[RFC PATCH v9 for 4.15 01/14] Restartable sequences system call
From: Mathieu Desnoyers
Date: Thu Oct 12 2017 - 19:08:03 EST
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. A locking-based
fall-back, purely implemented in user-space, is proposed here to deal
with debugger single-stepping. This fallback interacts with rseq_start()
and rseq_finish(), which force retries in response to concurrent
lock-based activity.
Here are benchmarks of counter increment in various scenarios compared
to restartable sequences. Those benchmarks were taken on v8 of the
patchset.
ARMv7 Processor rev 4 (v7l)
Machine model: Cubietruck
Counter increment speed (ns/increment)
1 thread 2 threads
global increment (baseline) 6 N/A
percpu rseq increment 50 52
percpu rseq spinlock 94 94
global atomic increment 48 74 (__sync_add_and_fetch_4)
global atomic CAS 50 172 (__sync_val_compare_and_swap_4)
global pthread mutex 148 862
ARMv7 Processor rev 10 (v7l)
Machine model: Wandboard
Counter increment speed (ns/increment)
1 thread 4 threads
global increment (baseline) 7 N/A
percpu rseq increment 50 50
percpu rseq spinlock 82 84
global atomic increment 44 262 (__sync_add_and_fetch_4)
global atomic CAS 46 316 (__sync_val_compare_and_swap_4)
global pthread mutex 146 1400
x86-64 Intel(R) Xeon(R) CPU E5-2630 v3 @ 2.40GHz:
Counter increment speed (ns/increment)
1 thread 8 threads
global increment (baseline) 3.0 N/A
percpu rseq increment 3.6 3.8
percpu rseq spinlock 5.6 6.2
global LOCK; inc 8.0 166.4
global LOCK; cmpxchg 13.4 435.2
global pthread mutex 25.2 1363.6
* 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
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
2855 bytes, and the data size increase of vmlinux is 1024 bytes.
* CONFIG_RSEQ=n
text data bss dec hex filename
9964559 4256280 962560 15183399 e7ae27 vmlinux.norseq
* CONFIG_RSEQ=y
text data bss dec hex filename
9967414 4257304 962560 15187278 e7bd4e vmlinux.rseq
[1] https://lwn.net/Articles/650333/
[2] http://www.linuxplumbersconf.org/2013/ocw/system/presentations/1695/original/LPC%20-%20PerCpu%20Atomics.pdf
Link: http://lkml.kernel.org/r/20151027235635.16059.11630.stgit@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
Link: http://lkml.kernel.org/r/20150624222609.6116.86035.stgit@xxxxxxxxxxxxxxxxxxxxxxxxxx
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@xxxxxxxxxxxx>
CC: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
CC: Paul Turner <pjt@xxxxxxxxxx>
CC: Andrew Hunter <ahh@xxxxxxxxxx>
CC: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
CC: Andy Lutomirski <luto@xxxxxxxxxxxxxx>
CC: Andi Kleen <andi@xxxxxxxxxxxxxx>
CC: Dave Watson <davejwatson@xxxxxx>
CC: Chris Lameter <cl@xxxxxxxxx>
CC: Ingo Molnar <mingo@xxxxxxxxxx>
CC: "H. Peter Anvin" <hpa@xxxxxxxxx>
CC: Ben Maurer <bmaurer@xxxxxx>
CC: Steven Rostedt <rostedt@xxxxxxxxxxx>
CC: "Paul E. McKenney" <paulmck@xxxxxxxxxxxxxxxxxx>
CC: Josh Triplett <josh@xxxxxxxxxxxxxxxx>
CC: Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxxx>
CC: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
CC: Russell King <linux@xxxxxxxxxxxxxxxx>
CC: Catalin Marinas <catalin.marinas@xxxxxxx>
CC: Will Deacon <will.deacon@xxxxxxx>
CC: Michael Kerrisk <mtk.manpages@xxxxxxxxx>
CC: Boqun Feng <boqun.feng@xxxxxxxxx>
CC: Alexander Viro <viro@xxxxxxxxxxxxxxxxxx>
CC: linux-api@xxxxxxxxxxxxxxx
---
Changes since v1:
- Return -1, errno=EINVAL if cpu_cache pointer is not aligned on
sizeof(int32_t).
- Update man page to describe the pointer alignement requirements and
update atomicity guarantees.
- Add MAINTAINERS file GETCPU_CACHE entry.
- Remove dynamic memory allocation: go back to having a single
getcpu_cache entry per thread. Update documentation accordingly.
- Rebased on Linux 4.4.
Changes since v2:
- Introduce a "cmd" argument, along with an enum with GETCPU_CACHE_GET
and GETCPU_CACHE_SET. Introduce a uapi header linux/getcpu_cache.h
defining this enumeration.
- Split resume notifier architecture implementation from the system call
wire up in the following arch-specific patches.
- Man pages updates.
- Handle 32-bit compat pointers.
- Simplify handling of getcpu_cache GETCPU_CACHE_SET compiler barrier:
set the current cpu cache pointer before doing the cache update, and
set it back to NULL if the update fails. Setting it back to NULL on
error ensures that no resume notifier will trigger a SIGSEGV if a
migration happened concurrently.
Changes since v3:
- Fix __user annotations in compat code,
- Update memory ordering comments.
- Rebased on kernel v4.5-rc5.
Changes since v4:
- Inline getcpu_cache_fork, getcpu_cache_execve, and getcpu_cache_exit.
- Add new line between if() and switch() to improve readability.
- Added sched switch benchmarks (hackbench) and size overhead comparison
to change log.
