[PATCH 0/4] V2, Move stuff out of xenfs

[PATCH 0/4] V2, Move stuff out of xenfs

[Bastian Blank]

In <20100605162947.GA31336@wavehammer.waldi.eu.org> I started the
discussion about xenfs and the usage of it. This is the second try to
move stuff over into normal device drivers.

Changes:
- Remove guest_capabilities stuff. It is only used to detect if the init
  scripts should start anything.
- Fix error values.

Bastian

[PATCH 1/4] xen: Add privcmd device driver

[Bastian Blank]

Access to arbitrary hypercalls is currently provided via xenfs. This
adds a standard character device to handle this. The support in xenfs
remains for backward compatibility and uses the device driver code.

Signed-off-by: Bastian Blank <waldi@debian.org>
---
 drivers/xen/Kconfig               |    7 ++++++
 drivers/xen/Makefile              |    2 +
 drivers/xen/{xenfs => }/privcmd.c |   39 ++++++++++++++++++++++++++++++++++++-
 drivers/xen/privcmd.h             |    3 ++
 drivers/xen/xenfs/Makefile        |    2 +-
 drivers/xen/xenfs/super.c         |    3 +-
 drivers/xen/xenfs/xenfs.h         |    1 -
 7 files changed, 53 insertions(+), 4 deletions(-)
 rename drivers/xen/{xenfs => }/privcmd.c (92%)
 create mode 100644 drivers/xen/privcmd.h

diff --git a/drivers/xen/Kconfig b/drivers/xen/Kconfig
index 8795480..a1ced52 100644
--- a/drivers/xen/Kconfig
+++ b/drivers/xen/Kconfig
@@ -86,6 +86,7 @@ config XEN_BACKEND
 
 config XENFS
 	tristate "Xen filesystem"
+	select XEN_PRIVCMD
 	default y
 	help
 	  The xen filesystem provides a way for domains to share
@@ -171,4 +172,10 @@ config XEN_PCIDEV_BACKEND
 	  xen-pciback.hide=(03:00.0)(04:00.0)
 
 	  If in doubt, say m.
+
+config XEN_PRIVCMD
+	tristate
+	depends on XEN
+	default m
+
 endmenu
diff --git a/drivers/xen/Makefile b/drivers/xen/Makefile
index 974fffd..aa31337 100644
--- a/drivers/xen/Makefile
+++ b/drivers/xen/Makefile
@@ -19,7 +19,9 @@ obj-$(CONFIG_XEN_TMEM)			+= tmem.o
 obj-$(CONFIG_SWIOTLB_XEN)		+= swiotlb-xen.o
 obj-$(CONFIG_XEN_DOM0)			+= pci.o
 obj-$(CONFIG_XEN_PCIDEV_BACKEND)	+= xen-pciback/
+obj-$(CONFIG_XEN_PRIVCMD)		+= xen-privcmd.o
 
 xen-evtchn-y				:= evtchn.o
 xen-gntdev-y				:= gntdev.o
 xen-gntalloc-y				:= gntalloc.o
+xen-privcmd-y				:= privcmd.o
diff --git a/drivers/xen/xenfs/privcmd.c b/drivers/xen/privcmd.c
similarity index 92%
rename from drivers/xen/xenfs/privcmd.c
rename to drivers/xen/privcmd.c
index dbd3b16..863fbd0 100644
--- a/drivers/xen/xenfs/privcmd.c
+++ b/drivers/xen/privcmd.c
@@ -7,6 +7,7 @@
  */
 
 #include <linux/kernel.h>
+#include <linux/module.h>
 #include <linux/sched.h>
 #include <linux/slab.h>
 #include <linux/string.h>
@@ -18,6 +19,7 @@
 #include <linux/highmem.h>
 #include <linux/pagemap.h>
 #include <linux/seq_file.h>
+#include <linux/miscdevice.h>
 
 #include <asm/pgalloc.h>
 #include <asm/pgtable.h>
@@ -32,6 +34,10 @@
 #include <xen/page.h>
 #include <xen/xen-ops.h>
 
+#include "privcmd.h"
+
+MODULE_LICENSE("GPL");
+
 #ifndef HAVE_ARCH_PRIVCMD_MMAP
 static int privcmd_enforce_singleshot_mapping(struct vm_area_struct *vma);
 #endif
@@ -394,7 +400,38 @@ static int privcmd_enforce_singleshot_mapping(struct vm_area_struct *vma)
 }
 #endif
 
