Use hostmatcher to replace matchlist, improve security (#17605)

Use hostmacher to replace matchlist.

And we introduce a better DialContext to do a full host/IP check, otherwise the attackers can still bypass the allow/block list by a 302 redirection.

custom/conf/app.example.ini

@@ -2114,7 +2114,7 @@ PATH =
 ;ALLOWED_DOMAINS =
 ;;
 ;; Blocklist for migrating, default is blank. Multiple domains could be separated by commas.
-;; When ALLOWED_DOMAINS is not blank, this option will be ignored.
+;; When ALLOWED_DOMAINS is not blank, this option has a higher priority to deny domains.
 ;BLOCKED_DOMAINS =
 ;;
 ;; Allow private addresses defined by RFC 1918, RFC 1122, RFC 4632 and RFC 4291 (false by default)

docs/content/doc/advanced/config-cheat-sheet.en-us.md

@@ -1045,7 +1045,7 @@ Task queue configuration has been moved to `queue.task`. However, the below conf
 - `MAX_ATTEMPTS`: **3**: Max attempts per http/https request on migrations.
 - `RETRY_BACKOFF`: **3**: Backoff time per http/https request retry (seconds)
 - `ALLOWED_DOMAINS`: **\<empty\>**: Domains allowlist for migrating repositories, default is blank. It means everything will be allowed. Multiple domains could be separated by commas.
-- `BLOCKED_DOMAINS`: **\<empty\>**: Domains blocklist for migrating repositories, default is blank. Multiple domains could be separated by commas. When `ALLOWED_DOMAINS` is not blank, this option will be ignored.
+- `BLOCKED_DOMAINS`: **\<empty\>**: Domains blocklist for migrating repositories, default is blank. Multiple domains could be separated by commas. When `ALLOWED_DOMAINS` is not blank, this option has a higher priority to deny domains.
 - `ALLOW_LOCALNETWORKS`: **false**: Allow private addresses defined by RFC 1918, RFC 1122, RFC 4632 and RFC 4291
 - `SKIP_TLS_VERIFY`: **false**: Allow skip tls verify
 

docs/content/doc/advanced/config-cheat-sheet.zh-cn.md

@@ -335,7 +335,7 @@ IS_INPUT_FILE = false
 - `MAX_ATTEMPTS`: **3**: 在迁移过程中的 http/https 请求重试次数。
 - `RETRY_BACKOFF`: **3**: 等待下一次重试的时间，单位秒。
 - `ALLOWED_DOMAINS`: **\<empty\>**: 迁移仓库的域名白名单，默认为空，表示允许从任意域名迁移仓库，多个域名用逗号分隔。
-- `BLOCKED_DOMAINS`: **\<empty\>**: 迁移仓库的域名黑名单，默认为空，多个域名用逗号分隔。如果 `ALLOWED_DOMAINS` 不为空，此选项将会被忽略。
+- `BLOCKED_DOMAINS`: **\<empty\>**: 迁移仓库的域名黑名单，默认为空，多个域名用逗号分隔。如果 `ALLOWED_DOMAINS` 不为空，此选项有更高的优先级拒绝这里的域名。
 - `ALLOW_LOCALNETWORKS`: **false**: Allow private addresses defined by RFC 1918
 - `SKIP_TLS_VERIFY`: **false**: 允许忽略 TLS 认证
 

integrations/api_repo_lfs_migrate_test.go

@@ -14,6 +14,7 @@ import (
 	"code.gitea.io/gitea/modules/lfs"
 	"code.gitea.io/gitea/modules/setting"
 	api "code.gitea.io/gitea/modules/structs"
+	"code.gitea.io/gitea/services/migrations"
 
 	"github.com/stretchr/testify/assert"
 )
@@ -25,6 +26,7 @@ func TestAPIRepoLFSMigrateLocal(t *testing.T) {
 	oldAllowLocalNetworks := setting.Migrations.AllowLocalNetworks
 	setting.ImportLocalPaths = true
 	setting.Migrations.AllowLocalNetworks = true
+	assert.NoError(t, migrations.Init())
 
 	user := unittest.AssertExistsAndLoadBean(t, &models.User{ID: 1}).(*models.User)
 	session := loginUser(t, user.Name)
@@ -47,4 +49,5 @@ func TestAPIRepoLFSMigrateLocal(t *testing.T) {
 
 	setting.ImportLocalPaths = oldImportLocalPaths
 	setting.Migrations.AllowLocalNetworks = oldAllowLocalNetworks
+	assert.NoError(t, migrations.Init()) // reset old migration settings
 }

integrations/api_repo_test.go

@@ -331,10 +331,10 @@ func TestAPIRepoMigrate(t *testing.T) {
 			switch respJSON["message"] {
 			case "Remote visit addressed rate limitation.":
 				t.Log("test hit github rate limitation")
-			case "You are not allowed to import from private IPs.":
+			case "You can not import from disallowed hosts.":
 				assert.EqualValues(t, "private-ip", testCase.repoName)
 			default:
-				t.Errorf("unexpected error '%v' on url '%s'", respJSON["message"], testCase.cloneURL)
+				assert.Fail(t, "unexpected error '%v' on url '%s'", respJSON["message"], testCase.cloneURL)
 			}
 		} else {
 			assert.EqualValues(t, testCase.expectedStatus, resp.Code)

integrations/mirror_pull_test.go

@@ -47,7 +47,7 @@ func TestMirrorPull(t *testing.T) {
 
 	ctx := context.Background()
 
-	mirror, err := repository.MigrateRepositoryGitData(ctx, user, mirrorRepo, opts)
+	mirror, err := repository.MigrateRepositoryGitData(ctx, user, mirrorRepo, opts, nil)
 	assert.NoError(t, err)
 
 	gitRepo, err := git.OpenRepository(repoPath)

integrations/mirror_push_test.go

@@ -16,6 +16,7 @@ import (
 	"code.gitea.io/gitea/modules/git"
 	"code.gitea.io/gitea/modules/repository"
 	"code.gitea.io/gitea/modules/setting"
+	"code.gitea.io/gitea/services/migrations"
 	mirror_service "code.gitea.io/gitea/services/mirror"
 
 	"github.com/stretchr/testify/assert"
@@ -29,6 +30,7 @@ func testMirrorPush(t *testing.T, u *url.URL) {
 	defer prepareTestEnv(t)()
 
 	setting.Migrations.AllowLocalNetworks = true
+	assert.NoError(t, migrations.Init())
 
 	user := unittest.AssertExistsAndLoadBean(t, &models.User{ID: 2}).(*models.User)
 	srcRepo := unittest.AssertExistsAndLoadBean(t, &models.Repository{ID: 1}).(*models.Repository)

models/error.go

@@ -797,7 +797,6 @@ type ErrInvalidCloneAddr struct {
 	IsPermissionDenied bool
 	LocalPath          bool
 	NotResolvedIP      bool
-	PrivateNet         string
 }
 
 // IsErrInvalidCloneAddr checks if an error is a ErrInvalidCloneAddr.
@@ -810,9 +809,6 @@ func (err *ErrInvalidCloneAddr) Error() string {
 	if err.NotResolvedIP {
 		return fmt.Sprintf("migration/cloning from '%s' is not allowed: unknown hostname", err.Host)
 	}
-	if len(err.PrivateNet) != 0 {
-		return fmt.Sprintf("migration/cloning from '%s' is not allowed: the host resolve to a private ip address '%s'", err.Host, err.PrivateNet)
-	}
 	if err.IsInvalidPath {
 		return fmt.Sprintf("migration/cloning from '%s' is not allowed: the provided path is invalid", err.Host)
 	}

modules/hostmatcher/hostmatcher.go

@@ -13,15 +13,18 @@ import (
 )
 
