path: root/man2/ioctl_userfaultfd.2

.\" Copyright (c) 2016, IBM Corporation.
.\" Written by Mike Rapoport <rppt@linux.vnet.ibm.com>
.\" and Copyright (C) 2016 Michael Kerrisk <mtk.manpages@gmail.com>
.\"
.\" SPDX-License-Identifier: Linux-man-pages-copyleft
.\"
.\"
.TH ioctl_userfaultfd 2 (date) "Linux man-pages (unreleased)"
.SH NAME
ioctl_userfaultfd \- create a file descriptor for handling page faults in user
space
.SH LIBRARY
Standard C library
.RI ( libc ", " \-lc )
.SH SYNOPSIS
.nf
.BR "#include <linux/userfaultfd.h>" "  /* Definition of " UFFD* " constants */"
.B #include <sys/ioctl.h>
.PP
.BI "int ioctl(int " fd ", int " cmd ", ...);"
.fi
.SH DESCRIPTION
Various
.BR ioctl (2)
operations can be performed on a userfaultfd object (created by a call to
.BR userfaultfd (2))
using calls of the form:
.PP
.in +4n
.EX
ioctl(fd, cmd, argp);
.EE
.in
In the above,
.I fd
is a file descriptor referring to a userfaultfd object,
.I cmd
is one of the commands listed below, and
.I argp
is a pointer to a data structure that is specific to
.IR cmd .
.PP
The various
.BR ioctl (2)
operations are described below.
The
.BR UFFDIO_API ,
.BR UFFDIO_REGISTER ,
and
.B UFFDIO_UNREGISTER
operations are used to
.I configure
userfaultfd behavior.
These operations allow the caller to choose what features will be enabled and
what kinds of events will be delivered to the application.
The remaining operations are
.I range
operations.
These operations enable the calling application to resolve page-fault
events.
.\"
.SS UFFDIO_API
(Since Linux 4.3.)
Enable operation of the userfaultfd and perform API handshake.
.PP
The
.I argp
argument is a pointer to a
.I uffdio_api
structure, defined as:
.PP
.in +4n
.EX
struct uffdio_api {
    __u64 api;        /* Requested API version (input) */
    __u64 features;   /* Requested features (input/output) */
    __u64 ioctls;     /* Available ioctl() operations (output) */
};
.EE
.in
.PP
The
.I api
field denotes the API version requested by the application.
.PP
The kernel verifies that it can support the requested API version,
and sets the
.I features
and
.I ioctls
fields to bit masks representing all the available features and the generic
.BR ioctl (2)
operations available.
.PP
Before Linux 4.11, the
.I features
field must be initialized to zero before the call to
.BR UFFDIO_API ,
and zero (i.e., no feature bits) is placed in the
.I features
field by the kernel upon return from
.BR ioctl (2).
.PP
Starting from Linux 4.11, the
.I features
field can be used to ask whether particular features are supported
and explicitly enable userfaultfd features that are disabled by default.
The kernel always reports all the available features in the
.I features
field.
.PP
To enable userfaultfd features the application should set
a bit corresponding to each feature it wants to enable in the
.I features
field.
If the kernel supports all the requested features it will enable them.
Otherwise it will zero out the returned
.I uffdio_api
structure and return
.BR EINVAL .
.\" FIXME add more details about feature negotiation and enablement
.PP
The following feature bits may be set:
.TP
.BR UFFD_FEATURE_EVENT_FORK " (since Linux 4.11)"
When this feature is enabled,
the userfaultfd objects associated with a parent process are duplicated
into the child process during
.BR fork (2)
and a
.B UFFD_EVENT_FORK
event is delivered to the userfaultfd monitor
.TP
.BR UFFD_FEATURE_EVENT_REMAP " (since Linux 4.11)"
If this feature is enabled,
when the faulting process invokes
.BR mremap (2),
the userfaultfd monitor will receive an event of type
.BR UFFD_EVENT_REMAP .
