cascardo/linux.git
7 years agocheckpatch: improve 'bare use of' signed/unsigned types warning
Joe Perches [Tue, 2 Aug 2016 21:04:42 +0000 (14:04 -0700)]
checkpatch: improve 'bare use of' signed/unsigned types warning

Fix false positive warning of identifiers ending in signed with an =
assignment of WARNING: Prefer 'signed int' to bare use of 'signed'.

Link: http://lkml.kernel.org/r/6a0e24c3e9102337528ecfcbbe91a0eb5b4820ed.1469529497.git.joe@perches.com
Signed-off-by: Joe Perches <joe@perches.com>
Reported-by: Alan Douglas <alanjhd@gmail.com>
Acked-by: Andy Whitcroft <apw@canonical.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agocheckpatch: don't complain about BIT macro in uapi
Tomas Winkler [Tue, 2 Aug 2016 21:04:39 +0000 (14:04 -0700)]
checkpatch: don't complain about BIT macro in uapi

BIT macro cannot be exported to UAPI, don't complain about it.

Link: http://lkml.kernel.org/r/1468707033-16173-1-git-send-email-tomas.winkler@intel.com
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Acked-by: Joe Perches <joe@perches.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agocheckpatch: yet another commit id improvement
Joe Perches [Tue, 2 Aug 2016 21:04:36 +0000 (14:04 -0700)]
checkpatch: yet another commit id improvement

Using \b isn't good enough to isolate what appears to be a commit id in
a commit message.

Make sure there is a space or a quote like character after a continuous
run of hexadecimal characters that could be a commit id.

Link: http://lkml.kernel.org/r/fdd22b47463a21c21132edbb8aa35e372950a1e6.1468869915.git.joe@perches.com
Signed-off-by: Joe Perches <joe@perches.com>
Cc: "Zhuo, Qiuxu" <qiuxu.zhuo@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agocheckpatch: allow c99 style // comments
Joe Perches [Tue, 2 Aug 2016 21:04:33 +0000 (14:04 -0700)]
checkpatch: allow c99 style // comments

Sanitise the lines that contain c99 comments so that the error doesn't
get emitted.

Link: http://lkml.kernel.org/r/d4d22c34ad7bcc1bceb52f0742f76b7a6d585235.1468368420.git.joe@perches.com
Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agocheckpatch: skip long lines that use an EFI_GUID macro
Joe Perches [Tue, 2 Aug 2016 21:04:31 +0000 (14:04 -0700)]
checkpatch: skip long lines that use an EFI_GUID macro

These are also possible single line uses that exceed the generic maximum
line length (typically 80 columns)

Link: http://lkml.kernel.org/r/32a6a85fbd6161f1bb55ce176a464e44591afc5b.1468368420.git.joe@perches.com
Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agofirmware: support loading into a pre-allocated buffer
Stephen Boyd [Tue, 2 Aug 2016 21:04:28 +0000 (14:04 -0700)]
firmware: support loading into a pre-allocated buffer

Some systems are memory constrained but they need to load very large
firmwares.  The firmware subsystem allows drivers to request this
firmware be loaded from the filesystem, but this requires that the
entire firmware be loaded into kernel memory first before it's provided
to the driver.  This can lead to a situation where we map the firmware
twice, once to load the firmware into kernel memory and once to copy the
firmware into the final resting place.

This creates needless memory pressure and delays loading because we have
to copy from kernel memory to somewhere else.  Let's add a
request_firmware_into_buf() API that allows drivers to request firmware
be loaded directly into a pre-allocated buffer.  This skips the
intermediate step of allocating a buffer in kernel memory to hold the
firmware image while it's read from the filesystem.  It also requires
that drivers know how much memory they'll require before requesting the
firmware and negates any benefits of firmware caching because the
firmware layer doesn't manage the buffer lifetime.

For a 16MB buffer, about half the time is spent performing a memcpy from
the buffer to the final resting place.  I see loading times go from
0.081171 seconds to 0.047696 seconds after applying this patch.  Plus
the vmalloc pressure is reduced.

This is based on a patch from Vikram Mulukutla on codeaurora.org:
  https://www.codeaurora.org/cgit/quic/la/kernel/msm-3.18/commit/drivers/base/firmware_class.c?h=rel/msm-3.18&id=0a328c5f6cd999f5c591f172216835636f39bcb5

Link: http://lkml.kernel.org/r/20160607164741.31849-4-stephen.boyd@linaro.org
Signed-off-by: Stephen Boyd <stephen.boyd@linaro.org>
Cc: Mimi Zohar <zohar@linux.vnet.ibm.com>
Cc: Vikram Mulukutla <markivx@codeaurora.org>
Cc: Mark Brown <broonie@kernel.org>
Cc: Ming Lei <ming.lei@canonical.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agofirmware: provide infrastructure to make fw caching optional
Vikram Mulukutla [Tue, 2 Aug 2016 21:04:25 +0000 (14:04 -0700)]
firmware: provide infrastructure to make fw caching optional

Some low memory systems with complex peripherals cannot afford to have
the relatively large firmware images taking up valuable memory during
suspend and resume.  Change the internal implementation of
firmware_class to disallow caching based on a configurable option.  In
the near future, variants of request_firmware will take advantage of
this feature.

Link: http://lkml.kernel.org/r/20160607164741.31849-3-stephen.boyd@linaro.org
[stephen.boyd@linaro.org: Drop firmware_desc design and use flags]
Signed-off-by: Vikram Mulukutla <markivx@codeaurora.org>
Signed-off-by: Stephen Boyd <stephen.boyd@linaro.org>
Cc: Mimi Zohar <zohar@linux.vnet.ibm.com>
Cc: Mark Brown <broonie@kernel.org>
Cc: Ming Lei <ming.lei@canonical.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agofirmware: consolidate kmap/read/write logic
Stephen Boyd [Tue, 2 Aug 2016 21:04:22 +0000 (14:04 -0700)]
firmware: consolidate kmap/read/write logic

Some systems are memory constrained but they need to load very large
firmwares.  The firmware subsystem allows drivers to request this
firmware be loaded from the filesystem, but this requires that the
entire firmware be loaded into kernel memory first before it's provided
to the driver.  This can lead to a situation where we map the firmware
twice, once to load the firmware into kernel memory and once to copy the
firmware into the final resting place.

This design creates needless memory pressure and delays loading because
we have to copy from kernel memory to somewhere else.  This patch sets
adds support to the request firmware API to load the firmware directly
into a pre-allocated buffer, skipping the intermediate copying step and
alleviating memory pressure during firmware loading.  The drawback is
that we can't use the firmware caching feature because the memory for
the firmware cache is not managed by the firmware layer.

This patch (of 3):

We use similar structured code to read and write the kmapped firmware
pages.  The only difference is read copies from the kmap region and
write copies to it.  Consolidate this into one function to reduce
duplication.

Link: http://lkml.kernel.org/r/20160607164741.31849-2-stephen.boyd@linaro.org
Signed-off-by: Stephen Boyd <stephen.boyd@linaro.org>
Cc: Vikram Mulukutla <markivx@codeaurora.org>
Cc: Mimi Zohar <zohar@linux.vnet.ibm.com>
Cc: Mark Brown <broonie@kernel.org>
Cc: Ming Lei <ming.lei@canonical.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agoradix-tree: fix comment about "exceptional" bits
Ross Zwisler [Tue, 2 Aug 2016 21:04:19 +0000 (14:04 -0700)]
radix-tree: fix comment about "exceptional" bits

The bottom two bits of radix tree entries are reserved for special use
by the radix tree code itself.  A comment detailing their usage was
added by commit 3bcadd6fa6c4 ("radix-tree: free up the bottom bit of
exceptional entries for reuse")

This comment states that if the bottom two bits are '11', this means
that this is a locked exceptional entry.

It turns out that this bit combination was never actually used.  Radix
tree locking for DAX was indeed implemented, but it actually used the
third LSB:

  /* We use lowest available exceptional entry bit for locking */
  #define RADIX_DAX_ENTRY_LOCK (1 << RADIX_TREE_EXCEPTIONAL_SHIFT)

This locking code was also made specific to the DAX code instead of
being generally implemented in radix-tree.h.

So, fix the comment.

Link: http://lkml.kernel.org/r/1468997731-2155-1-git-send-email-ross.zwisler@linux.intel.com
Signed-off-by: Ross Zwisler <ross.zwisler@linux.intel.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Jan Kara <jack@suse.com>
Cc: Konstantin Khlebnikov <koct9i@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agocrc32: use ktime_get_ns() for measurement
Arnd Bergmann [Tue, 2 Aug 2016 21:04:16 +0000 (14:04 -0700)]
crc32: use ktime_get_ns() for measurement

The crc32 test function measures the elapsed time in nanoseconds, but
uses 'struct timespec' for that.  We want to remove timespec from the
kernel for y2038 compatibility, and ktime_get_ns() also helps make the
code simpler here.

It is also slightly better to use monontonic time, as we are only
interested in the time difference.

Link: http://lkml.kernel.org/r/20160617143932.3289626-1-arnd@arndb.de
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Cc: "David S . Miller" <davem@davemloft.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agolib/iommu-helper: skip to next segment
Sebastian Ott [Tue, 2 Aug 2016 21:04:13 +0000 (14:04 -0700)]
lib/iommu-helper: skip to next segment

When a large enough area in the iommu bitmap is found but would span a
boundary we continue the search starting from the next bit position.
For large allocations this can lead to several useless invocations of
bitmap_find_next_zero_area() and iommu_is_span_boundary().

Continue the search from the start of the next segment (which is the
next bit position such that we'll not cross the same segment boundary
again).

Link: http://lkml.kernel.org/r/alpine.LFD.2.20.1606081910070.3211@schleppi
Signed-off-by: Sebastian Ott <sebott@linux.vnet.ibm.com>
Reviewed-by: Gerald Schaefer <gerald.schaefer@de.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agoget_maintainer.pl: reduce need for command-line option -f
Joe Perches [Tue, 2 Aug 2016 21:04:10 +0000 (14:04 -0700)]
get_maintainer.pl: reduce need for command-line option -f

If a vcs is used, look to see if the vcs tracks the file specified and
so the -f option becomes optional.

Link: http://lkml.kernel.org/r/7c86a8df0d48770c45778a43b6b3e4627b2a90ee.1469746395.git.joe@perches.com
Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agoprintk: add kernel parameter to control writes to /dev/kmsg
Borislav Petkov [Tue, 2 Aug 2016 21:04:07 +0000 (14:04 -0700)]
printk: add kernel parameter to control writes to /dev/kmsg

Add a "printk.devkmsg" kernel command line parameter which controls how
userspace writes into /dev/kmsg.  It has three options:

 * ratelimit - ratelimit logging from userspace.
 * on  - unlimited logging from userspace
 * off - logging from userspace gets ignored

The default setting is to ratelimit the messages written to it.

This changes the kernel default setting of "on" to "ratelimit" and we do
that because we want to keep userspace spamming /dev/kmsg to sane
levels.  This is especially moot when a small kernel log buffer wraps
around and messages get lost.  So the ratelimiting setting should be a
sane setting where kernel messages should have a bit higher chance of
survival from all the spamming.

It additionally does not limit logging to /dev/kmsg while the system is
booting if we haven't disabled it on the command line.

Furthermore, we can control the logging from a lower priority sysctl
interface - kernel.printk_devkmsg.

That interface will succeed only if printk.devkmsg *hasn't* been
supplied on the command line.  If it has, then printk.devkmsg is a
one-time setting which remains for the duration of the system lifetime.
This "locking" of the setting is to prevent userspace from changing the
logging on us through sysctl(2).

This patch is based on previous patches from Linus and Steven.

[bp@suse.de: fixes]
Link: http://lkml.kernel.org/r/20160719072344.GC25563@nazgul.tnic
Link: http://lkml.kernel.org/r/20160716061745.15795-3-bp@alien8.de
Signed-off-by: Borislav Petkov <bp@suse.de>
Cc: Dave Young <dyoung@redhat.com>
Cc: Franck Bui <fbui@suse.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agoratelimit: extend to print suppressed messages on release
Borislav Petkov [Tue, 2 Aug 2016 21:04:04 +0000 (14:04 -0700)]
ratelimit: extend to print suppressed messages on release

Extend the ratelimiting facility to print the amount of suppressed lines
when it is being released.

This use case is aimed at short-termed, burst-like users for which we
want to output the suppressed lines stats only once, after it has been
disposed of.  For an example, see /dev/kmsg usage in a follow-on patch.

Also, change the printk() line we issue on release to not use
"callbacks" as it is misleading: we're not suppressing callbacks but
printk() calls.

This has been separated from a previous patch by Linus.

Link: http://lkml.kernel.org/r/20160716061745.15795-2-bp@alien8.de
Signed-off-by: Borislav Petkov <bp@suse.de>
Cc: Dave Young <dyoung@redhat.com>
Cc: Franck Bui <fbui@suse.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agofbdev/bfin_adv7393fb: move DRIVER_NAME before its first use
Borislav Petkov [Tue, 2 Aug 2016 21:04:01 +0000 (14:04 -0700)]
fbdev/bfin_adv7393fb: move DRIVER_NAME before its first use

Move the DRIVER_NAME macro definition before the first usage site and
fix build error.

Link: http://lkml.kernel.org/r/20160801163937.GA28119@nazgul.tnic
Signed-off-by: Borislav Petkov <bp@suse.de>
Reported-by: kbuild test robot <fengguang.wu@intel.com>
Cc: Jean-Christophe Plagniol-Villard <plagnioj@jcrosoft.com>
Cc: Tomi Valkeinen <tomi.valkeinen@ti.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agoprintk: include <asm/sections.h> instead of <asm-generic/sections.h>
Christoph Hellwig [Tue, 2 Aug 2016 21:03:59 +0000 (14:03 -0700)]
printk: include <asm/sections.h> instead of <asm-generic/sections.h>

asm-generic headers are generic implementations for architecture
specific code and should not be included by common code.  Thus use the
asm/ version of sections.h to get at the linker sections.

Link: http://lkml.kernel.org/r/1468285008-7331-1-git-send-email-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agoprintk: introduce suppress_message_printing()
Sergey Senozhatsky [Tue, 2 Aug 2016 21:03:56 +0000 (14:03 -0700)]
printk: introduce suppress_message_printing()

Messages' levels and console log level are inspected when the actual
printing occurs, which may provoke console_unlock() and
console_cont_flush() to waste CPU cycles on every message that has
loglevel above the current console_loglevel.

