All vm_object_page_remove() callers, except
linux_invalidate_mapping_pages() in the LinuxKPI, free swap space when
removing a range of pages from an object. The LinuxKPI case appears to
be an unintentional omission that could result in leaked swap blocks, so
unconditionally free swap space in vm_object_page_remove() to protect
against similar bugs in the future.
Reviewed by: alc, kib
Tested by: pho
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D25329
On 64-bit platforms, the two short fields in `struct tmpfs_fid` are padded to
the 64-bit alignment of the long field. This pushes the offsets of the
subsequent fields by 4 bytes and makes `struct tmpfs_fid` bigger than
`struct fid`. `tmpfs_vptofh()` casts a `struct fid *` to `struct tmpfs_fid *`,
causing 4 bytes of adjacent memory to be overwritten when the struct fields are
set. Through several layers of indirection and embedded structs, the adjacent
memory for one particular call to `tmpfs_vptofh()` happens to be the stack
canary for `nfsrvd_compound()`. Half of the canary ends up being clobbered,
going unnoticed until eventually the stack check fails when `nfsrvd_compound()`
returns and a panic is triggered.
Instead of duplicating fields of `struct fid` in `struct tmpfs_fid`, narrow the
struct to cover only the unique fields for tmpfs and assert at compile time
that the struct fits in the allotted space. This way we don't have to
replicate the offsets of `struct fid` fields, we just use them directly.
Reviewed by: kib, mav, rmacklem
Approved by: mav (mentor)
MFC after: 1 week
Sponsored by: iXsystems, Inc.
Differential Revision: https://reviews.freebsd.org/D25077
r357614 added CTLFLAG_NEEDGIANT to make it easier to find nodes that are
still not MPSAFE (or already are but aren’t properly marked).
Use it in preparation for a general review of all nodes.
This is non-functional change that adds annotations to SYSCTL_NODE and
SYSCTL_PROC nodes using one of the soon-to-be-required flags.
Mark all obvious cases as MPSAFE. All entries that haven't been marked
as MPSAFE before are by default marked as NEEDGIANT
Approved by: kib (mentor, blanket)
Commented by: kib, gallatin, melifaro
Differential Revision: https://reviews.freebsd.org/D23718
At the time opt-in was introduced adding yourself as a writer was esrializing
across the mount point. Nowadays it is fully per-cpu, the only impact being
a small single-threaded hit on top of what's there right now.
Vast majority of the overhead stems from the call to VOP_GETWRITEMOUNT which
has is done regardless.
Should someone want to microoptimize this single-threaded they can coalesce
looking the mount up with adding a write to it.
which disables tracking mtime updates due to writes through the shared
mapped areas backed by tmpfs files. This removes periodic scans which
downgrades rw mapped pages to ro to note the writes.
Suggested by: mjg
Reviewed by: markj
Tested by: pho
Sponsored by: The FreeBSD Foundation
MFC after: 1 week
Differential revision: https://reviews.freebsd.org/D23432
It was generated to be just a jumping off point to tmpfs_itimes.
While here provide a dedicated variant for getattr since we normally don't
expect to need to the update from that caller.
O_SEARCH is defined by POSIX [0] to open a directory for searching, skipping
permissions checks on the directory itself after the initial open(). This is
close to the semantics we've historically applied for O_EXEC on a directory,
which is UB according to POSIX. Conveniently, O_SEARCH on a file is also
explicitly undefined behavior according to POSIX, so O_EXEC would be a fine
choice. The spec goes on to state that O_SEARCH and O_EXEC need not be
distinct values, but they're not defined to be the same value.
This was pointed out as an incompatibility with other systems that had made
its way into libarchive, which had assumed that O_EXEC was an alias for
O_SEARCH.
This defines compatibility O_SEARCH/FSEARCH (equivalent to O_EXEC and FEXEC
respectively) and expands our UB for O_EXEC on a directory. O_EXEC on a
directory is checked in vn_open_vnode already, so for completeness we add a
NOEXECCHECK when O_SEARCH has been specified on the top-level fd and do not
re-check that when descending in namei.
[0] https://pubs.opengroup.org/onlinepubs/9699919799/
Reviewed by: kib
Differential Revision: https://reviews.freebsd.org/D23247
The vnode pager does not want the object lock held. Moving this out allows
further object lock scope reduction in callers. While here add some missing
paging in progress calls and an assert. The object handle is now protected
explicitly with pip.
