This manpage written for the case where xfs or other filesystems will
implement the same ioctl.
Signed-off-by: Jeremy Bongio <[email protected]>
---
man2/ioctl_fsuuid.2 | 110 ++++++++++++++++++++++++++++++++++++++++++++
1 file changed, 110 insertions(+)
create mode 100644 man2/ioctl_fsuuid.2
diff --git a/man2/ioctl_fsuuid.2 b/man2/ioctl_fsuuid.2
new file mode 100644
index 000000000..db414bf59
--- /dev/null
+++ b/man2/ioctl_fsuuid.2
@@ -0,0 +1,110 @@
+.\" Copyright (c) 2022 Google, Inc., written by Jeremy Bongio <[email protected]>
+.\"
+.\" SPDX-License-Identifier: GPL-2.0-or-later
+.TH IOCTL_FSUUID 2 2022-07-08 "Linux" "Linux Programmer's Manual"
+.SH NAME
+ioctl_fsuuid \- get or set a filesystem label
+.SH LIBRARY
+Standard C library
+.RI ( libc ", " \-lc )
+.SH SYNOPSIS
+.nf
+.BR "#include <linux/fs.h>" " /* Definition of " *FSLABEL* " constants */"
+.BR "#include <uuid/uuid.h>" " /* Definition of " *UUID_MAX* " constants */"
+.B #include <sys/ioctl.h>
+.PP
+.BI "int ioctl(int " fd ", FS_IOC_GETFSUUID, struct " fsuuid ");"
+.BI "int ioctl(int " fd ", FS_IOC_SETFSUUID, struct " fsuuid ");"
+.fi
+.SH DESCRIPTION
+If a filesystem supports online uuid manipulation, these
+.BR ioctl (2)
+operations can be used to get or set the uuid for the filesystem
+on which
+.I fd
+resides.
+.PP
+The
+.B FS_IOC_GETFSUUID
+operation will read the filesystem uuid into
+.I fu_uuid.
+.I fu_len
+must be set to the number of bytes allocated for the uuid.
+.I fu_uuid
+must be initialized to the size of the filesystem uuid.
+The maximmum number of bytes for a uuid is
+.BR UUID_MAX.
+.I fu_flags
+must be set to 0.
+.PP
+The
+.B FS_IOC_SETFSUUID
+operation will set the filesystem uuid according to
+.I fu_uuid.
+.I fu_len
+must be set to the number of bytes in the uuid.
+.I fu_flags
+must be set to 0. The
+.B FS_IOC_SETFSUUID
+operation requires privilege
+.RB ( CAP_SYS_ADMIN ).
+.PP
+This information is conveyed in a structure of
+the following form:
+.PP
+.in +4n
+.EX
+struct fsuuid {
+ __u32 fu_len;
+ __u32 fu_flags;
+ __u8 fu_uuid[];
+};
+.EE
+.in
+.PP
+.SH RETURN VALUE
+On success zero is returned.
+On error, \-1 is returned, and
+.I errno
+is set to indicate the error.
+.SH ERRORS
+Possible errors include (but are not limited to) the following:
+.TP
+.B EFAULT
+Either the pointer to the
+.I fsuuid
+structure is invalid or
+.I fu_uuid
+has not been initialized properly.
+.TP
+.B EINVAL
+The specified arguments are invalid.
+.I fu_len
+does not match the filesystem uuid length or
+.I fu_flags
+has bits set that are not implemented.
+.TP
+.B ENOTTY
+The filesystem does not support the ioctl.
+.TP
+.B EOPNOTSUPP
+The filesystem does not currently support changing the uuid through this
+ioctl. This may be due to incompatible feature flags.
+.TP
+.B EPERM
+The calling process does not have sufficient permissions to set the uuid.
+.SH VERSIONS
+These
+.BR ioctl (2)
+operations first appeared in Linux 5.19.
+They were previously known as
+.B EXT4_IOC_GETFSUUID
+and
+.B EXT4_IOC_SETFSUUID
+and were private to ext4.
+.SH CONFORMING TO
+This API is Linux-specific.
+.BR UUID_MAX.
+.SH SEE ALSO
+.BR ioctl (2)
+
--
2.37.0.144.g8ac04bfd2-goog
On Tue, Jul 12, 2022 at 03:56:53PM -0700, Jeremy Bongio wrote:
> This manpage written for the case where xfs or other filesystems will
> implement the same ioctl.
You might want to cc the manpages project ([email protected]),
since this is likely to end up in there some day.
