syncthing/lib/config/folderconfiguration.go

402 lines
15 KiB
Go
Raw Permalink Normal View History

// Copyright (C) 2014 The Syncthing Authors.
//
// This Source Code Form is subject to the terms of the Mozilla Public
// License, v. 2.0. If a copy of the MPL was not distributed with this file,
// You can obtain one at https://mozilla.org/MPL/2.0/.
package config
import (
"bytes"
"crypto/sha256"
"errors"
"fmt"
"path"
"path/filepath"
"sort"
"strings"
"time"
"github.com/shirou/gopsutil/v4/disk"
"github.com/syncthing/syncthing/lib/build"
"github.com/syncthing/syncthing/lib/db"
"github.com/syncthing/syncthing/lib/fs"
"github.com/syncthing/syncthing/lib/protocol"
)
var (
2018-03-18 00:42:31 +00:00
ErrPathNotDirectory = errors.New("folder path not a directory")
ErrPathMissing = errors.New("folder path missing")
ErrMarkerMissing = errors.New("folder marker missing (this indicates potential data loss, search docs/forum to get information about how to proceed)")
)
const (
DefaultMarkerName = ".stfolder"
refactor: use modern Protobuf encoder (#9817) At a high level, this is what I've done and why: - I'm moving the protobuf generation for the `protocol`, `discovery` and `db` packages to the modern alternatives, and using `buf` to generate because it's nice and simple. - After trying various approaches on how to integrate the new types with the existing code, I opted for splitting off our own data model types from the on-the-wire generated types. This means we can have a `FileInfo` type with nicer ergonomics and lots of methods, while the protobuf generated type stays clean and close to the wire protocol. It does mean copying between the two when required, which certainly adds a small amount of inefficiency. If we want to walk this back in the future and use the raw generated type throughout, that's possible, this however makes the refactor smaller (!) as it doesn't change everything about the type for everyone at the same time. - I have simply removed in cold blood a significant number of old database migrations. These depended on previous generations of generated messages of various kinds and were annoying to support in the new fashion. The oldest supported database version now is the one from Syncthing 1.9.0 from Sep 7, 2020. - I changed config structs to be regular manually defined structs. For the sake of discussion, some things I tried that turned out not to work... ### Embedding / wrapping Embedding the protobuf generated structs in our existing types as a data container and keeping our methods and stuff: ``` package protocol type FileInfo struct { *generated.FileInfo } ``` This generates a lot of problems because the internal shape of the generated struct is quite different (different names, different types, more pointers), because initializing it doesn't work like you'd expect (i.e., you end up with an embedded nil pointer and a panic), and because the types of child types don't get wrapped. That is, even if we also have a similar wrapper around a `Vector`, that's not the type you get when accessing `someFileInfo.Version`, you get the `*generated.Vector` that doesn't have methods, etc. ### Aliasing ``` package protocol type FileInfo = generated.FileInfo ``` Doesn't help because you can't attach methods to it, plus all the above. ### Generating the types into the target package like we do now and attaching methods This fails because of the different shape of the generated type (as in the embedding case above) plus the generated struct already has a bunch of methods that we can't necessarily override properly (like `String()` and a bunch of getters). ### Methods to functions I considered just moving all the methods we attach to functions in a specific package, so that for example ``` package protocol func (f FileInfo) Equal(other FileInfo) bool ``` would become ``` package fileinfos func Equal(a, b *generated.FileInfo) bool ``` and this would mostly work, but becomes quite verbose and cumbersome, and somewhat limits discoverability (you can't see what methods are available on the type in auto completions, etc). In the end I did this in some cases, like in the database layer where a lot of things like `func (fv *FileVersion) IsEmpty() bool` becomes `func fvIsEmpty(fv *generated.FileVersion)` because they were anyway just internal methods. Fixes #8247
