platform/kernel/linux-rpi.git
6 years agobdi: use refcount_t for reference counting instead atomic_t
Sebastian Andrzej Siewior [Wed, 22 Aug 2018 04:55:31 +0000 (21:55 -0700)]
bdi: use refcount_t for reference counting instead atomic_t

refcount_t type and corresponding API should be used instead of atomic_t
when the variable is used as a reference counter.  This permits avoiding
accidental refcounter overflows that might lead to use-after-free
situations.

Link: http://lkml.kernel.org/r/20180703200141.28415-4-bigeasy@linutronix.de
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Suggested-by: Peter Zijlstra <peterz@infradead.org>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agokernel.h: documentation for roundup() vs round_up()
Kees Cook [Wed, 22 Aug 2018 04:55:28 +0000 (21:55 -0700)]
kernel.h: documentation for roundup() vs round_up()

Things like 3619dec5103d ("dh key: fix rounding up KDF output length")
expose the lack of explicit documentation for roundup() vs round_up().  At
least we can try to document it better if anyone goes looking.

Link: http://lkml.kernel.org/r/20180703041950.GA43464@beast
Signed-off-by: Kees Cook <keescook@chromium.org>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agoinclude/asm-generic/bug.h: clarify valid uses of WARN()
Dmitry Vyukov [Wed, 22 Aug 2018 04:55:24 +0000 (21:55 -0700)]
include/asm-generic/bug.h: clarify valid uses of WARN()

Explicitly state that WARN*() should be used only for recoverable kernel
issues/bugs and that it should not be used for any kind of invalid
external inputs or transient conditions.

Motivation: it's a very useful capability to be able to understand if a
particular kernel splat means a kernel bug or simply an invalid user-space
program.  For the former one wants to notify kernel developers, while
notifying kernel developers for the latter is annoying.  Even a kernel
developer may not know what to do with a WARNING in an unfamiliar
subsystem.  This is especially critical for any automated testing systems
that may use panic_on_warn and mail kernel developers.

The clear separation also serves as an additional documentation: is it a
condition that must never occur because of additional checks/logic
elsewhere?  or is it simply a check for invalid inputs or unfortunate
conditions?

Use of pr_err() for user messages also leads to better error messages.
"Something is wrong in file foo on line X" is not particularly useful
message for end user.  pr_err() forces developers to write more meaningful
error messages for user.

As of now we are almost there.  We are doing systematic kernel testing
with panic_on_warn and are not seeing massive amounts of false positives.
But every now and then another WARN on ENOMEM or invalid inputs pops up
and leads to a lengthy argument each time.  The goal of this change is to
officially document the rules.

Link: http://lkml.kernel.org/r/20180620103716.61636-1-dvyukov@gmail.com
Signed-off-by: Dmitry Vyukov <dvyukov@google.com>
Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agoproc/kcore: add vmcoreinfo note to /proc/kcore
Omar Sandoval [Wed, 22 Aug 2018 04:55:20 +0000 (21:55 -0700)]
proc/kcore: add vmcoreinfo note to /proc/kcore

The vmcoreinfo information is useful for runtime debugging tools, not just
for crash dumps.  A lot of this information can be determined by other
means, but this is much more convenient, and it only adds a page at most
to the file.

Link: http://lkml.kernel.org/r/fddbcd08eed76344863303878b12de1c1e2a04b6.1531953780.git.osandov@fb.com
Signed-off-by: Omar Sandoval <osandov@fb.com>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Bhupesh Sharma <bhsharma@redhat.com>
Cc: Eric Biederman <ebiederm@xmission.com>
Cc: James Morse <james.morse@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agocrash_core: use VMCOREINFO_SYMBOL_ARRAY() for swapper_pg_dir
Omar Sandoval [Wed, 22 Aug 2018 04:55:17 +0000 (21:55 -0700)]
crash_core: use VMCOREINFO_SYMBOL_ARRAY() for swapper_pg_dir

This is preparation for allowing CRASH_CORE to be enabled for any
architecture.

swapper_pg_dir is always either an array or a macro expanding to NULL.
In the latter case, VMCOREINFO_SYMBOL() won't work, as it tries to take
the address of the given symbol:

#define VMCOREINFO_SYMBOL(name) \
vmcoreinfo_append_str("SYMBOL(%s)=%lx\n", #name, (unsigned long)&name)

Instead, use VMCOREINFO_SYMBOL_ARRAY(), which uses the value:

#define VMCOREINFO_SYMBOL_ARRAY(name) \
vmcoreinfo_append_str("SYMBOL(%s)=%lx\n", #name, (unsigned long)name)

This is the same thing for the array case but isn't an error for the macro
case.

Link: http://lkml.kernel.org/r/c05f9781ec204f40fc96f95086e7b6de6a3eb2c3.1532563124.git.osandov@fb.com
Signed-off-by: Omar Sandoval <osandov@fb.com>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Bhupesh Sharma <bhsharma@redhat.com>
Cc: Eric Biederman <ebiederm@xmission.com>
Cc: James Morse <james.morse@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agoproc/kcore: optimize multiple page reads
Omar Sandoval [Wed, 22 Aug 2018 04:55:13 +0000 (21:55 -0700)]
proc/kcore: optimize multiple page reads

The current code does a full search of the segment list every time for
every page.  This is wasteful, since it's almost certain that the next
page will be in the same segment.  Instead, check if the previous segment
covers the current page before doing the list search.

Link: http://lkml.kernel.org/r/fd346c11090cf93d867e01b8d73a6567c5ac6361.1531953780.git.osandov@fb.com
Signed-off-by: Omar Sandoval <osandov@fb.com>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Bhupesh Sharma <bhsharma@redhat.com>
Cc: Eric Biederman <ebiederm@xmission.com>
Cc: James Morse <james.morse@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agoproc/kcore: clean up ELF header generation
Omar Sandoval [Wed, 22 Aug 2018 04:55:09 +0000 (21:55 -0700)]
proc/kcore: clean up ELF header generation

Currently, the ELF file header, program headers, and note segment are
allocated all at once, in some icky code dating back to 2.3.  Programs
tend to read the file header, then the program headers, then the note
segment, all separately, so this is a waste of effort.  It's cleaner and
more efficient to handle the three separately.

Link: http://lkml.kernel.org/r/19c92cbad0e11f6103ff3274b2e7a7e51a1eb74b.1531953780.git.osandov@fb.com
Signed-off-by: Omar Sandoval <osandov@fb.com>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Bhupesh Sharma <bhsharma@redhat.com>
Cc: Eric Biederman <ebiederm@xmission.com>
Cc: James Morse <james.morse@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agoproc/kcore: hold lock during read
Omar Sandoval [Wed, 22 Aug 2018 04:55:06 +0000 (21:55 -0700)]
proc/kcore: hold lock during read

Now that we're using an rwsem, we can hold it during the entirety of
read_kcore() and have a common return path.  This is preparation for the
next change.

[akpm@linux-foundation.org: fix locking bug reported by Tetsuo Handa]
Link: http://lkml.kernel.org/r/d7cfbc1e8a76616f3b699eaff9df0a2730380534.1531953780.git.osandov@fb.com
Signed-off-by: Omar Sandoval <osandov@fb.com>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Bhupesh Sharma <bhsharma@redhat.com>
Cc: Eric Biederman <ebiederm@xmission.com>
Cc: James Morse <james.morse@arm.com>
Cc: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agoproc/kcore: fix memory hotplug vs multiple opens race
Omar Sandoval [Wed, 22 Aug 2018 04:55:02 +0000 (21:55 -0700)]
proc/kcore: fix memory hotplug vs multiple opens race

There's a theoretical race condition that will cause /proc/kcore to miss
a memory hotplug event:

CPU0                              CPU1
// hotplug event 1
kcore_need_update = 1

open_kcore()                      open_kcore()
    kcore_update_ram()                kcore_update_ram()
        // Walk RAM                       // Walk RAM
        __kcore_update_ram()              __kcore_update_ram()
            kcore_need_update = 0

// hotplug event 2
kcore_need_update = 1
                                              kcore_need_update = 0

Note that CPU1 set up the RAM kcore entries with the state after hotplug
event 1 but cleared the flag for hotplug event 2.  The RAM entries will
therefore be stale until there is another hotplug event.

This is an extremely unlikely sequence of events, but the fix makes the
synchronization saner, anyways: we serialize the entire update sequence,
which means that whoever clears the flag will always succeed in replacing
the kcore list.

Link: http://lkml.kernel.org/r/6106c509998779730c12400c1b996425df7d7089.1531953780.git.osandov@fb.com
Signed-off-by: Omar Sandoval <osandov@fb.com>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Bhupesh Sharma <bhsharma@redhat.com>
Cc: Eric Biederman <ebiederm@xmission.com>
Cc: James Morse <james.morse@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agoproc/kcore: replace kclist_lock rwlock with rwsem
Omar Sandoval [Wed, 22 Aug 2018 04:54:59 +0000 (21:54 -0700)]
proc/kcore: replace kclist_lock rwlock with rwsem

Now we only need kclist_lock from user context and at fs init time, and
the following changes need to sleep while holding the kclist_lock.

Link: http://lkml.kernel.org/r/521ba449ebe921d905177410fee9222d07882f0d.1531953780.git.osandov@fb.com
Signed-off-by: Omar Sandoval <osandov@fb.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Bhupesh Sharma <bhsharma@redhat.com>
Cc: Eric Biederman <ebiederm@xmission.com>
Cc: James Morse <james.morse@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agoproc/kcore: don't grab lock for memory hotplug notifier
Omar Sandoval [Wed, 22 Aug 2018 04:54:55 +0000 (21:54 -0700)]
proc/kcore: don't grab lock for memory hotplug notifier

The memory hotplug notifier kcore_callback() only needs kclist_lock to
prevent races with __kcore_update_ram(), but we can easily eliminate that
race by using an atomic xchg() in __kcore_update_ram().  This is
preparation for converting kclist_lock to an rwsem.

Link: http://lkml.kernel.org/r/0a4bc89f4dbde8b5b2ea309f7b4fb6a85fe29df2.1531953780.git.osandov@fb.com
Signed-off-by: Omar Sandoval <osandov@fb.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Bhupesh Sharma <bhsharma@redhat.com>
Cc: Eric Biederman <ebiederm@xmission.com>
Cc: James Morse <james.morse@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agoproc/kcore: don't grab lock for kclist_add()
Omar Sandoval [Wed, 22 Aug 2018 04:54:51 +0000 (21:54 -0700)]
proc/kcore: don't grab lock for kclist_add()

Patch series "/proc/kcore improvements", v4.

This series makes a few improvements to /proc/kcore.  It fixes a couple of
small issues in v3 but is otherwise the same.  Patches 1, 2, and 3 are
prep patches.  Patch 4 is a fix/cleanup.  Patch 5 is another prep patch.
Patches 6 and 7 are optimizations to ->read().  Patch 8 makes it possible
to enable CRASH_CORE on any architecture, which is needed for patch 9.
Patch 9 adds vmcoreinfo to /proc/kcore.

This patch (of 9):

kclist_add() is only called at init time, so there's no point in grabbing
any locks.  We're also going to replace the rwlock with a rwsem, which we
don't want to try grabbing during early boot.

While we're here, mark kclist_add() with __init so that we'll get a
warning if it's called from non-init code.

Link: http://lkml.kernel.org/r/98208db1faf167aa8b08eebfa968d95c70527739.1531953780.git.osandov@fb.com
Signed-off-by: Omar Sandoval <osandov@fb.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Bhupesh Sharma <bhsharma@redhat.com>
Tested-by: Bhupesh Sharma <bhsharma@redhat.com>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Bhupesh Sharma <bhsharma@redhat.com>
Cc: Eric Biederman <ebiederm@xmission.com>
Cc: James Morse <james.morse@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agofs/proc/kcore.c: use __pa_symbol() for KCORE_TEXT list entries
James Morse [Wed, 22 Aug 2018 04:54:48 +0000 (21:54 -0700)]
fs/proc/kcore.c: use __pa_symbol() for KCORE_TEXT list entries

elf_kcore_store_hdr() uses __pa() to find the physical address of
KCORE_RAM or KCORE_TEXT entries exported as program headers.

This trips CONFIG_DEBUG_VIRTUAL's checks, as the KCORE_TEXT entries are
not in the linear map.

Handle these two cases separately, using __pa_symbol() for the KCORE_TEXT
entries.

Link: http://lkml.kernel.org/r/20180711131944.15252-1-james.morse@arm.com
Signed-off-by: James Morse <james.morse@arm.com>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Omar Sandoval <osandov@fb.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agofs/proc/vmcore.c: use new typedef vm_fault_t
Souptick Joarder [Wed, 22 Aug 2018 04:54:44 +0000 (21:54 -0700)]
fs/proc/vmcore.c: use new typedef vm_fault_t

Use new return type vm_fault_t for fault handler in struct
vm_operations_struct.  For now, this is just documenting that the function
returns a VM_FAULT value rather than an errno.  Once all instances are
converted, vm_fault_t will become a distinct type.

See 1c8f422059ae ("mm: change return type to vm_fault_t") for reference.

Link: http://lkml.kernel.org/r/20180702153325.GA3875@jordon-HP-15-Notebook-PC
Signed-off-by: Souptick Joarder <jrdr.linux@gmail.com>
Reviewed-by: Matthew Wilcox <mawilcox@microsoft.com>
Cc: Ganesh Goudar <ganeshgr@chelsio.com>
Cc: Rahul Lakkireddy <rahul.lakkireddy@chelsio.com>
Cc: David S. Miller <davem@davemloft.net>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Matthew Wilcox <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agoproc: use "unsigned int" in /proc/stat hook
Alexey Dobriyan [Wed, 22 Aug 2018 04:54:41 +0000 (21:54 -0700)]
proc: use "unsigned int" in /proc/stat hook

Number of CPUs is never high enough to force 64-bit arithmetic.
Save couple of bytes on x86_64.

Link: http://lkml.kernel.org/r/20180627200710.GC18434@avx2
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agoproc: spread "const" a bit
Alexey Dobriyan [Wed, 22 Aug 2018 04:54:37 +0000 (21:54 -0700)]
proc: spread "const" a bit

Link: http://lkml.kernel.org/r/20180627200614.GB18434@avx2
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agoproc: use macro in /proc/latency hook
Alexey Dobriyan [Wed, 22 Aug 2018 04:54:34 +0000 (21:54 -0700)]
proc: use macro in /proc/latency hook

->latency_record is defined as

struct latency_record[LT_SAVECOUNT];

so use the same macro whie iterating.

Link: http://lkml.kernel.org/r/20180627200534.GA18434@avx2
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agoproc: save 2 atomic ops on write to "/proc/*/attr/*"
Alexey Dobriyan [Wed, 22 Aug 2018 04:54:30 +0000 (21:54 -0700)]
proc: save 2 atomic ops on write to "/proc/*/attr/*"

Code checks if write is done by current to its own attributes.
For that get/put pair is unnecessary as it can be done under RCU.

