linux/fs/eventfd.c
Paolo Bonzini a484c3dd94 eventfd: document lockless access in eventfd_poll
Since commit e22553e2a2 ("eventfd: don't take the spinlock in
eventfd_poll", 2015-02-17), eventfd is reading ctx->count outside
ctx->wqh.lock.

However, things aren't as simple as the read barrier in eventfd_poll
would suggest.  In fact, the read barrier, besides lacking a comment, is
not paired in any obvious manner with another read barrier, and it is
pointless because it is sitting between a write (deep in poll_wait) and
the read of ctx->count.  The read barrier is acting just as a compiler
barrier, for which we can use READ_ONCE instead.  This is what the code
change in this patch does.

The documentation change is just as important, however.  The question,
posed by Andrea Arcangeli, is then why the thing is safe on
architectures where spin_unlock does not imply a store-load memory
barrier.  The answer is that it's safe because writes of ctx->count use
the same lock as poll_wait, and hence an acquire barrier implicit in
poll_wait provides the necessary synchronization between eventfd_poll
and callers of wake_up_locked_poll.  This is sort of mentioned in the
commit message with respect to eventfd_ctx_read ("eventfd_read is
similar, it will do a single decrement with the lock held") but it
applies to all other callers too.  It's tricky enough that it should be
documented in the code.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Andrea Arcangeli <aarcange@redhat.com>
Cc: Chris Mason <clm@fb.com>
Cc: Davide Libenzi <davidel@xmailserver.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-22 15:36:02 -07:00

488 lines
13 KiB
C

/*
* fs/eventfd.c
*
* Copyright (C) 2007 Davide Libenzi <davidel@xmailserver.org>
*
*/
#include <linux/file.h>
#include <linux/poll.h>
#include <linux/init.h>
#include <linux/fs.h>
#include <linux/sched.h>
#include <linux/kernel.h>
#include <linux/slab.h>
#include <linux/list.h>
#include <linux/spinlock.h>
#include <linux/anon_inodes.h>
#include <linux/syscalls.h>
#include <linux/export.h>
#include <linux/kref.h>
#include <linux/eventfd.h>
#include <linux/proc_fs.h>
#include <linux/seq_file.h>
struct eventfd_ctx {
struct kref kref;
wait_queue_head_t wqh;
/*
* Every time that a write(2) is performed on an eventfd, the
* value of the __u64 being written is added to "count" and a
* wakeup is performed on "wqh". A read(2) will return the "count"
* value to userspace, and will reset "count" to zero. The kernel
* side eventfd_signal() also, adds to the "count" counter and
* issue a wakeup.
*/
__u64 count;
unsigned int flags;
};
/**
* eventfd_signal - Adds @n to the eventfd counter.
* @ctx: [in] Pointer to the eventfd context.
* @n: [in] Value of the counter to be added to the eventfd internal counter.
* The value cannot be negative.
*
* This function is supposed to be called by the kernel in paths that do not
* allow sleeping. In this function we allow the counter to reach the ULLONG_MAX
* value, and we signal this as overflow condition by returning a POLLERR
* to poll(2).
*
* Returns the amount by which the counter was incremented. This will be less
* than @n if the counter has overflowed.
*/
__u64 eventfd_signal(struct eventfd_ctx *ctx, __u64 n)
{
unsigned long flags;
spin_lock_irqsave(&ctx->wqh.lock, flags);
if (ULLONG_MAX - ctx->count < n)
n = ULLONG_MAX - ctx->count;
ctx->count += n;
if (waitqueue_active(&ctx->wqh))
wake_up_locked_poll(&ctx->wqh, POLLIN);
spin_unlock_irqrestore(&ctx->wqh.lock, flags);
return n;
}
EXPORT_SYMBOL_GPL(eventfd_signal);
static void eventfd_free_ctx(struct eventfd_ctx *ctx)
{
kfree(ctx);
}
static void eventfd_free(struct kref *kref)
{
struct eventfd_ctx *ctx = container_of(kref, struct eventfd_ctx, kref);
eventfd_free_ctx(ctx);
}
/**
* eventfd_ctx_get - Acquires a reference to the internal eventfd context.
* @ctx: [in] Pointer to the eventfd context.
*
* Returns: In case of success, returns a pointer to the eventfd context.