Schematically, console_unlock() does the following:

console_unlock()
{
        ...
        for (;;) {
                ...
                raw_spin_lock_irqsave(&logbuf_lock, flags);
skip:
                msg = log_from_idx(console_idx);

                if (msg->flags & LOG_NOCONS) {
                        ...
                        goto skip;
                }

                level = msg->level;
                len += msg_print_text();                        >> sprintfs
                                                                   memcpy,
                                                                   etc.

                if (nr_ext_console_drivers) {
                        ext_len = msg_print_ext_header();       >> scnprintf
                        ext_len += msg_print_ext_body();        >> scnprintfs
                                                                   etc.
                }
                ...
                raw_spin_unlock(&logbuf_lock);

                call_console_drivers(level, ext_text, ext_len, text, len)
                {
                        if (level >= console_loglevel &&        >> drop the message
                                        !ignore_loglevel)
                                return;

                        console->write(...);
                }

                local_irq_restore(flags);
        }
        ...
}

The thing here is this deferred `level >= console_loglevel' check.  We
are wasting CPU cycles on sprintfs/memcpy/etc.  preparing the messages
that we will eventually drop.

This can be huge when we register a new CON_PRINTBUFFER console, for
instance.  For every such a console register_console() resets the

        console_seq, console_idx, console_prev

and sets a `exclusive console' pointer to replay the log buffer to that
just-registered console.  And there can be a lot of messages to replay,
in the worst case most of which can be dropped after console_loglevel
test.

We know messages' levels long before we call msg_print_text() and
friends, so we can just move console_loglevel check out of
call_console_drivers() and format a new message only if we are sure that
it won't be dropped.

The patch factors out loglevel check into suppress_message_printing()
function and tests message->level and console_loglevel before formatting
functions in console_unlock() and console_cont_flush() are getting
executed.  This improves things not only for exclusive CON_PRINTBUFFER
consoles, but for every console_unlock() that attempts to print a
message of level above the console_loglevel.

Link: http://lkml.kernel.org/r/20160627135012.8229-1-sergey.senozhatsky@gmail.com
Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
Reviewed-by: Petr Mladek <pmladek@suse.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: Jan Kara <jack@suse.cz>
Cc: Calvin Owens <calvinowens@fb.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agoprintk: create pr_<level> functions
Joe Perches [Tue, 2 Aug 2016 21:03:53 +0000 (14:03 -0700)]
printk: create pr_<level> functions

Using functions instead of macros can reduce overall code size by
eliminating unnecessary "KERN_SOH<digit>" prefixes from format strings.

  defconfig x86-64:

  $ size vmlinux*
     text    data     bss      dec     hex  filename
  10193570 4331464 1105920 15630954  ee826a vmlinux.new
  10192623 4335560 1105920 15634103  ee8eb7 vmlinux.old

As the return value are unimportant and unused in the kernel tree, these
new functions return void.

Miscellanea:

 - change pr_<level> macros to call new __pr_<level> functions
 - change vprintk_nmi and vprintk_default to add LOGLEVEL_<level> argument

[akpm@linux-foundation.org: fix LOGLEVEL_INFO, per Joe]
Link: http://lkml.kernel.org/r/e16cc34479dfefcae37c98b481e6646f0f69efc3.1466718827.git.joe@perches.com
Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agoprintk: do not include interrupt.h
Sergey Senozhatsky [Tue, 2 Aug 2016 21:03:50 +0000 (14:03 -0700)]
printk: do not include interrupt.h

A trivial cosmetic change: interrupt.h header is redundant since commit
6b898c07cb1d ("console: use might_sleep in console_lock").

Link: http://lkml.kernel.org/r/20160620132847.21930-1-sergey.senozhatsky@gmail.com
Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agodynamic_debug: only add header when used
Luis de Bethencourt [Tue, 2 Aug 2016 21:03:47 +0000 (14:03 -0700)]
dynamic_debug: only add header when used

kernel.h header doesn't directly use dynamic debug, instead we can
include it in module.c (which used it via kernel.h).  printk.h only uses
it if CONFIG_DYNAMIC_DEBUG is on, changing the inclusion to only happen
in that case.

Link: http://lkml.kernel.org/r/1468429793-16917-1-git-send-email-luisbg@osg.samsung.com
[luisbg@osg.samsung.com: include dynamic_debug.h in drb_int.h]
Link: http://lkml.kernel.org/r/1468447828-18558-2-git-send-email-luisbg@osg.samsung.com
Signed-off-by: Luis de Bethencourt <luisbg@osg.samsung.com>
Cc: Rusty Russell <rusty@rustcorp.com.au>
Cc: Hidehiro Kawai <hidehiro.kawai.ez@hitachi.com>
Cc: Borislav Petkov <bp@suse.de>
Cc: Michal Nazarewicz <mina86@mina86.com>
Cc: Rasmus Villemoes <linux@rasmusvillemoes.dk>
Cc: Joe Perches <joe@perches.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agotask_work: use READ_ONCE/lockless_dereference, avoid pi_lock if !task_works
Oleg Nesterov [Tue, 2 Aug 2016 21:03:44 +0000 (14:03 -0700)]
task_work: use READ_ONCE/lockless_dereference, avoid pi_lock if !task_works

Change task_work_cancel() to use lockless_dereference(), this is what
the code really wants but we didn't have this helper when it was
written.

Also add the fast-path task->task_works == NULL check, in the likely
case this task has no pending works and we can avoid
spin_lock(task->pi_lock).

While at it, change other users of ACCESS_ONCE() to use READ_ONCE().

Link: http://lkml.kernel.org/r/20160610150042.GA13868@redhat.com
Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Cc: Andrea Parri <parri.andrea@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agoinclude: mman: use bool instead of int for the return value of arch_validate_prot
Chen Gang [Tue, 2 Aug 2016 21:03:42 +0000 (14:03 -0700)]
include: mman: use bool instead of int for the return value of arch_validate_prot

For pure bool function's return value, bool is a little better more or
less than int.

Link: http://lkml.kernel.org/r/1469331815-2026-1-git-send-email-chengang@emindsoft.com.cn
Signed-off-by: Chen Gang <gang.chen.5i5j@gmail.com>
Acked-by: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agomailmap: add Linus Lüssing
Linus Lüssing [Tue, 2 Aug 2016 21:03:39 +0000 (14:03 -0700)]
mailmap: add Linus Lüssing

For one thing, summarizes all non-umlaut versions into the umlaut one
(Linus Luessing -> Linus Lüssing).

For another, maps obsolete email addresses to the current @c0d3.blue
one.

Link: http://lkml.kernel.org/r/1467805371-2773-1-git-send-email-linus.luessing@c0d3.blue
Signed-off-by: Linus Lüssing <linus.luessing@c0d3.blue>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agouapi: move forward declarations of internal structures
Alexey Dobriyan [Tue, 2 Aug 2016 21:03:36 +0000 (14:03 -0700)]
uapi: move forward declarations of internal structures

Don't user forward declarations of internal kernel structures in headers
exported to userspace.

Move "struct completion;".
Move "struct task_struct;".

Link: http://lkml.kernel.org/r/20160713215808.GA22486@p183.telecom.by
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agotreewide: replace obsolete _refok by __ref
Fabian Frederick [Tue, 2 Aug 2016 21:03:33 +0000 (14:03 -0700)]
treewide: replace obsolete _refok by __ref

There was only one use of __initdata_refok and __exit_refok

__init_refok was used 46 times against 82 for __ref.

Those definitions are obsolete since commit 312b1485fb50 ("Introduce new
section reference annotations tags: __ref, __refdata, __refconst")

This patch removes the following compatibility definitions and replaces
them treewide.

/* compatibility defines */
#define __init_refok     __ref
#define __initdata_refok __refdata
#define __exit_refok     __ref

I can also provide separate patches if necessary.
(One patch per tree and check in 1 month or 2 to remove old definitions)

[akpm@linux-foundation.org: coding-style fixes]
Link: http://lkml.kernel.org/r/1466796271-3043-1-git-send-email-fabf@skynet.be
Signed-off-by: Fabian Frederick <fabf@skynet.be>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agomemstick: don't allocate unused major for ms_block
NeilBrown [Tue, 2 Aug 2016 21:03:30 +0000 (14:03 -0700)]
memstick: don't allocate unused major for ms_block

When alloc_disk(0) is used the ->major number is completely ignored.
All devices are allocated with a major of BLOCK_EXT_MAJOR.

So remove registration and deregistration of 'major'.

Link: http://lkml.kernel.org/r/20160602064318.4403.49955.stgit@noble
Signed-off-by: NeilBrown <neilb@suse.com>
Cc: Keith Busch <keith.busch@intel.com>
Cc: Jens Axboe <axboe@fb.com>
Cc: Maxim Levitsky <maximlevitsky@gmail.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agoinit/Kconfig: make COMPILE_TEST depend on !UML
Richard Weinberger [Tue, 2 Aug 2016 21:03:27 +0000 (14:03 -0700)]
init/Kconfig: make COMPILE_TEST depend on !UML

UML is a bit special since it does not have iomem nor dma.  That means a
lot of drivers will not build if they miss a dependency on HAS_IOMEM.
s390 used to have the same issues but since it gained PCI support UML is
the only stranger.

We are tired of patching dozens of new drivers after every merge window
just to un-break allmod/yesconfig UML builds.  One could argue that a
decent driver has to know on what it depends and therefore a missing
HAS_IOMEM dependency is a clear driver bug.  But the dependency not
obvious and not everyone does UML builds with COMPILE_TEST enabled when
developing a device driver.

A possible solution to make these builds succeed on UML would be
providing stub functions for ioremap() and friends which fail upon
runtime.  Another one is simply disabling COMPILE_TEST for UML.  Since
it is the least hassle and does not force use to fake iomem support
let's do the latter.

Link: http://lkml.kernel.org/r/1466152995-28367-1-git-send-email-richard@nod.at
Signed-off-by: Richard Weinberger <richard@nod.at>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agofs/proc/task_mmu.c: suppress compilation warnings with W=1
Valdis Kletnieks [Tue, 2 Aug 2016 21:03:25 +0000 (14:03 -0700)]
fs/proc/task_mmu.c: suppress compilation warnings with W=1

Suppress a bunch of warnings of the form:

  fs/proc/task_mmu.c: In function 'show_smap_vma_flags':
  fs/proc/task_mmu.c:635:22: warning: initialized field overwritten [-Wt override-init]
     [ilog2(VM_READ)] = "rd",
                        ^~~~
  fs/proc/task_mmu.c:635:22: note: (near initialization for 'mnemonics[0]')

They happen because of the way we intentionally build the table, so
silence the warning when building with 'make W=1'.

Link: http://lkml.kernel.org/r/8727.1470022083@turing-police.cc.vt.edu
Signed-off-by: Valdis Kletnieks <valdis.kletnieks@vt.edu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agoprocfs: avoid 32-bit time_t in /proc/*/stat
Arnd Bergmann [Tue, 2 Aug 2016 21:03:22 +0000 (14:03 -0700)]
procfs: avoid 32-bit time_t in /proc/*/stat

/proc/stat shows (among lots of other things) the current boottime (i.e.
number of seconds since boot).  While a 32-bit number is sufficient for
this particular case, we want to get rid of the 'struct timespec'
suffers from a 32-bit overflow in 2038.

This changes the code to use a struct timespec64, which is known to be
safe in all cases.

Link: http://lkml.kernel.org/r/20160617201247.2292101-1-arnd@arndb.de
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agoproc_oom_score: remove tasklist_lock and pid_alive()
Oleg Nesterov [Tue, 2 Aug 2016 21:03:19 +0000 (14:03 -0700)]
proc_oom_score: remove tasklist_lock and pid_alive()

This was needed before to ensure that ->signal != 0 and do_each_thread()
is safe, see commit b95c35e76b29b ("oom: fix the unsafe usage of
badness() in proc_oom_score()") for details.

Today tsk->signal can't go away and for_each_thread(tsk) is always safe.

Link: http://lkml.kernel.org/r/20160608211921.GA15508@redhat.com
Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Acked-by: David Rientjes <rientjes@google.com>
Acked-by: Michal Hocko <mhocko@suse.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>
7 years agoMAINTAINERS: befs: add new maintainers
Luis de Bethencourt [Tue, 2 Aug 2016 21:03:16 +0000 (14:03 -0700)]
MAINTAINERS: befs: add new maintainers

Salah Triki and Luis de Bethencourt are taking over maintainership of
befs.

Link: http://lkml.kernel.org/r/1469651079-32455-1-git-send-email-luisbg@osg.samsung.com
Signed-off-by: Luis de Bethencourt <luisbg@osg.samsung.com>
Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Acked-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agocgroup: update cgroup's document path
seokhoon.yoon [Tue, 2 Aug 2016 21:03:13 +0000 (14:03 -0700)]
cgroup: update cgroup's document path

cgroup's document path is changed to "cgroup-v1".  update it.

Link: http://lkml.kernel.org/r/1470148443-6509-1-git-send-email-iamyooon@gmail.com
Signed-off-by: seokhoon.yoon <iamyooon@gmail.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agoUBSAN: fix typo in format string
Nicolas Iooss [Tue, 2 Aug 2016 21:03:10 +0000 (14:03 -0700)]
UBSAN: fix typo in format string

handle_object_size_mismatch() used %pk to format a kernel pointer with
pr_err().  This seemed to be a misspelling for %pK, but using this to
format a kernel pointer does not make much sence here.

Therefore use %p instead, like in handle_missaligned_access().

Link: http://lkml.kernel.org/r/20160730083010.11569-1-nicolas.iooss_linux@m4x.org
Signed-off-by: Nicolas Iooss <nicolas.iooss_linux@m4x.org>
Acked-by: Andrey Ryabinin <aryabinin@virtuozzo.com>
Cc: Joe Perches <joe@perches.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agosysv, ipc: fix security-layer leaking
Fabian Frederick [Tue, 2 Aug 2016 21:03:07 +0000 (14:03 -0700)]
sysv, ipc: fix security-layer leaking

Commit 53dad6d3a8e5 ("ipc: fix race with LSMs") updated ipc_rcu_putref()
to receive rcu freeing function but used generic ipc_rcu_free() instead
of msg_rcu_free() which does security cleaning.

