platform/kernel/linux-rpi.git
5 years agomm/slab: validate cache membership under freelist hardening
Kees Cook [Fri, 12 Jul 2019 03:53:23 +0000 (20:53 -0700)]
mm/slab: validate cache membership under freelist hardening

Patch series "mm/slab: Improved sanity checking".

This adds defenses against slab cache confusion (as seen in real-world
exploits[1]) and gracefully handles type confusions when trying to look
up slab caches from an arbitrary page.  (Also is patch 3: new LKDTM
tests for these defenses as well as for the existing double-free
detection.

This patch (of 3):

When building under CONFIG_SLAB_FREELIST_HARDENING, it makes sense to
perform sanity-checking on the assumed slab cache during
kmem_cache_free() to make sure the kernel doesn't mix freelists across
slab caches and corrupt memory (as seen in the exploitation of flaws
like CVE-2018-9568[1]).  Note that the prior code might WARN() but still
corrupt memory (i.e.  return the assumed cache instead of the owned
cache).

There is no noticeable performance impact (changes are within noise).
Measuring parallel kernel builds, I saw the following with
CONFIG_SLAB_FREELIST_HARDENED, before and after this patch:

before:

Run times: 288.85 286.53 287.09 287.07 287.21
Min: 286.53 Max: 288.85 Mean: 287.35 Std Dev: 0.79

after:

Run times: 289.58 287.40 286.97 287.20 287.01
Min: 286.97 Max: 289.58 Mean: 287.63 Std Dev: 0.99

Delta: 0.1% which is well below the standard deviation

[1] https://github.com/ThomasKing2014/slides/raw/master/Building%20universal%20Android%20rooting%20with%20a%20type%20confusion%20vulnerability.pdf

Link: http://lkml.kernel.org/r/20190530045017.15252-2-keescook@chromium.org
Signed-off-by: Kees Cook <keescook@chromium.org>
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>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Alexander Popov <alex.popov@linux.com>
Cc: Alexander Potapenko <glider@google.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>
5 years agoocfs2: use kmemdup rather than duplicating its implementation
Fuqian Huang [Fri, 12 Jul 2019 03:53:19 +0000 (20:53 -0700)]
ocfs2: use kmemdup rather than duplicating its implementation

kmemdup is introduced to duplicate a region of memory in a neat way.

Rather than kmalloc/kzalloc + memcpy, which the programmer needs to
write the size twice (sometimes lead to mistakes), kmemdup improves
readability, leads to smaller code and also reduce the chances of
mistakes.

Suggestion to use kmemdup rather than using kmalloc/kzalloc + memcpy.

[akpm@linux-foundation.org: coding style fixes]
Link: http://lkml.kernel.org/r/20190703163147.881-1-huangfq.daxian@gmail.com
Signed-off-by: Fuqian Huang <huangfq.daxian@gmail.com>
Reviewed-by: Joseph Qi <joseph.qi@linux.alibaba.com>
Cc: Mark Fasheh <mark@fasheh.com>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Junxiao Bi <junxiao.bi@oracle.com>
Cc: Changwei Ge <gechangwei@live.cn>
Cc: Gang He <ghe@suse.com>
Cc: Jun Piao <piaojun@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agofs/ocfs2/dlmglue.c: unneeded variable: "status"
Hariprasad Kelam [Fri, 12 Jul 2019 03:53:16 +0000 (20:53 -0700)]
fs/ocfs2/dlmglue.c: unneeded variable: "status"

fix below issue reported by coccicheck

  fs/ocfs2/dlmglue.c:4410:5-11: Unneeded variable: "status". Return "0" on line 4428

We can not change return type of ocfs2_downconvert_thread as its
registered as callback of kthread_create.

Link: http://lkml.kernel.org/r/20190702183237.GA13975@hari-Inspiron-1545
Signed-off-by: Hariprasad Kelam <hariprasad.kelam@gmail.com>
Reviewed-by: Joseph Qi <joseph.qi@linux.alibaba.com>
Cc: Mark Fasheh <mark@fasheh.com>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Junxiao Bi <junxiao.bi@oracle.com>
Cc: Changwei Ge <gechangwei@live.cn>
Cc: Gang He <ghe@suse.com>
Cc: Jun Piao <piaojun@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoocfs: no need to check return value of debugfs_create functions
Greg Kroah-Hartman [Fri, 12 Jul 2019 03:53:12 +0000 (20:53 -0700)]
ocfs: no need to check return value of debugfs_create functions

When calling debugfs functions, there is no need to ever check the
return value.  The function can work or not, but the code logic should
never do something different based on this.

Also, because there is no need to save the file dentry, remove all of
the variables that were being saved, and just recursively delete the
whole directory when shutting down, saving a lot of logic and local
variables.

[gregkh@linuxfoundation.org: v2]
Link: http://lkml.kernel.org/r/20190613055455.GE19717@kroah.com
Link: http://lkml.kernel.org/r/20190612152912.GA19151@kroah.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Reviewed-by: Joseph Qi <joseph.qi@linux.alibaba.com>
Cc: Mark Fasheh <mark@fasheh.com>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Joseph Qi <joseph.qi@linux.alibaba.com>
Cc: Jia Guo <guojia12@huawei.com>
Cc: Junxiao Bi <junxiao.bi@oracle.com>
Cc: Changwei Ge <gechangwei@live.cn>
Cc: Gang He <ghe@suse.com>
Cc: Jun Piao <piaojun@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoocfs2: add first lock wait time in locking_state
Gang He [Fri, 12 Jul 2019 03:53:09 +0000 (20:53 -0700)]
ocfs2: add first lock wait time in locking_state

ocfs2 file system uses locking_state file under debugfs to dump each
ocfs2 file system's dlm lock resources, but the users ever encountered
some hang(deadlock) problems in ocfs2 file system.  I'd like to add
first lock wait time in locking_state file, which can help the upper
scripts detect these deadlock problems via comparing the first lock wait
time with the current time.

Link: http://lkml.kernel.org/r/20190611015414.27754-3-ghe@suse.com
Signed-off-by: Gang He <ghe@suse.com>
Reviewed-by: Joseph Qi <joseph.qi@linux.alibaba.com>
Cc: Mark Fasheh <mark@fasheh.com>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Junxiao Bi <junxiao.bi@oracle.com>
Cc: Changwei Ge <gechangwei@live.cn>
Cc: Gang He <ghe@suse.com>
Cc: Jun Piao <piaojun@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoocfs2: add locking filter debugfs file
Gang He [Fri, 12 Jul 2019 03:53:05 +0000 (20:53 -0700)]
ocfs2: add locking filter debugfs file

Add locking filter debugfs file, which is used to filter lock resources
dump from locking_state debugfs file.  We use d_filter_secs field to
filter lock resources dump, the default d_filter_secs(0) value filters
nothing, otherwise, only dump the last N seconds active lock resources.
This enhancement can avoid dumping lots of old records.  The
d_filter_secs value can be changed via locking_filter file.

[akpm@linux-foundation.org: fix undefined reference to `__udivdi3']
Link: http://lkml.kernel.org/r/20190611015414.27754-2-ghe@suse.com
Signed-off-by: Gang He <ghe@suse.com>
Reviewed-by: Joseph Qi <joseph.qi@linux.alibaba.com>
Acked-by: Randy Dunlap <rdunlap@infradead.org> [build-tested]
Cc: Mark Fasheh <mark@fasheh.com>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Junxiao Bi <junxiao.bi@oracle.com>
Cc: Changwei Ge <gechangwei@live.cn>
Cc: Gang He <ghe@suse.com>
Cc: Jun Piao <piaojun@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoocfs2: add last unlock times in locking_state
Gang He [Fri, 12 Jul 2019 03:53:02 +0000 (20:53 -0700)]
ocfs2: add last unlock times in locking_state

ocfs2 file system uses locking_state file under debugfs to dump each
ocfs2 file system's dlm lock resources, but the dlm lock resources in
memory are becoming more and more after the files were touched by the
user.  it will become a bit difficult to analyze these dlm lock resource
records in locking_state file by the upper scripts, though some files
are not active for now, which were accessed long time ago.

Then, I'd like to add last pr/ex unlock times in locking_state file for
each dlm lock resource record, the the upper scripts can use last unlock
time to filter inactive dlm lock resource record.

Link: http://lkml.kernel.org/r/20190611015414.27754-1-ghe@suse.com
Signed-off-by: Gang He <ghe@suse.com>
Reviewed-by: Joseph Qi <joseph.qi@linux.alibaba.com>
Cc: Mark Fasheh <mark@fasheh.com>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Junxiao Bi <junxiao.bi@oracle.com>
Cc: Changwei Ge <gechangwei@live.cn>
Cc: Gang He <ghe@suse.com>
Cc: Jun Piao <piaojun@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoocfs2/dlm: use struct_size() helper
Gustavo A. R. Silva [Fri, 12 Jul 2019 03:52:58 +0000 (20:52 -0700)]
ocfs2/dlm: use struct_size() helper

One of the more common cases of allocation size calculations is finding
the size of a structure that has a zero-sized array at the end, along
with memory for some number of elements for that array.  For example:

  struct dlm_migratable_lockres
  {
          ...
          struct dlm_migratable_lock ml[0];  // 16 bytes each, begins at byte 112
  };

Make use of the struct_size() helper instead of an open-coded version in
order to avoid any potential type mistakes.

So, replace the following form:

   sizeof(struct dlm_migratable_lockres) + (mres->num_locks * sizeof(struct dlm_migratable_lock))

with:

   struct_size(mres, ml, mres->num_locks)

Notice that, in this case, variable sz is not necessary, hence it is
removed.

This code was detected with the help of Coccinelle.

Link: http://lkml.kernel.org/r/20190605204926.GA24467@embeddedor
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Reviewed-by: Joseph Qi <joseph.qi@linux.alibaba.com>
Cc: Mark Fasheh <mark@fasheh.com>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Junxiao Bi <junxiao.bi@oracle.com>
Cc: Changwei Ge <gechangwei@live.cn>
Cc: Gang He <ghe@suse.com>
Cc: Jun Piao <piaojun@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agofs: ocfs: fix spelling mistake "hearbeating" -> "heartbeat"
ChenGang [Fri, 12 Jul 2019 03:52:55 +0000 (20:52 -0700)]
fs: ocfs: fix spelling mistake "hearbeating" -> "heartbeat"

There are some spelling mistakes in ocfs, fix it.

Link: http://lkml.kernel.org/r/1558964623-106628-1-git-send-email-cg.chen@huawei.com
Signed-off-by: ChenGang <cg.chen@huawei.com>
Cc: Mark Fasheh <mark@fasheh.com>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Junxiao Bi <junxiao.bi@oracle.com>
Cc: Joseph Qi <jiangqi903@gmail.com>
Cc: Changwei Ge <gechangwei@live.cn>
Cc: Gang He <ghe@suse.com>
Cc: Jun Piao <piaojun@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agosh: prevent warnings when using iounmap
Sam Ravnborg [Fri, 12 Jul 2019 03:52:52 +0000 (20:52 -0700)]
sh: prevent warnings when using iounmap

When building drm/exynos for sh, as part of an allmodconfig build, the
following warning triggered:

  exynos7_drm_decon.c: In function `decon_remove':
  exynos7_drm_decon.c:769:24: warning: unused variable `ctx'
    struct decon_context *ctx = dev_get_drvdata(&pdev->dev);

The ctx variable is only used as argument to iounmap().

In sh - allmodconfig CONFIG_MMU is not defined
so it ended up in:

\#define __iounmap(addr) do { } while (0)
\#define iounmap __iounmap

Fix the warning by introducing a static inline function for iounmap.

This is similar to several other architectures.

Link: http://lkml.kernel.org/r/20190622114208.24427-1-sam@ravnborg.org
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Reviewed-by: Geert Uytterhoeven <geert+renesas@glider.be>
Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Cc: Rich Felker <dalias@libc.org>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Mark Brown <broonie@kernel.org>
Cc: Inki Dae <inki.dae@samsung.com>
Cc: Krzysztof Kozlowski <krzk@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agosh: config: remove left-over BACKLIGHT_LCD_SUPPORT
Krzysztof Kozlowski [Fri, 12 Jul 2019 03:52:48 +0000 (20:52 -0700)]
sh: config: remove left-over BACKLIGHT_LCD_SUPPORT

CONFIG_BACKLIGHT_LCD_SUPPORT was removed in 8c5dc8d9f19c ("video:
backlight: Remove useless BACKLIGHT_LCD_SUPPORT kernel symbol").
Options protected by CONFIG_BACKLIGHT_LCD_SUPPORT are now available
directly.

Link: http://lkml.kernel.org/r/20190603191925.20659-1-krzk@kernel.org
Signed-off-by: Krzysztof Kozlowski <krzk@kernel.org>
Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Cc: Rich Felker <dalias@libc.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoarch/sh/configs/sdk7786_defconfig: remove CONFIG_LOGFS
Kefeng Wang [Fri, 12 Jul 2019 03:52:45 +0000 (20:52 -0700)]
arch/sh/configs/sdk7786_defconfig: remove CONFIG_LOGFS

After commit 1d0fd57a50aa ("logfs: remove from tree"), logfs was
removed, drop CONFIG_LOGFS from all defconfigs.

