The issue was reported by
Alexander Lochmann <alexander.lochmann@tu-dortmund.de>,
who found the problem by performing lock analysis using LockDoc,
see https://doi.org/10.1145/3302424.3303948.
Reviewed by: mckusick
Tested by: pho
Sponsored by: The FreeBSD Foundation
MFC after: 1 week
Differential revision: https://reviews.freebsd.org/D31780
Fix a few 'if(' to be 'if (' in a few places, per style(9) and
overwhelming usage in the rest of the kernel / tree.
MFC After: 3 days
Sponsored by: Netflix
The data is only needed by filesystems that
1. use buffer cache
2. utilize clustering write support.
Requested by: mjg
Reviewed by: asomers (previous version), fsu (ext2 parts), mckusick
Tested by: pho
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D28679
Replace MAXPHYS by runtime variable maxphys. It is initialized from
MAXPHYS by default, but can be also adjusted with the tunable kern.maxphys.
Make b_pages[] array in struct buf flexible. Size b_pages[] for buffer
cache buffers exactly to atop(maxbcachebuf) (currently it is sized to
atop(MAXPHYS)), and b_pages[] for pbufs is sized to atop(maxphys) + 1.
The +1 for pbufs allow several pbuf consumers, among them vmapbuf(),
to use unaligned buffers still sized to maxphys, esp. when such
buffers come from userspace (*). Overall, we save significant amount
of otherwise wasted memory in b_pages[] for buffer cache buffers,
while bumping MAXPHYS to desired high value.
Eliminate all direct uses of the MAXPHYS constant in kernel and driver
sources, except a place which initialize maxphys. Some random (and
arguably weird) uses of MAXPHYS, e.g. in linuxolator, are converted
straight. Some drivers, which use MAXPHYS to size embeded structures,
get private MAXPHYS-like constant; their convertion is out of scope
for this work.
Changes to cam/, dev/ahci, dev/ata, dev/mpr, dev/mpt, dev/mvs,
dev/siis, where either submitted by, or based on changes by mav.
Suggested by: mav (*)
Reviewed by: imp, mav, imp, mckusick, scottl (intermediate versions)
Tested by: pho
Sponsored by: The FreeBSD Foundation
Differential revision: https://reviews.freebsd.org/D27225
Previously just ensuring that we do not sleep when clustering for
md(4) vnode was enough. Now, with the switch of the pbuf allocator to
uma and completely broken per-subsystem pbuf limits, it might cause
unbounded sleep even for non-md(4) vnodes.
Reported and tested by: pho
Reviewed by: markj
Sponsored by: The FreeBSD Foundation
MFC after: 1 week
Differential revision: https://reviews.freebsd.org/D22899
the vnode, logical block number, and size of data block that is
being requested. They then use the VOP_BMAP function to calculate
the mapping from logical block number to physical block number from
which to access the data. This change expands the interface to also
pass the physical block number in cases where the VOP_MAP function
may no longer work, for example when a file is being truncated.
No functional change.
Reviewed by: kib
Tested by: Peter Holm
Sponsored by: Netflix
Recent changes to busy/valid/dirty have enabled page based synchronization
and the object lock is no longer required in many cases.
Reviewed by: kib
Sponsored by: Netflix, Intel
Differential Revision: https://reviews.freebsd.org/D21597
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
by issuing delayed-writes (bdwrite()) until a non-sequential block
is written or the maximum cluster size is reached. At that point
it collects the delayed buffers together (using bread()) to write
them in a single operation. The assumption was that since we just
looked at them they will still be in memory so there is no need to
check for a read error from bread(). Very occationally (apparently
every 10-hours or so when being pounded by Peter Holm's tests)
this assumption is wrong.
The fix is to check for errors from bread() and fail the cluster
write thus falling back to the default individual flushing of any
still dirty buffers.