Running LTP msgsnd06 with kmemleak gives the following:

  cat /sys/kernel/debug/kmemleak

  unreferenced object 0xffff88003c0a11f8 (size 8):
    comm "msgsnd06", pid 1645, jiffies 4294672526 (age 6.549s)
    hex dump (first 8 bytes):
      1b 00 00 00 01 00 00 00                          ........
    backtrace:
      kmemleak_alloc+0x23/0x40
      kmem_cache_alloc_trace+0xe1/0x180
      selinux_msg_queue_alloc_security+0x3f/0xd0
      security_msg_queue_alloc+0x2e/0x40
      newque+0x4e/0x150
      ipcget+0x159/0x1b0
      SyS_msgget+0x39/0x40
      entry_SYSCALL_64_fastpath+0x13/0x8f

Manfred Spraul suggested to fix sem.c as well and Davidlohr Bueso to
only use ipc_rcu_free in case of security allocation failure in newary()

Fixes: 53dad6d3a8e ("ipc: fix race with LSMs")
Link: http://lkml.kernel.org/r/1470083552-22966-1-git-send-email-fabf@skynet.be
Signed-off-by: Fabian Frederick <fabf@skynet.be>
Cc: Davidlohr Bueso <dbueso@suse.de>
Cc: Manfred Spraul <manfred@colorfullife.com>
Cc: <stable@vger.kernel.org> [3.12+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agomm: vmscan: fix memcg-aware shrinkers not called on global reclaim
Vladimir Davydov [Tue, 2 Aug 2016 21:03:04 +0000 (14:03 -0700)]
mm: vmscan: fix memcg-aware shrinkers not called on global reclaim

We must call shrink_slab() for each memory cgroup on both global and
memcg reclaim in shrink_node_memcg().  Commit d71df22b55099 accidentally
changed that so that now shrink_slab() is only called with memcg != NULL
on memcg reclaim.  As a result, memcg-aware shrinkers (including
dentry/inode) are never invoked on global reclaim.  Fix that.

Fixes: b2e18757f2c9 ("mm, vmscan: begin reclaiming pages on a per-node basis")
Link: http://lkml.kernel.org/r/1470056590-7177-1-git-send-email-vdavydov@virtuozzo.com
Signed-off-by: Vladimir Davydov <vdavydov@virtuozzo.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Hillf Danton <hillf.zj@alibaba-inc.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@surriel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agoradix-tree: account nodes to memcg only if explicitly requested
Vladimir Davydov [Tue, 2 Aug 2016 21:03:01 +0000 (14:03 -0700)]
radix-tree: account nodes to memcg only if explicitly requested

Radix trees may be used not only for storing page cache pages, so
unconditionally accounting radix tree nodes to the current memory cgroup
is bad: if a radix tree node is used for storing data shared among
different cgroups we risk pinning dead memory cgroups forever.

So let's only account radix tree nodes if it was explicitly requested by
passing __GFP_ACCOUNT to INIT_RADIX_TREE.  Currently, we only want to
account page cache entries, so mark mapping->page_tree so.

Fixes: 58e698af4c63 ("radix-tree: account radix_tree_node to memory cgroup")
Link: http://lkml.kernel.org/r/1470057188-7864-1-git-send-email-vdavydov@virtuozzo.com
Signed-off-by: Vladimir Davydov <vdavydov@virtuozzo.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: <stable@vger.kernel.org> [4.6+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agokasan: avoid overflowing quarantine size on low memory systems
Alexander Potapenko [Tue, 2 Aug 2016 21:02:58 +0000 (14:02 -0700)]
kasan: avoid overflowing quarantine size on low memory systems

If the total amount of memory assigned to quarantine is less than the
amount of memory assigned to per-cpu quarantines, |new_quarantine_size|
may overflow.  Instead, set it to zero.

[akpm@linux-foundation.org: cleanup: use WARN_ONCE return value]
Link: http://lkml.kernel.org/r/1470063563-96266-1-git-send-email-glider@google.com
Fixes: 55834c59098d ("mm: kasan: initial memory quarantine implementation")
Signed-off-by: Alexander Potapenko <glider@google.com>
Reported-by: Dmitry Vyukov <dvyukov@google.com>
Cc: Andrey Ryabinin <aryabinin@virtuozzo.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agokasan: improve double-free reports
Andrey Ryabinin [Tue, 2 Aug 2016 21:02:55 +0000 (14:02 -0700)]
kasan: improve double-free reports

Currently we just dump stack in case of double free bug.
Let's dump all info about the object that we have.

[aryabinin@virtuozzo.com: change double free message per Alexander]
Link: http://lkml.kernel.org/r/1470153654-30160-1-git-send-email-aryabinin@virtuozzo.com
Link: http://lkml.kernel.org/r/1470062715-14077-6-git-send-email-aryabinin@virtuozzo.com
Signed-off-by: Andrey Ryabinin <aryabinin@virtuozzo.com>
Cc: Alexander Potapenko <glider@google.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agomm/kasan: get rid of ->state in struct kasan_alloc_meta
Andrey Ryabinin [Tue, 2 Aug 2016 21:02:52 +0000 (14:02 -0700)]
mm/kasan: get rid of ->state in struct kasan_alloc_meta

The state of object currently tracked in two places - shadow memory, and
the ->state field in struct kasan_alloc_meta.  We can get rid of the
latter.  The will save us a little bit of memory.  Also, this allow us
to move free stack into struct kasan_alloc_meta, without increasing
memory consumption.  So now we should always know when the last time the
object was freed.  This may be useful for long delayed use-after-free
bugs.

As a side effect this fixes following UBSAN warning:
UBSAN: Undefined behaviour in mm/kasan/quarantine.c:102:13
member access within misaligned address ffff88000d1efebc for type 'struct qlist_node'
which requires 8 byte alignment

Link: http://lkml.kernel.org/r/1470062715-14077-5-git-send-email-aryabinin@virtuozzo.com
Reported-by: kernel test robot <xiaolong.ye@intel.com>
Signed-off-by: Andrey Ryabinin <aryabinin@virtuozzo.com>
Cc: Alexander Potapenko <glider@google.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: Pekka Enberg <penberg@kernel.org>
Cc: David Rientjes <rientjes@google.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agomm/kasan: get rid of ->alloc_size in struct kasan_alloc_meta
Andrey Ryabinin [Tue, 2 Aug 2016 21:02:49 +0000 (14:02 -0700)]
mm/kasan: get rid of ->alloc_size in struct kasan_alloc_meta

Size of slab object already stored in cache->object_size.

Note, that kmalloc() internally rounds up size of allocation, so
object_size may be not equal to alloc_size, but, usually we don't need
to know the exact size of allocated object.  In case if we need that
information, we still can figure it out from the report.  The dump of
shadow memory allows to identify the end of allocated memory, and
thereby the exact allocation size.

Link: http://lkml.kernel.org/r/1470062715-14077-4-git-send-email-aryabinin@virtuozzo.com
Signed-off-by: Andrey Ryabinin <aryabinin@virtuozzo.com>
Cc: Alexander Potapenko <glider@google.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agomm/kasan, slub: don't disable interrupts when object leaves quarantine
Andrey Ryabinin [Tue, 2 Aug 2016 21:02:46 +0000 (14:02 -0700)]
mm/kasan, slub: don't disable interrupts when object leaves quarantine

SLUB doesn't require disabled interrupts to call ___cache_free().

Link: http://lkml.kernel.org/r/1470062715-14077-3-git-send-email-aryabinin@virtuozzo.com
Signed-off-by: Andrey Ryabinin <aryabinin@virtuozzo.com>
Acked-by: Alexander Potapenko <glider@google.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agomm/kasan: don't reduce quarantine in atomic contexts
Andrey Ryabinin [Tue, 2 Aug 2016 21:02:43 +0000 (14:02 -0700)]
mm/kasan: don't reduce quarantine in atomic contexts

Currently we call quarantine_reduce() for ___GFP_KSWAPD_RECLAIM (implied
by __GFP_RECLAIM) allocation.  So, basically we call it on almost every
allocation.  quarantine_reduce() sometimes is heavy operation, and
calling it with disabled interrupts may trigger hard LOCKUP:

 NMI watchdog: Watchdog detected hard LOCKUP on cpu 2irq event stamp: 1411258
 Call Trace:
  <NMI>   dump_stack+0x68/0x96
   watchdog_overflow_callback+0x15b/0x190
   __perf_event_overflow+0x1b1/0x540
   perf_event_overflow+0x14/0x20
   intel_pmu_handle_irq+0x36a/0xad0
   perf_event_nmi_handler+0x2c/0x50
   nmi_handle+0x128/0x480
   default_do_nmi+0xb2/0x210
   do_nmi+0x1aa/0x220
   end_repeat_nmi+0x1a/0x1e
  <<EOE>>   __kernel_text_address+0x86/0xb0
   print_context_stack+0x7b/0x100
   dump_trace+0x12b/0x350
   save_stack_trace+0x2b/0x50
   set_track+0x83/0x140
   free_debug_processing+0x1aa/0x420
   __slab_free+0x1d6/0x2e0
   ___cache_free+0xb6/0xd0
   qlist_free_all+0x83/0x100
   quarantine_reduce+0x177/0x1b0
   kasan_kmalloc+0xf3/0x100

Reduce the quarantine_reduce iff direct reclaim is allowed.

Fixes: 55834c59098d("mm: kasan: initial memory quarantine implementation")
Link: http://lkml.kernel.org/r/1470062715-14077-2-git-send-email-aryabinin@virtuozzo.com
Signed-off-by: Andrey Ryabinin <aryabinin@virtuozzo.com>
Reported-by: Dave Jones <davej@codemonkey.org.uk>
Acked-by: Alexander Potapenko <glider@google.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agomm/kasan: fix corruptions and false positive reports
Andrey Ryabinin [Tue, 2 Aug 2016 21:02:40 +0000 (14:02 -0700)]
mm/kasan: fix corruptions and false positive reports

Once an object is put into quarantine, we no longer own it, i.e.  object
could leave the quarantine and be reallocated.  So having set_track()
call after the quarantine_put() may corrupt slab objects.

 BUG kmalloc-4096 (Not tainted): Poison overwritten
 -----------------------------------------------------------------------------
 Disabling lock debugging due to kernel taint
 INFO: 0xffff8804540de850-0xffff8804540de857. First byte 0xb5 instead of 0x6b
...
 INFO: Freed in qlist_free_all+0x42/0x100 age=75 cpu=3 pid=24492
  __slab_free+0x1d6/0x2e0
  ___cache_free+0xb6/0xd0
  qlist_free_all+0x83/0x100
  quarantine_reduce+0x177/0x1b0
  kasan_kmalloc+0xf3/0x100
  kasan_slab_alloc+0x12/0x20
  kmem_cache_alloc+0x109/0x3e0
  mmap_region+0x53e/0xe40
  do_mmap+0x70f/0xa50
  vm_mmap_pgoff+0x147/0x1b0
  SyS_mmap_pgoff+0x2c7/0x5b0
  SyS_mmap+0x1b/0x30
  do_syscall_64+0x1a0/0x4e0
  return_from_SYSCALL_64+0x0/0x7a
 INFO: Slab 0xffffea0011503600 objects=7 used=7 fp=0x          (null) flags=0x8000000000004080
 INFO: Object 0xffff8804540de848 @offset=26696 fp=0xffff8804540dc588
 Redzone ffff8804540de840: bb bb bb bb bb bb bb bb                          ........
 Object ffff8804540de848: 6b 6b 6b 6b 6b 6b 6b 6b b5 52 00 00 f2 01 60 cc  kkkkkkkk.R....`.

Similarly, poisoning after the quarantine_put() leads to false positive
use-after-free reports:

 BUG: KASAN: use-after-free in anon_vma_interval_tree_insert+0x304/0x430 at addr ffff880405c540a0
 Read of size 8 by task trinity-c0/3036
 CPU: 0 PID: 3036 Comm: trinity-c0 Not tainted 4.7.0-think+ #9
 Call Trace:
   dump_stack+0x68/0x96
   kasan_report_error+0x222/0x600
   __asan_report_load8_noabort+0x61/0x70
   anon_vma_interval_tree_insert+0x304/0x430
   anon_vma_chain_link+0x91/0xd0
   anon_vma_clone+0x136/0x3f0
   anon_vma_fork+0x81/0x4c0
   copy_process.part.47+0x2c43/0x5b20
   _do_fork+0x16d/0xbd0
   SyS_clone+0x19/0x20
   do_syscall_64+0x1a0/0x4e0
   entry_SYSCALL64_slow_path+0x25/0x25

Fix this by putting an object in the quarantine after all other
operations.

Fixes: 80a9201a5965 ("mm, kasan: switch SLUB to stackdepot, enable memory quarantine for SLUB")
Link: http://lkml.kernel.org/r/1470062715-14077-1-git-send-email-aryabinin@virtuozzo.com
Signed-off-by: Andrey Ryabinin <aryabinin@virtuozzo.com>
Reported-by: Dave Jones <davej@codemonkey.org.uk>
Reported-by: Vegard Nossum <vegard.nossum@oracle.com>
Reported-by: Sasha Levin <alexander.levin@verizon.com>
Acked-by: Alexander Potapenko <glider@google.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agomemcg: put soft limit reclaim out of way if the excess tree is empty
Michal Hocko [Tue, 2 Aug 2016 21:02:37 +0000 (14:02 -0700)]
memcg: put soft limit reclaim out of way if the excess tree is empty

We've had a report about soft lockups caused by lock bouncing in the
soft reclaim path:

  BUG: soft lockup - CPU#0 stuck for 22s! [kav4proxy-kavic:3128]
  RIP: 0010:[<ffffffff81469798>]  [<ffffffff81469798>] _raw_spin_lock+0x18/0x20
  Call Trace:
    mem_cgroup_soft_limit_reclaim+0x25a/0x280
    shrink_zones+0xed/0x200
    do_try_to_free_pages+0x74/0x320
    try_to_free_pages+0x112/0x180
    __alloc_pages_slowpath+0x3ff/0x820
    __alloc_pages_nodemask+0x1e9/0x200
    alloc_pages_vma+0xe1/0x290
    do_wp_page+0x19f/0x840
    handle_pte_fault+0x1cd/0x230
    do_page_fault+0x1fd/0x4c0
    page_fault+0x25/0x30

There are no memcgs created so there cannot be any in the soft limit
excess obviously:

  [...]
  memory  0       1       1

so all this just seems to be mem_cgroup_largest_soft_limit_node trying
to get spin_lock_irq(&mctz->lock) just to find out that the soft limit
excess tree is empty.  This is just pointless wasting of cycles and
cache line bouncing during heavy parallel reclaim on large machines.
The particular machine wasn't very healthy and most probably suffering
from a memory leak which just caused the memory reclaim to trash
heavily.  But bouncing on the lock certainly didn't help...

Fix this by optimistic lockless check and bail out early if the tree is
empty.  This is theoretically racy but that shouldn't matter all that
much.  First of all soft limit is a best effort feature and it is slowly
getting deprecated and its usage should be really scarce.  Bouncing on a
lock without a good reason is surely much bigger problem, especially on
large CPU machines.