2024-12-01 15:50:17 +00:00
EncryptionTokenName = "syncthing-encryption_password_token" //nolint: gosec
maxConcurrentWritesDefault = 2
maxConcurrentWritesLimit = 64
)
refactor: use modern Protobuf encoder (#9817) At a high level, this is what I've done and why: - I'm moving the protobuf generation for the `protocol`, `discovery` and `db` packages to the modern alternatives, and using `buf` to generate because it's nice and simple. - After trying various approaches on how to integrate the new types with the existing code, I opted for splitting off our own data model types from the on-the-wire generated types. This means we can have a `FileInfo` type with nicer ergonomics and lots of methods, while the protobuf generated type stays clean and close to the wire protocol. It does mean copying between the two when required, which certainly adds a small amount of inefficiency. If we want to walk this back in the future and use the raw generated type throughout, that's possible, this however makes the refactor smaller (!) as it doesn't change everything about the type for everyone at the same time. - I have simply removed in cold blood a significant number of old database migrations. These depended on previous generations of generated messages of various kinds and were annoying to support in the new fashion. The oldest supported database version now is the one from Syncthing 1.9.0 from Sep 7, 2020. - I changed config structs to be regular manually defined structs. For the sake of discussion, some things I tried that turned out not to work... ### Embedding / wrapping Embedding the protobuf generated structs in our existing types as a data container and keeping our methods and stuff: ``` package protocol type FileInfo struct { *generated.FileInfo } ``` This generates a lot of problems because the internal shape of the generated struct is quite different (different names, different types, more pointers), because initializing it doesn't work like you'd expect (i.e., you end up with an embedded nil pointer and a panic), and because the types of child types don't get wrapped. That is, even if we also have a similar wrapper around a `Vector`, that's not the type you get when accessing `someFileInfo.Version`, you get the `*generated.Vector` that doesn't have methods, etc. ### Aliasing ``` package protocol type FileInfo = generated.FileInfo ``` Doesn't help because you can't attach methods to it, plus all the above. ### Generating the types into the target package like we do now and attaching methods This fails because of the different shape of the generated type (as in the embedding case above) plus the generated struct already has a bunch of methods that we can't necessarily override properly (like `String()` and a bunch of getters). ### Methods to functions I considered just moving all the methods we attach to functions in a specific package, so that for example ``` package protocol func (f FileInfo) Equal(other FileInfo) bool ``` would become ``` package fileinfos func Equal(a, b *generated.FileInfo) bool ``` and this would mostly work, but becomes quite verbose and cumbersome, and somewhat limits discoverability (you can't see what methods are available on the type in auto completions, etc). In the end I did this in some cases, like in the database layer where a lot of things like `func (fv *FileVersion) IsEmpty() bool` becomes `func fvIsEmpty(fv *generated.FileVersion)` because they were anyway just internal methods. Fixes #8247
2024-12-01 15:50:17 +00:00
type FolderDeviceConfiguration struct {
DeviceID protocol.DeviceID `json:"deviceID" xml:"id,attr"`
IntroducedBy protocol.DeviceID `json:"introducedBy" xml:"introducedBy,attr"`
EncryptionPassword string `json:"encryptionPassword" xml:"encryptionPassword"`
}
type FolderConfiguration struct {
ID string `json:"id" xml:"id,attr" nodefault:"true"`
Label string `json:"label" xml:"label,attr" restart:"false"`
FilesystemType FilesystemType `json:"filesystemType" xml:"filesystemType"`
Path string `json:"path" xml:"path,attr" default:"~"`
Type FolderType `json:"type" xml:"type,attr"`