Reported by: Peter Holm and Chuck Silvers
Reviewed by: kib
MFC after: 3 days
o In vm_pager_bufferinit() create pbuf_zone and start accounting on how many
pbufs are we going to have set.
In various subsystems that are going to utilize pbufs create private zones
via call to pbuf_zsecond_create(). The latter calls uma_zsecond_create(),
and sets a limit on created zone. After startup preallocate pbufs according
to requirements of all pbuf zones.
Subsystems that used to have a private limit with old allocator now have
private pbuf zones: md(4), fusefs, NFS client, smbfs, VFS cluster, FFS,
swap, vnode pager.
The following subsystems use shared pbuf zone: cam(4), nvme(4), physio(9),
aio(4). They should have their private limits, but changing that is out of
scope of this commit.
o Fetch tunable value of kern.nswbuf from init_param2() and while here move
NSWBUF_MIN to opt_param.h and eliminate opt_swap.h, that was holding only
this option.
Default values aren't touched by this commit, but they probably should be
reviewed wrt to modern hardware.
This change removes a tight bottleneck from sendfile(2) operation, that
uses pbufs in vnode pager. Other pagers also would benefit from faster
allocation.
Together with: gallatin
Tested by: pho
- Create getblkx(9) variant of getblk(9) which can return error.
- Add GB_NOSPARSE flag for getblk()/getblkx() which requests that BMAP
was performed before the buffer is created, and EJUSTRETURN returned
in case the requested block does not exist.
- Make ffs_read() use GB_NOSPARSE to avoid instantiating buffer (and
allocating the pages for it), copying from zero_region instead.
The end result is less page allocations and buffer recycling when a
hole is read, which is important for some benchmarks.
Requested and reviewed by: jeff
Tested by: pho
Sponsored by: The FreeBSD Foundation
MFC after: 2 weeks
Differential revision: https://reviews.freebsd.org/D14917
Mainly focus on files that use BSD 3-Clause license.
The Software Package Data Exchange (SPDX) group provides a specification
to make it easier for automated tools to detect and summarize well known
opensource licenses. We are gradually adopting the specification, noting
that the tags are considered only advisory and do not, in any way,
superceed or replace the license texts.
Special thanks to Wind River for providing access to "The Duke of
Highlander" tool: an older (2014) run over FreeBSD tree was useful as a
starting point.
Upstream the BUF_TRACKING and FULL_BUF_TRACKING buffer debugging code.
This can be handy in tracking down what code touched hung bios and bufs
last. The full history is especially useful, but adds enough bloat that
it shouldn't be enabled in release builds.
Function names (or arbitrary string constants) are tracked in a
fixed-size ring in bufs. Bios gain a pointer to the upper buf for
tracking. SCSI CCBs gain a pointer to the upper bio for tracking.
Reviewed by: markj
Sponsored by: Dell EMC Isilon
Differential Revision: https://reviews.freebsd.org/D8366
It was added in r153192 for XFS and doesn't appear to have been used for
anything else. XFS was disconnected in r241607 and removed entirely in
r247631.
Reported by: mlaier
Reviewed by: imp, kib
Differential Revision: https://reviews.freebsd.org/D7468
for limiting disk (actually filesystem) IO.
Note that in some cases these limits are not quite precise. It's ok,
as long as it's within some reasonable bounds.
Testing - and review of the code, in particular the VFS and VM parts - is
very welcome.
MFC after: 1 month
Relnotes: yes
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D5080
a buffer pointer in the event of an error (for some errors it would
return a buffer pointer and for other errors it would not return a
buffer pointer). The cluster_read() function was similarly inconsistent.
Clients of these functions were inconsistent in handling errors.
Some would assume that no buffer was returned after an error and
would thus lose buffers under certain error conditions. Others would
assume that brelse() should always be called after an error and
would thus panic the system under certain error conditions.