Link: http://lkml.kernel.org/r/1470073277-1056-1-git-send-email-mhocko@kernel.org
Signed-off-by: Michal Hocko <mhocko@suse.com>
Acked-by: Vladimir Davydov <vdavydov@virtuozzo.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agomm, hugetlb: fix huge_pte_alloc BUG_ON
Michal Hocko [Tue, 2 Aug 2016 21:02:34 +0000 (14:02 -0700)]
mm, hugetlb: fix huge_pte_alloc BUG_ON

Zhong Jiang has reported a BUG_ON from huge_pte_alloc hitting when he
runs his database load with memory online and offline running in
parallel.  The reason is that huge_pmd_share might detect a shared pmd
which is currently migrated and so it has migration pte which is
!pte_huge.

There doesn't seem to be any easy way to prevent from the race and in
fact seeing the migration swap entry is not harmful.  Both callers of
huge_pte_alloc are prepared to handle them.  copy_hugetlb_page_range
will copy the swap entry and make it COW if needed.  hugetlb_fault will
back off and so the page fault is retries if the page is still under
migration and waits for its completion in hugetlb_fault.

That means that the BUG_ON is wrong and we should update it.  Let's
simply check that all present ptes are pte_huge instead.

Link: http://lkml.kernel.org/r/20160721074340.GA26398@dhcp22.suse.cz
Signed-off-by: Michal Hocko <mhocko@suse.com>
Reported-by: zhongjiang <zhongjiang@huawei.com>
Acked-by: 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>
7 years agomm/hugetlb: avoid soft lockup in set_max_huge_pages()
Jia He [Tue, 2 Aug 2016 21:02:31 +0000 (14:02 -0700)]
mm/hugetlb: avoid soft lockup in set_max_huge_pages()

In powerpc servers with large memory(32TB), we watched several soft
lockups for hugepage under stress tests.

The call traces are as follows:
1.
get_page_from_freelist+0x2d8/0xd50
__alloc_pages_nodemask+0x180/0xc20
alloc_fresh_huge_page+0xb0/0x190
set_max_huge_pages+0x164/0x3b0

2.
prep_new_huge_page+0x5c/0x100
alloc_fresh_huge_page+0xc8/0x190
set_max_huge_pages+0x164/0x3b0

This patch fixes such soft lockups.  It is safe to call cond_resched()
there because it is out of spin_lock/unlock section.

Link: http://lkml.kernel.org/r/1469674442-14848-1-git-send-email-hejianet@gmail.com
Signed-off-by: Jia He <hejianet@gmail.com>
Reviewed-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
Cc: Paul Gortmaker <paul.gortmaker@windriver.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agotools/testing/radix-tree/linux/gfp.h: fix bitrotted value
Valdis Kletnieks [Tue, 2 Aug 2016 21:02:28 +0000 (14:02 -0700)]
tools/testing/radix-tree/linux/gfp.h: fix bitrotted value

Apparently, the tools/testing version dates to a few flags ago, and
we've sprouted 4 new ones since.  Keep in sync with the value in the
main tree...

Link: http://lkml.kernel.org/r/23400.1469702675@turing-police.cc.vt.edu
Signed-off-by: Valdis Kletnieks <valdis.kletnieks@vt.edu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agomm: move swap-in anonymous page into active list
Minchan Kim [Tue, 2 Aug 2016 21:02:25 +0000 (14:02 -0700)]
mm: move swap-in anonymous page into active list

Every swap-in anonymous page starts from inactive lru list's head.  It
should be activated unconditionally when VM decide to reclaim because
page table entry for the page always usually has marked accessed bit.
Thus, their window size for getting a new referece is 2 * NR_inactive +
NR_active while others is NR_inactive + NR_active.

It's not fair that it has more chance to be referenced compared to other
newly allocated page which starts from active lru list's head.

Johannes:

: The page can still have a valid copy on the swap device, so prefering to
: reclaim that page over a fresh one could make sense.  But as you point
: out, having it start inactive instead of active actually ends up giving it
: *more* LRU time, and that seems to be without justification.

Rik:

: The reason newly read in swap cache pages start on the inactive list is
: that we do some amount of read-around, and do not know which pages will
: get used.
:
: However, immediately activating the ones that DO get used, like your patch
: does, is the right thing to do.

Link: http://lkml.kernel.org/r/1469762740-17860-1-git-send-email-minchan@kernel.org
Signed-off-by: Minchan Kim <minchan@kernel.org>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Rik van Riel <riel@redhat.com>
Cc: Nadav Amit <nadav.amit@gmail.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>
7 years agomm: fail prefaulting if page table allocation fails
Vegard Nossum [Tue, 2 Aug 2016 21:02:22 +0000 (14:02 -0700)]
mm: fail prefaulting if page table allocation fails

I ran into this:

    BUG: sleeping function called from invalid context at mm/page_alloc.c:3784
    in_atomic(): 0, irqs_disabled(): 0, pid: 1434, name: trinity-c1
    2 locks held by trinity-c1/1434:
     #0:  (&mm->mmap_sem){......}, at: [<ffffffff810ce31e>] __do_page_fault+0x1ce/0x8f0
     #1:  (rcu_read_lock){......}, at: [<ffffffff81378f86>] filemap_map_pages+0xd6/0xdd0

    CPU: 0 PID: 1434 Comm: trinity-c1 Not tainted 4.7.0+ #58
    Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Ubuntu-1.8.2-1ubuntu1 04/01/2014
    Call Trace:
      dump_stack+0x65/0x84
      panic+0x185/0x2dd
      ___might_sleep+0x51c/0x600
      __might_sleep+0x90/0x1a0
      __alloc_pages_nodemask+0x5b1/0x2160
      alloc_pages_current+0xcc/0x370
      pte_alloc_one+0x12/0x90
      __pte_alloc+0x1d/0x200
      alloc_set_pte+0xe3e/0x14a0
      filemap_map_pages+0x42b/0xdd0
      handle_mm_fault+0x17d5/0x28b0
      __do_page_fault+0x310/0x8f0
      trace_do_page_fault+0x18d/0x310
      do_async_page_fault+0x27/0xa0
      async_page_fault+0x28/0x30

The important bits from the above is that filemap_map_pages() is calling
into the page allocator while holding rcu_read_lock (sleeping is not
allowed inside RCU read-side critical sections).

According to Kirill Shutemov, the prefaulting code in do_fault_around()
is supposed to take care of this, but missing error handling means that
the allocation failure can go unnoticed.

We don't need to return VM_FAULT_OOM (or any other error) here, since we
can just let the normal fault path try again.

Fixes: 7267ec008b5c ("mm: postpone page table allocation until we have page to map")
Link: http://lkml.kernel.org/r/1469708107-11868-1-git-send-email-vegard.nossum@oracle.com
Signed-off-by: Vegard Nossum <vegard.nossum@oracle.com>
Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: "Hillf Danton" <hillf.zj@alibaba-inc.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agoocfs2/dlm: continue to purge recovery lockres when recovery master goes down
piaojun [Tue, 2 Aug 2016 21:02:19 +0000 (14:02 -0700)]
ocfs2/dlm: continue to purge recovery lockres when recovery master goes down

We found a dlm-blocked situation caused by continuous breakdown of
recovery masters described below.  To solve this problem, we should
purge recovery lock once detecting recovery master goes down.

N3                      N2                   N1(reco master)
                        go down
                                             pick up recovery lock and
                                             begin recoverying for N2

                                             go down

pick up recovery
lock failed, then
purge it:
dlm_purge_lockres
  ->DROPPING_REF is set

send deref to N1 failed,
recovery lock is not purged

find N1 go down, begin
recoverying for N1, but
blocked in dlm_do_recovery
as DROPPING_REF is set:
dlm_do_recovery
  ->dlm_pick_recovery_master
    ->dlmlock
      ->dlm_get_lock_resource
        ->__dlm_wait_on_lockres_flags(tmpres,
   DLM_LOCK_RES_DROPPING_REF);

Fixes: 8c0343968163 ("ocfs2/dlm: clear DROPPING_REF flag when the master goes down")
Link: http://lkml.kernel.org/r/578453AF.8030404@huawei.com
Signed-off-by: Jun Piao <piaojun@huawei.com>
Reviewed-by: Joseph Qi <joseph.qi@huawei.com>
Reviewed-by: Jiufei Xue <xuejiufei@huawei.com>
Reviewed-by: Mark Fasheh <mfasheh@suse.de>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Junxiao Bi <junxiao.bi@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agoocfs2/dlm: solve a BUG when deref failed in dlm_drop_lockres_ref
piaojun [Tue, 2 Aug 2016 21:02:16 +0000 (14:02 -0700)]
ocfs2/dlm: solve a BUG when deref failed in dlm_drop_lockres_ref

We found a BUG situation that lockres is migrated during deref described
below.  To solve the BUG, we could purge lockres directly when other
node says I did not have a ref.  Additionally, we'd better purge lockres
if master goes down, as no one will response deref done.

Node 1                  Node 2(old master)             Node3(new master)
dlm_purge_lockres
send deref to N2

                        leave domain
                        migrate lockres to N3
                                                       finish migration
                                                       send do assert
                                                       master to N1

receive do assert msg
form N3, but can not
find lockres because
DROPPING_REF is set,
so the owner is still
N2.

                        receive deref from N1
                        and response -EINVAL
                        because lockres is migrated

BUG when receive -EINVAL
in dlm_drop_lockres_ref

Fixes: 842b90b62461d ("ocfs2/dlm: return in progress if master can not clear the refmap bit right now")

Link: http://lkml.kernel.org/r/57845103.3070406@huawei.com
Signed-off-by: Jun Piao <piaojun@huawei.com>
Reviewed-by: Joseph Qi <joseph.qi@huawei.com>
Reviewed-by: Jiufei Xue <xuejiufei@huawei.com>
Reviewed-by: Mark Fasheh <mfasheh@suse.de>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Junxiao Bi <junxiao.bi@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agoocfs2/dlm: disable BUG_ON when DLM_LOCK_RES_DROPPING_REF is cleared before dlm_deref_...
piaojun [Tue, 2 Aug 2016 21:02:13 +0000 (14:02 -0700)]
ocfs2/dlm: disable BUG_ON when DLM_LOCK_RES_DROPPING_REF is cleared before dlm_deref_lockres_done_handler

We found a BUG situation in which DLM_LOCK_RES_DROPPING_REF is cleared
unexpected that described below.  To solve the bug, we disable the
BUG_ON and purge lockres in dlm_do_local_recovery_cleanup.

Node 1                               Node 2(master)
dlm_purge_lockres
                                     dlm_deref_lockres_handler

                                     DLM_LOCK_RES_SETREF_INPROG is set
                                     response DLM_DEREF_RESPONSE_INPROG

receive DLM_DEREF_RESPONSE_INPROG
stop puring in dlm_purge_lockres
and wait for DLM_DEREF_RESPONSE_DONE

                                     dispatch dlm_deref_lockres_worker
                                     response DLM_DEREF_RESPONSE_DONE

receive DLM_DEREF_RESPONSE_DONE and
prepare to purge lockres

                                     Node 2 goes down

find Node2 down and do local
clean up for Node2:
dlm_do_local_recovery_cleanup
  -> clear DLM_LOCK_RES_DROPPING_REF

when purging lockres, BUG_ON happens
because DLM_LOCK_RES_DROPPING_REF is clear:
dlm_deref_lockres_done_handler
  ->BUG_ON(!(res->state & DLM_LOCK_RES_DROPPING_REF));

[akpm@linux-foundation.org: fix duplicated write to `ret']
Fixes: 60d663cb5273 ("ocfs2/dlm: add DEREF_DONE message")
Link: http://lkml.kernel.org/r/57845055.9080702@huawei.com
Signed-off-by: Jun Piao <piaojun@huawei.com>
Reviewed-by: Joseph Qi <joseph.qi@huawei.com>
Reviewed-by: Jiufei Xue <xuejiufei@huawei.com>
Reviewed-by: Mark Fasheh <mfasheh@suse.de>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Junxiao Bi <junxiao.bi@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agoocfs2: retry on ENOSPC if sufficient space in truncate log
Eric Ren [Tue, 2 Aug 2016 21:02:10 +0000 (14:02 -0700)]
ocfs2: retry on ENOSPC if sufficient space in truncate log

The testcase "mmaptruncate" in ocfs2 test suite always fails with ENOSPC
error on small volume (say less than 10G).  This testcase repeatedly
performs "extend" and "truncate" on a file.  Continuously, it truncates
the file to 1/2 of the size, and then extends to 100% of the size.  The
main bitmap will quickly run out of space because the "truncate" code
prevent truncate log from being flushed by
ocfs2_schedule_truncate_log_flush(osb, 1), while truncate log may have
cached lots of clusters.

So retry to allocate after flushing truncate log when ENOSPC is
returned.  And we cannot reuse the deleted blocks before the transaction
committed.  Fortunately, we already have a function to do this -
ocfs2_try_to_free_truncate_log().  Just need to remove the "static"
modifier and put it into the right place.

The "unlock"/"lock" code isn't elegant, but there seems to be no better
option.

[zren@suse.com: locking fix]
Link: http://lkml.kernel.org/r/1468031546-4797-1-git-send-email-zren@suse.com
Link: http://lkml.kernel.org/r/1466586469-5541-1-git-send-email-zren@suse.com
Signed-off-by: Eric Ren <zren@suse.com>
Reviewed-by: Gang He <ghe@suse.com>
Reviewed-by: Joseph Qi <joseph.qi@huawei.com>
Reviewed-by: Mark Fasheh <mfasheh@suse.de>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Junxiao Bi <junxiao.bi@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agoocfs2: ensure that dlm lockspace is created by kernel module
Gang He [Tue, 2 Aug 2016 21:02:07 +0000 (14:02 -0700)]
ocfs2: ensure that dlm lockspace is created by kernel module

We encountered a bug from the customer, the user did a fsck.ocfs2 on the
file system and exited unusually, the lockspace (with LVB size = 32) was
left in the kernel space, next, the user mounted this file system, the
kernel module did not create a new lockspace (LVB size = 64) via calling
dlm_new_lockspace() function in mounting stage, just used the existing
lockspace, created by the user space tool, this would lead the user was
not able to mount this file system from the other nodes, with the error
message like:

  dlm: 032F5......: config mismatch: 64,0 nodeid 177127961: 32,0
  (mount.ocfs2,26981,46):ocfs2_dlm_init:2995 ERROR: status = -71
  ocfs2_mount_volume:1881 ERROR: status = -71
  ocfs2_fill_super:1236 ERROR: status = -71

The user found it very difficult to find the root cause, then, we
brought out this patch to relieve such problem.