.TP
.BR UFFD_FEATURE_EVENT_REMOVE " (since Linux 4.11)"
If this feature is enabled,
when the faulting process calls
.BR madvise (2)
with the
.B MADV_DONTNEED
or
.B MADV_REMOVE
advice value to free a virtual memory area
the userfaultfd monitor will receive an event of type
.BR UFFD_EVENT_REMOVE .
.TP
.BR UFFD_FEATURE_EVENT_UNMAP " (since Linux 4.11)"
If this feature is enabled,
when the faulting process unmaps virtual memory either explicitly with
.BR munmap (2),
or implicitly during either
.BR mmap (2)
or
.BR mremap (2),
the userfaultfd monitor will receive an event of type
.BR UFFD_EVENT_UNMAP .
.TP
.BR UFFD_FEATURE_MISSING_HUGETLBFS " (since Linux 4.11)"
If this feature bit is set,
the kernel supports registering userfaultfd ranges on hugetlbfs
virtual memory areas
.TP
.BR UFFD_FEATURE_MISSING_SHMEM " (since Linux 4.11)"
If this feature bit is set,
the kernel supports registering userfaultfd ranges on shared memory areas.
This includes all kernel shared memory APIs:
System V shared memory,
.BR tmpfs (5),
shared mappings of
.IR /dev/zero ,
.BR mmap (2)
with the
.B MAP_SHARED
flag set,
.BR memfd_create (2),
and so on.
.TP
.BR UFFD_FEATURE_SIGBUS " (since Linux 4.14)"
.\" commit 2d6d6f5a09a96cc1fec7ed992b825e05f64cb50e
If this feature bit is set, no page-fault events
.RB ( UFFD_EVENT_PAGEFAULT )
will be delivered.
Instead, a
.B SIGBUS
signal will be sent to the faulting process.
Applications using this
feature will not require the use of a userfaultfd monitor for processing
memory accesses to the regions registered with userfaultfd.
.TP
.BR UFFD_FEATURE_THREAD_ID " (since Linux 4.14)"
If this feature bit is set,
.I uffd_msg.pagefault.feat.ptid
will be set to the faulted thread ID for each page-fault message.
.TP
.BR UFFD_FEATURE_MINOR_HUGETLBFS " (since Linux 5.13)"
If this feature bit is set,
the kernel supports registering userfaultfd ranges
in minor mode on hugetlbfs-backed memory areas.
.TP
.BR UFFD_FEATURE_MINOR_SHMEM " (since Linux 5.14)"
If this feature bit is set,
the kernel supports registering userfaultfd ranges
in minor mode on shmem-backed memory areas.
.TP
.BR UFFD_FEATURE_EXACT_ADDRESS " (since Linux 5.18)"
If this feature bit is set,
.I uffd_msg.pagefault.address
will be set to the exact page-fault address that was reported by the hardware,
and will not mask the offset within the page.
Note that old Linux versions might indicate the exact address as well,
even though the feature bit is not set.
.PP
The returned
.I ioctls
field can contain the following bits:
.\" FIXME This user-space API seems not fully polished. Why are there
.\" not constants defined for each of the bit-mask values listed below?
.TP
.B 1 << _UFFDIO_API
The
.B UFFDIO_API
operation is supported.
.TP
.B 1 << _UFFDIO_REGISTER
The
.B UFFDIO_REGISTER
operation is supported.
.TP
.B 1 << _UFFDIO_UNREGISTER
The
.B UFFDIO_UNREGISTER
operation is supported.
.PP
This
.BR ioctl (2)
operation returns 0 on success.
On error, \-1 is returned and
.I errno
is set to indicate the error.
Possible errors include:
.TP
.B EFAULT
.I argp
refers to an address that is outside the calling process's
accessible address space.
.TP
.B EINVAL
The userfaultfd has already been enabled by a previous
.B UFFDIO_API
operation.
.TP
.B EINVAL
The API version requested in the
.I api
field is not supported by this kernel, or the
.I features
field passed to the kernel includes feature bits that are not supported
by the current kernel version.
