This code merge adds a pNFS service to the NFSv4.1 server. Although it is
a large commit it should not affect behaviour for a non-pNFS NFS server.
Some documentation on how this works can be found at:
http://people.freebsd.org/~rmacklem/pnfs-planb-setup.txt
and will hopefully be turned into a proper document soon.
This is a merge of the kernel code. Userland and man page changes will
come soon, once the dust settles on this merge.
It has passed a "make universe", so I hope it will not cause build problems.
It also adds NFSv4.1 server support for the "current stateid".
Here is a brief overview of the pNFS service:
A pNFS service separates the Read/Write oeprations from all the other NFSv4.1
Metadata operations. It is hoped that this separation allows a pNFS service
to be configured that exceeds the limits of a single NFS server for either
storage capacity and/or I/O bandwidth.
It is possible to configure mirroring within the data servers (DSs) so that
the data storage file for an MDS file will be mirrored on two or more of
the DSs.
When this is used, failure of a DS will not stop the pNFS service and a
failed DS can be recovered once repaired while the pNFS service continues
to operate. Although two way mirroring would be the norm, it is possible
to set a mirroring level of up to four or the number of DSs, whichever is
less.
The Metadata server will always be a single point of failure,
just as a single NFS server is.
A Plan B pNFS service consists of a single MetaData Server (MDS) and K
Data Servers (DS), all of which are recent FreeBSD systems.
Clients will mount the MDS as they would a single NFS server.
When files are created, the MDS creates a file tree identical to what a
single NFS server creates, except that all the regular (VREG) files will
be empty. As such, if you look at the exported tree on the MDS directly
on the MDS server (not via an NFS mount), the files will all be of size 0.
Each of these files will also have two extended attributes in the system
attribute name space:
pnfsd.dsfile - This extended attrbute stores the information that
the MDS needs to find the data storage file(s) on DS(s) for this file.
pnfsd.dsattr - This extended attribute stores the Size, AccessTime, ModifyTime
and Change attributes for the file, so that the MDS doesn't need to
acquire the attributes from the DS for every Getattr operation.
For each regular (VREG) file, the MDS creates a data storage file on one
(or more if mirroring is enabled) of the DSs in one of the "dsNN"
subdirectories. The name of this file is the file handle
of the file on the MDS in hexadecimal so that the name is unique.
The DSs use subdirectories named "ds0" to "dsN" so that no one directory
gets too large. The value of "N" is set via the sysctl vfs.nfsd.dsdirsize
on the MDS, with the default being 20.
For production servers that will store a lot of files, this value should
probably be much larger.
It can be increased when the "nfsd" daemon is not running on the MDS,
once the "dsK" directories are created.
For pNFS aware NFSv4.1 clients, the FreeBSD server will return two pieces
of information to the client that allows it to do I/O directly to the DS.
DeviceInfo - This is relatively static information that defines what a DS
is. The critical bits of information returned by the FreeBSD
server is the IP address of the DS and, for the Flexible
File layout, that NFSv4.1 is to be used and that it is
"tightly coupled".
There is a "deviceid" which identifies the DeviceInfo.
Layout - This is per file and can be recalled by the server when it
is no longer valid. For the FreeBSD server, there is support
for two types of layout, call File and Flexible File layout.
Both allow the client to do I/O on the DS via NFSv4.1 I/O
operations. The Flexible File layout is a more recent variant
that allows specification of mirrors, where the client is
expected to do writes to all mirrors to maintain them in a
consistent state. The Flexible File layout also allows the
client to report I/O errors for a DS back to the MDS.
The Flexible File layout supports two variants referred to as
"tightly coupled" vs "loosely coupled". The FreeBSD server always
uses the "tightly coupled" variant where the client uses the
same credentials to do I/O on the DS as it would on the MDS.
For the "loosely coupled" variant, the layout specifies a
synthetic user/group that the client uses to do I/O on the DS.
The FreeBSD server does not do striping and always returns
layouts for the entire file. The critical information in a layout
is Read vs Read/Writea and DeviceID(s) that identify which
DS(s) the data is stored on.
At this time, the MDS generates File Layout layouts to NFSv4.1 clients
that know how to do pNFS for the non-mirrored DS case unless the sysctl
vfs.nfsd.default_flexfile is set non-zero, in which case Flexible File
layouts are generated.
The mirrored DS configuration always generates Flexible File layouts.
For NFS clients that do not support NFSv4.1 pNFS, all I/O operations
are done against the MDS which acts as a proxy for the appropriate DS(s).
When the MDS receives an I/O RPC, it will do the RPC on the DS as a proxy.
If the DS is on the same machine, the MDS/DS will do the RPC on the DS as
a proxy and so on, until the machine runs out of some resource, such as
session slots or mbufs.
As such, DSs must be separate systems from the MDS.