 // HostMatchList is used to check if a host or IP is in a list.
-// If you only need to do wildcard matching, consider to use modules/matchlist
 type HostMatchList struct {
-	hosts  []string
+	SettingKeyHint string
+	SettingValue   string
+
+	// builtins networks
+	builtins []string
+	// patterns for host names (with wildcard support)
+	patterns []string
+	// ipNets is the CIDR network list
 	ipNets []*net.IPNet
 }
 
-// MatchBuiltinAll all hosts are matched
-const MatchBuiltinAll = "*"
-
 // MatchBuiltinExternal A valid non-private unicast IP, all hosts on public internet are matched
 const MatchBuiltinExternal = "external"
 
@@ -31,9 +34,13 @@ const MatchBuiltinPrivate = "private"
 // MatchBuiltinLoopback 127.0.0.0/8 for IPv4 and ::1/128 for IPv6, localhost is included.
 const MatchBuiltinLoopback = "loopback"
 
+func isBuiltin(s string) bool {
+	return s == MatchBuiltinExternal || s == MatchBuiltinPrivate || s == MatchBuiltinLoopback
+}
+
 // ParseHostMatchList parses the host list HostMatchList
-func ParseHostMatchList(hostList string) *HostMatchList {
+func ParseHostMatchList(settingKeyHint string, hostList string) *HostMatchList {
-	hl := &HostMatchList{}
+	hl := &HostMatchList{SettingKeyHint: settingKeyHint, SettingValue: hostList}
 	for _, s := range strings.Split(hostList, ",") {
 		s = strings.ToLower(strings.TrimSpace(s))
 		if s == "" {
@@ -42,53 +49,106 @@ func ParseHostMatchList(hostList string) *HostMatchList {
 		_, ipNet, err := net.ParseCIDR(s)
 		if err == nil {
 			hl.ipNets = append(hl.ipNets, ipNet)
+		} else if isBuiltin(s) {
+			hl.builtins = append(hl.builtins, s)
 		} else {
-			hl.hosts = append(hl.hosts, s)
+			hl.patterns = append(hl.patterns, s)
 		}
 	}
 	return hl
 }
 
-// MatchesHostOrIP checks if the host or IP matches an allow/deny(block) list
+// ParseSimpleMatchList parse a simple matchlist (no built-in networks, no CIDR support, only wildcard pattern match)
-func (hl *HostMatchList) MatchesHostOrIP(host string, ip net.IP) bool {
+func ParseSimpleMatchList(settingKeyHint string, matchList string) *HostMatchList {
-	var matched bool
+	hl := &HostMatchList{
-	host = strings.ToLower(host)
+		SettingKeyHint: settingKeyHint,
-	ipStr := ip.String()
+		SettingValue:   matchList,
-loop:
+	}
-	for _, hostInList := range hl.hosts {
+	for _, s := range strings.Split(matchList, ",") {
-		switch hostInList {
+		s = strings.ToLower(strings.TrimSpace(s))
-		case "":
+		if s == "" {
 			continue
-		case MatchBuiltinAll:
+		}
-			matched = true
+		// we keep the same result as old `matchlist`, so no builtin/CIDR support here, we only match wildcard patterns
-			break loop
+		hl.patterns = append(hl.patterns, s)
+	}
+	return hl
+}
+
+// AppendBuiltin appends more builtins to match
+func (hl *HostMatchList) AppendBuiltin(builtin string) {
+	hl.builtins = append(hl.builtins, builtin)
+}
+
+// IsEmpty checks if the checklist is empty
+func (hl *HostMatchList) IsEmpty() bool {
+	return hl == nil || (len(hl.builtins) == 0 && len(hl.patterns) == 0 && len(hl.ipNets) == 0)
+}
+
+func (hl *HostMatchList) checkPattern(host string) bool {
+	host = strings.ToLower(strings.TrimSpace(host))
+	for _, pattern := range hl.patterns {
+		if matched, _ := filepath.Match(pattern, host); matched {
+			return true
+		}
+	}
+	return false
+}
+
+func (hl *HostMatchList) checkIP(ip net.IP) bool {
+	for _, pattern := range hl.patterns {
+		if pattern == "*" {
+			return true
+		}
+	}
+	for _, builtin := range hl.builtins {
+		switch builtin {
 		case MatchBuiltinExternal:
-			if matched = ip.IsGlobalUnicast() && !util.IsIPPrivate(ip); matched {
+			if ip.IsGlobalUnicast() && !util.IsIPPrivate(ip) {
-				break loop
+				return true
 			}
 		case MatchBuiltinPrivate:
-			if matched = util.IsIPPrivate(ip); matched {
+			if util.IsIPPrivate(ip) {
-				break loop
+				return true
 			}
 		case MatchBuiltinLoopback:
-			if matched = ip.IsLoopback(); matched {
+			if ip.IsLoopback() {
-				break loop
+				return true
-			}
-		default:
-			if matched, _ = filepath.Match(hostInList, host); matched {
-				break loop
-			}
-			if matched, _ = filepath.Match(hostInList, ipStr); matched {
-				break loop
 			}
 		}
 	}
-	if !matched {
+	for _, ipNet := range hl.ipNets {
-		for _, ipNet := range hl.ipNets {
+		if ipNet.Contains(ip) {
-			if matched = ipNet.Contains(ip); matched {
+			return true
-				break
-			}
 		}
 	}
-	return matched
+	return false
+}
+
+// MatchHostName checks if the host matches an allow/deny(block) list
+func (hl *HostMatchList) MatchHostName(host string) bool {
+	if hl == nil {
+		return false
+	}
+	if hl.checkPattern(host) {
+		return true
+	}
+	if ip := net.ParseIP(host); ip != nil {
+		return hl.checkIP(ip)
+	}
+	return false
+}
+
+// MatchIPAddr checks if the IP matches an allow/deny(block) list, it's safe to pass `nil` to `ip`
+func (hl *HostMatchList) MatchIPAddr(ip net.IP) bool {
+	if hl == nil {
+		return false
+	}
+	host := ip.String() // nil-safe, we will get "<nil>" if ip is nil
+	return hl.checkPattern(host) || hl.checkIP(ip)
+}
+
+// MatchHostOrIP checks if the host or IP matches an allow/deny(block) list
+func (hl *HostMatchList) MatchHostOrIP(host string, ip net.IP) bool {
+	return hl.MatchHostName(host) || hl.MatchIPAddr(ip)
 }

modules/hostmatcher/hostmatcher_test.go

@@ -20,17 +20,28 @@ func TestHostOrIPMatchesList(t *testing.T) {
 
 	// for IPv6: "::1" is loopback, "fd00::/8" is private
 
-	hl := ParseHostMatchList("private, External, *.myDomain.com, 169.254.1.0/24")
+	hl := ParseHostMatchList("", "private, External, *.myDomain.com, 169.254.1.0/24")
+
+	test := func(cases []tc) {
+		for _, c := range cases {
+			assert.Equalf(t, c.expected, hl.MatchHostOrIP(c.host, c.ip), "case domain=%s, ip=%v, expected=%v", c.host, c.ip, c.expected)
+		}
+	}
+
 	cases := []tc{
 		{"", net.IPv4zero, false},
 		{"", net.IPv6zero, false},
 
 		{"", net.ParseIP("127.0.0.1"), false},
+		{"127.0.0.1", nil, false},
 		{"", net.ParseIP("::1"), false},
 
 		{"", net.ParseIP("10.0.1.1"), true},
+		{"10.0.1.1", nil, true},
 		{"", net.ParseIP("192.168.1.1"), true},
+		{"192.168.1.1", nil, true},
 		{"", net.ParseIP("fd00::1"), true},
+		{"fd00::1", nil, true},
 