First, we add one more flag in calling dlm_new_lockspace() function, to
make sure the lockspace is created by kernel module itself, and this
change will not affect the backward compatibility.

Second, the obvious error message is reported in the kernel log, let the
user be more easy to find the root cause.

This patch will be used to insure the dlm lockspace is created by kernel
module when mounting a ocfs2 file system.  There are two ways to create
a lockspace, from user space and kernel space, but the same name
lockspaces probably have different lvblen lengths/flags.

To avoid this mix using, we add one more flag DLM_LSFL_NEWEXCL, it will
make sure the dlm lockspace is created by kernel module when mounting.
Secondly, if a user space program (ocfs2-tools) is running on a file
system, the user tries to mount this file system in the cluster, DLM
module will return a -EEXIST or -EPROTO errno, we should give the user a
obvious error message, then, the user can let that user space tool exit
before mounting the file system again.

Link: http://lkml.kernel.org/r/1463731940-13044-2-git-send-email-ghe@suse.com
Signed-off-by: Gang He <ghe@suse.com>
Reviewed-by: Goldwyn Rodrigues <rgoldwyn@suse.com>
Reviewed-by: Mark Fasheh <mfasheh@suse.de>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Junxiao Bi <junxiao.bi@oracle.com>
Cc: Joseph Qi <joseph.qi@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux
Linus Torvalds [Tue, 2 Aug 2016 16:41:13 +0000 (12:41 -0400)]
Merge branch 'for-linus' of git://git./linux/kernel/git/s390/linux

Pull more s390 updates from Martin Schwidefsky:

 - some cleanup for the hugetlbfs pte/pmd conversion functions

 - the code to check for the minimum CPU type is converted from
   assembler to C and an informational message is added in case the CPU
   is not new enough to run the kernel

 - bug fixes

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux:
  s390/ftrace/jprobes: Fix conflict between jprobes and function graph tracing
  s390: Define AT_VECTOR_SIZE_ARCH for ARCH_DLINFO
  s390/zcrypt: fix possible memory leak in ap_module_init()
  s390/numa: only set possible nodes within node_possible_map
  s390/als: fix compile with gcov enabled
  s390/facilities: do not generate DWORDS define anymore
  s390/als: print missing facilities on facility mismatch
  s390/als: print machine type on facility mismatch
  s390/als: convert architecture level set code to C
  s390/sclp: move uninitialized data to data section
  s390/zcrypt: Fix zcrypt suspend/resume behavior
  s390/cio: fix premature wakeup during chp configure
  s390/cio: convert cfg_lock mutex to spinlock
  s390/mm: clean up pte/pmd encoding

7 years agoMerge tag 'drm-for-v4.8' of git://people.freedesktop.org/~airlied/linux
Linus Torvalds [Tue, 2 Aug 2016 01:44:08 +0000 (21:44 -0400)]
Merge tag 'drm-for-v4.8' of git://people.freedesktop.org/~airlied/linux

Merge drm updates from Dave Airlie:
 "This is the main drm pull request for 4.8.

  I'm down with a cold at the moment so hopefully this isn't in too bad
  a state, I finished pulling stuff last week mostly (nouveau fixes just
  went in today), so only this message should be influenced by illness.
  Apologies to anyone who's major feature I missed :-)

  Core:
        Lockless GEM BO freeing
        Non-blocking atomic work
        Documentation changes (rst/sphinx)
        Prep for new fencing changes
        Simple display helpers
        Master/auth changes
        Register/unregister rework
        Loads of trivial patches/fixes.

  New stuff:
        ARM Mali display driver (not the 3D chip)
        sii902x RGB->HDMI bridge

  Panel:
        Support for new panels
        Improved backlight support

  Bridge:
        Convert ADV7511 to bridge driver
        ADV7533 support
        TC358767 (DSI/DPI to eDP) encoder chip support

  i915:
        BXT support enabled by default
        GVT-g infrastructure
        GuC command submission and fixes
        BXT workarounds
        SKL/BKL workarounds
        Demidlayering device registration
        Thundering herd fixes
        Missing pci ids
        Atomic updates

  amdgpu/radeon:
        ATPX improvements for better dGPU power control on PX systems
        New power features for CZ/BR/ST
        Pipelined BO moves and evictions in TTM
        GPU scheduler improvements
        GPU reset improvements
        Overclocking on dGPUs with amdgpu
        Polaris powermanagement enabled

  nouveau:
        GK20A/GM20B volt and clock improvements.
        Initial support for GP100/GP104 GPUs, GP104 will not yet support
        acceleration due to NVIDIA having not released firmware for them as of yet.

  exynos:
        Exynos5433 SoC with IOMMU support.

  vc4:
        Shader validation for branching

  imx-drm:
        Atomic mode setting conversion
        Reworked DMFC FIFO allocation
        External bridge support

  analogix-dp:
        RK3399 eDP support
        Lots of fixes.

  rockchip:
        Lots of small fixes.

  msm:
        DT bindings cleanups
        Shrinker and madvise support
        ASoC HDMI codec support

  tegra:
        Host1x driver cleanups
        SOR reworking for DP support
        Runtime PM support

  omapdrm:
        PLL enhancements
        Header refactoring
        Gamma table support

  arcgpu:
        Simulator support

  virtio-gpu:
        Atomic modesetting fixes.

  rcar-du:
        Misc fixes.

  mediatek:
        MT8173 HDMI support

  sti:
        ASOC HDMI codec support
        Minor fixes

  fsl-dcu:
        Suspend/resume support
        Bridge support

  amdkfd:
        Minor fixes.

  etnaviv:
        Enable GPU clock gating

  hisilicon:
        Vblank and other fixes"

* tag 'drm-for-v4.8' of git://people.freedesktop.org/~airlied/linux: (1575 commits)
  drm/nouveau/gr/nv3x: fix instobj write offsets in gr setup
  drm/nouveau/acpi: fix lockup with PCIe runtime PM
  drm/nouveau/acpi: check for function 0x1B before using it
  drm/nouveau/acpi: return supported DSM functions
  drm/nouveau/acpi: ensure matching ACPI handle and supported functions
  drm/nouveau/fbcon: fix font width not divisible by 8
  drm/amd/powerplay: remove enable_clock_power_gatings_tasks from initialize and resume events
  drm/amd/powerplay: move clockgating to after ungating power in pp for uvd/vce
  drm/amdgpu: add query device id and revision id into system info entry at CGS
  drm/amdgpu: add new definition in bif header
  drm/amd/powerplay: rename smum header guards
  drm/amdgpu: enable UVD context buffer for older HW
  drm/amdgpu: fix default UVD context size
  drm/amdgpu: fix incorrect type of info_id
  drm/amdgpu: make amdgpu_cgs_call_acpi_method as static
  drm/amdgpu: comment out unused defaults_staturn_pro static const structure to fix the build
  drm/amdgpu: enable UVD VM only on polaris
  drm/amdgpu: increase timeout of IB test
  drm/amdgpu: add destroy session when generate VCE destroy msg.
  drm/amd: fix deadlock of job_list_lock V2
  ...

7 years agoclocksource/drivers/clps_711x: fixup for "ARM: clps711x:
Stephen Rothwell [Thu, 7 Jul 2016 03:59:06 +0000 (13:59 +1000)]
clocksource/drivers/clps_711x: fixup for "ARM: clps711x:

 Switch to MULTIPLATFORM"

Missed conflict between commit c86f51737f8d ("ARM: clps711x: Switch to
MULTIPLATFORM") from the arm-soc tree and commit 250e46aa3bb3
("clocksource/drivers/clps_711x: Add the COMPILE_TEST option") from the
clockevents tree.

Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
7 years agoMerge branch 'linux-4.8' of git://github.com/skeggsb/linux into drm-next
Dave Airlie [Tue, 2 Aug 2016 01:16:02 +0000 (11:16 +1000)]
Merge branch 'linux-4.8' of git://github.com/skeggsb/linux into drm-next

Runtime PM fixes, fbcon and nv30 fix.
* 'linux-4.8' of git://github.com/skeggsb/linux:
  drm/nouveau/gr/nv3x: fix instobj write offsets in gr setup
  drm/nouveau/acpi: fix lockup with PCIe runtime PM
  drm/nouveau/acpi: check for function 0x1B before using it
  drm/nouveau/acpi: return supported DSM functions
  drm/nouveau/acpi: ensure matching ACPI handle and supported functions
  drm/nouveau/fbcon: fix font width not divisible by 8

7 years agoMerge tag 'please-pull-misc-4.8' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Mon, 1 Aug 2016 22:55:31 +0000 (18:55 -0400)]
Merge tag 'please-pull-misc-4.8' of git://git./linux/kernel/git/aegl/linux

Pull ia64 updates from Tony Luck:
 "Miscellaneous ia64 cleanups"

* tag 'please-pull-misc-4.8' of git://git.kernel.org/pub/scm/linux/kernel/git/aegl/linux:
  ia64: salinfo: use a waitqueue instead a sema down/up combo
  ia64: efi: use timespec64 for persistent clock

7 years agoMerge tag 'armsoc-dt64' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc
Linus Torvalds [Mon, 1 Aug 2016 22:47:01 +0000 (18:47 -0400)]
Merge tag 'armsoc-dt64' of git://git./linux/kernel/git/arm/arm-soc

Pull 64-bit ARM DT updates from Olof Johansson:
 "Just as the 32-bit contents, the 64-bit device tree branch also
  contains a number of additions this release cycle.

  New platforms:
   - LG LG1313
   - Mediatek MT6755
   - Renesas r8a7796
   - Broadcom 2837

  Other platforms with larger updates are:
   - Nvidia X1 platforms (USB 3.0, regulators, display subsystem)
   - Mediatek MT8173 (display subsystem added)
   - Rockchip RK3399 (a lot of new peripherals)
   - ARM Juno reference implementation (SCPI power domains, coresight,
     thermal)"

* tag 'armsoc-dt64' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc: (118 commits)
  arm64: tegra: Enable HDMI on Jetson TX1
  arm64: tegra: Add sor1_src clock
  arm64: tegra: Add XUSB powergates on Tegra210
  arm64: tegra: Add DPAUX pinctrl bindings
  arm64: tegra: Add ACONNECT bus node for Tegra210
  arm64: tegra: Add audio powergate node for Tegra210
  arm64: tegra: Add regulators for Tegra210 Smaug
  arm64: tegra: Correct Tegra210 XUSB mailbox interrupt
  arm64: tegra: Enable XUSB controller on Jetson TX1
  arm64: tegra: Enable debug serial on Jetson TX1
  arm64: tegra: Add Tegra210 XUSB controller
  arm64: tegra: Add Tegra210 XUSB pad controller
  arm64: tegra: Add DSI panel on Jetson TX1
  arm64: tegra: p2597: Add SDMMC power supplies
  arm64: tegra: Add PMIC support on Jetson TX1
  Revert "ARM64: DTS: meson-gxbb: switch ethernet to real clock"
  arm64: dts: hi6220: Add pl031 RTC support
  arm64: dts: r8a7796/salvator-x: Enable watchdog timer
  arm64: dts: r8a7796: Add RWDT node
  arm64: dts: r8a7796: Use SYSC "always-on" PM Domain
  ...

7 years agoMerge tag 'armsoc-dt' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc
Linus Torvalds [Mon, 1 Aug 2016 22:37:45 +0000 (18:37 -0400)]
Merge tag 'armsoc-dt' of git://git./linux/kernel/git/arm/arm-soc

Pull ARM DT updates from Olof Johansson:
 "Device tree contents continue to be the largest branches we submit.
  This time around, some of the contents worth pointing out is:

  New SoC platforms:
   - Freescale i.MX 7Solo
   - Broadcom BCM23550
   - Cirrus Logic EP7209 and EP7211 (clps711x platforms)_
   - Hisilicon HI3519
   - Renesas R8A7792

  Some of the other delta that is sticking out, line-count wise:
   - Exynos moves of IP blocks under an SoC bus, which causes a large
     delta due to indentation changes
   - a new Tegra K1 board: Apalis
   - a bunch of small updates to many Allwinner platforms; new hardware
     support, some cleanup, etc"

* tag 'armsoc-dt' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc: (426 commits)
  ARM: dts: sun8i: Add dts file for inet86dz board
  ARM: dts: sun8i: Add dts file for Polaroid MID2407PXE03 tablet
  ARM: dts: sun8i: Use sun8i-reference-design-tablet for ga10h dts
  ARM: dts: sun8i: Use sun8i-reference-design-tablet for polaroid mid2809pxe04
  ARM: dts: sun8i: reference-design-tablet: Add drivevbus-supply
  ARM: dts: Copy sun8i-q8-common.dtsi sun8i-reference-design-tablet.dtsi
  ARM: dts: sun5i: Use sun5i-reference-design-tablet.dtsi for utoo p66 dts
  ARM: dts: sun5i: Use sun5i-reference-design-tablet.dtsi for dit4350 dts
  ARM: dts: sun5i: reference-design-tablet: Remove mention of q8
  ARM: dts: sun5i: reference-design-tablet: Set lradc vref to avcc
  ARM: dts: sun5i: Rename sun5i-q8-common.dtsi sun5i-reference-design-tablet.dtsi
  ARM: dts: sun5i: Move q8 display bits to sun5i-a13-q8-tablet.dts
  ARM: dts: sunxi: Rename sunxi-q8-common.dtsi sunxi-reference-design-tablet.dtsi
  ARM: dts: at91: Don't build unnecessary dtbs
  ARM: dts: at91: sama5d3x: separate motherboard gmac and emac definitions
  ARM: dts: at91: at91sam9g25ek: fix isi endpoint node
  ARM: dts: at91: move isi definition to at91sam9g25ek
  ARM: dts: at91: fix i2c-gpio node name
  ARM: dts: at91: vinco: fix regulator name
  ARM: dts: at91: ariag25 : fix onewire node
  ...

7 years agoMerge tag 'armsoc-drivers' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc
Linus Torvalds [Mon, 1 Aug 2016 22:36:01 +0000 (18:36 -0400)]
Merge tag 'armsoc-drivers' of git://git./linux/kernel/git/arm/arm-soc

Pull ARM SoC driver updates from Olof Johansson:
 "Driver updates for ARM SoCs.

  A slew of changes this release cycle.  The reset driver tree, that we
  merge through arm-soc for historical reasons, is also sizable this
  time around.