Tested by: james.rose@framestore.com
Relnotes: yes
gcc8 warns that "verf" was set but not used. This was because the code
that uses it is disabled via a "#if 0".
This patch adds a "#if 0" to the variable's declaration and assignment
to get rid of the warning.
This way the code could be re-enabled without difficulty.
Requested by: mmacy
MFC after: 2 weeks
Mechanically replace uses of MALLOC/FREE with appropriate invocations of
malloc(9) / free(9) (a series of sed expressions). Something like:
* MALLOC(a, b, ... -> a = malloc(...
* FREE( -> free(
* free((caddr_t) -> free(
No functional change.
For now, punt on modifying contrib ipfilter code, leaving a definition of
the macro in its KMALLOC().
Reported by: jhb
Reviewed by: cy, imp, markj, rmacklem
Sponsored by: Dell EMC Isilon
Differential Revision: https://reviews.freebsd.org/D14035
This reduces noise when kernel is compiled by newer GCC versions,
such as one used by external toolchain ports.
Reviewed by: kib, andrew(sys/arm and sys/arm64), emaste(partial), erj(partial)
Reviewed by: jhb (sys/dev/pci/* sys/kern/vfs_aio.c and sys/kern/kern_synch.c)
Differential Revision: https://reviews.freebsd.org/D10385
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.
It applies to both NFS client and NFS server, and is useful for both.
This is different from vfs.nfsd.enable_stringtouid, which is specific
to server side.
Reviewed by: rmacklem@
MFC after: 2 weeks
Sponsored by: DARPA, AFRL
vfs.nfsd.nfsd_enable_stringtouid, but in reverse - when set to 1,
it forces the NFSv4 server to return numeric UIDs and GIDs instead
of "user@domain" strings. This helps with clients that can't
translate returned identifiers, eg when rerooting.
The same can be achieved by just never running nfsuserd(8),
but the sysctl is useful to toggle the behaviour back and forth
without rebooting.
Reviewed by: rmacklem (earlier version)
MFC after: 2 weeks
Sponsored by: DARPA, AFRL
Differential Revision: https://reviews.freebsd.org/D11326
Extend the ino_t, dev_t, nlink_t types to 64-bit ints. Modify
struct dirent layout to add d_off, increase the size of d_fileno
to 64-bits, increase the size of d_namlen to 16-bits, and change
the required alignment. Increase struct statfs f_mntfromname[] and
f_mntonname[] array length MNAMELEN to 1024.
ABI breakage is mitigated by providing compatibility using versioned
symbols, ingenious use of the existing padding in structures, and
by employing other tricks. Unfortunately, not everything can be
fixed, especially outside the base system. For instance, third-party
APIs which pass struct stat around are broken in backward and
forward incompatible ways.
Kinfo sysctl MIBs ABI is changed in backward-compatible way, but
there is no general mechanism to handle other sysctl MIBS which
return structures where the layout has changed. It was considered
that the breakage is either in the management interfaces, where we
usually allow ABI slip, or is not important.
Struct xvnode changed layout, no compat shims are provided.
For struct xtty, dev_t tty device member was reduced to uint32_t.
It was decided that keeping ABI compat in this case is more useful
than reporting 64-bit dev_t, for the sake of pstat.
Update note: strictly follow the instructions in UPDATING. Build
and install the new kernel with COMPAT_FREEBSD11 option enabled,
then reboot, and only then install new world.
Credits: The 64-bit inode project, also known as ino64, started life
many years ago as a project by Gleb Kurtsou (gleb). Kirk McKusick
(mckusick) then picked up and updated the patch, and acted as a
flag-waver. Feedback, suggestions, and discussions were carried
by Ed Maste (emaste), John Baldwin (jhb), Jilles Tjoelker (jilles),
and Rick Macklem (rmacklem). Kris Moore (kris) performed an initial
ports investigation followed by an exp-run by Antoine Brodin (antoine).
Essential and all-embracing testing was done by Peter Holm (pho).
The heavy lifting of coordinating all these efforts and bringing the
project to completion were done by Konstantin Belousov (kib).
Sponsored by: The FreeBSD Foundation (emaste, kib)
Differential revision: https://reviews.freebsd.org/D10439
The FreeBSD NFSv4 server did not set the attribute bit for TimeAccess in
the reply to an Open with exclusive_create, as required by the RFCs.
(This is required since the FreeBSD NFS server stores the create_verifier
in the va_atime attribute.)
As such, the Linux NFSv4 client did not set the TimeAccess (atime) in
the Setattr done in an RPC after the one with the Open/exclusive_create.
This patch fixes the server to set the TimeAccess bit in the reply.
I believe that storing the create_verifier in an extended attribute for
file systems that support extended attributes might be a good idea,
but I will wait for a discussion of this on the freebsd-fs@ email list
before considering committing a patch to do this.
