Hi Suren, On Wed, Apr 23, 2025 at 12:53:09PM -0700, Suren Baghdasaryan wrote: > Documentation was extracted from the original patch written by Andrea > Arcangeli and upstreamed in [1]. Minor edits were made to maintain > the same documentation style as other userfaultfd ioctl commands. > > [1] > > Signed-off-by: Suren Baghdasaryan I've formatted references to fork(2). I also fixed a few places with s/BR/B/. Other than that, the patch was fine; I've applied it. Thanks! Below is the diff with which I amended the patch. Have a lovely day! Alex diff --git i/man/man2const/UFFDIO_MOVE.2const w/man/man2const/UFFDIO_MOVE.2const index 77b0ca781..daa122b30 100644 --- i/man/man2const/UFFDIO_MOVE.2const +++ w/man/man2const/UFFDIO_MOVE.2const @@ -108,36 +108,42 @@ .SH ERRORS .I .mode field. .TP -.BR ENOENT +.B ENOENT The source virtual memory range has unmapped holes and .B UFFDIO_MOVE_MODE_ALLOW_SRC_HOLES is not set. .TP -.BR EEXIST +.B EEXIST The destination virtual memory range is fully or partially mapped. .TP -.BR EBUSY +.B EBUSY The pages in the source virtual memory range are either pinned or not exclusive to the process. Once KSM deduplicates pages -or fork() COW-shares pages during fork() with child processes, +or +.BR fork (2) +COW-shares pages during +.BR fork (2) +with child processes, they are no longer exclusive. The kernel might only perform lightweight checks for detecting whether the pages are exclusive. To make the operation more likely to succeed, KSM should be disabled, -fork() should be avoided +.BR fork (2) +should be avoided or .B MADV_DONTFORK should be configured for the source virtual memory area -before fork(). +before +.BR fork (2). .TP -.BR ENOMEM +.B ENOMEM Allocating memory needed for the operation failed. .TP -.BR ESRCH +.B ESRCH The target process has exited at the time of a .B UFFDIO_MOVE operation. > --- > Changes since v1[1] > - removed '\" t, per Alejandro Colomar > - reformated sentences to use semantic newlines, per Alejandro Colomar > - changed field names to use '.' prefix, per Alejandro Colomar > - changed EBUSY, UFFDIO_MOVE and MADV_DONTFORK to be bold, > per Alejandro Colomar > - folded duplicate information into EBUSY error description, > per Alejandro Colomar > > [1] https://lore.kernel.org/all/20250423011203.2559210-1-surenb@google.com/ > > man/man2/ioctl_userfaultfd.2 | 2 + > man/man2const/UFFDIO_MOVE.2const | 153 +++++++++++++++++++++++++++++++ > 2 files changed, 155 insertions(+) > create mode 100644 man/man2const/UFFDIO_MOVE.2const > > diff --git a/man/man2/ioctl_userfaultfd.2 b/man/man2/ioctl_userfaultfd.2 > index 3cb1b8305..5ec08ca55 100644 > --- a/man/man2/ioctl_userfaultfd.2 > +++ b/man/man2/ioctl_userfaultfd.2 > @@ -69,6 +69,8 @@ events. > .TQ > .BR UFFDIO_COPY (2const) > .TQ > +.BR UFFDIO_MOVE (2const) > +.TQ > .BR UFFDIO_ZEROPAGE (2const) > .TQ > .BR UFFDIO_WAKE (2const) > diff --git a/man/man2const/UFFDIO_MOVE.2const b/man/man2const/UFFDIO_MOVE.2const > new file mode 100644 > index 000000000..77b0ca781 > --- /dev/null > +++ b/man/man2const/UFFDIO_MOVE.2const > @@ -0,0 +1,153 @@ > +.\" Written by Andrea Arcangeli > +.\" > +.\" SPDX-License-Identifier: Linux-man-pages-copyleft > +.\" > +.TH UFFDIO_MOVE 2const (date) "Linux man-pages (unreleased)" > +.SH NAME > +UFFDIO_MOVE > +\- > +atomically move a continuous memory chunk into the userfault registered range > +.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 ", UFFDIO_MOVE, struct uffdio_move *" argp ); > +.P > +.B #include > +.P > +.fi > +.EX > +.B struct uffdio_move { > +.BR " __u64 dst;" " /* Destination of move */" > +.BR " __u64 src;" " /* Source of move */" > +.BR " __u64 len;" " /* Number of bytes to move */" > +.BR " __u64 mode;" " /* Flags controlling behavior of move */" > +.BR " __s64 move;" " /* Number of bytes moved, or negated error */" > +.B }; > +.EE > +.SH DESCRIPTION > +Atomically move a continuous memory chunk > +into the userfault registered range > +and optionally wake up the blocked thread. > +.P > +The following value may be bitwise ORed in > +.I .mode > +to change the behavior of the > +.B UFFDIO_MOVE > +operation: > +.TP > +.B UFFDIO_MOVE_MODE_DONTWAKE > +Do not wake up the thread that waits for page-fault resolution > +.TP > +.B UFFDIO_MOVE_MODE_ALLOW_SRC_HOLES > +Allow holes in the source virtual range that is being moved. > +When not specified, the holes will result in > +.B ENOENT > +error. > +When specified, > +the holes will be accounted as successfully moved memory. > +This is mostly useful > +to move hugepage aligned virtual regions > +without knowing if there are transparent hugepages in the regions or not, > +but preventing the risk of > +having to split the hugepage during the operation. > +.P > +The > +.I .move > +field is used by the kernel > +to return the number of bytes that was actually moved, > +or an error > +(a negated > +.IR errno -style > +value). > +The > +.I .move > +field is output-only; > +it is not read by the > +.B UFFDIO_MOVE > +operation. > +.SH RETURN VALUE > +On success, > +0 is returned. > +In this case, the entire area was moved. > +.P > +On error, \-1 is returned and > +.I errno > +is set to indicate the error. > +.SH ERRORS > +.TP > +.B EAGAIN > +The number of bytes moved (i.e., the value returned in the > +.I .move > +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 > +The source virtual memory range has unmapped holes and > +.B UFFDIO_MOVE_MODE_ALLOW_SRC_HOLES > +is not set. > +.TP > +.BR EEXIST > +The destination virtual memory range is fully or partially > +mapped. > +.TP > +.BR EBUSY > +The pages in the source virtual memory range are either > +pinned or not exclusive to the process. > +Once KSM deduplicates pages > +or fork() COW-shares pages during fork() with child processes, > +they are no longer exclusive. > +The kernel might only perform lightweight checks > +for detecting whether the pages are exclusive. > +To make the operation more likely to succeed, > +KSM should be disabled, > +fork() should be avoided > +or > +.B MADV_DONTFORK > +should be configured > +for the source virtual memory area > +before fork(). > +.TP > +.BR ENOMEM > +Allocating memory needed for the operation failed. > +.TP > +.BR ESRCH > +The target process has exited at the time of a > +.B UFFDIO_MOVE > +operation. > +.SH STANDARDS > +Linux. > +.SH HISTORY > +Linux 6.8. > +.SH SEE ALSO > +.BR ioctl (2), > +.BR ioctl_userfaultfd (2), > +.BR userfaultfd (2) > +.P > +.I linux.git/\:Documentation/\:admin\-guide/\:mm/\:userfaultfd.rst > > base-commit: 80e2715270fc05d5627c26f88e4c1ba8b093f510 > -- > 2.49.0.805.g082f7c87e0-goog > --