Link: http://lkml.kernel.org/r/20190530021032.190639-1-wangkefeng.wang@huawei.com
Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoscripts/spelling.txt: add more spellings to spelling.txt
Colin Ian King [Fri, 12 Jul 2019 03:52:42 +0000 (20:52 -0700)]
scripts/spelling.txt: add more spellings to spelling.txt

Here are some of the more common spelling mistakes and typos that I've
found while fixing up spelling mistakes in the kernel over the past few
months.  Developers keep on coming up with more inventive ways to spell
words.

Link: http://lkml.kernel.org/r/20190618134807.9729-1-colin.king@canonical.com
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoscripts/decode_stacktrace: Accept dash/underscore in modules
Evan Green [Fri, 12 Jul 2019 03:52:39 +0000 (20:52 -0700)]
scripts/decode_stacktrace: Accept dash/underscore in modules

The manpage for modprobe mentions that dashes and underscores are
treated interchangeably in module names.  The stack trace dumps seem to
print module names with underscores.  Use bash to replace _ with the
pattern [-_] so that file names with dashes or underscores can be found.

For example, this line:
[   27.919759]  hda_widget_sysfs_init+0x2b8/0x3a5 [snd_hda_core]

should find a module named snd-hda-core.ko.

Link: http://lkml.kernel.org/r/20190531205926.42474-1-evgreen@chromium.org
Signed-off-by: Evan Green <evgreen@chromium.org>
Reviewed-by: Douglas Anderson <dianders@chromium.org>
Acked-by: Konstantin Khlebnikov <khlebnikov@yandex-team.ru>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Cc: Douglas Anderson <dianders@chromium.org>
Cc: Evan Green <evgreen@chromium.org>
Cc: Nicolas Boichat <drinkcat@chromium.org>
Cc: Marc Zyngier <marc.zyngier@arm.com>
Cc: Manuel Traut <manut@linutronix.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoscripts/spelling.txt: add spelling fix for prohibited
Chris Paterson [Fri, 12 Jul 2019 03:52:36 +0000 (20:52 -0700)]
scripts/spelling.txt: add spelling fix for prohibited

Misspelling 'prohibited' is quite common in the real world, although
surprisingly not so much in the Linux Kernel.  In addition to fixing the
typo we may as well add it to the spelling checker.

Also adding the present participle (prohibiting).

Link: http://lkml.kernel.org/r/20190514153341.22540-1-chris.paterson2@renesas.com
Fixes: 5bf2fbbef50c ("clk: renesas: cpg-mssr: Add r8a77470 support")
Signed-off-by: Chris Paterson <chris.paterson2@renesas.com>
Acked-by: Stephen Boyd <sboyd@kernel.org>
Acked-by: Geert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoscripts/spelling.txt: drop "sepc" from the misspelling list
Paul Walmsley [Fri, 12 Jul 2019 03:52:33 +0000 (20:52 -0700)]
scripts/spelling.txt: drop "sepc" from the misspelling list

The RISC-V architecture has a register named the "Supervisor Exception
Program Counter", or "sepc".  This abbreviation triggers checkpatch.pl's
misspelling detector, resulting in noise in the checkpatch output.  The
risk that this noise could cause more useful warnings to be missed seems
to outweigh the harm of an occasional misspelling of "spec".  Thus drop
the "sepc" entry from the misspelling list.

[akpm@linux-foundation.org: fix existing "sepc" instances, per Joe]
Link: http://lkml.kernel.org/r/20190518210037.13674-1-paul.walmsley@sifive.com
Signed-off-by: Paul Walmsley <paul.walmsley@sifive.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>
5 years agoscripts/decode_stacktrace: look for modules with .ko.debug extension
Nicolas Boichat [Fri, 12 Jul 2019 03:52:30 +0000 (20:52 -0700)]
scripts/decode_stacktrace: look for modules with .ko.debug extension

In Chromium OS kernel builds, we split the debug information as .ko.debug
files, and that's what decode_stacktrace.sh needs to use.

Relax objfile matching rule to allow any .ko* file to be matched.

[drinkcat@chromium.org: add quotes around name pattern]
Link: http://lkml.kernel.org/r/20190528103346.42720-1-drinkcat@chromium.org
Link: http://lkml.kernel.org/r/20190521234148.64060-1-drinkcat@chromium.org
Signed-off-by: Nicolas Boichat <drinkcat@chromium.org>
Cc: Marc Zyngier <marc.zyngier@arm.com>
Cc: Konstantin Khlebnikov <khlebnikov@yandex-team.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoscripts/decode_stacktrace: match basepath using shell prefix operator, not regex
Nicolas Boichat [Fri, 12 Jul 2019 03:52:27 +0000 (20:52 -0700)]
scripts/decode_stacktrace: match basepath using shell prefix operator, not regex

The basepath may contain special characters, which would confuse the regex
matcher.  ${var#prefix} does the right thing.

Link: http://lkml.kernel.org/r/20190518055946.181563-1-drinkcat@chromium.org
Fixes: 67a28de47faa8358 ("scripts/decode_stacktrace: only strip base path when a prefix of the path")
Signed-off-by: Nicolas Boichat <drinkcat@chromium.org>
Reviewed-by: Stephen Boyd <swboyd@chromium.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoinclude/linux/dmar.h: replace single-char identifiers in macros
Qian Cai [Fri, 12 Jul 2019 03:52:24 +0000 (20:52 -0700)]
include/linux/dmar.h: replace single-char identifiers in macros

There are a few macros in IOMMU have single-char identifiers make the code
hard to read and debug.  Replace them with meaningful names.

Link: http://lkml.kernel.org/r/1559566783-13627-1-git-send-email-cai@lca.pw
Signed-off-by: Qian Cai <cai@lca.pw>
Suggested-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Joerg Roedel <jroedel@suse.de>
Cc: Robin Murphy <robin.murphy@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoMAINTAINERS: nilfs2: update email address
Ryusuke Konishi [Fri, 12 Jul 2019 03:52:21 +0000 (20:52 -0700)]
MAINTAINERS: nilfs2: update email address

Change my email since lab.ntt.co.jp email domain has been deprecated due
to company policy.

Link: http://lkml.kernel.org/r/1562495153-8166-1-git-send-email-konishi.ryusuke@gmail.com
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agonilfs2: do not use unexported cpu_to_le32()/le32_to_cpu() in uapi header
Masahiro Yamada [Fri, 12 Jul 2019 03:52:18 +0000 (20:52 -0700)]
nilfs2: do not use unexported cpu_to_le32()/le32_to_cpu() in uapi header

cpu_to_le32/le32_to_cpu is defined in include/linux/byteorder/generic.h,
which is not exported to user-space.

UAPI headers must use the ones prefixed with double-underscore.

Detected by compile-testing exported headers:

  include/linux/nilfs2_ondisk.h: In function `nilfs_checkpoint_set_snapshot':
  include/linux/nilfs2_ondisk.h:536:17: error: implicit declaration of function `cpu_to_le32' [-Werror=implicit-function-declaration]
    cp->cp_flags = cpu_to_le32(le32_to_cpu(cp->cp_flags) |  \
                   ^
  include/linux/nilfs2_ondisk.h:552:1: note: in expansion of macro `NILFS_CHECKPOINT_FNS'
   NILFS_CHECKPOINT_FNS(SNAPSHOT, snapshot)
   ^~~~~~~~~~~~~~~~~~~~
  include/linux/nilfs2_ondisk.h:536:29: error: implicit declaration of function `le32_to_cpu' [-Werror=implicit-function-declaration]
    cp->cp_flags = cpu_to_le32(le32_to_cpu(cp->cp_flags) |  \
                               ^
  include/linux/nilfs2_ondisk.h:552:1: note: in expansion of macro `NILFS_CHECKPOINT_FNS'
   NILFS_CHECKPOINT_FNS(SNAPSHOT, snapshot)
   ^~~~~~~~~~~~~~~~~~~~
  include/linux/nilfs2_ondisk.h: In function `nilfs_segment_usage_set_clean':
  include/linux/nilfs2_ondisk.h:622:19: error: implicit declaration of function `cpu_to_le64' [-Werror=implicit-function-declaration]
    su->su_lastmod = cpu_to_le64(0);
                     ^~~~~~~~~~~

Link: http://lkml.kernel.org/r/20190605053006.14332-1-yamada.masahiro@socionext.com
Fixes: e63e88bc53ba ("nilfs2: move ioctl interface and disk layout to uapi separately")
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Acked-by: Ryusuke Konishi <konishi.ryusuke@gmail.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Greg KH <gregkh@linuxfoundation.org>
Cc: Joe Perches <joe@perches.com>
Cc: <stable@vger.kernel.org> [4.9+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm/z3fold.c: lock z3fold page before __SetPageMovable()
Henry Burns [Fri, 12 Jul 2019 03:52:14 +0000 (20:52 -0700)]
mm/z3fold.c: lock z3fold page before __SetPageMovable()

Following zsmalloc.c's example we call trylock_page() and unlock_page().
Also make z3fold_page_migrate() assert that newpage is passed in locked,
as per the documentation.

[akpm@linux-foundation.org: fix trylock_page return value test, per Shakeel]
Link: http://lkml.kernel.org/r/20190702005122.41036-1-henryburns@google.com
Link: http://lkml.kernel.org/r/20190702233538.52793-1-henryburns@google.com
Signed-off-by: Henry Burns <henryburns@google.com>
Suggested-by: Vitaly Wool <vitalywool@gmail.com>
Acked-by: Vitaly Wool <vitalywool@gmail.com>
Acked-by: David Rientjes <rientjes@google.com>
Reviewed-by: Shakeel Butt <shakeelb@google.com>
Cc: Vitaly Vul <vitaly.vul@sony.com>
Cc: Mike Rapoport <rppt@linux.vnet.ibm.com>
Cc: Xidong Wang <wangxidong_97@163.com>
Cc: Jonathan Adams <jwadams@google.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm/memcontrol: fix wrong statistics in memory.stat
Yafang Shao [Fri, 12 Jul 2019 03:52:11 +0000 (20:52 -0700)]
mm/memcontrol: fix wrong statistics in memory.stat

When we calculate total statistics for memcg1_stats and memcg1_events,
we use the the index 'i' in the for loop as the events index.  Actually
we should use memcg1_stats[i] and memcg1_events[i] as the events index.

Link: http://lkml.kernel.org/r/1562116978-19539-1-git-send-email-laoar.shao@gmail.com
Fixes: 42a300353577 ("mm: memcontrol: fix recursive statistics correctness & scalabilty").
Signed-off-by: Yafang Shao <laoar.shao@gmail.com
Reviewed-by: Shakeel Butt <shakeelb@google.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Yafang Shao <shaoyafang@didiglobal.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm/nvdimm: add is_ioremap_addr and use that to check ioremap address
Aneesh Kumar K.V [Fri, 12 Jul 2019 03:52:08 +0000 (20:52 -0700)]
mm/nvdimm: add is_ioremap_addr and use that to check ioremap address

Architectures like powerpc use different address range to map ioremap
and vmalloc range.  The memunmap() check used by the nvdimm layer was
wrongly using is_vmalloc_addr() to check for ioremap range which fails
for ppc64.  This result in ppc64 not freeing the ioremap mapping.  The
side effect of this is an unbind failure during module unload with
papr_scm nvdimm driver

Link: http://lkml.kernel.org/r/20190701134038.14165-1-aneesh.kumar@linux.ibm.com
Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
Fixes: b5beae5e224f ("powerpc/pseries: Add driver for PAPR SCM regions")
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm: vmscan: scan anonymous pages on file refaults
Kuo-Hsin Yang [Fri, 12 Jul 2019 03:52:04 +0000 (20:52 -0700)]
mm: vmscan: scan anonymous pages on file refaults

When file refaults are detected and there are many inactive file pages,
the system never reclaim anonymous pages, the file pages are dropped
aggressively when there are still a lot of cold anonymous pages and
system thrashes.  This issue impacts the performance of applications
with large executable, e.g.  chrome.

With this patch, when file refault is detected, inactive_list_is_low()
always returns true for file pages in get_scan_count() to enable
scanning anonymous pages.

The problem can be reproduced by the following test program.