Reported by: jim@ks.uiuc.edu
Suggested by: dfr
MFC after: 2 weeks
Renumber cluase 4 to 3, per what everybody else did when BSD granted
them permission to remove clause 3. My insistance on keeping the same
numbering for legal reasons is too pedantic, so give up on that point.
Submitted by: Jan Schaumann <jschauma@stevens.edu>
Pull Request: https://github.com/freebsd/freebsd/pull/96
If dotdot lookup does not escape from the file descriptor passed as
the lookup root, we can allow the component traversal. Track the
directories traversed, and check the result of dotdot lookup against
the recorded list of the directory vnodes.
Dotdot lookups are enabled by sysctl vfs.lookup_cap_dotdot, currently
disabled by default until more verification of the approach is done.
Disallow non-local filesystems for dotdot, since remote server might
conspire with the local process to allow it to escape the namespace.
This might be too cautious, provide the knob
vfs.lookup_cap_dotdot_nonlocal to override as well.
Idea by: rwatson
Discussed with: emaste, jonathan, rwatson
Reviewed by: mjg (previous version)
Tested by: pho (previous version)
Sponsored by: The FreeBSD Foundation
MFC after: 2 week
Differential revision: https://reviews.freebsd.org/D8110
the patch in D1626 plus changes so that it includes counts for
NFSv4.1 (and the draft of NFSv4.2).
Also, make all the counts uint64_t and add a vers field at the
beginning, so that future revisions can easily be implemented.
There is code in place to handle the old vesion of the nfsstats
structure for backwards binary compatibility.
Subsequent commits will update nfsstat(8) to use the new fields.
Submitted by: will (earlier version)
Reviewed by: ken
MFC after: 1 month
Relnotes: yes
Differential Revision: https://reviews.freebsd.org/D1626
the NFS server would leave the newly created vnode locked. This could
result in a file system that would not unmount and processes wedged,
waiting for the file to be unlocked.
Since this VOP_SETATTR() never fails for most file systems, this bug
doesn't normally manifest itself. I found it during testing of an
exported GlusterFS file system, which can fail.
This patch adds the vput() and changes the error to the correct NFS one.
MFC after: 2 weeks
option that will be added to the nfsuserd daemon in a future
commit. It modifies the cache used by NFSv4 for name<-->id
translation (both username/uid and group/gid) to support this.
When "-manage-gids" is set, the server looks up each uid
for the RPC and uses the list of groups cached in the server
instead of the list of groups provided in the RPC request.
The cached group list is acquired for the cache by the nfsuserd
daemon via getgrouplist(3).
This avoids the 16 groups limit for the list in the RPC request.
Since the cache is now used for every RPC when "-manage-gids"
is enabled, the code also modifies the cache to use a separate
mutex for each hash list instead of a single global mutex.
Suggested by: jpaetzel
Tested by: jpaetzel
MFC after: 2 weeks
No appreciable change in performance was observed after increasing
the sizes of these tables and then testing with a single client.
However, there was an email that indicated high CPU overheads for
a heavily loaded NFSv4 and it is hoped that increasing the sizes
of the hash tables via these tunables might help.
The tables remain the same size by default.
Differential Revision: https://reviews.freebsd.org/D2596
MFC after: 2 weeks
use VOP_FSYNC() to perform the NFS server's Commit operation.
This patch adds a mnt_kern_flag called MNTK_USES_BCACHE which
is set by file systems that use the buffer cache. If this flag
is not set, the NFS server always does a VOP_FSYNC().
This should be ok for old file system modules that do not set
MNTK_USES_BCACHE, since calling VOP_FSYNC() is correct, although
it might not be optimal for file systems that use the buffer cache.
Reviewed by: kib
MFC after: 2 weeks
in the NFS server; garbage collect now-unused NFSMSIZ() and M_HASCL()
macros. Also garbage collect now-unused versions in headers for the
removed previous NFS client and server.
Reviewed by: rmacklem
Sponsored by: EMC / Isilon Storage Division
was reported via email. This was caused by a LOR between the
sleep lock used to serialize the local locking (nfsrv_locklf())
and locking the vnode. I believe this patch fixes the problem
by delaying relocking of the vnode until the sleep lock is
unlocked (nfsrv_unlocklf()). To avoid nfsvno_advlock() having the side
effect of unlocking the vnode, unlocking the vnode was moved to before
the functions that call nfsvno_advlock().
It shouldn't affect the execution of the default case where
vfs.nfsd.enable_locallocks=0.
Reported by: loic.blot@unix-experience.fr
Discussed with: kib
MFC after: 1 week
This fix addresses only issues with the pynfs reports, none of these
issues are know to create problems for extant real clients.
Submitted by: Bart Hsiao <bart.hsiao@gmail.com>
Reworked by: myself
Reviewed by: rmacklem
Approved by: rmacklem
Sponsored by: QNAP Systems Inc.
It overflows witness.
Shorten the names of some nfs mutexes.