-const struct file_operations privcmd_file_ops = {
+const struct file_operations xen_privcmd_fops = {
+	.owner = THIS_MODULE,
 	.unlocked_ioctl = privcmd_ioctl,
 	.mmap = privcmd_mmap,
 };
+EXPORT_SYMBOL_GPL(xen_privcmd_fops);
+
+static struct miscdevice privcmd_dev = {
+	.minor = MISC_DYNAMIC_MINOR,
+	.name = "xen/privcmd",
+	.fops = &xen_privcmd_fops,
+};
+
+static int __init privcmd_init(void)
+{
+	int err;
+
+	if (!xen_domain())
+		return -ENODEV;
+
+	err = misc_register(&privcmd_dev);
+	if (err != 0) {
+		printk(KERN_ERR "Could not register privcmd device\n");
+		return err;
+	}
+	return 0;
+}
+
+static void __exit privcmd_exit(void)
+{
+	misc_deregister(&privcmd_dev);
+}
+
+module_init(privcmd_init);
+module_exit(privcmd_exit);
diff --git a/drivers/xen/privcmd.h b/drivers/xen/privcmd.h
new file mode 100644
index 0000000..14facae
--- /dev/null
+++ b/drivers/xen/privcmd.h
@@ -0,0 +1,3 @@
+#include <linux/fs.h>
+
+extern const struct file_operations xen_privcmd_fops;
diff --git a/drivers/xen/xenfs/Makefile b/drivers/xen/xenfs/Makefile
index 4fde944..5d45ff1 100644
--- a/drivers/xen/xenfs/Makefile
+++ b/drivers/xen/xenfs/Makefile
@@ -1,4 +1,4 @@
 obj-$(CONFIG_XENFS) += xenfs.o
 
-xenfs-y			  = super.o xenbus.o privcmd.o
+xenfs-y			  = super.o xenbus.o
 xenfs-$(CONFIG_XEN_DOM0) += xenstored.o
diff --git a/drivers/xen/xenfs/super.c b/drivers/xen/xenfs/super.c
index 1aa3897..a55fbf9 100644
--- a/drivers/xen/xenfs/super.c
+++ b/drivers/xen/xenfs/super.c
@@ -16,6 +16,7 @@
 #include <xen/xen.h>
 
 #include "xenfs.h"
+#include "../privcmd.h"
 
 #include <asm/xen/hypervisor.h>
 
@@ -84,7 +85,7 @@ static int xenfs_fill_super(struct super_block *sb, void *data, int silent)
 		[1] = {},
 		{ "xenbus", &xenbus_file_ops, S_IRUSR|S_IWUSR },
 		{ "capabilities", &capabilities_file_ops, S_IRUGO },
-		{ "privcmd", &privcmd_file_ops, S_IRUSR|S_IWUSR },
+		{ "privcmd", &xen_privcmd_fops, S_IRUSR|S_IWUSR },
 		{""},
 	};
 	int rc;
diff --git a/drivers/xen/xenfs/xenfs.h b/drivers/xen/xenfs/xenfs.h
index b68aa62..5056306 100644
--- a/drivers/xen/xenfs/xenfs.h
+++ b/drivers/xen/xenfs/xenfs.h
@@ -2,7 +2,6 @@
 #define _XENFS_XENBUS_H
 
 extern const struct file_operations xenbus_file_ops;
-extern const struct file_operations privcmd_file_ops;
 extern const struct file_operations xsd_kva_file_ops;
 extern const struct file_operations xsd_port_file_ops;
 
-- 
1.7.7.3

Re: [Xen-devel] [PATCH 1/4] xen: Add privcmd device driver

[Ian Campbell]

On Wed, 2011-12-07 at 21:13 +0000, Bastian Blank wrote:
> +static int __init privcmd_init(void)
> +{
> +       int err;
> +
> +       if (!xen_domain())
> +               return -ENODEV;
> +
> +       err = misc_register(&privcmd_dev);
> +       if (err != 0) {
> +               printk(KERN_ERR "Could not register privcmd device\n"); 

Should say "...register Xen privcmd device". Otherwise:

Acked-by: Ian Campbell <ian.campbell@citrix.com>

[PATCH 2/4] xen: Add xenbus device driver

[Bastian Blank]

Access to xenbus is currently handled via xenfs. This adds a device
driver for xenbus and makes xenfs use this code.

Signed-off-by: Bastian Blank <waldi@debian.org>
---
 drivers/xen/xenbus/Makefile                        |    1 +
 drivers/xen/xenbus/xenbus_comms.h                  |    4 ++
 .../xenbus.c => xenbus/xenbus_dev_frontend.c}      |   37 ++++++++++++++++++--
 drivers/xen/xenfs/Makefile                         |    2 +-
 drivers/xen/xenfs/super.c                          |    3 +-
 drivers/xen/xenfs/xenfs.h                          |    1 -
 6 files changed, 42 insertions(+), 6 deletions(-)
 rename drivers/xen/{xenfs/xenbus.c => xenbus/xenbus_dev_frontend.c} (95%)