Devices []FolderDeviceConfiguration `json:"devices" xml:"device"`
RescanIntervalS int `json:"rescanIntervalS" xml:"rescanIntervalS,attr" default:"3600"`
FSWatcherEnabled bool `json:"fsWatcherEnabled" xml:"fsWatcherEnabled,attr" default:"true"`
FSWatcherDelayS float64 `json:"fsWatcherDelayS" xml:"fsWatcherDelayS,attr" default:"10"`
FSWatcherTimeoutS float64 `json:"fsWatcherTimeoutS" xml:"fsWatcherTimeoutS,attr"`
IgnorePerms bool `json:"ignorePerms" xml:"ignorePerms,attr"`
AutoNormalize bool `json:"autoNormalize" xml:"autoNormalize,attr" default:"true"`
MinDiskFree Size `json:"minDiskFree" xml:"minDiskFree" default:"1 %"`
Versioning VersioningConfiguration `json:"versioning" xml:"versioning"`
Copiers int `json:"copiers" xml:"copiers"`
PullerMaxPendingKiB int `json:"pullerMaxPendingKiB" xml:"pullerMaxPendingKiB"`
Hashers int `json:"hashers" xml:"hashers"`
Order PullOrder `json:"order" xml:"order"`
IgnoreDelete bool `json:"ignoreDelete" xml:"ignoreDelete"`
ScanProgressIntervalS int `json:"scanProgressIntervalS" xml:"scanProgressIntervalS"`
PullerPauseS int `json:"pullerPauseS" xml:"pullerPauseS"`
MaxConflicts int `json:"maxConflicts" xml:"maxConflicts" default:"10"`
DisableSparseFiles bool `json:"disableSparseFiles" xml:"disableSparseFiles"`
DisableTempIndexes bool `json:"disableTempIndexes" xml:"disableTempIndexes"`
Paused bool `json:"paused" xml:"paused"`
WeakHashThresholdPct int `json:"weakHashThresholdPct" xml:"weakHashThresholdPct"`
MarkerName string `json:"markerName" xml:"markerName"`
CopyOwnershipFromParent bool `json:"copyOwnershipFromParent" xml:"copyOwnershipFromParent"`
RawModTimeWindowS int `json:"modTimeWindowS" xml:"modTimeWindowS"`
MaxConcurrentWrites int `json:"maxConcurrentWrites" xml:"maxConcurrentWrites" default:"2"`
DisableFsync bool `json:"disableFsync" xml:"disableFsync"`
BlockPullOrder BlockPullOrder `json:"blockPullOrder" xml:"blockPullOrder"`
CopyRangeMethod CopyRangeMethod `json:"copyRangeMethod" xml:"copyRangeMethod" default:"standard"`
CaseSensitiveFS bool `json:"caseSensitiveFS" xml:"caseSensitiveFS"`
JunctionsAsDirs bool `json:"junctionsAsDirs" xml:"junctionsAsDirs"`
SyncOwnership bool `json:"syncOwnership" xml:"syncOwnership"`
SendOwnership bool `json:"sendOwnership" xml:"sendOwnership"`
SyncXattrs bool `json:"syncXattrs" xml:"syncXattrs"`
SendXattrs bool `json:"sendXattrs" xml:"sendXattrs"`
XattrFilter XattrFilter `json:"xattrFilter" xml:"xattrFilter"`
// Legacy deprecated
DeprecatedReadOnly bool `json:"-" xml:"ro,attr,omitempty"` // Deprecated: Do not use.
DeprecatedMinDiskFreePct float64 `json:"-" xml:"minDiskFreePct,omitempty"` // Deprecated: Do not use.
DeprecatedPullers int `json:"-" xml:"pullers,omitempty"` // Deprecated: Do not use.
DeprecatedScanOwnership bool `json:"-" xml:"scanOwnership,omitempty"` // Deprecated: Do not use.
}
// Extended attribute filter. This is a list of patterns to match (glob
// style), each with an action (permit or deny). First match is used. If the
// filter is empty, all strings are permitted. If the filter is non-empty,
// the default action becomes deny. To counter this, you can use the "*"
// pattern to match all strings at the end of the filter. There are also
// limits on the size of accepted attributes.
type XattrFilter struct {
Entries []XattrFilterEntry `json:"entries" xml:"entry"`
MaxSingleEntrySize int `json:"maxSingleEntrySize" xml:"maxSingleEntrySize" default:"1024"`
MaxTotalSize int `json:"maxTotalSize" xml:"maxTotalSize" default:"4096"`
}
type XattrFilterEntry struct {
Match string `json:"match" xml:"match,attr"`
Permit bool `json:"permit" xml:"permit,attr"`
}
func (f FolderConfiguration) Copy() FolderConfiguration {
c := f
c.Devices = make([]FolderDeviceConfiguration, len(f.Devices))
copy(c.Devices, f.Devices)
c.Versioning = f.Versioning.Copy()
return c
}
// Filesystem creates a filesystem for the path and options of this folder.
// The fset parameter may be nil, in which case no mtime handling on top of
// the filesystem is provided.
func (f FolderConfiguration) Filesystem(fset *db.FileSet) fs.Filesystem {
// This is intentionally not a pointer method, because things like
// cfg.Folders["default"].Filesystem(nil) should be valid.