 		{"", net.ParseIP("8.8.8.8"), true},
 		{"", net.ParseIP("1001::1"), true},
@@ -39,13 +50,13 @@ func TestHostOrIPMatchesList(t *testing.T) {
 		{"sub.mydomain.com", net.IPv4zero, true},
 
 		{"", net.ParseIP("169.254.1.1"), true},
+		{"169.254.1.1", nil, true},
 		{"", net.ParseIP("169.254.2.2"), false},
+		{"169.254.2.2", nil, false},
 	}
-	for _, c := range cases {
+	test(cases)
-		assert.Equalf(t, c.expected, hl.MatchesHostOrIP(c.host, c.ip), "case %s(%v)", c.host, c.ip)
-	}
 
-	hl = ParseHostMatchList("loopback")
+	hl = ParseHostMatchList("", "loopback")
 	cases = []tc{
 		{"", net.IPv4zero, false},
 		{"", net.ParseIP("127.0.0.1"), true},
@@ -59,11 +70,9 @@ func TestHostOrIPMatchesList(t *testing.T) {
 
 		{"mydomain.com", net.IPv4zero, false},
 	}
-	for _, c := range cases {
+	test(cases)
-		assert.Equalf(t, c.expected, hl.MatchesHostOrIP(c.host, c.ip), "case %s(%v)", c.host, c.ip)
-	}
 
-	hl = ParseHostMatchList("private")
+	hl = ParseHostMatchList("", "private")
 	cases = []tc{
 		{"", net.IPv4zero, false},
 		{"", net.ParseIP("127.0.0.1"), false},
@@ -77,11 +86,9 @@ func TestHostOrIPMatchesList(t *testing.T) {
 
 		{"mydomain.com", net.IPv4zero, false},
 	}
-	for _, c := range cases {
+	test(cases)
-		assert.Equalf(t, c.expected, hl.MatchesHostOrIP(c.host, c.ip), "case %s(%v)", c.host, c.ip)
-	}
 
-	hl = ParseHostMatchList("external")
+	hl = ParseHostMatchList("", "external")
 	cases = []tc{
 		{"", net.IPv4zero, false},
 		{"", net.ParseIP("127.0.0.1"), false},
@@ -95,11 +102,9 @@ func TestHostOrIPMatchesList(t *testing.T) {
 
 		{"mydomain.com", net.IPv4zero, false},
 	}
-	for _, c := range cases {
+	test(cases)
-		assert.Equalf(t, c.expected, hl.MatchesHostOrIP(c.host, c.ip), "case %s(%v)", c.host, c.ip)
-	}
 
-	hl = ParseHostMatchList("*")
+	hl = ParseHostMatchList("", "*")
 	cases = []tc{
 		{"", net.IPv4zero, true},
 		{"", net.ParseIP("127.0.0.1"), true},
@@ -113,7 +118,43 @@ func TestHostOrIPMatchesList(t *testing.T) {
 
 		{"mydomain.com", net.IPv4zero, true},
 	}
-	for _, c := range cases {
+	test(cases)
-		assert.Equalf(t, c.expected, hl.MatchesHostOrIP(c.host, c.ip), "case %s(%v)", c.host, c.ip)
+
+	// built-in network names can be escaped (warping the first char with `[]`) to be used as a real host name
+	// this mechanism is reversed for internal usage only (maybe for some rare cases), it's not supposed to be used by end users
+	// a real user should never use loopback/private/external as their host names
+	hl = ParseHostMatchList("", "loopback, [p]rivate")
+	cases = []tc{
+		{"loopback", nil, false},
+		{"", net.ParseIP("127.0.0.1"), true},
+		{"private", nil, true},
+		{"", net.ParseIP("192.168.1.1"), false},
 	}
+	test(cases)
+
+	hl = ParseSimpleMatchList("", "loopback, *.domain.com")
+	cases = []tc{
+		{"loopback", nil, true},
+		{"", net.ParseIP("127.0.0.1"), false},
+		{"sub.domain.com", nil, true},
+		{"other.com", nil, false},
+		{"", net.ParseIP("1.1.1.1"), false},
+	}
+	test(cases)
+
+	hl = ParseSimpleMatchList("", "external")
+	cases = []tc{
+		{"", net.ParseIP("192.168.1.1"), false},
+		{"", net.ParseIP("1.1.1.1"), false},
+		{"external", nil, true},
+	}
+	test(cases)
+
+	hl = ParseSimpleMatchList("", "")
+	cases = []tc{
+		{"", net.ParseIP("192.168.1.1"), false},
+		{"", net.ParseIP("1.1.1.1"), false},
+		{"external", nil, false},
+	}
+	test(cases)
 }

modules/hostmatcher/http.go

@@ -0,0 +1,58 @@
+// Copyright 2021 The Gitea Authors. All rights reserved.
+// Use of this source code is governed by a MIT-style
+// license that can be found in the LICENSE file.
+
+package hostmatcher
+
+import (
+	"context"
+	"fmt"
+	"net"
+	"syscall"
+	"time"
+)
+
+// NewDialContext returns a DialContext for Transport, the DialContext will do allow/block list check
+func NewDialContext(usage string, allowList *HostMatchList, blockList *HostMatchList) func(ctx context.Context, network, addr string) (net.Conn, error) {
+	// How Go HTTP Client works with redirection:
+	//   transport.RoundTrip URL=http://domain.com, Host=domain.com
+	//   transport.DialContext addrOrHost=domain.com:80
+	//   dialer.Control tcp4:11.22.33.44:80
+	//   transport.RoundTrip URL=http://www.domain.com/, Host=(empty here, in the direction, HTTP client doesn't fill the Host field)
+	//   transport.DialContext addrOrHost=domain.com:80
+	//   dialer.Control tcp4:11.22.33.44:80
+	return func(ctx context.Context, network, addrOrHost string) (net.Conn, error) {
+		dialer := net.Dialer{
+			// default values comes from http.DefaultTransport
+			Timeout:   30 * time.Second,
+			KeepAlive: 30 * time.Second,
+
+			Control: func(network, ipAddr string, c syscall.RawConn) (err error) {
+				var host string
+				if host, _, err = net.SplitHostPort(addrOrHost); err != nil {
+					return err
+				}
+				// in Control func, the addr was already resolved to IP:PORT format, there is no cost to do ResolveTCPAddr here
+				tcpAddr, err := net.ResolveTCPAddr(network, ipAddr)
+				if err != nil {
+					return fmt.Errorf("%s can only call HTTP servers via TCP, deny '%s(%s:%s)', err=%v", usage, host, network, ipAddr, err)
+				}
+
+				var blockedError error
+				if blockList.MatchHostOrIP(host, tcpAddr.IP) {
+					blockedError = fmt.Errorf("%s can not call blocked HTTP servers (check your %s setting), deny '%s(%s)'", usage, blockList.SettingKeyHint, host, ipAddr)
+				}
+
+				// if we have an allow-list, check the allow-list first
+				if !allowList.IsEmpty() {
+					if !allowList.MatchHostOrIP(host, tcpAddr.IP) {
+						return fmt.Errorf("%s can only call allowed HTTP servers (check your %s setting), deny '%s(%s)'", usage, allowList.SettingKeyHint, host, ipAddr)
+					}
+				}
+				// otherwise, we always follow the blocked list
+				return blockedError
+			},
+		}
+		return dialer.DialContext(ctx, network, addrOrHost)
+	}
+}

modules/lfs/client.go

@@ -7,6 +7,7 @@ package lfs
 import (
 	"context"
 	"io"
+	"net/http"
 	"net/url"
 )
 