Changes since v5:
- Rename "getcpu_cache" to "thread_local_abi", allowing to extend
this system call to cover future features such as restartable critical
sections. Generalizing this system call ensures that we can add
features similar to the cpu_id field within the same cache-line
without having to track one pointer per feature within the task
struct.
- Add a tlabi_nr parameter to the system call, thus allowing to extend
the ABI beyond the initial 64-byte structure by registering structures
with tlabi_nr greater than 0. The initial ABI structure is associated
with tlabi_nr 0.
- Rebased on kernel v4.5.
Changes since v6:
- Integrate "restartable sequences" v2 patchset from Paul Turner.
- Add handling of single-stepping purely in user-space, with a
fallback to locking after 2 rseq failures to ensure progress, and
by exposing a __rseq_table section to debuggers so they know where
to put breakpoints when dealing with rseq assembly blocks which
can be aborted at any point.
- make the code and ABI generic: porting the kernel implementation
simply requires to wire up the signal handler and return to user-space
hooks, and allocate the syscall number.
- extend testing with a fully configurable test program. See
param_spinlock_test -h for details.
- handling of rseq ENOSYS in user-space, also with a fallback
to locking.
- modify Paul Turner's rseq ABI to only require a single TLS store on
the user-space fast-path, removing the need to populate two additional
registers. This is made possible by introducing struct rseq_cs into
the ABI to describe a critical section start_ip, post_commit_ip, and
abort_ip.
- Rebased on kernel v4.7-rc7.
Changes since v7:
- Documentation updates.
- Integrated powerpc architecture support.
- Compare rseq critical section start_ip, allows shriking the user-space
fast-path code size.
- Added Peter Zijlstra, Paul E. McKenney and Boqun Feng as
co-maintainers.
- Added do_rseq2 and do_rseq_memcpy to test program helper library.
- Code cleanup based on review from Peter Zijlstra, Andy Lutomirski and
Boqun Feng.
- Rebase on kernel v4.8-rc2.
Changes since v8:
- clear rseq_cs even if non-nested. Speeds up user-space fast path by
removing the final "rseq_cs=NULL" assignment.
- add enum rseq_flags: critical sections and threads can set migration,
preemption and signal "disable" flags to inhibit rseq behavior.
- rseq_event_counter needs to be updated with a pre-increment: Otherwise
misses an increment after exec (when TLS and in-kernel states are
initially 0).
Man page associated:
RSEQ(2) Linux Programmer's Manual RSEQ(2)
NAME
rseq - Restartable sequences and cpu number cache
SYNOPSIS
#include <linux/rseq.h>
int rseq(struct rseq * rseq, int flags);
DESCRIPTION
The rseq() ABI accelerates user-space operations on per-cpu data
by defining a shared data structure ABI between each user-space
thread and the kernel.
It allows user-space to perform update operations on per-cpu data
without requiring heavy-weight atomic operations.
Restartable sequences are atomic with respect to preemption (makâ
ing it atomic with respect to other threads running on the same
CPU), as well as signal delivery (user-space execution contexts
nested over the same thread).
It is suited for update operations on per-cpu data.
It can be used on data structures shared between threads within a
process, and on data structures shared between threads across difâ
ferent processes.
Some examples of operations that can be accelerated by this ABI:
 Querying the current CPU number,
 Incrementing per-CPU counters,
 Modifying data protected by per-CPU spinlocks,
 Inserting/removing elements in per-CPU linked-lists,
 Writing/reading per-CPU ring buffers content.
The rseq argument is a pointer to the thread-local rseq structure
to be shared between kernel and user-space. A NULL rseq value
unregisters the current thread rseq structure.
The layout of struct rseq is as follows:
Structure alignment
This structure is aligned on multiples of 128 bytes.
Structure size
This structure has a fixed size of 128 bytes.
Fields
cpu_id
Cache of the CPU number on which the current thread is runâ
ning.
event_counter
Counter guaranteed to be incremented when the current
thread is preempted or when a signal is delivered to the
current thread.
rseq_cs
The rseq_cs field is a pointer to a struct rseq_cs. Is is
NULL when no rseq assembly block critical section is active
for the current thread. Setting it to point to a critical
section descriptor (struct rseq_cs) marks the beginning of
the critical section. It is cleared after the end of the
critical section.
The layout of struct rseq_cs is as follows:
Structure alignment
This structure is aligned on multiples of 256 bytes.
Structure size
This structure has a fixed size of 256 bytes.
Fields
start_ip
Instruction pointer address of the first instruction of the
sequence of consecutive assembly instructions.
post_commit_ip
Instruction pointer address after the last instruction of
the sequence of consecutive assembly instructions.
abort_ip
Instruction pointer address where to move the execution
flow in case of abort of the sequence of consecutive assemâ
bly instructions.
Upon registration, the flags argument is currently unused and must
be specified as 0. Upon unregistration, the flags argument can be
either specified as 0, or as RSEQ_FORCE_UNREGISTER, which will
force unregistration of the current rseq address rather than
requiring each registration to be matched by an unregistration.
Libraries and applications should keep the rseq structure in a
thread-local storage variable. Since only one rseq address can be
registered per thread, applications and libraries should define
their struct rseq as a volatile thread-local storage variable with
the weak symbol __rseq_abi. This allows using rseq from an appliâ
cation executable and from multiple shared libraries linked to the
same executable. The cpu_id field should be initialized to -1.