Note: rcu_read_unlock() can be done even earlier since pointer to a task
is not dereferenced. It depends if /proc code should look scary or not:

rcu_read_lock();
task = pid_task(...);
rcu_read_unlock();
if (!task)
return -ESRCH;
if (task != current)
return -EACCESS:

P.S.: rename "length" variable. Code like this

length = -EINVAL;

should not exist.

Link: http://lkml.kernel.org/r/20180627200218.GF18113@avx2
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agoproc: put task earlier in /proc/*/fail-nth
Alexey Dobriyan [Wed, 22 Aug 2018 04:54:27 +0000 (21:54 -0700)]
proc: put task earlier in /proc/*/fail-nth

Link: http://lkml.kernel.org/r/20180627195427.GE18113@avx2
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agoproc: smaller readlock section in readdir("/proc")
Alexey Dobriyan [Wed, 22 Aug 2018 04:54:23 +0000 (21:54 -0700)]
proc: smaller readlock section in readdir("/proc")

Readdir context is thread local, so ->pos is thread local,
move it out of readlock.

Link: http://lkml.kernel.org/r/20180627195339.GD18113@avx2
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agoproc: test /proc/thread-self symlink
Alexey Dobriyan [Wed, 22 Aug 2018 04:54:20 +0000 (21:54 -0700)]
proc: test /proc/thread-self symlink

Same story: I have WIP patch to make it faster, so better have a test
as well.

Link: http://lkml.kernel.org/r/20180627195209.GC18113@avx2
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Shuah Khan <shuahkh@osg.samsung.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agoproc: test /proc/self symlink
Alexey Dobriyan [Wed, 22 Aug 2018 04:54:16 +0000 (21:54 -0700)]
proc: test /proc/self symlink

There are plans to change how /proc/self result is calculated,
for that a test is necessary.

Use direct system call because of this whole getpid caching story.

Link: http://lkml.kernel.org/r/20180627195103.GB18113@avx2
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Shuah Khan <shuahkh@osg.samsung.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agofs/proc/uptime.c: use ktime_get_boottime_ts64
Arnd Bergmann [Wed, 22 Aug 2018 04:54:13 +0000 (21:54 -0700)]
fs/proc/uptime.c: use ktime_get_boottime_ts64

get_monotonic_boottime() is deprecated and uses the old timespec type.
Let's convert /proc/uptime to use ktime_get_boottime_ts64().

Link: http://lkml.kernel.org/r/20180620081746.282742-1-arnd@arndb.de
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Deepa Dinamani <deepa.kernel@gmail.com>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agoproc: fixup PDE allocation bloat
Alexey Dobriyan [Wed, 22 Aug 2018 04:54:09 +0000 (21:54 -0700)]
proc: fixup PDE allocation bloat

24074a35c5c975 ("proc: Make inline name size calculation automatic")
started to put PDE allocations into kmalloc-256 which is unnecessary as
~40 character names are very rare.

Put allocation back into kmalloc-192 cache for 64-bit non-debug builds.

Put BUILD_BUG_ON to know when PDE size has gotten out of control.

[adobriyan@gmail.com: fix BUILD_BUG_ON breakage on powerpc64]
Link: http://lkml.kernel.org/r/20180703191602.GA25521@avx2
Link: http://lkml.kernel.org/r/20180617215732.GA24688@avx2
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Cc: David Howells <dhowells@redhat.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agomm: fix comment for NODEMASK_ALLOC
Oscar Salvador [Wed, 22 Aug 2018 04:54:06 +0000 (21:54 -0700)]
mm: fix comment for NODEMASK_ALLOC

Currently, NODEMASK_ALLOC allocates a nodemask_t with kmalloc when
NODES_SHIFT is higher than 8, otherwise it declares it within the stack.

The comment says that the reasoning behind this, is that nodemask_t will
be 256 bytes when NODES_SHIFT is higher than 8, but this is not true.  For
example, NODES_SHIFT = 9 will give us a 64 bytes nodemask_t.  Let us fix
up the comment for that.

Another thing is that it might make sense to let values lower than
128bytes be allocated in the stack.  Although this all depends on the
depth of the stack (and this changes from function to function), I think
that 64 bytes is something we can easily afford.  So we could even bump
the limit by 1 (from > 8 to > 9).

Link: http://lkml.kernel.org/r/20180820085516.9687-1-osalvador@techadventures.net
Signed-off-by: Oscar Salvador <osalvador@suse.de>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agodrivers/block/zram/zram_drv.c: fix bug storing backing_dev
Peter Kalauskas [Wed, 22 Aug 2018 04:54:02 +0000 (21:54 -0700)]
drivers/block/zram/zram_drv.c: fix bug storing backing_dev

The call to strlcpy in backing_dev_store is incorrect. It should take
the size of the destination buffer instead of the size of the source
buffer.  Additionally, ignore the newline character (\n) when reading
the new file_name buffer. This makes it possible to set the backing_dev
as follows:

echo /dev/sdX > /sys/block/zram0/backing_dev

The reason it worked before was the fact that strlcpy() copies 'len - 1'
bytes, which is strlen(buf) - 1 in our case, so it accidentally didn't
copy the trailing new line symbol.  Which also means that "echo -n
/dev/sdX" most likely was broken.

Signed-off-by: Peter Kalauskas <peskal@google.com>
Link: http://lkml.kernel.org/r/20180813061623.GC64836@rodete-desktop-imager.corp.google.com
Acked-by: Minchan Kim <minchan@kernel.org>
Reviewed-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
Cc: <stable@vger.kernel.org> [4.14+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years ago/proc/meminfo: add percpu populated pages count
Dennis Zhou (Facebook) [Wed, 22 Aug 2018 04:53:58 +0000 (21:53 -0700)]
/proc/meminfo: add percpu populated pages count

Currently, percpu memory only exposes allocation and utilization
information via debugfs.  This more or less is only really useful for
understanding the fragmentation and allocation information at a per-chunk
level with a few global counters.  This is also gated behind a config.
BPF and cgroup, for example, have seen an increase in use causing
increased use of percpu memory.  Let's make it easier for someone to
identify how much memory is being used.

This patch adds the "Percpu" stat to meminfo to more easily look up how
much percpu memory is in use.  This number includes the cost for all
allocated backing pages and not just insight at the per a unit, per chunk
level.  Metadata is excluded.  I think excluding metadata is fair because
the backing memory scales with the numbere of cpus and can quickly
outweigh the metadata.  It also makes this calculation light.

Link: http://lkml.kernel.org/r/20180807184723.74919-1-dennisszhou@gmail.com
Signed-off-by: Dennis Zhou <dennisszhou@gmail.com>
Acked-by: Tejun Heo <tj@kernel.org>
Acked-by: Roman Gushchin <guro@fb.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: David Rientjes <rientjes@google.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Christoph Lameter <cl@linux.com>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agomm, oom: introduce memory.oom.group
Roman Gushchin [Wed, 22 Aug 2018 04:53:54 +0000 (21:53 -0700)]
mm, oom: introduce memory.oom.group

For some workloads an intervention from the OOM killer can be painful.
Killing a random task can bring the workload into an inconsistent state.

Historically, there are two common solutions for this
problem:
1) enabling panic_on_oom,
2) using a userspace daemon to monitor OOMs and kill
   all outstanding processes.

Both approaches have their downsides: rebooting on each OOM is an obvious
waste of capacity, and handling all in userspace is tricky and requires a
userspace agent, which will monitor all cgroups for OOMs.

In most cases an in-kernel after-OOM cleaning-up mechanism can eliminate
the necessity of enabling panic_on_oom.  Also, it can simplify the cgroup
management for userspace applications.

This commit introduces a new knob for cgroup v2 memory controller:
memory.oom.group.  The knob determines whether the cgroup should be
treated as an indivisible workload by the OOM killer.  If set, all tasks
belonging to the cgroup or to its descendants (if the memory cgroup is not
a leaf cgroup) are killed together or not at all.

To determine which cgroup has to be killed, we do traverse the cgroup
hierarchy from the victim task's cgroup up to the OOMing cgroup (or root)
and looking for the highest-level cgroup with memory.oom.group set.

Tasks with the OOM protection (oom_score_adj set to -1000) are treated as
an exception and are never killed.

This patch doesn't change the OOM victim selection algorithm.

Link: http://lkml.kernel.org/r/20180802003201.817-4-guro@fb.com
Signed-off-by: Roman Gushchin <guro@fb.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: David Rientjes <rientjes@google.com>
Cc: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Cc: Tejun Heo <tj@kernel.org>
Cc: Vladimir Davydov <vdavydov.dev@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agomm, oom: refactor oom_kill_process()
Roman Gushchin [Wed, 22 Aug 2018 04:53:50 +0000 (21:53 -0700)]
mm, oom: refactor oom_kill_process()

Patch series "introduce memory.oom.group", v2.

This is a tiny implementation of cgroup-aware OOM killer, which adds an
ability to kill a cgroup as a single unit and so guarantee the integrity
of the workload.

Although it has only a limited functionality in comparison to what now
resides in the mm tree (it doesn't change the victim task selection
algorithm, doesn't look at memory stas on cgroup level, etc), it's also
much simpler and more straightforward.  So, hopefully, we can avoid having
long debates here, as we had with the full implementation.

As it doesn't prevent any futher development, and implements an useful and
complete feature, it looks as a sane way forward.

This patch (of 2):

oom_kill_process() consists of two logical parts: the first one is
responsible for considering task's children as a potential victim and
printing the debug information.  The second half is responsible for
sending SIGKILL to all tasks sharing the mm struct with the given victim.

This commit splits oom_kill_process() with an intention to re-use the the
second half: __oom_kill_process().

The cgroup-aware OOM killer will kill multiple tasks belonging to the
victim cgroup.  We don't need to print the debug information for the each
task, as well as play with task selection (considering task's children),
so we can't use the existing oom_kill_process().

Link: http://lkml.kernel.org/r/20171130152824.1591-2-guro@fb.com
Link: http://lkml.kernel.org/r/20180802003201.817-3-guro@fb.com
Signed-off-by: Roman Gushchin <guro@fb.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: David Rientjes <rientjes@google.com>
Cc: Vladimir Davydov <vdavydov.dev@gmail.com>
Cc: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Cc: David Rientjes <rientjes@google.com>
Cc: Tejun Heo <tj@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agotools/testing/selftests/vm/: add MAP_POPULATE test
Dmitry Safonov [Wed, 22 Aug 2018 04:53:47 +0000 (21:53 -0700)]
tools/testing/selftests/vm/: add MAP_POPULATE test

As with many other projects, we use some shmalloc allocator.  At some
point we need to make a part of allocated pages back private to process.
And it should be populated straight away.  Check that (MAP_PRIVATE |
MAP_POPULATE) actually copies the private page.

[akpm@linux-foundation.org: change message, per review discussion]
Link: http://lkml.kernel.org/r/20180801233636.29354-1-dima@arista.com
Signed-off-by: Dmitry Safonov <dima@arista.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Dmitry Safonov <0x7f454c46@gmail.com>
Cc: Hua Zhong <hzhong@arista.com>
Cc: Shuah Khan <shuah@kernel.org>
Cc: Stuart Ritchie <sritchie@arista.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agomm/page_alloc: Introduce free_area_init_core_hotplug
Oscar Salvador [Wed, 22 Aug 2018 04:53:43 +0000 (21:53 -0700)]
mm/page_alloc: Introduce free_area_init_core_hotplug

Currently, whenever a new node is created/re-used from the memhotplug
path, we call free_area_init_node()->free_area_init_core().  But there is
some code that we do not really need to run when we are coming from such
path.

free_area_init_core() performs the following actions:

1) Initializes pgdat internals, such as spinlock, waitqueues and more.
2) Account # nr_all_pages and # nr_kernel_pages. These values are used later on
   when creating hash tables.
3) Account number of managed_pages per zone, substracting dma_reserved and
   memmap pages.
4) Initializes some fields of the zone structure data
5) Calls init_currently_empty_zone to initialize all the freelists
6) Calls memmap_init to initialize all pages belonging to certain zone

When called from memhotplug path, free_area_init_core() only performs
actions #1 and #4.

Action #2 is pointless as the zones do not have any pages since either the
node was freed, or we are re-using it, eitherway all zones belonging to
this node should have 0 pages.  For the same reason, action #3 results
always in manages_pages being 0.

Action #5 and #6 are performed later on when onlining the pages:
 online_pages()->move_pfn_range_to_zone()->init_currently_empty_zone()
 online_pages()->move_pfn_range_to_zone()->memmap_init_zone()

This patch does two things:

First, moves the node/zone initializtion to their own function, so it
allows us to create a small version of free_area_init_core, where we only
perform:

1) Initialization of pgdat internals, such as spinlock, waitqueues and more
4) Initialization of some fields of the zone structure data

These two functions are: pgdat_init_internals() and zone_init_internals().

The second thing this patch does, is to introduce
free_area_init_core_hotplug(), the memhotplug version of
free_area_init_core():

Currently, we call free_area_init_node() from the memhotplug path.  In
there, we set some pgdat's fields, and call calculate_node_totalpages().
calculate_node_totalpages() calculates the # of pages the node has.

Since the node is either new, or we are re-using it, the zones belonging
to this node should not have any pages, so there is no point to calculate
this now.

Actually, we re-set these values to 0 later on with the calls to:

reset_node_managed_pages()
reset_node_present_pages()

The # of pages per node and the # of pages per zone will be calculated when
onlining the pages:

online_pages()->move_pfn_range()->move_pfn_range_to_zone()->resize_zone_range()
online_pages()->move_pfn_range()->move_pfn_range_to_zone()->resize_pgdat_range()

Also, since free_area_init_core/free_area_init_node will now only get called during early init, let us replace
__paginginit with __init, so their code gets freed up.

[osalvador@techadventures.net: fix section usage]
Link: http://lkml.kernel.org/r/20180731101752.GA473@techadventures.net
[osalvador@suse.de: v6]
Link: http://lkml.kernel.org/r/20180801122348.21588-6-osalvador@techadventures.net
Link: http://lkml.kernel.org/r/20180730101757.28058-5-osalvador@techadventures.net
Signed-off-by: Oscar Salvador <osalvador@suse.de>
Reviewed-by: Pavel Tatashin <pasha.tatashin@oracle.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Pasha Tatashin <Pavel.Tatashin@microsoft.com>
Cc: Aaron Lu <aaron.lu@intel.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agomm/page_alloc: inline function to handle CONFIG_DEFERRED_STRUCT_PAGE_INIT
Oscar Salvador [Wed, 22 Aug 2018 04:53:39 +0000 (21:53 -0700)]
mm/page_alloc: inline function to handle CONFIG_DEFERRED_STRUCT_PAGE_INIT

Let us move the code between CONFIG_DEFERRED_STRUCT_PAGE_INIT to an inline
function.  Not having an ifdef in the function makes the code more
readable.