---8<---
void fallocate_file(const char *filename, off_t size)
{
struct stat st;
int fd;

if (!stat(filename, &st) && st.st_size >= size)
return;

fd = open(filename, O_WRONLY | O_CREAT, 0600);
if (fd < 0) {
perror("create file");
exit(1);
}
if (posix_fallocate(fd, 0, size)) {
perror("fallocate");
exit(1);
}
close(fd);
}

long *alloc_anon(long size)
{
long *start = malloc(size);
memset(start, 1, size);
return start;
}

long access_file(const char *filename, long size, long rounds)
{
int fd, i;
volatile char *start1, *end1, *start2;
const int page_size = getpagesize();
long sum = 0;

fd = open(filename, O_RDONLY);
if (fd == -1) {
perror("open");
exit(1);
}

/*
 * Some applications, e.g. chrome, use a lot of executable file
 * pages, map some of the pages with PROT_EXEC flag to simulate
 * the behavior.
 */
start1 = mmap(NULL, size / 2, PROT_READ | PROT_EXEC, MAP_SHARED,
      fd, 0);
if (start1 == MAP_FAILED) {
perror("mmap");
exit(1);
}
end1 = start1 + size / 2;

start2 = mmap(NULL, size / 2, PROT_READ, MAP_SHARED, fd, size / 2);
if (start2 == MAP_FAILED) {
perror("mmap");
exit(1);
}

for (i = 0; i < rounds; ++i) {
struct timeval before, after;
volatile char *ptr1 = start1, *ptr2 = start2;
gettimeofday(&before, NULL);
for (; ptr1 < end1; ptr1 += page_size, ptr2 += page_size)
sum += *ptr1 + *ptr2;
gettimeofday(&after, NULL);
printf("File access time, round %d: %f (sec)
", i,
       (after.tv_sec - before.tv_sec) +
       (after.tv_usec - before.tv_usec) / 1000000.0);
}
return sum;
}

int main(int argc, char *argv[])
{
const long MB = 1024 * 1024;
long anon_mb, file_mb, file_rounds;
const char filename[] = "large";
long *ret1;
long ret2;

if (argc != 4) {
printf("usage: thrash ANON_MB FILE_MB FILE_ROUNDS
");
exit(0);
}
anon_mb = atoi(argv[1]);
file_mb = atoi(argv[2]);
file_rounds = atoi(argv[3]);

fallocate_file(filename, file_mb * MB);
printf("Allocate %ld MB anonymous pages
", anon_mb);
ret1 = alloc_anon(anon_mb * MB);
printf("Access %ld MB file pages
", file_mb);
ret2 = access_file(filename, file_mb * MB, file_rounds);
printf("Print result to prevent optimization: %ld
",
       *ret1 + ret2);
return 0;
}
---8<---

Running the test program on 2GB RAM VM with kernel 5.2.0-rc5, the program
fills ram with 2048 MB memory, access a 200 MB file for 10 times.  Without
this patch, the file cache is dropped aggresively and every access to the
file is from disk.

  $ ./thrash 2048 200 10
  Allocate 2048 MB anonymous pages
  Access 200 MB file pages
  File access time, round 0: 2.489316 (sec)
  File access time, round 1: 2.581277 (sec)
  File access time, round 2: 2.487624 (sec)
  File access time, round 3: 2.449100 (sec)
  File access time, round 4: 2.420423 (sec)
  File access time, round 5: 2.343411 (sec)
  File access time, round 6: 2.454833 (sec)
  File access time, round 7: 2.483398 (sec)
  File access time, round 8: 2.572701 (sec)
  File access time, round 9: 2.493014 (sec)

With this patch, these file pages can be cached.

  $ ./thrash 2048 200 10
  Allocate 2048 MB anonymous pages
  Access 200 MB file pages
  File access time, round 0: 2.475189 (sec)
  File access time, round 1: 2.440777 (sec)
  File access time, round 2: 2.411671 (sec)
  File access time, round 3: 1.955267 (sec)
  File access time, round 4: 0.029924 (sec)
  File access time, round 5: 0.000808 (sec)
  File access time, round 6: 0.000771 (sec)
  File access time, round 7: 0.000746 (sec)
  File access time, round 8: 0.000738 (sec)
  File access time, round 9: 0.000747 (sec)

Checked the swap out stats during the test [1], 19006 pages swapped out
with this patch, 3418 pages swapped out without this patch. There are
more swap out, but I think it's within reasonable range when file backed
data set doesn't fit into the memory.

$ ./thrash 2000 100 2100 5 1 # ANON_MB FILE_EXEC FILE_NOEXEC ROUNDS
PROCESSES Allocate 2000 MB anonymous pages active_anon: 1613644,
inactive_anon: 348656, active_file: 892, inactive_file: 1384 (kB)
pswpout: 7972443, pgpgin: 478615246 Access 100 MB executable file pages
Access 2100 MB regular file pages File access time, round 0: 12.165,
(sec) active_anon: 1433788, inactive_anon: 478116, active_file: 17896,
inactive_file: 24328 (kB) File access time, round 1: 11.493, (sec)
active_anon: 1430576, inactive_anon: 477144, active_file: 25440,
inactive_file: 26172 (kB) File access time, round 2: 11.455, (sec)
active_anon: 1427436, inactive_anon: 476060, active_file: 21112,
inactive_file: 28808 (kB) File access time, round 3: 11.454, (sec)
active_anon: 1420444, inactive_anon: 473632, active_file: 23216,
inactive_file: 35036 (kB) File access time, round 4: 11.479, (sec)
active_anon: 1413964, inactive_anon: 471460, active_file: 31728,
inactive_file: 32224 (kB) pswpout: 7991449 (+ 19006), pgpgin: 489924366
(+ 11309120)

With 4 processes accessing non-overlapping parts of a large file, 30316
pages swapped out with this patch, 5152 pages swapped out without this
patch.  The swapout number is small comparing to pgpgin.

[1]: https://github.com/vovo/testing/blob/master/mem_thrash.c

Link: http://lkml.kernel.org/r/20190701081038.GA83398@google.com
Fixes: e9868505987a ("mm,vmscan: only evict file pages when we have plenty")
Fixes: 7c5bd705d8f9 ("mm: memcg: only evict file pages when we have plenty")
Signed-off-by: Kuo-Hsin Yang <vovoy@chromium.org>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Sonny Rao <sonnyrao@chromium.org>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Rik van Riel <riel@redhat.com>
Cc: Vladimir Davydov <vdavydov.dev@gmail.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: <stable@vger.kernel.org> [4.12+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoMerge branch 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Thu, 11 Jul 2019 20:54:00 +0000 (13:54 -0700)]
Merge branch 'x86-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull x86 fixes from Thomas Gleixner:
 "A collection of assorted fixes:

   - Fix for the pinned cr0/4 fallout which escaped all testing efforts
     because the kvm-intel module was never loaded when the kernel was
     compiled with CONFIG_PARAVIRT=n. The cr0/4 accessors are moved out
     of line and static key is now solely used in the core code and
     therefore can stay in the RO after init section. So the kvm-intel
     and other modules do not longer reference the (read only) static
     key which the module loader tried to update.

   - Prevent an infinite loop in arch_stack_walk_user() by breaking out
     of the loop once the return address is detected to be 0.

   - Prevent the int3_emulate_call() selftest from corrupting the stack
     when KASAN is enabled. KASASN clobbers more registers than covered
     by the emulated call implementation. Convert the int3_magic()
     selftest to a ASM function so the compiler cannot KASANify it.

   - Unbreak the build with old GCC versions and with the Gold linker by
     reverting the 'Move of _etext to the actual end of .text'. In both
     cases the build fails with 'Invalid absolute R_X86_64_32S
     relocation: _etext'

   - Initialize the context lock for init_mm, which was never an issue
     until the alternatives code started to use a temporary mm for
     patching.

   - Fix a build warning vs. the LOWMEM_PAGES constant where clang
     complains rightfully about a signed integer overflow in the shift
     operation by converting the operand to an ULL.

   - Adjust the misnamed ENDPROC() of common_spurious in the 32bit entry
     code"

* 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/stacktrace: Prevent infinite loop in arch_stack_walk_user()
  x86/asm: Move native_write_cr0/4() out of line
  x86/pgtable/32: Fix LOWMEM_PAGES constant
  x86/alternatives: Fix int3_emulate_call() selftest stack corruption
  x86/entry/32: Fix ENDPROC of common_spurious
  Revert "x86/build: Move _etext to actual end of .text"
  x86/ldt: Initialize the context lock for init_mm

5 years agoMerge branch 'timers-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Thu, 11 Jul 2019 20:52:23 +0000 (13:52 -0700)]
Merge branch 'timers-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull timer fixes from Thomas Gleixner:
 "Two small fixes from the timer departement:

   - Prevent the compiler from converting the nanoseconds adjustment
     loop in the VDSO update function to a division (__udivdi3) by using
     the __iter_div_u64_rem() inline function which exists to prevent
     exactly that problem.

   - Fix the wrong argument order of the GENMASK macro in the NPCM timer
     driver"

* 'timers-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  timekeeping/vsyscall: Use __iter_div_u64_rem()
  clocksource/drivers/npcm: Fix misuse of GENMASK macro

5 years agoMerge branch 'core-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Thu, 11 Jul 2019 20:50:44 +0000 (13:50 -0700)]
Merge branch 'core-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull stacktrace fix from Thomas Gleixner:
 "Fix yet another instance of kernel thread check which ignores that
  kernel threads can call use_mm()"

* 'core-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  stacktrace: Use PF_KTHREAD to check for kernel threads

5 years agoMerge branch 'irq-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Thu, 11 Jul 2019 20:48:36 +0000 (13:48 -0700)]
Merge branch 'irq-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull irq fixes from Thomas Gleixner:
 "Two small fixes for interrupt chip drivers:

   - Prevent UAF in the new RZA1 chip driver

   - Fix the wrong argument order of the GENMASK macro in the GIC code"

* 'irq-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  irqchip/gic-v3-its: Fix misuse of GENMASK macro
  irqchip/renesas-rza1: Prevent use-after-free in rza1_irqc_probe()

5 years agoMerge tag 'acpi-5.3-rc1-2' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael...
Linus Torvalds [Thu, 11 Jul 2019 18:17:09 +0000 (11:17 -0700)]
Merge tag 'acpi-5.3-rc1-2' of git://git./linux/kernel/git/rafael/linux-pm

Pull ACPI fix from Rafael Wysocki:
 "Revert a recent ACPICA commit causing systems to hang at boot time"

* tag 'acpi-5.3-rc1-2' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
  Revert "ACPICA: Update table load object initialization"

5 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next
Linus Torvalds [Thu, 11 Jul 2019 17:55:49 +0000 (10:55 -0700)]
Merge git://git./linux/kernel/git/davem/net-next

Pull networking updates from David Miller:
 "Some highlights from this development cycle:

   1) Big refactoring of ipv6 route and neigh handling to support
      nexthop objects configurable as units from userspace. From David
      Ahern.

   2) Convert explored_states in BPF verifier into a hash table,
      significantly decreased state held for programs with bpf2bpf
      calls, from Alexei Starovoitov.

   3) Implement bpf_send_signal() helper, from Yonghong Song.

   4) Various classifier enhancements to mvpp2 driver, from Maxime
      Chevallier.

   5) Add aRFS support to hns3 driver, from Jian Shen.

   6) Fix use after free in inet frags by allocating fqdirs dynamically
      and reworking how rhashtable dismantle occurs, from Eric Dumazet.

   7) Add act_ctinfo packet classifier action, from Kevin
      Darbyshire-Bryant.

   8) Add TFO key backup infrastructure, from Jason Baron.

   9) Remove several old and unused ISDN drivers, from Arnd Bergmann.

  10) Add devlink notifications for flash update status to mlxsw driver,
      from Jiri Pirko.

  11) Lots of kTLS offload infrastructure fixes, from Jakub Kicinski.

  12) Add support for mv88e6250 DSA chips, from Rasmus Villemoes.

  13) Various enhancements to ipv6 flow label handling, from Eric
      Dumazet and Willem de Bruijn.

  14) Support TLS offload in nfp driver, from Jakub Kicinski, Dirk van
      der Merwe, and others.

  15) Various improvements to axienet driver including converting it to
      phylink, from Robert Hancock.

  16) Add PTP support to sja1105 DSA driver, from Vladimir Oltean.

  17) Add mqprio qdisc offload support to dpaa2-eth, from Ioana
      Radulescu.

  18) Add devlink health reporting to mlx5, from Moshe Shemesh.

  19) Convert stmmac over to phylink, from Jose Abreu.

  20) Add PTP PHC (Physical Hardware Clock) support to mlxsw, from
      Shalom Toledo.

  21) Add nftables SYNPROXY support, from Fernando Fernandez Mancera.

  22) Convert tcp_fastopen over to use SipHash, from Ard Biesheuvel.

  23) Track spill/fill of constants in BPF verifier, from Alexei
      Starovoitov.

  24) Support bounded loops in BPF, from Alexei Starovoitov.

  25) Various page_pool API fixes and improvements, from Jesper Dangaard
      Brouer.

  26) Just like ipv4, support ref-countless ipv6 route handling. From
      Wei Wang.

  27) Support VLAN offloading in aquantia driver, from Igor Russkikh.

  28) Add AF_XDP zero-copy support to mlx5, from Maxim Mikityanskiy.

  29) Add flower GRE encap/decap support to nfp driver, from Pieter
      Jansen van Vuuren.

  30) Protect against stack overflow when using act_mirred, from John
      Hurley.

  31) Allow devmap map lookups from eBPF, from Toke Høiland-Jørgensen.

  32) Use page_pool API in netsec driver, Ilias Apalodimas.

  33) Add Google gve network driver, from Catherine Sullivan.

  34) More indirect call avoidance, from Paolo Abeni.

  35) Add kTLS TX HW offload support to mlx5, from Tariq Toukan.

  36) Add XDP_REDIRECT support to bnxt_en, from Andy Gospodarek.

  37) Add MPLS manipulation actions to TC, from John Hurley.

  38) Add sending a packet to connection tracking from TC actions, and
      then allow flower classifier matching on conntrack state. From
      Paul Blakey.