*/
struct eventfd_ctx *eventfd_ctx_get(struct eventfd_ctx *ctx)
{
kref_get(&ctx->kref);
return ctx;
}
EXPORT_SYMBOL_GPL(eventfd_ctx_get);
/**
* eventfd_ctx_put - Releases a reference to the internal eventfd context.
* @ctx: [in] Pointer to eventfd context.
*
* The eventfd context reference must have been previously acquired either
* with eventfd_ctx_get() or eventfd_ctx_fdget().
*/
void eventfd_ctx_put(struct eventfd_ctx *ctx)
{
kref_put(&ctx->kref, eventfd_free);
}
EXPORT_SYMBOL_GPL(eventfd_ctx_put);
static int eventfd_release(struct inode *inode, struct file *file)
{
struct eventfd_ctx *ctx = file->private_data;
wake_up_poll(&ctx->wqh, POLLHUP);
eventfd_ctx_put(ctx);
return 0;
}
static unsigned int eventfd_poll(struct file *file, poll_table *wait)
{
struct eventfd_ctx *ctx = file->private_data;
unsigned int events = 0;
u64 count;
poll_wait(file, &ctx->wqh, wait);
/*
* All writes to ctx->count occur within ctx->wqh.lock. This read
* can be done outside ctx->wqh.lock because we know that poll_wait
* takes that lock (through add_wait_queue) if our caller will sleep.
*
* The read _can_ therefore seep into add_wait_queue's critical
* section, but cannot move above it! add_wait_queue's spin_lock acts
* as an acquire barrier and ensures that the read be ordered properly
* against the writes. The following CAN happen and is safe:
*
* poll write
* ----------------- ------------
* lock ctx->wqh.lock (in poll_wait)
* count = ctx->count
* __add_wait_queue
* unlock ctx->wqh.lock
* lock ctx->qwh.lock
* ctx->count += n
* if (waitqueue_active)
* wake_up_locked_poll
* unlock ctx->qwh.lock
* eventfd_poll returns 0
*
* but the following, which would miss a wakeup, cannot happen:
*
* poll write
* ----------------- ------------
* count = ctx->count (INVALID!)
* lock ctx->qwh.lock
* ctx->count += n
* **waitqueue_active is false**
* **no wake_up_locked_poll!**
* unlock ctx->qwh.lock
* lock ctx->wqh.lock (in poll_wait)
* __add_wait_queue
* unlock ctx->wqh.lock
* eventfd_poll returns 0
*/
count = READ_ONCE(ctx->count);
if (count > 0)
events |= POLLIN;
if (count == ULLONG_MAX)
events |= POLLERR;
if (ULLONG_MAX - 1 > count)
events |= POLLOUT;
return events;
}
static void eventfd_ctx_do_read(struct eventfd_ctx *ctx, __u64 *cnt)
{
*cnt = (ctx->flags & EFD_SEMAPHORE) ? 1 : ctx->count;
ctx->count -= *cnt;
}
/**
* eventfd_ctx_remove_wait_queue - Read the current counter and removes wait queue.
* @ctx: [in] Pointer to eventfd context.
* @wait: [in] Wait queue to be removed.
* @cnt: [out] Pointer to the 64-bit counter value.
*
* Returns %0 if successful, or the following error codes:
*
* -EAGAIN : The operation would have blocked.
*
* This is used to atomically remove a wait queue entry from the eventfd wait
* queue head, and read/reset the counter value.
*/
int eventfd_ctx_remove_wait_queue(struct eventfd_ctx *ctx, wait_queue_t *wait,
__u64 *cnt)
{
unsigned long flags;
spin_lock_irqsave(&ctx->wqh.lock, flags);
eventfd_ctx_do_read(ctx, cnt);
__remove_wait_queue(&ctx->wqh, wait);
if (*cnt != 0 && waitqueue_active(&ctx->wqh))
wake_up_locked_poll(&ctx->wqh, POLLOUT);
spin_unlock_irqrestore(&ctx->wqh.lock, flags);
return *cnt != 0 ? 0 : -EAGAIN;
}
EXPORT_SYMBOL_GPL(eventfd_ctx_remove_wait_queue);
/**
* eventfd_ctx_read - Reads the eventfd counter or wait if it is zero.
* @ctx: [in] Pointer to eventfd context.