Reviewed by: kib, markj
Differential Revision: https://reviews.freebsd.org/D23033
The current notion of an active vnode is eliminated.
Vnodes transition between 0<->1 hold counts all the time and the
associated traversal between different lists induces significant
scalability problems in certain workloads.
Introduce a global list containing all allocated vnodes. They get
unlinked only when UMA reclaims memory and are only requeued when
hold count reaches 0.
Sample result from an incremental make -s -j 104 bzImage on tmpfs:
stock: 118.55s user 3649.73s system 7479% cpu 50.382 total
patched: 122.38s user 1780.45s system 6242% cpu 30.480 total
Reviewed by: jeff
Tested by: pho (in a larger patch, previous version)
Differential Revision: https://reviews.freebsd.org/D22997
Filesystems which want to use it in limited capacity can employ the
VOP_UNLOCK_FLAGS macro.
Reviewed by: kib (previous version)
Differential Revision: https://reviews.freebsd.org/D21427
With the previous reviews, the page lock is no longer required in order
to perform queue operations on a page. It is also no longer needed in
the page queue scans. This change effectively eliminates remaining uses
of the page lock and also the false sharing caused by multiple pages
sharing a page lock.
Reviewed by: jeff
Tested by: pho
Sponsored by: Netflix, Intel
Differential Revision: https://reviews.freebsd.org/D22885
compilation error because, without _KERNEL defined, the macro
TMPFS_VALIDATE_DIR is invoked, but never defined. User-level software
that includes sys/tmpfs.h must define _KERNEL to make the definition
of TMPFS_VALIDATE_DIR visible.
This change puts all the inline functions that, directly or
indirectly, invoke MPASS into the scope of the _KERNEL block, allowing
many user-space includers of <sys/tmpfs.h> to stop defining _KERNEL.
Reviewed by: alc, kib
MFC after: 1 week
Differential Revision: https://reviews.freebsd.org/D22874
an exclusive object lock.
Previously swap space was freed on a best effort basis when a page that
had valid swap was dirtied, thus invalidating the swap copy. This may be
done inconsistently and requires the object lock which is not always
convenient.
Instead, track when swap space is present. The first dirty is responsible
for deleting space or setting PGA_SWAP_FREE which will trigger background
scans to free the swap space.
Simplify the locking in vm_fault_dirty() now that we can reliably identify
the first dirty.
Discussed with: alc, kib, markj
Differential Revision: https://reviews.freebsd.org/D22654
Both of these features are not needed by many consumers and result in avoidable
reads which in turn puts them on profiles due to cache-line ping ponging.
On top of that the current lockgmr entry point is slower than necessary
single-threaded. As an attempted clean up preparing for other changes,
provide new routines which don't support any of the aforementioned features.
With these patches in place vop_stdlock and vop_stdunlock disappear from
flamegraphs during -j 104 buildkernel.
Reviewed by: jeff (previous version)
Tested by: pho
Differential Revision: https://reviews.freebsd.org/D22665
The current vnode layout is not smp-friendly by having frequently read data
avoidably sharing cachelines with very frequently modified fields. In
particular v_iflag inspected for VI_DOOMED can be found in the same line with
v_usecount. Instead make it available in the same cacheline as the v_op, v_data
and v_type which all get read all the time.
v_type is avoidably 4 bytes while the necessary data will easily fit in 1.
Shrinking it frees up 3 bytes, 2 of which get used here to introduce a new
flag field with a new value: VIRF_DOOMED.
Reviewed by: kib, jeff
Differential Revision: https://reviews.freebsd.org/D22715
Top-level kern_renameat() increases the writecount on the mount point,
which, together with tmpfs unmount suspending the mount, already
ensures that unmount cannot proceed while rename unlocks and relocks
all operated vnodes.
Remove vfs_busy() call from tmpfs_rename() which was done while
holding a vnode lock, creating the deadlock. The only intent of the
busy operation seems to be the prevention of unmount, which is already
ensured.
