Thread (3 messages) flat view 3 messages, 2 authors, 2014-10-23

Re: [PATCH RFC v3 01/16] virtio: memory access APIs

From: Cornelia Huck <hidden>
Date: 2014-10-23 07:54:20
Also in: lkml, virtualization

On Wed, 22 Oct 2014 21:44:08 +0300
"Michael S. Tsirkin" [off-list ref] wrote:
quoted hunk ↗ jump to hunk
virtio 1.0 makes all memory structures LE, so
we need APIs to conditionally do a byteswap on BE
architectures.

To make it easier to check code statically,
add virtio specific types for multi-byte integers
in memory.

Add low level wrappers that do a byteswap conditionally, these will be
useful e.g. for vhost.  Add high level wrappers that will (in the
future) query device endian-ness and act accordingly.

At the moment, stub them out and assume native endian-ness everywhere.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
---
 include/linux/virtio_byteorder.h | 29 ++++++++++++++++++++++++
 include/linux/virtio_config.h    | 16 +++++++++++++
 include/uapi/linux/virtio_ring.h | 49 ++++++++++++++++++++--------------------
 include/uapi/linux/Kbuild        |  1 +
 4 files changed, 71 insertions(+), 24 deletions(-)
 create mode 100644 include/linux/virtio_byteorder.h
diff --git a/include/linux/virtio_byteorder.h b/include/linux/virtio_byteorder.h
new file mode 100644
index 0000000..7afdd8a
--- /dev/null
+++ b/include/linux/virtio_byteorder.h
@@ -0,0 +1,29 @@
+#ifndef _LINUX_VIRTIO_BYTEORDER_H
+#define _LINUX_VIRTIO_BYTEORDER_H
+#include <linux/types.h>
+#include <uapi/linux/virtio_types.h>
+
+/* Memory accessors for handling virtio in modern little endian and in
+ * compatibility big endian format. */
s/big/native/
+
+#define __DEFINE_VIRTIO_XX_TO_CPU(bits) \
+static inline u##bits __virtio##bits##_to_cpu(bool little_endian, __virtio##bits val) \
+{ \
+	if (little_endian) \
+		return le##bits##_to_cpu((__force __le##bits)val); \
+	else \
+		return (__force u##bits)val; \
+} \
+static inline __virtio##bits __cpu_to_virtio##bits(bool little_endian, u##bits val) \
+{ \
+	if (little_endian) \
+		return (__force __virtio##bits)cpu_to_le##bits(val); \
+	else \
+		return val; \
+}
+
+__DEFINE_VIRTIO_XX_TO_CPU(16)
+__DEFINE_VIRTIO_XX_TO_CPU(32)
+__DEFINE_VIRTIO_XX_TO_CPU(64)
...although I'm still not too happy with macro-generated helpers.
+
+#endif /* _LINUX_VIRTIO_BYTEORDER */
quoted hunk ↗ jump to hunk
diff --git a/include/uapi/linux/virtio_ring.h b/include/uapi/linux/virtio_ring.h
index a99f9b7..6c00632 100644
--- a/include/uapi/linux/virtio_ring.h
+++ b/include/uapi/linux/virtio_ring.h
quoted hunk ↗ jump to hunk
@@ -61,32 +62,32 @@
 /* Virtio ring descriptors: 16 bytes.  These can chain together via "next". */
 struct vring_desc {
 	/* Address (guest-physical). */
-	__u64 addr;
+	__virtio64 addr;
 	/* Length. */
-	__u32 len;
+	__virtio32 len;
 	/* The flags as indicated above. */
-	__u16 flags;
+	__virtio16 flags;
 	/* We chain unused descriptors via this, too */
-	__u16 next;
+	__virtio16 next;
 };
I think all of these __virtio types need an explanation somewhere as to
what they mean, e.g.:

/*
 * __virtio{16,32,64} have the following meaning:
 * - __u{16,32,64} for virtio devices in legacy mode,
 *   accessed in native endian
 * - __le{16,32,64} for standard-compliant virtio devices
 */
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help