Reported and tested by: pho
No objections from: rmacklem, mav
Sponsored by: The FreeBSD Foundation
MFC after: 1 week
the reply to ReaddirPlus when the server failed within the loop
that calls VFS_VGET(). This failure is most likely an error
return from VFS_VGET() caused by a bogus d_fileno that was
truncated to 32bits.
This patch fixes the server so that it will return directory postop
attributes for the failure. It does not fix the underlying issue caused
by d_fileno being uint32_t when a file system like ZFS generates
a fileno that is greater than 32bits.
Reported by: jpaetzel
Reviewed by: jpaetzel
MFC after: 1 month
into head. The code is not believed to have any effect
on the semantics of non-NFSv4.1 server behaviour.
It is a rather large merge, but I am hoping that there will
not be any regressions for the NFS server.
MFC after: 1 month
UFS rather than for all but ZFS. This code was assuming that offsets were
monotonically increasing for all file systems except ZFS and that the
cookies from a previous call may have been rewound to a block boundary.
According to mckusick@ only UFS is known to do this, so only requests against
UFS file systems should remove cookies smaller than the given offset. This
fixes serving TMPFS over NFS as it too does not have monotonically increasing
offsets. The comment around the code also indicated it was specific to UFS.
Some of the code using 'not_zfs' is specific to ZFS snapshot handling, so
add a 'is_zfs' variable for those cases.
It's possible that 'is_zfs' check for VFS_VGET() support may not be
specific to ZFS. This needs more research and testing.
After this fix TMPFS and other file systems can be served over NFS.
To test I compared the results of syncing a /usr/src tree into a tmpfs and
serving that over NFS. Before the fix 3589 files were missing on the remote
view. After the fix all files were successfully found.
Reviewed by: rmacklem
Discussed with: mckusick, rmacklem via fs@
Discussed at: http://lists.freebsd.org/pipermail/freebsd-fs/2014-April/019264.html
MFC after: 2 weeks
Sponsored by: EMC / Isilon Storage Division
should either accept owner and owner_group strings that are just
the digits of the uid/gid or return NFS4ERR_BADOWNER.
This patch adds a sysctl vfs.nfsd.enable_stringtouid, which can
be set to enable the server w.r.t. accepting numeric string. It
also ensures that NFS4ERR_BADOWNER is returned if numeric uid/gid
strings are not enabled. This fixes the server for recent Linux
nfs4 clients that use numeric uid/gid strings by default.
Reported and tested by: craigyk@gmail.com
MFC after: 2 weeks
for the VOP_READ() call. This patch fixes both the old and new
server for this case.
PR: 185232
Submitted by: PR had patch for old server
Reviewed by: kib
MFC after: 2 weeks
further refinement is required as some device drivers intended to be
portable over FreeBSD versions rely on __FreeBSD_version to decide whether
to include capability.h.
MFC after: 3 weeks
- Introduce additional hash to group requests by hash of sockref. This
allows to process TCP acknowledgements without looping though all the cache,
and as result allows to do it every time.
- Indroduce additional callbacks to notify application layer about sockets
disconnection. Without this last few requests processed just before socket
disconnection never processed their ACKs and stuck in cache for many hours.
- Implement transport-specific method for tracking reply acknowledgements.
New implementation does not cross multiple stack layers to get the data and
does not have race conditions that previously made some requests stuck
in cache. This could be done more efficiently at sockbuf layer, but that
would broke some KBIs, while I don't know other consumers for it aside NFS.
- Instead of traversing all DRC twice per request, run cleaning only once
per request, and except in some conditions traverse only single hash slot
at a time.
Together this limits NFS DRC growth only to situations of real connectivity
problems. If network is working well, and so all replies are acknowledged,
cache remains almost empty even after hours of heavy load. Without this
change on the same test cache was growing to many thousand requests even
with perfectly working local network.
As another result this reduces CPU time spent on the DRC handling during
SPEC NFS benchmark from about 10% to 0.5%.
Sponsored by: iXsystems, Inc.
when a Getattr for a file is done by a client other than the one that
holds the file's delegation. This would only happen when delegations
are enabled and the problem is fixed by this patch.
MFC after: 1 week
reported to the freebsd-fs mailing list. I believe the problem was
caused by the Readdir operation using VFS_VGET() for a snapshot file entry
instead of VOP_LOOKUP(). This would not occur for NFSv3, since it
will do a VFS_VGET() of "." which fails with ENOTSUPP at the beginning
of the directory, whereas NFSv4 does not check "." or "..". This
patch adds a call to VFS_VGET() for the directory being read to check
for ENOTSUPP.
I also observed that the mount_on_fileid and fsid attributes were
not correct at the snapshot's auto mountpoints when looking at packet
traces for the Readdir. This patch fixes the attributes by doing a check
for different v_mount structure, even if the vnode v_mountedhere is not
set.
