Some notable patches, nothing major:

- 36b35911 "fuse: fix deadlock with parallel mounts"

- f91e3081 "fuse: debug log: print PID"

- 334a5a5f "fuse: update for kernel API version 40"

- afb9051a "fuse: provide InitIn/InitOut.Flags64()"

There are other changes which add e.g. passthrough support and other
features, but they apply mostly to fs package, not nodefs, and I
reviewed them to be safe and orthogonal to what WCFS uses and needs.

* master: (53 commits)
  fs: skip passthrough test for older kernels
  fuse: provide InitIn/InitOut.Flags64()
  fs: reorganize code
  fs: support passthrough mode
  fuse: expose BackingFdMap
  fuse: fix printing of new constant
  fuse: remove deprecated CAP_DIRECT_IO_RELAX
  fuse: print tweaks for passthrough support
  fs: return fileEntry from registerFile and addNewChild
  benchmark: add a test reading from libfuse passthrough_hp
  benchmark: modernize setupFS
  benchmark: introduce benchmark for loopback reads
  benchmark: simplify ReadLines
  fs: fix stray comment
  fuse: wire up passthrough support
  fuse: point comment to fs API rather than nodefs/pathfs
  fuse: move Flags2 handling to platform specific code
  fuse: update for kernel API version 40
  fuse: fix CreateOut print
  fuse: document singleReader background
  ...

Change-Id: I8e23afdfeb6ff65211f6ac3413e8e86e66341313

Change-Id: I7e23afdfeb6ff65211f6ac3413e8e86e66341313

Change-Id: I91a375cc2956b2ead0ad3a4a5b2a0c66a4c327e2

If a returned file implements the FilePassthroughFder interface, we
try to register the file in the kernel.

Implement this for loopbackFile, and test the behavior.

For benchmarking, use a single reader. With multiple readers, contents
are served out of kernel cache, and do not reflect FUSE performance. 

Benchmark (CPU i5-8350U pinned at 2Ghz):

$ go build -v && go test -run "abc" -bench '(Libfuse|FD)' --passthrough_hp ~/vc/libfuse/build/example/passthrough_hp -test.cpu=1
BenchmarkGoFuseFDRead 	   27444	     45997 ns/op	1424.80 MB/s	      87 B/op	       1 allocs/op
BenchmarkLibfuseHP    	   35377	     32198 ns/op	2035.43 MB/s	       0 B/op	       0 allocs/op

$ go build -v && sudo go test -run "abc" -bench '(Libfuse|FD)' --passthrough_hp ~/vc/libfuse/build/example/passthrough_hp -test.cpu=1
BenchmarkGoFuseFDRead 	   91788	     11902 ns/op	5506.23 MB/s	       3 B/op	       0 allocs/op
BenchmarkLibfuseHP    	  100556	     11831 ns/op	5539.38 MB/s	       0 B/op	       0 allocs/op

Change-Id: If8bde502a3450028f4d87ba61fa9c76ea3ea6c63

This prepares for extension to the passthrough support, in particular,
the ops_mask to be added to `struct fuse_backing_map`.

Change-Id: If4d7c321a5d09e9b2a3b804bd5e434aab394fe30

Even trivial changes need testing (sigh).

Change-Id: I1d009e97b51037544fc90392999587d2472ff13a

This is technically a backward incompatible change, but this feature
is currently not supported.

Change-Id: Ifcea9a337a06277f70a23eeee66db8d4b661b046

Change-Id: I513d6344ff0d22649c240bbf31ef901dca70f4e2

Also create a fresh fileEntry even if we reuse the file handle number.

Change-Id: Ic3f37f8ad6d916f84b0dbcb8268f195161aac4ad

passthrough_hp is the high-performance loopback filesystem. By
exercising it, we can distinguish between kernel bugs (shared between
go-fuse and libfuse) and go-fuse bugs.

Passthrough mode does not work with iflag=direct, because
passthrough_hp does not check if the file is marked "directio",
assuming read() calls in passthrough mode are some kind of error.

Change-Id: I2b6b4c943031e1a2864e15e5bf396aac28c63e31

Change-Id: I0dd07cf43fded2aa28ff83b2797e9222ae6e1295

Loopback (file backed by some other file on disk) is an important
use-case, and FUSE features such as passthrough and splicing aim to
reduce overhead.

Example: 

    Splice disabled:

    $ go test -bench FD -test.cpu=1,2
    goos: linux
    goarch: amd64
    pkg: github.com/hanwen/go-fuse/v2/benchmark
    cpu: Intel(R) Core(TM) i5-8350U CPU @ 1.70GHz
    BenchmarkGoFuseFDRead     	    1929	    597942 ns/op	3507.28 MB/s	    4290 B/op	     129 allocs/op
    BenchmarkGoFuseFDRead-2   	    2497	    421856 ns/op	4971.26 MB/s	   62500 B/op	     177 allocs/op
    PASS
    ok  	github.com/hanwen/go-fuse/v2/benchmark	3.510s

    Splice enabled:

    $ go test -bench FD -test.cpu=1,2
    goos: linux
    goarch: amd64
    pkg: github.com/hanwen/go-fuse/v2/benchmark
    cpu: Intel(R) Core(TM) i5-8350U CPU @ 1.70GHz
    BenchmarkGoFuseFDRead     	    1524	    750249 ns/op	2795.28 MB/s	    4132 B/op	      98 allocs/op
    BenchmarkGoFuseFDRead-2   	    2364	    463522 ns/op	4524.39 MB/s	   11719 B/op	     106 allocs/op
    PASS

in other words, the API complexity that splicing introduced was not
worth it: it actually slows down things.