.\" FIXME In the above error case, the returned 'uffdio_api' structure is
.\" zeroed out. Why is this done? This should be explained in the manual page.
.\"
.\" Mike Rapoport:
.\"     In my understanding the uffdio_api
.\"     structure is zeroed to allow the caller
.\"     to distinguish the reasons for -EINVAL.
.\"
.SS UFFDIO_REGISTER
(Since Linux 4.3.)
Register a memory address range with the userfaultfd object.
The pages in the range must be "compatible".
Please refer to the list of register modes below
for the compatible memory backends for each mode.
.PP
The
.I argp
argument is a pointer to a
.I uffdio_register
structure, defined as:
.PP
.in +4n
.EX
struct uffdio_range {
    __u64 start;    /* Start of range */
    __u64 len;      /* Length of range (bytes) */
};

struct uffdio_register {
    struct uffdio_range range;
    __u64 mode;     /* Desired mode of operation (input) */
    __u64 ioctls;   /* Available ioctl() operations (output) */
};
.EE
.in
.PP
The
.I range
field defines a memory range starting at
.I start
and continuing for
.I len
bytes that should be handled by the userfaultfd.
.PP
The
.I mode
field defines the mode of operation desired for this memory region.
The following values may be bitwise ORed to set the userfaultfd mode for
the specified range:
.TP
.B UFFDIO_REGISTER_MODE_MISSING
Track page faults on missing pages.
Since Linux 4.3,
only private anonymous ranges are compatible.
Since Linux 4.11,
hugetlbfs and shared memory ranges are also compatible.
.TP
.B UFFDIO_REGISTER_MODE_WP
Track page faults on write-protected pages.
Since Linux 5.7,
only private anonymous ranges are compatible.
.TP
.B UFFDIO_REGISTER_MODE_MINOR
Track minor page faults.
Since Linux 5.13,
only hugetlbfs ranges are compatible.
Since Linux 5.14,
compatibility with shmem ranges was added.
.PP
If the operation is successful, the kernel modifies the
.I ioctls
bit-mask field to indicate which
.BR ioctl (2)
operations are available for the specified range.
This returned bit mask can contain the following bits:
.TP
.B 1 << _UFFDIO_COPY
The
.B UFFDIO_COPY
operation is supported.
.TP
.B 1 << _UFFDIO_WAKE
The
.B UFFDIO_WAKE
operation is supported.
.TP
.B 1 << _UFFDIO_WRITEPROTECT
The
.B UFFDIO_WRITEPROTECT
.TP
.B 1 << _UFFDIO_ZEROPAGE
The
.B UFFDIO_ZEROPAGE
operation is supported.
.TP
.B 1 << _UFFDIO_CONTINUE
The
.B UFFDIO_CONTINUE
operation is supported.
.PP
This
.BR ioctl (2)
operation returns 0 on success.
On error, \-1 is returned and
.I errno
is set to indicate the error.
Possible errors include:
.\" FIXME Is the following error list correct?
.\"
.TP
.B EBUSY
A mapping in the specified range is registered with another
userfaultfd object.
.TP
.B EFAULT
.I argp
refers to an address that is outside the calling process's
accessible address space.
.TP
.B EINVAL
An invalid or unsupported bit was specified in the
.I mode
field; or the
.I mode
field was zero.
.TP
.B EINVAL
There is no mapping in the specified address range.
.TP
.B EINVAL
.I range.start
or
.I range.len
is not a multiple of the system page size; or,
.I range.len
is zero; or these fields are otherwise invalid.
.TP
.B EINVAL
There as an incompatible mapping in the specified address range.
.\" Mike Rapoport:
.\" ENOMEM if the process is exiting and the
.\" mm_struct has gone by the time userfault grabs it.
.SS UFFDIO_UNREGISTER
(Since Linux 4.3.)
Unregister a memory address range from userfaultfd.
The pages in the range must be "compatible" (see the description of
.BR  UFFDIO_REGISTER .)
