Skip to content

cloud-kernel
cloud-kernel

openanolis / cloud-kernel
接近 2 年 前同步成功

通知 170
Star 36
Fork 7
cloud-kernel
cloud-kernel
切换分支/标签
  1. 06 6月, 2018 2 次提交
    • M
      rseq: Introduce restartable sequences system call · d7822b1e
      由 Mathieu Desnoyers 提交于 
      Expose a new system call allowing each thread to register one userspace
memory area to be used as an ABI between kernel and user-space for two
purposes: user-space restartable sequences and quick access to read the
current CPU number value from user-space.

* Restartable sequences (per-cpu atomics)

Restartables sequences allow user-space to perform update operations on
per-cpu data without requiring heavy-weight atomic operations.

The restartable critical sections (percpu atomics) work has been started
by Paul Turner and Andrew Hunter. It lets the kernel handle restart of
critical sections. [1] [2] The re-implementation proposed here brings a
few simplifications to the ABI which facilitates porting to other
architectures and speeds up the user-space fast path.

Here are benchmarks of various rseq use-cases.

Test hardware:

arm32: ARMv7 Processor rev 4 (v7l) "Cubietruck", 2-core
x86-64: Intel E5-2630 v3@2.40GHz, 16-core, hyperthreading

The following benchmarks were all performed on a single thread.

* Per-CPU statistic counter increment

                getcpu+atomic (ns/op)    rseq (ns/op)    speedup
arm32:                344.0                 31.4          11.0
x86-64:                15.3                  2.0           7.7

* LTTng-UST: write event 32-bit header, 32-bit payload into tracer
             per-cpu buffer

                getcpu+atomic (ns/op)    rseq (ns/op)    speedup
arm32:               2502.0                 2250.0         1.1
x86-64:               117.4                   98.0         1.2

* liburcu percpu: lock-unlock pair, dereference, read/compare word

                getcpu+atomic (ns/op)    rseq (ns/op)    speedup
arm32:                751.0                 128.5          5.8
x86-64:                53.4                  28.6          1.9

* jemalloc memory allocator adapted to use rseq

Using rseq with per-cpu memory pools in jemalloc at Facebook (based on
rseq 2016 implementation):

The production workload response-time has 1-2% gain avg. latency, and
the P99 overall latency drops by 2-3%.

* Reading the current CPU number

Speeding up reading the current CPU number on which the caller thread is
running is done by keeping the current CPU number up do date within the
cpu_id field of the memory area registered by the thread. This is done
by making scheduler preemption set the TIF_NOTIFY_RESUME flag on the
current thread. Upon return to user-space, a notify-resume handler
updates the current CPU value within the registered user-space memory
area. User-space can then read the current CPU number directly from
memory.

Keeping the current cpu id in a memory area shared between kernel and
user-space is an improvement over current mechanisms available to read
the current CPU number, which has the following benefits over
alternative approaches:

- 35x speedup on ARM vs system call through glibc
- 20x speedup on x86 compared to calling glibc, which calls vdso
  executing a "lsl" instruction,
- 14x speedup on x86 compared to inlined "lsl" instruction,
- Unlike vdso approaches, this cpu_id value can be read from an inline
  assembly, which makes it a useful building block for restartable
  sequences.
- The approach of reading the cpu id through memory mapping shared
  between kernel and user-space is portable (e.g. ARM), which is not the
  case for the lsl-based x86 vdso.

On x86, yet another possible approach would be to use the gs segment
selector to point to user-space per-cpu data. This approach performs
similarly to the cpu id cache, but it has two disadvantages: it is
not portable, and it is incompatible with existing applications already
using the gs segment selector for other purposes.

Benchmarking various approaches for reading the current CPU number:

ARMv7 Processor rev 4 (v7l)
Machine model: Cubietruck
- Baseline (empty loop):                                    8.4 ns
- Read CPU from rseq cpu_id:                               16.7 ns
- Read CPU from rseq cpu_id (lazy register):               19.8 ns
- glibc 2.19-0ubuntu6.6 getcpu:                           301.8 ns
- getcpu system call:                                     234.9 ns

x86-64 Intel(R) Xeon(R) CPU E5-2630 v3 @ 2.40GHz:
- Baseline (empty loop):                                    0.8 ns
- Read CPU from rseq cpu_id:                                0.8 ns
- Read CPU from rseq cpu_id (lazy register):                0.8 ns
- Read using gs segment selector:                           0.8 ns
- "lsl" inline assembly:                                   13.0 ns
- glibc 2.19-0ubuntu6 getcpu:                              16.6 ns
- getcpu system call:                                      53.9 ns