> Signed-off-by: Jeremy Bongio <[email protected]>
> ---
> man2/ioctl_fsuuid.2 | 110 ++++++++++++++++++++++++++++++++++++++++++++
> 1 file changed, 110 insertions(+)
> create mode 100644 man2/ioctl_fsuuid.2
>
> diff --git a/man2/ioctl_fsuuid.2 b/man2/ioctl_fsuuid.2
> new file mode 100644
> index 000000000..db414bf59
> --- /dev/null
> +++ b/man2/ioctl_fsuuid.2
> @@ -0,0 +1,110 @@
> +.\" Copyright (c) 2022 Google, Inc., written by Jeremy Bongio <[email protected]>
> +.\"
> +.\" SPDX-License-Identifier: GPL-2.0-or-later
> +.TH IOCTL_FSUUID 2 2022-07-08 "Linux" "Linux Programmer's Manual"
> +.SH NAME
> +ioctl_fsuuid \- get or set a filesystem label
> +.SH LIBRARY
> +Standard C library
> +.RI ( libc ", " \-lc )
> +.SH SYNOPSIS
> +.nf
> +.BR "#include <linux/fs.h>" " /* Definition of " *FSLABEL* " constants */"
> +.BR "#include <uuid/uuid.h>" " /* Definition of " *UUID_MAX* " constants */"
Hmm... that's in libuuid, and not everyone's going to want to link with
that. You might consider defining EXT4_UUID_SIZE in the same place as
the ioctl definitions, or putting it in libext2fs somewhere.
> +.B #include <sys/ioctl.h>
> +.PP
> +.BI "int ioctl(int " fd ", FS_IOC_GETFSUUID, struct " fsuuid ");"
> +.BI "int ioctl(int " fd ", FS_IOC_SETFSUUID, struct " fsuuid ");"
> +.fi
> +.SH DESCRIPTION
> +If a filesystem supports online uuid manipulation, these
No need to say 'online', ioctls don't work on unmounted filesystems.
> +.BR ioctl (2)
> +operations can be used to get or set the uuid for the filesystem
> +on which
> +.I fd
> +resides.
> +.PP
> +The
> +.B FS_IOC_GETFSUUID
> +operation will read the filesystem uuid into
> +.I fu_uuid.
> +.I fu_len
> +must be set to the number of bytes allocated for the uuid.
> +.I fu_uuid
> +must be initialized to the size of the filesystem uuid.
Huh?
> +The maximmum number of bytes for a uuid is
> +.BR UUID_MAX.
"The number of bytes to allocate for the UUID is filesystem-dependent."?
> +.I fu_flags
> +must be set to 0.
> +.PP
> +The
> +.B FS_IOC_SETFSUUID
> +operation will set the filesystem uuid according to
> +.I fu_uuid.
> +.I fu_len
> +must be set to the number of bytes in the uuid.
> +.I fu_flags
> +must be set to 0. The
> +.B FS_IOC_SETFSUUID
> +operation requires privilege
> +.RB ( CAP_SYS_ADMIN ).
> +.PP
> +This information is conveyed in a structure of
> +the following form:
> +.PP
> +.in +4n
> +.EX
> +struct fsuuid {
> + __u32 fu_len;
> + __u32 fu_flags;
> + __u8 fu_uuid[];
> +};
Question: Would it perhaps make more sense to describe the struct and
what goes in fu_len/fu_uuid first, and then describe what get and set
do?
> +.EE
> +.in
> +.PP
> +.SH RETURN VALUE
> +On success zero is returned.
> +On error, \-1 is returned, and
> +.I errno
> +is set to indicate the error.
> +.SH ERRORS
> +Possible errors include (but are not limited to) the following:
> +.TP
> +.B EFAULT
> +Either the pointer to the
> +.I fsuuid
> +structure is invalid or
> +.I fu_uuid
> +has not been initialized properly.
> +.TP
> +.B EINVAL
> +The specified arguments are invalid.
> +.I fu_len
> +does not match the filesystem uuid length or
> +.I fu_flags
> +has bits set that are not implemented.
> +.TP
> +.B ENOTTY
> +The filesystem does not support the ioctl.
> +.TP
> +.B EOPNOTSUPP
> +The filesystem does not currently support changing the uuid through this
> +ioctl. This may be due to incompatible feature flags.
> +.TP
> +.B EPERM
> +The calling process does not have sufficient permissions to set the uuid.
> +.SH VERSIONS
> +These
> +.BR ioctl (2)
> +operations first appeared in Linux 5.19.
5.20 at this point...
> +They were previously known as
> +.B EXT4_IOC_GETFSUUID
> +and
> +.B EXT4_IOC_SETFSUUID
Oh! I had assumed you'd just leave these as "ext4" ioctls for now and
let whoever does the next filesystem port add these bits.
> +and were private to ext4.
> +.SH CONFORMING TO
> +This API is Linux-specific.
> +.BR UUID_MAX.
> +.SH SEE ALSO
> +.BR ioctl (2)
> +
> --
> 2.37.0.144.g8ac04bfd2-goog
>