.PP
The address range to unregister is specified in the
.I uffdio_range
structure pointed to by
.IR argp .
.PP
This
.BR ioctl (2)
operation returns 0 on success.
On error, \-1 is returned and
.I errno
is set to indicate the error.
Possible errors include:
.TP
.B EINVAL
Either the
.I start
or the
.I len
field of the
.I ufdio_range
structure was not a multiple of the system page size; or the
.I len
field was zero; or these fields were otherwise invalid.
.TP
.B EINVAL
There as an incompatible mapping in the specified address range.
.TP
.B EINVAL
There was no mapping in the specified address range.
.\"
.SS UFFDIO_COPY
(Since Linux 4.3.)
Atomically copy a continuous memory chunk into the userfault registered
range and optionally wake up the blocked thread.
The source and destination addresses and the number of bytes to copy are
specified by the
.IR src ", " dst ", and " len
fields of the
.I uffdio_copy
structure pointed to by
.IR argp :
.PP
.in +4n
.EX
struct uffdio_copy {
    __u64 dst;    /* Destination of copy */
    __u64 src;    /* Source of copy */
    __u64 len;    /* Number of bytes to copy */
    __u64 mode;   /* Flags controlling behavior of copy */
    __s64 copy;   /* Number of bytes copied, or negated error */
};
.EE
.in
.PP
The following value may be bitwise ORed in
.I mode
to change the behavior of the
.B UFFDIO_COPY
operation:
.TP
.B UFFDIO_COPY_MODE_DONTWAKE
Do not wake up the thread that waits for page-fault resolution
.TP
.B UFFDIO_COPY_MODE_WP
Copy the page with read-only permission.
This allows the user to trap the next write to the page,
which will block and generate another write-protect userfault message.
This is used only when both
.B UFFDIO_REGISTER_MODE_MISSING
and
.B UFFDIO_REGISTER_MODE_WP
modes are enabled for the registered range.
.PP
The
.I copy
field is used by the kernel to return the number of bytes
that was actually copied, or an error (a negated
.IR errno -style
value).
.\" FIXME Above: Why is the 'copy' field used to return error values?
.\" This should be explained in the manual page.
If the value returned in
.I copy
doesn't match the value that was specified in
.IR len ,
the operation fails with the error
.BR EAGAIN .
The
.I copy
field is output-only;
it is not read by the
.B UFFDIO_COPY
operation.
.PP
This
.BR ioctl (2)
operation returns 0 on success.
In this case, the entire area was copied.
On error, \-1 is returned and
.I errno
is set to indicate the error.
Possible errors include:
.TP
.B EAGAIN
The number of bytes copied (i.e., the value returned in the
.I copy
field)
does not equal the value that was specified in the
.I len
field.
.TP
.B EINVAL
Either
.I dst
or
.I len
was not a multiple of the system page size, or the range specified by
.I src
and
.I len
or
.I dst
and
.I len
was invalid.
.TP
.B EINVAL
An invalid bit was specified in the
.I mode
field.
.TP
.BR ENOENT " (since Linux 4.11)"
The faulting process has changed
its virtual memory layout simultaneously with an outstanding
.B UFFDIO_COPY
operation.
.TP
.BR ENOSPC " (from Linux 4.11 until Linux 4.13)"
The faulting process has exited at the time of a
.B UFFDIO_COPY
operation.
.TP
.BR ESRCH " (since Linux 4.13)"
The faulting process has exited at the time of a
.B UFFDIO_COPY
operation.
.\"
.SS UFFDIO_ZEROPAGE
(Since Linux 4.3.)
Zero out a memory range registered with userfaultfd.
.PP
The requested range is specified by the
.I range
field of the
.I uffdio_zeropage
structure pointed to by
.IR argp :
.PP
.in +4n
.EX
struct uffdio_zeropage {
    struct uffdio_range range;
    __u64 mode;     /* Flags controlling behavior of copy */
    __s64 zeropage; /* Number of bytes zeroed, or negated error */
};
.EE
.in
.PP
The following value may be bitwise ORed in
.I mode
to change the behavior of the
.B UFFDIO_ZEROPAGE
operation:
.TP
.B UFFDIO_ZEROPAGE_MODE_DONTWAKE
Do not wake up the thread that waits for page-fault resolution.