  Among the changes:

   - clps711x: Treewide changes to compatible strings, merged here for simplicity.
   - Qualcomm: SCM firmware driver cleanups, move to platform driver
   - ux500: Major cleanups, removal of old mach-specific infrastructure.
   - Atmel external bus memory driver
   - Move of brcmstb platform to the rest of bcm
   - PMC driver updates for tegra, various fixes and improvements
   - Samsung platform driver updates to support 64-bit Exynos platforms
   - Reset controller cleanups moving to devm_reset_controller_register() APIs
   - Reset controller driver for Amlogic Meson
   - Reset controller driver for Hisilicon hi6220
   - ARM SCPI power domain support"

* tag 'armsoc-drivers' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc: (100 commits)
  ARM: ux500: consolidate base platform files
  ARM: ux500: move soc_id driver to drivers/soc
  ARM: ux500: call ux500_setup_id later
  ARM: ux500: consolidate soc_device code in id.c
  ARM: ux500: remove cpu_is_u* helpers
  ARM: ux500: use CLK_OF_DECLARE()
  ARM: ux500: move l2x0 init to .init_irq
  mfd: db8500 stop passing around platform data
  ASoC: ab8500-codec: remove platform data based probe
  ARM: ux500: move ab8500_regulator_plat_data into driver
  ARM: ux500: remove unused regulator data
  soc: raspberrypi-power: add CONFIG_OF dependency
  firmware: scpi: add CONFIG_OF dependency
  video: clps711x-fb: Changing the compatibility string to match with the smallest supported chip
  input: clps711x-keypad: Changing the compatibility string to match with the smallest supported chip
  pwm: clps711x: Changing the compatibility string to match with the smallest supported chip
  serial: clps711x: Changing the compatibility string to match with the smallest supported chip
  irqchip: clps711x: Changing the compatibility string to match with the smallest supported chip
  clocksource: clps711x: Changing the compatibility string to match with the smallest supported chip
  clk: clps711x: Changing the compatibility string to match with the smallest supported chip
  ...

7 years agoMerge tag 'armsoc-defconfig' of git://git.kernel.org/pub/scm/linux/kernel/git/arm...
Linus Torvalds [Mon, 1 Aug 2016 22:30:44 +0000 (18:30 -0400)]
Merge tag 'armsoc-defconfig' of git://git./linux/kernel/git/arm/arm-soc

Pull ARM SoC defconfig updates from Olof Johansson:
 "Defconfig additions, removals, etc.

  Most of these are small changes adding the options for newly
  upstreamed drivers, or drivers needed for new board support.

  - removed Broadcom bcm_defconfig, targets are all covered in
    multi_v7_defconfig

  - new defconfig: multi_v4t_defconfig"

* tag 'armsoc-defconfig' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc: (43 commits)
  ARM: multi_v7_defconfig: enable STMicroelectronics DRM Support
  ARM: multi_v7_defconfig: enable STMicroelectronics BDISP 2D blitter driver
  ARM: omap2plus_defconfig: Fix USB selection to keep Ethernet working
  ARM: configs: lpc18xx: enable adc and dac iio drivers
  ARM: config: Add a multi_v4t_defconfig
  ARM: multi_v7_defconfig: add ACT8945A
  ARM: at91/defconfig: Add ACT8945A driver
  ARM: imx_v6_v7_defconfig: Select ADS7846 support
  ARM: configs: qualcomm: Add MDM9615 missing defconfigs
  ARM: multi_v7_defconfig: Enable vivid driver as a module
  ARM: exynos_defconfig: Enable vivid driver as a module
  ARM: defconfig: enable the MSM8660 pin controller
  ARM: imx_v6_v7_defconfig: enable USB FFS gadget
  ARM: socfpga: enable PL330 DMA in socfpga_defconfig
  ARM: socfpga: add PCIe to socfpga_defconfig
  ARM: multi_v7_defconfig: defconfig: Enable r8a7792 SoC
  ARM: shmobile: defconfig: Enable r8a7792 SoC
  ARM: Remove bcm_defconfig
  ARM: multi_v7_defconfig: Enable Broadcom Kona watchdog
  ARM: multi_v7_defconfig: Enable Broadcom STB PWM
  ...

7 years agoMerge tag 'armsoc-arm64' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc
Linus Torvalds [Mon, 1 Aug 2016 22:29:28 +0000 (18:29 -0400)]
Merge tag 'armsoc-arm64' of git://git./linux/kernel/git/arm/arm-soc

Pull 64-bit ARM SoC updates from Olof Johansson:
 "Changes to platform code for 64-bit ARM platforms.

  Nearly all of these are defconfig updates to enable new drivers or old
  drivers still used on these 64-bit platforms.

  Added platforms for this release are:

   - Broadcom BCM2837
   - Renesas R8A7796"

* tag 'armsoc-arm64' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc: (22 commits)
  arm64: remove duplicate PWM entry in defconfig
  arm64: Update default configuration
  arm64: defconfig: Enable more IP blocks for Exynos7 and Exynos5433
  arm64: amlogic: select gxbb clk driver
  arm64: defconfig: Enable S2MPS11 clock and S3C RTC driver
  arm64: marvell: enable Armada 3700 clock drivers
  arm64: defconfig: enable msm8996 pinctrl support
  arm64: defconfig: Enable qcom msm8996 clk drivers
  arm: defconfig: Enable PM8941 pwr key
  arm64: defconfig: enable stmmac and realtek PHY as modules
  arm64: Kconfig: select PM{,_GENERIC_DOMAINS} for ARCH_VEXPRESS
  arm64: defconfig: enable SENSORS_ARM_SCPI
  arm64: defconfig: enable Generic on-chip SRAM driver
  arm64: configs: enable PCIe driver for Armada 7K/8K
  arm64: Add platform selection for BCM2835.
  arm64: defconfig: disable plain NEON implementation of AES
  arm64: Allow for different DMA and CPU bus offsets
  arm64: defconfig: enable Renesas R8A7796 SoC
  arm64: defconfig: Enable Cadence MACB/GEM support
  ARM64: Kconfig: Select the Amlogic Meson pin controller driver
  ...

7 years agoMerge tag 'armsoc-soc' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc
Linus Torvalds [Mon, 1 Aug 2016 22:27:08 +0000 (18:27 -0400)]
Merge tag 'armsoc-soc' of git://git./linux/kernel/git/arm/arm-soc

Pull ARM SoC platform updates from Olof Johansson:
 "Improved and new platform support for various SoCs:

  New SoC support:
   - Broadcom BCM23550
   - Freescale i.MX7Solo
   - Qualcomm MDM9615
   - Renesas r8a7792

  Improvements:
   - convert clps711x to multiplatform
   - debug uart improvements for Atmel platforms
   - Tango platform improvements: HOTPLUG_CPU, Suspend-to-ram
   - OMAP tweaks and improvements to hwmod
   - OMAP support for kexec on SMP"

* tag 'armsoc-soc' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc: (109 commits)
  ARM: davinci: fix build break because of undeclared dm365_evm_snd_data
  ARM: s3c64xx: smartq: Avoid sparse warnings
  ARM: sti: Implement dummy L2 cache's write_sec
  ARM: STi: Update machine _namestr to be more generic.
  arm: meson: explicitly select clk drivers
  ARM: tango: add Suspend-to-RAM support
  ARM: hisi: consolidate the hisilicon machine entries
  ARM: tango: fix CONFIG_HOTPLUG_CPU=n build
  MAINTAINERS: Update BCM281XX/BCM11XXX/BCM216XX entry
  MAINTAINERS: Update BCM63XX entry
  MAINTAINERS: Add NS2 entry
  MAINTAINERS: Fix nsp false-positives
  MAINTAINERS: Change L to M for Broadcom ARM/ARM64 SoC entries
  ARM: debug: Enable DEBUG_BCM_5301X for Northstar Plus SoCs
  ARM: clps711x: Switch to MULTIPLATFORM
  ARM: clps711x: Remove boards support
  ARM: clps711x: Add basic DT support
  ARM: clps711x: Reduce static map size
  ARM: SAMSUNG: Constify iomem address passed to s5p_init_cpu
  ARM: oxnas: Change OX810SE default driver config
  ...

7 years agoMerge tag 'armsoc-cleanup' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc
Linus Torvalds [Mon, 1 Aug 2016 22:21:13 +0000 (18:21 -0400)]
Merge tag 'armsoc-cleanup' of git://git./linux/kernel/git/arm/arm-soc

Pull ARM SoC cleanups from Olof Johansson:
 "The cleanup branch keeps going down in size as we've completed a lot
  of the major legacy platform removals and conversions.

  A handful of changes this time around, some of the themes or larger
  sets are:

   - A bunch of i.MX cleanups around platform detection, init call cleanups
   - Misc fixes of missing/implicit includes
   - Removal of ARCH_[WANT_OPTIONAL|REQUIRE]_GPIOLIB"

* tag 'armsoc-cleanup' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc: (40 commits)
  ARM: mps2: fix typo
  ARM: s3c64xx: avoid warning about 'struct device_node'
  bus: mvebu-mbus: make mvebu_mbus_syscore_ops static
  bus: mvebu-mbus: fix __iomem on register pointers
  ARM: tegra: Remove board_init_funcs array
  ARM: iop: Fix indentation
  ARM: imx: remove cpu_is_mx*()
  ARM: imx: remove last call to cpu_is_mx5*
  ARM: imx: rework mx27_pm_init() call
  ARM: imx: deconstruct mx3_idle
  ARM: imx: deconstruct mxc_rnga initialization
  ARM: imx: remove cpu_is_mx1 check
  ARM: i.MX: Do not explicitly call l2x0_of_init()
  ARM: i.MX: system.c: Tweak prefetch settings for performance
  ARM: i.MX: system.c: Replace magic numbers
  ARM: i.MX: system.c: Remove redundant errata 752271 code
  ARM: i.MX: system.c: Convert goto to if statement
  ARM: Kirkwood: fix kirkwood_pm_init() declaration/type
  ARM: Kirkwood: make kirkwood_disable_mbus_error_propagation() static
  ARM: orion5x: make orion5x_legacy_handle_irq static
  ...

7 years agoMerge tag 'linux-kselftest-4.8-rc1-update' of git://git.kernel.org/pub/scm/linux...
Linus Torvalds [Mon, 1 Aug 2016 20:52:08 +0000 (16:52 -0400)]
Merge tag 'linux-kselftest-4.8-rc1-update' of git://git./linux/kernel/git/shuah/linux-kselftest

Pull kselftest updates from Shuah Khan:
 "This contains new tests and fixes:

   - a few fixes to existing tests

   - new media tests for testing driver unbind, and device removal paths
     while an user application is actively making system calls and
     ioctls"

* tag 'linux-kselftest-4.8-rc1-update' of git://git.kernel.org/pub/scm/linux/kernel/git/shuah/linux-kselftest:
  selftests: media_tests add a new video device test
  selftests: media_tests - Add media_device_open to .gitignore
  selftests: add media controller regression test scripts and document
  selftests: add media_device_open test
  selftests: media_device_test change it to randomize loop count
  selftests/vm: Don't mlockall MCL_CURRENT in on-fault-limit test
  selftests/vm: write strlen length instead of sizeof to nr_hugepages
  selftests/lib: set printf.sh executable

7 years agoMerge tag 'linux-kselftest-4.8-rc1-fixes' of git://git.kernel.org/pub/scm/linux/kerne...
Linus Torvalds [Mon, 1 Aug 2016 20:51:02 +0000 (16:51 -0400)]
Merge tag 'linux-kselftest-4.8-rc1-fixes' of git://git./linux/kernel/git/shuah/linux-kselftest

Pull kselftest fixes from Shuah Khan:

 - Add a new timer set-tz test case

 - Fix a bug in exec test Makefile dependency list

* tag 'linux-kselftest-4.8-rc1-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/shuah/linux-kselftest:
  selftests/exec: Makefile is a run-time dependency, add it to the install list
  kselftests: timers: Add set-tz test case

7 years agoMerge tag 'hwmon-for-linus-v4.8-2' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Mon, 1 Aug 2016 20:49:13 +0000 (16:49 -0400)]
Merge tag 'hwmon-for-linus-v4.8-2' of git://git./linux/kernel/git/groeck/linux-staging

Pull more hwmon updates from Guenter Roeck:

 - Improved error handling in tmp102, lm75, and lm90 drivers

 - Bug fixes in sht3x, ftsteutates, iio_hwmon, and adt7411 drivers

* tag 'hwmon-for-linus-v4.8-2' of git://git.kernel.org/pub/scm/linux/kernel/git/groeck/linux-staging:
  hwmon: (adt7411) set sane values for CFG1 and CFG3
  hwmon: (iio_hwmon) fix memory leak in name attribute
  hwmon: (ftsteutates) Fix potential memory access error
  hwmon: (tmp102) Improve error handling
  hwmon: (lm75) Improve error handling
  hwmon: (lm90) Improve error handling
  hwmon: (lm90) Add missing assignment
  hwmon: (sht3x) set initial jiffies to last_update

7 years agoMerge branch 'linus' of git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6
Linus Torvalds [Mon, 1 Aug 2016 18:28:42 +0000 (14:28 -0400)]
Merge branch 'linus' of git://git./linux/kernel/git/herbert/crypto-2.6

Pull crypto fixes from Herbert Xu:
 "This fixes a number of regressions in the marvell cesa driver caused
  by the chaining work, and a regression in lib/mpi that leads to a
  GFP_KERNEL allocation with preemption disabled"

* 'linus' of git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6:
  crypto: marvell - Don't copy IV vectors from the _process op for ciphers
  lib/mpi: Fix SG miter leak
  crypto: marvell - Update cache with input sg only when it is unmapped
  crypto: marvell - Don't chain at DMA level when backlog is disabled
  crypto: marvell - Fix memory leaks in TDMA chain for cipher requests

7 years agoMerge branch 'x86-headers-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Mon, 1 Aug 2016 18:23:42 +0000 (14:23 -0400)]
Merge branch 'x86-headers-for-linus' of git://git./linux/kernel/git/tip/tip

Pull x86 header cleanups from Ingo Molnar:
 "This tree is a cleanup of the x86 tree reducing spurious uses of
  module.h - which should improve build performance a bit"