Link: http://lkml.kernel.org/r/20180730101757.28058-4-osalvador@techadventures.net
Signed-off-by: Oscar Salvador <osalvador@suse.de>
Acked-by: Michal Hocko <mhocko@suse.com>
Reviewed-by: Pavel Tatashin <pasha.tatashin@oracle.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Aaron Lu <aaron.lu@intel.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Pasha Tatashin <Pavel.Tatashin@microsoft.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agomm: remove __paginginit
Pavel Tatashin [Wed, 22 Aug 2018 04:53:36 +0000 (21:53 -0700)]
mm: remove __paginginit

__paginginit is the same thing as __meminit except for platforms without
sparsemem, there it is defined as __init.

Remove __paginginit and use __meminit.  Use __ref in one single function
that merges __meminit and __init sections: setup_usemap().

Link: http://lkml.kernel.org/r/20180801122348.21588-4-osalvador@techadventures.net
Signed-off-by: Pavel Tatashin <pasha.tatashin@oracle.com>
Signed-off-by: Oscar Salvador <osalvador@suse.de>
Reviewed-by: Oscar Salvador <osalvador@suse.de>
Cc: Pasha Tatashin <Pavel.Tatashin@microsoft.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agomm: access zone->node via zone_to_nid() and zone_set_nid()
Pavel Tatashin [Wed, 22 Aug 2018 04:53:32 +0000 (21:53 -0700)]
mm: access zone->node via zone_to_nid() and zone_set_nid()

zone->node is configured only when CONFIG_NUMA=y, so it is a good idea to
have inline functions to access this field in order to avoid ifdef's in c
files.

Link: http://lkml.kernel.org/r/20180730101757.28058-3-osalvador@techadventures.net
Signed-off-by: Pavel Tatashin <pasha.tatashin@oracle.com>
Signed-off-by: Oscar Salvador <osalvador@suse.de>
Reviewed-by: Oscar Salvador <osalvador@suse.de>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Aaron Lu <aaron.lu@intel.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Pasha Tatashin <Pavel.Tatashin@microsoft.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agomm/page_alloc.c: move ifdefery out of free_area_init_core
Oscar Salvador [Wed, 22 Aug 2018 04:53:29 +0000 (21:53 -0700)]
mm/page_alloc.c: move ifdefery out of free_area_init_core

Patch series "Refactor free_area_init_core and add
free_area_init_core_hotplug", v6.

This patchset does three things:

 1) Clean up/refactor free_area_init_core/free_area_init_node
    by moving the ifdefery out of the functions.
 2) Move the pgdat/zone initialization in free_area_init_core to its
    own function.
 3) Introduce free_area_init_core_hotplug, a small subset of
    free_area_init_core, which is only called from memhotlug code path. In this
    way, we have:

    free_area_init_core: called during early initialization
    free_area_init_core_hotplug: called whenever a new node is allocated/re-used (memhotplug path)

This patch (of 5):

Moving the #ifdefs out of the function makes it easier to follow.

Link: http://lkml.kernel.org/r/20180730101757.28058-2-osalvador@techadventures.net
Signed-off-by: Oscar Salvador <osalvador@suse.de>
Acked-by: Michal Hocko <mhocko@suse.com>
Reviewed-by: Pavel Tatashin <pasha.tatashin@oracle.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Pasha Tatashin <Pavel.Tatashin@microsoft.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Aaron Lu <aaron.lu@intel.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: David Hildenbrand <david@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agomm: remove zone_id() and make use of zone_idx() in is_dev_zone()
Oscar Salvador [Wed, 22 Aug 2018 04:53:24 +0000 (21:53 -0700)]
mm: remove zone_id() and make use of zone_idx() in is_dev_zone()

is_dev_zone() is using zone_id() to check if the zone is ZONE_DEVICE.
zone_id() looks pretty much the same as zone_idx(), and while the use of
zone_idx() is quite spread in the kernel, zone_id() is only being used by
is_dev_zone().

This patch removes zone_id() and makes is_dev_zone() use zone_idx() to
check the zone, so we do not have two things with the same functionality
around.

Link: http://lkml.kernel.org/r/20180730133718.28683-1-osalvador@techadventures.net
Signed-off-by: Oscar Salvador <osalvador@suse.de>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Reviewed-by: Pavel Tatashin <pasha.tatashin@oracle.com>
Cc: Pasha Tatashin <Pavel.Tatashin@microsoft.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agoDocumentation/sysctl/vm.txt: update __vm_enough_memory()'s path
juviliu [Wed, 22 Aug 2018 04:53:20 +0000 (21:53 -0700)]
Documentation/sysctl/vm.txt: update __vm_enough_memory()'s path

__vm_enough_memory has moved to mm/util.c.

Link: http://lkml.kernel.org/r/E18EDF4A4FA4A04BBFA824B6D7699E532A7E5913@EXMBX-SZMAIL013.tencent.com
Signed-off-by: Juvi Liu <juviliu@tencent.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agomemcg: reduce memcg tree traversals for stats collection
Shakeel Butt [Wed, 22 Aug 2018 04:53:17 +0000 (21:53 -0700)]
memcg: reduce memcg tree traversals for stats collection

Currently cgroup-v1's memcg_stat_show traverses the memcg tree ~17 times
to collect the stats while cgroup-v2's memory_stat_show traverses the
memcg tree thrice.  On a large machine, a couple thousand memcgs is very
normal and if the churn is high and memcgs stick around during to several
reasons, tens of thousands of nodes in memcg tree can exist.  This patch
has refactored and shared the stat collection code between cgroup-v1 and
cgroup-v2 and has reduced the tree traversal to just one.

I ran a simple benchmark which reads the root_mem_cgroup's stat file
1000 times in the presense of 2500 memcgs on cgroup-v1. The results are:

Without the patch:
$ time ./read-root-stat-1000-times

real    0m1.663s
user    0m0.000s
sys     0m1.660s

With the patch:
$ time ./read-root-stat-1000-times

real    0m0.468s
user    0m0.000s
sys     0m0.467s

Link: http://lkml.kernel.org/r/20180724224635.143944-1-shakeelb@google.com
Signed-off-by: Shakeel Butt <shakeelb@google.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Vladimir Davydov <vdavydov.dev@gmail.com>
Cc: Greg Thelen <gthelen@google.com>
Cc: Bruce Merry <bmerry@ska.ac.za>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agomm: fix page_freeze_refs and page_unfreeze_refs in comments
Jiang Biao [Wed, 22 Aug 2018 04:53:13 +0000 (21:53 -0700)]
mm: fix page_freeze_refs and page_unfreeze_refs in comments

page_freeze_refs/page_unfreeze_refs have already been relplaced by
page_ref_freeze/page_ref_unfreeze , but they are not modified in the
comments.

Link: http://lkml.kernel.org/r/1532590226-106038-1-git-send-email-jiang.biao2@zte.com.cn
Signed-off-by: Jiang Biao <jiang.biao2@zte.com.cn>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agomm: clarify CONFIG_PAGE_POISONING and usage
Kees Cook [Wed, 22 Aug 2018 04:53:10 +0000 (21:53 -0700)]
mm: clarify CONFIG_PAGE_POISONING and usage

The Kconfig text for CONFIG_PAGE_POISONING doesn't mention that it has to
be enabled explicitly.  This updates the documentation for that and adds a
note about CONFIG_PAGE_POISONING to the "page_poison" command line docs.
While here, change description of CONFIG_PAGE_POISONING_ZERO too, as it's
not "random" data, but rather the fixed debugging value that would be used
when not zeroing.  Additionally removes a stray "bool" in the Kconfig.

Link: http://lkml.kernel.org/r/20180725223832.GA43733@beast
Signed-off-by: Kees Cook <keescook@chromium.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Laura Abbott <labbott@redhat.com>
Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agomm: zero out the vma in vma_init()
Andrew Morton [Wed, 22 Aug 2018 04:53:06 +0000 (21:53 -0700)]
mm: zero out the vma in vma_init()

Rather than in vm_area_alloc().  To ensure that the various oddball
stack-based vmas are in a good state.  Some of the callers were zeroing
them out, others were not.

Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Russell King <rmk+kernel@arm.linux.org.uk>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agomm/mempool.c: add missing parameter description
Mike Rapoport [Wed, 22 Aug 2018 04:53:03 +0000 (21:53 -0700)]
mm/mempool.c: add missing parameter description

The kernel-doc for mempool_init function is missing the description of the
pool parameter.  Add it.

Link: http://lkml.kernel.org/r/1532336274-26228-1-git-send-email-rppt@linux.vnet.ibm.com
Signed-off-by: Mike Rapoport <rppt@linux.vnet.ibm.com>
Acked-by: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agomm: /proc/pid/smaps_rollup: convert to single value seq_file
Vlastimil Babka [Wed, 22 Aug 2018 04:52:59 +0000 (21:52 -0700)]
mm: /proc/pid/smaps_rollup: convert to single value seq_file

The /proc/pid/smaps_rollup file is currently implemented via the
m_start/m_next/m_stop seq_file iterators shared with the other maps files,
that iterate over vma's.  However, the rollup file doesn't print anything
for each vma, only accumulate the stats.

There are some issues with the current code as reported in [1] - the
accumulated stats can get skewed if seq_file start()/stop() op is called
multiple times, if show() is called multiple times, and after seeks to
non-zero position.

Patch [1] fixed those within existing design, but I believe it is
fundamentally wrong to expose the vma iterators to the seq_file mechanism
when smaps_rollup shows logically a single set of values for the whole
address space.

This patch thus refactors the code to provide a single "value" at offset
0, with vma iteration to gather the stats done internally.  This fixes the
situations where results are skewed, and simplifies the code, especially
in show_smap(), at the expense of somewhat less code reuse.

[1] https://marc.info/?l=linux-mm&m=151927723128134&w=2

[vbabka@suse.c: use seq_file infrastructure]
Link: http://lkml.kernel.org/r/bf4525b0-fd5b-4c4c-2cb3-adee3dd95a48@suse.cz
Link: http://lkml.kernel.org/r/20180723111933.15443-5-vbabka@suse.cz
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Reported-by: Daniel Colascione <dancol@google.com>
Reviewed-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agomm: /proc/pid/smaps: factor out common stats printing
Vlastimil Babka [Wed, 22 Aug 2018 04:52:56 +0000 (21:52 -0700)]
mm: /proc/pid/smaps: factor out common stats printing

To prepare for handling /proc/pid/smaps_rollup differently from
/proc/pid/smaps factor out from show_smap() printing the parts of output
that are common for both variants, which is the bulk of the gathered
memory stats.

[vbabka@suse.cz: add const, per Alexey]
Link: http://lkml.kernel.org/r/b45f319f-cd04-337b-37f8-77f99786aa8a@suse.cz
Link: http://lkml.kernel.org/r/20180723111933.15443-4-vbabka@suse.cz
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Reviewed-by: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Daniel Colascione <dancol@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agomm: /proc/pid/smaps: factor out mem stats gathering
Vlastimil Babka [Wed, 22 Aug 2018 04:52:52 +0000 (21:52 -0700)]
mm: /proc/pid/smaps: factor out mem stats gathering

To prepare for handling /proc/pid/smaps_rollup differently from
/proc/pid/smaps factor out vma mem stats gathering from show_smap() - it
will be used by both.

Link: http://lkml.kernel.org/r/20180723111933.15443-3-vbabka@suse.cz
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Reviewed-by: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Daniel Colascione <dancol@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agomm: /proc/pid/*maps remove is_pid and related wrappers
Vlastimil Babka [Wed, 22 Aug 2018 04:52:48 +0000 (21:52 -0700)]
mm: /proc/pid/*maps remove is_pid and related wrappers

Patch series "cleanups and refactor of /proc/pid/smaps*".

The recent regression in /proc/pid/smaps made me look more into the code.
Especially the issues with smaps_rollup reported in [1] as explained in
Patch 4, which fixes them by refactoring the code.  Patches 2 and 3 are
preparations for that.  Patch 1 is me realizing that there's a lot of
boilerplate left from times where we tried (unsuccessfuly) to mark thread
stacks in the output.

Originally I had also plans to rework the translation from
/proc/pid/*maps* file offsets to the internal structures.  Now the offset
means "vma number", which is not really stable (vma's can come and go
between read() calls) and there's an extra caching of last vma's address.
My idea was that offsets would be interpreted directly as addresses, which
would also allow meaningful seeks (see the ugly seek_to_smaps_entry() in
tools/testing/selftests/vm/mlock2.h).  However loff_t is (signed) long
long so that might be insufficient somewhere for the unsigned long
addresses.

So the result is fixed issues with skewed /proc/pid/smaps_rollup results,
simpler smaps code, and a lot of unused code removed.

[1] https://marc.info/?l=linux-mm&m=151927723128134&w=2

This patch (of 4):

Commit b76437579d13 ("procfs: mark thread stack correctly in
proc/<pid>/maps") introduced differences between /proc/PID/maps and
/proc/PID/task/TID/maps to mark thread stacks properly, and this was
also done for smaps and numa_maps.  However it didn't work properly and
was ultimately removed by commit b18cb64ead40 ("fs/proc: Stop trying to
report thread stacks").

Now the is_pid parameter for the related show_*() functions is unused
and we can remove it together with wrapper functions and ops structures
that differ for PID and TID cases only in this parameter.

Link: http://lkml.kernel.org/r/20180723111933.15443-2-vbabka@suse.cz
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Reviewed-by: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Daniel Colascione <dancol@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agomm/oom_kill.c: clean up oom_reap_task_mm()
Michal Hocko [Wed, 22 Aug 2018 04:52:45 +0000 (21:52 -0700)]
mm/oom_kill.c: clean up oom_reap_task_mm()

Andrew has noticed some inconsistencies in oom_reap_task_mm.  Notably

 - Undocumented return value.

 - comment "failed to reap part..." is misleading - sounds like it's
   referring to something which happened in the past, is in fact
   referring to something which might happen in the future.

 - fails to call trace_finish_task_reaping() in one case

 - code duplication.

 - Increases mmap_sem hold time a little by moving
   trace_finish_task_reaping() inside the locked region.  So sue me ;)

 - Sharing the finish: path means that the trace event won't
   distinguish between the two sources of finishing.

Add a short explanation for the return value and fix the rest by
reorganizing the function a bit to have unified function exit paths.

Link: http://lkml.kernel.org/r/20180724141747.GP28386@dhcp22.suse.cz
Suggested-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Michal Hocko <mhocko@suse.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Tetsuo Handa <penguin-kernel@i-love.sakura.ne.jp>
Cc: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agomm, oom: describe task memory unit, larger PID pad
Rodrigo Freire [Wed, 22 Aug 2018 04:52:41 +0000 (21:52 -0700)]
mm, oom: describe task memory unit, larger PID pad

The default page memory unit of OOM task dump events might not be
intuitive and potentially misleading for the non-initiated when debugging
OOM events: These are pages and not kBs.  Add a small printk prior to the
task dump informing that the memory units are actually memory _pages_.

