Re: [PATCH 0/2] eventfd: new EFD_STATE flag

From: Davide Libenzi <davidel@xmailserver.org>
To: Avi Kivity <avi@redhat.com>
Cc: "Michael S. Tsirkin" <mst@redhat.com>,
	gleb@redhat.com, kvm@vger.kernel.org,
	Linux Kernel Mailing List <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH 0/2] eventfd: new EFD_STATE flag
Date: Wed, 26 Aug 2009 16:30:23 -0700 (PDT)	[thread overview]
Message-ID: <alpine.DEB.2.00.0908261626410.7780@makko.or.mcafeemobile.com> (raw)
In-Reply-To: <alpine.DEB.2.00.0908261242530.7780@makko.or.mcafeemobile.com>

On Wed, 26 Aug 2009, Davide Libenzi wrote:

> > I see no kernel equivalent to read(), but that's easily done.
> 
> Adding an in-kernel read based on "ctx", that is no problem at all.

Something like the untested below.
I had thought you said the eventfd readers where in userspace, but I might 
have misunderstood you.



- Davide



---
 fs/eventfd.c            |   51 +++++++++++++++++++++++++++++++++---------------
 include/linux/eventfd.h |    7 ++++++
 2 files changed, 43 insertions(+), 15 deletions(-)

Index: linux-2.6.mod/fs/eventfd.c
===================================================================

--- linux-2.6.mod.orig/fs/eventfd.c	2009-08-26 15:58:03.000000000 -0700
+++ linux-2.6.mod/fs/eventfd.c	2009-08-26 16:20:03.000000000 -0700
@@ -130,26 +130,33 @@ static unsigned int eventfd_poll(struct 
 	return events;
 }
 
-static ssize_t eventfd_read(struct file *file, char __user *buf, size_t count,
-			    loff_t *ppos)
+/**
+ * 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 64bit conter value.
+ *
+ * Returns zero if successful, or the following error codes:
+ *
+ * -EAGAIN      : The operation would have blocked but @no_wait was nonzero.
+ * -ERESTARTSYS : A signal interrupted the wait operation.
+ */
+ssize_t eventfd_ctx_read(struct eventfd_ctx *ctx, int no_wait, __u64 *cnt)
 {
-	struct eventfd_ctx *ctx = file->private_data;
 	ssize_t res;
-	__u64 ucnt = 0;
 	DECLARE_WAITQUEUE(wait, current);
 
-	if (count < sizeof(ucnt))
-		return -EINVAL;
 	spin_lock_irq(&ctx->wqh.lock);
+	*cnt = 0;
 	res = -EAGAIN;
 	if (ctx->count > 0)
-		res = sizeof(ucnt);
-	else if (!(file->f_flags & O_NONBLOCK)) {
+		res = 0;
+	else if (!no_wait) {
 		__add_wait_queue(&ctx->wqh, &wait);
-		for (res = 0;;) {
+		for (;;) {
 			set_current_state(TASK_INTERRUPTIBLE);
 			if (ctx->count > 0) {
-				res = sizeof(ucnt);
+				res = 0;
 				break;
 			}
 			if (signal_pending(current)) {
@@ -163,18 +170,32 @@ static ssize_t eventfd_read(struct file 
 		__remove_wait_queue(&ctx->wqh, &wait);
 		__set_current_state(TASK_RUNNING);
 	}
-	if (likely(res > 0)) {
-		ucnt = (ctx->flags & EFD_SEMAPHORE) ? 1 : ctx->count;
-		ctx->count -= ucnt;
+	if (likely(res == 0)) {
+		*cnt = (ctx->flags & EFD_SEMAPHORE) ? 1 : ctx->count;
+		ctx->count -= *cnt;
 		if (waitqueue_active(&ctx->wqh))
 			wake_up_locked_poll(&ctx->wqh, POLLOUT);
 	}
 	spin_unlock_irq(&ctx->wqh.lock);
-	if (res > 0 && put_user(ucnt, (__u64 __user *) buf))
-		return -EFAULT;
 
 	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;
+	if ((res = eventfd_ctx_read(ctx, file->f_flags & O_NONBLOCK, &cnt)) < 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)
Index: linux-2.6.mod/include/linux/eventfd.h
===================================================================
--- linux-2.6.mod.orig/include/linux/eventfd.h	2009-08-26 15:58:03.000000000 -0700
+++ linux-2.6.mod/include/linux/eventfd.h	2009-08-26 16:17:01.000000000 -0700
@@ -33,6 +33,7 @@ struct file *eventfd_fget(int fd);
 struct eventfd_ctx *eventfd_ctx_fdget(int fd);
 struct eventfd_ctx *eventfd_ctx_fileget(struct file *file);
 int eventfd_signal(struct eventfd_ctx *ctx, int n);
+ssize_t eventfd_ctx_read(struct eventfd_ctx *ctx, int no_wait, __u64 *cnt);
 
 #else /* CONFIG_EVENTFD */
 
@@ -55,6 +56,12 @@ static inline void eventfd_ctx_put(struc
 
 }
 
+static inline ssize_t eventfd_ctx_read(struct eventfd_ctx *ctx, int no_wait,
+				       __u64 *cnt)
+{
+	return -ENOSYS;
+}
+
 #endif
 
 #endif /* _LINUX_EVENTFD_H */
