Pull zstd support from Chris Mason:
"Nick Terrell's patch series to add zstd support to the kernel has been
floating around for a while. After talking with Dave Sterba, Herbert
and Phillip, we decided to send the whole thing in as one pull
request.
zstd is a big win in speed over zlib and in compression ratio over
lzo, and the compression team here at FB has gotten great results
using it in production. Nick will continue to update the kernel side
with new improvements from the open source zstd userland code.
Nick has a number of benchmarks for the main zstd code in his lib/zstd
commit:
I ran the benchmarks on a Ubuntu 14.04 VM with 2 cores and 4 GiB
of RAM. The VM is running on a MacBook Pro with a 3.1 GHz Intel
Core i7 processor, 16 GB of RAM, and a SSD. I benchmarked using
`silesia.tar` [3], which is 211,988,480 B large. Run the following
commands for the benchmark:
sudo modprobe zstd_compress_test
sudo mknod zstd_compress_test c 245 0
sudo cp silesia.tar zstd_compress_test
The time is reported by the time of the userland `cp`.
The MB/s is computed with
1,536,217,008 B / time(buffer size, hash)
which includes the time to copy from userland.
The Adjusted MB/s is computed with
1,536,217,088 B / (time(buffer size, hash) - time(buffer size, none)).
The memory reported is the amount of memory the compressor
requests.
| Method | Size (B) | Time (s) | Ratio | MB/s | Adj MB/s | Mem (MB) |
|----------|----------|----------|-------|---------|----------|----------|
| none | 11988480 | 0.100 | 1 | 2119.88 | - | - |
| zstd -1 | 73645762 | 1.044 | 2.878 | 203.05 | 224.56 | 1.23 |
| zstd -3 | 66988878 | 1.761 | 3.165 | 120.38 | 127.63 | 2.47 |
| zstd -5 | 65001259 | 2.563 | 3.261 | 82.71 | 86.07 | 2.86 |
| zstd -10 | 60165346 | 13.242 | 3.523 | 16.01 | 16.13 | 13.22 |
| zstd -15 | 58009756 | 47.601 | 3.654 | 4.45 | 4.46 | 21.61 |
| zstd -19 | 54014593 | 102.835 | 3.925 | 2.06 | 2.06 | 60.15 |
| zlib -1 | 77260026 | 2.895 | 2.744 | 73.23 | 75.85 | 0.27 |
| zlib -3 | 72972206 | 4.116 | 2.905 | 51.50 | 52.79 | 0.27 |
| zlib -6 | 68190360 | 9.633 | 3.109 | 22.01 | 22.24 | 0.27 |
| zlib -9 | 67613382 | 22.554 | 3.135 | 9.40 | 9.44 | 0.27 |
I benchmarked zstd decompression using the same method on the same
machine. The benchmark file is located in the upstream zstd repo
under `contrib/linux-kernel/zstd_decompress_test.c` [4]. The
memory reported is the amount of memory required to decompress
data compressed with the given compression level. If you know the
maximum size of your input, you can reduce the memory usage of
decompression irrespective of the compression level.
| Method | Time (s) | MB/s | Adjusted MB/s | Memory (MB) |
|----------|----------|---------|---------------|-------------|
| none | 0.025 | 8479.54 | - | - |
| zstd -1 | 0.358 | 592.15 | 636.60 | 0.84 |
| zstd -3 | 0.396 | 535.32 | 571.40 | 1.46 |
| zstd -5 | 0.396 | 535.32 | 571.40 | 1.46 |
| zstd -10 | 0.374 | 566.81 | 607.42 | 2.51 |
| zstd -15 | 0.379 | 559.34 | 598.84 | 4.61 |
| zstd -19 | 0.412 | 514.54 | 547.77 | 8.80 |
| zlib -1 | 0.940 | 225.52 | 231.68 | 0.04 |
| zlib -3 | 0.883 | 240.08 | 247.07 | 0.04 |
| zlib -6 | 0.844 | 251.17 | 258.84 | 0.04 |
| zlib -9 | 0.837 | 253.27 | 287.64 | 0.04 |
I ran a long series of tests and benchmarks on the btrfs side and the
gains are very similar to the core benchmarks Nick ran"
* 'zstd-minimal' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs:
squashfs: Add zstd support
btrfs: Add zstd support
lib: Add zstd modules
lib: Add xxhash module
GFP_TEMPORARY was introduced by commit e12ba74d8f ("Group short-lived
and reclaimable kernel allocations") along with __GFP_RECLAIMABLE. It's
primary motivation was to allow users to tell that an allocation is
short lived and so the allocator can try to place such allocations close
together and prevent long term fragmentation. As much as this sounds
like a reasonable semantic it becomes much less clear when to use the
highlevel GFP_TEMPORARY allocation flag. How long is temporary? Can the
context holding that memory sleep? Can it take locks? It seems there is
no good answer for those questions.
