Age | Commit message (Collapse) | Author |
|
https://docs.github.com/en/actions/writing-workflows/choosing-when-your-workflow-runs/events-that-trigger-workflows#workflow_dispatch
|
|
Complain loudly when the underlying storage does not support
byte-range locks.
https://github.com/rfjakob/gocryptfs/issues/754
|
|
With -sharedstorage, when we get a decryption error, we lock the
byte range and try again.
This makes concurrent R/W safe agains torn writes.
https://github.com/rfjakob/gocryptfs/issues/754
|
|
Add the -sharedstorage flag to the mounts and make it more
severe by adjusting the I/O and file sizes to arbitrary unaligned
values.
It passes *most of the time*.
If it fails, it's like this, and the reason is that there is no
fcntl locks in the READ path yet.
gocryptfs/tests/cluster$ go test -run TestClusterConcurrentRW -v
=== RUN TestClusterConcurrentRW
doRead 2895433: corrupt block #0: cipher: message authentication failed
doRead 2895433: corrupt block #0: cipher: message authentication failed
doRead 2895433: corrupt block #0: cipher: message authentication failed
cluster_test.go:98: readThread iteration 9165: ReadAt failed: read /var/tmp/gocryptfs-test-parent-1026/3021427391/TestClusterConcurrentRW.366469887.mnt1/foo: input/output error
--- FAIL: TestClusterConcurrentRW (0.40s)
FAIL
exit status 1
FAIL github.com/rfjakob/gocryptfs/v2/tests/cluster 0.406s
|
|
As we must write complete ciphertext blocks (except at EOF), non-overlapping
plaintext writes can overlap in the ciphertext.
And because overlapping writes can turn the data into data soup (see
TestPoCTornWrite) we serialize them using fcntl locking.
|
|
|
|
|
|
Turns out we can get the same error with a simpler flow.
|
|
Scary. But explains why TestConcurrentCreate fails.
gocryptfs/tests/cluster$ go test -run TestPoCTornWrite
--- FAIL: TestPoCTornWrite (0.00s)
poc_test.go:210: iteration 214: inconsistent block: d6d6d6d6d6d6d6d6d1d1d1d1d1d1d1d1d1d1d1d1d1d1d1d1d1d1d1d1d1d1d1d1d1d1d1d1d1d1d1d1d1d1
FAIL
|
|
|
|
With -sharedstorage, we keep the on-disk file header.
Other mounts may have the file ID cached so we cannot mess with it.
This makes TestOpenTruncate pass.
|
|
This fails right now:
1 jakob@brikett:~/go/src/github.com/rfjakob/gocryptfs/tests/cluster$ go test -run TestOpenTruncate -v
=== RUN TestOpenTruncate
cluster_test.go:235: POSIX compliance issue: non-exlusive create failed with err=file exists
doWrite 2898550: corrupt header: ParseHeader: header is all-zero. Header hexdump: 000000000000000000000000000000000000
cluster_test.go:240: iteration 1: WriteAt: write /var/tmp/gocryptfs-test-parent-1026/1896094179/TestOpenTruncate.4202105280.mnt2/foo: input/output error
--- FAIL: TestOpenTruncate (0.10s)
FAIL
exit status 1
FAIL github.com/rfjakob/gocryptfs/v2/tests/cluster 0.099s
1 jakob@brikett:~/go/src/github.com/rfjakob/gocryptfs/tests/cluster$ go test -run TestOpenTruncate -v
=== RUN TestOpenTruncate
cluster_test.go:235: POSIX compliance issue: non-exlusive create failed with err=file exists
doRead 2898565: corrupt block #0: cipher: message authentication failed
ino2898565 fh9: RMW read failed: errno=5
cluster_test.go:240: iteration 8: WriteAt: write /var/tmp/gocryptfs-test-parent-1026/652691834/TestOpenTruncate.281532388.mnt1/foo: input/output error
--- FAIL: TestOpenTruncate (0.09s)
FAIL
exit status 1
FAIL github.com/rfjakob/gocryptfs/v2/tests/cluster 0.095s
|
|
|
|
|
|
This exercises the byte-range locks we just added.
|
|
Multiple hosts creating the same file at the same time could
have overwritten each other's file header, leading to data
corruption.
Fix the race by placing a byte-range lock on the file when
creating the file header.
|
|
finds out what happens if multiple
gocryptfs mounts write to one file concurrently
(usually, nothing good).
This use case is relevant for HPC clusters.
|
|
I maybe should have noted that this is xfstests generic/013.
|
|
|
|
Not having Access() means go-fuse emulates it by looking at Getattr().