- Speed (benchmark taken on v8 of patchset)

Running 10 runs of hackbench -l 100000 seems to indicate, contrary to
expectations, that enabling CONFIG_RSEQ slightly accelerates the
scheduler:

Configuration: 2 sockets * 8-core Intel(R) Xeon(R) CPU E5-2630 v3 @
2.40GHz (directly on hardware, hyperthreading disabled in BIOS, energy
saving disabled in BIOS, turboboost disabled in BIOS, cpuidle.off=1
kernel parameter), with a Linux v4.6 defconfig+localyesconfig,
restartable sequences series applied.

* CONFIG_RSEQ=n

avg.:      41.37 s
std.dev.:   0.36 s

* CONFIG_RSEQ=y

avg.:      40.46 s
std.dev.:   0.33 s

- Size

On x86-64, between CONFIG_RSEQ=n/y, the text size increase of vmlinux is
567 bytes, and the data size increase of vmlinux is 5696 bytes.

[1] https://lwn.net/Articles/650333/
[2] http://www.linuxplumbersconf.org/2013/ocw/system/presentations/1695/original/LPC%20-%20PerCpu%20Atomics.pdfSigned-off-by: NMathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: NThomas Gleixner <tglx@linutronix.de>
Acked-by: NPeter Zijlstra (Intel) <peterz@infradead.org>
Cc: Joel Fernandes <joelaf@google.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Dave Watson <davejwatson@fb.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: "H . Peter Anvin" <hpa@zytor.com>
Cc: Chris Lameter <cl@linux.com>
Cc: Russell King <linux@arm.linux.org.uk>
Cc: Andrew Hunter <ahh@google.com>
Cc: Michael Kerrisk <mtk.manpages@gmail.com>
Cc: "Paul E . McKenney" <paulmck@linux.vnet.ibm.com>
Cc: Paul Turner <pjt@google.com>
Cc: Boqun Feng <boqun.feng@gmail.com>
Cc: Josh Triplett <josh@joshtriplett.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Ben Maurer <bmaurer@fb.com>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Cc: linux-api@vger.kernel.org
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Link: http://lkml.kernel.org/r/20151027235635.16059.11630.stgit@pjt-glaptop.roam.corp.google.com
Link: http://lkml.kernel.org/r/20150624222609.6116.86035.stgit@kitami.mtv.corp.google.com
Link: https://lkml.kernel.org/r/20180602124408.8430-3-mathieu.desnoyers@efficios.com
      d7822b1e
    • M
      uapi/headers: Provide types_32_64.h · b575e837
      由 Mathieu Desnoyers 提交于 
      Provide helper macros for fields which represent pointers in
kernel-userspace ABI. This facilitates handling of 32-bit
user-space by 64-bit kernels by defining those fields as
32-bit 0-padding and 32-bit integer on 32-bit architectures,
which allows the kernel to treat those as 64-bit integers.
The order of padding and 32-bit integer depends on the
endianness.
Signed-off-by: NMathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: NThomas Gleixner <tglx@linutronix.de>
Cc: Joel Fernandes <joelaf@google.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Dave Watson <davejwatson@fb.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: "H . Peter Anvin" <hpa@zytor.com>
Cc: Chris Lameter <cl@linux.com>
Cc: Russell King <linux@arm.linux.org.uk>
Cc: Andrew Hunter <ahh@google.com>
Cc: Michael Kerrisk <mtk.manpages@gmail.com>
Cc: "Paul E . McKenney" <paulmck@linux.vnet.ibm.com>
Cc: Paul Turner <pjt@google.com>
Cc: Boqun Feng <boqun.feng@gmail.com>
Cc: Josh Triplett <josh@joshtriplett.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Ben Maurer <bmaurer@fb.com>
Cc: linux-api@vger.kernel.org
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Link: https://lkml.kernel.org/r/20180602124408.8430-2-mathieu.desnoyers@efficios.com
      b575e837
  3. 05 6月, 2018 1 次提交
    • L
      swait: strengthen language to discourage use · c5e7a7ea
      由 Linus Torvalds 提交于 
      We already earlier discouraged people from using this interface in
commit 88796e7e ("sched/swait: Document it clearly that the swait
facilities are special and shouldn't be used"), but I just got a pull
request with a new broken user.

So make the comment *really* clear.