  39) Netfilter hw offload support, from Pablo Neira Ayuso"

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next: (2080 commits)
  net/mlx5e: Return in default case statement in tx_post_resync_params
  mlx5: Return -EINVAL when WARN_ON_ONCE triggers in mlx5e_tls_resync().
  net: dsa: add support for BRIDGE_MROUTER attribute
  pkt_sched: Include const.h
  net: netsec: remove static declaration for netsec_set_tx_de()
  net: netsec: remove superfluous if statement
  netfilter: nf_tables: add hardware offload support
  net: flow_offload: rename tc_cls_flower_offload to flow_cls_offload
  net: flow_offload: add flow_block_cb_is_busy() and use it
  net: sched: remove tcf block API
  drivers: net: use flow block API
  net: sched: use flow block API
  net: flow_offload: add flow_block_cb_{priv, incref, decref}()
  net: flow_offload: add list handling functions
  net: flow_offload: add flow_block_cb_alloc() and flow_block_cb_free()
  net: flow_offload: rename TCF_BLOCK_BINDER_TYPE_* to FLOW_BLOCK_BINDER_TYPE_*
  net: flow_offload: rename TC_BLOCK_{UN}BIND to FLOW_BLOCK_{UN}BIND
  net: flow_offload: add flow_block_cb_setup_simple()
  net: hisilicon: Add an tx_desc to adapt HI13X1_GMAC
  net: hisilicon: Add an rx_desc to adapt HI13X1_GMAC
  ...

5 years agoMerge tag 'clone3-v5.3' of git://git.kernel.org/pub/scm/linux/kernel/git/brauner...
Linus Torvalds [Thu, 11 Jul 2019 17:09:44 +0000 (10:09 -0700)]
Merge tag 'clone3-v5.3' of git://git./linux/kernel/git/brauner/linux

Pull clone3 system call from Christian Brauner:
 "This adds the clone3 syscall which is an extensible successor to clone
  after we snagged the last flag with CLONE_PIDFD during the 5.2 merge
  window for clone(). It cleanly supports all of the flags from clone()
  and thus all legacy workloads.

  There are few user visible differences between clone3 and clone.
  First, CLONE_DETACHED will cause EINVAL with clone3 so we can reuse
  this flag. Second, the CSIGNAL flag is deprecated and will cause
  EINVAL to be reported. It is superseeded by a dedicated "exit_signal"
  argument in struct clone_args thus freeing up even more flags. And
  third, clone3 gives CLONE_PIDFD a dedicated return argument in struct
  clone_args instead of abusing CLONE_PARENT_SETTID's parent_tidptr
  argument.

  The clone3 uapi is designed to be easy to handle on 32- and 64 bit:

    /* uapi */
    struct clone_args {
            __aligned_u64 flags;
            __aligned_u64 pidfd;
            __aligned_u64 child_tid;
            __aligned_u64 parent_tid;
            __aligned_u64 exit_signal;
            __aligned_u64 stack;
            __aligned_u64 stack_size;
            __aligned_u64 tls;
    };

  and a separate kernel struct is used that uses proper kernel typing:

    /* kernel internal */
    struct kernel_clone_args {
            u64 flags;
            int __user *pidfd;
            int __user *child_tid;
            int __user *parent_tid;
            int exit_signal;
            unsigned long stack;
            unsigned long stack_size;
            unsigned long tls;
    };

  The system call comes with a size argument which enables the kernel to
  detect what version of clone_args userspace is passing in. clone3
  validates that any additional bytes a given kernel does not know about
  are set to zero and that the size never exceeds a page.

  A nice feature is that this patchset allowed us to cleanup and
  simplify various core kernel codepaths in kernel/fork.c by making the
  internal _do_fork() function take struct kernel_clone_args even for
  legacy clone().

  This patch also unblocks the time namespace patchset which wants to
  introduce a new CLONE_TIMENS flag.

  Note, that clone3 has only been wired up for x86{_32,64}, arm{64}, and
  xtensa. These were the architectures that did not require special
  massaging.

  Other architectures treat fork-like system calls individually and
  after some back and forth neither Arnd nor I felt confident that we
  dared to add clone3 unconditionally to all architectures. We agreed to
  leave this up to individual architecture maintainers. This is why
  there's an additional patch that introduces __ARCH_WANT_SYS_CLONE3
  which any architecture can set once it has implemented support for
  clone3. The patch also adds a cond_syscall(clone3) for architectures
  such as nios2 or h8300 that generate their syscall table by simply
  including asm-generic/unistd.h. The hope is to get rid of
  __ARCH_WANT_SYS_CLONE3 and cond_syscall() rather soon"

* tag 'clone3-v5.3' of git://git.kernel.org/pub/scm/linux/kernel/git/brauner/linux:
  arch: handle arches who do not yet define clone3
  arch: wire-up clone3() syscall
  fork: add clone3

5 years agox86/stacktrace: Prevent infinite loop in arch_stack_walk_user()
Eiichi Tsukata [Thu, 11 Jul 2019 02:35:01 +0000 (11:35 +0900)]
x86/stacktrace: Prevent infinite loop in arch_stack_walk_user()

arch_stack_walk_user() checks `if (fp == frame.next_fp)` to prevent a
infinite loop by self reference but it's not enogh for circular reference.

Once a lack of return address is found, there is no point to continue the
loop, so break out.

Fixes: 02b67518e2b1 ("tracing: add support for userspace stacktraces in tracing/iter_ctrl")
Signed-off-by: Eiichi Tsukata <devel@etsukata.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Link: https://lkml.kernel.org/r/20190711023501.963-1-devel@etsukata.com
5 years agoMerge tag 'pidfd-updates-v5.3' of git://git.kernel.org/pub/scm/linux/kernel/git/braun...
Linus Torvalds [Thu, 11 Jul 2019 05:17:21 +0000 (22:17 -0700)]
Merge tag 'pidfd-updates-v5.3' of git://git./linux/kernel/git/brauner/linux

Pull pidfd updates from Christian Brauner:
 "This adds two main features.

   - First, it adds polling support for pidfds. This allows process
     managers to know when a (non-parent) process dies in a race-free
     way.

     The notification mechanism used follows the same logic that is
     currently used when the parent of a task is notified of a child's
     death. With this patchset it is possible to put pidfds in an
     {e}poll loop and get reliable notifications for process (i.e.
     thread-group) exit.

   - The second feature compliments the first one by making it possible
     to retrieve pollable pidfds for processes that were not created
     using CLONE_PIDFD.

     A lot of processes get created with traditional PID-based calls
     such as fork() or clone() (without CLONE_PIDFD). For these
     processes a caller can currently not create a pollable pidfd. This
     is a problem for Android's low memory killer (LMK) and service
     managers such as systemd.

  Both patchsets are accompanied by selftests.

  It's perhaps worth noting that the work done so far and the work done
  in this branch for pidfd_open() and polling support do already see
  some adoption:

   - Android is in the process of backporting this work to all their LTS
     kernels [1]

   - Service managers make use of pidfd_send_signal but will need to
     wait until we enable waiting on pidfds for full adoption.

   - And projects I maintain make use of both pidfd_send_signal and
     CLONE_PIDFD [2] and will use polling support and pidfd_open() too"

[1] https://android-review.googlesource.com/q/topic:%22pidfd+polling+support+4.9+backport%22
    https://android-review.googlesource.com/q/topic:%22pidfd+polling+support+4.14+backport%22
    https://android-review.googlesource.com/q/topic:%22pidfd+polling+support+4.19+backport%22

[2] https://github.com/lxc/lxc/blob/aab6e3eb73c343231cdde775db938994fc6f2803/src/lxc/start.c#L1753

* tag 'pidfd-updates-v5.3' of git://git.kernel.org/pub/scm/linux/kernel/git/brauner/linux:
  tests: add pidfd_open() tests
  arch: wire-up pidfd_open()
  pid: add pidfd_open()
  pidfd: add polling selftests
  pidfd: add polling support

5 years agoMerge tag 'm68k-for-v5.3-tag2' of git://git.kernel.org/pub/scm/linux/kernel/git/geert...
Linus Torvalds [Thu, 11 Jul 2019 04:44:07 +0000 (21:44 -0700)]
Merge tag 'm68k-for-v5.3-tag2' of git://git./linux/kernel/git/geert/linux-m68k

Pull m68k fix from Geert Uytterhoeven:
 "Don't select ARCH_HAS_DMA_PREP_COHERENT for nommu or coldfire.

  This is a fix for an issue detected in next, to avoid introducing
  build failures when merging Christoph's dma-mapping tree later"

* tag 'm68k-for-v5.3-tag2' of git://git.kernel.org/pub/scm/linux/kernel/git/geert/linux-m68k:
  m68k: Don't select ARCH_HAS_DMA_PREP_COHERENT for nommu or coldfire

5 years agoMerge branch 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/gerg/m68knommu
Linus Torvalds [Thu, 11 Jul 2019 04:42:03 +0000 (21:42 -0700)]
Merge branch 'for-next' of git://git./linux/kernel/git/gerg/m68knommu

Pull m68nommu updates from Greg Ungerer:
 "A series of cleanups for the FLAT format binary loader, binfmt_flat,
  from Christoph.

  The end goal is to support no-MMU on RISC-V, and the last patch
  enables that"

* 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/gerg/m68knommu:
  riscv: add binfmt_flat support
  binfmt_flat: don't offset the data start
  binfmt_flat: move the MAX_SHARED_LIBS definition to binfmt_flat.c
  binfmt_flat: remove the persistent argument from flat_get_addr_from_rp
  binfmt_flat: provide an asm-generic/flat.h
  binfmt_flat: make support for old format binaries optional
  binfmt_flat: add a ARCH_HAS_BINFMT_FLAT option
  binfmt_flat: add endianess annotations
  binfmt_flat: use fixed size type for the on-disk format
  binfmt_flat: consolidate two version of flat_v2_reloc_t
  binfmt_flat: remove the unused OLD_FLAT_FLAG_RAM definition
  binfmt_flat: remove the uapi <linux/flat.h> header
  binfmt_flat: replace flat_argvp_envp_on_stack with a Kconfig variable
  binfmt_flat: remove flat_old_ram_flag
  binfmt_flat: provide a default version of flat_get_relocate_addr
  binfmt_flat: remove flat_set_persistent
  binfmt_flat: remove flat_reloc_valid

5 years agoMerge tag 'nfsd-5.3' of git://linux-nfs.org/~bfields/linux
Linus Torvalds [Thu, 11 Jul 2019 04:22:43 +0000 (21:22 -0700)]
Merge tag 'nfsd-5.3' of git://linux-nfs.org/~bfields/linux

Pull nfsd updates from Bruce Fields:
 "Highlights:

   - Add a new /proc/fs/nfsd/clients/ directory which exposes some
     long-requested information about NFSv4 clients (like open files)
     and allows forced revocation of client state.

   - Replace the global duplicate reply cache by a cache per network
     namespace; previously, a request in one network namespace could
     incorrectly match an entry from another, though we haven't seen
     this in production. This is the last remaining container bug that
     I'm aware of; at this point you should be able to run separate
     nfsd's in each network namespace, each with their own set of
     exports, and everything should work.

   - Cleanup and modify lock code to show the pid of lockd as the owner
     of NLM locks. This is the correct version of the bugfix originally
     attempted in b8eee0e90f97 ("lockd: Show pid of lockd for remote
     locks")"

* tag 'nfsd-5.3' of git://linux-nfs.org/~bfields/linux: (34 commits)
  nfsd: Make __get_nfsdfs_client() static
  nfsd: Make two functions static
  nfsd: Fix misuse of strlcpy
  sunrpc/cache: remove the exporting of cache_seq_next
  nfsd: decode implementation id
  nfsd: create xdr_netobj_dup helper
  nfsd: allow forced expiration of NFSv4 clients
  nfsd: create get_nfsdfs_clp helper
  nfsd4: show layout stateids
  nfsd: show lock and deleg stateids
  nfsd4: add file to display list of client's opens
  nfsd: add more information to client info file
  nfsd: escape high characters in binary data
  nfsd: copy client's address including port number to cl_addr
  nfsd4: add a client info file
  nfsd: make client/ directory names small ints
  nfsd: add nfsd/clients directory
  nfsd4: use reference count to free client
  nfsd: rename cl_refcount
  nfsd: persist nfsd filesystem across mounts
  ...

5 years agoMerge tag 'gfs2-for-5.3' of git://git.kernel.org/pub/scm/linux/kernel/git/gfs2/linux...
Linus Torvalds [Thu, 11 Jul 2019 04:20:05 +0000 (21:20 -0700)]
Merge tag 'gfs2-for-5.3' of git://git./linux/kernel/git/gfs2/linux-gfs2

Pull gfs2 updates from Andreas Gruenbacher:
 "Some relatively minor changes for gfs2:

   - An initial batch of obvious cleanups and fixes from Bob's recovery
     patch queue.

   - Two iomap conversion patches and some cleanups from Christoph
     Hellwig.

   - A cosmetic cleanup from Kefeng Wang (Huawei).