* 'x86-headers-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86, crypto: Restore MODULE_LICENSE() to glue_helper.c so it loads
  x86/apic: Remove duplicated include from probe_64.c
  x86/ce4100: Remove duplicated include from ce4100.c
  x86/headers: Include spinlock_types.h in x8664_ksyms_64.c for missing spinlock_t
  x86/platform: Delete extraneous MODULE_* tags fromm ts5500
  x86: Audit and remove any remaining unnecessary uses of module.h
  x86/kvm: Audit and remove any unnecessary uses of module.h
  x86/xen: Audit and remove any unnecessary uses of module.h
  x86/platform: Audit and remove any unnecessary uses of module.h
  x86/lib: Audit and remove any unnecessary uses of module.h
  x86/kernel: Audit and remove any unnecessary uses of module.h
  x86/mm: Audit and remove any unnecessary uses of module.h
  x86: Don't use module.h just for AUTHOR / LICENSE tags

7 years agoMerge tag 'backlight-for-linus-4.8' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Mon, 1 Aug 2016 11:32:32 +0000 (07:32 -0400)]
Merge tag 'backlight-for-linus-4.8' of git://git./linux/kernel/git/lee/backlight

Pull backlight updates from Lee Jones:
 "Add support for an enable regulator to lp855x_bl"

* tag 'backlight-for-linus-4.8' of git://git.kernel.org/pub/scm/linux/kernel/git/lee/backlight:
  backlight: lp855x: Add enable regulator

7 years agoMerge tag 'mfd-for-linus-4.8' of git://git.kernel.org/pub/scm/linux/kernel/git/lee/mfd
Linus Torvalds [Mon, 1 Aug 2016 11:28:14 +0000 (07:28 -0400)]
Merge tag 'mfd-for-linus-4.8' of git://git./linux/kernel/git/lee/mfd

Pull MFD updates from Lee Jones:
 "Core Framework:
   - New API to call bespoke pre/post IRQ handlers; Regmap

  New Device Support:
   - Add support for RN5T567 to rn5t618
   - Add support for COMe-cSL6 and COMe-mAL10 to kempld-core

  New Functionality:
   - Add support for USB Power Supply to axp20x
   - Add support for Power Key to hi655x-pmic

  Fix-ups:
   - Update MAINTAINERS; Dialog, Altera
   - Remove module support; max77843, max77620, max8998, max8997, max8925-i2c
   - Add module support; max14577
   - Constifying; max77620
   - Allow bespoke IRQ masking/unmasking; max77620
   - Remove superfluous code; arizona, qcom_rpm, smsc-ece1099
   - Power Management fixups; arizona-core
   - Error-path improvement; twl-core, dm355evm_msp, smsc-ece1099, hi655x
   - Clocking fixups; twl6040
   - Trivial (spelling, headers, coding-style, whitespace, (re)naming);
       si476x-i2c, omap-usb-tll, ti_am335x_tscadc, tps6507, hi655x-pmic

  Bug Fixes:
   - Fix offset error for MSM8660; qcom_rpm
   - Fix possible spurious IRQs; arizona, hi655x-pmic"

* tag 'mfd-for-linus-4.8' of git://git.kernel.org/pub/scm/linux/kernel/git/lee/mfd: (42 commits)
  mfd: qcom_rpm: Parametrize also ack selector size
  mfd: twl6040: Handle mclk used for HPPLL and optional internal clock source
  mfd: Add support for COMe-cSL6 and COMe-mAL10 to Kontron PLD driver
  mfd: hi655x: Fix return value check in hi655x_pmic_probe()
  mfd: smsc-ece1099: Return directly after a function failure in smsc_i2c_probe()
  mfd: smsc-ece1099: Delete an unnecessary variable initialisation in smsc_i2c_probe()
  mfd: dm355evm_msp: Return directly after a failed platform_device_alloc() in add_child()
  mfd: twl-core: Refactoring for add_numbered_child()
  mfd: twl-core: Return directly after a failed platform_device_alloc() in add_numbered_child()
  mfd: arizona: Add missing disable of PM runtime on probe error path
  mfd: stmpe: Move platform data into MFD driver
  mfd: max14577: Allow driver to be built as a module
  mfd: max14577: Use module_init() instead of subsys_initcall()
  mfd: arizona: Remove some duplicate defines
  mfd: qcom_rpm: Remove unused define
  mfd: hi655x-pmic: Add powerkey device to hi655x PMIC driver
  mfd: hi655x-pmic: Rename some interrupt macro names
  mfd: hi655x-pmic: Fixup issue with un-acked interrupts
  mfd: arizona: Check if AOD interrupts are pending before dispatching
  mfd: qcom_rpm: Fix offset error for msm8660
  ...

7 years agoMerge tag 'iommu-updates-v4.8' of git://git.kernel.org/pub/scm/linux/kernel/git/joro...
Linus Torvalds [Mon, 1 Aug 2016 11:25:10 +0000 (07:25 -0400)]
Merge tag 'iommu-updates-v4.8' of git://git./linux/kernel/git/joro/iommu

Pull IOMMU updates from Joerg Roedel:

 - big-endian support and preparation for defered probing for the Exynos
   IOMMU driver

 - simplifications in iommu-group id handling

 - support for Mediatek generation one IOMMU hardware

 - conversion of the AMD IOMMU driver to use the generic IOVA allocator.
   This driver now also benefits from the recent scalability
   improvements in the IOVA code.

 - preparations to use generic DMA mapping code in the Rockchip IOMMU
   driver

 - device tree adaption and conversion to use generic page-table code
   for the MSM IOMMU driver

 - an iova_to_phys optimization in the ARM-SMMU driver to greatly
   improve page-table teardown performance with VFIO

 - various other small fixes and conversions

* tag 'iommu-updates-v4.8' of git://git.kernel.org/pub/scm/linux/kernel/git/joro/iommu: (59 commits)
  iommu/amd: Initialize dma-ops domains with 3-level page-table
  iommu/amd: Update Alias-DTE in update_device_table()
  iommu/vt-d: Return error code in domain_context_mapping_one()
  iommu/amd: Use container_of to get dma_ops_domain
  iommu/amd: Flush iova queue before releasing dma_ops_domain
  iommu/amd: Handle IOMMU_DOMAIN_DMA in ops->domain_free call-back
  iommu/amd: Use dev_data->domain in get_domain()
  iommu/amd: Optimize map_sg and unmap_sg
  iommu/amd: Introduce dir2prot() helper
  iommu/amd: Implement timeout to flush unmap queues
  iommu/amd: Implement flush queue
  iommu/amd: Allow NULL pointer parameter for domain_flush_complete()
  iommu/amd: Set up data structures for flush queue
  iommu/amd: Remove align-parameter from __map_single()
  iommu/amd: Remove other remains of old address allocator
  iommu/amd: Make use of the generic IOVA allocator
  iommu/amd: Remove special mapping code for dma_ops path
  iommu/amd: Pass gfp-flags to iommu_map_page()
  iommu/amd: Implement apply_dm_region call-back
  iommu/amd: Create a list of reserved iova addresses
  ...

7 years agoMerge branch 'mailbox-for-next' of git://git.linaro.org/landing-teams/working/fujitsu...
Linus Torvalds [Mon, 1 Aug 2016 11:23:29 +0000 (07:23 -0400)]
Merge branch 'mailbox-for-next' of git://git.linaro.org/landing-teams/working/fujitsu/integration

Pull mailbox updates from Jussi Brar:
 "Broadcom:
   - New PDC controller driver and bindings

  Misc:
   - PL320 - Convert from 'raw' IO to 'relaxed' version
   - Test - fix dangling pointer"

* 'mailbox-for-next' of git://git.linaro.org/landing-teams/working/fujitsu/integration:
  mailbox: Fix format and type mismatches in Broadcom PDC driver
  mailbox: Add Broadcom PDC mailbox driver
  dt-bindings: add bindings documentation for PDC driver.
  mailbox: pl320: remove __raw IO
  mailbox: mailbox-test: set tdev->signal to NULL after freeing

7 years agoMerge tag 'mmc-v4.8' of git://git.linaro.org/people/ulf.hansson/mmc
Linus Torvalds [Mon, 1 Aug 2016 01:36:58 +0000 (21:36 -0400)]
Merge tag 'mmc-v4.8' of git://git.linaro.org/people/ulf.hansson/mmc

Pull MMC updates from Ulf Hansson:
 "MMC core:
   - A couple of changes to improve the support for erase/discard/trim cmds
   - Add eMMC HS400 enhanced strobe support
   - Show OCR and DSR registers in SYSFS for MMC/SD cards
   - Correct and improve busy detection logic for MMC switch (CMD6) cmds
   - Disable HPI cmds for certain broken Hynix eMMC cards
   - Allow MMC hosts to specify non-support for SD and MMC cmds
   - Some minor additional fixes

  MMC host:
   - sdhci: Re-works, fixes and clean-ups
   - sdhci: Add HW auto re-tuning support
   - sdhci: Re-factor code to prepare for adding support for eMMC CMDQ
   - sdhci-esdhc-imx: Fixes and clean-ups
   - sdhci-esdhc-imx: Update system PM support
   - sdhci-esdhc-imx: Enable HW auto re-tuning
   - sdhci-bcm2835: Remove driver as sdhci-iproc is used instead
   - sdhci-brcmstb: Add new driver for Broadcom BRCMSTB SoCs
   - sdhci-msm: Add support for UHS cards
   - sdhci-tegra: Improve support for UHS cards
   - sdhci-of-arasan: Update phy support for Rockchip SoCs
   - sdhci-of-arasan: Deploy enhanced strobe support
   - dw_mmc: Some fixes and clean-ups
   - dw_mmc: Enable support for erase/discard/trim cmds
   - dw_mmc: Enable CMD23 support
   - mediatek: Some fixes related to the eMMC HS400 support
   - sh_mmcif: Improve support for HW busy detection
   - rtsx_pci: Enable support for erase/discard/trim cmds"

* tag 'mmc-v4.8' of git://git.linaro.org/people/ulf.hansson/mmc: (135 commits)
  mmc: rtsx_pci: Remove deprecated create_singlethread_workqueue
  mmc: rtsx_pci: Enable MMC_CAP_ERASE to allow erase/discard/trim requests
  mmc: rtsx_pci: Use the provided busy timeout from the mmc core
  mmc: sdhci-pltfm: Drop define for SDHCI_PLTFM_PMOPS
  mmc: sdhci-pltfm: Convert to use the SET_SYSTEM_SLEEP_PM_OPS
  mmc: sdhci-pltfm: Make sdhci_pltfm_suspend|resume() static
  mmc: sdhci-esdhc-imx: Use common sdhci_suspend|resume_host()
  mmc: sdhci-esdhc-imx: Assign system PM ops within #ifdef CONFIG_PM_SLEEP
  mmc: sdhci-sirf: Remove non needed #ifdef CONFIG_PM* for dev_pm_ops
  mmc: sdhci-s3c: Remove non needed #ifdef CONFIG_PM for dev_pm_ops
  mmc: sdhci-pxav3: Remove non needed #ifdef CONFIG_PM for dev_pm_ops
  mmc: sdhci-of-esdhc: Simplify code by using SIMPLE_DEV_PM_OPS
  mmc: sdhci-acpi: Simplify code by using SET_SYSTEM_SLEEP_PM_OPS
  mmc: sdhci-pci-core: Simplify code by using SET_SYSTEM_SLEEP_PM_OPS
  mmc: Change the max discard sectors and erase response when HW busy detect
  phy: rockchip-emmc: Wait even longer for the DLL to lock
  phy: rockchip-emmc: Be tolerant to card clock of 0 in power on
  mmc: sdhci-of-arasan: Revert: Always power the PHY off/on when clock changes
  mmc: sdhci-msm: Add support for UHS cards
  mmc: sdhci-msm: Add set_uhs_signaling() implementation
  ...

7 years agoMerge git://www.linux-watchdog.org/linux-watchdog
Linus Torvalds [Mon, 1 Aug 2016 01:32:22 +0000 (21:32 -0400)]
Merge git://www.linux-watchdog.org/linux-watchdog

Pull watchdog updates from Wim Van Sebroeck:
 "Core:
   - min and max timeout improvements, WDOG_HW_RUNNING improvements,
     status funtionality
   - Add a device managed API for watchdog_register_device()

  New watchdog drivers:
   -  Aspeed SoCs
   -  Maxim PMIC MAX77620
   -  Amlogic Meson GXBB SoC

  Enhancements:
   - support for the r8a7796 watchdog device
   - support for F81866 watchdog device
   - support for 5th variation of Apollo Lake
   - support for MCP78S chipset
   - clean-up of softdog.c watchdog device driver
   - pic32-wdt and pic32-dmt fixes
   - Documentation/watchdog: watchdog-test improvements
   - several other fixes and improvements"

* git://www.linux-watchdog.org/linux-watchdog: (50 commits)
  watchdog: gpio_wdt: Fix missing platform_set_drvdata() in gpio_wdt_probe()
  watchdog: core: Clear WDOG_HW_RUNNING before calling the stop function
  watchdog: core: Fix error handling of watchdog_dev_init()
  watchdog: pic32-wdt: Fix return value check in pic32_wdt_drv_probe()
  watchdog: pic32-dmt: Remove .owner field for driver
  watchdog: pic32-wdt: Remove .owner field for driver
  watchdog: renesas-wdt: Add support for the r8a7796 wdt
  Documentation/watchdog: check return value for magic close
  watchdog: sbsa: Drop status function
  watchdog: Implement status function in watchdog core
  watchdog: tangox: Set max_hw_heartbeat_ms instead of max_timeout
  watchdog: change watchdog_need_worker logic
  watchdog: add support for MCP78S chipset in nv_tco
  watchdog: bcm2835_wdt: remove redundant ->set_timeout callback
  watchdog: bcm2835_wdt: constify _ops and _info structures
  dt-bindings: watchdog: Add Meson GXBB Watchdog bindings
  watchdog: Add Meson GXBB Watchdog Driver
  watchdog: qcom: configure BARK time in addition to BITE time
  watchdog: qcom: add option for standalone watchdog not in timer block
  watchdog: qcom: update device tree bindings
  ...

7 years agoMerge branch 'for-linus-4.8' of git://git.kernel.org/pub/scm/linux/kernel/git/mason...
Linus Torvalds [Mon, 1 Aug 2016 01:27:32 +0000 (21:27 -0400)]
Merge branch 'for-linus-4.8' of git://git./linux/kernel/git/mason/linux-btrfs

Pull btrfs updates from Chris Mason:
 "This pull is dedicated to Josef's enospc rework, which we've been
  testing for a few releases now.  It fixes some early enospc problems
  and is dramatically faster.