Also extends PID field to align on up to 7 characters.
Reference https://lkml.org/lkml/2018/7/3/1201

Link: http://lkml.kernel.org/r/c795eb5129149ed8a6345c273aba167ff1bbd388.1530715938.git.rfreire@redhat.com
Signed-off-by: Rodrigo Freire <rfreire@redhat.com>
Acked-by: David Rientjes <rientjes@google.com>
Acked-by: Rafael Aquini <aquini@redhat.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Tetsuo Handa <penguin-kernel@i-love.sakura.ne.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agomm, oom: remove oom_lock from oom_reaper
Michal Hocko [Wed, 22 Aug 2018 04:52:37 +0000 (21:52 -0700)]
mm, oom: remove oom_lock from oom_reaper

oom_reaper used to rely on the oom_lock since e2fe14564d33 ("oom_reaper:
close race with exiting task").  We do not really need the lock anymore
though.  212925802454 ("mm: oom: let oom_reap_task and exit_mmap run
concurrently") has removed serialization with the exit path based on the
mm reference count and so we do not really rely on the oom_lock anymore.

Tetsuo was arguing that at least MMF_OOM_SKIP should be set under the lock
to prevent from races when the page allocator didn't manage to get the
freed (reaped) memory in __alloc_pages_may_oom but it sees the flag later
on and move on to another victim.  Although this is possible in principle
let's wait for it to actually happen in real life before we make the
locking more complex again.

Therefore remove the oom_lock for oom_reaper paths (both exit_mmap and
oom_reap_task_mm).  The reaper serializes with exit_mmap by mmap_sem +
MMF_OOM_SKIP flag.  There is no synchronization with out_of_memory path
now.

[mhocko@kernel.org: oom_reap_task_mm should return false when __oom_reap_task_mm did]
Link: http://lkml.kernel.org/r/20180724141747.GP28386@dhcp22.suse.cz
Link: http://lkml.kernel.org/r/20180719075922.13784-1-mhocko@kernel.org
Signed-off-by: Michal Hocko <mhocko@suse.com>
Suggested-by: David Rientjes <rientjes@google.com>
Acked-by: David Rientjes <rientjes@google.com>
Cc: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agomm, oom: distinguish blockable mode for mmu notifiers
Michal Hocko [Wed, 22 Aug 2018 04:52:33 +0000 (21:52 -0700)]
mm, oom: distinguish blockable mode for mmu notifiers

There are several blockable mmu notifiers which might sleep in
mmu_notifier_invalidate_range_start and that is a problem for the
oom_reaper because it needs to guarantee a forward progress so it cannot
depend on any sleepable locks.

Currently we simply back off and mark an oom victim with blockable mmu
notifiers as done after a short sleep.  That can result in selecting a new
oom victim prematurely because the previous one still hasn't torn its
memory down yet.

We can do much better though.  Even if mmu notifiers use sleepable locks
there is no reason to automatically assume those locks are held.  Moreover
majority of notifiers only care about a portion of the address space and
there is absolutely zero reason to fail when we are unmapping an unrelated
range.  Many notifiers do really block and wait for HW which is harder to
handle and we have to bail out though.

This patch handles the low hanging fruit.
__mmu_notifier_invalidate_range_start gets a blockable flag and callbacks
are not allowed to sleep if the flag is set to false.  This is achieved by
using trylock instead of the sleepable lock for most callbacks and
continue as long as we do not block down the call chain.

I think we can improve that even further because there is a common pattern
to do a range lookup first and then do something about that.  The first
part can be done without a sleeping lock in most cases AFAICS.

The oom_reaper end then simply retries if there is at least one notifier
which couldn't make any progress in !blockable mode.  A retry loop is
already implemented to wait for the mmap_sem and this is basically the
same thing.

The simplest way for driver developers to test this code path is to wrap
userspace code which uses these notifiers into a memcg and set the hard
limit to hit the oom.  This can be done e.g.  after the test faults in all
the mmu notifier managed memory and set the hard limit to something really
small.  Then we are looking for a proper process tear down.

[akpm@linux-foundation.org: coding style fixes]
[akpm@linux-foundation.org: minor code simplification]
Link: http://lkml.kernel.org/r/20180716115058.5559-1-mhocko@kernel.org
Signed-off-by: Michal Hocko <mhocko@suse.com>
Acked-by: Christian König <christian.koenig@amd.com> # AMD notifiers
Acked-by: Leon Romanovsky <leonro@mellanox.com> # mlx and umem_odp
Reported-by: David Rientjes <rientjes@google.com>
Cc: "David (ChunMing) Zhou" <David1.Zhou@amd.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: David Airlie <airlied@linux.ie>
Cc: Jani Nikula <jani.nikula@linux.intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Doug Ledford <dledford@redhat.com>
Cc: Jason Gunthorpe <jgg@ziepe.ca>
Cc: Mike Marciniszyn <mike.marciniszyn@intel.com>
Cc: Dennis Dalessandro <dennis.dalessandro@intel.com>
Cc: Sudeep Dutt <sudeep.dutt@intel.com>
Cc: Ashutosh Dixit <ashutosh.dixit@intel.com>
Cc: Dimitri Sivanich <sivanich@sgi.com>
Cc: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Cc: Juergen Gross <jgross@suse.com>
Cc: "Jérôme Glisse" <jglisse@redhat.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Felix Kuehling <felix.kuehling@amd.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agomm/swapfile.c: put_swap_page: share more between huge/normal code path
Huang Ying [Wed, 22 Aug 2018 04:52:29 +0000 (21:52 -0700)]
mm/swapfile.c: put_swap_page: share more between huge/normal code path

In this patch, locking related code is shared between huge/normal code
path in put_swap_page() to reduce code duplication. The `free_entries == 0`
case is merged into the more general `free_entries != SWAPFILE_CLUSTER`
case, because the new locking method makes it easy.

The added lines is same as the removed lines.  But the code size is
increased when CONFIG_TRANSPARENT_HUGEPAGE=n.

text    data     bss     dec     hex filename
base:        24123    2004     340   26467    6763 mm/swapfile.o
unified:       24485    2004     340   26829    68cd mm/swapfile.o

Dig on step deeper with `size -A mm/swapfile.o` for base and unified
kernel and compare the result, yields,

  -.text                                17723      0
  +.text                                17835      0
  -.orc_unwind_ip                        1380      0
  +.orc_unwind_ip                        1480      0
  -.orc_unwind                           2070      0
  +.orc_unwind                           2220      0
  -Total                                26686
  +Total                                27048

The total difference is the same.  The text segment difference is much
smaller: 112.  More difference comes from the ORC unwinder segments:
(1480 + 2220) - (1380 + 2070) = 250.  If the frame pointer unwinder is
used, this costs nothing.

Link: http://lkml.kernel.org/r/20180720071845.17920-9-ying.huang@intel.com
Signed-off-by: "Huang, Ying" <ying.huang@intel.com>
Reviewed-by: Daniel Jordan <daniel.m.jordan@oracle.com>
Acked-by: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Shaohua Li <shli@kernel.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@redhat.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Matthew Wilcox <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agomm/swapfile.c: add __swap_entry_free_locked()
Huang Ying [Wed, 22 Aug 2018 04:52:24 +0000 (21:52 -0700)]
mm/swapfile.c: add __swap_entry_free_locked()

The part of __swap_entry_free() with lock held is separated into a new
function __swap_entry_free_locked().  Because we want to reuse that
piece of code in some other places.

Just mechanical code refactoring, there is no any functional change in
this function.

Link: http://lkml.kernel.org/r/20180720071845.17920-8-ying.huang@intel.com
Signed-off-by: "Huang, Ying" <ying.huang@intel.com>
Reviewed-by: Daniel Jordan <daniel.m.jordan@oracle.com>
Acked-by: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Shaohua Li <shli@kernel.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@redhat.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Matthew Wilcox <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agomm, swap, get_swap_pages: use entry_size instead of cluster in parameter
Huang Ying [Wed, 22 Aug 2018 04:52:20 +0000 (21:52 -0700)]
mm, swap, get_swap_pages: use entry_size instead of cluster in parameter

As suggested by Matthew Wilcox, it is better to use "int entry_size"
instead of "bool cluster" as parameter to specify whether to operate for
huge or normal swap entries.  Because this improve the flexibility to
support other swap entry size.  And Dave Hansen thinks that this
improves code readability too.

So in this patch, the "bool cluster" parameter of get_swap_pages() is
replaced by "int entry_size".

And nr_swap_entries() trick is used to reduce the binary size when
!CONFIG_TRANSPARENT_HUGE_PAGE.

       text    data     bss     dec     hex filename
base  24215    2028     340   26583    67d7 mm/swapfile.o
head  24123    2004     340   26467    6763 mm/swapfile.o

Link: http://lkml.kernel.org/r/20180720071845.17920-7-ying.huang@intel.com
Signed-off-by: "Huang, Ying" <ying.huang@intel.com>
Suggested-by: Matthew Wilcox <willy@infradead.org>
Acked-by: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Daniel Jordan <daniel.m.jordan@oracle.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Shaohua Li <shli@kernel.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@redhat.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agomm/swapfile.c: unify normal/huge code path in put_swap_page()
Huang Ying [Wed, 22 Aug 2018 04:52:17 +0000 (21:52 -0700)]
mm/swapfile.c: unify normal/huge code path in put_swap_page()

In this patch, the normal/huge code path in put_swap_page() and several
helper functions are unified to avoid duplicated code, bugs, etc.  and
make it easier to review the code.

The removed lines are more than added lines.  And the binary size is
kept exactly same when CONFIG_TRANSPARENT_HUGEPAGE=n.

Link: http://lkml.kernel.org/r/20180720071845.17920-6-ying.huang@intel.com
Signed-off-by: "Huang, Ying" <ying.huang@intel.com>
Suggested-by: Dave Hansen <dave.hansen@linux.intel.com>
Acked-by: Dave Hansen <dave.hansen@linux.intel.com>
Reviewed-by: Daniel Jordan <daniel.m.jordan@oracle.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Shaohua Li <shli@kernel.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@redhat.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Matthew Wilcox <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agomm/swapfile.c: unify normal/huge code path in swap_page_trans_huge_swapped()
Huang Ying [Wed, 22 Aug 2018 04:52:13 +0000 (21:52 -0700)]
mm/swapfile.c: unify normal/huge code path in swap_page_trans_huge_swapped()

As suggested by Dave, we should unify the code path for normal and huge
swap support if possible to avoid duplicated code, bugs, etc.  and make
it easier to review code.

In this patch, the normal/huge code path in
swap_page_trans_huge_swapped() is unified, the added and removed lines
are same.  And the binary size is kept almost same when
CONFIG_TRANSPARENT_HUGEPAGE=n.

 text    data     bss     dec     hex filename
base: 24179    2028     340   26547    67b3 mm/swapfile.o
unified: 24215    2028     340   26583    67d7 mm/swapfile.o

Link: http://lkml.kernel.org/r/20180720071845.17920-5-ying.huang@intel.com
Signed-off-by: "Huang, Ying" <ying.huang@intel.com>
Suggested-and-acked-by: Dave Hansen <dave.hansen@linux.intel.com>
Reviewed-by: Daniel Jordan <daniel.m.jordan@oracle.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Shaohua Li <shli@kernel.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@redhat.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Matthew Wilcox <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agomm/swapfile.c: use swap_count() in swap_page_trans_huge_swapped()
Huang Ying [Wed, 22 Aug 2018 04:52:09 +0000 (21:52 -0700)]
mm/swapfile.c: use swap_count() in swap_page_trans_huge_swapped()

In swap_page_trans_huge_swapped(), to identify whether there's any page
table mapping for a 4k sized swap entry, "si->swap_map[i] !=
SWAP_HAS_CACHE" is used.  This works correctly now, because all users of
the function will only call it after checking SWAP_HAS_CACHE.  But as
pointed out by Daniel, it is better to use "swap_count(map[i])" here,
because it works for "map[i] == 0" case too.

And this makes the implementation more consistent between normal and
huge swap entry.

Link: http://lkml.kernel.org/r/20180720071845.17920-4-ying.huang@intel.com
Signed-off-by: "Huang, Ying" <ying.huang@intel.com>
Suggested-and-reviewed-by: Daniel Jordan <daniel.m.jordan@oracle.com>
Acked-by: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Shaohua Li <shli@kernel.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@redhat.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Matthew Wilcox <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agomm/swapfile.c: replace some #ifdef with IS_ENABLED()
Huang Ying [Wed, 22 Aug 2018 04:52:05 +0000 (21:52 -0700)]
mm/swapfile.c: replace some #ifdef with IS_ENABLED()

In mm/swapfile.c, THP (Transparent Huge Page) swap specific code is
enclosed by #ifdef CONFIG_THP_SWAP/#endif to avoid code dilating when
THP isn't enabled.  But #ifdef/#endif in .c file hurt the code
readability, so Dave suggested to use IS_ENABLED(CONFIG_THP_SWAP)
instead and let compiler to do the dirty job for us.  This has potential
to remove some duplicated code too.  From output of `size`,

text    data     bss     dec     hex filename
THP=y:         26269    2076     340   28685    700d mm/swapfile.o
ifdef/endif:   24115    2028     340   26483    6773 mm/swapfile.o
IS_ENABLED:    24179    2028     340   26547    67b3 mm/swapfile.o

IS_ENABLED() based solution works quite well, almost as good as that of
#ifdef/#endif.  And from the diffstat, the removed lines are more than
added lines.

One #ifdef for split_swap_cluster() is kept.  Because it is a public
function with a stub implementation for CONFIG_THP_SWAP=n in swap.h.

Link: http://lkml.kernel.org/r/20180720071845.17920-3-ying.huang@intel.com
Signed-off-by: "Huang, Ying" <ying.huang@intel.com>
Suggested-and-acked-by: Dave Hansen <dave.hansen@linux.intel.com>
Reviewed-by: Daniel Jordan <daniel.m.jordan@oracle.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Shaohua Li <shli@kernel.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@redhat.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Matthew Wilcox <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agomm: swap: add comments to lock_cluster_or_swap_info()
Huang Ying [Wed, 22 Aug 2018 04:52:01 +0000 (21:52 -0700)]
mm: swap: add comments to lock_cluster_or_swap_info()

Patch series "swap: THP optimizing refactoring", v4.

Now the THP (Transparent Huge Page) swap optimizing is implemented in the
way like below,

  #ifdef CONFIG_THP_SWAP
  huge_function(...)
  {
  }
  #else
  normal_function(...)
  {
  }
  #endif

  general_function(...)
  {
   if (huge)
   return thp_function(...);
else
   return normal_function(...);
  }

As pointed out by Dave Hansen, this will,

1. Create a new, wholly untested code path for huge page
2. Create two places to patch bugs
3. Are not reusing code when possible

This patchset is to address these problems via merging huge/normal code
path/functions if possible.

One concern is that this may cause code size to dilate when
!CONFIG_TRANSPARENT_HUGEPAGE.  The data shows that most refactoring will
only cause quite slight code size increase.

This patch (of 8):

To improve code readability.