Each thread is responsible for registering and unregistering its
rseq structure. No more than one rseq structure address can be
registered per thread at a given time. The same address can be
registered more than once for a thread, and each registration
needs to have a matching unregistration before the address is
effectively unregistered. After the rseq address is effectively
unregistered for a thread, a new address can be registered. Unregâ
istration of associated rseq structure is implicitly performed
when a thread or process exits.
In a typical usage scenario, the thread registering the rseq
structure will be performing loads and stores from/to that strucâ
ture. It is however also allowed to read that structure from other
threads. The rseq field updates performed by the kernel provide
relaxed atomicity semantics, which guarantee that other threads
performing relaxed atomic reads of the cpu number cache will
always observe a consistent value.
RETURN VALUE
A return value of 0 indicates success. On error, -1 is returned,
and errno is set appropriately.
ERRORS
EINVAL Either flags contains an invalid value, or rseq contains an
address which is not appropriately aligned.
ENOSYS The rseq() system call is not implemented by this kernel.
EFAULT rseq is an invalid address.
EBUSY The rseq argument contains a non-NULL address which differs
from the memory location already registered for this
thread.
EOVERFLOW
Registering the rseq address is not allowed because it
would cause a reference counter overflow.
ENOENT The rseq argument is NULL, but no memory location is curâ
rently registered for this thread.
VERSIONS
The rseq() system call was added in Linux 4.X (TODO).
CONFORMING TO
rseq() is Linux-specific.
ALGORITHM
The restartable sequences mechanism is the overlap of two distinct
restart mechanisms: a sequence counter tracking preemption and
signal delivery for high-level code, and an ip-fixup-based mechaâ
nism for the final assembly instruction sequence.
A high-level summary of the algorithm to use rseq from user-space
is as follows:
The high-level code between rseq_start() and rseq_finish() loads
the current value of the sequence counter in rseq_start(), and
then it gets compared with the new current value within the
rseq_finish() restartable instruction sequence. Between
rseq_start() and rseq_finish(), the high-level code can perform
operations that do not have side-effects, such as getting the curâ
rent CPU number, and loading from variables.
Stores are performed at the very end of the restartable sequence
assembly block. Each assembly block defines a struct rseq_cs
structure which describes the start_ip and post_commit_ip
addresses, as well as the abort_ip address where the kernel should
move the thread instruction pointer if a rseq critical section
assembly block is preempted or if a signal is delivered on top of
a rseq critical section assembly block.
Detailed algorithm of rseq use:
rseq_start()
0. Userspace loads the current event counter value from the
event_counter field of the registered struct rseq TLS area,
rseq_finish()
Steps [1]-[3] (inclusive) need to be a sequence of instrucâ
tions in userspace that can handle being moved to the
abort_ip between any of those instructions.
The abort_ip address needs to be less than start_ip, or
greater-or-equal the post_commit_ip. Step [4] and the
failure code step [F1] need to be at addresses lesser than
start_ip, or greater-or-equal the post_commit_ip.
[ start_ip ]
1. Userspace stores the address of the struct rseq_cs assembly
block descriptor into the rseq_cs field of the registered
struct rseq TLS area.
2. Userspace tests to see whether the current event_counter
value match the value loaded at [0]. Manually jumping to
[F1] in case of a mismatch.
Note that if we are preempted or interrupted by a signal
after [1] and before post_commit_ip, then the kernel also
performs the comparison performed in [2], and conditionally
clears the rseq_cs field of struct rseq, then jumps us to
abort_ip.
3. Userspace critical section final instruction before
post_commit_ip is the commit. The critical section is self-
terminating.
[ post_commit_ip ]
4. Userspace clears the rseq_cs field of the struct rseq TLS
area.
5. Return true.
On failure at [2]:
F1.
Userspace clears the rseq_cs field of the struct rseq TLS
area. Followed by step [F2].
[ abort_ip ]
F2.
Return false.
EXAMPLE
The following code uses the rseq() system call to keep a thread-local
storage variable up to date with the current CPU number, with a fallâ
back on sched_getcpu(3) if the cache is not available. For example
simplicity, it is done in main(), but multithreaded programs would
need to invoke rseq() from each program thread.