Reported by: jas@cse.yorku.ca
Tested by: jas@cse.yorku.ca
Reviewed by: asomers
MFC after: 1 week
in the future in a backward compatible (API and ABI) way.
The cap_rights_t represents capability rights. We used to use one bit to
represent one right, but we are running out of spare bits. Currently the new
structure provides place for 114 rights (so 50 more than the previous
cap_rights_t), but it is possible to grow the structure to hold at least 285
rights, although we can make it even larger if 285 rights won't be enough.
The structure definition looks like this:
struct cap_rights {
uint64_t cr_rights[CAP_RIGHTS_VERSION + 2];
};
The initial CAP_RIGHTS_VERSION is 0.
The top two bits in the first element of the cr_rights[] array contain total
number of elements in the array - 2. This means if those two bits are equal to
0, we have 2 array elements.
The top two bits in all remaining array elements should be 0.
The next five bits in all array elements contain array index. Only one bit is
used and bit position in this five-bits range defines array index. This means
there can be at most five array elements in the future.
To define new right the CAPRIGHT() macro must be used. The macro takes two
arguments - an array index and a bit to set, eg.
#define CAP_PDKILL CAPRIGHT(1, 0x0000000000000800ULL)
We still support aliases that combine few rights, but the rights have to belong
to the same array element, eg:
#define CAP_LOOKUP CAPRIGHT(0, 0x0000000000000400ULL)
#define CAP_FCHMOD CAPRIGHT(0, 0x0000000000002000ULL)
#define CAP_FCHMODAT (CAP_FCHMOD | CAP_LOOKUP)
There is new API to manage the new cap_rights_t structure:
cap_rights_t *cap_rights_init(cap_rights_t *rights, ...);
void cap_rights_set(cap_rights_t *rights, ...);
void cap_rights_clear(cap_rights_t *rights, ...);
bool cap_rights_is_set(const cap_rights_t *rights, ...);
bool cap_rights_is_valid(const cap_rights_t *rights);
void cap_rights_merge(cap_rights_t *dst, const cap_rights_t *src);
void cap_rights_remove(cap_rights_t *dst, const cap_rights_t *src);
bool cap_rights_contains(const cap_rights_t *big, const cap_rights_t *little);
Capability rights to the cap_rights_init(), cap_rights_set(),
cap_rights_clear() and cap_rights_is_set() functions are provided by
separating them with commas, eg:
cap_rights_t rights;
cap_rights_init(&rights, CAP_READ, CAP_WRITE, CAP_FSTAT);
There is no need to terminate the list of rights, as those functions are
actually macros that take care of the termination, eg:
#define cap_rights_set(rights, ...) \
__cap_rights_set((rights), __VA_ARGS__, 0ULL)
void __cap_rights_set(cap_rights_t *rights, ...);
Thanks to using one bit as an array index we can assert in those functions that
there are no two rights belonging to different array elements provided
together. For example this is illegal and will be detected, because CAP_LOOKUP
belongs to element 0 and CAP_PDKILL to element 1:
cap_rights_init(&rights, CAP_LOOKUP | CAP_PDKILL);
Providing several rights that belongs to the same array's element this way is
correct, but is not advised. It should only be used for aliases definition.
This commit also breaks compatibility with some existing Capsicum system calls,
but I see no other way to do that. This should be fine as Capsicum is still
experimental and this change is not going to 9.x.
Sponsored by: The FreeBSD Foundation
DRC for NFS over TCP.
- Increase the size of the hash tables.
- Create a separate mutex for each hash list of the TCP hash table.
- Single thread the code that deletes stale cache entries.
- Add a tunable called vfs.nfsd.tcphighwater, which can be increased
to allow the cache to grow larger, avoiding the overhead of frequent
scans to delete stale cache entries.
(The default value will result in frequent scans to delete stale cache
entries, analagous to what the pre-patched code does.)
- Add a tunable called vfs.nfsd.cachetcp that can be used to disable
DRC caching for NFS over TCP, since the old NFS server didn't DRC cache TCP.
It also adjusts the size of nfsrc_floodlevel dynamically, so that it is
always greater than vfs.nfsd.tcphighwater.
For UDP the algorithm remains the same as the pre-patched code, but the
tunable vfs.nfsd.udphighwater can be used to allow the cache to grow
larger and reduce the overhead caused by frequent scans for stale entries.
UDP also uses a larger hash table size than the pre-patched code.
Reported by: wollman
Tested by: wollman (earlier version of patch)
Submitted by: ivoras (earlier patch)
Reviewed by: jhb (earlier version of patch)
MFC after: 1 month
- 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
it will work with either the old or new server.
The FHA code keeps a cache of currently active file handles for
NFSv2 and v3 requests, so that read and write requests for the same
file are directed to the same group of threads (reads) or thread
(writes). It does not currently work for NFSv4 requests. They are
more complex, and will take more work to support.