   - Another minor fix and cleanup by me"

* tag 'gfs2-for-5.3' of git://git.kernel.org/pub/scm/linux/kernel/git/gfs2/linux-gfs2:
  gfs2: Remove unused gfs2_iomap_alloc argument
  gfs2: don't use buffer_heads in gfs2_allocate_page_backing
  gfs2: use iomap_bmap instead of generic_block_bmap
  gfs2: mark stuffed_readpage static
  gfs2: merge gfs2_writepage_common into gfs2_writepage
  gfs2: merge gfs2_writeback_aops and gfs2_ordered_aops
  gfs2: remove the unused gfs2_stuffed_write_end function
  gfs2: use page_offset in gfs2_page_mkwrite
  gfs2: replace more printk with calls to fs_info and friends
  gfs2: dump fsid when dumping glock problems
  gfs2: simplify gfs2_freeze by removing case
  gfs2: Rename SDF_SHUTDOWN to SDF_WITHDRAWN
  gfs2: Warn when a journal replay overwrites a rgrp with buffers
  gfs2: log which portion of the journal is replayed
  gfs2: eliminate tr_num_revoke_rm
  gfs2: kthread and remount improvements
  gfs2: Use IS_ERR_OR_NULL
  gfs2: Clean up freeing struct gfs2_sbd

5 years agoMerge tag 'ext4_for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso...
Linus Torvalds [Thu, 11 Jul 2019 04:06:01 +0000 (21:06 -0700)]
Merge tag 'ext4_for_linus' of git://git./linux/kernel/git/tytso/ext4

Pull ext4 updates from Ted Ts'o:
 "Many bug fixes and cleanups, and an optimization for case-insensitive
  lookups"

* tag 'ext4_for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4:
  ext4: fix coverity warning on error path of filename setup
  ext4: replace ktype default_attrs with default_groups
  ext4: rename htree_inline_dir_to_tree() to ext4_inlinedir_to_tree()
  ext4: refactor initialize_dirent_tail()
  ext4: rename "dirent_csum" functions to use "dirblock"
  ext4: allow directory holes
  jbd2: drop declaration of journal_sync_buffer()
  ext4: use jbd2_inode dirty range scoping
  jbd2: introduce jbd2_inode dirty range scoping
  mm: add filemap_fdatawait_range_keep_errors()
  ext4: remove redundant assignment to node
  ext4: optimize case-insensitive lookups
  ext4: make __ext4_get_inode_loc plug
  ext4: clean up kerneldoc warnigns when building with W=1
  ext4: only set project inherit bit for directory
  ext4: enforce the immutable flag on open files
  ext4: don't allow any modifications to an immutable file
  jbd2: fix typo in comment of journal_submit_inode_data_buffers
  jbd2: fix some print format mistakes
  ext4: gracefully handle ext4_break_layouts() failure during truncate

5 years agoMerge tag 'afs-next-20190628' of git://git.kernel.org/pub/scm/linux/kernel/git/dhowel...
Linus Torvalds [Thu, 11 Jul 2019 03:55:33 +0000 (20:55 -0700)]
Merge tag 'afs-next-20190628' of git://git./linux/kernel/git/dhowells/linux-fs

Pull afs updates from David Howells:
 "A set of minor changes for AFS:

   - Remove an unnecessary check in afs_unlink()

   - Add a tracepoint for tracking callback management

   - Add a tracepoint for afs_server object usage

   - Use struct_size()

   - Add mappings for AFS UAE abort codes to Linux error codes, using
     symbolic names rather than hex numbers in the .c file"

* tag 'afs-next-20190628' of git://git.kernel.org/pub/scm/linux/kernel/git/dhowells/linux-fs:
  afs: Add support for the UAE error table
  fs/afs: use struct_size() in kzalloc()
  afs: Trace afs_server usage
  afs: Add some callback management tracepoints
  afs: afs_unlink() doesn't need to check dentry->d_inode

5 years agoMerge tag 'fscrypt-for-linus' of git://git.kernel.org/pub/scm/fs/fscrypt/fscrypt
Linus Torvalds [Thu, 11 Jul 2019 03:51:03 +0000 (20:51 -0700)]
Merge tag 'fscrypt-for-linus' of git://git./fs/fscrypt/fscrypt

Pull fscrypt updates from Eric Biggers:

 - Preparations for supporting encryption on ext4 filesystems where the
   filesystem block size is smaller than PAGE_SIZE.

 - Don't allow setting encryption policies on dead directories.

 - Various cleanups.

* tag 'fscrypt-for-linus' of git://git.kernel.org/pub/scm/fs/fscrypt/fscrypt:
  fscrypt: document testing with xfstests
  fscrypt: remove selection of CONFIG_CRYPTO_SHA256
  fscrypt: remove unnecessary includes of ratelimit.h
  fscrypt: don't set policy for a dead directory
  ext4: encrypt only up to last block in ext4_bio_write_page()
  ext4: decrypt only the needed block in __ext4_block_zero_page_range()
  ext4: decrypt only the needed blocks in ext4_block_write_begin()
  ext4: clear BH_Uptodate flag on decryption error
  fscrypt: decrypt only the needed blocks in __fscrypt_decrypt_bio()
  fscrypt: support decrypting multiple filesystem blocks per page
  fscrypt: introduce fscrypt_decrypt_block_inplace()
  fscrypt: handle blocksize < PAGE_SIZE in fscrypt_zeroout_range()
  fscrypt: support encrypting multiple filesystem blocks per page
  fscrypt: introduce fscrypt_encrypt_block_inplace()
  fscrypt: clean up some BUG_ON()s in block encryption/decryption
  fscrypt: rename fscrypt_do_page_crypto() to fscrypt_crypt_block()
  fscrypt: remove the "write" part of struct fscrypt_ctx
  fscrypt: simplify bounce page handling

5 years agoMerge tag 'copy-file-range-fixes-1' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux
Linus Torvalds [Thu, 11 Jul 2019 03:32:37 +0000 (20:32 -0700)]
Merge tag 'copy-file-range-fixes-1' of git://git./fs/xfs/xfs-linux

Pull copy_file_range updates from Darrick Wong:
 "This fixes numerous parameter checking problems and inconsistent
  behaviors in the new(ish) copy_file_range system call.

  Now the system call will actually check its range parameters
  correctly; refuse to copy into files for which the caller does not
  have sufficient privileges; update mtime and strip setuid like file
  writes are supposed to do; and allows copying up to the EOF of the
  source file instead of failing the call like we used to.

  Summary:

   - Create a generic copy_file_range handler and make individual
     filesystems responsible for calling it (i.e. no more assuming that
     do_splice_direct will work or is appropriate)

   - Refactor copy_file_range and remap_range parameter checking where
     they are the same

   - Install missing copy_file_range parameter checking(!)

   - Remove suid/sgid and update mtime like any other file write

   - Change the behavior so that a copy range crossing the source file's
     eof will result in a short copy to the source file's eof instead of
     EINVAL

   - Permit filesystems to decide if they want to handle
     cross-superblock copy_file_range in their local handlers"

* tag 'copy-file-range-fixes-1' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux:
  fuse: copy_file_range needs to strip setuid bits and update timestamps
  vfs: allow copy_file_range to copy across devices
  xfs: use file_modified() helper
  vfs: introduce file_modified() helper
  vfs: add missing checks to copy_file_range
  vfs: remove redundant checks from generic_remap_checks()
  vfs: introduce generic_file_rw_checks()
  vfs: no fallback for ->copy_file_range
  vfs: introduce generic_copy_file_range()

5 years agoMerge tag 'iomap-5.3-merge-1' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux
Linus Torvalds [Thu, 11 Jul 2019 03:29:45 +0000 (20:29 -0700)]
Merge tag 'iomap-5.3-merge-1' of git://git./fs/xfs/xfs-linux

Pull iomap updates from Darrick Wong:
 "There are a few fixes for gfs2 but otherwise it's pretty quiet so far.

   - Only mark inode dirty at the end of writing to a file (instead of
     once for every page written).

   - Fix for an accounting error in the page_done callback"

* tag 'iomap-5.3-merge-1' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux:
  iomap: fix page_done callback for short writes
  fs: fold __generic_write_end back into generic_write_end
  iomap: don't mark the inode dirty in iomap_write_end

5 years agoMerge tag 'for_v5.3-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/jack/linux-fs
Linus Torvalds [Thu, 11 Jul 2019 03:27:07 +0000 (20:27 -0700)]
Merge tag 'for_v5.3-rc1' of git://git./linux/kernel/git/jack/linux-fs

Pull ext2, udf and quota updates from Jan Kara:

 - some ext2 fixes and cleanups

 - a fix of udf bug when extending files

 - a fix of quota Q_XGETQSTAT[V] handling

* tag 'for_v5.3-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/jack/linux-fs:
  udf: Fix incorrect final NOT_ALLOCATED (hole) extent length
  ext2: Use kmemdup rather than duplicating its implementation
  quota: honor quota type in Q_XGETQSTAT[V] calls
  ext2: Always brelse bh on failure in ext2_iget()
  ext2: add missing brelse() in ext2_iget()
  ext2: Fix a typo in ext2_getattr argument
  ext2: fix a typo in comment
  ext2: add missing brelse() in ext2_new_inode()
  ext2: optimize ext2_xattr_get()
  ext2: introduce new helper for xattr entry comparison
  ext2: merge xattr next entry check to ext2_xattr_entry_valid()
  ext2: code cleanup for ext2_preread_inode()
  ext2: code cleanup by using test_opt() and clear_opt()
  doc: ext2: update description of quota options for ext2
  ext2: Strengthen xattr block checks
  ext2: Merge loops in ext2_xattr_set()
  ext2: introduce helper for xattr entry validation
  ext2: introduce helper for xattr header validation
  quota: add dqi_dirty_list description to comment of Dquot List Management

5 years agoMerge tag 'fsnotify_for_v5.3-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Thu, 11 Jul 2019 03:09:17 +0000 (20:09 -0700)]
Merge tag 'fsnotify_for_v5.3-rc1' of git://git./linux/kernel/git/jack/linux-fs

Pull fsnotify updates from Jan Kara:
 "This contains cleanups of the fsnotify name removal hook and also a
  patch to disable fanotify permission events for 'proc' filesystem"

* tag 'fsnotify_for_v5.3-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/jack/linux-fs:
  fsnotify: get rid of fsnotify_nameremove()
  fsnotify: move fsnotify_nameremove() hook out of d_delete()
  configfs: call fsnotify_rmdir() hook
  debugfs: call fsnotify_{unlink,rmdir}() hooks
  debugfs: simplify __debugfs_remove_file()
  devpts: call fsnotify_unlink() hook
  tracefs: call fsnotify_{unlink,rmdir}() hooks
  rpc_pipefs: call fsnotify_{unlink,rmdir}() hooks
  btrfs: call fsnotify_rmdir() hook
  fsnotify: add empty fsnotify_{unlink,rmdir}() hooks
  fanotify: Disallow permission events for proc filesystem

5 years agoMerge tag 'locks-v5.3-1' of git://git.kernel.org/pub/scm/linux/kernel/git/jlayton...
Linus Torvalds [Thu, 11 Jul 2019 02:21:38 +0000 (19:21 -0700)]
Merge tag 'locks-v5.3-1' of git://git./linux/kernel/git/jlayton/linux

Pull file locking updates from Jeff Layton:
 "Just a couple of small lease-related patches this cycle.

  One from Ira to add a new tracepoint that fires during lease conflict
  checks, and another patch from Amir to reduce false positives when
  checking for lease conflicts"

* tag 'locks-v5.3-1' of git://git.kernel.org/pub/scm/linux/kernel/git/jlayton/linux:
  locks: eliminate false positive conflicts for write lease
  locks: Add trace_leases_conflict

5 years agoRevert "Merge tag 'keys-acl-20190703' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Thu, 11 Jul 2019 01:43:43 +0000 (18:43 -0700)]
Revert "Merge tag 'keys-acl-20190703' of git://git./linux/kernel/git/dhowells/linux-fs"

This reverts merge 0f75ef6a9cff49ff612f7ce0578bced9d0b38325 (and thus
effectively commits

   7a1ade847596 ("keys: Provide KEYCTL_GRANT_PERMISSION")
   2e12256b9a76 ("keys: Replace uid/gid/perm permissions checking with an ACL")

that the merge brought in).

It turns out that it breaks booting with an encrypted volume, and Eric
biggers reports that it also breaks the fscrypt tests [1] and loading of
in-kernel X.509 certificates [2].

The root cause of all the breakage is likely the same, but David Howells
is off email so rather than try to work it out it's getting reverted in
order to not impact the rest of the merge window.

 [1] https://lore.kernel.org/lkml/20190710011559.GA7973@sol.localdomain/
 [2] https://lore.kernel.org/lkml/20190710013225.GB7973@sol.localdomain/

Link: https://lore.kernel.org/lkml/CAHk-=wjxoeMJfeBahnWH=9zShKp2bsVy527vo3_y8HfOdhwAAw@mail.gmail.com/
Reported-by: Eric Biggers <ebiggers@kernel.org>
Cc: David Howells <dhowells@redhat.com>
Cc: James Morris <jmorris@namei.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agox86/asm: Move native_write_cr0/4() out of line
Thomas Gleixner [Wed, 10 Jul 2019 19:42:46 +0000 (21:42 +0200)]
x86/asm: Move native_write_cr0/4() out of line

The pinning of sensitive CR0 and CR4 bits caused a boot crash when loading
the kvm_intel module on a kernel compiled with CONFIG_PARAVIRT=n.

