org>
Cc: Stephen Hemminger <sthemmin@microsoft.com>
Cc: Wei Liu <wei.liu@kernel.org>
Link: http://lkml.kernel.org/r/20200319131221.14044-1-david@redhat.com
Link: http://lkml.kernel.org/r/20200317104942.11178-2-david@redhat.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/sparse.c: move subsection_map related functions together

No functional change.

[bhe@redhat.com: move functions into CONFIG_MEMORY_HOTPLUG ifdeffery scope]
Link: http://lkml.kernel.org/r/20200316045804.GC3486@MiWiFi-R3L-srv
Signed-off-by: Baoquan He <bhe@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Wei Yang <richard.weiyang@gmail.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Link: http://lkml.kernel.org/r/20200312124414.439-6-bhe@redhat.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/sparse.c: add note about only VMEMMAP supporting sub-section hotplug

And tell check_pfn_span() gating the porper alignment and size of hot
added memory region.

And also move the code comments from inside section_deactivate() to being
above it.  The code comments are reasonable for the whole function, and
the moving makes code cleaner.

Signed-off-by: Baoquan He <bhe@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: David Hildenbrand <david@redhat.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Cc: Wei Yang <richard.weiyang@gmail.com>
Link: http://lkml.kernel.org/r/20200312124414.439-5-bhe@redhat.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/sparse.c: only use subsection map in VMEMMAP case

Currently, to support subsection aligned memory region adding for pmem,
subsection map is added to track which subsection is present.

However, config ZONE_DEVICE depends on SPARSEMEM_VMEMMAP.  It means
subsection map only makes sense when SPARSEMEM_VMEMMAP enabled.  For the
classic sparse, it's meaningless.  Even worse, it may confuse people when
checking code related to the classic sparse.

About the classic sparse which doesn't support subsection hotplug, Dan
said it's more because the effort and maintenance burden outweighs the
benefit.  Besides, the current 64 bit ARCHes all enable
SPARSEMEM_VMEMMAP_ENABLE by default.

Combining the above reasons, no need to provide subsection map and the
relevant handling for the classic sparse.  Let's remove them.

Signed-off-by: Baoquan He <bhe@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: David Hildenbrand <david@redhat.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Cc: Wei Yang <richard.weiyang@gmail.com>
Link: http://lkml.kernel.org/r/20200312124414.439-4-bhe@redhat.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/sparse.c: introduce a new function clear_subsection_map()

Factor out the code which clear subsection map of one memory region from
section_deactivate() into clear_subsection_map().

And also add helper function is_subsection_map_empty() to check if the
current subsection map is empty or not.

Signed-off-by: Baoquan He <bhe@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: David Hildenbrand <david@redhat.com>
Acked-by: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Wei Yang <richard.weiyang@gmail.com>
Link: http://lkml.kernel.org/r/20200312124414.439-3-bhe@redhat.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/sparse.c: introduce new function fill_subsection_map()

Patch series "mm/hotplug: Only use subsection map for VMEMMAP", v4.

Memory sub-section hotplug was added to fix the issue that nvdimm could be
mapped at non-section aligned starting address.  A subsection map is added
into struct mem_section_usage to implement it.

However, config ZONE_DEVICE depends on SPARSEMEM_VMEMMAP.  It means
subsection map only makes sense when SPARSEMEM_VMEMMAP enabled.  For the
classic sparse, subsection map is meaningless and confusing.

About the classic sparse which doesn't support subsection hotplug, Dan
said it's more because the effort and maintenance burden outweighs the
benefit.  Besides, the current 64 bit ARCHes all enable
SPARSEMEM_VMEMMAP_ENABLE by default.

This patch (of 5):

Factor out the code that fills the subsection map from section_activate()
into fill_subsection_map(), this makes section_activate() cleaner and
easier to follow.

Signed-off-by: Baoquan He <bhe@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Wei Yang <richard.weiyang@gmail.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Acked-by: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Michal Hocko <mhocko@suse.com>
Link: http://lkml.kernel.org/r/20200312124414.439-2-bhe@redhat.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/memory_hotplug.c: cleanup __add_pages()

Let's drop the basically unused section stuff and simplify.  The logic now
matches the logic in __remove_pages().

Signed-off-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Baoquan He <bhe@redhat.com>
Reviewed-by: Wei Yang <richard.weiyang@gmail.com>
Cc: Segher Boessenkool <segher@kernel.crashing.org>
Cc: Oscar Salvador <osalvador@suse.de>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Dan Williams <dan.j.williams@intel.com>
Link: http://lkml.kernel.org/r/20200228095819.10750-3-david@redhat.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/memory_hotplug.c: simplify calculation of number of pages in __remove_pages()

In commit 52fb87c81f11 ("mm/memory_hotplug: cleanup __remove_pages()"), we
cleaned up __remove_pages(), and introduced a shorter variant to calculate
the number of pages to the next section boundary.

Turns out we can make this calculation easier to read.  We always want to
have the number of pages (> 0) to the next section boundary, starting from
the current pfn.

We'll clean up __remove_pages() in a follow-up patch and directly make use
of this computation.

Suggested-by: Segher Boessenkool <segher@kernel.crashing.org>
Signed-off-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Baoquan He <bhe@redhat.com>
Reviewed-by: Wei Yang <richard.weiyang@gmail.com>
Cc: Oscar Salvador <osalvador@suse.de>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Dan Williams <dan.j.williams@intel.com>
Link: http://lkml.kernel.org/r/20200228095819.10750-2-david@redhat.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>