diff --git a/drivers/xen/xenbus/Makefile b/drivers/xen/xenbus/Makefile
index 8dca685..a2ea363 100644
--- a/drivers/xen/xenbus/Makefile
+++ b/drivers/xen/xenbus/Makefile
@@ -1,4 +1,5 @@
 obj-y	+= xenbus.o
+obj-y	+= xenbus_dev_frontend.o
 
 xenbus-objs =
 xenbus-objs += xenbus_client.o
diff --git a/drivers/xen/xenbus/xenbus_comms.h b/drivers/xen/xenbus/xenbus_comms.h
index c21db75..6e42800 100644
--- a/drivers/xen/xenbus/xenbus_comms.h
+++ b/drivers/xen/xenbus/xenbus_comms.h
@@ -31,6 +31,8 @@
 #ifndef _XENBUS_COMMS_H
 #define _XENBUS_COMMS_H
 
+#include <linux/fs.h>
+
 int xs_init(void);
 int xb_init_comms(void);
 
@@ -43,4 +45,6 @@ int xs_input_avail(void);
 extern struct xenstore_domain_interface *xen_store_interface;
 extern int xen_store_evtchn;
 
+extern const struct file_operations xen_xenbus_fops;
+
 #endif /* _XENBUS_COMMS_H */
diff --git a/drivers/xen/xenfs/xenbus.c b/drivers/xen/xenbus/xenbus_dev_frontend.c
similarity index 95%
rename from drivers/xen/xenfs/xenbus.c
rename to drivers/xen/xenbus/xenbus_dev_frontend.c
index bbd000f..fb30cff 100644
--- a/drivers/xen/xenfs/xenbus.c
+++ b/drivers/xen/xenbus/xenbus_dev_frontend.c
@@ -52,13 +52,16 @@
 #include <linux/namei.h>
 #include <linux/string.h>
 #include <linux/slab.h>
+#include <linux/miscdevice.h>
+#include <linux/module.h>
 
-#include "xenfs.h"
-#include "../xenbus/xenbus_comms.h"
+#include "xenbus_comms.h"
 
 #include <xen/xenbus.h>
 #include <asm/xen/hypervisor.h>
 
+MODULE_LICENSE("GPL");
+
 /*
  * An element of a list of outstanding transactions, for which we're
  * still waiting a reply.
@@ -583,7 +586,7 @@ static unsigned int xenbus_file_poll(struct file *file, poll_table *wait)
 	return 0;
 }
 
-const struct file_operations xenbus_file_ops = {
+const struct file_operations xen_xenbus_fops = {
 	.read = xenbus_file_read,
 	.write = xenbus_file_write,
 	.open = xenbus_file_open,
@@ -591,3 +594,31 @@ const struct file_operations xenbus_file_ops = {
 	.poll = xenbus_file_poll,
 	.llseek = no_llseek,
 };
+EXPORT_SYMBOL_GPL(xen_xenbus_fops);
+
+static struct miscdevice xenbus_dev = {
+	.minor = MISC_DYNAMIC_MINOR,
+	.name = "xen/xenbus",
+	.fops = &xen_xenbus_fops,
+};
+
+static int __init xenbus_init(void)
+{
+	int err;
+
+	if (!xen_domain())
+		return -ENODEV;
+
+	err = misc_register(&xenbus_dev);
+	if (err)
+		printk(KERN_ERR "Could not register xenbus device\n");
+	return err;
+}
+
+static void __exit xenbus_exit(void)
+{
+	misc_deregister(&xenbus_dev);
+}
+
+module_init(xenbus_init);
+module_exit(xenbus_exit);
diff --git a/drivers/xen/xenfs/Makefile b/drivers/xen/xenfs/Makefile
index 5d45ff1..b019865 100644
--- a/drivers/xen/xenfs/Makefile
+++ b/drivers/xen/xenfs/Makefile
@@ -1,4 +1,4 @@
 obj-$(CONFIG_XENFS) += xenfs.o
 
-xenfs-y			  = super.o xenbus.o
+xenfs-y			  = super.o
 xenfs-$(CONFIG_XEN_DOM0) += xenstored.o
diff --git a/drivers/xen/xenfs/super.c b/drivers/xen/xenfs/super.c
index a55fbf9..a84b53c 100644
--- a/drivers/xen/xenfs/super.c
+++ b/drivers/xen/xenfs/super.c
@@ -17,6 +17,7 @@
 
 #include "xenfs.h"
 #include "../privcmd.h"
+#include "../xenbus/xenbus_comms.h"
 