opts := make([]fs.Option, 0, 3)
refactor: use modern Protobuf encoder (#9817) At a high level, this is what I've done and why: - I'm moving the protobuf generation for the `protocol`, `discovery` and `db` packages to the modern alternatives, and using `buf` to generate because it's nice and simple. - After trying various approaches on how to integrate the new types with the existing code, I opted for splitting off our own data model types from the on-the-wire generated types. This means we can have a `FileInfo` type with nicer ergonomics and lots of methods, while the protobuf generated type stays clean and close to the wire protocol. It does mean copying between the two when required, which certainly adds a small amount of inefficiency. If we want to walk this back in the future and use the raw generated type throughout, that's possible, this however makes the refactor smaller (!) as it doesn't change everything about the type for everyone at the same time. - I have simply removed in cold blood a significant number of old database migrations. These depended on previous generations of generated messages of various kinds and were annoying to support in the new fashion. The oldest supported database version now is the one from Syncthing 1.9.0 from Sep 7, 2020. - I changed config structs to be regular manually defined structs. For the sake of discussion, some things I tried that turned out not to work... ### Embedding / wrapping Embedding the protobuf generated structs in our existing types as a data container and keeping our methods and stuff: ``` package protocol type FileInfo struct { *generated.FileInfo } ``` This generates a lot of problems because the internal shape of the generated struct is quite different (different names, different types, more pointers), because initializing it doesn't work like you'd expect (i.e., you end up with an embedded nil pointer and a panic), and because the types of child types don't get wrapped. That is, even if we also have a similar wrapper around a `Vector`, that's not the type you get when accessing `someFileInfo.Version`, you get the `*generated.Vector` that doesn't have methods, etc. ### Aliasing ``` package protocol type FileInfo = generated.FileInfo ``` Doesn't help because you can't attach methods to it, plus all the above. ### Generating the types into the target package like we do now and attaching methods This fails because of the different shape of the generated type (as in the embedding case above) plus the generated struct already has a bunch of methods that we can't necessarily override properly (like `String()` and a bunch of getters). ### Methods to functions I considered just moving all the methods we attach to functions in a specific package, so that for example ``` package protocol func (f FileInfo) Equal(other FileInfo) bool ``` would become ``` package fileinfos func Equal(a, b *generated.FileInfo) bool ``` and this would mostly work, but becomes quite verbose and cumbersome, and somewhat limits discoverability (you can't see what methods are available on the type in auto completions, etc). In the end I did this in some cases, like in the database layer where a lot of things like `func (fv *FileVersion) IsEmpty() bool` becomes `func fvIsEmpty(fv *generated.FileVersion)` because they were anyway just internal methods. Fixes #8247
2024-12-01 15:50:17 +00:00
if f.FilesystemType == FilesystemTypeBasic && f.JunctionsAsDirs {
opts = append(opts, new(fs.OptionJunctionsAsDirs))
}
if !f.CaseSensitiveFS {
opts = append(opts, new(fs.OptionDetectCaseConflicts))
}
if fset != nil {
opts = append(opts, fset.MtimeOption())
}
refactor: use modern Protobuf encoder (#9817) At a high level, this is what I've done and why: - I'm moving the protobuf generation for the `protocol`, `discovery` and `db` packages to the modern alternatives, and using `buf` to generate because it's nice and simple. - After trying various approaches on how to integrate the new types with the existing code, I opted for splitting off our own data model types from the on-the-wire generated types. This means we can have a `FileInfo` type with nicer ergonomics and lots of methods, while the protobuf generated type stays clean and close to the wire protocol. It does mean copying between the two when required, which certainly adds a small amount of inefficiency. If we want to walk this back in the future and use the raw generated type throughout, that's possible, this however makes the refactor smaller (!) as it doesn't change everything about the type for everyone at the same time. - I have simply removed in cold blood a significant number of old database migrations. These depended on previous generations of generated messages of various kinds and were annoying to support in the new fashion. The oldest supported database version now is the one from Syncthing 1.9.0 from Sep 7, 2020. - I changed config structs to be regular manually defined structs. For the sake of discussion, some things I tried that turned out not to work... ### Embedding / wrapping Embedding the protobuf generated structs in our existing types as a data container and keeping our methods and stuff: ``` package protocol type FileInfo struct { *generated.FileInfo } ``` This generates a lot of problems because the internal shape of the generated struct is quite different (different names, different types, more pointers), because initializing it doesn't work like you'd expect (i.e., you end up with an embedded nil pointer and a panic), and because the types of child types don't get wrapped. That is, even if we also have a similar wrapper around a `Vector`, that's not the type you get when accessing `someFileInfo.Version`, you get the `*generated.Vector` that doesn't have methods, etc. ### Aliasing ``` package protocol type FileInfo = generated.FileInfo ``` Doesn't help because you can't attach methods to it, plus all the above. ### Generating the types into the target package like we do now and attaching methods This fails because of the different shape of the generated type (as in the embedding case above) plus the generated struct already has a bunch of methods that we can't necessarily override properly (like `String()` and a bunch of getters). ### Methods to functions I considered just moving all the methods we attach to functions in a specific package, so that for example ``` package protocol func (f FileInfo) Equal(other FileInfo) bool ``` would become ``` package fileinfos func Equal(a, b *generated.FileInfo) bool ``` and this would mostly work, but becomes quite verbose and cumbersome, and somewhat limits discoverability (you can't see what methods are available on the type in auto completions, etc). In the end I did this in some cases, like in the database layer where a lot of things like `func (fv *FileVersion) IsEmpty() bool` becomes `func fvIsEmpty(fv *generated.FileVersion)` because they were anyway just internal methods. Fixes #8247
2024-12-01 15:50:17 +00:00
return fs.NewFilesystem(f.FilesystemType.ToFS(), f.Path, opts...)