#define _GNU_SOURCE
#include <stdlib.h>
#include <stdio.h>
#include <unistd.h>
#include <stdint.h>
#include <sched.h>
#include <stddef.h>
#include <errno.h>
#include <string.h>
#include <stdbool.h>
#include <sys/syscall.h>
#include <linux/rseq.h>
__attribute__((weak)) __thread volatile struct rseq __rseq_abi = {
.u.e.cpu_id = -1,
};
static int
sys_rseq(volatile struct rseq *rseq_abi, int flags)
{
return syscall(__NR_rseq, rseq_abi, flags);
}
static int32_t
rseq_current_cpu_raw(void)
{
return __rseq_abi.u.e.cpu_id;
}
static int32_t
rseq_current_cpu(void)
{
int32_t cpu;
cpu = rseq_current_cpu_raw();
if (cpu < 0)
cpu = sched_getcpu();
return cpu;
}
static int
rseq_register_current_thread(void)
{
int rc;
rc = sys_rseq(&__rseq_abi, 0);
if (rc) {
fprintf(stderr,
"Error: sys_rseq(...) register failed(%d): %s\n",
errno, strerror(errno));
return -1;
}
return 0;
}
static int
rseq_unregister_current_thread(void)
{
int rc;
rc = sys_rseq(NULL, 0);
if (rc) {
fprintf(stderr,
"Error: sys_rseq(...) unregister failed(%d): %s\n",
errno, strerror(errno));
return -1;
}
return 0;
}
int
main(int argc, char **argv)
{
bool rseq_registered = false;
if (!rseq_register_current_thread()) {
rseq_registered = true;
} else {
fprintf(stderr,
"Unable to register restartable sequences.\n");
fprintf(stderr, "Using sched_getcpu() as fallback.\n");
}
printf("Current CPU number: %d\n", rseq_current_cpu());
if (rseq_registered && rseq_unregister_current_thread()) {
exit(EXIT_FAILURE);
}
exit(EXIT_SUCCESS);
}
SEE ALSO
sched_getcpu(3)
Linux 2016-08-19 RSEQ(2)
---
MAINTAINERS | 10 ++
arch/Kconfig | 7 +
fs/exec.c | 1 +
include/linux/sched.h | 89 ++++++++++++
include/uapi/linux/rseq.h | 131 +++++++++++++++++
init/Kconfig | 13 ++
kernel/Makefile | 1 +
kernel/fork.c | 2 +
kernel/rseq.c | 347 ++++++++++++++++++++++++++++++++++++++++++++++
kernel/sched/core.c | 4 +
kernel/sys_ni.c | 3 +
11 files changed, 608 insertions(+)
create mode 100644 include/uapi/linux/rseq.h
create mode 100644 kernel/rseq.c
diff --git a/MAINTAINERS b/MAINTAINERS
index 1c3feffb1c1c..f05c526fe1e8 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -11224,6 +11224,16 @@ F: include/dt-bindings/reset/
F: include/linux/reset.h
F: include/linux/reset-controller.h
+RESTARTABLE SEQUENCES SUPPORT
+M: Mathieu Desnoyers <mathieu.desnoyers@xxxxxxxxxxxx>
+M: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
+M: "Paul E. McKenney" <paulmck@xxxxxxxxxxxxxxxxxx>
+M: Boqun Feng <boqun.feng@xxxxxxxxx>
+L: linux-kernel@xxxxxxxxxxxxxxx
+S: Supported
+F: kernel/rseq.c
+F: include/uapi/linux/rseq.h
+
RFKILL
M: Johannes Berg <johannes@xxxxxxxxxxxxxxxx>
L: linux-wireless@xxxxxxxxxxxxxxx
diff --git a/arch/Kconfig b/arch/Kconfig
index 21d0089117fe..6f1203612403 100644
--- a/arch/Kconfig
+++ b/arch/Kconfig
@@ -257,6 +257,13 @@ config HAVE_REGS_AND_STACK_ACCESS_API
declared in asm/ptrace.h
For example the kprobes-based event tracer needs this API.
+config HAVE_RSEQ
+ bool
+ depends on HAVE_REGS_AND_STACK_ACCESS_API
+ help
+ This symbol should be selected by an architecture if it
+ supports an implementation of restartable sequences.
+
config HAVE_CLK
bool
help
diff --git a/fs/exec.c b/fs/exec.c
index 62175cbcc801..75fcbaeb0206 100644
--- a/fs/exec.c
+++ b/fs/exec.c
@@ -1794,6 +1794,7 @@ static int do_execveat_common(int fd, struct filename *filename,
/* execve succeeded */
current->fs->in_exec = 0;
current->in_execve = 0;
+ rseq_execve(current);
acct_update_integrals(current);
task_numa_free(current);
free_bprm(bprm);
diff --git a/include/linux/sched.h b/include/linux/sched.h
index c05ac5f5aa03..203abf387a14 100644
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -26,6 +26,7 @@
#include <linux/signal_types.h>
#include <linux/mm_types_task.h>
#include <linux/task_io_accounting.h>
+#include <linux/rseq.h>
/* task_struct member predeclarations (sorted alphabetically): */
struct audit_context;
@@ -966,6 +967,13 @@ struct task_struct {
unsigned long numa_pages_migrated;
#endif /* CONFIG_NUMA_BALANCING */
+#ifdef CONFIG_RSEQ
+ struct rseq __user *rseq;
+ u32 rseq_event_counter;
+ unsigned int rseq_refcount;
+ bool rseq_preempt, rseq_signal, rseq_migrate;
+#endif
+
struct tlbflush_unmap_batch tlb_ubc;
struct rcu_head rcu;
@@ -1626,4 +1634,85 @@ extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
#define TASK_SIZE_OF(tsk) TASK_SIZE
#endif
+#ifdef CONFIG_RSEQ
+static inline void rseq_set_notify_resume(struct task_struct *t)
+{
+ if (t->rseq)
+ set_tsk_thread_flag(t, TIF_NOTIFY_RESUME);
+}
+void __rseq_handle_notify_resume(struct pt_regs *regs);
+static inline void rseq_handle_notify_resume(struct pt_regs *regs)
+{
+ if (current->rseq)
+ __rseq_handle_notify_resume(regs);
+}
+/*
+ * If parent process has a registered restartable sequences area, the
+ * child inherits. Only applies when forking a process, not a thread. In
+ * case a parent fork() in the middle of a restartable sequence, set the
+ * resume notifier to force the child to retry.