 #include <asm/xen/hypervisor.h>
 
@@ -83,7 +84,7 @@ static int xenfs_fill_super(struct super_block *sb, void *data, int silent)
 {
 	static struct tree_descr xenfs_files[] = {
 		[1] = {},
-		{ "xenbus", &xenbus_file_ops, S_IRUSR|S_IWUSR },
+		{ "xenbus", &xen_xenbus_fops, S_IRUSR|S_IWUSR },
 		{ "capabilities", &capabilities_file_ops, S_IRUGO },
 		{ "privcmd", &xen_privcmd_fops, S_IRUSR|S_IWUSR },
 		{""},
diff --git a/drivers/xen/xenfs/xenfs.h b/drivers/xen/xenfs/xenfs.h
index 5056306..6b80c77 100644
--- a/drivers/xen/xenfs/xenfs.h
+++ b/drivers/xen/xenfs/xenfs.h
@@ -1,7 +1,6 @@
 #ifndef _XENFS_XENBUS_H
 #define _XENFS_XENBUS_H
 
-extern const struct file_operations xenbus_file_ops;
 extern const struct file_operations xsd_kva_file_ops;
 extern const struct file_operations xsd_port_file_ops;
 
-- 
1.7.7.3

Re: [Xen-devel] [PATCH 2/4] xen: Add xenbus device driver

[Ian Campbell]

On Wed, 2011-12-07 at 21:13 +0000, Bastian Blank wrote:
> Access to xenbus is currently handled via xenfs. This adds a device
> driver for xenbus and makes xenfs use this code.
> 
> Signed-off-by: Bastian Blank <waldi@debian.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>

> ---
>  drivers/xen/xenbus/Makefile                        |    1 +
>  drivers/xen/xenbus/xenbus_comms.h                  |    4 ++
>  .../xenbus.c => xenbus/xenbus_dev_frontend.c}      |   37 ++++++++++++++++++--
>  drivers/xen/xenfs/Makefile                         |    2 +-
>  drivers/xen/xenfs/super.c                          |    3 +-
>  drivers/xen/xenfs/xenfs.h                          |    1 -
>  6 files changed, 42 insertions(+), 6 deletions(-)
>  rename drivers/xen/{xenfs/xenbus.c => xenbus/xenbus_dev_frontend.c} (95%)
> 
> diff --git a/drivers/xen/xenbus/Makefile b/drivers/xen/xenbus/Makefile
> index 8dca685..a2ea363 100644
> --- a/drivers/xen/xenbus/Makefile
> +++ b/drivers/xen/xenbus/Makefile
> @@ -1,4 +1,5 @@
>  obj-y	+= xenbus.o
> +obj-y	+= xenbus_dev_frontend.o
>  
>  xenbus-objs =
>  xenbus-objs += xenbus_client.o
> diff --git a/drivers/xen/xenbus/xenbus_comms.h b/drivers/xen/xenbus/xenbus_comms.h
> index c21db75..6e42800 100644
> --- a/drivers/xen/xenbus/xenbus_comms.h
> +++ b/drivers/xen/xenbus/xenbus_comms.h
> @@ -31,6 +31,8 @@
>  #ifndef _XENBUS_COMMS_H
>  #define _XENBUS_COMMS_H
>  
> +#include <linux/fs.h>
> +
>  int xs_init(void);
>  int xb_init_comms(void);
>  
> @@ -43,4 +45,6 @@ int xs_input_avail(void);
>  extern struct xenstore_domain_interface *xen_store_interface;
>  extern int xen_store_evtchn;
>  
> +extern const struct file_operations xen_xenbus_fops;
> +
>  #endif /* _XENBUS_COMMS_H */
> diff --git a/drivers/xen/xenfs/xenbus.c b/drivers/xen/xenbus/xenbus_dev_frontend.c
> similarity index 95%
> rename from drivers/xen/xenfs/xenbus.c
> rename to drivers/xen/xenbus/xenbus_dev_frontend.c
> index bbd000f..fb30cff 100644
> --- a/drivers/xen/xenfs/xenbus.c
> +++ b/drivers/xen/xenbus/xenbus_dev_frontend.c
> @@ -52,13 +52,16 @@
