Device names

Device names

[H. Peter Anvin]

I just found out that all the device handling in btrfs is based on
pathnames, but shorter pathnames (1024) that PATH_MAX (4096).

This is confusing, and concerning for multiple reasons:

1. pathnames are namespace-specific; what is a pathname in one namespace
might not be in another.
2. different truncation rules in the rest of the kernel.

There seem to be no way one can do the equivalent of BTRFS_IOC_DEV_INFO
but actually get a device number or other stable information for that
device (that could be compared to a file descriptor for verification);
is that an accurate observation?

	-hpa

-- 
H. Peter Anvin, Intel Open Source Technology Center
I work for Intel.  I don't speak on their behalf.

Re: Device names

[Chris Mason]

On Mon, Jun 18, 2012 at 06:29:59PM -0600, H. Peter Anvin wrote:
> I just found out that all the device handling in btrfs is based on
> pathnames, but shorter pathnames (1024) that PATH_MAX (4096).
> 
> This is confusing, and concerning for multiple reasons:
> 
> 1. pathnames are namespace-specific; what is a pathname in one namespace
> might not be in another.
> 2. different truncation rules in the rest of the kernel.
> 
> There seem to be no way one can do the equivalent of BTRFS_IOC_DEV_INFO
> but actually get a device number or other stable information for that
> device (that could be compared to a file descriptor for verification);
> is that an accurate observation?

At mount time, we go through and verify the path names still belong to
the filesystem you thought they belonged to.  The bdev is locked during
the verification, so it won't be able to go away or change.

This is a long way of saying right we don't spit out device numbers.
Even device numbers can change.  We can easily add a uuid based listing,
which I think is what you want.

-chris

Re: Device names

[H. Peter Anvin]

On 06/19/2012 04:51 PM, Chris Mason wrote:
> 
> At mount time, we go through and verify the path names still belong to
> the filesystem you thought they belonged to.  The bdev is locked during
> the verification, so it won't be able to go away or change.
> 
> This is a long way of saying right we don't spit out device numbers.
> Even device numbers can change.  We can easily add a uuid based listing,
> which I think is what you want.
> 

No, I want to find the actual devices.  I know I can get the UUID, but
scanning all the block devices in the system looking for that UUID is a
nonstarter.

Device path names can change while the system is operating (and, worse,
are dependent on namespace changes and chroot); device *numbers* cannot
as long as the device is in use (e.g. mounted.)  They can indeed change
while not in use, of course.

	-hpa

Re: Device names

[Chris Mason]

On Tue, Jun 19, 2012 at 06:00:11PM -0600, H. Peter Anvin wrote:
> On 06/19/2012 04:51 PM, Chris Mason wrote:
> > 
> > At mount time, we go through and verify the path names still belong to
> > the filesystem you thought they belonged to.  The bdev is locked during
> > the verification, so it won't be able to go away or change.
> > 
> > This is a long way of saying right we don't spit out device numbers.
> > Even device numbers can change.  We can easily add a uuid based listing,
> > which I think is what you want.
> > 
> 
> No, I want to find the actual devices.  I know I can get the UUID, but
> scanning all the block devices in the system looking for that UUID is a
> nonstarter.
> 
> Device path names can change while the system is operating (and, worse,
> are dependent on namespace changes and chroot); device *numbers* cannot
> as long as the device is in use (e.g. mounted.)  They can indeed change
> while not in use, of course.

Ok, my two choices for exporting this to you are a /sys/btrfs kind of
directory (representing the mounted filesystems) or an ioctl.  Which one
is most usable for you?

You want to map from /some_dir to a definitive list of devices you need
to find in syslinux to later boot off that FS, right?

-chris

Re: Device names

[Goffredo Baroncelli]

On 06/20/2012 03:37 PM, Chris Mason wrote:
> On Tue, Jun 19, 2012 at 06:00:11PM -0600, H. Peter Anvin wrote:
>> On 06/19/2012 04:51 PM, Chris Mason wrote:
>>>
>>> At mount time, we go through and verify the path names still belong to
>>> the filesystem you thought they belonged to.  The bdev is locked during
>>> the verification, so it won't be able to go away or change.
>>>
>>> This is a long way of saying right we don't spit out device numbers.
>>> Even device numbers can change.  We can easily add a uuid based listing,
>>> which I think is what you want.
>>>
>>
>> No, I want to find the actual devices.  I know I can get the UUID, but
>> scanning all the block devices in the system looking for that UUID is a
>> nonstarter.
>>
>> Device path names can change while the system is operating (and, worse,
>> are dependent on namespace changes and chroot); device *numbers* cannot
>> as long as the device is in use (e.g. mounted.)  They can indeed change
>> while not in use, of course.
> 
> Ok, my two choices for exporting this to you are a /sys/btrfs kind of
> directory (representing the mounted filesystems) or an ioctl.  Which one
> is most usable for you?

As short term solution, I suggest to update the BTRFS_IOC_DEV_INFO to
export also the major:minor pair. This should be a minor change and
also should be backward compatible (there is a lot of space in the
struct btrfs_ioctl_dev_info_args )

diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c
index 14f8e1f..79fdd83 100644
--- a/fs/btrfs/ioctl.c
+++ b/fs/btrfs/ioctl.c
@@ -2261,6 +2261,8 @@ static long btrfs_ioctl_dev_info(struct btrfs_root
*root,
        di_args->devid = dev->devid;
        di_args->bytes_used = dev->bytes_used;
        di_args->total_bytes = dev->total_bytes;
+       di_args->major = imajor(dev->bdev->bd_inode);
+       di_args->minor = iminor(dev->bdev->bd_inode);
        memcpy(di_args->uuid, dev->uuid, sizeof(di_args->uuid));
        if (dev->name)
                strncpy(di_args->path, dev->name, sizeof(di_args->path));
diff --git a/fs/btrfs/ioctl.h b/fs/btrfs/ioctl.h
index 086e6bd..7afa688 100644
--- a/fs/btrfs/ioctl.h
+++ b/fs/btrfs/ioctl.h
@@ -98,7 +98,9 @@ struct btrfs_ioctl_dev_info_args {
        __u8 uuid[BTRFS_UUID_SIZE];             /* in/out */
        __u64 bytes_used;                       /* out */
        __u64 total_bytes;                      /* out */
-       __u64 unused[379];                      /* pad to 4k */
+       __u64 major;                            /* out */
+       __u64 minor;                            /* out */
+       __u64 unused[377];                      /* pad to 4k */
        __u8 path[BTRFS_DEVICE_PATH_NAME_MAX];  /* out */
 };



As long term solution, exporting all this kind of information in
/sys/btrfs/... could be a more robust solution, which could simplify the
"backward compatible" problem.

Only my 2¢...

GB

> 
> You want to map from /some_dir to a definitive list of devices you need
> to find in syslinux to later boot off that FS, right?
> 
> -chris
> --
> To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> .
>