Link: http://lkml.kernel.org/r/20180720071845.17920-2-ying.huang@intel.com
Signed-off-by: "Huang, Ying" <ying.huang@intel.com>
Suggested-and-acked-by: Dave Hansen <dave.hansen@linux.intel.com>
Reviewed-by: Daniel Jordan <daniel.m.jordan@oracle.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Shaohua Li <shli@kernel.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@redhat.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Matthew Wilcox <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agomm: struct shrinker: make flags of unsigned type
Kirill Tkhai [Wed, 22 Aug 2018 04:51:57 +0000 (21:51 -0700)]
mm: struct shrinker: make flags of unsigned type

Currently, there are two flags only, so unsigned is more then enough.
Also, move int seeks to keep these fields together.

Link: http://lkml.kernel.org/r/153199748720.21131.6476256940113102483.stgit@localhost.localdomain
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Vladimir Davydov <vdavydov.dev@gmail.com>
Cc: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agomm: struct shrink_control: keep int fields together
Kirill Tkhai [Wed, 22 Aug 2018 04:51:53 +0000 (21:51 -0700)]
mm: struct shrink_control: keep int fields together

Patch series "Reorderings in struct shrinker and struct shrink_control".

These structures are intensively used during reclaim and, displace other
data in cache, so there is no a reason they have int fields not grouped
together.

This patch (of 2):

gfp_t is of unsigned type, so let's move nid to keep them together.

Link: http://lkml.kernel.org/r/153199747930.21131.861043607301997810.stgit@localhost.localdomain
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Vladimir Davydov <vdavydov.dev@gmail.com>
Cc: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agomm: check shrinker is memcg-aware in register_shrinker_prepared()
Kirill Tkhai [Wed, 22 Aug 2018 04:51:49 +0000 (21:51 -0700)]
mm: check shrinker is memcg-aware in register_shrinker_prepared()

There is a sad BUG introduced in patch adding SHRINKER_REGISTERING.
shrinker_idr business is only for memcg-aware shrinkers.  Only such type
of shrinkers have id and they must be finaly installed via idr_replace()
in this function.  For !memcg-aware shrinkers we never initialize
shrinker->id field.

But there are all types of shrinkers passed to idr_replace(), and every
!memcg-aware shrinker with random ID (most probably, its id is 0)
replaces memcg-aware shrinker pointed by the ID in IDR.

This patch fixes the problem.

Link: http://lkml.kernel.org/r/8ff8a793-8211-713a-4ed9-d6e52390c2fc@virtuozzo.com
Fixes: 7e010df53c80 "mm: use special value SHRINKER_REGISTERING instead of list_empty() check"
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Reported-by: <syzbot+d5f648a1bfe15678786b@syzkaller.appspotmail.com>
Cc: Andrey Ryabinin <aryabinin@virtuozzo.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Josef Bacik <jbacik@fb.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Cc: Shakeel Butt <shakeelb@google.com>
Cc: <syzkaller-bugs@googlegroups.com>
Cc: Huang Ying <ying.huang@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agoautofs: fix autofs_sbi() does not check super block type
Ian Kent [Wed, 22 Aug 2018 04:51:45 +0000 (21:51 -0700)]
autofs: fix autofs_sbi() does not check super block type

autofs_sbi() does not check the superblock magic number to verify it has
been given an autofs super block.

Link: http://lkml.kernel.org/r/153475422934.17131.7563724552005298277.stgit@pluto.themaw.net
Reported-by: <syzbot+87c3c541582e56943277@syzkaller.appspotmail.com>
Signed-off-by: Ian Kent <raven@themaw.net>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agoMerge tag 'please-pull-noboot' of git://git.kernel.org/pub/scm/linux/kernel/git/aegl...
Linus Torvalds [Tue, 21 Aug 2018 02:37:09 +0000 (19:37 -0700)]
Merge tag 'please-pull-noboot' of git://git./linux/kernel/git/aegl/linux

Pull ia64 NO_BOOTMEM conversion from Tony Luck:
 "Mike Rapoport kindly fixed up ia64 to work with NO_BOOTMEM"

* tag 'please-pull-noboot' of git://git.kernel.org/pub/scm/linux/kernel/git/aegl/linux:
  ia64: switch to NO_BOOTMEM
  ia64: use mem_data to detect nodes' minimal and maximal PFNs
  ia64: remove unused num_dma_physpages member from 'struct early_node_data'
  ia64: contig/paging_init: reduce code duplication

6 years agoMerge tag 'linux-kselftest-4.19-rc1' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Tue, 21 Aug 2018 02:02:17 +0000 (19:02 -0700)]
Merge tag 'linux-kselftest-4.19-rc1' of git://git./linux/kernel/git/shuah/linux-kselftest

Pull Kselftest update from Shuah Khan:

 - add cgroup core selftests

 - fix compile warnings in android ion test

 - fix to bugs in exclude and skip paths in vDSO test

 - remove obsolete config options

 - add missing .gitignore file

* tag 'linux-kselftest-4.19-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/shuah/linux-kselftest:
  selftests/ftrace: Fix kprobe string testcase to not probe notrace function
  selftests: mount: remove no longer needed config option
  selftests: cgroup: add gitignore file
  Add cgroup core selftests
  selftests: vDSO - fix to return KSFT_SKIP when test couldn't be run
  selftests: vDSO - fix to exclude x86 test on non-x86 platforms
  selftests/android: initialize heap_type to avoid compiling warning

6 years agoMerge tag 'trace-v4.19' of git://git.kernel.org/pub/scm/linux/kernel/git/rostedt...
Linus Torvalds [Tue, 21 Aug 2018 01:32:00 +0000 (18:32 -0700)]
Merge tag 'trace-v4.19' of git://git./linux/kernel/git/rostedt/linux-trace

Pull tracing updates from Steven Rostedt:

 - Restructure of lockdep and latency tracers

   This is the biggest change. Joel Fernandes restructured the hooks
   from irqs and preemption disabling and enabling. He got rid of a lot
   of the preprocessor #ifdef mess that they caused.

   He turned both lockdep and the latency tracers to use trace events
   inserted in the preempt/irqs disabling paths. But unfortunately,
   these started to cause issues in corner cases. Thus, parts of the
   code was reverted back to where lockdep and the latency tracers just
   get called directly (without using the trace events). But because the
   original change cleaned up the code very nicely we kept that, as well
   as the trace events for preempt and irqs disabling, but they are
   limited to not being called in NMIs.

 - Have trace events use SRCU for "rcu idle" calls. This was required
   for the preempt/irqs off trace events. But it also had to not allow
   them to be called in NMI context. Waiting till Paul makes an NMI safe
   SRCU API.

 - New notrace SRCU API to allow trace events to use SRCU.

 - Addition of mcount-nop option support

 - SPDX headers replacing GPL templates.

 - Various other fixes and clean ups.

 - Some fixes are marked for stable, but were not fully tested before
   the merge window opened.

* tag 'trace-v4.19' of git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/linux-trace: (44 commits)
  tracing: Fix SPDX format headers to use C++ style comments
  tracing: Add SPDX License format tags to tracing files
  tracing: Add SPDX License format to bpf_trace.c
  blktrace: Add SPDX License format header
  s390/ftrace: Add -mfentry and -mnop-mcount support
  tracing: Add -mcount-nop option support
  tracing: Avoid calling cc-option -mrecord-mcount for every Makefile
  tracing: Handle CC_FLAGS_FTRACE more accurately
  Uprobe: Additional argument arch_uprobe to uprobe_write_opcode()
  Uprobes: Simplify uprobe_register() body
  tracepoints: Free early tracepoints after RCU is initialized
  uprobes: Use synchronize_rcu() not synchronize_sched()
  tracing: Fix synchronizing to event changes with tracepoint_synchronize_unregister()
  ftrace: Remove unused pointer ftrace_swapper_pid
  tracing: More reverting of "tracing: Centralize preemptirq tracepoints and unify their usage"
  tracing/irqsoff: Handle preempt_count for different configs
  tracing: Partial revert of "tracing: Centralize preemptirq tracepoints and unify their usage"
  tracing: irqsoff: Account for additional preempt_disable
  trace: Use rcu_dereference_raw for hooks from trace-event subsystem
  tracing/kprobes: Fix within_notrace_func() to check only notrace functions
  ...

6 years agoMerge tag 'ceph-for-4.19-rc1' of git://github.com/ceph/ceph-client
Linus Torvalds [Tue, 21 Aug 2018 01:26:55 +0000 (18:26 -0700)]
Merge tag 'ceph-for-4.19-rc1' of git://github.com/ceph/ceph-client

Pull ceph updates from Ilya Dryomov:
 "The main things are support for cephx v2 authentication protocol and
  basic support for rbd images within namespaces (myself).

  Also included are y2038 conversion patches from Arnd, a pile of
  miscellaneous fixes from Chengguang and Zheng's feature bit
  infrastructure for the filesystem"

* tag 'ceph-for-4.19-rc1' of git://github.com/ceph/ceph-client: (40 commits)
  ceph: don't drop message if it contains more data than expected
  ceph: support cephfs' own feature bits
  crush: fix using plain integer as NULL warning
  libceph: remove unnecessary non NULL check for request_key
  ceph: refactor error handling code in ceph_reserve_caps()
  ceph: refactor ceph_unreserve_caps()
  ceph: change to void return type for __do_request()
  ceph: compare fsc->max_file_size and inode->i_size for max file size limit
  ceph: add additional size check in ceph_setattr()
  ceph: add additional offset check in ceph_write_iter()
  ceph: add additional range check in ceph_fallocate()
  ceph: add new field max_file_size in ceph_fs_client
  libceph: weaken sizeof check in ceph_x_verify_authorizer_reply()
  libceph: check authorizer reply/challenge length before reading
  libceph: implement CEPHX_V2 calculation mode
  libceph: add authorizer challenge
  libceph: factor out encrypt_authorizer()
  libceph: factor out __ceph_x_decrypt()
  libceph: factor out __prepare_write_connect()
  libceph: store ceph_auth_handshake pointer in ceph_connection
  ...

6 years agoMerge tag 'rtc-4.19' of git://git.kernel.org/pub/scm/linux/kernel/git/abelloni/linux
Linus Torvalds [Mon, 20 Aug 2018 23:30:27 +0000 (16:30 -0700)]
Merge tag 'rtc-4.19' of git://git./linux/kernel/git/abelloni/linux

Pull RTC updates from Alexandre Belloni:
 "It is now possible to add custom sysfs attributes while avoiding a
  possible race condition. Unused code has been removed resulting in a
  nice reduction of the code base. And more drivers have been switched
  to SPDX by their maintainers.

 Summary:

  Subsystem:
   - new helpers to add custom sysfs attributes
   - struct rtc_task removal along with rtc_irq_[un]register()
   - rtc_irq_set_state and rtc_irq_set_freq are not exported anymore

  Drivers:
   - armada38x: reset after rtc power loss
   - ds1307: now supports m41t11
   - isl1208: now supports isl1219 and tamper detection
   - pcf2127: internal SRAM support"

* tag 'rtc-4.19' of git://git.kernel.org/pub/scm/linux/kernel/git/abelloni/linux: (34 commits)
  rtc: ds1307: simplify hwmon config
  rtc: s5m: Add SPDX license identifier
  rtc: maxim: Add SPDX license identifiers
  rtc: isl1219: add device tree documentation
  rtc: isl1208: set ev-evienb bit from device tree
  rtc: isl1208: Add "evdet" interrupt source for isl1219
  rtc: isl1208: add support for isl1219 with tamper detection
  rtc: sysfs: facilitate attribute add to rtc device
  rtc: remove struct rtc_task
  char: rtc: remove task handling
  rtc: pcf85063: preserve control register value between stop and start
  rtc: sh: remove unused variable rtc_dev
  rtc: unexport rtc_irq_set_*
  rtc: simplify rtc_irq_set_state/rtc_irq_set_freq
  rtc: remove irq_task and irq_task_lock
  rtc: remove rtc_irq_register/rtc_irq_unregister
  rtc: sh: remove dead code
  rtc: sa1100: don't set PIE frequency
  rtc: ds1307: support m41t11 variant
  rtc: ds1307: fix data pointer to m41t0
  ...

6 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jikos/livep...
Linus Torvalds [Mon, 20 Aug 2018 23:10:47 +0000 (16:10 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/jikos/livepatching

Pull livepatching updates from Jiri Kosina:
 "Code cleanups from Kamalesh Babulal"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jikos/livepatching:
  livepatch: Validate module/old func name length
  livepatch: Remove reliable stacktrace check in klp_try_switch_task()

6 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jikos/hid
Linus Torvalds [Mon, 20 Aug 2018 22:59:01 +0000 (15:59 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/jikos/hid

Pull HID updates from Jiri Kosina:

 - touch_max detection improvements and quirk handling fixes in wacom
   driver from Jason Gerecke and Ping Cheng

 - Palm rejection from Dmitry Torokhov and _dial support from Benjamin
   Tissoires for hid-multitouch driver

 - Low voltage support for i2c-hid driver from Stephen Boyd

 - Guitar-Hero support from Nicolas Adenis-Lamarre

 - other assorted small fixes and device ID additions

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jikos/hid: (40 commits)
  HID: intel_ish-hid: tx_buf memory leak on probe/remove
  HID: intel-ish-hid: Prevent loading of driver on Mehlow
  HID: cougar: Add support for the Cougar 500k Gaming Keyboard
  HID: cougar: make compare_device_paths reusable
  HID: intel-ish-hid: remove redundant variable num_frags
  HID: multitouch: handle palm for touchscreens
  HID: multitouch: touchscreens also use confidence reports
  HID: multitouch: report MT_TOOL_PALM for non-confident touches
  HID: microsoft: support the Surface Dial
  HID: core: do not upper bound the collection stack
  HID: input: enable Totem on the Dell Canvas 27
  HID: multitouch: remove one copy of values
  HID: multitouch: ditch mt_report_id
  HID: multitouch: store a per application quirks value
  HID: multitouch: Store per collection multitouch data
  HID: multitouch: make sure the static list of class is not changed
  input: add MT_TOOL_DIAL
  HID: elan: Add support for touchpad on the Toshiba Click Mini L9W
  HID: elan: Add USB-id for HP x2 10-n000nd touchpad
  HID: elan: Add a flag for selecting if the touchpad has a LED
  ...