+ */
+static inline void rseq_fork(struct task_struct *t, unsigned long clone_flags)
+{
+ if (clone_flags & CLONE_THREAD) {
+ t->rseq = NULL;
+ t->rseq_event_counter = 0;
+ t->rseq_refcount = 0;
+ } else {
+ t->rseq = current->rseq;
+ t->rseq_event_counter = current->rseq_event_counter;
+ t->rseq_refcount = current->rseq_refcount;
+ rseq_set_notify_resume(t);
+ }
+}
+static inline void rseq_execve(struct task_struct *t)
+{
+ t->rseq = NULL;
+ t->rseq_event_counter = 0;
+ t->rseq_refcount = 0;
+}
+static inline void rseq_sched_out(struct task_struct *t)
+{
+ rseq_set_notify_resume(t);
+}
+static inline void rseq_signal_deliver(struct pt_regs *regs)
+{
+ current->rseq_signal = true;
+ rseq_handle_notify_resume(regs);
+}
+static inline void rseq_preempt(struct task_struct *t)
+{
+ t->rseq_preempt = true;
+}
+static inline void rseq_migrate(struct task_struct *t)
+{
+ t->rseq_migrate = true;
+}
+#else
+static inline void rseq_set_notify_resume(struct task_struct *t)
+{
+}
+static inline void rseq_handle_notify_resume(struct pt_regs *regs)
+{
+}
+static inline void rseq_fork(struct task_struct *t, unsigned long clone_flags)
+{
+}
+static inline void rseq_execve(struct task_struct *t)
+{
+}
+static inline void rseq_sched_out(struct task_struct *t)
+{
+}
+static inline void rseq_signal_deliver(struct pt_regs *regs)
+{
+}
+static inline void rseq_preempt(struct task_struct *t)
+{
+}
+static inline void rseq_migrate(struct task_struct *t)
+{
+}
+#endif
+
#endif
diff --git a/include/uapi/linux/rseq.h b/include/uapi/linux/rseq.h
new file mode 100644
index 000000000000..8abd8b638ce0
--- /dev/null
+++ b/include/uapi/linux/rseq.h
@@ -0,0 +1,131 @@
+#ifndef _UAPI_LINUX_RSEQ_H
+#define _UAPI_LINUX_RSEQ_H
+
+/*
+ * linux/rseq.h
+ *
+ * Restartable sequences system call API
+ *
+ * Copyright (c) 2015-2016 Mathieu Desnoyers <mathieu.desnoyers@xxxxxxxxxxxx>
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifdef __KERNEL__
+# include <linux/types.h>
+#else /* #ifdef __KERNEL__ */
+# include <stdint.h>
+#endif /* #else #ifdef __KERNEL__ */
+
+#include <asm/byteorder.h>
+
+#ifdef __LP64__
+# define RSEQ_FIELD_u32_u64(field) uint64_t field
+#elif defined(__BYTE_ORDER) ? \
+ __BYTE_ORDER == __BIG_ENDIAN : defined(__BIG_ENDIAN)
+# define RSEQ_FIELD_u32_u64(field) uint32_t _padding ## field, field
+#else
+# define RSEQ_FIELD_u32_u64(field) uint32_t field, _padding ## field
+#endif
+
+enum rseq_flags {
+ RSEQ_FORCE_UNREGISTER = (1 << 0),
+};
+
+enum rseq_cs_flags {
+ RSEQ_CS_FLAG_NO_RESTART_ON_PREEMPT = (1U << 0),
+ RSEQ_CS_FLAG_NO_RESTART_ON_SIGNAL = (1U << 1),
+ RSEQ_CS_FLAG_NO_RESTART_ON_MIGRATE = (1U << 2),
+};
+
+/*
+ * struct rseq_cs is aligned on 4 * 8 bytes to ensure it is always
+ * contained within a single cache-line. It is usually declared as
+ * link-time constant data.
+ */
+struct rseq_cs {
+ RSEQ_FIELD_u32_u64(start_ip);
+ RSEQ_FIELD_u32_u64(post_commit_ip);
+ RSEQ_FIELD_u32_u64(abort_ip);
+ uint32_t flags;
+} __attribute__((aligned(4 * sizeof(uint64_t))));
+
+union rseq_cpu_event {
+ struct {
+ /*
+ * Restartable sequences cpu_id field.
+ * Updated by the kernel, and read by user-space with
+ * single-copy atomicity semantics. Aligned on 32-bit.
+ * Negative values are reserved for user-space.
+ */
+ int32_t cpu_id;
+ /*
+ * Restartable sequences event_counter field.
+ * Updated by the kernel, and read by user-space with
+ * single-copy atomicity semantics. Aligned on 32-bit.
+ */
+ uint32_t event_counter;
+ } e;
+ /*
+ * On architectures with 64-bit aligned reads, both cpu_id and
+ * event_counter can be read with single-copy atomicity
+ * semantics.
+ */
+ uint64_t v;
+};
+
+/*
+ * struct rseq is aligned on 4 * 8 bytes to ensure it is always
+ * contained within a single cache-line.
+ */
+struct rseq {
+ union rseq_cpu_event u;
+ /*
+ * Restartable sequences rseq_cs field.
+ * Contains NULL when no critical section is active for the
+ * current thread, or holds a pointer to the currently active
+ * struct rseq_cs.
+ * Updated by user-space at the beginning and end of assembly
+ * instruction sequence block, and by the kernel when it
+ * restarts an assembly instruction sequence block. Read by the
+ * kernel with single-copy atomicity semantics. Aligned on
+ * 64-bit.
+ */
+ RSEQ_FIELD_u32_u64(rseq_cs);
+ /*
+ * - RSEQ_DISABLE flag:
+ *
+ * Fallback fast-track flag for single-stepping.
+ * Set by user-space if lack of progress is detected.
+ * Cleared by user-space after rseq finish.
+ * Read by the kernel.
+ * - RSEQ_CS_FLAG_NO_RESTART_ON_PREEMPT
+ * Inhibit instruction sequence block restart and event
+ * counter increment on preemption for this thread.