The reason is that the static key which controls the pinning is marked RO
after init. The kvm_intel module contains a CR4 write which requires to
update the static key entry list. That obviously does not work when the key
is in a RO section.

With CONFIG_PARAVIRT enabled this does not happen because the CR4 write
uses the paravirt indirection and the actual write function is built in.

As the key is intended to be immutable after init, move
native_write_cr0/4() out of line.

While at it consolidate the update of the cr4 shadow variable and store the
value right away when the pinning is initialized on a booting CPU. No point
in reading it back 20 instructions later. This allows to confine the static
key and the pinning variable to cpu/common and allows to mark them static.

Fixes: 8dbec27a242c ("x86/asm: Pin sensitive CR0 bits")
Fixes: 873d50d58f67 ("x86/asm: Pin sensitive CR4 bits")
Reported-by: Linus Torvalds <torvalds@linux-foundation.org>
Reported-by: Xi Ruoyao <xry111@mengyan1223.wang>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Xi Ruoyao <xry111@mengyan1223.wang>
Acked-by: Kees Cook <keescook@chromium.org>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/alpine.DEB.2.21.1907102140340.1758@nanos.tec.linutronix.de
5 years agotimekeeping/vsyscall: Use __iter_div_u64_rem()
Arnd Bergmann [Wed, 10 Jul 2019 13:01:53 +0000 (15:01 +0200)]
timekeeping/vsyscall: Use __iter_div_u64_rem()

On 32-bit x86 when building with clang-9, the 'division' loop gets turned
back into an inefficient division that causes a link error:

kernel/time/vsyscall.o: In function `update_vsyscall':
vsyscall.c:(.text+0xe3): undefined reference to `__udivdi3'

Use the existing __iter_div_u64_rem() function which is used to address the
same issue in other places.

Fixes: 44f57d788e7d ("timekeeping: Provide a generic update_vsyscall() implementation")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Nathan Chancellor <natechancellor@gmail.com>
Tested-by: Nathan Chancellor <natechancellor@gmail.com>
Link: https://lkml.kernel.org/r/20190710130206.1670830-1-arnd@arndb.de
5 years agox86/pgtable/32: Fix LOWMEM_PAGES constant
Arnd Bergmann [Wed, 10 Jul 2019 13:04:55 +0000 (15:04 +0200)]
x86/pgtable/32: Fix LOWMEM_PAGES constant

clang points out that the computation of LOWMEM_PAGES causes a signed
integer overflow on 32-bit x86:

arch/x86/kernel/head32.c:83:20: error: signed shift result (0x100000000) requires 34 bits to represent, but 'int' only has 32 bits [-Werror,-Wshift-overflow]
                (PAGE_TABLE_SIZE(LOWMEM_PAGES) << PAGE_SHIFT);
                                 ^~~~~~~~~~~~
arch/x86/include/asm/pgtable_32.h:109:27: note: expanded from macro 'LOWMEM_PAGES'
 #define LOWMEM_PAGES ((((2<<31) - __PAGE_OFFSET) >> PAGE_SHIFT))
                         ~^ ~~
arch/x86/include/asm/pgtable_32.h:98:34: note: expanded from macro 'PAGE_TABLE_SIZE'
 #define PAGE_TABLE_SIZE(pages) ((pages) / PTRS_PER_PGD)

Use the _ULL() macro to make it a 64-bit constant.

Fixes: 1e620f9b23e5 ("x86/boot/32: Convert the 32-bit pgtable setup code from assembly to C")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Link: https://lkml.kernel.org/r/20190710130522.1802800-1-arnd@arndb.de
5 years agoRevert "ACPICA: Update table load object initialization"
Rafael J. Wysocki [Wed, 10 Jul 2019 09:52:50 +0000 (11:52 +0200)]
Revert "ACPICA: Update table load object initialization"

Revert commit c522ad0637ca ("ACPICA: Update table load object
initialization") as it causes systems to hang on attempts to load
OEM ACPI tables.

Reported-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
5 years agoclocksource/drivers/npcm: Fix misuse of GENMASK macro
Joe Perches [Wed, 10 Jul 2019 05:04:15 +0000 (22:04 -0700)]
clocksource/drivers/npcm: Fix misuse of GENMASK macro

Arguments are supposed to be ordered high then low.

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Link: https://lkml.kernel.org/r/d6a9d49c9837d38816b71d783f5aed7235e8ca94.1562734889.git.joe@perches.com
5 years agoirqchip/gic-v3-its: Fix misuse of GENMASK macro
Joe Perches [Wed, 10 Jul 2019 05:04:18 +0000 (22:04 -0700)]
irqchip/gic-v3-its: Fix misuse of GENMASK macro

Arguments are supposed to be ordered high then low.

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Marc Zyngier <marc.zyngier@arm.com>
Link: https://lkml.kernel.org/r/ab5deb4fc3cd604cb620054770b7d00016d736bc.1562734889.git.joe@perches.com
5 years agoudf: Fix incorrect final NOT_ALLOCATED (hole) extent length
Steven J. Magnani [Mon, 1 Jul 2019 02:39:35 +0000 (21:39 -0500)]
udf: Fix incorrect final NOT_ALLOCATED (hole) extent length

In some cases, using the 'truncate' command to extend a UDF file results
in a mismatch between the length of the file's extents (specifically, due
to incorrect length of the final NOT_ALLOCATED extent) and the information
(file) length. The discrepancy can prevent other operating systems
(i.e., Windows 10) from opening the file.

Two particular errors have been observed when extending a file:

1. The final extent is larger than it should be, having been rounded up
   to a multiple of the block size.

B. The final extent is not shorter than it should be, due to not having
   been updated when the file's information length was increased.

[JK: simplified udf_do_extend_final_block(), fixed up some types]

Fixes: 2c948b3f86e5 ("udf: Avoid IO in udf_clear_inode")
CC: stable@vger.kernel.org
Signed-off-by: Steven J. Magnani <steve@digidescorp.com>
Link: https://lore.kernel.org/r/1561948775-5878-1-git-send-email-steve@digidescorp.com
Signed-off-by: Jan Kara <jack@suse.cz>
5 years agonet/mlx5e: Return in default case statement in tx_post_resync_params
Nathan Chancellor [Mon, 8 Jul 2019 23:11:55 +0000 (16:11 -0700)]
net/mlx5e: Return in default case statement in tx_post_resync_params

clang warns:

drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_tx.c:251:2:
warning: variable 'rec_seq_sz' is used uninitialized whenever switch
default is taken [-Wsometimes-uninitialized]
        default:
        ^~~~~~~
drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_tx.c:255:46: note:
uninitialized use occurs here
        skip_static_post = !memcmp(rec_seq, &rn_be, rec_seq_sz);
                                                    ^~~~~~~~~~
drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_tx.c:239:16: note:
initialize the variable 'rec_seq_sz' to silence this warning
        u16 rec_seq_sz;
                      ^
                       = 0
1 warning generated.

This case statement was clearly designed to be one that should not be
hit during runtime because of the WARN_ON statement so just return early
to prevent copying uninitialized memory up into rn_be.

Fixes: d2ead1f360e8 ("net/mlx5e: Add kTLS TX HW offload support")
Link: https://github.com/ClangBuiltLinux/linux/issues/590
Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agomlx5: Return -EINVAL when WARN_ON_ONCE triggers in mlx5e_tls_resync().
David S. Miller [Wed, 10 Jul 2019 04:35:08 +0000 (21:35 -0700)]
mlx5: Return -EINVAL when WARN_ON_ONCE triggers in mlx5e_tls_resync().

Return value was changes to 'int' from void but this return statement
was not updated, or it slipped in via a merge.

Fixes: b5d9a834f4fd ("net/tls: don't clear TX resync flag on error")
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonfsd: Make __get_nfsdfs_client() static
YueHaibing [Mon, 8 Jul 2019 12:54:08 +0000 (20:54 +0800)]
nfsd: Make __get_nfsdfs_client() static

Fix sparse warning:

fs/nfsd/nfsctl.c:1221:22: warning:
 symbol '__get_nfsdfs_client' was not declared. Should it be static?

Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
5 years agonfsd: Make two functions static
YueHaibing [Mon, 8 Jul 2019 07:29:33 +0000 (15:29 +0800)]
nfsd: Make two functions static

Fix sparse warnings:

fs/nfsd/nfs4state.c:1908:6: warning: symbol 'drop_client' was not declared. Should it be static?
fs/nfsd/nfs4state.c:2518:6: warning: symbol 'force_expire_client' was not declared. Should it be static?

Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
5 years agonet: dsa: add support for BRIDGE_MROUTER attribute
Vivien Didelot [Tue, 9 Jul 2019 03:31:13 +0000 (23:31 -0400)]
net: dsa: add support for BRIDGE_MROUTER attribute

This patch adds support for enabling or disabling the flooding of
unknown multicast traffic on the CPU ports, depending on the value
of the switchdev SWITCHDEV_ATTR_ID_BRIDGE_MROUTER attribute.

The current behavior is kept unchanged but a user can now prevent
the CPU conduit to be flooded with a lot of unregistered traffic that
the network stack needs to filter in software with e.g.:

    echo 0 > /sys/class/net/br0/multicast_router

Signed-off-by: Vivien Didelot <vivien.didelot@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agopkt_sched: Include const.h
David Ahern [Tue, 9 Jul 2019 21:45:17 +0000 (14:45 -0700)]
pkt_sched: Include const.h

Commit 9903c8dc7342 changed TC_ETF defines to use _BITUL instead of BIT
but did not add the dependecy on linux/const.h. As a consequence,
importing the uapi headers into iproute2 causes builds to fail. Add
the dependency.

Fixes: 9903c8dc7342 ("etf: Don't use BIT() in UAPI headers.")
Cc: Vedang Patel <vedang.patel@intel.com>
Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: netsec: remove static declaration for netsec_set_tx_de()
Ilias Apalodimas [Tue, 9 Jul 2019 21:14:49 +0000 (00:14 +0300)]
net: netsec: remove static declaration for netsec_set_tx_de()

On commit ba2b232108d3 ("net: netsec: add XDP support") a static
declaration for netsec_set_tx_de() was added to make the diff easier
to read.  Now that the patch is merged let's move the functions around
and get rid of that

Signed-off-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: netsec: remove superfluous if statement
Ilias Apalodimas [Tue, 9 Jul 2019 21:14:48 +0000 (00:14 +0300)]
net: netsec: remove superfluous if statement

While freeing tx buffers the memory has to be unmapped if the packet was
an skb or was used for .ndo_xdp_xmit using the same arguments. Get rid
of the unneeded extra 'else if' statement

Signed-off-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMerge branch 'nf-hw-offload'
David S. Miller [Tue, 9 Jul 2019 21:38:51 +0000 (14:38 -0700)]
Merge branch 'nf-hw-offload'

Pablo Neira Ayuso says:

====================
netfilter: add hardware offload infrastructure

This patchset adds support for Netfilter hardware offloads.

This patchset reuses the existing block infrastructure, the
netdev_ops->ndo_setup_tc() interface, TC_SETUP_CLSFLOWER classifier and
the flow rule API.

Patch #1 adds flow_block_cb_setup_simple(), most drivers do the same thing
         to set up flow blocks, to reduce the number of changes, consolidate
         codebase. Use _simple() postfix as requested by Jakub Kicinski.
         This new function resides in net/core/flow_offload.c

Patch #2 renames TC_BLOCK_{UN}BIND to FLOW_BLOCK_{UN}BIND.

Patch #3 renames TCF_BLOCK_BINDER_TYPE_* to FLOW_BLOCK_BINDER_TYPE_*.

Patch #4 adds flow_block_cb_alloc() and flow_block_cb_free() helper
         functions, this is the first patch of the flow block API.

Patch #5 adds the helper to deal with list operations in the flow block API.
         This includes flow_block_cb_lookup(), flow_block_cb_add() and
 flow_block_cb_remove().

Patch #6 adds flow_block_cb_priv(), flow_block_cb_incref() and
         flow_block_cb_decref() which completes the flow block API.

Patch #7 updates the cls_api to use the flow block API from the new
         tcf_block_setup(). This infrastructure transports these objects
         via list (through the tc_block_offload object) back to the core
 for registration.

            CLS_API                           DRIVER
        TC_SETUP_BLOCK    ---------->  setup flow_block_cb object &
                                 it adds object to flow_block_offload->cb_list
                                                |
            CLS_API     <-----------------------'
           registers                     list with flow blocks
         flow_block_cb &                   travels back to
       calls ->reoffload               the core for registration

         drivers allocate and sets up (configure the blocks), then
 registration happens from the core (cls_api and netfilter).

Patch #8 updates drivers to use the flow block API.

Patch #9 removes the tcf block callback API, which is replaced by the
         flow block API.

Patch #10 adds the flow_block_cb_is_busy() helper to check if the block
  is already used by a subsystem. This helper is invoked from
  drivers. Once drivers are updated to support for multiple
  subsystems, they can remove this check.

Patch #11 rename tc structure and definitions for the block bind/unbind
  path.