}
func (f FolderConfiguration) ModTimeWindow() time.Duration {
dur := time.Duration(f.RawModTimeWindowS) * time.Second
if f.RawModTimeWindowS < 1 && build.IsAndroid {
if usage, err := disk.Usage(f.Filesystem(nil).URI()); err != nil {
dur = 2 * time.Second
l.Debugf(`Detecting FS at "%v" on android: Setting mtime window to 2s: err == "%v"`, f.Path, err)
} else if strings.HasPrefix(strings.ToLower(usage.Fstype), "ext2") || strings.HasPrefix(strings.ToLower(usage.Fstype), "ext3") || strings.HasPrefix(strings.ToLower(usage.Fstype), "ext4") {
l.Debugf(`Detecting FS at %v on android: Leaving mtime window at 0: usage.Fstype == "%v"`, f.Path, usage.Fstype)
} else {
dur = 2 * time.Second
l.Debugf(`Detecting FS at "%v" on android: Setting mtime window to 2s: usage.Fstype == "%v"`, f.Path, usage.Fstype)
}
}
return dur
}
func (f *FolderConfiguration) CreateMarker() error {
2018-03-18 00:42:31 +00:00
if err := f.CheckPath(); err != ErrMarkerMissing {
return err
}
if f.MarkerName != DefaultMarkerName {
// Folder uses a non-default marker so we shouldn't mess with it.
// Pretend we created it and let the subsequent health checks sort
// out the actual situation.
return nil
}
ffs := f.Filesystem(nil)
// Create the marker as a directory
err := ffs.Mkdir(DefaultMarkerName, 0o755)
if err != nil {
return err
}
// Create a file inside it, reducing the risk of the marker directory
// being removed by automated cleanup tools.
markerFile := filepath.Join(DefaultMarkerName, f.markerFilename())
if err := fs.WriteFile(ffs, markerFile, f.markerContents(), 0o644); err != nil {
return err
}
// Sync & hide the containing directory
if dir, err := ffs.Open("."); err != nil {
l.Debugln("folder marker: open . failed:", err)
} else if err := dir.Sync(); err != nil {
l.Debugln("folder marker: fsync . failed:", err)
}
ffs.Hide(DefaultMarkerName)
return nil
}
func (f *FolderConfiguration) RemoveMarker() error {
ffs := f.Filesystem(nil)
_ = ffs.Remove(filepath.Join(DefaultMarkerName, f.markerFilename()))
return ffs.Remove(DefaultMarkerName)
}
func (f *FolderConfiguration) markerFilename() string {
h := sha256.Sum256([]byte(f.ID))
return fmt.Sprintf("syncthing-folder-%x.txt", h[:3])
}
func (f *FolderConfiguration) markerContents() []byte {
var buf bytes.Buffer
buf.WriteString("# This directory is a Syncthing folder marker.\n# Do not delete.\n\n")
fmt.Fprintf(&buf, "folderID: %s\n", f.ID)
fmt.Fprintf(&buf, "created: %s\n", time.Now().Format(time.RFC3339))
return buf.Bytes()
}
// CheckPath returns nil if the folder root exists and contains the marker file
func (f *FolderConfiguration) CheckPath() error {
return f.checkFilesystemPath(f.Filesystem(nil), ".")