* @no_wait: [in] Different from zero if the operation should not block.
* @cnt: [out] Pointer to the 64-bit counter value.
*
* Returns %0 if successful, or the following error codes:
*
* -EAGAIN : The operation would have blocked but @no_wait was non-zero.
* -ERESTARTSYS : A signal interrupted the wait operation.
*
* If @no_wait is zero, the function might sleep until the eventfd internal
* counter becomes greater than zero.
*/
ssize_t eventfd_ctx_read(struct eventfd_ctx *ctx, int no_wait, __u64 *cnt)
{
ssize_t res;
DECLARE_WAITQUEUE(wait, current);
spin_lock_irq(&ctx->wqh.lock);
*cnt = 0;
res = -EAGAIN;
if (ctx->count > 0)
res = 0;
else if (!no_wait) {
__add_wait_queue(&ctx->wqh, &wait);
for (;;) {
set_current_state(TASK_INTERRUPTIBLE);
if (ctx->count > 0) {
res = 0;
break;
}
if (signal_pending(current)) {
res = -ERESTARTSYS;
break;
}
spin_unlock_irq(&ctx->wqh.lock);
schedule();
spin_lock_irq(&ctx->wqh.lock);
}
__remove_wait_queue(&ctx->wqh, &wait);
__set_current_state(TASK_RUNNING);
}
if (likely(res == 0)) {
eventfd_ctx_do_read(ctx, cnt);
if (waitqueue_active(&ctx->wqh))
wake_up_locked_poll(&ctx->wqh, POLLOUT);
}
spin_unlock_irq(&ctx->wqh.lock);
return res;
}
EXPORT_SYMBOL_GPL(eventfd_ctx_read);
static ssize_t eventfd_read(struct file *file, char __user *buf, size_t count,
loff_t *ppos)
{
struct eventfd_ctx *ctx = file->private_data;
ssize_t res;
__u64 cnt;
if (count < sizeof(cnt))
return -EINVAL;
res = eventfd_ctx_read(ctx, file->f_flags & O_NONBLOCK, &cnt);
if (res < 0)
return res;
return put_user(cnt, (__u64 __user *) buf) ? -EFAULT : sizeof(cnt);
}
static ssize_t eventfd_write(struct file *file, const char __user *buf, size_t count,
loff_t *ppos)
{
struct eventfd_ctx *ctx = file->private_data;
ssize_t res;
__u64 ucnt;
DECLARE_WAITQUEUE(wait, current);
if (count < sizeof(ucnt))
return -EINVAL;
if (copy_from_user(&ucnt, buf, sizeof(ucnt)))
return -EFAULT;
if (ucnt == ULLONG_MAX)
return -EINVAL;
spin_lock_irq(&ctx->wqh.lock);
res = -EAGAIN;
if (ULLONG_MAX - ctx->count > ucnt)
res = sizeof(ucnt);
else if (!(file->f_flags & O_NONBLOCK)) {
__add_wait_queue(&ctx->wqh, &wait);
for (res = 0;;) {
set_current_state(TASK_INTERRUPTIBLE);
if (ULLONG_MAX - ctx->count > ucnt) {
res = sizeof(ucnt);
break;
}
if (signal_pending(current)) {
res = -ERESTARTSYS;
break;
}
spin_unlock_irq(&ctx->wqh.lock);
schedule();
spin_lock_irq(&ctx->wqh.lock);
}
__remove_wait_queue(&ctx->wqh, &wait);
__set_current_state(TASK_RUNNING);
}
if (likely(res > 0)) {
ctx->count += ucnt;
if (waitqueue_active(&ctx->wqh))
wake_up_locked_poll(&ctx->wqh, POLLIN);
}
spin_unlock_irq(&ctx->wqh.lock);
return res;
}
#ifdef CONFIG_PROC_FS
static void eventfd_show_fdinfo(struct seq_file *m, struct file *f)
{
struct eventfd_ctx *ctx = f->private_data;
spin_lock_irq(&ctx->wqh.lock);
seq_printf(m, "eventfd-count: %16llx\n",
(unsigned long long)ctx->count);
spin_unlock_irq(&ctx->wqh.lock);
}
#endif
static const struct file_operations eventfd_fops = {
#ifdef CONFIG_PROC_FS
.show_fdinfo = eventfd_show_fdinfo,
#endif
.release = eventfd_release,
.poll = eventfd_poll,
.read = eventfd_read,
.write = eventfd_write,
.llseek = noop_llseek,
};
/**
* eventfd_fget - Acquire a reference of an eventfd file descriptor.