Patch #12 introduces basic netfilter hardware offload infrastructure
          for the ingress chain. This includes 5-tuple exact matching
          and accept / drop rule actions. Only basechains are supported
          at this stage, no .reoffload callback is implemented either.
          Default policy to "accept" is only supported for now.

        table netdev filter {
                chain ingress {
                        type filter hook ingress device eth0 priority 0; flags offload;

                        ip daddr 192.168.0.10 tcp dport 22 drop
                }
        }

This patchset reuses the existing tcf block callback API and it places it
in the flow block callback API in net/core/flow_offload.c.

This series aims to address Jakub and Jiri's feedback, please see specific
patches in this batch for changelog in this v4.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonetfilter: nf_tables: add hardware offload support
Pablo Neira Ayuso [Tue, 9 Jul 2019 21:00:43 +0000 (23:00 +0200)]
netfilter: nf_tables: add hardware offload support

This patch adds hardware offload support for nftables through the
existing netdev_ops->ndo_setup_tc() interface, the TC_SETUP_CLSFLOWER
classifier and the flow rule API. This hardware offload support is
available for the NFPROTO_NETDEV family and the ingress hook.

Each nftables expression has a new ->offload interface, that is used to
populate the flow rule object that is attached to the transaction
object.

There is a new per-table NFT_TABLE_F_HW flag, that is set on to offload
an entire table, including all of its chains.

This patch supports for basic metadata (layer 3 and 4 protocol numbers),
5-tuple payload matching and the accept/drop actions; this also includes
basechain hardware offload only.

Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: flow_offload: rename tc_cls_flower_offload to flow_cls_offload
Pablo Neira Ayuso [Tue, 9 Jul 2019 20:55:49 +0000 (22:55 +0200)]
net: flow_offload: rename tc_cls_flower_offload to flow_cls_offload

And any other existing fields in this structure that refer to tc.
Specifically:

* tc_cls_flower_offload_flow_rule() to flow_cls_offload_flow_rule().
* TC_CLSFLOWER_* to FLOW_CLS_*.
* tc_cls_common_offload to tc_cls_common_offload.

Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: flow_offload: add flow_block_cb_is_busy() and use it
Pablo Neira Ayuso [Tue, 9 Jul 2019 20:55:48 +0000 (22:55 +0200)]
net: flow_offload: add flow_block_cb_is_busy() and use it

This patch adds a function to check if flow block callback is already in
use.  Call this new function from flow_block_cb_setup_simple() and from
drivers.

Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: sched: remove tcf block API
Pablo Neira Ayuso [Tue, 9 Jul 2019 20:55:47 +0000 (22:55 +0200)]
net: sched: remove tcf block API

Unused, now replaced by flow block API.

Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agodrivers: net: use flow block API
Pablo Neira Ayuso [Tue, 9 Jul 2019 20:55:46 +0000 (22:55 +0200)]
drivers: net: use flow block API

This patch updates flow_block_cb_setup_simple() to use the flow block API.
Several drivers are also adjusted to use it.

This patch introduces the per-driver list of flow blocks to account for
blocks that are already in use.

Remove tc_block_offload alias.

Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: sched: use flow block API
Pablo Neira Ayuso [Tue, 9 Jul 2019 20:55:45 +0000 (22:55 +0200)]
net: sched: use flow block API

This patch adds tcf_block_setup() which uses the flow block API.

This infrastructure takes the flow block callbacks coming from the
driver and register/unregister to/from the cls_api core.

Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: flow_offload: add flow_block_cb_{priv, incref, decref}()
Pablo Neira Ayuso [Tue, 9 Jul 2019 20:55:44 +0000 (22:55 +0200)]
net: flow_offload: add flow_block_cb_{priv, incref, decref}()

This patch completes the flow block API to introduce:

* flow_block_cb_priv() to access callback private data.
* flow_block_cb_incref() to bump reference counter on this flow block.
* flow_block_cb_decref() to decrement the reference counter.

These functions are taken from the existing tcf_block_cb_priv(),
tcf_block_cb_incref() and tcf_block_cb_decref().

Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: flow_offload: add list handling functions
Pablo Neira Ayuso [Tue, 9 Jul 2019 20:55:43 +0000 (22:55 +0200)]
net: flow_offload: add list handling functions

This patch adds the list handling functions for the flow block API:

* flow_block_cb_lookup() allows drivers to look up for existing flow blocks.
* flow_block_cb_add() adds a flow block to the per driver list to be registered
  by the core.
* flow_block_cb_remove() to remove a flow block from the list of existing
  flow blocks per driver and to request the core to unregister this.

The flow block API also annotates the netns this flow block belongs to.

Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: flow_offload: add flow_block_cb_alloc() and flow_block_cb_free()
Pablo Neira Ayuso [Tue, 9 Jul 2019 20:55:42 +0000 (22:55 +0200)]
net: flow_offload: add flow_block_cb_alloc() and flow_block_cb_free()

Add a new helper function to allocate flow_block_cb objects.

Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: flow_offload: rename TCF_BLOCK_BINDER_TYPE_* to FLOW_BLOCK_BINDER_TYPE_*
Pablo Neira Ayuso [Tue, 9 Jul 2019 20:55:41 +0000 (22:55 +0200)]
net: flow_offload: rename TCF_BLOCK_BINDER_TYPE_* to FLOW_BLOCK_BINDER_TYPE_*

Rename from TCF_BLOCK_BINDER_TYPE_* to FLOW_BLOCK_BINDER_TYPE_* and
remove temporary tcf_block_binder_type alias.

Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: flow_offload: rename TC_BLOCK_{UN}BIND to FLOW_BLOCK_{UN}BIND
Pablo Neira Ayuso [Tue, 9 Jul 2019 20:55:40 +0000 (22:55 +0200)]
net: flow_offload: rename TC_BLOCK_{UN}BIND to FLOW_BLOCK_{UN}BIND

Rename from TC_BLOCK_{UN}BIND to FLOW_BLOCK_{UN}BIND and remove
temporary tc_block_command alias.

Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: flow_offload: add flow_block_cb_setup_simple()
Pablo Neira Ayuso [Tue, 9 Jul 2019 20:55:39 +0000 (22:55 +0200)]
net: flow_offload: add flow_block_cb_setup_simple()

Most drivers do the same thing to set up the flow block callbacks, this
patch adds a helper function to do this.

This preparation patch reduces the number of changes to adapt the
existing drivers to use the flow block callback API.

This new helper function takes a flow block list per-driver, which is
set to NULL until this driver list is used.

This patch also introduces the flow_block_command and
flow_block_binder_type enumerations, which are renamed to use
FLOW_BLOCK_* in follow up patches.

There are three definitions (aliases) in order to reduce the number of
updates in this patch, which go away once drivers are fully adapted to
use this flow block API.

Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Reviewed-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMerge branch 'net-hisilicon-Add-support-for-HI13X1-to-hip04_eth'
David S. Miller [Tue, 9 Jul 2019 21:29:27 +0000 (14:29 -0700)]
Merge branch 'net-hisilicon-Add-support-for-HI13X1-to-hip04_eth'

Jiangfeng Xiao says:

====================
net: hisilicon: Add support for HI13X1 to hip04_eth

The main purpose of this patch series is to extend the
hip04_eth driver to support HI13X1_GMAC.

The offset and bitmap of some registers of HI13X1_GMAC
are different from hip04_eth common soc. In addition,
the definition of send descriptor and parsing descriptor
are different from hip04_eth common soc. So the macro
of the register offset is redefined to adapt the HI13X1_GMAC.

Clean up the sparse warning by the way.

Change since v1:
* Add a cover letter.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: hisilicon: Add an tx_desc to adapt HI13X1_GMAC
Jiangfeng Xiao [Tue, 9 Jul 2019 03:31:11 +0000 (11:31 +0800)]
net: hisilicon: Add an tx_desc to adapt HI13X1_GMAC

HI13X1 changed the offsets and bitmaps for tx_desc
registers in the same peripheral device on different
models of the hip04_eth.

Signed-off-by: Jiangfeng Xiao <xiaojiangfeng@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: hisilicon: Add an rx_desc to adapt HI13X1_GMAC
Jiangfeng Xiao [Tue, 9 Jul 2019 03:31:10 +0000 (11:31 +0800)]
net: hisilicon: Add an rx_desc to adapt HI13X1_GMAC

HI13X1 changed the offsets and bitmaps for rx_desc
registers in the same peripheral device on different
models of the hip04_eth.

Signed-off-by: Jiangfeng Xiao <xiaojiangfeng@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: hisilicon: Offset buf address to adapt HI13X1_GMAC
Jiangfeng Xiao [Tue, 9 Jul 2019 03:31:09 +0000 (11:31 +0800)]
net: hisilicon: Offset buf address to adapt HI13X1_GMAC

The buf unit size of HI13X1_GMAC is cache_line_size,
which is 64, so the address we write to the buf register
needs to be shifted right by 6 bits.

The 31st bit of the PPE_CFG_CPU_ADD_ADDR register
of HI13X1_GMAC indicates whether to release the buffer
of the message, and the low indicates that it is valid.

Signed-off-by: Jiangfeng Xiao <xiaojiangfeng@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: hisilicon: Add group field to adapt HI13X1_GMAC
Jiangfeng Xiao [Tue, 9 Jul 2019 03:31:08 +0000 (11:31 +0800)]
net: hisilicon: Add group field to adapt HI13X1_GMAC

In general, group is the same as the port, but some
boards specify a special group for better load
balancing of each processing unit.

Signed-off-by: Jiangfeng Xiao <xiaojiangfeng@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: hisilicon: dt-bindings: Add an field of port-handle
Jiangfeng Xiao [Tue, 9 Jul 2019 03:31:07 +0000 (11:31 +0800)]
net: hisilicon: dt-bindings: Add an field of port-handle

In general, group is the same as the port, but some
boards specify a special group for better load
balancing of each processing unit.

Signed-off-by: Jiangfeng Xiao <xiaojiangfeng@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: hisilicon: HI13X1_GMAX need dreq reset at first
Jiangfeng Xiao [Tue, 9 Jul 2019 03:31:06 +0000 (11:31 +0800)]
net: hisilicon: HI13X1_GMAX need dreq reset at first

HI13X1_GMAC delete request for soft reset at first,
otherwise, the subsequent initialization will not
take effect.

Signed-off-by: Jiangfeng Xiao <xiaojiangfeng@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: hisilicon: HI13X1_GMAX skip write LOCAL_PAGE_REG
Jiangfeng Xiao [Tue, 9 Jul 2019 03:31:05 +0000 (11:31 +0800)]
net: hisilicon: HI13X1_GMAX skip write LOCAL_PAGE_REG

HI13X1_GMAC changed the offsets and bitmaps for
GE_TX_LOCAL_PAGE_REG registers in the same peripheral
device on different models of the hip04_eth. With the
default configuration, HI13X1_GMAC can also work without
any writes to the GE_TX_LOCAL_PAGE_REG register.

Signed-off-by: Jiangfeng Xiao <xiaojiangfeng@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: hisilicon: Cleanup for cast to restricted __be32
Jiangfeng Xiao [Tue, 9 Jul 2019 03:31:04 +0000 (11:31 +0800)]
net: hisilicon: Cleanup for cast to restricted __be32

This patch fixes the following warning from sparse:
hip04_eth.c:533:23: warning: cast to restricted __be16
hip04_eth.c:533:23: warning: cast to restricted __be16
hip04_eth.c:533:23: warning: cast to restricted __be16
hip04_eth.c:533:23: warning: cast to restricted __be16
hip04_eth.c:534:23: warning: cast to restricted __be32
hip04_eth.c:534:23: warning: cast to restricted __be32
hip04_eth.c:534:23: warning: cast to restricted __be32
hip04_eth.c:534:23: warning: cast to restricted __be32
hip04_eth.c:534:23: warning: cast to restricted __be32
hip04_eth.c:534:23: warning: cast to restricted __be32

Signed-off-by: Jiangfeng Xiao <xiaojiangfeng@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: hisilicon: Cleanup for got restricted __be32
Jiangfeng Xiao [Tue, 9 Jul 2019 03:31:03 +0000 (11:31 +0800)]
net: hisilicon: Cleanup for got restricted __be32

This patch fixes the following warning from sparse:
hip04_eth.c:468:25: warning: incorrect type in assignment
hip04_eth.c:468:25:    expected unsigned int [usertype] send_addr
hip04_eth.c:468:25:    got restricted __be32 [usertype]
hip04_eth.c:469:25: warning: incorrect type in assignment
hip04_eth.c:469:25:    expected unsigned int [usertype] send_size
hip04_eth.c:469:25:    got restricted __be32 [usertype]
hip04_eth.c:470:19: warning: incorrect type in assignment
hip04_eth.c:470:19:    expected unsigned int [usertype] cfg
hip04_eth.c:470:19:    got restricted __be32 [usertype]
hip04_eth.c:472:23: warning: incorrect type in assignment
hip04_eth.c:472:23:    expected unsigned int [usertype] wb_addr
hip04_eth.c:472:23:    got restricted __be32 [usertype]

Signed-off-by: Jiangfeng Xiao <xiaojiangfeng@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: hisilicon: Add support for HI13X1 to hip04_eth
Jiangfeng Xiao [Tue, 9 Jul 2019 03:31:02 +0000 (11:31 +0800)]
net: hisilicon: Add support for HI13X1 to hip04_eth

Extend the hip04_eth driver to support HI13X1_GMAC.
Enable it with CONFIG_HI13X1_GMAC option.