}
func (f *FolderConfiguration) checkFilesystemPath(ffs fs.Filesystem, path string) error {
fi, err := ffs.Stat(path)
if err != nil {
if !fs.IsNotExist(err) {
return err
}
2018-03-18 00:42:31 +00:00
return ErrPathMissing
}
// Users might have the root directory as a symlink or reparse point.
// Furthermore, OneDrive bullcrap uses a magic reparse point to the cloudz...
// Yet it's impossible for this to happen, as filesystem adds a trailing
// path separator to the root, so even if you point the filesystem at a file
// Stat ends up calling stat on C:\dir\file\ which, fails with "is not a directory"
// in the error check above, and we don't even get to here.
if !fi.IsDir() && !fi.IsSymlink() {
2018-03-18 00:42:31 +00:00
return ErrPathNotDirectory
}
_, err = ffs.Stat(filepath.Join(path, f.MarkerName))
if err != nil {
if !fs.IsNotExist(err) {
return err
}
2018-03-18 00:42:31 +00:00
return ErrMarkerMissing
}
return nil
}
func (f *FolderConfiguration) CreateRoot() (err error) {
// Directory permission bits. Will be filtered down to something
// sane by umask on Unixes.
permBits := fs.FileMode(0o777)
if build.IsWindows {
// Windows has no umask so we must chose a safer set of bits to
// begin with.
permBits = 0o700
}
filesystem := f.Filesystem(nil)
if _, err = filesystem.Stat("."); fs.IsNotExist(err) {
err = filesystem.MkdirAll(".", permBits)
}
return err
}
func (f FolderConfiguration) Description() string {
if f.Label == "" {
return f.ID
}
return fmt.Sprintf("%q (%s)", f.Label, f.ID)
}
func (f *FolderConfiguration) DeviceIDs() []protocol.DeviceID {
deviceIDs := make([]protocol.DeviceID, len(f.Devices))
for i, n := range f.Devices {
deviceIDs[i] = n.DeviceID
}
return deviceIDs
}
func (f *FolderConfiguration) prepare(myID protocol.DeviceID, existingDevices map[protocol.DeviceID]*DeviceConfiguration) {
// Ensure that
// - any loose devices are not present in the wrong places
// - there are no duplicate devices
// - we are part of the devices
// - folder is not shared in trusted mode with an untrusted device
f.Devices = ensureExistingDevices(f.Devices, existingDevices)
f.Devices = ensureNoDuplicateFolderDevices(f.Devices)
f.Devices = ensureDevicePresent(f.Devices, myID)
f.Devices = ensureNoUntrustedTrustingSharing(f, f.Devices, existingDevices)
sort.Slice(f.Devices, func(a, b int) bool {
return f.Devices[a].DeviceID.Compare(f.Devices[b].DeviceID) == -1
})
if f.RescanIntervalS > MaxRescanIntervalS {
f.RescanIntervalS = MaxRescanIntervalS
} else if f.RescanIntervalS < 0 {
f.RescanIntervalS = 0
}
if f.FSWatcherDelayS <= 0 {
f.FSWatcherEnabled = false
f.FSWatcherDelayS = 10
} else if f.FSWatcherDelayS < 0.01 {
f.FSWatcherDelayS = 0.01
}
if f.Versioning.CleanupIntervalS > MaxRescanIntervalS {
f.Versioning.CleanupIntervalS = MaxRescanIntervalS
} else if f.Versioning.CleanupIntervalS < 0 {
f.Versioning.CleanupIntervalS = 0
}
if f.WeakHashThresholdPct == 0 {
f.WeakHashThresholdPct = 25
}
if f.MarkerName == "" {
f.MarkerName = DefaultMarkerName
}
if f.MaxConcurrentWrites <= 0 {
f.MaxConcurrentWrites = maxConcurrentWritesDefault
} else if f.MaxConcurrentWrites > maxConcurrentWritesLimit {
f.MaxConcurrentWrites = maxConcurrentWritesLimit
}
if f.Type == FolderTypeReceiveEncrypted {
f.DisableTempIndexes = true
f.IgnorePerms = true
}
}
// RequiresRestartOnly returns a copy with only the attributes that require
// restart on change.
func (f FolderConfiguration) RequiresRestartOnly() FolderConfiguration {
copy := f
// Manual handling for things that are not taken care of by the tag
// copier, yet should not cause a restart.