Reported and tested by: pho
Sponsored by: The FreeBSD Foundation
MFC after: 1 week
reudundant complicated checks and additional locking required only for
anonymous memory. Introduce vm_object_allocate_anon() to create these
objects. DEFAULT and SWAP objects now have the correct settings for
non-anonymous consumers and so individual consumers need not modify the
default flags to create super-pages and avoid ONEMAPPING/NOSPLIT.
Reviewed by: alc, dougm, kib, markj
Tested by: pho
Differential Revision: https://reviews.freebsd.org/D22119
flag and use the same system.
This enables further fault locking improvements by allowing more faults to
proceed with a shared lock.
Reviewed by: kib
Tested by: pho
Differential Revision: https://reviews.freebsd.org/D22116
Atomics are used for page busy and valid state when the shared busy is
held. The details of the locking protocol and valid and dirty
synchronization are in the updated vm_page.h comments.
Reviewed by: kib, markj
Tested by: pho
Sponsored by: Netflix, Intel
Differential Revision: https://reviews.freebsd.org/D21594
This is the first in a series of patches that promotes the page busy field
to a first class lock that no longer requires the object lock for
consistency.
Reviewed by: kib, markj
Tested by: pho
Sponsored by: Netflix, Intel
Differential Revision: https://reviews.freebsd.org/D21548
In case the implementation ever changes from using a chain of next pointers,
then changing the macro definition will be necessary, but changing all the
files that iterate over vm_map entries will not.
Drop a counter in vm_object.c that would have an effect only if the
vm_map entry count was wrong.
Discussed with: alc
Reviewed by: markj
Tested by: pho (earlier version)
Differential Revision: https://reviews.freebsd.org/D21882
During readdir() we guarantee that the tn_dir.tn_parent does not go
away, but it might be replaced by a parallel rename. Read tn_parent
only once, then use the cached value.
Reported and tested by: pho
Sponsored by: The FreeBSD Foundation
MFC after: 1 week
There are 3 counters modified all the time in this structure - one for
keeping the structure alive, one for preventing unmount and one for
tracking active writers. Exact values of these counters are very rarely
needed, which makes them a prime candidate for conversion to a per-cpu
scheme, resulting in much better performance.
Sample benchmark performing fstatfs (modifying 2 out of 3 counters) on
a 104-way 2 socket Skylake system:
before: 852393 ops/s
after: 76682077 ops/s
Reviewed by: kib, jeff
Tested by: pho
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D21637
There are several mechanisms by which a vm_page reference is held,
preventing the page from being freed back to the page allocator. In
particular, holding the page's object lock is sufficient to prevent the
page from being freed; holding the busy lock or a wiring is sufficent as
well. These references are protected by the page lock, which must
therefore be acquired for many per-page operations. This results in
false sharing since the page locks are external to the vm_page
structures themselves and each lock protects multiple structures.
Transition to using an atomically updated per-page reference counter.
The object's reference is counted using a flag bit in the counter. A
second flag bit is used to atomically block new references via
pmap_extract_and_hold() while removing managed mappings of a page.
Thus, the reference count of a page is guaranteed not to increase if the
page is unbusied, unmapped, and the object's write lock is held. As
a consequence of this, the page lock no longer protects a page's
identity; operations which move pages between objects are now
synchronized solely by the objects' locks.
The vm_page_wire() and vm_page_unwire() KPIs are changed. The former
requires that either the object lock or the busy lock is held. The
latter no longer has a return value and may free the page if it releases
the last reference to that page. vm_page_unwire_noq() behaves the same
as before; the caller is responsible for checking its return value and
freeing or enqueuing the page as appropriate. vm_page_wire_mapped() is
introduced for use in pmap_extract_and_hold(). It fails if the page is
concurrently being unmapped, typically triggering a fallback to the
fault handler. vm_page_wire() no longer requires the page lock and
vm_page_unwire() now internally acquires the page lock when releasing
the last wiring of a page (since the page lock still protects a page's
queue state). In particular, synchronization details are no longer
leaked into the caller.
The change excises the page lock from several frequently executed code
paths. In particular, vm_object_terminate() no longer bounces between
page locks as it releases an object's pages, and direct I/O and
sendfile(SF_NOCACHE) completions no longer require the page lock. In
these latter cases we now get linear scalability in the common scenario
where different threads are operating on different files.