@@ -24,9 +25,9 @@ type Client interface {
 }
 
 // NewClient creates a LFS client
-func NewClient(endpoint *url.URL, skipTLSVerify bool) Client {
+func NewClient(endpoint *url.URL, httpTransport *http.Transport) Client {
 	if endpoint.Scheme == "file" {
 		return newFilesystemClient(endpoint)
 	}
-	return newHTTPClient(endpoint, skipTLSVerify)
+	return newHTTPClient(endpoint, httpTransport)
 }

modules/lfs/client_test.go

@@ -13,10 +13,10 @@ import (
 
 func TestNewClient(t *testing.T) {
 	u, _ := url.Parse("file:///test")
-	c := NewClient(u, true)
+	c := NewClient(u, nil)
 	assert.IsType(t, &FilesystemClient{}, c)
 
 	u, _ = url.Parse("https://test.com/lfs")
-	c = NewClient(u, true)
+	c = NewClient(u, nil)
 	assert.IsType(t, &HTTPClient{}, c)
 }

modules/lfs/http_client.go

@@ -7,7 +7,6 @@ package lfs
 import (
 	"bytes"
 	"context"
-	"crypto/tls"
 	"errors"
 	"fmt"
 	"net/http"
@@ -34,12 +33,15 @@ func (c *HTTPClient) BatchSize() int {
 	return batchSize
 }
 
-func newHTTPClient(endpoint *url.URL, skipTLSVerify bool) *HTTPClient {
+func newHTTPClient(endpoint *url.URL, httpTransport *http.Transport) *HTTPClient {
+	if httpTransport == nil {
+		httpTransport = &http.Transport{
+			Proxy: proxy.Proxy(),
+		}
+	}
+
 	hc := &http.Client{
-		Transport: &http.Transport{
+		Transport: httpTransport,
-			TLSClientConfig: &tls.Config{InsecureSkipVerify: skipTLSVerify},
-			Proxy:           proxy.Proxy(),
-		},
 	}
 
 	client := &HTTPClient{

modules/matchlist/matchlist.go

@@ -1,46 +0,0 @@
-// Copyright 2019 The Gitea Authors. All rights reserved.
-// Use of this source code is governed by a MIT-style
-// license that can be found in the LICENSE file.
-
-package matchlist
-
-import (
-	"strings"
-
-	"github.com/gobwas/glob"
-)
-
-// Matchlist represents a block or allow list
-type Matchlist struct {
-	ruleGlobs []glob.Glob
-}
-
-// NewMatchlist creates a new block or allow list
-func NewMatchlist(rules ...string) (*Matchlist, error) {
-	for i := range rules {
-		rules[i] = strings.ToLower(rules[i])
-	}
-	list := Matchlist{
-		ruleGlobs: make([]glob.Glob, 0, len(rules)),
-	}
-
-	for _, rule := range rules {
-		rg, err := glob.Compile(rule)
-		if err != nil {
-			return nil, err
-		}
-		list.ruleGlobs = append(list.ruleGlobs, rg)
-	}
-
-	return &list, nil
-}
-
-// Match will matches
-func (b *Matchlist) Match(u string) bool {
-	for _, r := range b.ruleGlobs {
-		if r.Match(u) {
-			return true
-		}
-	}
-	return false
-}

modules/repository/repo.go

@@ -8,7 +8,7 @@ import (
 	"context"
 	"fmt"
 	"io"
-	"net/url"
+	"net/http"
 	"path"
 	"strings"
 	"time"
@@ -46,7 +46,10 @@ func WikiRemoteURL(remote string) string {
 }
 
 // MigrateRepositoryGitData starts migrating git related data after created migrating repository
-func MigrateRepositoryGitData(ctx context.Context, u *models.User, repo *models.Repository, opts migration.MigrateOptions) (*models.Repository, error) {
+func MigrateRepositoryGitData(ctx context.Context, u *models.User,
+	repo *models.Repository, opts migration.MigrateOptions,
+	httpTransport *http.Transport,
+) (*models.Repository, error) {
 	repoPath := models.RepoPath(u.Name, opts.RepoName)
 
 	if u.IsOrganization() {
@@ -141,8 +144,9 @@ func MigrateRepositoryGitData(ctx context.Context, u *models.User, repo *models.
 		}
 
 		if opts.LFS {
-			ep := lfs.DetermineEndpoint(opts.CloneAddr, opts.LFSEndpoint)
+			endpoint := lfs.DetermineEndpoint(opts.CloneAddr, opts.LFSEndpoint)
-			if err = StoreMissingLfsObjectsInRepository(ctx, repo, gitRepo, ep, setting.Migrations.SkipTLSVerify); err != nil {
+			lfsClient := lfs.NewClient(endpoint, httpTransport)
+			if err = StoreMissingLfsObjectsInRepository(ctx, repo, gitRepo, lfsClient); err != nil {
 				log.Error("Failed to store missing LFS objects for repository: %v", err)
 			}
 		}
@@ -336,8 +340,7 @@ func PushUpdateAddTag(repo *models.Repository, gitRepo *git.Repository, tagName
 }
 
 // StoreMissingLfsObjectsInRepository downloads missing LFS objects
-func StoreMissingLfsObjectsInRepository(ctx context.Context, repo *models.Repository, gitRepo *git.Repository, endpoint *url.URL, skipTLSVerify bool) error {
+func StoreMissingLfsObjectsInRepository(ctx context.Context, repo *models.Repository, gitRepo *git.Repository, lfsClient lfs.Client) error {
-	client := lfs.NewClient(endpoint, skipTLSVerify)
 	contentStore := lfs.NewContentStore()
 
 	pointerChan := make(chan lfs.PointerBlob)
@@ -345,7 +348,7 @@ func StoreMissingLfsObjectsInRepository(ctx context.Context, repo *models.Reposi
 	go lfs.SearchPointerBlobs(ctx, gitRepo, pointerChan, errChan)
 
 	downloadObjects := func(pointers []lfs.Pointer) error {
-		err := client.Download(ctx, pointers, func(p lfs.Pointer, content io.ReadCloser, objectError error) error {
+		err := lfsClient.Download(ctx, pointers, func(p lfs.Pointer, content io.ReadCloser, objectError error) error {
 			if objectError != nil {
 				return objectError
 			}
@@ -411,7 +414,7 @@ func StoreMissingLfsObjectsInRepository(ctx context.Context, repo *models.Reposi
 			}
 
 			batch = append(batch, pointerBlob.Pointer)
-			if len(batch) >= client.BatchSize() {
+			if len(batch) >= lfsClient.BatchSize() {
 				if err := downloadObjects(batch); err != nil {
 					return err
 				}

modules/setting/migrations.go

@@ -4,17 +4,13 @@
 
 package setting
 
-import (
-	"strings"
-)
-
 var (
 	// Migrations settings
 	Migrations = struct {
 		MaxAttempts        int
 		RetryBackoff       int
-		AllowedDomains     []string
+		AllowedDomains     string
-		BlockedDomains     []string
+		BlockedDomains     string
 		AllowLocalNetworks bool
 		SkipTLSVerify      bool
 	}{
@@ -28,15 +24,8 @@ func newMigrationsService() {
 	Migrations.MaxAttempts = sec.Key("MAX_ATTEMPTS").MustInt(Migrations.MaxAttempts)
 	Migrations.RetryBackoff = sec.Key("RETRY_BACKOFF").MustInt(Migrations.RetryBackoff)
 
-	Migrations.AllowedDomains = sec.Key("ALLOWED_DOMAINS").Strings(",")
+	Migrations.AllowedDomains = sec.Key("ALLOWED_DOMAINS").MustString("")
-	for i := range Migrations.AllowedDomains {
+	Migrations.BlockedDomains = sec.Key("BLOCKED_DOMAINS").MustString("")
-		Migrations.AllowedDomains[i] = strings.ToLower(Migrations.AllowedDomains[i])
-	}
-	Migrations.BlockedDomains = sec.Key("BLOCKED_DOMAINS").Strings(",")
-	for i := range Migrations.BlockedDomains {
-		Migrations.BlockedDomains[i] = strings.ToLower(Migrations.BlockedDomains[i])
-	}
-
 	Migrations.AllowLocalNetworks = sec.Key("ALLOW_LOCALNETWORKS").MustBool(false)
 	Migrations.SkipTLSVerify = sec.Key("SKIP_TLS_VERIFY").MustBool(false)
 }

modules/setting/webhook.go

@@ -7,7 +7,6 @@ package setting
 import (
 	"net/url"
 