To correct both of these problems with minimal code churn, bread()
and cluster_write() now always free the buffer when returning an
error thus ensuring that buffers will never be lost. The brelse()
routine checks for being passed a NULL buffer pointer and silently
returns to avoid panics. Thus both approaches to handling error
returns from bread() and cluster_read() will work correctly.
Future code should be written assuming that bread() and cluster_read()
will never return a buffer with an error, so should not attempt to
brelse() the buffer when an error is returned.
Reviewed by: kib
- Use pointer assignment rather than a combination of pointers and
flags to switch buffers between unmapped and mapped. This eliminates
multiple flags and generally simplifies the logic.
- Eliminate b_saveaddr since it is only used with pager bufs which have
their b_data re-initialized on each allocation.
- Gather up some convenience routines in the buffer cache for
manipulating buf space and buf malloc space.
- Add an inline, buf_mapped(), to standardize checks around unmapped
buffers.
In collaboration with: mlaier
Reviewed by: kib
Tested by: pho (many small revisions ago)
Sponsored by: EMC / Isilon Storage Division
buildkernel run.
Some of them were write-only under some kernel options, e.g. variables
keeping values only used by CTR() macros. It costs nothing to the
code readability and correctness to eliminate the warnings in those
cases too by removing the local cached values used only for
single-access.
Review: https://reviews.freebsd.org/D2665
Reviewed by: rodrigc
Looked at by: bjk
Sponsored by: The FreeBSD Foundation
MFC after: 1 week
for the available pbuf when passed vnode is backing md(4). Other i/o
directed to the same md device might already hold pbufs, and then we
could deadlock since only our progress can free a pbuf needed for
wakeup.
Obtained from: projects/vm6
Reminded and tested by: pho
MFC after: 1 week
shared busy without first draining the hard busy state. Previously it
went unnoticed since VPO_BUSY and m->busy fields were distinct, and
vm_page_io_start() did not verified that the passed page has VPO_BUSY
flag cleared, but such page state is wrong. New implementation is
more strict and catched this case.
Drain the busy state as needed, before calling vm_page_sbusy().
Tested by: pho, jkim
Sponsored by: The FreeBSD Foundation
Unify the 2 concept into a real, minimal, sxlock where the shared
acquisition represent the soft busy and the exclusive acquisition
represent the hard busy.
The old VPO_WANTED mechanism becames the hard-path for this new lock
and it becomes per-page rather than per-object.
The vm_object lock becames an interlock for this functionality:
it can be held in both read or write mode.
However, if the vm_object lock is held in read mode while acquiring
or releasing the busy state, the thread owner cannot make any
assumption on the busy state unless it is also busying it.
Also:
- Add a new flag to directly shared busy pages while vm_page_alloc
and vm_page_grab are being executed. This will be very helpful
once these functions happen under a read object lock.
- Move the swapping sleep into its own per-object flag
The KPI is heavilly changed this is why the version is bumped.
It is very likely that some VM ports users will need to change
their own code.
Sponsored by: EMC / Isilon storage division
Discussed with: alc
Reviewed by: jeff, kib
Tested by: gavin, bapt (older version)
Tested by: pho, scottl
- Use a shared bufobj lock in getblk() and inmem().
- Convert softdep's lk to rwlock to match the bufobj lock.
- Move INFREECNT to b_flags and protect it with the buf lock.
- Remove unnecessary locking around bremfree() and BKGRDINPROG.
Sponsored by: EMC / Isilon Storage Division
Discussed with: mckusick, kib, mdf
defaults to 1, meaning that it's off.
When read-ahead is enabled on a file, the vfs cluster code deliberately
breaks a read into 2 I/O transactions; one to satisfy the actual read,
and one to perform read-ahead. This makes sense in low-latency
circumstances, but often produces unbalanced i/o transactions that
penalize disks. By setting vfs.read_min, we can tell the algorithm to
fetch a larger transaction that what we asked for, achieving the same
effect as the read-ahead but without the doubled, unbalanced transaction
and the slightly lower latency. This significantly helps our workloads
with video streaming.