The swait interfaces are bad, and should not be used unless you have
some *very* strong reasons that include tons of hard performance numbers
on just why you want to use them, and you show that you actually
understand that they aren't at all like the normal wait/wakeup
interfaces.

So far, every single user has been suspect.  The main user is KVM, which
is completely pointless (there is only ever one waiter, which avoids the
interface subtleties, but also means that having a queue instead of a
pointer is counter-productive and certainly not an "optimization").

So make the comments much stronger.

Not that anybody likely reads them anyway, but there's always some
slight hope that it will cause somebody to think twice.

I'd like to remove this interface entirely, but there is the theoretical
possibility that it's actually the right thing to use in some situation,
most likely some deep RT use.
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>
      c5e7a7ea
  5. 03 6月, 2018 1 次提交
    • J
      block: don't use blocking queue entered for recursive bio submits · cd4a4ae4
      由 Jens Axboe 提交于 
      If we end up splitting a bio and the queue goes away between
the initial submission and the later split submission, then we
can block forever in blk_queue_enter() waiting for the reference
to drop to zero. This will never happen, since we already hold
a reference.

Mark a split bio as already having entered the queue, so we can
just use the live non-blocking queue enter variant.

Thanks to Tetsuo Handa for the analysis.

Reported-by: syzbot+c4f9cebf9d651f6e54de@syzkaller.appspotmail.com
Signed-off-by: NJens Axboe <axboe@kernel.dk>
      cd4a4ae4
  7. 02 6月, 2018 1 次提交
    • D
      bpf: fix uapi hole for 32 bit compat applications · 36f9814a
      由 Daniel Borkmann 提交于 
      In 64 bit, we have a 4 byte hole between ifindex and netns_dev in the
case of struct bpf_map_info but also struct bpf_prog_info. In net-next
commit b85fab0e ("bpf: Add gpl_compatible flag to struct bpf_prog_info")
added a bitfield into it to expose some flags related to programs. Thus,
add an unnamed __u32 bitfield for both so that alignment keeps the same
in both 32 and 64 bit cases, and can be naturally extended from there
as in b85fab0e.

Before:

  # file test.o
  test.o: ELF 32-bit LSB relocatable, Intel 80386, version 1 (SYSV), not stripped
  # pahole test.o
  struct bpf_map_info {
	__u32                      type;                 /*     0     4 */
	__u32                      id;                   /*     4     4 */
	__u32                      key_size;             /*     8     4 */
	__u32                      value_size;           /*    12     4 */
	__u32                      max_entries;          /*    16     4 */
	__u32                      map_flags;            /*    20     4 */
	char                       name[16];             /*    24    16 */
	__u32                      ifindex;              /*    40     4 */
	__u64                      netns_dev;            /*    44     8 */
	__u64                      netns_ino;            /*    52     8 */

	/* size: 64, cachelines: 1, members: 10 */
	/* padding: 4 */
  };

After (same as on 64 bit):

  # file test.o
  test.o: ELF 32-bit LSB relocatable, Intel 80386, version 1 (SYSV), not stripped
  # pahole test.o
  struct bpf_map_info {
	__u32                      type;                 /*     0     4 */
	__u32                      id;                   /*     4     4 */
	__u32                      key_size;             /*     8     4 */
	__u32                      value_size;           /*    12     4 */
	__u32                      max_entries;          /*    16     4 */
	__u32                      map_flags;            /*    20     4 */
	char                       name[16];             /*    24    16 */
	__u32                      ifindex;              /*    40     4 */

	/* XXX 4 bytes hole, try to pack */

	__u64                      netns_dev;            /*    48     8 */
	__u64                      netns_ino;            /*    56     8 */
	/* --- cacheline 1 boundary (64 bytes) --- */

	/* size: 64, cachelines: 1, members: 10 */
	/* sum members: 60, holes: 1, sum holes: 4 */
  };
Reported-by: NDmitry V. Levin <ldv@altlinux.org>
Reported-by: NEugene Syromiatnikov <esyr@redhat.com>
Fixes: 52775b33 ("bpf: offload: report device information about offloaded maps")
Fixes: 675fc275 ("bpf: offload: report device information for offloaded programs")
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>
Acked-by: NAlexei Starovoitov <ast@kernel.org>
Signed-off-by: NAlexei Starovoitov <ast@kernel.org>
      36f9814a
  9. 01 6月, 2018 5 次提交
  11. 31 5月, 2018 9 次提交
  13. 30 5月, 2018 1 次提交
  15. 29 5月, 2018 18 次提交
  17. 28 5月, 2018 2 次提交