-	"code.gitea.io/gitea/modules/hostmatcher"
 	"code.gitea.io/gitea/modules/log"
 )
 
@@ -17,7 +16,7 @@ var (
 		QueueLength     int
 		DeliverTimeout  int
 		SkipTLSVerify   bool
-		AllowedHostList *hostmatcher.HostMatchList
+		AllowedHostList string
 		Types           []string
 		PagingNum       int
 		ProxyURL        string
@@ -38,7 +37,7 @@ func newWebhookService() {
 	Webhook.QueueLength = sec.Key("QUEUE_LENGTH").MustInt(1000)
 	Webhook.DeliverTimeout = sec.Key("DELIVER_TIMEOUT").MustInt(5)
 	Webhook.SkipTLSVerify = sec.Key("SKIP_TLS_VERIFY").MustBool()
-	Webhook.AllowedHostList = hostmatcher.ParseHostMatchList(sec.Key("ALLOWED_HOST_LIST").MustString(hostmatcher.MatchBuiltinExternal))
+	Webhook.AllowedHostList = sec.Key("ALLOWED_HOST_LIST").MustString("")
 	Webhook.Types = []string{"gitea", "gogs", "slack", "discord", "dingtalk", "telegram", "msteams", "feishu", "matrix", "wechatwork"}
 	Webhook.PagingNum = sec.Key("PAGING_NUM").MustInt(10)
 	Webhook.ProxyURL = sec.Key("PROXY_URL").MustString("")

options/locale/locale_en-US.ini

@@ -899,8 +899,7 @@ migrate.clone_address_desc = The HTTP(S) or Git 'clone' URL of an existing repos
 migrate.github_token_desc = You can put one or more tokens with comma separated here to make migrating faster because of Github API rate limit. WARN: Abusing this feature may violate the service provider's policy and lead to account blocking.
 migrate.clone_local_path = or a local server path
 migrate.permission_denied = You are not allowed to import local repositories.
-migrate.permission_denied_blocked = You are not allowed to import from blocked hosts.
+migrate.permission_denied_blocked = You can not import from disallowed hosts, please ask the admin to check ALLOWED_DOMAINS/ALLOW_LOCALNETWORKS/BLOCKED_DOMAINS settings.
-migrate.permission_denied_private_ip = You are not allowed to import from private IPs.
 migrate.invalid_local_path = "The local path is invalid. It does not exist or is not a directory."
 migrate.invalid_lfs_endpoint = The LFS endpoint is not valid.
 migrate.failed = Migration failed: %v

routers/api/v1/repo/migrate.go

@@ -253,10 +253,8 @@ func handleRemoteAddrError(ctx *context.APIContext, err error) {
 		case addrErr.IsPermissionDenied:
 			if addrErr.LocalPath {
 				ctx.Error(http.StatusUnprocessableEntity, "", "You are not allowed to import local repositories.")
-			} else if len(addrErr.PrivateNet) == 0 {
-				ctx.Error(http.StatusUnprocessableEntity, "", "You are not allowed to import from blocked hosts.")
 			} else {
-				ctx.Error(http.StatusUnprocessableEntity, "", "You are not allowed to import from private IPs.")
+				ctx.Error(http.StatusUnprocessableEntity, "", "You can not import from disallowed hosts.")
 			}
 		case addrErr.IsInvalidPath:
 			ctx.Error(http.StatusUnprocessableEntity, "", "Invalid local path, it does not exist or not a directory.")

routers/web/repo/migrate.go

@@ -128,10 +128,8 @@ func handleMigrateRemoteAddrError(ctx *context.Context, err error, tpl base.TplN
 		case addrErr.IsPermissionDenied:
 			if addrErr.LocalPath {
 				ctx.RenderWithErr(ctx.Tr("repo.migrate.permission_denied"), tpl, form)
-			} else if len(addrErr.PrivateNet) == 0 {
-				ctx.RenderWithErr(ctx.Tr("repo.migrate.permission_denied_blocked"), tpl, form)
 			} else {
-				ctx.RenderWithErr(ctx.Tr("repo.migrate.permission_denied_private_ip"), tpl, form)
+				ctx.RenderWithErr(ctx.Tr("repo.migrate.permission_denied_blocked"), tpl, form)
 			}
 		case addrErr.IsInvalidPath:
 			ctx.RenderWithErr(ctx.Tr("repo.migrate.invalid_local_path"), tpl, form)

routers/web/repo/setting.go

@@ -750,10 +750,8 @@ func handleSettingRemoteAddrError(ctx *context.Context, err error, form *forms.R
 		case addrErr.IsPermissionDenied:
 			if addrErr.LocalPath {
 				ctx.RenderWithErr(ctx.Tr("repo.migrate.permission_denied"), tplSettingsOptions, form)
-			} else if len(addrErr.PrivateNet) == 0 {
-				ctx.RenderWithErr(ctx.Tr("repo.migrate.permission_denied_blocked"), tplSettingsOptions, form)
 			} else {
-				ctx.RenderWithErr(ctx.Tr("repo.migrate.permission_denied_private_ip"), tplSettingsOptions, form)
+				ctx.RenderWithErr(ctx.Tr("repo.migrate.permission_denied_blocked"), tplSettingsOptions, form)
 			}
 		case addrErr.IsInvalidPath:
 			ctx.RenderWithErr(ctx.Tr("repo.migrate.invalid_local_path"), tplSettingsOptions, form)

services/migrations/gitea_downloader.go

@@ -6,7 +6,6 @@ package migrations
 
 import (
 	"context"
-	"crypto/tls"
 	"errors"
 	"fmt"
 	"io"
@@ -18,8 +17,6 @@ import (
 	admin_model "code.gitea.io/gitea/models/admin"
 	"code.gitea.io/gitea/modules/log"
 	base "code.gitea.io/gitea/modules/migration"
-	"code.gitea.io/gitea/modules/proxy"
-	"code.gitea.io/gitea/modules/setting"
 	"code.gitea.io/gitea/modules/structs"
 
 	gitea_sdk "code.gitea.io/sdk/gitea"
@@ -90,12 +87,7 @@ func NewGiteaDownloader(ctx context.Context, baseURL, repoPath, username, passwo
 		gitea_sdk.SetToken(token),
 		gitea_sdk.SetBasicAuth(username, password),
 		gitea_sdk.SetContext(ctx),
-		gitea_sdk.SetHTTPClient(&http.Client{
+		gitea_sdk.SetHTTPClient(NewMigrationHTTPClient()),
-			Transport: &http.Transport{
-				TLSClientConfig: &tls.Config{InsecureSkipVerify: setting.Migrations.SkipTLSVerify},
-				Proxy:           proxy.Proxy(),
-			},
-		}),
 	)
 	if err != nil {
 		log.Error(fmt.Sprintf("Failed to create NewGiteaDownloader for: %s. Error: %v", baseURL, err))
@@ -275,12 +267,7 @@ func (g *GiteaDownloader) convertGiteaRelease(rel *gitea_sdk.Release) *base.Rele
 		Created:         rel.CreatedAt,
 	}
 