>  #include <linux/namei.h>
>  #include <linux/string.h>
>  #include <linux/slab.h>
> +#include <linux/miscdevice.h>
> +#include <linux/module.h>
>  
> -#include "xenfs.h"
> -#include "../xenbus/xenbus_comms.h"
> +#include "xenbus_comms.h"
>  
>  #include <xen/xenbus.h>
>  #include <asm/xen/hypervisor.h>
>  
> +MODULE_LICENSE("GPL");
> +
>  /*
>   * An element of a list of outstanding transactions, for which we're
>   * still waiting a reply.
> @@ -583,7 +586,7 @@ static unsigned int xenbus_file_poll(struct file *file, poll_table *wait)
>  	return 0;
>  }
>  
> -const struct file_operations xenbus_file_ops = {
> +const struct file_operations xen_xenbus_fops = {
>  	.read = xenbus_file_read,
>  	.write = xenbus_file_write,
>  	.open = xenbus_file_open,
> @@ -591,3 +594,31 @@ const struct file_operations xenbus_file_ops = {
>  	.poll = xenbus_file_poll,
>  	.llseek = no_llseek,
>  };
> +EXPORT_SYMBOL_GPL(xen_xenbus_fops);
> +
> +static struct miscdevice xenbus_dev = {
> +	.minor = MISC_DYNAMIC_MINOR,
> +	.name = "xen/xenbus",
> +	.fops = &xen_xenbus_fops,
> +};
> +
> +static int __init xenbus_init(void)
> +{
> +	int err;
> +
> +	if (!xen_domain())
> +		return -ENODEV;
> +
> +	err = misc_register(&xenbus_dev);
> +	if (err)
> +		printk(KERN_ERR "Could not register xenbus device\n");
> +	return err;
> +}
> +
> +static void __exit xenbus_exit(void)
> +{
> +	misc_deregister(&xenbus_dev);
> +}
> +
> +module_init(xenbus_init);
> +module_exit(xenbus_exit);
> diff --git a/drivers/xen/xenfs/Makefile b/drivers/xen/xenfs/Makefile
> index 5d45ff1..b019865 100644
> --- a/drivers/xen/xenfs/Makefile
> +++ b/drivers/xen/xenfs/Makefile
> @@ -1,4 +1,4 @@
>  obj-$(CONFIG_XENFS) += xenfs.o
>  
> -xenfs-y			  = super.o xenbus.o
> +xenfs-y			  = super.o
>  xenfs-$(CONFIG_XEN_DOM0) += xenstored.o
> diff --git a/drivers/xen/xenfs/super.c b/drivers/xen/xenfs/super.c
> index a55fbf9..a84b53c 100644
> --- a/drivers/xen/xenfs/super.c
> +++ b/drivers/xen/xenfs/super.c
> @@ -17,6 +17,7 @@
>  
>  #include "xenfs.h"
>  #include "../privcmd.h"
> +#include "../xenbus/xenbus_comms.h"
>  
>  #include <asm/xen/hypervisor.h>
>  
> @@ -83,7 +84,7 @@ static int xenfs_fill_super(struct super_block *sb, void *data, int silent)
>  {
>  	static struct tree_descr xenfs_files[] = {
>  		[1] = {},
> -		{ "xenbus", &xenbus_file_ops, S_IRUSR|S_IWUSR },
> +		{ "xenbus", &xen_xenbus_fops, S_IRUSR|S_IWUSR },
>  		{ "capabilities", &capabilities_file_ops, S_IRUGO },
>  		{ "privcmd", &xen_privcmd_fops, S_IRUSR|S_IWUSR },
>  		{""},
> diff --git a/drivers/xen/xenfs/xenfs.h b/drivers/xen/xenfs/xenfs.h
> index 5056306..6b80c77 100644
> --- a/drivers/xen/xenfs/xenfs.h
> +++ b/drivers/xen/xenfs/xenfs.h
> @@ -1,7 +1,6 @@
>  #ifndef _XENFS_XENBUS_H
>  #define _XENFS_XENBUS_H
>  
> -extern const struct file_operations xenbus_file_ops;
>  extern const struct file_operations xsd_kva_file_ops;
>  extern const struct file_operations xsd_port_file_ops;
>  