6 years agoMerge tag 'backlight-next-4.19' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Mon, 20 Aug 2018 22:41:37 +0000 (15:41 -0700)]
Merge tag 'backlight-next-4.19' of git://git./linux/kernel/git/lee/backlight

Pull backlight updates from Lee Jones:
 "Core Framework:
   - Remove unused/obsolete code/comments

  New Functionality:
   - Allow less granular brightness specification for high-res PWMs; pwm_bl
   - Align brightness {inc,dec}rements with that perceived by the human-eye; pwm_bl

  Fix-ups:
   - Prepare for the introduction of -Wimplicit-fall-through; adp8860_bl

  Bug Fixes:
   - Fix uninitialised variable; pwm_bl"

* tag 'backlight-next-4.19' of git://git.kernel.org/pub/scm/linux/kernel/git/lee/backlight:
  backlight: pwm_bl: Fix uninitialized variable
  backlight: adp8860: Mark expected switch fall-through
  backlight: Remove obsolete comment for ->state
  dt-bindings: pwm-backlight: Move brightness-levels to optional
  backlight: pwm_bl: Compute brightness of LED linearly to human eye
  dt-bindings: pwm-backlight: Add a num-interpolation-steps property
  backlight: pwm_bl: Linear interpolation between brightness-levels

6 years agoMerge tag 'mfd-next-4.19' of git://git.kernel.org/pub/scm/linux/kernel/git/lee/mfd
Linus Torvalds [Mon, 20 Aug 2018 22:38:44 +0000 (15:38 -0700)]
Merge tag 'mfd-next-4.19' of git://git./linux/kernel/git/lee/mfd

Pull MFD updates from Lee Jones:
 "New Drivers:
   - Add Cirrus Logic Madera Codec (CS47L35, CS47L85 and CS47L90/91) driver
   - Add ChromeOS EC CEC driver
   - Add ROHM BD71837 PMIC driver

  New Device Support:
   - Add support for Dialog Semi DA9063L PMIC variant to DA9063
   - Add support for Intel Ice Lake to Intel-PLSS-PCI
   - Add support for X-Powers AXP806 to AXP20x

  New Functionality:
   - Add support for USB Charging to the ChromeOS Embedded Controller
   - Add support for HDMI CEC to the ChromeOS Embedded Controller
   - Add support for HDMI CEC to Intel HDMI
   - Add support for accessory detection to Madera devices
   - Allow individual pins to be configured via DT' wlf,csnaddr-pd
   - Provide legacy platform specific EEPROM/Watchdog commands; rave-sp

  Fix-upsL
   - Trivial renaming/spelling fixes; cros_ec, da9063-*
   - Convert to Managed Resources (devm_*); da9063-*, ti_am335x_tscadc
   - Transition to helper macros/functions; da9063-*
   - Constify; kempld-core
   - Improve error path/messages; wm8994-core
   - Disable IRQs locally instead of relying on USB subsystem; dln2
   - Remove unused code; rave-sp
   - New exports; sec-core

  Bug Fixes:
   - Fix possible false I2C transaction error; arizona-core
   - Fix declared memory area size; hi655x-pmic
   - Fix checksum type; rave-sp
   - Fix incorrect default serial port configuration: rave-sp
   - Fix incorrect coherent DMA mask for sub-devices; sm501"

* tag 'mfd-next-4.19' of git://git.kernel.org/pub/scm/linux/kernel/git/lee/mfd: (60 commits)
  mfd: madera: Add register definitions for accessory detect
  mfd: sm501: Set coherent_dma_mask when creating subdevices
  mfd: bd71837: Devicetree bindings for ROHM BD71837 PMIC
  mfd: bd71837: Core driver for ROHM BD71837 PMIC
  media: platform: cros-ec-cec: Fix dependency on MFD_CROS_EC
  mfd: sec-core: Export OF module alias table
  mfd: as3722: Disable auto-power-on when AC OK
  mfd: axp20x: Support AXP806 in I2C mode
  mfd: axp20x: Add self-working mode support for AXP806
  dt-bindings: mfd: axp20x: Add "self-working" mode for AXP806
  mfd: wm8994: Allow to configure CS/ADDR Pulldown from dts
  mfd: wm8994: Allow to configure Speaker Mode Pullup from dts
  mfd: rave-sp: Emulate CMD_GET_STATUS on device that don't support it
  mfd: rave-sp: Add legacy watchdog ping command translation
  mfd: rave-sp: Add legacy EEPROM access command translation
  mfd: rave-sp: Initialize flow control and parity of the port
  mfd: rave-sp: Fix incorrectly specified checksum type
  mfd: rave-sp: Remove unused defines
  mfd: hi655x: Fix regmap area declared size for hi655x
  mfd: ti_am335x_tscadc: Fix struct clk memory leak
  ...

6 years agoMerge tag 'edac_fixes_for_4.19' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Mon, 20 Aug 2018 22:28:54 +0000 (15:28 -0700)]
Merge tag 'edac_fixes_for_4.19' of git://git./linux/kernel/git/bp/bp

Pull EDAC fix from Borislav Petkov:
 "An urgent fix for a NULL ptr deref on machines with LRDDR4 DIMMs, from
  Takashi Iwai"

* tag 'edac_fixes_for_4.19' of git://git.kernel.org/pub/scm/linux/kernel/git/bp/bp:
  EDAC: Add missing MEM_LRDDR4 entry in edac_mem_types[]

6 years agoRaise the minimum required gcc version to 4.6
Joe Perches [Mon, 20 Aug 2018 20:15:26 +0000 (13:15 -0700)]
Raise the minimum required gcc version to 4.6

Various architectures fail to build properly with older versions of the
gcc compiler.

An example from Guenter Roeck in thread [1]:
>
>   In file included from ./include/linux/mm.h:17:0,
>                    from ./include/linux/pid_namespace.h:7,
>                    from ./include/linux/ptrace.h:10,
>                    from arch/openrisc/kernel/asm-offsets.c:32:
>   ./include/linux/mm_types.h:497:16: error: flexible array member in otherwise empty struct
>
> This is just an example with gcc 4.5.1 for or32. I have seen the problem
> with gcc 4.4 (for unicore32) as well.

So update the minimum required version of gcc to 4.6.

[1] https://lore.kernel.org/lkml/20180814170904.GA12768@roeck-us.net/

Miscellanea:

 - Update Documentation/process/changes.rst

 - Remove and consolidate version test blocks in compiler-gcc.h for
   versions lower than 4.6

Signed-off-by: Joe Perches <joe@perches.com>
Reviewed-by: Kees Cook <keescook@chromium.org>
Reviewed-by: Nick Desaulniers <ndesaulniers@google.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agoia64: Fix kernel BUG at lib/ioremap.c:72!
Tony Luck [Mon, 20 Aug 2018 16:31:04 +0000 (09:31 -0700)]
ia64: Fix kernel BUG at lib/ioremap.c:72!

Commit 0bbf47eab469 ("ia64: use asm-generic/io.h") results in a BUG
while booting ia64.  This is because asm-generic/io.h defines
PCI_IOBASE, which results in the function acpi_pci_root_remap_iospace()
doing a lot of unnecessary (and wrong) things.

I'd suggested an #if !CONFIG_IA64 in the functon, but Arnd suggested
keeping the fix inside the arch/ia64 tree.

Fixes: 0bbf47eab469 ("ia64: use asm-generic/io.h")
Suggested-by: Arnd Bergman <arnd@arndb.de>
Signed-off-by: Tony Luck <tony.luck@intel.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agoMerge branch 'for-4.19/upstream' into for-linus
Jiri Kosina [Mon, 20 Aug 2018 16:33:50 +0000 (18:33 +0200)]
Merge branch 'for-4.19/upstream' into for-linus

6 years agoMerge branch 'for-4.19/wiimote' into for-linus
Jiri Kosina [Mon, 20 Aug 2018 16:13:57 +0000 (18:13 +0200)]
Merge branch 'for-4.19/wiimote' into for-linus

Guitar-Hero devices support for hid-wiimote

6 years agoMerge branch 'for-4.19/wacom' into for-linus
Jiri Kosina [Mon, 20 Aug 2018 16:12:42 +0000 (18:12 +0200)]
Merge branch 'for-4.19/wacom' into for-linus

Wacom driver updates:

- touch_max detection improvements
- quirk handling cleanup
- get rid of wacom custom usages

6 years agoMerge branch 'for-4.19/upstream' into for-linus
Jiri Kosina [Mon, 20 Aug 2018 16:11:20 +0000 (18:11 +0200)]
Merge branch 'for-4.19/upstream' into for-linus

Assorted small driver/core fixes.

6 years agoMerge branch 'for-4.19/sony' into for-linus
Jiri Kosina [Mon, 20 Aug 2018 16:10:33 +0000 (18:10 +0200)]
Merge branch 'for-4.19/sony' into for-linus

devm_* API conversion for hid-sony

6 years agoMerge branch 'for-4.19/multitouch-multiaxis' into for-linus
Jiri Kosina [Mon, 20 Aug 2018 16:09:06 +0000 (18:09 +0200)]
Merge branch 'for-4.19/multitouch-multiaxis' into for-linus

Multitouch updates:

- Dial support
- Palm rejection for touchscreens
- a few small assorted fixes

6 years agoMerge branch 'for-4.19/intel-ish' into for-linus
Jiri Kosina [Mon, 20 Aug 2018 16:07:36 +0000 (18:07 +0200)]
Merge branch 'for-4.19/intel-ish' into for-linus

Device-specific fixes for hid-intel-ish

6 years agoMerge branch 'for-4.19/i2c-hid' into for-linus
Jiri Kosina [Mon, 20 Aug 2018 16:07:01 +0000 (18:07 +0200)]
Merge branch 'for-4.19/i2c-hid' into for-linus

Low voltage support for i2c-hid

6 years agoMerge branch 'for-4.19/elan' into for-linus
Jiri Kosina [Mon, 20 Aug 2018 16:06:30 +0000 (18:06 +0200)]
Merge branch 'for-4.19/elan' into for-linus

Resolution/pressure fixes and new device support for hid-elan

6 years agoMerge branch 'for-4.19/cougar' into for-linus
Jiri Kosina [Mon, 20 Aug 2018 16:05:17 +0000 (18:05 +0200)]
Merge branch 'for-4.19/cougar' into for-linus

New device support for hid-cougar

6 years agoMerge branch 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/gerg/m68knommu
Linus Torvalds [Sun, 19 Aug 2018 23:23:03 +0000 (16:23 -0700)]
Merge branch 'for-next' of git://git./linux/kernel/git/gerg/m68knommu

Pull m68knommu updates from Greg Ungerer:
 "Only two changes.

  One cleans up warnings in the ColdFire DMA code, the other stubs out
  (with warnings) ColdFire clock api functions not normally used"

* 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/gerg/m68knommu:
  m68knommu: Fix typos in Coldfire 5272 DMA debug code
  m68k: coldfire: Normalize clk API

6 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net
Linus Torvalds [Sun, 19 Aug 2018 18:51:45 +0000 (11:51 -0700)]
Merge git://git./linux/kernel/git/davem/net

Pull networking fixes from David Miller:

 1) Fix races in IPVS, from Tan Hu.

 2) Missing unbind in matchall classifier, from Hangbin Liu.

 3) Missing act_ife action release, from Vlad Buslov.

 4) Cure lockdep splats in ila, from Cong Wang.

 5) veth queue leak on link delete, from Toshiaki Makita.

 6) Disable isdn's IIOCDBGVAR ioctl, it exposes kernel addresses. From
    Kees Cook.

 7) RCU usage fixup in XDP, from Tariq Toukan.

 8) Two TCP ULP fixes from Daniel Borkmann.

 9) r8169 needs REALTEK_PHY as a Kconfig dependency, from Heiner
    Kallweit.

10) Always take tcf_lock with BH disabled, otherwise we can deadlock
    with rate estimator code paths. From Vlad Buslov.

11) Don't use MSI-X on RTL8106e r8169 chips, they don't resume properly.
    From Jian-Hong Pan.

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (41 commits)
  ip6_vti: fix creating fallback tunnel device for vti6
  ip_vti: fix a null pointer deferrence when create vti fallback tunnel
  r8169: don't use MSI-X on RTL8106e
  net: lan743x_ptp: convert to ktime_get_clocktai_ts64
  net: sched: always disable bh when taking tcf_lock
  ip6_vti: simplify stats handling in vti6_xmit
  bpf: fix redirect to map under tail calls
  r8169: add missing Kconfig dependency
  tools/bpf: fix bpf selftest test_cgroup_storage failure
  bpf, sockmap: fix sock_map_ctx_update_elem race with exist/noexist
  bpf, sockmap: fix map elem deletion race with smap_stop_sock
  bpf, sockmap: fix leakage of smap_psock_map_entry
  tcp, ulp: fix leftover icsk_ulp_ops preventing sock from reattach
  tcp, ulp: add alias for all ulp modules
  bpf: fix a rcu usage warning in bpf_prog_array_copy_core()
  samples/bpf: all XDP samples should unload xdp/bpf prog on SIGTERM
  net/xdp: Fix suspicious RCU usage warning
  net/mlx5e: Delete unneeded function argument
  Documentation: networking: ti-cpsw: correct cbs parameters for Eth1 100Mb
  isdn: Disable IIOCDBGVAR
  ...

6 years agoip6_vti: fix creating fallback tunnel device for vti6
Haishuang Yan [Sun, 19 Aug 2018 07:05:05 +0000 (15:05 +0800)]
ip6_vti: fix creating fallback tunnel device for vti6

When set fb_tunnels_only_for_init_net to 1, don't create fallback tunnel
device for vti6 when a new namespace is created.