The current implementation of GFP_TEMPORARY is basically GFP_KERNEL |
__GFP_RECLAIMABLE which in itself is tricky because basically none of
the existing caller provide a way to reclaim the allocated memory. So
this is rather misleading and hard to evaluate for any benefits.
I have checked some random users and none of them has added the flag
with a specific justification. I suspect most of them just copied from
other existing users and others just thought it might be a good idea to
use without any measuring. This suggests that GFP_TEMPORARY just
motivates for cargo cult usage without any reasoning.
I believe that our gfp flags are quite complex already and especially
those with highlevel semantic should be clearly defined to prevent from
confusion and abuse. Therefore I propose dropping GFP_TEMPORARY and
replace all existing users to simply use GFP_KERNEL. Please note that
SLAB users with shrinkers will still get __GFP_RECLAIMABLE heuristic and
so they will be placed properly for memory fragmentation prevention.
I can see reasons we might want some gfp flag to reflect shorterm
allocations but I propose starting from a clear semantic definition and
only then add users with proper justification.
This was been brought up before LSF this year by Matthew [1] and it
turned out that GFP_TEMPORARY really doesn't have a clear semantic. It
seems to be a heuristic without any measured advantage for most (if not
all) its current users. The follow up discussion has revealed that
opinions on what might be temporary allocation differ a lot between
developers. So rather than trying to tweak existing users into a
semantic which they haven't expected I propose to simply remove the flag
and start from scratch if we really need a semantic for short term
allocations.
[1] http://lkml.kernel.org/r/20170118054945.GD18349@bombadil.infradead.org
[akpm@linux-foundation.org: fix typo]
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: drm/i915: fix up]
Link: http://lkml.kernel.org/r/20170816144703.378d4f4d@canb.auug.org.au
Link: http://lkml.kernel.org/r/20170728091904.14627-1-mhocko@kernel.org
Signed-off-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Acked-by: Mel Gorman <mgorman@suse.de>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Neil Brown <neilb@suse.de>
Cc: "Theodore Ts'o" <tytso@mit.edu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
IDR only supports non-negative IDs. There used to be a 'WARN_ON_ONCE(id <
0)' in idr_replace(), but it was intentionally removed by commit
2e1c9b2867 ("idr: remove WARN_ON_ONCE() on negative IDs").
Then it was added back by commit 0a835c4f09 ("Reimplement IDR and IDA
using the radix tree"). However it seems that adding it back was a
mistake, given that some users such as drm_gem_handle_delete()
(DRM_IOCTL_GEM_CLOSE) pass in a value from userspace to idr_replace(),
allowing the WARN_ON_ONCE to be triggered. drm_gem_handle_delete()
actually just wants idr_replace() to return an error code if the ID is
not allocated, including in the case where the ID is invalid (negative).
So once again remove the bogus WARN_ON_ONCE().
This bug was found by syzkaller, which encountered the following
warning:
WARNING: CPU: 3 PID: 3008 at lib/idr.c:157 idr_replace+0x1d8/0x240 lib/idr.c:157
Kernel panic - not syncing: panic_on_warn set ...