__FreeBSD_version is bumped. The DRM ports have been updated to
accomodate the KPI changes.
Reviewed by: jeff (earlier version)
Tested by: gallatin (earlier version), pho
Sponsored by: Netflix
Differential Revision: https://reviews.freebsd.org/D20486
Current implementation of vnode_create_vobject() and
vnode_destroy_vobject() is written so that it prepared to handle the
vm object destruction for live vnode. Practically, no filesystems use
this, except for some remnants that were present in UFS till today.
One of the consequences of that model is that each filesystem must
call vnode_destroy_vobject() in VOP_RECLAIM() or earlier, as result
all of them get rid of the v_object in reclaim.
Move the call to vnode_destroy_vobject() to vgonel() before
VOP_RECLAIM(). This makes v_object stable: either the object is NULL,
or it is valid vm object till the vnode reclamation. Remove code from
vnode_create_vobject() to handle races with the parallel destruction.
Reviewed by: markj
Tested by: pho
Sponsored by: The FreeBSD Foundation
Differential revision: https://reviews.freebsd.org/D21412
Suppose that a binary was executed from tmpfs mount, and the text
vnode was reclaimed while the binary was still running. It is
possible during even the normal operations since tmpfs vnode'
vm_object has swap type, and no references on the vnode is held. Also
assume that the text vnode was revived for some reason. Then, on the
process exit or exec, unmapping of the text mapping tries to remove
the text reference from the vnode, but since it went from
recycle/instantiation cycle, there is no reference kept, and assertion
in VOP_UNSET_TEXT_CHECKED() triggers.
Fix this by keeping a use reference on the tmpfs vnode for each exec
reference. This prevents the vnode reclamation while executable map
entry is active.
Do it by adding per-mount flag MNTK_TEXT_REFS that directs
vop_stdset_text() to add use ref on first vnode text use, and
per-vnode VI_TEXT_REF flag, to record the need on unref in
vop_stdunset_text() on last vnode text use going away. Set
MNTK_TEXT_REFS for tmpfs mounts.
Reported by: bdrewery
Tested by: sbruno, pho (previous version)
Sponsored by: The FreeBSD Foundation
MFC after: 1 week
Both filesystems do no use vnode_pager_dealloc() which would handle
this case otherwise. Nullfs because vnode vm_object handle never
points to nullfs vnode. Tmpfs because its vm_object is never vnode
object at all.
Tested by: pho
Sponsored by: The FreeBSD Foundation
MFC after: 1 week
Otherwise we might dereference NULL vp->v_data after
VP_TO_TMPFS_NODE().
Reported and tested by: pho
Sponsored by: The FreeBSD Foundation
MFC after: 1 week
In particular:
- suspend the mount around vflush() to avoid new writes come after the
vnode is processed;
- flush pending metadata updates (mostly node times);
- remap all rw mappings of files from the mount into ro.
It is not clear to me how to handle writeable mappings on rw->ro for
tmpfs best. Other filesystems, which use vnode vm object, call
vgone() on vnodes with writers, which sets the vm object type to
OBJT_DEAD, and keep the resident pages and installed ptes as is. In
particular, the existing mappings continue to work as far as
application only accesses resident pages, but changes are not flushed
to file.
For tmpfs the vm object of VREG vnodes also serves as the data pages
container, giving single copy of the mapped pages, so it cannot be set
to OBJT_DEAD. Alternatives for making rw mappings ro could be either
invalidating them at all, or marking as CoW.
Tested by: pho
Sponsored by: The FreeBSD Foundation
MFC after: 2 weeks
Differential revision: https://reviews.freebsd.org/D19737
In particular, this fixes atimes still changing for ro tmpfs.
tmpfs_set_status() gains tmpfs_mount * argument.
Tested by: pho
Sponsored by: The FreeBSD Foundation
MFC after: 2 weeks
Differential revision: https://reviews.freebsd.org/D19737
TMPFS_PAGES_MINRESERVED controls how much memory is reserved for the system
and not used by tmpfs.
On very small memory systems, the default value may be too high and this
prevents these small memory systems from using reroot, which is required
for them to install firmware updates.
Submitted by: Hiroki Mori <yamori813@yahoo.co.jp>
Reviewed by: mizhka
Differential Revision: https://reviews.freebsd.org/D13583