+ * - RSEQ_CS_FLAG_NO_RESTART_ON_SIGNAL
+ * Inhibit instruction sequence block restart and event
+ * counter increment on signal delivery for this thread.
+ * - RSEQ_CS_FLAG_NO_RESTART_ON_MIGRATE
+ * Inhibit instruction sequence block restart and event
+ * counter increment on migration for this thread.
+ */
+ uint32_t flags;
+} __attribute__((aligned(4 * sizeof(uint64_t))));
+
+#endif /* _UAPI_LINUX_RSEQ_H */
diff --git a/init/Kconfig b/init/Kconfig
index 8514b25db21c..b8aa41bd4f4f 100644
--- a/init/Kconfig
+++ b/init/Kconfig
@@ -1395,6 +1395,19 @@ config MEMBARRIER
If unsure, say Y.
+config RSEQ
+ bool "Enable rseq() system call" if EXPERT
+ default y
+ depends on HAVE_RSEQ
+ help
+ Enable the restartable sequences system call. It provides a
+ user-space cache for the current CPU number value, which
+ speeds up getting the current CPU number from user-space,
+ as well as an ABI to speed up user-space operations on
+ per-CPU data.
+
+ If unsure, say Y.
+
config EMBEDDED
bool "Embedded system"
option allnoconfig_y
diff --git a/kernel/Makefile b/kernel/Makefile
index 4cb8e8b23c6e..5c09592b3b9f 100644
--- a/kernel/Makefile
+++ b/kernel/Makefile
@@ -111,6 +111,7 @@ obj-$(CONFIG_TORTURE_TEST) += torture.o
obj-$(CONFIG_MEMBARRIER) += membarrier.o
obj-$(CONFIG_HAS_IOMEM) += memremap.o
+obj-$(CONFIG_RSEQ) += rseq.o
$(obj)/configs.o: $(obj)/config_data.h
diff --git a/kernel/fork.c b/kernel/fork.c
index b7e9e57b71ea..f311a99fb1d1 100644
--- a/kernel/fork.c
+++ b/kernel/fork.c
@@ -1849,6 +1849,8 @@ static __latent_entropy struct task_struct *copy_process(
*/
copy_seccomp(p);
+ rseq_fork(p, clone_flags);
+
/*
* Process group and session signals need to be delivered to just the
* parent before the fork or both the parent and the child after the
diff --git a/kernel/rseq.c b/kernel/rseq.c
new file mode 100644
index 000000000000..706a83bd885c
--- /dev/null
+++ b/kernel/rseq.c
@@ -0,0 +1,347 @@
+/*
+ * Restartable sequences system call
+ *
+ * Restartable sequences are a lightweight interface that allows
+ * user-level code to be executed atomically relative to scheduler
+ * preemption and signal delivery. Typically used for implementing
+ * per-cpu operations.
+ *
+ * It allows user-space to perform update operations on per-cpu data
+ * without requiring heavy-weight atomic operations.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * Copyright (C) 2015, Google, Inc.,
+ * Paul Turner <pjt@xxxxxxxxxx> and Andrew Hunter <ahh@xxxxxxxxxx>
+ * Copyright (C) 2015-2016, EfficiOS Inc.,
+ * Mathieu Desnoyers <mathieu.desnoyers@xxxxxxxxxxxx>
+ */
+
+#include <linux/sched.h>
+#include <linux/uaccess.h>
+#include <linux/syscalls.h>
+#include <linux/rseq.h>
+#include <linux/types.h>
+#include <asm/ptrace.h>
+
+/*
+ * The restartable sequences mechanism is the overlap of two distinct
+ * restart mechanisms: a sequence counter tracking preemption and signal
+ * delivery for high-level code, and an ip-fixup-based mechanism for the
+ * final assembly instruction sequence.
+ *
+ * A high-level summary of the algorithm to use rseq from user-space is
+ * as follows:
+ *
+ * The high-level code between rseq_start() and rseq_finish() loads the
+ * current value of the sequence counter in rseq_start(), and then it
+ * gets compared with the new current value within the rseq_finish()
+ * restartable instruction sequence. Between rseq_start() and
+ * rseq_finish(), the high-level code can perform operations that do not
+ * have side-effects, such as getting the current CPU number, and
+ * loading from variables.
+ *
+ * Stores are performed at the very end of the restartable sequence
+ * assembly block. Each assembly block within rseq_finish() defines a
+ * "struct rseq_cs" structure which describes the start_ip and
+ * post_commit_ip addresses, as well as the abort_ip address where the
+ * kernel should move the thread instruction pointer if a rseq critical
+ * section assembly block is preempted or if a signal is delivered on
+ * top of a rseq critical section assembly block.
+ *
+ * Detailed algorithm of rseq use:
+ *
+ * rseq_start()
+ *
+ * 0. Userspace loads the current event counter value from the
+ * event_counter field of the registered struct rseq TLS area,
+ *
+ * rseq_finish()
+ *
+ * Steps [1]-[3] (inclusive) need to be a sequence of instructions in
+ * userspace that can handle being moved to the abort_ip between any
+ * of those instructions.
+ *
+ * The abort_ip address needs to be less than start_ip, or
+ * greater-or-equal the post_commit_ip. Step [4] and the failure
+ * code step [F1] need to be at addresses lesser than start_ip, or
+ * greater-or-equal the post_commit_ip.
+ *
+ * [start_ip]
+ * 1. Userspace stores the address of the struct rseq_cs assembly
+ * block descriptor into the rseq_cs field of the registered
+ * struct rseq TLS area. This update is performed through a single
+ * store, followed by a compiler barrier which prevents the
+ * compiler from moving following loads or stores before this
+ * store.
