.\" Copyright (c) 2016, IBM Corporation. .\" Written by Mike Rapoport .\" and Copyright (C) 2016 Michael Kerrisk .\" .\" SPDX-License-Identifier: Linux-man-pages-copyleft .\" .\" .TH ioctl_userfaultfd 2 2024-03-03 "Linux man-pages 6.7" .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 " " /* Definition of " UFFD* " constants */" .B #include .P .BI "int ioctl(int " fd ", int " op ", ...);" .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: .P .in +4n .EX ioctl(fd, op, argp); .EE .in .P In the above, .I fd is a file descriptor referring to a userfaultfd object, .I op is one of the operations listed below, and .I argp is a pointer to a data structure that is specific to .IR op . .P 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. .P The .I argp argument is a pointer to a .I uffdio_api structure, defined as: .P .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 .P The .I api field denotes the API version requested by the application. 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. .P Since Linux 4.11, applications should use the .I features field to perform a two-step handshake. First, .B UFFDIO_API is called with the .I features field set to zero. The kernel responds by setting all supported feature bits. .P Applications which do not require any specific features can begin using the userfaultfd immediately. Applications which do need specific features should call .B UFFDIO_API again with a subset of the reported feature bits set to enable those features. .P 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). .P If the application sets unsupported feature bits, the kernel will zero out the returned .I uffdio_api structure and return .BR EINVAL . .P 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_PAGEFAULT_FLAG_WP " (since Linux 5.10)" If this feature bit is set, userfaultfd supports write-protect faults for anonymous memory. (Note that shmem / hugetlbfs support is indicated by a separate feature.) .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. .TP .BR UFFD_FEATURE_WP_HUGETLBFS_SHMEM " (since Linux 5.19)" If this feature bit is set, userfaultfd supports write-protect faults for hugetlbfs and shmem / tmpfs memory. .TP .BR UFFD_FEATURE_WP_UNPOPULATED " (since Linux 6.4)" If this feature bit is set, the kernel will handle anonymous memory the same way as file memory, by allowing the user to write-protect unpopulated page table entries. .TP .BR UFFD_FEATURE_POISON " (since Linux 6.6)" If this feature bit is set, the kernel supports resolving faults with the .B UFFDIO_POISON ioctl. .TP .BR UFFD_FEATURE_WP_ASYNC " (since Linux 6.7)" If this feature bit is set, the write protection faults would be asynchronously resolved by the kernel. .P 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. .P This .BR ioctl (2) operation returns 0 on success. On error, \-1 is returned and .I errno is set to indicate the error. If an error occurs, the kernel may zero the provided .I uffdio_api structure. The caller should treat its contents as unspecified, and reinitialize it before re-attempting another .B UFFDIO_API call. 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 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. .TP .B EINVAL A previous .B UFFDIO_API call already enabled one or more features for this userfaultfd. Calling .B UFFDIO_API twice, the first time with no features set, is explicitly allowed as per the two-step feature detection handshake. .TP .B EPERM The .B UFFD_FEATURE_EVENT_FORK feature was enabled, but the calling process doesn't have the .B CAP_SYS_PTRACE capability. .SS UFFDIO_REGISTER (Since Linux 4.3.) Register a memory address range with the userfaultfd object. The pages in the range must be \[lq]compatible\[rq]. Please refer to the list of register modes below for the compatible memory backends for each mode. .P The .I argp argument is a pointer to a .I uffdio_register structure, defined as: .P .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 .P 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. .P 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. .P 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 operation is supported. .TP .B 1 << _UFFDIO_ZEROPAGE The .B UFFDIO_ZEROPAGE operation is supported. .TP .B 1 << _UFFDIO_CONTINUE The .B UFFDIO_CONTINUE operation is supported. .TP .B 1 << _UFFDIO_POISON The .B UFFDIO_POISON operation is supported. .P 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 \[lq]compatible\[rq] (see the description of .BR UFFDIO_REGISTER .) .P The address range to unregister is specified in the .I uffdio_range structure pointed to by .IR argp . .P 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 , .IR dst , and .I len fields of the .I uffdio_copy structure pointed to by .IR argp : .P .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 .P 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. .P 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. .P 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. .P The requested range is specified by the .I range field of the .I uffdio_zeropage structure pointed to by .IR argp : .P .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 .P 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. .P 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. .P 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. .P 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 . .P The .I argp argument is a pointer to a .I uffdio_range structure (shown above) that specifies the address range. .P 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 . .P The .I argp argument is a pointer to a .I uffdio_range structure as shown below: .P .in +4n .EX struct uffdio_writeprotect { struct uffdio_range range; /* Range to change write permission*/ __u64 mode; /* Mode to change write permission */ }; .EE .in .P 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. .P 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. .P The .I argp argument is a pointer to a .I uffdio_continue structure as shown below: .P .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 .P 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. .P 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. .P 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 EEXIST One or more pages were already mapped in the given range. .TP .B EFAULT No existing page could be found in the page cache for the given range. .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 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 ESRCH The faulting process has exited at the time of a .B UFFDIO_CONTINUE operation. .\" .SS UFFDIO_POISON (Since Linux 6.6.) Mark an address range as "poisoned". Future accesses to these addresses will raise a .B SIGBUS signal. Unlike .B MADV_HWPOISON this works by installing page table entries, rather than "really" poisoning the underlying physical pages. This means it only affects this particular address space. .P The .I argp argument is a pointer to a .I uffdio_poison structure as shown below: .P .in +4n .EX struct uffdio_poison { struct uffdio_range range; /* Range to install poison PTE markers in */ __u64 mode; /* Flags controlling the behavior of poison */ __s64 updated; /* Number of bytes poisoned, or negated error */ }; .EE .in .P The following value may be bitwise ORed in .I mode to change the behavior of the .B UFFDIO_POISON operation: .TP .B UFFDIO_POISON_MODE_DONTWAKE Do not wake up the thread that waits for page-fault resolution. .P The .I updated field is used by the kernel to return the number of bytes that were actually poisoned, or an error in the same manner as .BR UFFDIO_COPY . If the value returned in the .I updated field doesn't match the value that was specified in .IR range.len , the operation fails with the error .BR EAGAIN . The .I updated field is output-only; it is not read by the .B UFFDIO_POISON operation. .P This .BR ioctl (2) operation returns 0 on success. In this case, the entire area was poisoned. 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 updated 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_POISON operation. .TP .B ENOMEM Allocating memory for page table entries failed. .TP .B ESRCH The faulting process has exited at the time of a .B UFFDIO_POISON 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) .P .I Documentation/admin\-guide/mm/userfaultfd.rst in the Linux kernel source tree