All of lore.kernel.org
 help / color / mirror / Atom feed
From: Rusty Russell <rusty@rustcorp.com.au>
To: linux-kernel@vger.kernel.org
Cc: netdev@vger.kernel.org,
	virtualization@lists.linux-foundation.org,
	Max Krasnyansky <maxk@qualcomm.com>
Subject: [PATCH RFC 2/5] vringfd base/offset
Date: Sat, 5 Apr 2008 22:04:28 +1000	[thread overview]
Message-ID: <200804052204.28518.rusty@rustcorp.com.au> (raw)
In-Reply-To: <200804052202.09157.rusty@rustcorp.com.au>

It turns out the lguest (and possibly kvm) want the addresses in the
ring buffer to only cover a certain part of memory, and be offset.

It makes sense that this be an ioctl.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

diff -r 08fb00b8acab Documentation/ioctl-number.txt
--- a/Documentation/ioctl-number.txt	Sat Apr 05 21:31:40 2008 +1100
+++ b/Documentation/ioctl-number.txt	Sat Apr 05 22:00:10 2008 +1100
@@ -183,6 +183,7 @@ 0xAC	00-1F	linux/raw.h
 0xAC	00-1F	linux/raw.h
 0xAD	00	Netfilter device	in development:
 					<mailto:rusty@rustcorp.com.au>	
+0xAE	00-01	linux/vring.h
 0xB0	all	RATIO devices		in development:
 					<mailto:vgo@ratio.de>
 0xB1	00-1F	PPPoX			<mailto:mostrows@styx.uwaterloo.ca>
diff -r 08fb00b8acab fs/vring.c
--- a/fs/vring.c	Sat Apr 05 21:31:40 2008 +1100
+++ b/fs/vring.c	Sat Apr 05 22:00:10 2008 +1100
@@ -38,6 +38,8 @@ struct vring_info
 	u16 mask;
 	u16 __user *last_used;
 	u16 last_avail;
+
+	unsigned long base, limit;
 
 	const struct vring_ops *ops;
 	void *ops_data;
@@ -120,10 +122,30 @@ static int vring_release(struct inode *i
 	return 0;
 }
 
+static int vring_ioctl(struct inode *in, struct file *filp,
+		       unsigned int cmd, unsigned long arg)
+{
+	struct vring_info *vr = filp->private_data;
+
+	switch (cmd) {
+	case VRINGSETBASE:
+		vr->base = arg;
+		break;
+	case VRINGSETLIMIT:
+		vr->limit = arg;
+		break;
+	default:
+		return -ENOTTY;
+	}
+
+	return 0;
+}
+
 static const struct file_operations vring_fops = {
 	.release	= vring_release,
 	.write		= vring_write,
 	.poll		= vring_poll,
+	.ioctl		= vring_ioctl,
 };
 
 asmlinkage long sys_vringfd(void __user *addr,
@@ -166,6 +188,8 @@ asmlinkage long sys_vringfd(void __user 
 	vr->mask = num_descs - 1;
 	vr->ops = NULL;
 	vr->used = NULL;
+	vr->limit = -1UL;
+	vr->base = 0;
 
 	err = get_user(vr->last_avail, &vr->ring.avail->idx);
 	if (err)
@@ -208,12 +232,15 @@ int vring_get_buffer(struct vring_info *
 		out_len = &dummy;
 
 	*in_len = *out_len = 0;
-	
-	if (unlikely(get_user(head, &vr->ring.avail->ring[head]) != 0))
+
+	if (unlikely(get_user(head, &vr->ring.avail->ring[vr->last_avail
+							  % vr->ring.num])))
 		return -EFAULT;
 
 	i = head;
 	do {
+		void __user *base;
+
 		if (unlikely(i >= vr->ring.num)) {
 			pr_debug("vring: bad index: %u\n", i);
 			return -EINVAL;
@@ -222,24 +249,38 @@ int vring_get_buffer(struct vring_info *
 		if (copy_from_user(&d, &vr->ring.desc[i], sizeof(d)) != 0)
 			return -EFAULT;
 
+		if (d.addr + d.len > vr->limit || (d.addr + d.len < d.addr)) {
+			pr_debug("vring: bad addr/len: %u@%p\n", 
+				 d.len, (void *)(unsigned long)d.addr);
+			return -EINVAL;
+		}
+
+		base = (void __user *)(unsigned long)d.addr + vr->base;
+
 		if (d.flags & VRING_DESC_F_WRITE) {
 			/* Check for length and iovec overflows */
-			if (!num_in)
+			if (!num_in) {
+				pr_debug("vring: writable desc %u in ring %p\n",
+				         i, vr->ring.desc);
 				return -EINVAL;
+			}
 			if (in == *num_in || *in_len + d.len < *in_len)
 				return -E2BIG;
 			in_iov[in].iov_len = d.len;
 			*in_len += d.len;
-			in_iov[in].iov_base = (void __user*)(long)d.addr;
+			in_iov[in].iov_base = base;
 			in++;
 		} else {
-			if (!num_out)
+			if (!num_out) {
+				pr_debug("vring: readable desc %u in ring %p\n",
+				         i, vr->ring.desc);
 				return -EINVAL;
+			}
 			if (out == *num_out || *out_len + d.len < *out_len)
 				return -E2BIG;
 			out_iov[out].iov_len = d.len;
 			*out_len += d.len;
-			out_iov[out].iov_base = (void __user*)(long)d.addr;
+			out_iov[out].iov_base = base;
 			out++;
 		}
 
diff -r 08fb00b8acab include/linux/vring.h
--- a/include/linux/vring.h	Sat Apr 05 21:31:40 2008 +1100
+++ b/include/linux/vring.h	Sat Apr 05 22:00:10 2008 +1100
@@ -18,7 +18,13 @@
  */
 #ifndef _LINUX_VRING_H
 #define _LINUX_VRING_H
+#include <linux/types.h>
 