Change-Id: I4f7519ab2cc5b0145d0f7f81f55ad20d544e9d87

Change-Id: I683c0f614f67e787229d0754354a25f225184995

Change-Id: Iad80862a8d13cb03511a005d385735594ff5e353

The passthrough feature lets a filesystem register an open file
descriptor with the kernel to avoid roundtrips for read/write
operations.

Change-Id: Ia8bde502a3450028f4d87ba61fa9c76ea3ea6c63

Change-Id: I18bde502a3450028f4d87ba61fa9c76ea3ea6c63

Document why some bits are platform dependent.

Change-Id: Ib8bde502a3450028f4d87ba61fa9c76ea3ea6c63

Add definitions and print routines for FUSE_NO_EXPORT_SUPPORT,
FUSE_NOTIFY_RESEND and FOPEN_PASSTHROUGH and associated fields.

Change-Id: I5bd52c44a91484ac78ea5413cbd4bf5dc927501e

Change-Id: I331f9f702fa0a850df2d6f1e3dd75f9f550750ed

Multiple reader code was introduced in
03081e9d (May 2012). Redo the
benchmark to verify that this still makes sense.

Change-Id: I03432b93285d3e3e9494858c844be1ff2dbd3147

This enables using go-fuse in a init binary that is spawned directly from
the kernel.

See also https://github.com/hanwen/go-fuse/pull/525.

Change-Id: Ibbaa04af993677536bd4ae8a173a5a6d93118940

mips64 uses 0x4000 for LARGEFILE. See also issue #514	

Change-Id: I46e83824f412131961318b16767bc05273fa2fbe

Add `isTest` to avoid panicking outside of test scenarios, in case we
mess up the flag definitions. The latter can easily happen, due to
subtle differences between architectures.

Change-Id: I4085de49e39eddf328269a7a3e694818c693e195

This will allow duplicate keys in the initialization.

Change-Id: I16969bddd1aad0232477379229a0afbe8faf1933

gvfsd-trash sets an inotify watch on mntDir and stat()s every file that is
created, racing with the test logic.

Create our files in a subdir to avoid that, as gvfsd-trash does not monitor
subdirs.

Fixes: https://github.com/hanwen/go-fuse/issues/478
Change-Id: I8e297a16af7fdd8b8c9574d456bbf43d36eddd67

It is sometimes helpful to know which process accesses
the filesystem.

Relates-to: https://github.com/hanwen/go-fuse/issues/478
Change-Id: Ie435d093f03a32c02f02b45df880a207c5d7cad1

Change-Id: I8f05010f64edaf1d41fdfca31ee4ae3079cd0ae3

As of commit 4f621f9a, we don't
support older versions

Change-Id: I9f05010f64edaf1d41fdfca31ee4ae3079cd0ae3

Change-Id: Ib094e7e55f922780408223626d88388d2cc5f8b6

Use UtimesNanoAt with AT_SYMLINK_NOFOLLOW. Add
PosixTest/SetattrSymlink to test this.

Change-Id: Ia3a00ad75afd1ba255965025f63fd047662885ec

posix_fallocate(2) returns error directly without setting errno[1]

[1]: https://man.freebsd.org/cgi/man.cgi?query=posix_fallocate&sektion=2

Change-Id: I90ac60ecab6f8814a579cc96d11d4d6660408e88

This test case is going to test whether the file lock works well on the
system. It firstly adds a lock and then get the lock through syscalls.
It uses F_OFD_GETLK command of fcntl(2) to get the lock info on an
opened file description[1], which is not defined on the FreeBSD. This
commit seperate the process of getting the information of lock for
different OSes, and mimic the behaviour of F_OFD_GETLK on FreeBSD.

[1]: https://www.gnu.org/software/libc/manual/html_node/Open-File-Description-Locks.html

Change-Id: I8abae0e7040fc667143eea028731e341be9b1baf

Fails to compile.
Example: https://github.com/hanwen/go-fuse/actions/runs/8909081785/job/24465823082

	Run GOMAXPROCS="1" ./all.bash
	+ go build ./...
	go: downloading golang.org/x/sys v0.0.0-20220520151302-bc2c85ada10a
	Error: fs/api.go:237:2: //go:build comment without // +build comment
	Error: fs/api.go:237:2: //go:build comment without // +build comment
	Error: Process completed with exit code 1.

Change-Id: I2e396906d9cf9fa2248b04f3963b1b2d32b6fd57

Calling Serve() multiple times leads to a panic on unmount and fun
debugging sessions ( https://github.com/hanwen/go-fuse/issues/512 ).
Catch it early.

Change-Id: I3c490448fa3b8407a6db1bd8ad94b8a4e1a93828

Per default, only 1000 FUSE mounts are allowed, then you get
> /usr/bin/fusermount3: too many FUSE filesystems mounted; mount_max=N can be set in /etc/fuse.conf

Let's stay well below 1000 as there may be other (unrelated)
FUSE mounts also occupying slots.

Change-Id: I2e2e094a88236ff803abba3722395c29207c8571

Change-Id: Iffa5314fbe171d182a4bdef42e0ab4781cc1d221

Prefer x/sys/unix over syscall package, which supports FreeBSD.

Change-Id: Ifd3a0fd6d2b90369a6e1e74d92606a655815a6a4

Change-Id: Ib1e1e1054db73df031e4be958206fc94bbea1236

test

Change-Id: Idcb3ee99c1afdde410e41e3a67c29c3143f602c3