[PATCH 3/4] xen: Add xenbusd device driver

[Bastian Blank]

Access for xenstored to the event channel and pre-allocated ring is
managed via xenfs.  This adds its own device driver featuring mmap for
the ring and an ioctl for the event channel.

Signed-off-by: Bastian Blank <waldi@debian.org>
---
 drivers/xen/xenbus/Makefile             |    1 +
 drivers/xen/xenbus/xenbus_dev_backend.c |   79 +++++++++++++++++++++++++++++++
 include/xen/xenbus_dev.h                |   41 ++++++++++++++++
 3 files changed, 121 insertions(+), 0 deletions(-)
 create mode 100644 drivers/xen/xenbus/xenbus_dev_backend.c
 create mode 100644 include/xen/xenbus_dev.h

diff --git a/drivers/xen/xenbus/Makefile b/drivers/xen/xenbus/Makefile
index a2ea363..31e2e90 100644
--- a/drivers/xen/xenbus/Makefile
+++ b/drivers/xen/xenbus/Makefile
@@ -10,4 +10,5 @@ xenbus-objs += xenbus_probe.o
 xenbus-be-objs-$(CONFIG_XEN_BACKEND) += xenbus_probe_backend.o
 xenbus-objs += $(xenbus-be-objs-y)
 
+obj-$(CONFIG_XEN_BACKEND) += xenbus_dev_backend.o
 obj-$(CONFIG_XEN_XENBUS_FRONTEND) += xenbus_probe_frontend.o
diff --git a/drivers/xen/xenbus/xenbus_dev_backend.c b/drivers/xen/xenbus/xenbus_dev_backend.c
new file mode 100644
index 0000000..58cf621
--- /dev/null
+++ b/drivers/xen/xenbus/xenbus_dev_backend.c
@@ -0,0 +1,79 @@
+#include <linux/slab.h>
+#include <linux/types.h>
+#include <linux/mm.h>
+#include <linux/fs.h>
+#include <linux/miscdevice.h>
+#include <linux/module.h>
+
+#include <xen/page.h>
+#include <xen/xenbus_dev.h>
+
+#include "xenbus_comms.h"
+
+MODULE_LICENSE("GPL");
+
+static long xenbusd_ioctl(struct file *file, unsigned int cmd, unsigned long data)
+{
+	if (!capable(CAP_SYS_ADMIN))
+		return -EACCES;
+
+	switch (cmd) {
+		case IOCTL_XENBUSD_EVTCHN:
+			if (xen_store_evtchn > 0)
+				return xen_store_evtchn;
+			return -ENODEV;
+
+		default:
+			return -ENOTTY;
+	}
+}
+
+static int xenbusd_mmap(struct file *file, struct vm_area_struct *vma)
+{
+	size_t size = vma->vm_end - vma->vm_start;
+
+	if (!capable(CAP_SYS_ADMIN))
+		return -EACCES;
+
+	if ((size > PAGE_SIZE) || (vma->vm_pgoff != 0))
+		return -EINVAL;
+
+	if (remap_pfn_range(vma, vma->vm_start,
+			    virt_to_pfn(xen_store_interface),
+			    size, vma->vm_page_prot))
+		return -EAGAIN;
+
+	return 0;
+}
+
+const struct file_operations xenbusd_fops = {
+	.mmap = xenbusd_mmap,
+	.unlocked_ioctl = xenbusd_ioctl,
+};
+
+static struct miscdevice xenbusd_dev = {
+	.minor = MISC_DYNAMIC_MINOR,
+	.name = "xen/xenbusd",
+	.fops = &xenbusd_fops,
+};
+
+static int __init xenbusd_init(void)
+{
+	int err;
+
+	if (!xen_initial_domain())
+		return -ENODEV;
+
+	err = misc_register(&xenbusd_dev);
+	if (err)
+		printk(KERN_ERR "Could not register xenbus device\n");
+	return err;
+}
+
+static void __exit xenbusd_exit(void)
+{
+	misc_deregister(&xenbusd_dev);
+}
+
+module_init(xenbusd_init);
+module_exit(xenbusd_exit);
diff --git a/include/xen/xenbus_dev.h b/include/xen/xenbus_dev.h
new file mode 100644
index 0000000..3008c03
--- /dev/null
+++ b/include/xen/xenbus_dev.h
@@ -0,0 +1,41 @@
+/******************************************************************************
+ * evtchn.h
+ *
+ * Interface to /dev/xen/xenbusd.
+ *
+ * Copyright (c) 2011 Bastian Blank <waldi@debian.org>
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License version 2
+ * as published by the Free Software Foundation; or, when distributed
+ * separately from the Linux kernel or incorporated into other
+ * software packages, subject to the following license:
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this source file (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.
+ */
+
+#ifndef __LINUX_XEN_XENBUS_DEV_H__
+#define __LINUX_XEN_XENBUS_DEV_H__
+
+#include <linux/ioctl.h>
+
+#define IOCTL_XENBUSD_EVTCHN				\
+	_IOC(_IOC_NONE, 'B', 0, 0)
+
+#endif /* __LINUX_XEN_XENBUS_DEV_H__ */
-- 
1.7.7.3

Re: [Xen-devel] [PATCH 3/4] xen: Add xenbusd device driver

[Ian Campbell]

On Wed, 2011-12-07 at 21:13 +0000, Bastian Blank wrote:
> Access for xenstored to the event channel and pre-allocated ring is
> managed via xenfs.  This adds its own device driver featuring mmap for
> the ring and an ioctl for the event channel.
> 
> Signed-off-by: Bastian Blank <waldi@debian.org>
> ---
>  drivers/xen/xenbus/Makefile             |    1 +
>  drivers/xen/xenbus/xenbus_dev_backend.c |   79 +++++++++++++++++++++++++++++++
>  include/xen/xenbus_dev.h                |   41 ++++++++++++++++
>  3 files changed, 121 insertions(+), 0 deletions(-)
>  create mode 100644 drivers/xen/xenbus/xenbus_dev_backend.c
>  create mode 100644 include/xen/xenbus_dev.h
> 
> diff --git a/drivers/xen/xenbus/Makefile b/drivers/xen/xenbus/Makefile
> index a2ea363..31e2e90 100644
> --- a/drivers/xen/xenbus/Makefile
> +++ b/drivers/xen/xenbus/Makefile
> @@ -10,4 +10,5 @@ xenbus-objs += xenbus_probe.o
>  xenbus-be-objs-$(CONFIG_XEN_BACKEND) += xenbus_probe_backend.o
>  xenbus-objs += $(xenbus-be-objs-y)
>  
> +obj-$(CONFIG_XEN_BACKEND) += xenbus_dev_backend.o

In principal this could be its own module which does not depend on any
of the backend stuff but I think this is fine for now.