-	httpClient := &http.Client{
+	httpClient := NewMigrationHTTPClient()
-		Transport: &http.Transport{
-			TLSClientConfig: &tls.Config{InsecureSkipVerify: setting.Migrations.SkipTLSVerify},
-			Proxy:           proxy.Proxy(),
-		},
-	}
 
 	for _, asset := range rel.Attachments {
 		size := int(asset.Size)

services/migrations/gitea_uploader.go

@@ -125,7 +125,7 @@ func (g *GiteaLocalUploader) CreateRepo(repo *base.Repository, opts base.Migrate
 		Wiki:           opts.Wiki,
 		Releases:       opts.Releases, // if didn't get releases, then sync them from tags
 		MirrorInterval: opts.MirrorInterval,
-	})
+	}, NewMigrationHTTPTransport())
 
 	g.repo = r
 	if err != nil {

services/migrations/github.go

@@ -7,7 +7,6 @@ package migrations
 
 import (
 	"context"
-	"crypto/tls"
 	"fmt"
 	"io"
 	"net/http"
@@ -19,7 +18,6 @@ import (
 	"code.gitea.io/gitea/modules/log"
 	base "code.gitea.io/gitea/modules/migration"
 	"code.gitea.io/gitea/modules/proxy"
-	"code.gitea.io/gitea/modules/setting"
 	"code.gitea.io/gitea/modules/structs"
 	"code.gitea.io/gitea/modules/util"
 
@@ -100,12 +98,7 @@ func NewGithubDownloaderV3(ctx context.Context, baseURL, userName, password, tok
 			)
 			var client = &http.Client{
 				Transport: &oauth2.Transport{
-					Base: &http.Transport{
+					Base:   NewMigrationHTTPTransport(),
-						TLSClientConfig: &tls.Config{InsecureSkipVerify: setting.Migrations.SkipTLSVerify},
-						Proxy: func(req *http.Request) (*url.URL, error) {
-							return proxy.Proxy()(req)
-						},
-					},
 					Source: oauth2.ReuseTokenSource(nil, ts),
 				},
 			}
@@ -113,14 +106,13 @@ func NewGithubDownloaderV3(ctx context.Context, baseURL, userName, password, tok
 			downloader.addClient(client, baseURL)
 		}
 	} else {
+		var transport = NewMigrationHTTPTransport()
+		transport.Proxy = func(req *http.Request) (*url.URL, error) {
+			req.SetBasicAuth(userName, password)
+			return proxy.Proxy()(req)
+		}
 		var client = &http.Client{
-			Transport: &http.Transport{
+			Transport: transport,
-				TLSClientConfig: &tls.Config{InsecureSkipVerify: setting.Migrations.SkipTLSVerify},
-				Proxy: func(req *http.Request) (*url.URL, error) {
-					req.SetBasicAuth(userName, password)
-					return proxy.Proxy()(req)
-				},
-			},
 		}
 		downloader.addClient(client, baseURL)
 	}
@@ -316,12 +308,7 @@ func (g *GithubDownloaderV3) convertGithubRelease(rel *github.RepositoryRelease)
 		r.Published = rel.PublishedAt.Time
 	}
 
-	httpClient := &http.Client{
+	httpClient := NewMigrationHTTPClient()
-		Transport: &http.Transport{
-			TLSClientConfig: &tls.Config{InsecureSkipVerify: setting.Migrations.SkipTLSVerify},
-			Proxy:           proxy.Proxy(),
-		},
-	}
 
 	for _, asset := range rel.Assets {
 		var assetID = *asset.ID // Don't optimize this, for closure we need a local variable

services/migrations/gitlab.go

@@ -6,7 +6,6 @@ package migrations
 
 import (
 	"context"
-	"crypto/tls"
 	"errors"
 	"fmt"
 	"io"
@@ -18,8 +17,6 @@ import (
 
 	"code.gitea.io/gitea/modules/log"
 	base "code.gitea.io/gitea/modules/migration"
-	"code.gitea.io/gitea/modules/proxy"
-	"code.gitea.io/gitea/modules/setting"
 	"code.gitea.io/gitea/modules/structs"
 
 	"github.com/xanzy/go-gitlab"
@@ -77,16 +74,11 @@ type GitlabDownloader struct {
 //   Use either a username/password, personal token entered into the username field, or anonymous/public access
 //   Note: Public access only allows very basic access
 func NewGitlabDownloader(ctx context.Context, baseURL, repoPath, username, password, token string) (*GitlabDownloader, error) {
-	gitlabClient, err := gitlab.NewClient(token, gitlab.WithBaseURL(baseURL), gitlab.WithHTTPClient(&http.Client{
+	gitlabClient, err := gitlab.NewClient(token, gitlab.WithBaseURL(baseURL), gitlab.WithHTTPClient(NewMigrationHTTPClient()))
-		Transport: &http.Transport{
-			TLSClientConfig: &tls.Config{InsecureSkipVerify: setting.Migrations.SkipTLSVerify},
-			Proxy:           proxy.Proxy(),
-		},
-	}))
 	// Only use basic auth if token is blank and password is NOT
 	// Basic auth will fail with empty strings, but empty token will allow anonymous public API usage
 	if token == "" && password != "" {
-		gitlabClient, err = gitlab.NewBasicAuthClient(username, password, gitlab.WithBaseURL(baseURL))
+		gitlabClient, err = gitlab.NewBasicAuthClient(username, password, gitlab.WithBaseURL(baseURL), gitlab.WithHTTPClient(NewMigrationHTTPClient()))
 	}
 
 	if err != nil {
@@ -300,12 +292,7 @@ func (g *GitlabDownloader) convertGitlabRelease(rel *gitlab.Release) *base.Relea
 		PublisherName:   rel.Author.Username,
 	}
 
-	httpClient := &http.Client{
+	httpClient := NewMigrationHTTPClient()
-		Transport: &http.Transport{
-			TLSClientConfig: &tls.Config{InsecureSkipVerify: setting.Migrations.SkipTLSVerify},
-			Proxy:           proxy.Proxy(),
-		},
-	}
 
 	for k, asset := range rel.Assets.Links {
 		r.Assets = append(r.Assets, &base.ReleaseAsset{

services/migrations/gogs.go

@@ -6,7 +6,6 @@ package migrations
 
 import (
 	"context"
-	"crypto/tls"
 	"fmt"
 	"net/http"
 	"net/url"
@@ -16,7 +15,6 @@ import (
 	"code.gitea.io/gitea/modules/log"
 	base "code.gitea.io/gitea/modules/migration"
 	"code.gitea.io/gitea/modules/proxy"
-	"code.gitea.io/gitea/modules/setting"
 	"code.gitea.io/gitea/modules/structs"
 
 	"github.com/gogs/go-gogs-client"
@@ -97,13 +95,12 @@ func NewGogsDownloader(ctx context.Context, baseURL, userName, password, token,
 		client = gogs.NewClient(baseURL, token)
 		downloader.userName = token
 	} else {
-		downloader.transport = &http.Transport{
+		var transport = NewMigrationHTTPTransport()
-			TLSClientConfig: &tls.Config{InsecureSkipVerify: setting.Migrations.SkipTLSVerify},
+		transport.Proxy = func(req *http.Request) (*url.URL, error) {
-			Proxy: func(req *http.Request) (*url.URL, error) {
+			req.SetBasicAuth(userName, password)
-				req.SetBasicAuth(userName, password)
+			return proxy.Proxy()(req)
-				return proxy.Proxy()(req)
-			},
 		}
+		downloader.transport = transport
 