This improves read-ahead performance, especially with ZFS, if the
FHA tuning parameters are configured appropriately. Without the
FHA code, concurrent reads that are part of a sequential read from
a file will be directed to separate NFS threads. This has the
effect of confusing the ZFS zfetch (prefetch) code and makes
sequential reads significantly slower with clients like Linux that
do a lot of prefetching.
The FHA code has also been updated to direct write requests to nearby
file offsets to the same thread in the same way it batches reads,
and the FHA code will now also send writes to multiple threads when
needed.
This improves sequential write performance in ZFS, because writes
to a file are now more ordered. Since NFS writes (generally
less than 64K) are smaller than the typical ZFS record size
(usually 128K), out of order NFS writes to the same block can
trigger a read in ZFS. Sending them down the same thread increases
the odds of their being in order.
In order for multiple write threads per file in the FHA code to be
useful, writes in the NFS server have been changed to use a LK_SHARED
vnode lock, and upgrade that to LK_EXCLUSIVE if the filesystem
doesn't allow multiple writers to a file at once. ZFS is currently
the only filesystem that allows multiple writers to a file, because
it has internal file range locking. This change does not affect the
NFSv4 code.
This improves random write performance to a single file in ZFS, since
we can now have multiple writers inside ZFS at one time.
I have changed the default tuning parameters to a 22 bit (4MB)
window size (from 256K) and unlimited commands per thread as a
result of my benchmarking with ZFS.
The FHA code has been updated to allow configuring the tuning
parameters from loader tunable variables in addition to sysctl
variables. The read offset window calculation has been slightly
modified as well. Instead of having separate bins, each file
handle has a rolling window of bin_shift size. This minimizes
glitches in throughput when shifting from one bin to another.
sys/conf/files:
Add nfs_fha_new.c and nfs_fha_old.c. Compile nfs_fha.c
when either the old or the new NFS server is built.
sys/fs/nfs/nfsport.h,
sys/fs/nfs/nfs_commonport.c:
Bring in changes from Rick Macklem to newnfs_realign that
allow it to operate in blocking (M_WAITOK) or non-blocking
(M_NOWAIT) mode.
sys/fs/nfs/nfs_commonsubs.c,
sys/fs/nfs/nfs_var.h:
Bring in a change from Rick Macklem to allow telling
nfsm_dissect() whether or not to wait for mallocs.
sys/fs/nfs/nfsm_subs.h:
Bring in changes from Rick Macklem to create a new
nfsm_dissect_nonblock() inline function and
NFSM_DISSECT_NONBLOCK() macro.
sys/fs/nfs/nfs_commonkrpc.c,
sys/fs/nfsclient/nfs_clkrpc.c:
Add the malloc wait flag to a newnfs_realign() call.
sys/fs/nfsserver/nfs_nfsdkrpc.c:
Setup the new NFS server's RPC thread pool so that it will
call the FHA code.
Add the malloc flag argument to newnfs_realign().
Unstaticize newnfs_nfsv3_procid[] so that we can use it in
the FHA code.
sys/fs/nfsserver/nfs_nfsdsocket.c:
In nfsrvd_dorpc(), add NFSPROC_WRITE to the list of RPC types
that use the LK_SHARED lock type.
sys/fs/nfsserver/nfs_nfsdport.c:
In nfsd_fhtovp(), if we're starting a write, check to see
whether the underlying filesystem supports shared writes.
If not, upgrade the lock type from LK_SHARED to LK_EXCLUSIVE.
sys/nfsserver/nfs_fha.c:
Remove all code that is specific to the NFS server
implementation. Anything that is server-specific is now
accessed through a callback supplied by that server's FHA
shim in the new softc.
There are now separate sysctls and tunables for the FHA
implementations for the old and new NFS servers. The new
NFS server has its tunables under vfs.nfsd.fha, the old
NFS server's tunables are under vfs.nfsrv.fha as before.
In fha_extract_info(), use callouts for all server-specific
code. Getting file handles and offsets is now done in the
individual server's shim module.
In fha_hash_entry_choose_thread(), change the way we decide
whether two reads are in proximity to each other.
Previously, the calculation was a simple shift operation to
see whether the offsets were in the same power of 2 bucket.
The issue was that there would be a bucket (and therefore
thread) transition, even if the reads were in close
proximity. When there is a thread transition, reads wind
up going somewhat out of order, and ZFS gets confused.
The new calculation simply tries to see whether the offsets
are within 1 << bin_shift of each other. If they are, the
reads will be sent to the same thread.
The effect of this change is that for sequential reads, if
the client doesn't exceed the max_reqs_per_nfsd parameter
and the bin_shift is set to a reasonable value (22, or
4MB works well in my tests), the reads in any sequential
stream will largely be confined to a single thread.
Change fha_assign() so that it takes a softc argument. It
is now called from the individual server's shim code, which
will pass in the softc.