+/* Ioctl defines, as in "ioctls are AEgly". */
+#define VRINGSETBASE	_IO(0xAE, 0)
+#define VRINGSETLIMIT	_IO(0xAE, 1)
+
+#ifdef __KERNEL__
 /* All members are optional */
 struct vring_ops
 {
@@ -51,4 +57,6 @@ void vring_used_buffer_atomic(struct vri
 void vring_used_buffer_atomic(struct vring_info *vr, int id, u32 len);
 
 void vring_wake(struct vring_info *vr);
+#endif /* __KERNEL__ */
+
 #endif /* _LINUX_VRING_H */

  reply	other threads:[~2008-04-05 12:05 UTC|newest]

Thread overview: 54+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2008-04-05 12:02 [PATCH RFC 1/5] vringfd syscall Rusty Russell
2008-04-05 12:04 ` Rusty Russell [this message]
2008-04-05 12:05   ` [PATCH RFC 3/5] tun: vringfd receive support Rusty Russell
2008-04-05 12:06     ` [PATCH RFC 4/5] tun: vringfd xmit support Rusty Russell
2008-04-05 12:06     ` Rusty Russell
2008-04-05 12:09       ` [PATCH RFC 5/5] lguest support Rusty Russell
2008-04-05 12:09       ` Rusty Russell
2008-04-07  5:13       ` [PATCH RFC 4/5] tun: vringfd xmit support Herbert Xu
2008-04-07  5:13       ` Herbert Xu
2008-04-07  7:24         ` Rusty Russell
2008-04-07  7:35           ` David Miller
2008-04-07  7:35           ` David Miller
2008-04-08  1:51             ` Rusty Russell
2008-04-08  1:51             ` Rusty Russell
2008-04-07  7:24         ` Rusty Russell
2008-04-08 19:49     ` [PATCH RFC 3/5] tun: vringfd receive support Max Krasnyansky
2008-04-09 12:46       ` Dor Laor
2008-04-10 17:02         ` Max Krasnyanskiy
2008-04-10 17:02         ` Max Krasnyanskiy
2008-04-09 12:46       ` Dor Laor
2008-04-10  5:44       ` Rusty Russell
2008-04-10 17:18         ` Max Krasnyanskiy
2008-04-10 17:18         ` Max Krasnyanskiy
2008-04-10  5:44       ` Rusty Russell
2008-04-08 19:49     ` Max Krasnyansky
2008-04-05 12:05   ` Rusty Russell
2008-04-05 17:26     ` Anthony Liguori
2008-04-05 17:26     ` Anthony Liguori
2008-04-05 12:44   ` [PATCH RFC 2/5] vringfd base/offset Avi Kivity
2008-04-06  2:54     ` Rusty Russell
2008-04-06  2:54     ` Rusty Russell
2008-04-05 12:44   ` Avi Kivity
2008-04-08  5:14   ` Arnd Bergmann
2008-04-08  5:14   ` Arnd Bergmann
2008-04-05 12:04 ` Rusty Russell
2008-04-05 17:18   ` Anthony Liguori
2008-04-06  3:23     ` Rusty Russell
2008-04-06  3:23     ` Rusty Russell
2008-04-05 17:18   ` Anthony Liguori
2008-04-07 17:54 ` [PATCH RFC 1/5] vringfd syscall Jonathan Corbet
2008-04-07 17:54 ` Jonathan Corbet
2008-04-07 22:34   ` Rusty Russell
2008-04-07 22:34   ` Rusty Russell
2008-04-08  2:35 ` Arnd Bergmann
2008-04-08  2:35   ` Arnd Bergmann
2008-04-08  2:35 ` Arnd Bergmann
2008-04-09 19:28 ` Jeremy Fitzhardinge
2008-04-09 19:28 ` Jeremy Fitzhardinge
2008-04-12 17:18 ` Marcelo Tosatti
2008-04-12 17:18 ` Marcelo Tosatti
2008-04-12 17:39   ` Marcelo Tosatti
2008-04-12 17:39   ` Marcelo Tosatti
2008-04-12 18:19   ` Rusty Russell
2008-04-12 18:19   ` Rusty Russell

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=200804052204.28518.rusty@rustcorp.com.au \
    --to=rusty@rustcorp.com.au \
    --cc=linux-kernel@vger.kernel.org \
    --cc=maxk@qualcomm.com \
    --cc=netdev@vger.kernel.org \
    --cc=virtualization@lists.linux-foundation.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.