Tested:
[root@builder2 ~]# modprobe ip6_tunnel
[root@builder2 ~]# modprobe ip6_vti
[root@builder2 ~]# echo 1 > /proc/sys/net/core/fb_tunnels_only_for_init_net
[root@builder2 ~]# unshare -n
[root@builder2 ~]# ip link
1: lo: <LOOPBACK> mtu 65536 qdisc noop state DOWN mode DEFAULT group
default qlen 1000
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00

Signed-off-by: Haishuang Yan <yanhaishuang@cmss.chinamobile.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoip_vti: fix a null pointer deferrence when create vti fallback tunnel
Haishuang Yan [Sun, 19 Aug 2018 07:05:04 +0000 (15:05 +0800)]
ip_vti: fix a null pointer deferrence when create vti fallback tunnel

After set fb_tunnels_only_for_init_net to 1, the itn->fb_tunnel_dev will
be NULL and will cause following crash:

[ 2742.849298] BUG: unable to handle kernel NULL pointer dereference at 0000000000000941
[ 2742.851380] PGD 800000042c21a067 P4D 800000042c21a067 PUD 42aaed067 PMD 0
[ 2742.852818] Oops: 0002 [#1] SMP PTI
[ 2742.853570] CPU: 7 PID: 2484 Comm: unshare Kdump: loaded Not tainted 4.18.0-rc8+ #2
[ 2742.855163] Hardware name: Fedora Project OpenStack Nova, BIOS seabios-1.7.5-11.el7 04/01/2014
[ 2742.856970] RIP: 0010:vti_init_net+0x3a/0x50 [ip_vti]
[ 2742.858034] Code: 90 83 c0 48 c7 c2 20 a1 83 c0 48 89 fb e8 6e 3b f6 ff 85 c0 75 22 8b 0d f4 19 00 00 48 8b 93 00 14 00 00 48 8b 14 ca 48 8b 12 <c6> 82 41 09 00 00 04 c6 82 38 09 00 00 45 5b c3 66 0f 1f 44 00 00
[ 2742.861940] RSP: 0018:ffff9be28207fde0 EFLAGS: 00010246
[ 2742.863044] RAX: 0000000000000000 RBX: ffff8a71ebed4980 RCX: 0000000000000013
[ 2742.864540] RDX: 0000000000000000 RSI: 0000000000000013 RDI: ffff8a71ebed4980
[ 2742.866020] RBP: ffff8a71ea717000 R08: ffffffffc083903c R09: ffff8a71ea717000
[ 2742.867505] R10: 0000000000000000 R11: 0000000000000000 R12: ffff8a71ebed4980
[ 2742.868987] R13: 0000000000000013 R14: ffff8a71ea5b49c0 R15: 0000000000000000
[ 2742.870473] FS:  00007f02266c9740(0000) GS:ffff8a71ffdc0000(0000) knlGS:0000000000000000
[ 2742.872143] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 2742.873340] CR2: 0000000000000941 CR3: 000000042bc20006 CR4: 00000000001606e0
[ 2742.874821] Call Trace:
[ 2742.875358]  ops_init+0x38/0xf0
[ 2742.876078]  setup_net+0xd9/0x1f0
[ 2742.876789]  copy_net_ns+0xb7/0x130
[ 2742.877538]  create_new_namespaces+0x11a/0x1d0
[ 2742.878525]  unshare_nsproxy_namespaces+0x55/0xa0
[ 2742.879526]  ksys_unshare+0x1a7/0x330
[ 2742.880313]  __x64_sys_unshare+0xe/0x20
[ 2742.881131]  do_syscall_64+0x5b/0x180
[ 2742.881933]  entry_SYSCALL_64_after_hwframe+0x44/0xa9

Reproduce:
echo 1 > /proc/sys/net/core/fb_tunnels_only_for_init_net
modprobe ip_vti
unshare -n

Fixes: 79134e6ce2c9 ("net: do not create fallback tunnels for non-default namespaces")
Cc: Eric Dumazet <edumazet@google.com>
Signed-off-by: Haishuang Yan <yanhaishuang@cmss.chinamobile.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agor8169: don't use MSI-X on RTL8106e
Jian-Hong Pan [Fri, 17 Aug 2018 05:07:35 +0000 (13:07 +0800)]
r8169: don't use MSI-X on RTL8106e

Found the ethernet network on ASUS X441UAR doesn't come back on resume
from suspend when using MSI-X.  The chip is RTL8106e - version 39.

[   21.848357] libphy: r8169: probed
[   21.848473] r8169 0000:02:00.0 eth0: RTL8106e, 0c:9d:92:32:67:b4, XID
44900000, IRQ 127
[   22.518860] r8169 0000:02:00.0 enp2s0: renamed from eth0
[   29.458041] Generic PHY r8169-200:00: attached PHY driver [Generic
PHY] (mii_bus:phy_addr=r8169-200:00, irq=IGNORE)
[   63.227398] r8169 0000:02:00.0 enp2s0: Link is Up - 100Mbps/Full -
flow control off
[  124.514648] Generic PHY r8169-200:00: attached PHY driver [Generic
PHY] (mii_bus:phy_addr=r8169-200:00, irq=IGNORE)

Here is the ethernet controller in detail:

02:00.0 Ethernet controller [0200]: Realtek Semiconductor Co., Ltd.
RTL8101/2/6E PCI Express Fast/Gigabit Ethernet controller [10ec:8136]
(rev 07)
Subsystem: ASUSTeK Computer Inc. RTL810xE PCI Express Fast
Ethernet controller [1043:200f]
Flags: bus master, fast devsel, latency 0, IRQ 16
I/O ports at e000 [size=256]
Memory at ef100000 (64-bit, non-prefetchable) [size=4K]
Memory at e0000000 (64-bit, prefetchable) [size=16K]
Capabilities: <access denied>
Kernel driver in use: r8169
Kernel modules: r8169

Falling back to MSI fixes the issue.

Fixes: 6c6aa15fdea5 ("r8169: improve interrupt handling")
Signed-off-by: Jian-Hong Pan <jian-hong@endlessm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: lan743x_ptp: convert to ktime_get_clocktai_ts64
Arnd Bergmann [Wed, 15 Aug 2018 17:49:49 +0000 (19:49 +0200)]
net: lan743x_ptp: convert to ktime_get_clocktai_ts64

timekeeping_clocktai64() has been renamed to ktime_get_clocktai_ts64()
for consistency with the other ktime_get_* access functions.

Rename the new caller that has come up as well.

Question: this is the only ptp driver that sets the hardware time
to the current system time in TAI. Why does it do that?

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: sched: always disable bh when taking tcf_lock
Vlad Buslov [Tue, 14 Aug 2018 18:46:16 +0000 (21:46 +0300)]
net: sched: always disable bh when taking tcf_lock

Recently, ops->init() and ops->dump() of all actions were modified to
always obtain tcf_lock when accessing private action state. Actions that
don't depend on tcf_lock for synchronization with their data path use
non-bh locking API. However, tcf_lock is also used to protect rate
estimator stats in softirq context by timer callback.

Change ops->init() and ops->dump() of all actions to disable bh when using
tcf_lock to prevent deadlock reported by following lockdep warning:

[  105.470398] ================================
[  105.475014] WARNING: inconsistent lock state
[  105.479628] 4.18.0-rc8+ #664 Not tainted
[  105.483897] --------------------------------
[  105.488511] inconsistent {SOFTIRQ-ON-W} -> {IN-SOFTIRQ-W} usage.
[  105.494871] swapper/16/0 [HC0[0]:SC1[1]:HE1:SE0] takes:
[  105.500449] 00000000f86c012e (&(&p->tcfa_lock)->rlock){+.?.}, at: est_fetch_counters+0x3c/0xa0
[  105.509696] {SOFTIRQ-ON-W} state was registered at:
[  105.514925]   _raw_spin_lock+0x2c/0x40
[  105.519022]   tcf_bpf_init+0x579/0x820 [act_bpf]
[  105.523990]   tcf_action_init_1+0x4e4/0x660
[  105.528518]   tcf_action_init+0x1ce/0x2d0
[  105.532880]   tcf_exts_validate+0x1d8/0x200
[  105.537416]   fl_change+0x55a/0x268b [cls_flower]
[  105.542469]   tc_new_tfilter+0x748/0xa20
[  105.546738]   rtnetlink_rcv_msg+0x56a/0x6d0
[  105.551268]   netlink_rcv_skb+0x18d/0x200
[  105.555628]   netlink_unicast+0x2d0/0x370
[  105.559990]   netlink_sendmsg+0x3b9/0x6a0
[  105.564349]   sock_sendmsg+0x6b/0x80
[  105.568271]   ___sys_sendmsg+0x4a1/0x520
[  105.572547]   __sys_sendmsg+0xd7/0x150
[  105.576655]   do_syscall_64+0x72/0x2c0
[  105.580757]   entry_SYSCALL_64_after_hwframe+0x49/0xbe
[  105.586243] irq event stamp: 489296
[  105.590084] hardirqs last  enabled at (489296): [<ffffffffb507e639>] _raw_spin_unlock_irq+0x29/0x40
[  105.599765] hardirqs last disabled at (489295): [<ffffffffb507e745>] _raw_spin_lock_irq+0x15/0x50
[  105.609277] softirqs last  enabled at (489292): [<ffffffffb413a6a3>] irq_enter+0x83/0xa0
[  105.618001] softirqs last disabled at (489293): [<ffffffffb413a800>] irq_exit+0x140/0x190
[  105.626813]
               other info that might help us debug this:
[  105.633976]  Possible unsafe locking scenario:

[  105.640526]        CPU0
[  105.643325]        ----
[  105.646125]   lock(&(&p->tcfa_lock)->rlock);
[  105.650747]   <Interrupt>
[  105.653717]     lock(&(&p->tcfa_lock)->rlock);
[  105.658514]
                *** DEADLOCK ***

[  105.665349] 1 lock held by swapper/16/0:
[  105.669629]  #0: 00000000a640ad99 ((&est->timer)){+.-.}, at: call_timer_fn+0x10b/0x550
[  105.678200]
               stack backtrace:
[  105.683194] CPU: 16 PID: 0 Comm: swapper/16 Not tainted 4.18.0-rc8+ #664
[  105.690249] Hardware name: Supermicro SYS-2028TP-DECR/X10DRT-P, BIOS 2.0b 03/30/2017
[  105.698626] Call Trace:
[  105.701421]  <IRQ>
[  105.703791]  dump_stack+0x92/0xeb
[  105.707461]  print_usage_bug+0x336/0x34c
[  105.711744]  mark_lock+0x7c9/0x980
[  105.715500]  ? print_shortest_lock_dependencies+0x2e0/0x2e0
[  105.721424]  ? check_usage_forwards+0x230/0x230
[  105.726315]  __lock_acquire+0x923/0x26f0
[  105.730597]  ? debug_show_all_locks+0x240/0x240
[  105.735478]  ? mark_lock+0x493/0x980
[  105.739412]  ? check_chain_key+0x140/0x1f0
[  105.743861]  ? __lock_acquire+0x836/0x26f0
[  105.748323]  ? lock_acquire+0x12e/0x290
[  105.752516]  lock_acquire+0x12e/0x290
[  105.756539]  ? est_fetch_counters+0x3c/0xa0
[  105.761084]  _raw_spin_lock+0x2c/0x40
[  105.765099]  ? est_fetch_counters+0x3c/0xa0
[  105.769633]  est_fetch_counters+0x3c/0xa0
[  105.773995]  est_timer+0x87/0x390
[  105.777670]  ? est_fetch_counters+0xa0/0xa0
[  105.782210]  ? lock_acquire+0x12e/0x290
[  105.786410]  call_timer_fn+0x161/0x550
[  105.790512]  ? est_fetch_counters+0xa0/0xa0
[  105.795055]  ? del_timer_sync+0xd0/0xd0
[  105.799249]  ? __lock_is_held+0x93/0x110
[  105.803531]  ? mark_held_locks+0x20/0xe0
[  105.807813]  ? _raw_spin_unlock_irq+0x29/0x40
[  105.812525]  ? est_fetch_counters+0xa0/0xa0
[  105.817069]  ? est_fetch_counters+0xa0/0xa0
[  105.821610]  run_timer_softirq+0x3c4/0x9f0
[  105.826064]  ? lock_acquire+0x12e/0x290
[  105.830257]  ? __bpf_trace_timer_class+0x10/0x10
[  105.835237]  ? __lock_is_held+0x25/0x110
[  105.839517]  __do_softirq+0x11d/0x7bf
[  105.843542]  irq_exit+0x140/0x190
[  105.847208]  smp_apic_timer_interrupt+0xac/0x3b0
[  105.852182]  apic_timer_interrupt+0xf/0x20
[  105.856628]  </IRQ>
[  105.859081] RIP: 0010:cpuidle_enter_state+0xd8/0x4d0
[  105.864395] Code: 46 ff 48 89 44 24 08 0f 1f 44 00 00 31 ff e8 cf ec 46 ff 80 7c 24 07 00 0f 85 1d 02 00 00 e8 9f 90 4b ff fb 66 0f 1f 44 00 00 <4c> 8b 6c 24 08 4d 29 fd 0f 80 36 03 00 00 4c 89 e8 48 ba cf f7 53
[  105.884288] RSP: 0018:ffff8803ad94fd20 EFLAGS: 00000246 ORIG_RAX: ffffffffffffff13
[  105.892494] RAX: 0000000000000000 RBX: ffffe8fb300829c0 RCX: ffffffffb41e19e1
[  105.899988] RDX: 0000000000000007 RSI: dffffc0000000000 RDI: ffff8803ad9358ac
[  105.907503] RBP: ffffffffb6636300 R08: 0000000000000004 R09: 0000000000000000
[  105.914997] R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000000004
[  105.922487] R13: ffffffffb6636140 R14: ffffffffb66362d8 R15: 000000188d36091b
[  105.929988]  ? trace_hardirqs_on_caller+0x141/0x2d0
[  105.935232]  do_idle+0x28e/0x320
[  105.938817]  ? arch_cpu_idle_exit+0x40/0x40
[  105.943361]  ? mark_lock+0x8c1/0x980
[  105.947295]  ? _raw_spin_unlock_irqrestore+0x32/0x60
[  105.952619]  cpu_startup_entry+0xc2/0xd0
[  105.956900]  ? cpu_in_idle+0x20/0x20
[  105.960830]  ? _raw_spin_unlock_irqrestore+0x32/0x60
[  105.966146]  ? trace_hardirqs_on_caller+0x141/0x2d0
[  105.971391]  start_secondary+0x2b5/0x360
[  105.975669]  ? set_cpu_sibling_map+0x1330/0x1330
[  105.980654]  secondary_startup_64+0xa5/0xb0

Taking tcf_lock in sample action with bh disabled causes lockdep to issue a
warning regarding possible irq lock inversion dependency between tcf_lock,
and psample_groups_lock that is taken when holding tcf_lock in sample init:

[  162.108959]  Possible interrupt unsafe locking scenario:

[  162.116386]        CPU0                    CPU1
[  162.121277]        ----                    ----
[  162.126162]   lock(psample_groups_lock);
[  162.130447]                                local_irq_disable();
[  162.136772]                                lock(&(&p->tcfa_lock)->rlock);
[  162.143957]                                lock(psample_groups_lock);
[  162.150813]   <Interrupt>
[  162.153808]     lock(&(&p->tcfa_lock)->rlock);
[  162.158608]
                *** DEADLOCK ***

In order to prevent potential lock inversion dependency between tcf_lock
and psample_groups_lock, extract call to psample_group_get() from tcf_lock
protected section in sample action init function.

Fixes: 4e232818bd32 ("net: sched: act_mirred: remove dependency on rtnl lock")
Fixes: 764e9a24480f ("net: sched: act_vlan: remove dependency on rtnl lock")
Fixes: 729e01260989 ("net: sched: act_tunnel_key: remove dependency on rtnl lock")
Fixes: d77284956656 ("net: sched: act_sample: remove dependency on rtnl lock")
Fixes: e8917f437006 ("net: sched: act_gact: remove dependency on rtnl lock")
Fixes: b6a2b971c0b0 ("net: sched: act_csum: remove dependency on rtnl lock")
Fixes: 2142236b4584 ("net: sched: act_bpf: remove dependency on rtnl lock")
Signed-off-by: Vlad Buslov <vladbu@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoMerge tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm
Linus Torvalds [Sun, 19 Aug 2018 17:38:36 +0000 (10:38 -0700)]
Merge tag 'for-linus' of git://git./virt/kvm/kvm

Pull first set of KVM updates from Paolo Bonzini:
 "PPC:
   - minor code cleanups

  x86:
   - PCID emulation and CR3 caching for shadow page tables
   - nested VMX live migration
   - nested VMCS shadowing
   - optimized IPI hypercall
   - some optimizations

  ARM will come next week"

* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm: (85 commits)
  kvm: x86: Set highest physical address bits in non-present/reserved SPTEs
  KVM/x86: Use CC_SET()/CC_OUT in arch/x86/kvm/vmx.c
  KVM: X86: Implement PV IPIs in linux guest
  KVM: X86: Add kvm hypervisor init time platform setup callback
  KVM: X86: Implement "send IPI" hypercall
  KVM/x86: Move X86_CR4_OSXSAVE check into kvm_valid_sregs()
  KVM: x86: Skip pae_root shadow allocation if tdp enabled
  KVM/MMU: Combine flushing remote tlb in mmu_set_spte()
  KVM: vmx: skip VMWRITE of HOST_{FS,GS}_BASE when possible
  KVM: vmx: skip VMWRITE of HOST_{FS,GS}_SEL when possible
  KVM: vmx: always initialize HOST_{FS,GS}_BASE to zero during setup
  KVM: vmx: move struct host_state usage to struct loaded_vmcs
  KVM: vmx: compute need to reload FS/GS/LDT on demand
  KVM: nVMX: remove a misleading comment regarding vmcs02 fields
  KVM: vmx: rename __vmx_load_host_state() and vmx_save_host_state()
  KVM: vmx: add dedicated utility to access guest's kernel_gs_base
  KVM: vmx: track host_state.loaded using a loaded_vmcs pointer
  KVM: vmx: refactor segmentation code in vmx_save_host_state()
  kvm: nVMX: Fix fault priority for VMX operations
  kvm: nVMX: Fix fault vector for VMX operation at CPL > 0
  ...