> [...]

> +static int __init xenbusd_init(void)
> +{
> +	int err;
> +
> +	if (!xen_initial_domain())
> +		return -ENODEV;
> +
> +	err = misc_register(&xenbusd_dev);
> +	if (err)
> +		printk(KERN_ERR "Could not register xenbus device\n");

"Could not register xenbus backend device" to distinguish from the f.e.
failure message, you could also add "frontend" there I suppose. Other
than that:

Acked-by: Ian Campbell <ian.campbell@citrix.com>

Re: [Xen-devel] [PATCH 3/4] xen: Add xenbusd device driver

[Bastian Blank]

On Thu, Dec 08, 2011 at 08:24:46AM +0000, Ian Campbell wrote:
> On Wed, 2011-12-07 at 21:13 +0000, Bastian Blank wrote:
> > +obj-$(CONFIG_XEN_BACKEND) += xenbus_dev_backend.o
> In principal this could be its own module which does not depend on any
> of the backend stuff but I think this is fine for now.

I even think about registering this device from xenstored_local_init in
xenbus_probe.c. Then it will be available only if the kernel also owns
the communcation ring.

> "Could not register xenbus backend device" to distinguish from the f.e.
> failure message, you could also add "frontend" there I suppose. Other
> than that:

Yep.

Will someone mind if I rename this device to xenbus_backend? It's more
clear.

Bastian

-- 
Deflector shields just came on, Captain.

Re: [Xen-devel] [PATCH 3/4] xen: Add xenbusd device driver

[Ian Campbell]

On Thu, 2011-12-08 at 09:27 +0000, Bastian Blank wrote:
> On Thu, Dec 08, 2011 at 08:24:46AM +0000, Ian Campbell wrote:
> > On Wed, 2011-12-07 at 21:13 +0000, Bastian Blank wrote:
> > > +obj-$(CONFIG_XEN_BACKEND) += xenbus_dev_backend.o
> > In principal this could be its own module which does not depend on any
> > of the backend stuff but I think this is fine for now.
> 
> I even think about registering this device from xenstored_local_init in
> xenbus_probe.c. Then it will be available only if the kernel also owns
> the communcation ring.

That sounds like a good idea.

> > "Could not register xenbus backend device" to distinguish from the f.e.
> > failure message, you could also add "frontend" there I suppose. Other
> > than that:
> 
> Yep.
> 
> Will someone mind if I rename this device to xenbus_backend? It's more
> clear.

It's ok by me. You could even use "xenstored"?

Ian.

Re: [Xen-devel] [PATCH 3/4] xen: Add xenbusd device driver

[Bastian Blank]

On Thu, Dec 08, 2011 at 10:27:58AM +0100, Bastian Blank wrote:
> I even think about registering this device from xenstored_local_init in
> xenbus_probe.c. Then it will be available only if the kernel also owns
> the communcation ring.

I'm not sure if I'm allowed to register character devices from a
postcore_initcall. Isn't this function called a bit early anyway?

Bastian

-- 
Each kiss is as the first.
		-- Miramanee, Kirk's wife, "The Paradise Syndrome",
		   stardate 4842.6

[PATCH 4/4] xen/privcmd: Remove unused support for arch specific privcmp mmap

[Bastian Blank]

---
 drivers/xen/privcmd.c |    2 --
 1 files changed, 0 insertions(+), 2 deletions(-)

diff --git a/drivers/xen/privcmd.c b/drivers/xen/privcmd.c
index 863fbd0..c13d26a 100644
--- a/drivers/xen/privcmd.c
+++ b/drivers/xen/privcmd.c
@@ -365,7 +365,6 @@ static long privcmd_ioctl(struct file *file,
 	return ret;
 }
 
-#ifndef HAVE_ARCH_PRIVCMD_MMAP
 static int privcmd_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
 {
 	printk(KERN_DEBUG "privcmd_fault: vma=%p %lx-%lx, pgoff=%lx, uv=%p\n",
@@ -398,7 +397,6 @@ static int privcmd_enforce_singleshot_mapping(struct vm_area_struct *vma)
 {
 	return (xchg(&vma->vm_private_data, (void *)1) == NULL);
 }
-#endif
 
 const struct file_operations xen_privcmd_fops = {
 	.owner = THIS_MODULE,
-- 
1.7.7.3

Re: [Xen-devel] [PATCH 4/4] xen/privcmd: Remove unused support for arch specific privcmp mmap

[Ian Campbell]

On Wed, 2011-12-07 at 21:13 +0000, Bastian Blank wrote:
> ---

You've forgotten the S-o-b for this patch.