Submitted by: emax
Reviewed by: kib
Obtained from: Netflix
do not map the b_pages pages into buffer_map KVA. The use of the
unmapped buffers eliminate the need to perform TLB shootdown for
mapping on the buffer creation and reuse, greatly reducing the amount
of IPIs for shootdown on big-SMP machines and eliminating up to 25-30%
of the system time on i/o intensive workloads.
The unmapped buffer should be explicitely requested by the GB_UNMAPPED
flag by the consumer. For unmapped buffer, no KVA reservation is
performed at all. The consumer might request unmapped buffer which
does have a KVA reserve, to manually map it without recursing into
buffer cache and blocking, with the GB_KVAALLOC flag.
When the mapped buffer is requested and unmapped buffer already
exists, the cache performs an upgrade, possibly reusing the KVA
reservation.
Unmapped buffer is translated into unmapped bio in g_vfs_strategy().
Unmapped bio carry a pointer to the vm_page_t array, offset and length
instead of the data pointer. The provider which processes the bio
should explicitely specify a readiness to accept unmapped bio,
otherwise g_down geom thread performs the transient upgrade of the bio
request by mapping the pages into the new bio_transient_map KVA
submap.
The bio_transient_map submap claims up to 10% of the buffer map, and
the total buffer_map + bio_transient_map KVA usage stays the
same. Still, it could be manually tuned by kern.bio_transient_maxcnt
tunable, in the units of the transient mappings. Eventually, the
bio_transient_map could be removed after all geom classes and drivers
can accept unmapped i/o requests.
Unmapped support can be turned off by the vfs.unmapped_buf_allowed
tunable, disabling which makes the buffer (or cluster) creation
requests to ignore GB_UNMAPPED and GB_KVAALLOC flags. Unmapped
buffers are only enabled by default on the architectures where
pmap_copy_page() was implemented and tested.
In the rework, filesystem metadata is not the subject to maxbufspace
limit anymore. Since the metadata buffers are always mapped, the
buffers still have to fit into the buffer map, which provides a
reasonable (but practically unreachable) upper bound on it. The
non-metadata buffer allocations, both mapped and unmapped, is
accounted against maxbufspace, as before. Effectively, this means that
the maxbufspace is forced on mapped and unmapped buffers separately.
The pre-patch bufspace limiting code did not worked, because
buffer_map fragmentation does not allow the limit to be reached.
By Jeff Roberson request, the getnewbuf() function was split into
smaller single-purpose functions.
Sponsored by: The FreeBSD Foundation
Discussed with: jeff (previous version)
Tested by: pho, scottl (previous version), jhb, bf
MFC after: 2 weeks
cluster_write() and cluster_wbuild() functions. The flags to be
allowed are a subset of the GB_* flags for getblk().
Sponsored by: The FreeBSD Foundation
Tested by: pho
future further optimizations where the vm_object lock will be held
in read mode most of the time the page cache resident pool of pages
are accessed for reading purposes.
The change is mostly mechanical but few notes are reported:
* The KPI changes as follow:
- VM_OBJECT_LOCK() -> VM_OBJECT_WLOCK()
- VM_OBJECT_TRYLOCK() -> VM_OBJECT_TRYWLOCK()
- VM_OBJECT_UNLOCK() -> VM_OBJECT_WUNLOCK()
- VM_OBJECT_LOCK_ASSERT(MA_OWNED) -> VM_OBJECT_ASSERT_WLOCKED()
(in order to avoid visibility of implementation details)
- The read-mode operations are added:
VM_OBJECT_RLOCK(), VM_OBJECT_TRYRLOCK(), VM_OBJECT_RUNLOCK(),
VM_OBJECT_ASSERT_RLOCKED(), VM_OBJECT_ASSERT_LOCKED()
* The vm/vm_pager.h namespace pollution avoidance (forcing requiring
sys/mutex.h in consumers directly to cater its inlining functions
using VM_OBJECT_LOCK()) imposes that all the vm/vm_pager.h
consumers now must include also sys/rwlock.h.