6 years agoMerge tag 'riscv-for-linus-4.19-mw0' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 19 Aug 2018 16:56:38 +0000 (09:56 -0700)]
Merge tag 'riscv-for-linus-4.19-mw0' of git://git./linux/kernel/git/palmer/riscv-linux

Pull RISC-V updates from Palmer Dabbelt:
 "This contains some major improvements to the RISC-V port, including
  the necessary interrupt controller and timer support to actually make
  it to userspace. Support for three devices has been added:

   - the ISA-mandated timers on RISC-V systems.

   - the ISA-mandated first-level interrupt controller on RISC-V
     systems, which is handled as part of our core arch code because
     it's very small and tightly tied to the ISA.

   - SiFive's platform-level interrupt controller, which talks to the
     actual devices.

  In addition to these new devices, there are a handful of cleanups all
  over the RISC-V tree:

   - build fixes for various configurations:
      * A fix to the vDSO build's makefile so it respects CFLAGS.
      * The addition of __lshrti3, a libgcc derived function necessary
        for some 32-bit configurations.
      * !SMP && PERF_EVENTS

   - Cleanups to the arch code to remove the remnants of old versions of
     the drivers that were just properly submitted.
      * Some dead code from the timer driver, most of which wasn't ever
        even compiled.
      * Cleanups of some interrupt #defines, which are now local to the
        interrupt handling code.

   - Fixes to ptrace(), which while not being sufficient to fully make
     GDB work are at least sufficient to get simple GDB tasks to work.

   - Early printk support via RISC-V's architecturally mandated SBI
     console device.

   - A fix to our early debug trap handler to ensure it's always
     aligned.

  These patches have all been through a fairly extensive review process,
  but as this enables a whole pile of functionality (ie, userspace) I'm
  confident we'll need to submit a few more patches. The only concrete
  issues I know about are the sys_riscv_flush_icache patches, but as I
  managed to screw those up on Friday I figured it'd be best to let them
  bake another week.

  This tag boots a Fedora root filesystem on QEMU's master branch for
  me, and before this morning's rebase (from 4.18-rc8 to 4.18) it booted
  on the HiFive Unleashed.

  Thanks to Christoph Hellwig and the other guys at WD for getting the
  new drivers in shape!"

* tag 'riscv-for-linus-4.19-mw0' of git://git.kernel.org/pub/scm/linux/kernel/git/palmer/riscv-linux:
  dt-bindings: interrupt-controller: SiFive Plaform Level Interrupt Controller
  dt-bindings: interrupt-controller: RISC-V local interrupt controller
  RISC-V: Fix !CONFIG_SMP compilation error
  irqchip: add a SiFive PLIC driver
  RISC-V: Add the directive for alignment of stvec's value
  clocksource: new RISC-V SBI timer driver
  RISC-V: implement low-level interrupt handling
  RISC-V: add a definition for the SIE SEIE bit
  RISC-V: remove INTERRUPT_CAUSE_* defines from asm/irq.h
  RISC-V: simplify software interrupt / IPI code
  RISC-V: remove timer leftovers
  RISC-V: Add early printk support via the SBI console
  RISC-V: Don't increment sepc after breakpoint.
  RISC-V: implement __lshrti3.
  RISC-V: Use KBUILD_CFLAGS instead of KCFLAGS when building the vDSO

6 years agoMerge tag 'char-misc-4.19-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/gregk...
Linus Torvalds [Sun, 19 Aug 2018 16:30:44 +0000 (09:30 -0700)]
Merge tag 'char-misc-4.19-rc1' of git://git./linux/kernel/git/gregkh/char-misc

Pull UIO fix from Greg KH:
 "Here is a single UIO fix that I forgot to send before 4.18-final came
  out. It reverts a UIO patch that went in the 4.18 development window
  that was causing problems.

  This patch has been in linux-next for a while with no problems, I just
  forgot to send it earlier, or as part of the larger char/misc patch
  series from yesterday, my fault"

* tag 'char-misc-4.19-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/char-misc:
  Revert "uio: use request_threaded_irq instead"

6 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input
Linus Torvalds [Sat, 18 Aug 2018 23:48:07 +0000 (16:48 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/dtor/input

Pull input updates from Dmitry Torokhov:

 - a new driver for Rohm BU21029 touch controller

 - new bitmap APIs: bitmap_alloc, bitmap_zalloc and bitmap_free

 - updates to Atmel, eeti. pxrc and iforce drivers

 - assorted driver cleanups and fixes.

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input: (57 commits)
  MAINTAINERS: Add PhoenixRC Flight Controller Adapter
  Input: do not use WARN() in input_alloc_absinfo()
  Input: mark expected switch fall-throughs
  Input: raydium_i2c_ts - use true and false for boolean values
  Input: evdev - switch to bitmap API
  Input: gpio-keys - switch to bitmap_zalloc()
  Input: elan_i2c_smbus - cast sizeof to int for comparison
  bitmap: Add bitmap_alloc(), bitmap_zalloc() and bitmap_free()
  md: Avoid namespace collision with bitmap API
  dm: Avoid namespace collision with bitmap API
  Input: pm8941-pwrkey - add resin entry
  Input: pm8941-pwrkey - abstract register offsets and event code
  Input: iforce - reorganize joystick configuration lists
  Input: atmel_mxt_ts - move completion to after config crc is updated
  Input: atmel_mxt_ts - don't report zero pressure from T9
  Input: atmel_mxt_ts - zero terminate config firmware file
  Input: atmel_mxt_ts - refactor config update code to add context struct
  Input: atmel_mxt_ts - config CRC may start at T71
  Input: atmel_mxt_ts - remove unnecessary debug on ENOMEM
  Input: atmel_mxt_ts - remove duplicate setup of ABS_MT_PRESSURE
  ...

6 years agoMerge tag 'hwlock-v4.19' of git://github.com/andersson/remoteproc
Linus Torvalds [Sat, 18 Aug 2018 23:45:27 +0000 (16:45 -0700)]
Merge tag 'hwlock-v4.19' of git://github.com/andersson/remoteproc

Pull hwspinlock updates from Bjorn Andersson:
 "This introduces devres helpers and an API to request a lock by name,
  then migrates the sprd SPI driver to use these"

* tag 'hwlock-v4.19' of git://github.com/andersson/remoteproc:
  hwspinlock: Fix incorrect return pointers
  spi: sprd: Change to use devm_hwspin_lock_request_specific()
  spi: sprd: Replace of_hwspin_lock_get_id() with of_hwspin_lock_get_id_byname()
  hwspinlock: Fix one comment mistake
  hwspinlock: Remove redundant config
  hwspinlock: Add devm_xxx() APIs to register/unregister one hwlock controller
  hwspinlock: Add devm_xxx() APIs to request/free hwlock
  hwspinlock: Add one new API to support getting a specific hwlock by the name

6 years agoMerge tag 'rpmsg-v4.19' of git://github.com/andersson/remoteproc
Linus Torvalds [Sat, 18 Aug 2018 23:43:57 +0000 (16:43 -0700)]
Merge tag 'rpmsg-v4.19' of git://github.com/andersson/remoteproc

Pull rpmsg updates from Bjorn Andersson:
 "This fixes a few compile and kerneldoc warnings, allows rpmsg devices
  to handle power domains, allow for labeling GLINK edges and supports
  compat for rpmsg_char"

* tag 'rpmsg-v4.19' of git://github.com/andersson/remoteproc:
  rpmsg: Add compat ioctl for rpmsg char driver
  rpmsg: glink: Store edge name for glink device
  dt-bindings: soc: qcom: Add label for GLINK bindings
  rpmsg: core: add support to power domains for devices
  rpmsg: smd: fix kerneldoc warnings
  rpmsg: glink: Fix various kerneldoc warnings.
  rpmsg: glink: correctly annotate intent members
  rpmsg: smd: Add missing include of sizes.h

6 years agoMerge tag 'rproc-v4.19' of git://github.com/andersson/remoteproc
Linus Torvalds [Sat, 18 Aug 2018 23:42:04 +0000 (16:42 -0700)]
Merge tag 'rproc-v4.19' of git://github.com/andersson/remoteproc

Pull remoteproc updates from Bjorn Andersson:
 "This adds support for pre-start and post-shutdown hooks for remoteproc
  subdevices, refactors the Qualcomm Hexagon support to allow reuse
  between several drivers, makes authentication in the MDT file loader
  optional, migrates a few format strings to use %pK and migrates the
  Davinci driver to use the reset framework"

* tag 'rproc-v4.19' of git://github.com/andersson/remoteproc:
  remoteproc/davinci: use the reset framework
  remoteproc/davinci: Mark error recovery as disabled
  remoteproc: st_slim: replace "%p" with "%pK"
  remoteproc: replace "%p" with "%pK"
  remoteproc: qcom: fix Q6V5_WCSS dependencies
  remoteproc: Reset table_ptr in rproc_start() failure paths
  remoteproc: qcom: q6v5-pil: fix modem hang on SDM845 after axis2 clk unvote
  remoteproc: qcom q6v5: fix modular build
  remoteproc: Introduce prepare and unprepare for subdevices
  remoteproc: rename subdev probe and remove functions
  remoteproc: Make client initialize ops in rproc_subdev
  remoteproc: Make start and stop in subdev optional
  remoteproc: Rename subdev functions to start/stop
  remoteproc: qcom: Introduce Hexagon V5 based WCSS driver
  remoteproc: qcom: q6v5-pil: Use common q6v5 helpers
  remoteproc: qcom: adsp: Use common q6v5 helpers
  remoteproc: q6v5: Extract common resource handling
  remoteproc: qcom: mdt_loader: Make the firmware authentication optional

6 years agoMerge tag 'linux-watchdog-4.19-rc1' of git://www.linux-watchdog.org/linux-watchdog
Linus Torvalds [Sat, 18 Aug 2018 23:16:57 +0000 (16:16 -0700)]
Merge tag 'linux-watchdog-4.19-rc1' of git://linux-watchdog.org/linux-watchdog

Pull watchdog updates from Wim Van Sebroeck:

 - add MEN 16z069 IP-Core driver

 - renesas-wdt: add support for the R8A77990 wdt

 - stm32_iwdg: Add stm32mp1 support and pclk feature

 - sp805_wdt, orion_wdt, sprd_wdt: several improvements

 - imx2_wdt, stmp3xxx: switch to SPDX identifier

* tag 'linux-watchdog-4.19-rc1' of git://www.linux-watchdog.org/linux-watchdog:
  watchdog: fix dependencies of menz69_wdt.o
  watchdog: sp805: Add clock-frequency property
  watchdog: add driver for the MEN 16z069 IP-Core
  watchdog: sprd_wdt: Remove redundant dev_err call in sprd_wdt_probe()
  watchdog: stmp3xxx: Switch to SPDX identifier
  watchdog: imx2_wdt: Switch to SPDX identifier
  watchdog: sp805: set WDOG_HW_RUNNING when appropriate
  watchdog: sp805: add 'timeout-sec' DT property support
  dt-bindings: watchdog: Add optional 'timeout-sec' property for sp805
  dt-bindings: watchdog: Consolidate SP805 binding docs
  watchdog: orion_wdt: Mark watchdog as active when running at probe
  watchdog: stm32: add pclk feature for stm32mp1
  dt-bindings: watchdog: add stm32mp1 support
  dt-bindings: watchdog: renesas-wdt: Add support for the R8A77990 wdt

6 years agoMerge tag 'dmaengine-4.19-rc1' of git://git.infradead.org/users/vkoul/slave-dma
Linus Torvalds [Sat, 18 Aug 2018 22:55:59 +0000 (15:55 -0700)]
Merge tag 'dmaengine-4.19-rc1' of git://git.infradead.org/users/vkoul/slave-dma

Pull DMAengine updates from Vinod Koul:
 "This round brings couple of framework changes, a new driver and usual
  driver updates:

   - new managed helper for dmaengine framework registration

   - split dmaengine pause capability to pause and resume and allow
     drivers to report that individually

   - update dma_request_chan_by_mask() to handle deferred probing

   - move imx-sdma to use virt-dma

   - new driver for Actions Semi Owl family S900 controller

   - minor updates to intel, renesas, mv_xor, pl330 etc"

* tag 'dmaengine-4.19-rc1' of git://git.infradead.org/users/vkoul/slave-dma: (46 commits)
  dmaengine: Add Actions Semi Owl family S900 DMA driver
  dt-bindings: dmaengine: Add binding for Actions Semi Owl SoCs
  dmaengine: sh: rcar-dmac: Should not stop the DMAC by rcar_dmac_sync_tcr()
  dmaengine: mic_x100_dma: use the new helper to simplify the code
  dmaengine: add a new helper dmaenginem_async_device_register
  dmaengine: imx-sdma: add memcpy interface
  dmaengine: imx-sdma: add SDMA_BD_MAX_CNT to replace '0xffff'
  dmaengine: dma_request_chan_by_mask() to handle deferred probing
  dmaengine: pl330: fix irq race with terminate_all
  dmaengine: Revert "dmaengine: mv_xor_v2: enable COMPILE_TEST"
  dmaengine: mv_xor_v2: use {lower,upper}_32_bits to configure HW descriptor address
  dmaengine: mv_xor_v2: enable COMPILE_TEST
  dmaengine: mv_xor_v2: move unmap to before callback
  dmaengine: mv_xor_v2: convert callback to helper function
  dmaengine: mv_xor_v2: kill the tasklets upon exit
  dmaengine: mv_xor_v2: explicitly freeup irq
  dmaengine: sh: rcar-dmac: Add dma_pause operation
  dmaengine: sh: rcar-dmac: add a new function to clear CHCR.DE with barrier
  dmaengine: idma64: Support dmaengine_terminate_sync()
  dmaengine: hsu: Support dmaengine_terminate_sync()
  ...