Convert all the NFS macros (in nfsm_subs.h) into inline functions with the
appropriate extra arguments.

This (hopefully) completes the unmessyfication work started by thib@ a long,
long time ago (in

Convert all the NFS macros (in nfsm_subs.h) into inline functions with the
appropriate extra arguments.

This (hopefully) completes the unmessyfication work started by thib@ a long,
long time ago (in a galaxy far away).

The conversion logic has been:
- nfsm_dissect has been turned into an rvalue expression, leaving the
cast operation up to its caller.
- macros which had three different exit paths (return, goto nfsmout or
fallthrough) have been split so that no macros have more than two exit paths.
- then they have been modified to return a value, which lets the caller
figure out what exit path is needed.
- local variables abused by the macros are now local variables of the new
inline functions.

This single commit is the sum of 25 intermediate diffs, which have all been
carefully reviewed by (at least) jsg@ and semarie@.

Tested with v2 and v3 servers and clients.

ok jsg@ semarie@

show more ...

spelling

Introduce a dedicated entry point data structure for file locks. This new data
structure allows for better tracking of pending lock operations which is
essential in order to prevent a use-after-free

Introduce a dedicated entry point data structure for file locks. This new data
structure allows for better tracking of pending lock operations which is
essential in order to prevent a use-after-free once the underlying vnode is
gone.

Inspired by the lockf implementation in FreeBSD.

ok visa@

Reported-by: syzbot+d5540a236382f50f1dac@syzkaller.appspotmail.com

show more ...

Implement proper locking for NFS nodes.

Tested in bulks by many. ok visa@, beck@

Make sillyrename not so damn silly. The kindergarten algorithm here for
picking a name meant that more than 58 sillys in a directory and we fail
with EINVAL, resulting in strange problems for nfs whi

Make sillyrename not so damn silly. The kindergarten algorithm here for
picking a name meant that more than 58 sillys in a directory and we fail
with EINVAL, resulting in strange problems for nfs which in turn causes
pain and stress in building, and PTSD in nfs and vfs hackers. Has bit us
in the butt since the vienna f2k7 hackathon.

good suggestions from deraadt@ guenther@ and otto@

ok deraadt@,oga@,blambert@,krw@,guenther@, and a "very special ok" tedu@

Oh god, I'm an nfs hacker..

show more ...

Backout the asyncio/aiod change, as it causes buf's to get hung.
problem noticed by deraadt@

ok beck@

introduce a flag member to struct nfs_aiod, and use flags instead of the exit
and worked members. nad_worked becomes NFSAIOD_WAKEUP, which is set after if
an aiod was removed from the idle list and w

introduce a flag member to struct nfs_aiod, and use flags instead of the exit
and worked members. nad_worked becomes NFSAIOD_WAKEUP, which is set after if
an aiod was removed from the idle list and woken up by nfs_asyncio().

don't rely on tsleep wchans being unique, that is keep going back to sleep if
woken up unless the NFSAIOD_WAKEUP flag is set.

fix a divide by zero crash if nfs.vfs.iothreads is set to 0, as that can happen
when we recalculate the maximum buf's to queue up for each aiod.

in nfs_asyncio() set the nad_mnt to NULL before returning the aiod back to the
idle list in the case where we have already queued up to many bufs, otherwise
we trip an assertion.

minimize the time we are holding the nfs_aiodl_mtx to only when we are inserting
or removing from the lists, with the exception of nfs_set_naiod() as it would
make the loops more complicated and its uncommon in any case.

tested by myself and deraadt@
"fine with me" deraadt@

show more ...

make sure that an aiod has been removed from the nfs_aiods_idle list
before inserting it back into the list.

crashes debugged with help from deraadt@ who also tested this fix.

Rework the way we do async I/O in nfs. Introduce separate buf queues for
each mount, and when work is "found", peg an aiod to that mount todo the
I/O. Make nfs_asyncio() a bit smarter when deciding w

Rework the way we do async I/O in nfs. Introduce separate buf queues for
each mount, and when work is "found", peg an aiod to that mount todo the
I/O. Make nfs_asyncio() a bit smarter when deciding when to do asyncio
and when to force it sync, this is done by keeping the aiod's one two lists,
an "idle" and an "all" list, so asyncio is only done when there are aiods
hanging around todo it for us or are already pegged to the mount.