Signed-off-by: Jiangfeng Xiao <xiaojiangfeng@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMerge branch 'stmmac-hash-table'
David S. Miller [Tue, 9 Jul 2019 21:17:59 +0000 (14:17 -0700)]
Merge branch 'stmmac-hash-table'

Biao Huang says:

====================
stmmac: fix out-of-boundary issue and add taller hash table support

Fix mac address out-of-boundary issue in net-next tree.
and resend the patch which was discussed in
https://lore.kernel.org/patchwork/patch/1082117
but with no further progress.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: stmmac: add support for hash table size 128/256 in dwmac4
Biao Huang [Tue, 9 Jul 2019 02:36:23 +0000 (10:36 +0800)]
net: stmmac: add support for hash table size 128/256 in dwmac4

1. get hash table size in hw feature reigster, and add support
for taller hash table(128/256) in dwmac4.
2. only clear GMAC_PACKET_FILTER bits used in this function,
to avoid side effect to functions of other bits.

stmmac selftests output log with flow control on:
ethtool -t eth0
The test result is PASS
The test extra info:
 1. MAC Loopback                 0
 2. PHY Loopback                 -95
 3. MMC Counters                 0
 4. EEE                          -95
 5. Hash Filter MC               0
 6. Perfect Filter UC            0
 7. MC Filter                    0
 8. UC Filter                    0
 9. Flow Control                 0

Signed-off-by: Biao Huang <biao.huang@mediatek.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: stmmac: dwmac4: mac address array boudary violation issue
Biao Huang [Tue, 9 Jul 2019 02:36:22 +0000 (10:36 +0800)]
net: stmmac: dwmac4: mac address array boudary violation issue

The mac address array size is GMAC_MAX_PERFECT_ADDRESSES,
so the 'reg' should be less than it, or will affect other registers.

Signed-off-by: Biao Huang <biao.huang@mediatek.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMerge branch 'tc-testing-Add-plugin-for-simple-traffic-generation'
David S. Miller [Tue, 9 Jul 2019 21:07:20 +0000 (14:07 -0700)]
Merge branch 'tc-testing-Add-plugin-for-simple-traffic-generation'

Lucas Bates says:

====================
tc-testing: Add plugin for simple traffic generation

This series supersedes the previous submission that included a patch for test
case verification using JSON output.  It adds a new tdc plugin, scapyPlugin, as
a way to send traffic to test tc filters and actions.

The first patch makes a change to the TdcPlugin module that will allow tdc
plugins to examine the test case currently being executed, so plugins can
play a more active role in testing by accepting information or commands from
the test case.  This is required for scapyPlugin to work.

The second patch adds scapyPlugin itself, and an example test case file to
demonstrate how the scapy block works in the test cases.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agotc-testing: introduce scapyPlugin for basic traffic
Lucas Bates [Tue, 9 Jul 2019 01:34:27 +0000 (21:34 -0400)]
tc-testing: introduce scapyPlugin for basic traffic

The scapyPlugin allows for simple traffic generation in tdc to
test various tc features. It was tested with scapy v2.4.2, but
should work with any successive version.

In order to use the plugin's functionality, scapy must be
installed. This can be done with:
   pip3 install scapy

or to install 2.4.2:
   pip3 install scapy==2.4.2

If the plugin is unable to import the scapy module, it will
terminate the tdc run.

The plugin makes use of a new key in the test case data, 'scapy'.
This block contains three other elements: 'iface', 'count', and
'packet':

        "scapy": {
            "iface": "$DEV0",
            "count": 1,
            "packet": "Ether(type=0x800)/IP(src='16.61.16.61')/ICMP()"
        },

* iface is the name of the device on the host machine from which
  the packet(s) will be sent. Values contained within tdc_config.py's
  NAMES dict can be used here - this is useful if paired with
  nsPlugin
* count is the number of copies of this packet to be sent
* packet is a string detailing the different layers of the packet
  to be sent. If a property isn't explicitly set, scapy will set
  default values for you.

Layers in the packet info are separated by slashes. For info about
common TCP and IP properties, see:
https://blogs.sans.org/pen-testing/files/2016/04/ScapyCheatSheet_v0.2.pdf

Caution is advised when running tests using the scapy functionality,
since the plugin blindly sends the packet as defined in the test case
data.

See creating-testcases/scapy-example.json for sample test cases;
the first test is intended to pass while the second is intended to
fail.

Signed-off-by: Lucas Bates <lucasb@mojatatu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agotc-testing: Allow tdc plugins to see test case data
Lucas Bates [Tue, 9 Jul 2019 01:34:26 +0000 (21:34 -0400)]
tc-testing: Allow tdc plugins to see test case data

Instead of only passing the test case name and ID, pass the
entire current test case down to the plugins. This change
allows plugins to start accepting commands and directives
from the test cases themselves, for greater flexibility
in testing.

Signed-off-by: Lucas Bates <lucasb@mojatatu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agox86/alternatives: Fix int3_emulate_call() selftest stack corruption
Peter Zijlstra [Mon, 8 Jul 2019 20:55:30 +0000 (15:55 -0500)]
x86/alternatives: Fix int3_emulate_call() selftest stack corruption

KASAN shows the following splat during boot:

  BUG: KASAN: unknown-crash in unwind_next_frame+0x3f6/0x490
  Read of size 8 at addr ffffffff84007db0 by task swapper/0

  CPU: 0 PID: 0 Comm: swapper Tainted: G                T 5.2.0-rc6-00013-g7457c0d #1
  Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1 04/01/2014
  Call Trace:
   dump_stack+0x19/0x1b
   print_address_description+0x1b0/0x2b2
   __kasan_report+0x10f/0x171
   kasan_report+0x12/0x1c
   __asan_load8+0x54/0x81
   unwind_next_frame+0x3f6/0x490
   unwind_next_frame+0x1b/0x23
   arch_stack_walk+0x68/0xa5
   stack_trace_save+0x7b/0xa0
   save_trace+0x3c/0x93
   mark_lock+0x1ef/0x9b1
   lock_acquire+0x122/0x221
   __mutex_lock+0xb6/0x731
   mutex_lock_nested+0x16/0x18
   _vm_unmap_aliases+0x141/0x183
   vm_unmap_aliases+0x14/0x16
   change_page_attr_set_clr+0x15e/0x2f2
   set_memory_4k+0x2a/0x2c
   check_bugs+0x11fd/0x1298
   start_kernel+0x793/0x7eb
   x86_64_start_reservations+0x55/0x76
   x86_64_start_kernel+0x87/0xaa
   secondary_startup_64+0xa4/0xb0

  Memory state around the buggy address:
   ffffffff84007c80: 00 00 00 00 00 00 00 00 00 00 00 00 00 f1 f1 f1
   ffffffff84007d00: f1 00 00 00 00 00 00 00 00 00 f2 f2 f2 f3 f3 f3
  >ffffffff84007d80: f3 79 be 52 49 79 be 00 00 00 00 00 00 00 00 f1

It turns out that int3_selftest() is corrupting the stack.  The problem is
that the KASAN-ified version of int3_magic() is much less trivial than the
C code appears.  It clobbers several unexpected registers.  So when the
selftest's INT3 is converted to an emulated call to int3_magic(), the
registers are clobbered and Bad Things happen when the function returns.

Fix this by converting int3_magic() to the trivial ASM function it should
be, avoiding all calling convention issues. Also add ASM_CALL_CONSTRAINT to
the INT3 ASM, since it contains a 'CALL'.

[peterz: cribbed changelog from josh]

Fixes: 7457c0da024b ("x86/alternatives: Add int3_emulate_call() selftest")
Reported-by: kernel test robot <rong.a.chen@intel.com>
Debugged-by: Josh Poimboeuf <jpoimboe@redhat.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Josh Poimboeuf <jpoimboe@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Andy Lutomirski <luto@kernel.org>
Link: https://lkml.kernel.org/r/20190709125744.GB3402@hirez.programming.kicks-ass.net
5 years agonet: dsa: vsc73xx: fix NET_DSA and OF dependencies
Arnd Bergmann [Tue, 9 Jul 2019 18:55:55 +0000 (20:55 +0200)]
net: dsa: vsc73xx: fix NET_DSA and OF dependencies

The restructuring of the driver got the dependencies wrong: without
CONFIG_NET_DSA we get this build failure:

WARNING: unmet direct dependencies detected for NET_DSA_VITESSE_VSC73XX
  Depends on [n]: NETDEVICES [=y] && HAVE_NET_DSA [=y] && OF [=y] && NET_DSA [=n]
  Selected by [m]:
  - NET_DSA_VITESSE_VSC73XX_PLATFORM [=m] && NETDEVICES [=y] && HAVE_NET_DSA [=y] && HAS_IOMEM [=y]

ERROR: "dsa_unregister_switch" [drivers/net/dsa/vitesse-vsc73xx-core.ko] undefined!
ERROR: "dsa_switch_alloc" [drivers/net/dsa/vitesse-vsc73xx-core.ko] undefined!
ERROR: "dsa_register_switch" [drivers/net/dsa/vitesse-vsc73xx-core.ko] undefined!

Add the appropriate dependencies.

Fixes: 95711cd5f0b4 ("net: dsa: vsc73xx: Split vsc73xx driver")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMerge branch 'Armada-8040-SoC-in-orion-mdio-hang'
David S. Miller [Tue, 9 Jul 2019 20:03:04 +0000 (13:03 -0700)]
Merge branch 'Armada-8040-SoC-in-orion-mdio-hang'

Josua Mayer says:

====================
Fix hang of Armada 8040 SoC in orion-mdio

With a modular kernel as configured by Debian a hang was observed with
the Armada 8040 SoC in the Clearfog GT and Macchiatobin boards.

The 8040 SoC actually requires four clocks to be enabled for the mdio
interface to function. All 4 clocks are already specified in
armada-cp110.dtsi. It has however been missed that the orion-mdio driver
only supports enabling up to three clocks.

This patch-set allows the orion-mdio driver to handle four clocks and
adds a warning when more clocks are specified to prevent this particular
oversight in the future.

Changes since v1:
- fixed condition for priting the warning (Andrew Lunn)
- rephrased commit description for deferred probing (Andrew Lunn)
- fixed compiler warnings (kbuild test robot)
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: mvmdio: defer probe of orion-mdio if a clock is not ready
Josua Mayer [Tue, 9 Jul 2019 13:01:01 +0000 (15:01 +0200)]
net: mvmdio: defer probe of orion-mdio if a clock is not ready

Defer probing of the orion-mdio interface when getting a clock returns
EPROBE_DEFER. This avoids locking up the Armada 8k SoC when mdio is used
before all clocks have been enabled.

Signed-off-by: Josua Mayer <josua@solid-run.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: mvmdio: print warning when orion-mdio has too many clocks
Josua Mayer [Tue, 9 Jul 2019 13:01:00 +0000 (15:01 +0200)]
net: mvmdio: print warning when orion-mdio has too many clocks

Print a warning when device tree specifies more than the maximum of four
clocks supported by orion-mdio. Because reading from mdio can lock up
the Armada 8k when a required clock is not initialized, it is important
to notify the user when a specified clock is ignored.

Signed-off-by: Josua Mayer <josua@solid-run.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: mvmdio: allow up to four clocks to be specified for orion-mdio
Josua Mayer [Tue, 9 Jul 2019 13:00:59 +0000 (15:00 +0200)]
net: mvmdio: allow up to four clocks to be specified for orion-mdio

Allow up to four clocks to be specified and enabled for the orion-mdio
interface, which are required by the Armada 8k and defined in
armada-cp110.dtsi.

Fixes a hang in probing the mvmdio driver that was encountered on the
Clearfog GT 8K with all drivers built as modules, but also affects other
boards such as the MacchiatoBIN.

Cc: stable@vger.kernel.org
Fixes: 96cb43423822 ("net: mvmdio: allow up to three clocks to be specified for orion-mdio")
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Josua Mayer <josua@solid-run.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agodt-bindings: allow up to four clocks for orion-mdio
Josua Mayer [Tue, 9 Jul 2019 13:00:58 +0000 (15:00 +0200)]
dt-bindings: allow up to four clocks for orion-mdio

Armada 8040 needs four clocks to be enabled for MDIO accesses to work.
Update the binding to allow the extra clock to be specified.

Cc: stable@vger.kernel.org
Fixes: 6d6a331f44a1 ("dt-bindings: allow up to three clocks for orion-mdio")
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Josua Mayer <josua@solid-run.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: netsec: start using buffers if page_pool registration succeeded
Ilias Apalodimas [Tue, 9 Jul 2019 12:35:53 +0000 (15:35 +0300)]
net: netsec: start using buffers if page_pool registration succeeded

The current driver starts using page_pool buffers before calling
xdp_rxq_info_reg_mem_model(). Start using the buffers after the
registration succeeded, so we won't have to call
page_pool_request_shutdown() in case of failure

Fixes: 5c67bf0ec4d0 ("net: netsec: Use page_pool API")
Signed-off-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>