CPU: 3 PID: 3008 Comm: syzkaller218828 Not tainted 4.13.0-rc4-next-20170811 #2
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Bochs 01/01/2011
Call Trace:
fixup_bug+0x40/0x90 arch/x86/kernel/traps.c:190
do_trap_no_signal arch/x86/kernel/traps.c:224 [inline]
do_trap+0x260/0x390 arch/x86/kernel/traps.c:273
do_error_trap+0x120/0x390 arch/x86/kernel/traps.c:310
do_invalid_op+0x1b/0x20 arch/x86/kernel/traps.c:323
invalid_op+0x1e/0x30 arch/x86/entry/entry_64.S:930
RIP: 0010:idr_replace+0x1d8/0x240 lib/idr.c:157
RSP: 0018:ffff8800394bf9f8 EFLAGS: 00010297
RAX: ffff88003c6c60c0 RBX: 1ffff10007297f43 RCX: 0000000000000000
RDX: 0000000000000000 RSI: 0000000000000000 RDI: ffff8800394bfa78
RBP: ffff8800394bfae0 R08: ffffffff82856487 R09: 0000000000000000
R10: ffff8800394bf9a8 R11: ffff88006c8bae28 R12: ffffffffffffffff
R13: ffff8800394bfab8 R14: dffffc0000000000 R15: ffff8800394bfbc8
drm_gem_handle_delete+0x33/0xa0 drivers/gpu/drm/drm_gem.c:297
drm_gem_close_ioctl+0xa1/0xe0 drivers/gpu/drm/drm_gem.c:671
drm_ioctl_kernel+0x1e7/0x2e0 drivers/gpu/drm/drm_ioctl.c:729
drm_ioctl+0x72e/0xa50 drivers/gpu/drm/drm_ioctl.c:825
vfs_ioctl fs/ioctl.c:45 [inline]
do_vfs_ioctl+0x1b1/0x1520 fs/ioctl.c:685
SYSC_ioctl fs/ioctl.c:700 [inline]
SyS_ioctl+0x8f/0xc0 fs/ioctl.c:691
entry_SYSCALL_64_fastpath+0x1f/0xbe
Here is a C reproducer:
#include <fcntl.h>
#include <stddef.h>
#include <stdint.h>
#include <sys/ioctl.h>
#include <drm/drm.h>
int main(void)
{
int cardfd = open("/dev/dri/card0", O_RDONLY);
ioctl(cardfd, DRM_IOCTL_GEM_CLOSE,
&(struct drm_gem_close) { .handle = -1 } );
}
Link: http://lkml.kernel.org/r/20170906235306.20534-1-ebiggers3@gmail.com
Fixes: 0a835c4f09 ("Reimplement IDR and IDA using the radix tree")
Signed-off-by: Eric Biggers <ebiggers@google.com>
Acked-by: Tejun Heo <tj@kernel.org>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Matthew Wilcox <mawilcox@microsoft.com>
Cc: <stable@vger.kernel.org> [v4.11+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Pull libnvdimm from Dan Williams:
"A rework of media error handling in the BTT driver and other updates.
It has appeared in a few -next releases and collected some late-
breaking build-error and warning fixups as a result.
Summary:
- Media error handling support in the Block Translation Table (BTT)
driver is reworked to address sleeping-while-atomic locking and
memory-allocation-context conflicts.
- The dax_device lookup overhead for xfs and ext4 is moved out of the
iomap hot-path to a mount-time lookup.
- A new 'ecc_unit_size' sysfs attribute is added to advertise the
read-modify-write boundary property of a persistent memory range.
- Preparatory fix-ups for arm and powerpc pmem support are included
along with other miscellaneous fixes"
* tag 'libnvdimm-for-4.14' of git://git.kernel.org/pub/scm/linux/kernel/git/nvdimm/nvdimm: (26 commits)
libnvdimm, btt: fix format string warnings
libnvdimm, btt: clean up warning and error messages
ext4: fix null pointer dereference on sbi
libnvdimm, nfit: move the check on nd_reserved2 to the endpoint
dax: fix FS_DAX=n BLOCK=y compilation
libnvdimm: fix integer overflow static analysis warning
libnvdimm, nd_blk: remove mmio_flush_range()
libnvdimm, btt: rework error clearing
libnvdimm: fix potential deadlock while clearing errors
libnvdimm, btt: cache sector_size in arena_info
libnvdimm, btt: ensure that flags were also unchanged during a map_read
libnvdimm, btt: refactor map entry operations with macros
libnvdimm, btt: fix a missed NVDIMM_IO_ATOMIC case in the write path
libnvdimm, nfit: export an 'ecc_unit_size' sysfs attribute
ext4: perform dax_device lookup at mount
ext2: perform dax_device lookup at mount
xfs: perform dax_device lookup at mount
dax: introduce a fs_dax_get_by_bdev() helper
libnvdimm, btt: check memory allocation failure
libnvdimm, label: fix index block size calculation
...