 		client = gogs.NewClient(baseURL, "")
 		client.SetHTTPClient(&http.Client{

services/migrations/http_client.go

@@ -0,0 +1,30 @@
+// Copyright 2021 The Gitea Authors. All rights reserved.
+// Use of this source code is governed by a MIT-style
+// license that can be found in the LICENSE file.
+
+package migrations
+
+import (
+	"crypto/tls"
+	"net/http"
+
+	"code.gitea.io/gitea/modules/hostmatcher"
+	"code.gitea.io/gitea/modules/proxy"
+	"code.gitea.io/gitea/modules/setting"
+)
+
+// NewMigrationHTTPClient returns a HTTP client for migration
+func NewMigrationHTTPClient() *http.Client {
+	return &http.Client{
+		Transport: NewMigrationHTTPTransport(),
+	}
+}
+
+// NewMigrationHTTPTransport returns a HTTP transport for migration
+func NewMigrationHTTPTransport() *http.Transport {
+	return &http.Transport{
+		TLSClientConfig: &tls.Config{InsecureSkipVerify: setting.Migrations.SkipTLSVerify},
+		Proxy:           proxy.Proxy(),
+		DialContext:     hostmatcher.NewDialContext("migration", allowList, blockList),
+	}
+}

services/migrations/migrate.go

@@ -15,8 +15,8 @@ import (
 
 	"code.gitea.io/gitea/models"
 	admin_model "code.gitea.io/gitea/models/admin"
+	"code.gitea.io/gitea/modules/hostmatcher"
 	"code.gitea.io/gitea/modules/log"
-	"code.gitea.io/gitea/modules/matchlist"
 	base "code.gitea.io/gitea/modules/migration"
 	"code.gitea.io/gitea/modules/setting"
 	"code.gitea.io/gitea/modules/util"
@@ -28,8 +28,8 @@ type MigrateOptions = base.MigrateOptions
 var (
 	factories []base.DownloaderFactory
 
-	allowList *matchlist.Matchlist
+	allowList *hostmatcher.HostMatchList
-	blockList *matchlist.Matchlist
+	blockList *hostmatcher.HostMatchList
 )
 
 // RegisterDownloaderFactory registers a downloader factory
@@ -73,30 +73,35 @@ func IsMigrateURLAllowed(remoteURL string, doer *models.User) error {
 		return &models.ErrInvalidCloneAddr{Host: u.Host, IsProtocolInvalid: true, IsPermissionDenied: true, IsURLError: true}
 	}
 
-	host := strings.ToLower(u.Host)
+	hostName, _, err := net.SplitHostPort(u.Host)
-	if len(setting.Migrations.AllowedDomains) > 0 {
+	if err != nil {
-		if !allowList.Match(host) {
+		// u.Host can be "host" or "host:port"
-			return &models.ErrInvalidCloneAddr{Host: u.Host, IsPermissionDenied: true}
+		err = nil //nolint
-		}
+		hostName = u.Host
-	} else {
+	}
-		if blockList.Match(host) {
+	addrList, err := net.LookupIP(hostName)
+	if err != nil {
+		return &models.ErrInvalidCloneAddr{Host: u.Host, NotResolvedIP: true}
+	}
+
+	var ipAllowed bool
+	var ipBlocked bool
+	for _, addr := range addrList {
+		ipAllowed = ipAllowed || allowList.MatchIPAddr(addr)
+		ipBlocked = ipBlocked || blockList.MatchIPAddr(addr)
+	}
+	var blockedError error
+	if blockList.MatchHostName(hostName) || ipBlocked {
+		blockedError = &models.ErrInvalidCloneAddr{Host: u.Host, IsPermissionDenied: true}
+	}
+	// if we have an allow-list, check the allow-list first
+	if !allowList.IsEmpty() {
+		if !allowList.MatchHostName(hostName) && !ipAllowed {
 			return &models.ErrInvalidCloneAddr{Host: u.Host, IsPermissionDenied: true}
 		}
 	}
-
+	// otherwise, we always follow the blocked list
-	if !setting.Migrations.AllowLocalNetworks {
+	return blockedError
-		addrList, err := net.LookupIP(strings.Split(u.Host, ":")[0])
-		if err != nil {
-			return &models.ErrInvalidCloneAddr{Host: u.Host, NotResolvedIP: true}
-		}
-		for _, addr := range addrList {
-			if util.IsIPPrivate(addr) || !addr.IsGlobalUnicast() {
-				return &models.ErrInvalidCloneAddr{Host: u.Host, PrivateNet: addr.String(), IsPermissionDenied: true}
-			}
-		}
-	}
-
-	return nil
 }
 
 // MigrateRepository migrate repository according MigrateOptions
@@ -462,16 +467,18 @@ func migrateRepository(downloader base.Downloader, uploader base.Uploader, opts
 
 // Init migrations service
 func Init() error {
-	var err error
+	// TODO: maybe we can deprecate these legacy ALLOWED_DOMAINS/ALLOW_LOCALNETWORKS/BLOCKED_DOMAINS, use ALLOWED_HOST_LIST/BLOCKED_HOST_LIST instead
-	allowList, err = matchlist.NewMatchlist(setting.Migrations.AllowedDomains...)
-	if err != nil {
-		return fmt.Errorf("init migration allowList domains failed: %v", err)
-	}
 
-	blockList, err = matchlist.NewMatchlist(setting.Migrations.BlockedDomains...)
+	blockList = hostmatcher.ParseSimpleMatchList("migrations.BLOCKED_DOMAINS", setting.Migrations.BlockedDomains)
-	if err != nil {
-		return fmt.Errorf("init migration blockList domains failed: %v", err)
-	}
 
+	allowList = hostmatcher.ParseSimpleMatchList("migrations.ALLOWED_DOMAINS/ALLOW_LOCALNETWORKS", setting.Migrations.AllowedDomains)
+	if allowList.IsEmpty() {
+		// the default policy is that migration module can access external hosts
+		allowList.AppendBuiltin(hostmatcher.MatchBuiltinExternal)
+	}
+	if setting.Migrations.AllowLocalNetworks {
+		allowList.AppendBuiltin(hostmatcher.MatchBuiltinPrivate)
+		allowList.AppendBuiltin(hostmatcher.MatchBuiltinLoopback)
+	}
 	return nil
 }

services/migrations/migrate_test.go

@@ -21,7 +21,8 @@ func TestMigrateWhiteBlocklist(t *testing.T) {
 	adminUser := unittest.AssertExistsAndLoadBean(t, &models.User{Name: "user1"}).(*models.User)
 	nonAdminUser := unittest.AssertExistsAndLoadBean(t, &models.User{Name: "user2"}).(*models.User)
 
-	setting.Migrations.AllowedDomains = []string{"github.com"}
+	setting.Migrations.AllowedDomains = "github.com"
+	setting.Migrations.AllowLocalNetworks = false
 	assert.NoError(t, Init())
 
 	err := IsMigrateURLAllowed("https://gitlab.com/gitlab/gitlab.git", nonAdminUser)
@@ -33,8 +34,8 @@ func TestMigrateWhiteBlocklist(t *testing.T) {
 	err = IsMigrateURLAllowed("https://gITHUb.com/go-gitea/gitea.git", nonAdminUser)
 	assert.NoError(t, err)
 