+ *
+ * 2. Userspace tests to see whether the current event counter value
+ * match the value loaded at [0]. Manually jumping to [F1] in case
+ * of a mismatch.
+ *
+ * Note that if we are preempted or interrupted by a signal
+ * after [1] and before post_commit_ip, then the kernel also
+ * performs the comparison performed in [2], and conditionally
+ * clears the rseq_cs field of struct rseq, then jumps us to
+ * abort_ip.
+ *
+ * 3. Userspace critical section final instruction before
+ * post_commit_ip is the commit. The critical section is
+ * self-terminating.
+ * [post_commit_ip]
+ *
+ * 4. Userspace clears the rseq_cs field of the struct rseq
+ * TLS area.
+ *
+ * 5. Return true.
+ *
+ * On failure at [2]:
+ *
+ * F1. Userspace clears the rseq_cs field of the struct rseq
+ * TLS area. Followed by step [F2].
+ *
+ * [abort_ip]
+ * F2. Return false.
+ */
+
+/*
+ * The rseq_event_counter allow user-space to detect preemption and
+ * signal delivery. It increments at least once before returning to
+ * user-space if a thread is preempted or has a signal delivered. It is
+ * not meant to be an exact counter of such events.
+ *
+ * Overflow of the event counter is not a problem in practice. It
+ * increments at most once between each user-space thread instruction
+ * executed, so we would need a thread to execute 2^32 instructions or
+ * more between rseq_start() and rseq_finish(), while single-stepping,
+ * for this to be an issue.
+ *
+ * On 64-bit architectures, both cpu_id and event_counter can be updated
+ * with a single 64-bit store. On 32-bit architectures, __put_user() is
+ * expected to perform two 32-bit single-copy stores to guarantee
+ * single-copy atomicity semantics for other threads.
+ */
+static bool rseq_update_cpu_id_event_counter(struct task_struct *t,
+ bool inc_event_counter)
+{
+ union rseq_cpu_event u;
+
+ u.e.cpu_id = raw_smp_processor_id();
+ u.e.event_counter = inc_event_counter ? ++t->rseq_event_counter :
+ t->rseq_event_counter;
+ if (__put_user(u.v, &t->rseq->u.v))
+ return false;
+ return true;
+}
+
+static bool rseq_get_rseq_cs(struct task_struct *t,
+ void __user **start_ip,
+ void __user **post_commit_ip,
+ void __user **abort_ip,
+ uint32_t *cs_flags)
+{
+ unsigned long ptr;
+ struct rseq_cs __user *urseq_cs;
+ struct rseq_cs rseq_cs;
+
+ if (__get_user(ptr, &t->rseq->rseq_cs))
+ return false;
+ if (!ptr)
+ return true;
+ urseq_cs = (struct rseq_cs __user *)ptr;
+ if (copy_from_user(&rseq_cs, urseq_cs, sizeof(rseq_cs)))
+ return false;
+ /*
+ * We need to clear rseq_cs upon entry into a signal handler
+ * nested on top of a rseq assembly block, so the signal handler
+ * will not be fixed up if itself interrupted by a nested signal
+ * handler or preempted. We also need to clear rseq_cs if we
+ * preempt or deliver a signal on top of code outside of the
+ * rseq assembly block, to ensure that a following preemption or
+ * signal delivery will not try to perform a fixup needlessly.
+ */
+ if (clear_user(&t->rseq->rseq_cs, sizeof(t->rseq->rseq_cs)))
+ return false;
+ *start_ip = (void __user *)rseq_cs.start_ip;
+ *post_commit_ip = (void __user *)rseq_cs.post_commit_ip;
+ *abort_ip = (void __user *)rseq_cs.abort_ip;
+ *cs_flags = rseq_cs.flags;
+ return true;
+}
+
+static int rseq_need_restart(struct task_struct *t, uint32_t cs_flags)
+{
+ bool need_restart = false;
+ uint32_t flags;
+
+ /* Get thread flags. */
+ if (__get_user(flags, &t->rseq->flags))
+ return -EFAULT;
+
+ /* Take into account critical section flags. */
+ flags |= cs_flags;
+
+ /*
+ * Restart on signal can only be inhibited when restart on
+ * preempt and restart on migrate are inhibited too. Otherwise,
+ * a preempted signal handler could fail to restart the prior
+ * execution context on sigreturn.
+ */
+ if (flags & RSEQ_CS_FLAG_NO_RESTART_ON_SIGNAL) {
+ if (!(flags & RSEQ_CS_FLAG_NO_RESTART_ON_MIGRATE))
+ return -EINVAL;
+ if (!(flags & RSEQ_CS_FLAG_NO_RESTART_ON_PREEMPT))
+ return -EINVAL;
+ }
+ if (t->rseq_migrate
+ && !(flags & RSEQ_CS_FLAG_NO_RESTART_ON_MIGRATE))
+ need_restart = true;
+ else if (t->rseq_preempt
+ && !(flags & RSEQ_CS_FLAG_NO_RESTART_ON_PREEMPT))
+ need_restart = true;
+ else if (t->rseq_signal
+ && !(flags & RSEQ_CS_FLAG_NO_RESTART_ON_SIGNAL))
+ need_restart = true;
+
+ t->rseq_preempt = false;
+ t->rseq_signal = false;
+ t->rseq_migrate = false;
+ if (need_restart)
+ return 1;
+ return 0;
+}
+
+static int rseq_ip_fixup(struct pt_regs *regs)
+{
+ struct task_struct *t = current;
+ void __user *start_ip = NULL;
+ void __user *post_commit_ip = NULL;
+ void __user *abort_ip = NULL;
+ uint32_t cs_flags = 0;
+ int ret;
+
+ ret = rseq_get_rseq_cs(t, &start_ip, &post_commit_ip, &abort_ip,
+ &cs_flags);
+ if (!ret)
+ return -EFAULT;
+
+ ret = rseq_need_restart(t, cs_flags);
+ if (ret < 0)
+ return -EFAULT;
+ if (!ret)
+ return 0;
+
+ /* Handle potentially not being within a critical section. */
+ if ((void __user *)instruction_pointer(regs) >= post_commit_ip ||
+ (void __user *)instruction_pointer(regs) < start_ip)
+ return 1;
+
+ /*
+ * We set this after potentially failing in
+ * clear_user so that the signal arrives at the
+ * faulting rip.