Every for_each_XXX_cpu() invocation calls cpumask_next() which is an
inline function:
static inline unsigned int cpumask_next(int n, const struct cpumask *srcp)
{
/* -1 is a legal arg here. */
if (n != -1)
cpumask_check(n);
return find_next_bit(cpumask_bits(srcp), nr_cpumask_bits, n + 1);
}
However!
find_next_bit() is regular out-of-line function which means "nr_cpu_ids"
load and increment happen at the caller resulting in a lot of bloat
x86_64 defconfig:
add/remove: 3/0 grow/shrink: 8/373 up/down: 155/-5668 (-5513)
x86_64 allyesconfig-ish:
add/remove: 3/1 grow/shrink: 57/634 up/down: 3515/-28177 (-24662) !!!
Some archs redefine find_next_bit() but it is OK:
m68k inline but SMP is not supported
arm out-of-line
unicore32 out-of-line
Function call will happen anyway, so move load and increment into callee.
Link: http://lkml.kernel.org/r/20170824230010.GA1593@avx2
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The sprint_oid() utility function doesn't properly check the buffer size
that it causes that the warning in vsnprintf() be triggered. For
example on v4.1 kernel:
------------[ cut here ]------------
WARNING: CPU: 0 PID: 2357 at lib/vsprintf.c:1867 vsnprintf+0x5a7/0x5c0()
...
We can trigger this issue by injecting maliciously crafted x509 cert in
DER format. Just using hex editor to change the length of OID to over
the length of the SEQUENCE container. For example:
0:d=0 hl=4 l= 980 cons: SEQUENCE
4:d=1 hl=4 l= 700 cons: SEQUENCE
8:d=2 hl=2 l= 3 cons: cont [ 0 ]
10:d=3 hl=2 l= 1 prim: INTEGER :02
13:d=2 hl=2 l= 9 prim: INTEGER :9B47FAF791E7D1E3
24:d=2 hl=2 l= 13 cons: SEQUENCE
26:d=3 hl=2 l= 9 prim: OBJECT :sha256WithRSAEncryption
37:d=3 hl=2 l= 0 prim: NULL
39:d=2 hl=2 l= 121 cons: SEQUENCE
41:d=3 hl=2 l= 22 cons: SET
43:d=4 hl=2 l= 20 cons: SEQUENCE <=== the SEQ length is 20
45:d=5 hl=2 l= 3 prim: OBJECT :organizationName
<=== the original length is 3, change the length of OID to over the length of SEQUENCE
Pawel Wieczorkiewicz reported this problem and Takashi Iwai provided
patch to fix it by checking the bufsize in sprint_oid().
Link: http://lkml.kernel.org/r/20170903021646.2080-1-jlee@suse.com
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: "Lee, Chun-Yi" <jlee@suse.com>
Reported-by: Pawel Wieczorkiewicz <pwieczorkiewicz@suse.com>
Cc: David Howells <dhowells@redhat.com>
Cc: Rusty Russell <rusty@rustcorp.com.au>
Cc: Pawel Wieczorkiewicz <pwieczorkiewicz@suse.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Current implementation of bitmap_parselist() uses a static variable to
save local state while setting bits in the bitmap. It is obviously wrong
if we assume execution in multiprocessor environment. Fortunately, it's
possible to rewrite this portion of code to avoid using the static
variable.
It is also possible to set bits in the mask per-range with bitmap_set(),
not per-bit, as it is implemented now, with set_bit(); which is way
faster.