-	setting.Migrations.AllowedDomains = []string{}
+	setting.Migrations.AllowedDomains = ""
-	setting.Migrations.BlockedDomains = []string{"github.com"}
+	setting.Migrations.BlockedDomains = "github.com"
 	assert.NoError(t, Init())
 
 	err = IsMigrateURLAllowed("https://gitlab.com/gitlab/gitlab.git", nonAdminUser)
@@ -47,6 +48,7 @@ func TestMigrateWhiteBlocklist(t *testing.T) {
 	assert.Error(t, err)
 
 	setting.Migrations.AllowLocalNetworks = true
+	assert.NoError(t, Init())
 	err = IsMigrateURLAllowed("https://10.0.0.1/go-gitea/gitea.git", nonAdminUser)
 	assert.NoError(t, err)
 

services/mirror/mirror_pull.go

@@ -261,8 +261,9 @@ func runSync(ctx context.Context, m *models.Mirror) ([]*mirrorSyncResult, bool)
 
 	if m.LFS && setting.LFS.StartServer {
 		log.Trace("SyncMirrors [repo: %-v]: syncing LFS objects...", m.Repo)
-		ep := lfs.DetermineEndpoint(remoteAddr.String(), m.LFSEndpoint)
+		endpoint := lfs.DetermineEndpoint(remoteAddr.String(), m.LFSEndpoint)
-		if err = repo_module.StoreMissingLfsObjectsInRepository(ctx, m.Repo, gitRepo, ep, false); err != nil {
+		lfsClient := lfs.NewClient(endpoint, nil)
+		if err = repo_module.StoreMissingLfsObjectsInRepository(ctx, m.Repo, gitRepo, lfsClient); err != nil {
 			log.Error("Failed to synchronize LFS objects for repository: %v", err)
 		}
 	}

services/mirror/mirror_push.go

@@ -8,7 +8,6 @@ import (
 	"context"
 	"errors"
 	"io"
-	"net/url"
 	"regexp"
 	"time"
 
@@ -133,8 +132,9 @@ func runPushSync(ctx context.Context, m *models.PushMirror) error {
 			}
 			defer gitRepo.Close()
 
-			ep := lfs.DetermineEndpoint(remoteAddr.String(), "")
+			endpoint := lfs.DetermineEndpoint(remoteAddr.String(), "")
-			if err := pushAllLFSObjects(ctx, gitRepo, ep, false); err != nil {
+			lfsClient := lfs.NewClient(endpoint, nil)
+			if err := pushAllLFSObjects(ctx, gitRepo, lfsClient); err != nil {
 				return util.NewURLSanitizedError(err, remoteAddr, true)
 			}
 		}
@@ -176,8 +176,7 @@ func runPushSync(ctx context.Context, m *models.PushMirror) error {
 	return nil
 }
 
-func pushAllLFSObjects(ctx context.Context, gitRepo *git.Repository, endpoint *url.URL, skipTLSVerify bool) error {
+func pushAllLFSObjects(ctx context.Context, gitRepo *git.Repository, lfsClient lfs.Client) error {
-	client := lfs.NewClient(endpoint, skipTLSVerify)
 	contentStore := lfs.NewContentStore()
 
 	pointerChan := make(chan lfs.PointerBlob)
@@ -185,7 +184,7 @@ func pushAllLFSObjects(ctx context.Context, gitRepo *git.Repository, endpoint *u
 	go lfs.SearchPointerBlobs(ctx, gitRepo, pointerChan, errChan)
 
 	uploadObjects := func(pointers []lfs.Pointer) error {
-		err := client.Upload(ctx, pointers, func(p lfs.Pointer, objectError error) (io.ReadCloser, error) {
+		err := lfsClient.Upload(ctx, pointers, func(p lfs.Pointer, objectError error) (io.ReadCloser, error) {
 			if objectError != nil {
 				return nil, objectError
 			}
@@ -219,7 +218,7 @@ func pushAllLFSObjects(ctx context.Context, gitRepo *git.Repository, endpoint *u
 		}
 
 		batch = append(batch, pointerBlob.Pointer)
-		if len(batch) >= client.BatchSize() {
+		if len(batch) >= lfsClient.BatchSize() {
 			if err := uploadObjects(batch); err != nil {
 				return err
 			}

services/webhook/deliver.go

@@ -13,17 +13,16 @@ import (
 	"encoding/hex"
 	"fmt"
 	"io"
-	"net"
 	"net/http"
 	"net/url"
 	"strconv"
 	"strings"
 	"sync"
-	"syscall"
 	"time"
 
 	webhook_model "code.gitea.io/gitea/models/webhook"
 	"code.gitea.io/gitea/modules/graceful"
+	"code.gitea.io/gitea/modules/hostmatcher"
 	"code.gitea.io/gitea/modules/log"
 	"code.gitea.io/gitea/modules/proxy"
 	"code.gitea.io/gitea/modules/setting"
@@ -31,8 +30,6 @@ import (
 	"github.com/gobwas/glob"
 )
 
-var contextKeyWebhookRequest interface{} = "contextKeyWebhookRequest"
-
 // Deliver deliver hook task
 func Deliver(t *webhook_model.HookTask) error {
 	w, err := webhook_model.GetWebhookByID(t.HookID)
@@ -98,10 +95,10 @@ func Deliver(t *webhook_model.HookTask) error {
 				return err
 			}
 		default:
-			return fmt.Errorf("Invalid http method for webhook: [%d] %v", t.ID, w.HTTPMethod)
+			return fmt.Errorf("invalid http method for webhook: [%d] %v", t.ID, w.HTTPMethod)
 		}
 	default:
-		return fmt.Errorf("Invalid http method for webhook: [%d] %v", t.ID, w.HTTPMethod)
+		return fmt.Errorf("invalid http method for webhook: [%d] %v", t.ID, w.HTTPMethod)
 	}
 
 	var signatureSHA1 string
@@ -172,10 +169,10 @@ func Deliver(t *webhook_model.HookTask) error {
 	}()
 
 	if setting.DisableWebhooks {
-		return fmt.Errorf("Webhook task skipped (webhooks disabled): [%d]", t.ID)
+		return fmt.Errorf("webhook task skipped (webhooks disabled): [%d]", t.ID)
 	}
 
-	resp, err := webhookHTTPClient.Do(req.WithContext(context.WithValue(req.Context(), contextKeyWebhookRequest, req)))
+	resp, err := webhookHTTPClient.Do(req.WithContext(graceful.GetManager().ShutdownContext()))
 	if err != nil {
 		t.ResponseInfo.Body = fmt.Sprintf("Delivery: %v", err)
 		return err
@@ -296,29 +293,18 @@ func webhookProxy() func(req *http.Request) (*url.URL, error) {
 func InitDeliverHooks() {
 	timeout := time.Duration(setting.Webhook.DeliverTimeout) * time.Second
 
+	allowedHostListValue := setting.Webhook.AllowedHostList
+	if allowedHostListValue == "" {
+		allowedHostListValue = hostmatcher.MatchBuiltinExternal
+	}
+	allowedHostMatcher := hostmatcher.ParseHostMatchList("webhook.ALLOWED_HOST_LIST", allowedHostListValue)
+
 	webhookHTTPClient = &http.Client{
 		Timeout: timeout,
 		Transport: &http.Transport{
 			TLSClientConfig: &tls.Config{InsecureSkipVerify: setting.Webhook.SkipTLSVerify},
 			Proxy:           webhookProxy(),
-			DialContext: func(ctx context.Context, network, addrOrHost string) (net.Conn, error) {
+			DialContext:     hostmatcher.NewDialContext("webhook", allowedHostMatcher, nil),
-				dialer := net.Dialer{
-					Timeout: timeout,
-					Control: func(network, ipAddr string, c syscall.RawConn) error {
-						// in Control func, the addr was already resolved to IP:PORT format, there is no cost to do ResolveTCPAddr here
-						tcpAddr, err := net.ResolveTCPAddr(network, ipAddr)
-						req := ctx.Value(contextKeyWebhookRequest).(*http.Request)
-						if err != nil {
-							return fmt.Errorf("webhook can only call HTTP servers via TCP, deny '%s(%s:%s)', err=%v", req.Host, network, ipAddr, err)
-						}
-						if !setting.Webhook.AllowedHostList.MatchesHostOrIP(req.Host, tcpAddr.IP) {
-							return fmt.Errorf("webhook can only call allowed HTTP servers (check your webhook.ALLOWED_HOST_LIST setting), deny '%s(%s)'", req.Host, ipAddr)
-						}
-						return nil
-					},
-				}
-				return dialer.DialContext(ctx, network, addrOrHost)
-			},
 		},
 	}