blank := FolderConfiguration{}
copyMatchingTag(&blank, &copy, "restart", func(v string) bool {
if len(v) > 0 && v != "false" {
panic(fmt.Sprintf(`unexpected tag value: %s. expected untagged or "false"`, v))
}
return v == "false"
})
return copy
}
func (f *FolderConfiguration) Device(device protocol.DeviceID) (FolderDeviceConfiguration, bool) {
for _, dev := range f.Devices {
if dev.DeviceID == device {
return dev, true
}
}
return FolderDeviceConfiguration{}, false
}
func (f *FolderConfiguration) SharedWith(device protocol.DeviceID) bool {
_, ok := f.Device(device)
return ok
}
func (f *FolderConfiguration) CheckAvailableSpace(req uint64) error {
val := f.MinDiskFree.BaseValue()
if val <= 0 {
return nil
}
fs := f.Filesystem(nil)
usage, err := fs.Usage(".")
if err != nil {
refactor: use modern Protobuf encoder (#9817) At a high level, this is what I've done and why: - I'm moving the protobuf generation for the `protocol`, `discovery` and `db` packages to the modern alternatives, and using `buf` to generate because it's nice and simple. - After trying various approaches on how to integrate the new types with the existing code, I opted for splitting off our own data model types from the on-the-wire generated types. This means we can have a `FileInfo` type with nicer ergonomics and lots of methods, while the protobuf generated type stays clean and close to the wire protocol. It does mean copying between the two when required, which certainly adds a small amount of inefficiency. If we want to walk this back in the future and use the raw generated type throughout, that's possible, this however makes the refactor smaller (!) as it doesn't change everything about the type for everyone at the same time. - I have simply removed in cold blood a significant number of old database migrations. These depended on previous generations of generated messages of various kinds and were annoying to support in the new fashion. The oldest supported database version now is the one from Syncthing 1.9.0 from Sep 7, 2020. - I changed config structs to be regular manually defined structs. For the sake of discussion, some things I tried that turned out not to work... ### Embedding / wrapping Embedding the protobuf generated structs in our existing types as a data container and keeping our methods and stuff: ``` package protocol type FileInfo struct { *generated.FileInfo } ``` This generates a lot of problems because the internal shape of the generated struct is quite different (different names, different types, more pointers), because initializing it doesn't work like you'd expect (i.e., you end up with an embedded nil pointer and a panic), and because the types of child types don't get wrapped. That is, even if we also have a similar wrapper around a `Vector`, that's not the type you get when accessing `someFileInfo.Version`, you get the `*generated.Vector` that doesn't have methods, etc. ### Aliasing ``` package protocol type FileInfo = generated.FileInfo ``` Doesn't help because you can't attach methods to it, plus all the above. ### Generating the types into the target package like we do now and attaching methods This fails because of the different shape of the generated type (as in the embedding case above) plus the generated struct already has a bunch of methods that we can't necessarily override properly (like `String()` and a bunch of getters). ### Methods to functions I considered just moving all the methods we attach to functions in a specific package, so that for example ``` package protocol func (f FileInfo) Equal(other FileInfo) bool ``` would become ``` package fileinfos func Equal(a, b *generated.FileInfo) bool ``` and this would mostly work, but becomes quite verbose and cumbersome, and somewhat limits discoverability (you can't see what methods are available on the type in auto completions, etc). In the end I did this in some cases, like in the database layer where a lot of things like `func (fv *FileVersion) IsEmpty() bool` becomes `func fvIsEmpty(fv *generated.FileVersion)` because they were anyway just internal methods. Fixes #8247
2024-12-01 15:50:17 +00:00
return nil //nolint: nilerr
}
if err := checkAvailableSpace(req, f.MinDiskFree, usage); err != nil {
return fmt.Errorf("insufficient space in folder %v (%v): %w", f.Description(), fs.URI(), err)
}
return nil
}
func (f XattrFilter) Permit(s string) bool {
if len(f.Entries) == 0 {
return true
}
for _, entry := range f.Entries {
if ok, _ := path.Match(entry.Match, s); ok {
return entry.Permit
}
}
return false
}
func (f XattrFilter) GetMaxSingleEntrySize() int {
return f.MaxSingleEntrySize
}
func (f XattrFilter) GetMaxTotalSize() int {
return f.MaxTotalSize
}