This works fine most of the time, but breaks down on sshfs, where
sshfs-benchmark.bash shows this:
gocryptfs/tests$ ./sshfs-benchmark.bash nuetzlich.net
working directory: /tmp/sshfs-benchmark.bash.JQC
sshfs mounted: nuetzlich.net:/tmp -> sshfs.mnt
gocryptfs mounted: sshfs.mnt/sshfs-benchmark.bash.Wrz/gocryptfs.crypt -> gocryptfs.mnt
sshfs-benchmark.bash: sshfs gocryptfs-on-sshfs
git init 3.98 6.80
rsync 7.71 10.84
rm -R 4.30rm: descend into write-protected directory 'gocryptfs.mnt/git1'?
The go-fuse emulation gets it wrong here because sshfs reports
permissions but does not enforce them.
Implement it ourselves properly.
|
|
|
|
Attempt to directly call mount(2) before trying fusermount. This means we
can do without fusermount if running as root.
https://github.com/rfjakob/gocryptfs/issues/697
|
|
|
|
go-fuse now sets this internally.
Regression-tested in TestDirectMount.
|
|
go-fuse now handles setting FsName, including DirectMount,
so use that instead of our own solution.
Regression-tested in TestDirectMount.
|
|
This is in preparation of adding directmount capability.
It also check that FsName is set correctly, which is
in preparation for the next patch.
|
|
Otherwise we fail like this on my Fedora 38 box:
=== RUN TestOverlay
DetectQuirks: tmpfs detected, no extended attributes except acls will work.
root_test.go:379: No user xattrs! overlay mount will likely fail.
15:15:57.957960 Unimplemented opcode OPCODE-51
root_test.go:398: mount: /tmp/gocryptfs-test-parent-0/3652394902/TestOverlay.2374697046.mnt/merged: wrong fs type, bad option, bad superblock on overlay, missing codepage or helper program, or other error.
dmesg(1) may have more information after failed mount system call.
root_test.go:399: exit status 32
--- FAIL: TestOverlay (0.04s)
FAIL
Also fix the messed-up DetectQuirks bit test.
|
|
For the streaming read benchmark, we don't want to benchmark
the page cache.
|
|
Looks like I used StringSliceVar (which splits on comma)
where I should have always used StringArrayVar (which does not).
Bug report contains this example of misbehavoir:
#gocryptfs -extpass 'echo abc,123' -init testdir
Reading password from extpass program "echo abc", arguments: ["123"]
extpass cmd start failed: exec: "echo abc": executable file not found in $PATH
Fixes https://github.com/rfjakob/gocryptfs/issues/730
|
|
|
|
|
|
go get github.com/hanwen/go-fuse/v2
|
|
And add a test for it.
Fixes https://github.com/rfjakob/gocryptfs/issues/724
|
|
BenchmarkGoGCMBlockSize/16-4 5499200 219.7 ns/op 72.83 MB/s
BenchmarkGoGCMBlockSize/32-4 4497284 266.2 ns/op 120.22 MB/s
BenchmarkGoGCMBlockSize/64-4 3296336 363.4 ns/op 176.10 MB/s
BenchmarkGoGCMBlockSize/128-4 4204794 285.5 ns/op 448.36 MB/s
BenchmarkGoGCMBlockSize/256-4 2928472 409.7 ns/op 624.83 MB/s
BenchmarkGoGCMBlockSize/512-4 1825164 658.0 ns/op 778.09 MB/s
BenchmarkGoGCMBlockSize/1024-4 1000000 1151 ns/op 889.98 MB/s
BenchmarkGoGCMBlockSize/2048-4 560275 2135 ns/op 959.47 MB/s
BenchmarkGoGCMBlockSize/4096-4 291906 4099 ns/op 999.28 MB/s
BenchmarkGoGCMBlockSize/8192-4 148916 8033 ns/op 1019.83 MB/s
BenchmarkGoGCMBlockSize/16384-4 75337 15911 ns/op 1029.75 MB/s
BenchmarkGoGCMBlockSize/32768-4 37912 31651 ns/op 1035.30 MB/s
BenchmarkGoGCMBlockSize/65536-4 19000 64287 ns/op 1019.43 MB/s
BenchmarkGoGCMBlockSize/131072-4 9225 127636 ns/op 1026.92 MB/s
BenchmarkGoGCMBlockSize/262144-4 4752 252300 ns/op 1039.02 MB/s
BenchmarkGoGCMBlockSize/524288-4 2377 504612 ns/op 1038.99 MB/s
BenchmarkGoGCMBlockSize/1048576-4 1183 1011637 ns/op 1036.51 MB/s
|
|
Only visible when you run "go test -bench" like this:
$ cd gocryptfs/internal/speed
$ go test -bench .