Change fhe_stats_sysctl() so that it takes a softc
parameter. It is now called from the individual server's
shim code. Add the current offset to the list of things
printed out about each active thread.
Change the num_reads and num_writes counters in the
fha_hash_entry structure to 32-bit values, and rename them
num_rw and num_exclusive, respectively, to reflect their
changed usage.
Add an enable sysctl and tunable that allows the user to
disable the FHA code (when vfs.XXX.fha.enable = 0). This
is useful for before/after performance comparisons.
nfs_fha.h:
Move most structure definitions out of nfs_fha.c and into
the header file, so that the individual server shims can
see them.
Change the default bin_shift to 22 (4MB) instead of 18
(256K). Allow unlimited commands per thread.
sys/nfsserver/nfs_fha_old.c,
sys/nfsserver/nfs_fha_old.h,
sys/fs/nfsserver/nfs_fha_new.c,
sys/fs/nfsserver/nfs_fha_new.h:
Add shims for the old and new NFS servers to interface with
the FHA code, and callbacks for the
The shims contain all of the code and definitions that are
specific to the NFS servers.
They setup the server-specific callbacks and set the server
name for the sysctl and loader tunable variables.
sys/nfsserver/nfs_srvkrpc.c:
Configure the RPC code to call fhaold_assign() instead of
fha_assign().
sys/modules/nfsd/Makefile:
Add nfs_fha.c and nfs_fha_new.c.
sys/modules/nfsserver/Makefile:
Add nfs_fha_old.c.
Reviewed by: rmacklem
Sponsored by: Spectra Logic
MFC after: 2 weeks
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
- Capability is no longer separate descriptor type. Now every descriptor
has set of its own capability rights.
- The cap_new(2) system call is left, but it is no longer documented and
should not be used in new code.
- The new syscall cap_rights_limit(2) should be used instead of
cap_new(2), which limits capability rights of the given descriptor
without creating a new one.
- The cap_getrights(2) syscall is renamed to cap_rights_get(2).
- If CAP_IOCTL capability right is present we can further reduce allowed
ioctls list with the new cap_ioctls_limit(2) syscall. List of allowed
ioctls can be retrived with cap_ioctls_get(2) syscall.
- If CAP_FCNTL capability right is present we can further reduce fcntls
that can be used with the new cap_fcntls_limit(2) syscall and retrive
them with cap_fcntls_get(2).
- To support ioctl and fcntl white-listing the filedesc structure was
heavly modified.
- The audit subsystem, kdump and procstat tools were updated to
recognize new syscalls.
- Capability rights were revised and eventhough I tried hard to provide
backward API and ABI compatibility there are some incompatible changes
that are described in detail below:
CAP_CREATE old behaviour:
- Allow for openat(2)+O_CREAT.
- Allow for linkat(2).
- Allow for symlinkat(2).
CAP_CREATE new behaviour:
- Allow for openat(2)+O_CREAT.
Added CAP_LINKAT:
- Allow for linkat(2). ABI: Reuses CAP_RMDIR bit.
- Allow to be target for renameat(2).
Added CAP_SYMLINKAT:
- Allow for symlinkat(2).
Removed CAP_DELETE. Old behaviour:
- Allow for unlinkat(2) when removing non-directory object.
- Allow to be source for renameat(2).
Removed CAP_RMDIR. Old behaviour:
- Allow for unlinkat(2) when removing directory.
Added CAP_RENAMEAT:
- Required for source directory for the renameat(2) syscall.
Added CAP_UNLINKAT (effectively it replaces CAP_DELETE and CAP_RMDIR):
- Allow for unlinkat(2) on any object.
- Required if target of renameat(2) exists and will be removed by this
call.
Removed CAP_MAPEXEC.
CAP_MMAP old behaviour:
- Allow for mmap(2) with any combination of PROT_NONE, PROT_READ and
PROT_WRITE.
CAP_MMAP new behaviour:
- Allow for mmap(2)+PROT_NONE.
Added CAP_MMAP_R:
- Allow for mmap(PROT_READ).
Added CAP_MMAP_W:
- Allow for mmap(PROT_WRITE).
Added CAP_MMAP_X:
- Allow for mmap(PROT_EXEC).
Added CAP_MMAP_RW:
- Allow for mmap(PROT_READ | PROT_WRITE).
Added CAP_MMAP_RX:
- Allow for mmap(PROT_READ | PROT_EXEC).
Added CAP_MMAP_WX:
- Allow for mmap(PROT_WRITE | PROT_EXEC).
Added CAP_MMAP_RWX:
- Allow for mmap(PROT_READ | PROT_WRITE | PROT_EXEC).
Renamed CAP_MKDIR to CAP_MKDIRAT.
Renamed CAP_MKFIFO to CAP_MKFIFOAT.
Renamed CAP_MKNODE to CAP_MKNODEAT.
CAP_READ old behaviour:
- Allow pread(2).