.PP
The
.I zeropage
field is used by the kernel to return the number of bytes
that was actually zeroed,
or an error in the same manner as
.BR UFFDIO_COPY .
.\" FIXME Why is the 'zeropage' field used to return error values?
.\" This should be explained in the manual page.
If the value returned in the
.I zeropage
field doesn't match the value that was specified in
.IR range.len ,
the operation fails with the error
.BR EAGAIN .
The
.I zeropage
field is output-only;
it is not read by the
.B UFFDIO_ZEROPAGE
operation.
.PP
This
.BR ioctl (2)
operation returns 0 on success.
In this case, the entire area was zeroed.
On error, \-1 is returned and
.I errno
is set to indicate the error.
Possible errors include:
.TP
.B EAGAIN
The number of bytes zeroed (i.e., the value returned in the
.I zeropage
field)
does not equal the value that was specified in the
.I range.len
field.
.TP
.B EINVAL
Either
.I range.start
or
.I range.len
was not a multiple of the system page size; or
.I range.len
was zero; or the range specified was invalid.
.TP
.B EINVAL
An invalid bit was specified in the
.I mode
field.
.TP
.BR ESRCH " (since Linux 4.13)"
The faulting process has exited at the time of a
.B UFFDIO_ZEROPAGE
operation.
.\"
.SS UFFDIO_WAKE
(Since Linux 4.3.)
Wake up the thread waiting for page-fault resolution on
a specified memory address range.
.PP
The
.B UFFDIO_WAKE
operation is used in conjunction with
.B UFFDIO_COPY
and
.B UFFDIO_ZEROPAGE
operations that have the
.B UFFDIO_COPY_MODE_DONTWAKE
or
.B UFFDIO_ZEROPAGE_MODE_DONTWAKE
bit set in the
.I mode
field.
The userfault monitor can perform several
.B UFFDIO_COPY
and
.B UFFDIO_ZEROPAGE
operations in a batch and then explicitly wake up the faulting thread using
.BR UFFDIO_WAKE .
.PP
The
.I argp
argument is a pointer to a
.I uffdio_range
structure (shown above) that specifies the address range.
.PP
This
.BR ioctl (2)
operation returns 0 on success.
On error, \-1 is returned and
.I errno
is set to indicate the error.
Possible errors include:
.TP
.B EINVAL
The
.I start
or the
.I len
field of the
.I ufdio_range
structure was not a multiple of the system page size; or
.I len
was zero; or the specified range was otherwise invalid.
.SS UFFDIO_WRITEPROTECT (Since Linux 5.7)
Write-protect or write-unprotect a userfaultfd-registered memory range
registered with mode
.BR UFFDIO_REGISTER_MODE_WP .
.PP
The
.I argp
argument is a pointer to a
.I uffdio_range
structure as shown below:
.PP
.in +4n
.EX
struct uffdio_writeprotect {
    struct uffdio_range range; /* Range to change write permission*/
    __u64 mode;                /* Mode to change write permission */
};
.EE
.in
.PP
There are two mode bits that are supported in this structure:
.TP
.B UFFDIO_WRITEPROTECT_MODE_WP
When this mode bit is set,
the ioctl will be a write-protect operation upon the memory range specified by
.IR range .
Otherwise it will be a write-unprotect operation upon the specified range,
which can be used to resolve a userfaultfd write-protect page fault.
.TP
.B UFFDIO_WRITEPROTECT_MODE_DONTWAKE
When this mode bit is set,
do not wake up any thread that waits for
page-fault resolution after the operation.
This can be specified only if
.B UFFDIO_WRITEPROTECT_MODE_WP
is not specified.