goos: linux
goarch: amd64
pkg: github.com/rfjakob/gocryptfs/v2/internal/speed
cpu: Intel(R) Core(TM) i5-3470 CPU @ 3.20GHz
BenchmarkStupidGCM-4 202352 5937 ns/op 689.96 MB/s
BenchmarkStupidGCMDecrypt-4 206023 5782 ns/op 708.38 MB/s
BenchmarkGoGCM-4 291878 4098 ns/op 999.45 MB/s
BenchmarkGoGCMBlockSize/1024-4 1000000 1151 ns/op 889.88 MB/s
BenchmarkGoGCMBlockSize/2048-4 561182 2134 ns/op 959.60 MB/s
BenchmarkGoGCMBlockSize/4096-4 292057 4101 ns/op 998.87 MB/s
BenchmarkGoGCMBlockSize/8192-4 149216 8031 ns/op 1020.09 MB/s
BenchmarkGoGCMBlockSize/16384-4 75361 15917 ns/op 1029.34 MB/s
BenchmarkGoGCMBlockSize/32768-4 37916 31649 ns/op 1035.35 MB/s
BenchmarkGoGCMBlockSize/65536-4 19005 63117 ns/op 1038.33 MB/s
BenchmarkGoGCMBlockSize/131072-4 9498 126166 ns/op 1038.89 MB/s
BenchmarkGoGCMBlockSize/262144-4 4755 252149 ns/op 1039.64 MB/s
BenchmarkGoGCMBlockSize/524288-4 2377 504108 ns/op 1040.03 MB/s
BenchmarkGoGCMBlockSize/1048576-4 1188 1008675 ns/op 1039.56 MB/s
BenchmarkGoGCMDecrypt-4 294664 4059 ns/op 1009.02 MB/s
BenchmarkAESSIV-4 46498 25432 ns/op 161.05 MB/s
BenchmarkAESSIVDecrypt-4 46908 25509 ns/op 160.57 MB/s
BenchmarkXchacha-4 244473 4894 ns/op 836.97 MB/s
BenchmarkXchachaDecrypt-4 249710 4798 ns/op 853.75 MB/s
BenchmarkStupidXchacha-4 166988 7101 ns/op 576.79 MB/s
BenchmarkStupidXchachaDecrypt-4 163093 7240 ns/op 565.72 MB/s
BenchmarkStupidChacha-4 184172 6527 ns/op 627.58 MB/s
BenchmarkStupidChachaDecrypt-4 179796 6659 ns/op 615.11 MB/s
PASS
ok github.com/rfjakob/gocryptfs/v2/internal/speed 30.068s
|
|
|
|
Package has been removed from Fedora.
https://github.com/rfjakob/gocryptfs/issues/659
|
|
Commit 6196a5b5 got the logic inverted, hence we never
set the last position markers.
Fixes https://github.com/rfjakob/gocryptfs/issues/712
|
|
It used to be reported as "function not implemented", accompanied
with this log output:
go-fuse: can't convert error type: ParseHeader: header is all-zero. Header hexdump: 000000000000000000000000000000000000
Now we report EIO and log this:
doWrite 1372183: corrupt header: ParseHeader: header is all-zero. Header hexdump: 000000000000000000000000000000000000
|
|
Get rid of this eyesore.
|
|
|
|
Calculated acc. to https://words.filippo.io/the-scrypt-parameters/ ,
and add benchmarks to double-check the numbers. They match.
|
|
added which package on linux is needed to use fido2 stick
|
|
|
|
|
|
On Go 1.17 and older we get this:
Error: ./version.go:67:24: info.Settings undefined (type *debug.BuildInfo has no field or method Settings)
Fix the build error by shedding some nice-to-have features.
|
|
Run "make format" using
go version go1.19.4 linux/amd64
|
|
Simplify and move it into a new file version.go.
|
|
Go1.12 introduced BuildInfo which embeds build information. It does
not embed build date to facilitate reproducable builds by default.
If build information is embedded from build script, use the information
provided by the Go build system.
|
|
The logger should be in the new background session together
with the gocryptfs process.
Before:
$ xfce4-terminal -x gocryptfs a b
$ ps xao pid,ppid,pgid,sid,comm,args
PID PPID PGID SID COMMAND COMMAND
192272 1371 192272 192272 gocryptfs /ssd2/jakob.donotbackup/go/bin/gocryptfs -fg -notifypid=192265 a b
192292 192272 192265 192265 logge <defunct> [logger] <defunct>
After:
$ xfce4-terminal -x gocryptfs a b
$ ps xao pid,ppid,pgid,sid,comm,args
PID PPID PGID SID COMMAND COMMAND
211714 1371 211714 211714 gocryptfs /ssd2/jakob.donotbackup/go/bin/gocryptfs -fg -notifypid=211708 a b
211776 211714 211714 211714 logger logger -t gocryptfs-211714-logger
Fixes https://github.com/rfjakob/gocryptfs/issues/660
|