* @fd: [in] Eventfd file descriptor.
*
* Returns a pointer to the eventfd file structure in case of success, or the
* following error pointer:
*
* -EBADF : Invalid @fd file descriptor.
* -EINVAL : The @fd file descriptor is not an eventfd file.
*/
struct file *eventfd_fget(int fd)
{
struct file *file;
file = fget(fd);
if (!file)
return ERR_PTR(-EBADF);
if (file->f_op != &eventfd_fops) {
fput(file);
return ERR_PTR(-EINVAL);
}
return file;
}
EXPORT_SYMBOL_GPL(eventfd_fget);
/**
* eventfd_ctx_fdget - Acquires a reference to the internal eventfd context.
* @fd: [in] Eventfd file descriptor.
*
* Returns a pointer to the internal eventfd context, otherwise the error
* pointers returned by the following functions:
*
* eventfd_fget
*/
struct eventfd_ctx *eventfd_ctx_fdget(int fd)
{
struct eventfd_ctx *ctx;
struct fd f = fdget(fd);
if (!f.file)
return ERR_PTR(-EBADF);
ctx = eventfd_ctx_fileget(f.file);
fdput(f);
return ctx;
}
EXPORT_SYMBOL_GPL(eventfd_ctx_fdget);
/**
* eventfd_ctx_fileget - Acquires a reference to the internal eventfd context.
* @file: [in] Eventfd file pointer.
*
* Returns a pointer to the internal eventfd context, otherwise the error
* pointer:
*
* -EINVAL : The @fd file descriptor is not an eventfd file.
*/
struct eventfd_ctx *eventfd_ctx_fileget(struct file *file)
{
if (file->f_op != &eventfd_fops)
return ERR_PTR(-EINVAL);
return eventfd_ctx_get(file->private_data);
}
EXPORT_SYMBOL_GPL(eventfd_ctx_fileget);
/**
* eventfd_file_create - Creates an eventfd file pointer.
* @count: Initial eventfd counter value.
* @flags: Flags for the eventfd file.
*
* This function creates an eventfd file pointer, w/out installing it into
* the fd table. This is useful when the eventfd file is used during the
* initialization of data structures that require extra setup after the eventfd
* creation. So the eventfd creation is split into the file pointer creation
* phase, and the file descriptor installation phase.
* In this way races with userspace closing the newly installed file descriptor
* can be avoided.
* Returns an eventfd file pointer, or a proper error pointer.
*/
struct file *eventfd_file_create(unsigned int count, int flags)
{
struct file *file;
struct eventfd_ctx *ctx;
/* Check the EFD_* constants for consistency. */
BUILD_BUG_ON(EFD_CLOEXEC != O_CLOEXEC);
BUILD_BUG_ON(EFD_NONBLOCK != O_NONBLOCK);
if (flags & ~EFD_FLAGS_SET)
return ERR_PTR(-EINVAL);
ctx = kmalloc(sizeof(*ctx), GFP_KERNEL);
if (!ctx)
return ERR_PTR(-ENOMEM);
kref_init(&ctx->kref);
init_waitqueue_head(&ctx->wqh);
ctx->count = count;
ctx->flags = flags;
file = anon_inode_getfile("[eventfd]", &eventfd_fops, ctx,
O_RDWR | (flags & EFD_SHARED_FCNTL_FLAGS));
if (IS_ERR(file))
eventfd_free_ctx(ctx);
return file;
}
SYSCALL_DEFINE2(eventfd2, unsigned int, count, int, flags)
{
int fd, error;
struct file *file;
error = get_unused_fd_flags(flags & EFD_SHARED_FCNTL_FLAGS);
if (error < 0)
return error;
fd = error;
file = eventfd_file_create(count, flags);
if (IS_ERR(file)) {
error = PTR_ERR(file);
goto err_put_unused_fd;
}
fd_install(fd, file);
return fd;
err_put_unused_fd:
put_unused_fd(fd);
return error;
}
SYSCALL_DEFINE1(eventfd, unsigned int, count)
{
return sys_eventfd2(count, 0);
}