+ */
+ instruction_pointer_set(regs, (unsigned long)abort_ip);
+ return 1;
+}
+
+/*
+ * This resume handler should always be executed between any of:
+ * - preemption,
+ * - signal delivery,
+ * and return to user-space.
+ *
+ * This is how we can ensure that the entire rseq critical section,
+ * consisting of both the C part and the assembly instruction sequence,
+ * will issue the commit instruction only if executed atomically with
+ * respect to other threads scheduled on the same CPU, and with respect
+ * to signal handlers.
+ */
+void __rseq_handle_notify_resume(struct pt_regs *regs)
+{
+ struct task_struct *t = current;
+ int ret;
+
+ if (unlikely(t->flags & PF_EXITING))
+ return;
+ if (unlikely(!access_ok(VERIFY_WRITE, t->rseq, sizeof(*t->rseq))))
+ goto error;
+ ret = rseq_ip_fixup(regs);
+ if (unlikely(ret < 0))
+ goto error;
+ if (unlikely(!rseq_update_cpu_id_event_counter(t, ret)))
+ goto error;
+ return;
+
+error:
+ force_sig(SIGSEGV, t);
+}
+
+/*
+ * sys_rseq - setup restartable sequences for caller thread.
+ */
+SYSCALL_DEFINE2(rseq, struct rseq __user *, rseq, int, flags)
+{
+ if (!rseq) {
+ /* Unregister rseq for current thread. */
+ if (unlikely(flags & ~RSEQ_FORCE_UNREGISTER))
+ return -EINVAL;
+ if (flags & RSEQ_FORCE_UNREGISTER) {
+ current->rseq = NULL;
+ current->rseq_refcount = 0;
+ return 0;
+ }
+ if (!current->rseq_refcount)
+ return -ENOENT;
+ if (!--current->rseq_refcount)
+ current->rseq = NULL;
+ return 0;
+ }
+
+ if (unlikely(flags))
+ return -EINVAL;
+
+ if (current->rseq) {
+ /*
+ * If rseq is already registered, check whether
+ * the provided address differs from the prior
+ * one.
+ */
+ BUG_ON(!current->rseq_refcount);
+ if (current->rseq != rseq)
+ return -EBUSY;
+ if (current->rseq_refcount == UINT_MAX)
+ return -EOVERFLOW;
+ current->rseq_refcount++;
+ } else {
+ /*
+ * If there was no rseq previously registered,
+ * we need to ensure the provided rseq is
+ * properly aligned and valid.
+ */
+ BUG_ON(current->rseq_refcount);
+ if (!IS_ALIGNED((unsigned long)rseq, __alignof__(*rseq)))
+ return -EINVAL;
+ if (!access_ok(VERIFY_WRITE, rseq, sizeof(*rseq)))
+ return -EFAULT;
+ current->rseq = rseq;
+ current->rseq_refcount = 1;
+ /*
+ * If rseq was previously inactive, and has just
+ * been registered, ensure the cpu_id and
+ * event_counter fields are updated before
+ * returning to user-space.
+ */
+ rseq_set_notify_resume(current);
+ }
+
+ return 0;
+}
diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index 0869b20fba81..12da0f771d73 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -1170,6 +1170,8 @@ void set_task_cpu(struct task_struct *p, unsigned int new_cpu)
#endif
#endif
+ rseq_migrate(p);
+
trace_sched_migrate_task(p, new_cpu);
if (task_cpu(p) != new_cpu) {
@@ -2572,6 +2574,7 @@ prepare_task_switch(struct rq *rq, struct task_struct *prev,
{
sched_info_switch(rq, prev, next);
perf_event_task_sched_out(prev, next);
+ rseq_sched_out(prev);
fire_sched_out_preempt_notifiers(prev, next);
prepare_lock_switch(rq, next);
prepare_arch_switch(next);
@@ -3322,6 +3325,7 @@ static void __sched notrace __schedule(bool preempt)
clear_preempt_need_resched();
if (likely(prev != next)) {
+ rseq_preempt(prev);
rq->nr_switches++;
rq->curr = next;
++*switch_count;
diff --git a/kernel/sys_ni.c b/kernel/sys_ni.c
index 8acef8576ce9..c7b366ccf39c 100644
--- a/kernel/sys_ni.c
+++ b/kernel/sys_ni.c
@@ -258,3 +258,6 @@ cond_syscall(sys_membarrier);
cond_syscall(sys_pkey_mprotect);
cond_syscall(sys_pkey_alloc);
cond_syscall(sys_pkey_free);
+
+/* restartable sequence */
+cond_syscall(sys_rseq);
--
2.11.0