The important side effect of this change is that setting bits in this
function from now is not per-bit atomic and less memory-ordered. This is
because set_bit() guarantees the order of memory accesses, while
bitmap_set() does not. I think that it is the advantage of the new
approach, because the bitmap_parselist() is intended to initialise bit
arrays, and user should protect the whole bitmap during initialisation if
needed. So protecting individual bits looks expensive and useless. Also,
other range-oriented functions in lib/bitmap.c don't worry much about
atomicity.
With all that, setting 2k bits in map with the pattern like 0-2047:128/256
becomes ~50 times faster after applying the patch in my testing
environment (arm64 hosted on qemu).
The second patch of the series adds the test for bitmap_parselist(). It's
not intended to cover all tricky cases, just to make sure that I didn't
screw up during rework.
Link: http://lkml.kernel.org/r/20170807225438.16161-1-ynorov@caviumnetworks.com
Signed-off-by: Yury Norov <ynorov@caviumnetworks.com>
Cc: Noam Camus <noamca@mellanox.com>
Cc: Rasmus Villemoes <linux@rasmusvillemoes.dk>
Cc: Matthew Wilcox <mawilcox@microsoft.com>
Cc: Mauro Carvalho Chehab <mchehab@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The only times the nil-parent (root node) condition is true is when the
node is the first in the tree, or after fixing rbtree rule #4 and the
case 1 rebalancing made the node the root. Such conditions do not apply
most of the time:
(i) The common case in an rbtree is to have more than a single node,
so this is only true for the first rb_insert().
(ii) While there is a chance only one first rotation is needed, cases
where the node's uncle is black (cases 2,3) are more common as we can
have the following scenarios during the rotation looping:
case1 only, case1+1, case2+3, case1+2+3, case3 only, etc.
This patch, therefore, adds an unlikely() optimization to this
conditional. When profiling with CONFIG_PROFILE_ANNOTATED_BRANCHES, a
kernel build shows that the incorrect rate is less than 15%, and for
workloads that involve insert mostly trees overtime tend to have less
than 2% incorrect rate.
Link: http://lkml.kernel.org/r/20170719014603.19029-3-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Patch series "rbtree: Cache leftmost node internally", v4.
A series to extending rbtrees to internally cache the leftmost node such
that we can have fast overlap check optimization for all interval tree
users[1]. The benefits of this series are that:
(i) Unify users that do internal leftmost node caching.
(ii) Optimize all interval tree users.
(iii) Convert at least two new users (epoll and procfs) to the new interface.
This patch (of 16):
Red-black tree semantics imply that nodes with smaller or greater (or
equal for duplicates) keys always be to the left and right,
respectively. For the kernel this is extremely evident when considering
our rb_first() semantics. Enabling lookups for the smallest node in the
tree in O(1) can save a good chunk of cycles in not having to walk down
the tree each time. To this end there are a few core users that
explicitly do this, such as the scheduler and rtmutexes. There is also
the desire for interval trees to have this optimization allowing faster
overlap checking.
This patch introduces a new 'struct rb_root_cached' which is just the
root with a cached pointer to the leftmost node. The reason why the
regular rb_root was not extended instead of adding a new structure was
that this allows the user to have the choice between memory footprint
and actual tree performance. The new wrappers on top of the regular
rb_root calls are:
- rb_first_cached(cached_root) -- which is a fast replacement
for rb_first.
- rb_insert_color_cached(node, cached_root, new)
- rb_erase_cached(node, cached_root)
In addition, augmented cached interfaces are also added for basic
insertion and deletion operations; which becomes important for the
interval tree changes.
With the exception of the inserts, which adds a bool for updating the
new leftmost, the interfaces are kept the same. To this end, porting rb
users to the cached version becomes really trivial, and keeping current
rbtree semantics for users that don't care about the optimization
requires zero overhead.
Link: http://lkml.kernel.org/r/20170719014603.19029-2-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Reviewed-by: Jan Kara <jack@suse.cz>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Linus Torvalds
Michal Hocko
Geert Uytterhoeven
Eric Biggers
Alexey Dobriyan
Dan Carpenter
Takashi Iwai
Eric Dumazet
Baoquan He
Yury Norov
Florian Fainelli
Andy Shevchenko
Davidlohr Bueso
Matthew Wilcox