.PP
This
.BR ioctl (2)
operation returns 0 on success.
On error, \-1 is returned and
.I errno
is set to indicate the error.
Possible errors include:
.TP
.B EINVAL
The
.I start
or the
.I len
field of the
.I ufdio_range
structure was not a multiple of the system page size; or
.I len
was zero; or the specified range was otherwise invalid.
.TP
.B EAGAIN
The process was interrupted; retry this call.
.TP
.B ENOENT
The range specified in
.I range
is not valid.
For example, the virtual address does not exist,
or not registered with userfaultfd write-protect mode.
.TP
.B EFAULT
Encountered a generic fault during processing.
.\"
.SS UFFDIO_CONTINUE
(Since Linux 5.13.)
Resolve a minor page fault
by installing page table entries
for existing pages in the page cache.
.PP
The
.I argp
argument is a pointer to a
.I uffdio_continue
structure as shown below:
.PP
.in +4n
.EX
struct uffdio_continue {
    struct uffdio_range range;
                   /* Range to install PTEs for and continue */
    __u64 mode;    /* Flags controlling the behavior of continue */
    __s64 mapped;  /* Number of bytes mapped, or negated error */
};
.EE
.in
.PP
The following value may be bitwise ORed in
.I mode
to change the behavior of the
.B UFFDIO_CONTINUE
operation:
.TP
.B UFFDIO_CONTINUE_MODE_DONTWAKE
Do not wake up the thread that waits for page-fault resolution.
.PP
The
.I mapped
field is used by the kernel
to return the number of bytes that were actually mapped,
or an error in the same manner as
.BR UFFDIO_COPY .
If the value returned in the
.I mapped
field doesn't match the value that was specified in
.IR range.len ,
the operation fails with the error
.BR EAGAIN .
The
.I mapped
field is output-only;
it is not read by the
.B UFFDIO_CONTINUE
operation.
.PP
This
.BR ioctl (2)
operation returns 0 on success.
In this case,
the entire area was mapped.
On error, \-1 is returned and
.I errno
is set to indicate the error.
Possible errors include:
.TP
.B EAGAIN
The number of bytes mapped
(i.e., the value returned in the
.I mapped
field)
does not equal the value that was specified in the
.I range.len
field.
.TP
.B EINVAL
Either
.I range.start
or
.I range.len
was not a multiple of the system page size; or
.I range.len
was zero; or the range specified was invalid.
.TP
.B EINVAL
An invalid bit was specified in the
.I mode
field.
.TP
.B EEXIST
One or more pages were already mapped in the given range.
.TP
.B ENOENT
The faulting process has changed its virtual memory layout simultaneously with
an outstanding
.B UFFDIO_CONTINUE
operation.
.TP
.B ENOMEM
Allocating memory needed to setup the page table mappings failed.
.TP
.B EFAULT
No existing page could be found in the page cache for the given range.
.TP
.B ESRCH
The faulting process has exited at the time of a
.B UFFDIO_CONTINUE
operation.
.\"
.SH RETURN VALUE
See descriptions of the individual operations, above.
.SH ERRORS
See descriptions of the individual operations, above.
In addition, the following general errors can occur for all of the
operations described above:
.TP
.B EFAULT
.I argp
does not point to a valid memory address.
.TP
.B EINVAL
(For all operations except
.BR UFFDIO_API .)
The userfaultfd object has not yet been enabled (via the
.B UFFDIO_API
operation).
.SH STANDARDS
Linux.
.SH BUGS
In order to detect available userfault features and
enable some subset of those features
the userfaultfd file descriptor must be closed after the first
.B UFFDIO_API
operation that queries features availability and reopened before
the second
.B UFFDIO_API
operation that actually enables the desired features.
.SH EXAMPLES
See
.BR userfaultfd (2).
.SH SEE ALSO
.BR ioctl (2),
.BR mmap (2),
.BR userfaultfd (2)
.PP
.I Documentation/admin\-guide/mm/userfaultfd.rst
in the Linux kernel source tree