- Disallow read(2), readv(2) (if there is no CAP_SEEK).
CAP_READ new behaviour:
- Allow read(2), readv(2).
- Disallow pread(2) (CAP_SEEK was also required).
CAP_WRITE old behaviour:
- Allow pwrite(2).
- Disallow write(2), writev(2) (if there is no CAP_SEEK).
CAP_WRITE new behaviour:
- Allow write(2), writev(2).
- Disallow pwrite(2) (CAP_SEEK was also required).
Added convinient defines:
#define CAP_PREAD (CAP_SEEK | CAP_READ)
#define CAP_PWRITE (CAP_SEEK | CAP_WRITE)
#define CAP_MMAP_R (CAP_MMAP | CAP_SEEK | CAP_READ)
#define CAP_MMAP_W (CAP_MMAP | CAP_SEEK | CAP_WRITE)
#define CAP_MMAP_X (CAP_MMAP | CAP_SEEK | 0x0000000000000008ULL)
#define CAP_MMAP_RW (CAP_MMAP_R | CAP_MMAP_W)
#define CAP_MMAP_RX (CAP_MMAP_R | CAP_MMAP_X)
#define CAP_MMAP_WX (CAP_MMAP_W | CAP_MMAP_X)
#define CAP_MMAP_RWX (CAP_MMAP_R | CAP_MMAP_W | CAP_MMAP_X)
#define CAP_RECV CAP_READ
#define CAP_SEND CAP_WRITE
#define CAP_SOCK_CLIENT \
(CAP_CONNECT | CAP_GETPEERNAME | CAP_GETSOCKNAME | CAP_GETSOCKOPT | \
CAP_PEELOFF | CAP_RECV | CAP_SEND | CAP_SETSOCKOPT | CAP_SHUTDOWN)
#define CAP_SOCK_SERVER \
(CAP_ACCEPT | CAP_BIND | CAP_GETPEERNAME | CAP_GETSOCKNAME | \
CAP_GETSOCKOPT | CAP_LISTEN | CAP_PEELOFF | CAP_RECV | CAP_SEND | \
CAP_SETSOCKOPT | CAP_SHUTDOWN)
Added defines for backward API compatibility:
#define CAP_MAPEXEC CAP_MMAP_X
#define CAP_DELETE CAP_UNLINKAT
#define CAP_MKDIR CAP_MKDIRAT
#define CAP_RMDIR CAP_UNLINKAT
#define CAP_MKFIFO CAP_MKFIFOAT
#define CAP_MKNOD CAP_MKNODAT
#define CAP_SOCK_ALL (CAP_SOCK_CLIENT | CAP_SOCK_SERVER)
Sponsored by: The FreeBSD Foundation
Reviewed by: Christoph Mallon <christoph.mallon@gmx.de>
Many aspects discussed with: rwatson, benl, jonathan
ABI compatibility discussed with: kib
In particular, do not lock Giant conditionally when calling into the
filesystem module, remove the VFS_LOCK_GIANT() and related
macros. Stop handling buffers belonging to non-mpsafe filesystems.
The VFS_VERSION is bumped to indicate the interface change which does
not result in the interface signatures changes.
Conducted and reviewed by: attilio
Tested by: pho
processes running as root to suspend/resume execution
of the kernel nfsd threads. An earlier version of this
patch was tested by Vincent Hoffman (vince at unsane.co.uk)
and John Hickey (jh at deterlab.net).
Reviewed by: kib
MFC after: 2 weeks
If you have a binary on a filesystem which is also mounted over by
nullfs, you could execute the binary from the lower filesystem, or
from the nullfs mount. When executed from lower filesystem, the lower
vnode gets VV_TEXT flag set, and the file cannot be modified while the
binary is active. But, if executed as the nullfs alias, only the
nullfs vnode gets VV_TEXT set, and you still can open the lower vnode
for write.
Add a set of VOPs for the VV_TEXT query, set and clear operations,
which are correctly bypassed to lower vnode.
Tested by: pho (previous version)
MFC after: 2 weeks
and owner_group strings that consist entirely of
digits, interpreting them as the uid/gid number.
This change was needed since new (>= 3.3) Linux
servers reply with these strings by default.
This change is mandated by the rfc3530bis draft.
Reported on freebsd-stable@ under the Subject
heading "Problem with Linux >= 3.3 as NFSv4 server"
by Norbert Aschendorff on Aug. 20, 2012.
Tested by: norbert.aschendorff at yahoo.de
Reviewed by: jhb
MFC after: 2 weeks
ZFS volume is exported via the new NFS server. The leak occurred
because the new NFS server code didn't handle the case where
a file system sets the SAVENAME flag in its VOP_LOOKUP() and
ZFS does this for the DELETE case.
Tested by: Oliver Brandmueller (ob at gruft.de), hrs
PR: kern/167266
MFC after: 1 month
