Document the new flags for mmap() and mlockall() and their behavior.
Inlcude a change to getrlimit(2) to cover interactions with
RLIMIT_MEMLOCK. These new flags will be introduced with the 4.2 kernel.
Signed-off-by: Eric B Munson <[email protected]>
Cc: Andrew Morton <[email protected]>
Cc: [email protected]
Cc: [email protected]
Cc: [email protected]
Cc: [email protected]
---
man2/getrlimit.2 | 9 +++++++--
man2/mlock.2 | 28 ++++++++++++++++++++++++++--
man2/mmap.2 | 21 +++++++++++++++++++++
3 files changed, 54 insertions(+), 4 deletions(-)
diff --git a/man2/getrlimit.2 b/man2/getrlimit.2
index ec464fe..729197e 100644
--- a/man2/getrlimit.2
+++ b/man2/getrlimit.2
@@ -215,7 +215,9 @@ and
and the
.BR mmap (2)
.B MAP_LOCKED
-operation.
+or
+.B MAP_LOCKONFAULT
+operations.
Since Linux 2.6.9 it also affects the
.BR shmctl (2)
.B SHM_LOCK
@@ -232,7 +234,10 @@ locks established by
.BR mlockall (2),
and
.BR mmap (2)
-.BR MAP_LOCKED ;
+with
+.B MAP_LOCKED
+or
+.BR MAP_LOCKONFAULT ;
a process can lock bytes up to this limit in each of these
two categories.
In Linux kernels before 2.6.9, this limit controlled the amount of
diff --git a/man2/mlock.2 b/man2/mlock.2
index b8487ff..139a7be 100644
--- a/man2/mlock.2
+++ b/man2/mlock.2
@@ -96,9 +96,31 @@ process in the future.
These could be for instance new pages required
by a growing heap and stack as well as new memory-mapped files or
shared memory regions.
+.B MCL_FUTURE
+will attempt to make all pages present when the address
+space is allocated.
+.TP
+.BR MCL_ONFAULT " (since Linux 4.2)"
+Like
+.BR MCL_FUTURE ,
+but
+.B MCL_ONFAULT
+does not attempt to make all pages present when the address space is
+allocated, instead wait until each page is accessed for the first
+time before locking. Note that as with the difference between
+.B MAP_LOCKED
+and
+.B MAP_LOCKONFAULT
+for
+.BR mmap "(2),"
+the caller is charged for the entire allocated address space. See
+.BR setrlimit "(2)"
+for more details on resource limits.
.PP
If
.B MCL_FUTURE
+or
+.B MCL_ONFAULT
has been specified, then a later system call (e.g.,
.BR mmap (2),
.BR sbrk (2),
@@ -250,9 +272,11 @@ or when the process terminates.
The
.BR mlockall ()
.B MCL_FUTURE
-setting is not inherited by a child created via
+and
+.B MCL_ONFAULT
+settings are not inherited by a child created via
.BR fork (2)
-and is cleared during an
+and are cleared during an
.BR execve (2).
The memory lock on an address range is automatically removed
diff --git a/man2/mmap.2 b/man2/mmap.2
index a865612..5aa29e9 100644
--- a/man2/mmap.2
+++ b/man2/mmap.2
@@ -277,6 +277,26 @@ of the mapping.
This flag is ignored in older kernels.
.\" If set, the mapped pages will not be swapped out.
.TP
+.BR MAP_LOCKONFAULT " (since Linux 4.2)"
+Lock pages covered by this mapping after they are
+accessed for the first time. Unlike
+.BR MAP_LOCKED ,
+.B MAP_LOCKONFAULT
+does not attempt to populate the mapping immediately. Note that while
+.B MAP_LOCKONFAULT
+does not populate the mapping, the caller is charged for the full mapping
+against
+.B RLIMIT_MEMLOCK
+when the mapping is created. This allows
+.BR mmap "(2)"
+calls to fail the same way for
+.B MAP_LOCKED
+and
+.B MAP_LOCKONFAULT
+when the resource limit would be exceeded. See
+.BR setrlimit "(2)"
+for more details on resource limits.
+.TP
.BR MAP_NONBLOCK " (since Linux 2.5.46)"
Only meaningful in conjunction with
.BR MAP_POPULATE .
@@ -618,6 +638,7 @@ The relevant flags are:
.BR MAP_GROWSDOWN ,
.BR MAP_HUGETLB ,
.BR MAP_LOCKED ,
+.BR MAP_LOCKONFAULT ,
.BR MAP_NONBLOCK ,
.BR MAP_NORESERVE ,
.BR MAP_POPULATE ,
--
1.9.1