Once upon a time this was used by the ia64 privcmd support[0] but since
no one appears to be working on ia64 dom0 support for upstream I'm not
sure if this is an issue or not. I lean towards "not", this can always
be put back as necessary.

Acked-by: Ian Campbell <ian.campbell@citrix.com>

[0]
http://xenbits.xen.org/hg/linux-2.6.18-xen.hg/file/bffd07392da5/arch/ia64/xen/hypervisor.c#l659
http://xenbits.xen.org/hg/linux-2.6.18-xen.hg/file/bffd07392da5/include/asm-ia64/hypervisor.h#l132

>  drivers/xen/privcmd.c |    2 --
>  1 files changed, 0 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/xen/privcmd.c b/drivers/xen/privcmd.c
> index 863fbd0..c13d26a 100644
> --- a/drivers/xen/privcmd.c
> +++ b/drivers/xen/privcmd.c
> @@ -365,7 +365,6 @@ static long privcmd_ioctl(struct file *file,
>  	return ret;
>  }
>  
> -#ifndef HAVE_ARCH_PRIVCMD_MMAP
>  static int privcmd_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
>  {
>  	printk(KERN_DEBUG "privcmd_fault: vma=%p %lx-%lx, pgoff=%lx, uv=%p\n",
> @@ -398,7 +397,6 @@ static int privcmd_enforce_singleshot_mapping(struct vm_area_struct *vma)
>  {
>  	return (xchg(&vma->vm_private_data, (void *)1) == NULL);
>  }
> -#endif
>  
>  const struct file_operations xen_privcmd_fops = {
>  	.owner = THIS_MODULE,

Re: [Xen-devel] [PATCH 0/4] V2, Move stuff out of xenfs

[Ian Campbell]

On Wed, 2011-12-07 at 21:13 +0000, Bastian Blank wrote:
> In <20100605162947.GA31336@wavehammer.waldi.eu.org> I started the
> discussion about xenfs and the usage of it. This is the second try to
> move stuff over into normal device drivers.

I had a few nit picks about error messages but otherwise I'm happy with
this.

How are we doing on the corresponding toolstack changes?

Ian.

Re: [Xen-devel] [PATCH 0/4] V2, Move stuff out of xenfs

[Bastian Blank]

On Thu, Dec 08, 2011 at 08:34:19AM +0000, Ian Campbell wrote:
> How are we doing on the corresponding toolstack changes?

I have at least parts of it somewhere on my workstation. I'll polish and
submit them also.

Bastian

-- 
Witch!  Witch!  They'll burn ya!
		-- Hag, "Tomorrow is Yesterday", stardate unknown

Re: [Xen-devel] [PATCH 0/4] V2, Move stuff out of xenfs

[David Vrabel]

On 08/12/11 09:28, Bastian Blank wrote:
> On Thu, Dec 08, 2011 at 08:34:19AM +0000, Ian Campbell wrote:
>> How are we doing on the corresponding toolstack changes?
> 
> I have at least parts of it somewhere on my workstation. I'll polish and
> submit them also.

I assume the toolstack will fall back to xenfs if the devices don't exist?

David

Re: [Xen-devel] [PATCH 0/4] V2, Move stuff out of xenfs

[Bastian Blank]

On Thu, Dec 08, 2011 at 11:10:58AM +0000, David Vrabel wrote:
> On 08/12/11 09:28, Bastian Blank wrote:
> > On Thu, Dec 08, 2011 at 08:34:19AM +0000, Ian Campbell wrote:
> >> How are we doing on the corresponding toolstack changes?
> > I have at least parts of it somewhere on my workstation. I'll polish and
> > submit them also.
> I assume the toolstack will fall back to xenfs if the devices don't exist?

This was the plan.

Bastian

-- 
	"Get back to your stations!"
	"We're beaming down to the planet, sir."
		-- Kirk and Mr. Leslie, "This Side of Paradise",
		   stardate 3417.3

Re: [Xen-devel] [PATCH 0/4] V2, Move stuff out of xenfs

[David Vrabel]

On 07/12/2011 21:13, Bastian Blank wrote:
> In<20100605162947.GA31336@wavehammer.waldi.eu.org>  I started the
> discussion about xenfs and the usage of it. This is the second try to
> move stuff over into normal device drivers.

You still don't say why this is needed in the commit messages.  I can't
find the message you refer to in the previous thread (the mail archives
don't let you search by message-id).

David