Idea liked by at least beck@ (and I think art@).
Extensive testing done by myself and jasper and a few others on various
arch's.

Ideas/Code from Net/Free.

OK blambert@.

show more ...

Use an RB tree instead of a hashtable for fh/node lookups.

Idea from NetBSD.

OK blambert@

Use a timespec instead of a time_t for the clients nfsnode
mtime, gives us better granularity, helps with cache consistency.

Idea lifted from NetBSD.

OK blambert@

Introduce a macro to invalidate the attribute
cache instead of setting n_attrstamp to 0 directly.

Lift the macro name from NetBSD.
prompted by and OK blambert@

Ensure each nfsiod can actually enqueue more than one asynchio - this mirrors
the accidental situation that used to happen when it leaked buffers and allowed
the syncer to do it, however this puts a

Ensure each nfsiod can actually enqueue more than one asynchio - this mirrors
the accidental situation that used to happen when it leaked buffers and allowed
the syncer to do it, however this puts a limit on how much of the buffer cache
it is allowed to consume to a sensible amount - improves nfs write performance
since we don't have to do tons of them synch now.

Modifies the existing code to use wakeup_one instead of cruft, and now
all nfsiod's tsleep the same way.

ok thib@ art@

show more ...

o Avoid putting duplicate entries into the name cache,
add a function to handle cache_enter()'s for us since
we need to update the appropriate timestamps so we don't
miss on lookup;
o Do not pu

o Avoid putting duplicate entries into the name cache,
add a function to handle cache_enter()'s for us since
we need to update the appropriate timestamps so we don't
miss on lookup;
o Do not purge the dvp in nfs_rmdir(), since there's no need;
o cache the new entry created in nfs_mkdir();
o Do caching of the access modes, this drastically reduces the
amount of over the wire access RPCs we do in the NFSv3 case.

This diff was written by Pedro Martelletto, sometime in 2004.
A lot of people have been running with it at one time or another,
this includes at least markus and matthieu.

OK deraadt@, blambert@

show more ...

Garbage collect nfsdmap and leftover code. That structure
was used as a part of the "old-style directory caching"
that was removed in 2001;

ok beck@, blambert@

remove some unused members of struct nfsnode;
some spacing while there.

ok krw@

MALLOC/FREE -> malloc/free + M_ZERO.
Uneeded includes and casts...

ok krw@

remove a whole bunch of duplicate function
prototypes.
(survived build/release on macppc/amd64).

ok pedro@

remove unused members from struct nfsmount and nfsnode;

ok pedro@

get rid of noop vop_reallocblks; pedro@ miod@ ok

Retire VOP_LEASE(); It was a bit for NQNFS and has
effectively been a no-op for quite some time now,
without promise for future usage.

ok pedro@
Testing by krw@ (earlier diff)
and Johan Mson Lindman

Retire VOP_LEASE(); It was a bit for NQNFS and has
effectively been a no-op for quite some time now,
without promise for future usage.

ok pedro@
Testing by krw@ (earlier diff)
and Johan Mson Lindman (tybollt@solace.miun.se)

show more ...

NFS commit coalescion: instead of sending a commit for each block, coalesce
these into larger ranges wherever possible.

this should speed up NFS writes quite a bit.

ok art@ millert@ pedro@ tedu@

nfs_poll cannot simply be defined to seltrue so use a stub nfs_poll
duplicated from UFS. Closes PR 3757; OK deraadt@

silly fifos aren't a real fs, so we have to be sure to call reclaim
in the host fs.

Replace select backends with poll backends. selscan() and pollscan()
now call the poll backend. With this change we implement greater
poll(2) functionality instead of emulating it via the select ba

Replace select backends with poll backends. selscan() and pollscan()
now call the poll backend. With this change we implement greater
poll(2) functionality instead of emulating it via the select backend.
Adapted from NetBSD and including some changes from FreeBSD.
Tested by many, deraadt@ OK

show more ...