  This also includes an updated fix for the delalloc accounting that
  happens after a fault in copy_from_user.  My patch in v4.7 was almost
  but not quite enough"

* 'for-linus-4.8' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs:
  Btrfs: fix delalloc accounting after copy_from_user faults
  Btrfs: avoid deadlocks during reservations in btrfs_truncate_block
  Btrfs: use FLUSH_LIMIT for relocation in reserve_metadata_bytes
  Btrfs: fill relocation block rsv after allocation
  Btrfs: always use trans->block_rsv for orphans
  Btrfs: change how we calculate the global block rsv
  Btrfs: use root when checking need_async_flush
  Btrfs: don't bother kicking async if there's nothing to reclaim
  Btrfs: fix release reserved extents trace points
  Btrfs: add fsid to some tracepoints
  Btrfs: add tracepoints for flush events
  Btrfs: fix delalloc reservation amount tracepoint
  Btrfs: trace pinned extents
  Btrfs: introduce ticketed enospc infrastructure
  Btrfs: add tracepoint for adding block groups
  Btrfs: warn_on for unaccounted spaces
  Btrfs: change delayed reservation fallback behavior
  Btrfs: always reserve metadata for delalloc extents
  Btrfs: fix callers of btrfs_block_rsv_migrate
  Btrfs: add bytes_readonly to the spaceinfo at once

7 years agohwmon: (adt7411) set sane values for CFG1 and CFG3
Michael Walle [Mon, 25 Jul 2016 09:12:05 +0000 (11:12 +0200)]
hwmon: (adt7411) set sane values for CFG1 and CFG3

According to the datasheet we have to set some bits as 0 and others as 1.
Make sure we do this for CFG1 and CFG3.

Signed-off-by: Michael Walle <michael@walle.cc>
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
7 years agohwmon: (iio_hwmon) fix memory leak in name attribute
Quentin Schulz [Tue, 26 Jul 2016 07:47:09 +0000 (09:47 +0200)]
hwmon: (iio_hwmon) fix memory leak in name attribute

The "name" variable's memory is now freed when the device is destructed
thanks to devm function.

Signed-off-by: Quentin Schulz <quentin.schulz@free-electrons.com>
Reported-by: Guenter Roeck <linux@roeck-us.net>
Fixes: e0f8a24e0edfd ("staging:iio::hwmon interface client driver.")
Fixes: 61bb53bcbdd86 ("hwmon: (iio_hwmon) Add support for humidity sensors")
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
7 years agohwmon: (ftsteutates) Fix potential memory access error
Guenter Roeck [Mon, 25 Jul 2016 22:01:28 +0000 (15:01 -0700)]
hwmon: (ftsteutates) Fix potential memory access error

Using set_bit() to set a bit in an integer is not a good idea, since
the function expects an unsigned long as argument, which can be 64 bit
wide. Coverity reports this problem as

>>>     CID 1364488:  Memory - illegal accesses  (INCOMPATIBLE_CAST)
>>>     Pointer "&ret" points to an object whose effective type is "int"
>>> (32 bits, signed) but is dereferenced as a wider "unsigned
+long" (64 bits, unsigned).  This may lead to memory corruption.
245                     set_bit(1, (unsigned long *)&ret);

Just use BIT instead.

Cc: Thilo Cestonaro <thilo@cestona.ro>
Fixes: 08426eda58e0 ("hwmon: Add driver for FTS BMC chip "Teutates"")
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
7 years agohwmon: (tmp102) Improve error handling
Guenter Roeck [Mon, 25 Jul 2016 21:57:47 +0000 (14:57 -0700)]
hwmon: (tmp102) Improve error handling

Use devm_add_action_or_reset() instead of devm_add_action(), and
check its return code.

Signed-off-by: Guenter Roeck <linux@roeck-us.net>
7 years agohwmon: (lm75) Improve error handling
Guenter Roeck [Mon, 25 Jul 2016 21:56:00 +0000 (14:56 -0700)]
hwmon: (lm75) Improve error handling

Use devm_add_action_or_reset() instead of devm_add_action(), and
check its return value.

Signed-off-by: Guenter Roeck <linux@roeck-us.net>
7 years agohwmon: (lm90) Improve error handling
Guenter Roeck [Mon, 25 Jul 2016 21:53:23 +0000 (14:53 -0700)]
hwmon: (lm90) Improve error handling

Replace devm_add_action() with devm_add_action_or_reset(),
and check its return value.

Reviewed-by: Jean Delvare <jdelvare@suse.de>
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
7 years agohwmon: (lm90) Add missing assignment
Guenter Roeck [Mon, 25 Jul 2016 21:47:18 +0000 (14:47 -0700)]
hwmon: (lm90) Add missing assignment

Coverity reports the following error.

>>>     CID 1364474:  Error handling issues  (CHECKED_RETURN)
>>>     Calling "lm90_read_reg" without checking return value (as is done
>>>     elsewhere 28 out of 29 times).
532             lm90_read_reg(client, LM90_REG_R_REMOTE_LOWH);
533             if (val < 0)
534                     return val;

Fixes: 10bfef47bd259 ("hwmon: (lm90) Read limit registers only once")
Reviewed-by: Jean Delvare <jdelvare@suse.de>
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
7 years agohwmon: (sht3x) set initial jiffies to last_update
Matt Ranostay [Mon, 25 Jul 2016 00:05:32 +0000 (17:05 -0700)]
hwmon: (sht3x) set initial jiffies to last_update

Handling the wraparound requires the data->last_update to be set to an
initial jiffies value. Otherwise on 32-bit systems you will not be able
to request a reading till the 5 minute jiffies rollover happens.

Cc: Guenter Roeck <linux@roeck-us.net>
Cc: David Frey <david.frey@sensirion.com>
Signed-off-by: Matt Ranostay <mranostay@gmail.com>
Reviewed-by: Jean Delvare <jdelvare@suse.de>
Fixes: 7c84f7f80d6fc ("hwmon: add support for Sensirion SHT3x sensors")
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
7 years agos390/ftrace/jprobes: Fix conflict between jprobes and function graph tracing
Jiri Olsa [Sun, 31 Jul 2016 12:52:53 +0000 (14:52 +0200)]
s390/ftrace/jprobes: Fix conflict between jprobes and function graph tracing

This fixes the same issue Steven already fixed for x86
in following commit:

  237d28db036e ftrace/jprobes/x86: Fix conflict between jprobes and function graph tracing

It fixes the crash, that happens when function graph tracing
and jprobes are used simultaneously. Please refer to above
commit for details.

Signed-off-by: Jiri Olsa <jolsa@kernel.org>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Acked-by: Steven Rostedt <rostedt@goodmis.org>
7 years agos390: Define AT_VECTOR_SIZE_ARCH for ARCH_DLINFO
James Hogan [Sun, 31 Jul 2016 12:11:11 +0000 (14:11 +0200)]
s390: Define AT_VECTOR_SIZE_ARCH for ARCH_DLINFO

AT_VECTOR_SIZE_ARCH should be defined with the maximum number of
NEW_AUX_ENT entries that ARCH_DLINFO can contain, but it wasn't defined
for s390 at all even though ARCH_DLINFO can contain one NEW_AUX_ENT when
VDSO is enabled.

This shouldn't be a problem as AT_VECTOR_SIZE_BASE includes space for
AT_BASE_PLATFORM which s390 doesn't use, but lets define it now and add
the comment above ARCH_DLINFO as found in several other architectures to
remind future modifiers of ARCH_DLINFO to keep AT_VECTOR_SIZE_ARCH up to
date.

Fixes: b020632e40c3 ("[S390] introduce vdso on s390")
Signed-off-by: James Hogan <james.hogan@imgtec.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: linux-s390@vger.kernel.org
7 years agos390/zcrypt: fix possible memory leak in ap_module_init()
Wei Yongjun [Sun, 31 Jul 2016 09:47:06 +0000 (11:47 +0200)]
s390/zcrypt: fix possible memory leak in ap_module_init()

ap_configuration is malloced in ap_module_init() and should be freed
before leaving from the error handling cases, otherwise it may cause
memory leak.

Signed-off-by: Wei Yongjun <weiyj.lk@gmail.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
7 years agos390/numa: only set possible nodes within node_possible_map
Heiko Carstens [Thu, 28 Jul 2016 16:14:29 +0000 (18:14 +0200)]
s390/numa: only set possible nodes within node_possible_map

Make sure that only those nodes appear in the node_possible_map that
may actually be used. Usually that means that the node online and
possible maps are identical. For mode "plain" we only have one node,
for mode "emu" we have "emu_nodes" nodes.

Before this the possible map included (with default config) 16 nodes
while usually only one was used. That made a couple of loops that
iterated over all possible nodes do more work than necessary.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Acked-by: Michael Holzheu <holzheu@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
7 years agos390/als: fix compile with gcov enabled
Heiko Carstens [Fri, 29 Jul 2016 06:36:35 +0000 (08:36 +0200)]
s390/als: fix compile with gcov enabled

Fix this one when gcov is enabled:

arch/s390/kernel/als.o:(.data+0x118): undefined reference to `__gcov_merge_add'
arch/s390/kernel/als.o: In function `_GLOBAL__sub_I_65535_0_verify_facilities':
(.text.startup+0x8): undefined reference to `__gcov_init'

Please merge with "s390/als: convert architecture level set code to C".

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
7 years agos390/facilities: do not generate DWORDS define anymore
Heiko Carstens [Wed, 29 Jun 2016 22:48:42 +0000 (00:48 +0200)]
s390/facilities: do not generate DWORDS define anymore

The architecture level set code has been converted to C and doesn't
need a define to figure out array sizes. Since the old code was the
only user of the DWORDS define, we can get rid of it again.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Reviewed-by: Sascha Silbe <silbe@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
7 years agos390/als: print missing facilities on facility mismatch
Heiko Carstens [Wed, 29 Jun 2016 21:56:47 +0000 (23:56 +0200)]
s390/als: print missing facilities on facility mismatch

If the kernel needs more facilities to run than the machine provides
it is running on, print the facility bit numbers which are missing.

This allows to easily tell what went wrong and if simply the machine
does not provide a required facility or if either the kernel or the
hypervisor may have a bug.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Reviewed-by: Sascha Silbe <silbe@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
7 years agos390/als: print machine type on facility mismatch
Heiko Carstens [Tue, 28 Jun 2016 12:40:07 +0000 (14:40 +0200)]
s390/als: print machine type on facility mismatch

If we have a facility mismatch the kernel only emits a warning that
the processor is not recent enough and stops operating. This doesn't
give us a lot of an idea of what actually went wrong.

As a first step print the machine type in addition.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Reviewed-by: Sascha Silbe <silbe@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
7 years agos390/als: convert architecture level set code to C
Heiko Carstens [Mon, 27 Jun 2016 13:52:38 +0000 (15:52 +0200)]
s390/als: convert architecture level set code to C

There is no reason to have this code in assembly language. Therefore
convert it to C.

Note that this code needs special treatment: it is called very early
and one of the side effects is that e.g. the bss section is not
cleared. Therefore the preferred way for static variables is to put
them on the stack which has a size of 16KB.

There is no functional change with this patch.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Reviewed-by: Sascha Silbe <silbe@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
7 years agos390/sclp: move uninitialized data to data section
Heiko Carstens [Thu, 30 Jun 2016 08:24:18 +0000 (10:24 +0200)]
s390/sclp: move uninitialized data to data section

The early sclp code may be called before the bss section is
cleared. Therefore move all variables to the data section.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
7 years agos390/zcrypt: Fix zcrypt suspend/resume behavior
Ingo Tuchscherer [Mon, 25 Jul 2016 12:52:28 +0000 (14:52 +0200)]
s390/zcrypt: Fix zcrypt suspend/resume behavior

The device suspend call triggers all ap devices to fetch potentially
available response messages from the queues. Therefore the
corresponding zcrypt device, that is allocated asynchronously after
ap device probing, needs to be fully prepared. This race condition
could lead to uninitialized response buffers while trying to read
from the queues.

Introduce a new callback within the ap layer to get noticed when a
zcrypt device is fully prepared. Additional checks prevent reading
from devices that are not fully prepared.

Signed-off-by: Ingo Tuchscherer <ingo.tuchscherer@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
7 years agos390/cio: fix premature wakeup during chp configure
Sebastian Ott [Tue, 19 Jul 2016 08:53:35 +0000 (10:53 +0200)]
s390/cio: fix premature wakeup during chp configure

We store requests for channel path configure operations in an array but
maintain an additional cfg_busy variable (indicating if we have requests
stored in said array). When 2 tasks request a channel path configure
operation cfg_busy could be set to 0 even if we still have unprocessed
requests. This would lead to the second task being woken up although its
request was not processed yet.

Fix that by getting rid of cfg_busy and use the chp_cfg_task array
in the wake up condition.

Signed-off-by: Sebastian Ott <sebott@linux.vnet.ibm.com>
Reviewed-by: Peter Oberparleiter <oberpar@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
7 years agos390/cio: convert cfg_lock mutex to spinlock
Sebastian Ott [Tue, 19 Jul 2016 08:43:26 +0000 (10:43 +0200)]
s390/cio: convert cfg_lock mutex to spinlock

cfg_lock is never held long and we don't want to sleep while the
lock is being held. Thus it can be converted to a simple spinlock.
In addition we can now use the lock during the evaluation of a
wake_up condition.

Signed-off-by: Sebastian Ott <sebott@linux.vnet.ibm.com>
Reviewed-by: Peter Oberparleiter <oberpar@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
7 years agos390/mm: clean up pte/pmd encoding
Gerald Schaefer [Mon, 18 Jul 2016 12:35:13 +0000 (14:35 +0200)]
s390/mm: clean up pte/pmd encoding

The hugetlbfs pte<->pmd conversion functions currently assume that the pmd
bit layout is consistent with the pte layout, which is not really true.

The SW read and write bits are encoded as the sequence "wr" in a pte, but
in a pmd it is "rw". The hugetlbfs conversion assumes that the sequence
is identical in both cases, which results in swapped read and write bits
in the pmd. In practice this is not a problem, because those pmd bits are
only relevant for THP pmds and not for hugetlbfs pmds. The hugetlbfs code
works on (fake) ptes, and the converted pte bits are correct.

There is another variation in pte/pmd encoding which affects dirty
prot-none ptes/pmds. In this case, a pmd has both its HW read-only and
invalid bit set, while it is only the invalid bit for a pte. This also has
no effect in practice, but it should better be consistent.

This patch fixes both inconsistencies by changing the SW read/write bit
layout for pmds as well as the PAGE_NONE encoding for ptes. It also makes
the hugetlbfs conversion functions more robust by introducing a
move_set_bit() macro that uses the pte/pmd bit #defines instead of
constant shifts.

Signed-off-by: Gerald Schaefer <gerald.schaefer@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>