* zfs requires a quite convoluted fix to include FreeBSD rwlocks into
the compat layer because the name clash between FreeBSD and solaris
versions must be avoided.
At this purpose zfs redefines the vm_object locking functions
directly, isolating the FreeBSD components in specific compat stubs.
The KPI results heavilly broken by this commit. Thirdy part ports must
be updated accordingly (I can think off-hand of VirtualBox, for example).
Sponsored by: EMC / Isilon storage division
Reviewed by: jeff
Reviewed by: pjd (ZFS specific review)
Discussed with: alc
Tested by: pho
write is a disk write request that tells the disk that the buffer
being written must be committed to the media along with any writes
that preceeded it before any future blocks may be written to the drive.
Barrier writes are provided by adding the functions bbarrierwrite
(bwrite with barrier) and babarrierwrite (bawrite with barrier).
Following a bbarrierwrite the client knows that the requested buffer
is on the media. It does not ensure that buffers written before that
buffer are on the media. It only ensure that buffers written before
that buffer will get to the media before any buffers written after
that buffer. A flush command must be sent to the disk to ensure that
all earlier written buffers are on the media.
Reviewed by: kib
Tested by: Peter Holm
most system, based on benchmark results on a low-end fibre channel SAN
under VMWare:
vfs.read_max read performance
8 (historical default) 83 MB/s
16 (recent bump) 131 MB/s
32 (this version) 152 MB/s
64 157 MB/s
(results are +/- 3 MB/s)
As read-ahead is heuristic, based on past IO requests, it shouldn't be
problematic. The new default is still smaller then in other OSes.
the vfs.read_max default. For most systems this means going from 128 KiB
to 256 KiB, which is still very conservative and lower than what most
other operating systems use, but as a sane default should not
interfere much with existing systems.
For systems with RAID volumes and/or virtualization envirnments, where
read performance is very important, increasing this sysctl tunable to 32
or even more will demonstratively yield additional performance benefits.
If MAXPHYS ever gets bumped up, it will probably be a good idea to slave
read_max to it.
in the case of a file system with a block size that is less than the page
size, cluster_rbuild() looks at too many of the page's valid bits.
Consequently, it may terminate prematurely, resulting in poor performance.
Reported by: bde
Reviewed by: tegge
Approved by: re (kib)
BO_LOCK/UNLOCK/MTX when manipulating the bufobj.
- Create a new lock in the bufobj to lock bufobj fields independently.
This leaves the vnode interlock as an 'identity' lock while the bufobj
is an io lock. The bufobj lock is ordered before the vnode interlock
and also before the mnt ilock.
- Exploit this new lock order to simplify softdep_check_suspend().
- A few sync related functions are marked with a new XXX to note that
we may not properly interlock against a non-zero bv_cnt when
attempting to sync all vnodes on a mountlist. I do not believe this
race is important. If I'm wrong this will make these locations easier
to find.
Reviewed by: kib (earlier diff)
Tested by: kris, pho (earlier diff)
td_ru. This removes the requirement for per-process synchronization in
statclock() and mi_switch(). This was previously supported by
sched_lock which is going away. All modifications to rusage are now
done in the context of the owning thread. reads proceed without locks.
- Aggregate exiting threads rusage in thread_exit() such that the exiting
thread's rusage is not lost.
- Provide a new routine, rufetch() to fetch an aggregate of all rusage
structures from all threads in a process. This routine must be used
in any place requiring a rusage from a process prior to it's exit. The
exited process's rusage is still available via p_ru.
- Aggregate tick statistics only on demand via rufetch() or when a thread
exits. Tick statistics are kept in the thread and protected by sched_lock
until it exits.
Initial patch by: attilio
Reviewed by: attilio, bde (some objections), arch (mostly silent)
