wgcfg: new package #11

crawshaw · 2020-03-30T22:15:31Z

No description provided.

All atomic access must be aligned to 64 bits, even on 32-bit platforms. Go promises that the start of allocated structs is aligned to 64 bits. So, place the atomically-accessed things first in the struct so that they benefit from that alignment. As a side bonus, it cleanly separates fields that are accessed by atomic ops, and those that should be accessed under mu. Also adds a test that will fail consistently on 32-bit platforms if the struct ever changes again to violate the rules. This is likely not needed because unaligned access crashes reliably, but this will reliably fail even if tests accidentally pass due to lucky alignment. Signed-Off-By: David Anderson <[email protected]>

Signed-off-by: Avery Pennarun <[email protected]>

The sticky socket code stays in the device package for now, as it reaches deeply into the peer list. This is the first step in an effort to split some code out of the very busy device package. Signed-off-by: David Crawshaw <[email protected]>

Signed-off-by: Brad Fitzpatrick <[email protected]>

Based on types and config parser from wireguard-windows. Signed-off-by: David Crawshaw <[email protected]>

crawshaw · 2020-03-30T22:20:58Z

@danderson this is I think entirely self-contained so shouldn't need much review.

Signed-off-by: Tyler Kropp <[email protected]>

Signed-off-by: Brad Fitzpatrick <[email protected]>

zx2c4 · 2020-03-30T22:37:24Z

wgcfg/key.go

+	return key, nil
+}
+
+// SymmetricKey is a chacha20poly1305 key.


Note that this is not the case with wireguard's pre-shared keys, which are 32byte values that are hashed into noise's chainkey.

Pushed a light comment fix: d127a16

zx2c4 · 2020-04-05T03:34:45Z

wgcfg/ip.go

+func (ip IP) Is6() bool { return !ip.Is4() }
+
+// Is4 reports whether ip is an IPv4 address.
+func (ip IP) Is4() bool {


Typically throughout WireGuard, we've used [4]byte for v4 and [16]byte for v6, considering the Go standard library's choice of v6-mapped-v4 to be a mistake. Especially since somebody might legitimately put a ::ffff:1.2.3.4 address through WireGuard and expect it to be v6 and not punned into v4.

Using the newer net/netip from the Go standard library would resolve the v6-mapped-v4 issue. Seems like Go wanted to improve it too.

zx2c4 · 2020-04-05T03:34:56Z

wgcfg/ip.go

+// it's not IPv4.
+func (ip IP) To4() []byte {
+	if ip.Is4() {
+		return ip.Addr[12:16]


Ditto here.

zx2c4 · 2020-04-05T03:35:07Z

wgcfg/ip.go

+// IP is an IPv4 or an IPv6 address.
+//
+// Internally the address is always represented in its IPv6 form.
+// IPv4 addresses use the IPv4-in-IPv6 syntax.


See objection below.

zx2c4 · 2020-04-05T03:36:02Z

wgcfg/ip.go

+}
+
+func (ip IP) MarshalText() ([]byte, error) {
+	return []byte(ip.String()), nil


Not super familiar here -- what's with the function name and signature? Can't we just have String() like usual? Is this for some weird reflection serialization?

zx2c4 · 2020-04-05T03:36:35Z

wgcfg/ip.go

+}
+
+func IPv4(b0, b1, b2, b3 byte) (ip IP) {
+	ip.Addr[10], ip.Addr[11] = 0xff, 0xff // IPv4-in-IPv6 prefix


No. See objection above.

zx2c4 · 2020-04-05T03:37:30Z

wgcfg/ip.go

+	c := int8(r.Mask)
+	i := 0
+	if r.IP.Is4() {
+		i = 12


If you handle v4 properly, this becomes a single register-sized and/mask.

zx2c4 · 2020-04-05T03:38:02Z

wgcfg/ip.go

+	return &net.IPNet{IP: r.IP.IP(), Mask: net.CIDRMask(int(r.Mask), bits)}
+}
+
+func (r CIDR) Contains(ip IP) bool {


Is this actually used anywhere in WireGuard?

Do you plan on using it inside of wireguard-windows or something?

zx2c4 · 2020-04-05T03:39:19Z

wgcfg/key.go

+}
+
+func (a *Key) LessThan(b *Key) bool {
+	for i := range a {


What is the purpose of this? Seems dangerous, given it's not constant time.

Key lookup structures should use hashing rather than binary searches.

zx2c4 · 2020-04-05T03:39:40Z

wgcfg/key.go

+	return subtle.ConstantTimeCompare(zeros[:], k[:]) == 1
+}
+
+func (k *Key) MarshalJSON() ([]byte, error) {


Why json here? wireguard-go doesn't use json.

zx2c4 · 2020-04-05T03:40:11Z

wgcfg/name.go

+	"CON", "PRN", "AUX", "NUL",
+	"COM1", "COM2", "COM3", "COM4", "COM5", "COM6", "COM7", "COM8", "COM9",
+	"LPT1", "LPT2", "LPT3", "LPT4", "LPT5", "LPT6", "LPT7", "LPT8", "LPT9",
+}


Looks like you took this out of wireguard-windows, but does it really make sense here in the generic wireguard-go package?

zx2c4 · 2020-04-05T03:40:48Z

wgcfg/parser.go

+}
+
+func FromWgQuick(s string, name string) (*Config, error) {
+	if !TunnelNameIsValid(name) {


wireguard-go doesn't parse wg-quick for anything. What's this for?

zx2c4 · 2020-04-05T03:41:19Z

wgcfg/parser.go

+
+// TODO(apenwarr): This is incompatibe with current Device.IpcSetOperation.
+//  It duplicates all the parser stuff in there, but is missing some
+//  keywords. Nothing useful seems to need it anymore.


If it's broken and nothing uses it, why is this included? Seems like the TODO can be addressed immediately.

zx2c4 · 2020-04-05T03:44:14Z

@danderson this is I think entirely self-contained so shouldn't need much review.

Config semantics are important and this code seems pretty half-baked (there's a "TODO: this is broken, remove me" still in there). Reopening this.

Maintaining a code base does not mean committing stuff in a "throw it over the fence" manner. It means having a real attention to detail and care, which takes focus before committing.

I didn't want to have to comment on any of this and let you handle everything, but the quality here is questionable.

Please fix this stuff up.

Update script & readme

danderson and others added 6 commits March 30, 2020 20:10

wintun: split error message for create vs open namespace.

6679323

Signed-off-by: Avery Pennarun <[email protected]>

tun: remove unused isUp method

900ae64

Signed-off-by: Brad Fitzpatrick <[email protected]>

tun: fix data race on name field

d5d7075

Signed-off-by: Brad Fitzpatrick <[email protected]>

wgcfg: new config package

bad6ad6

Based on types and config parser from wireguard-windows. Signed-off-by: David Crawshaw <[email protected]>

crawshaw force-pushed the crawshaw/up12 branch from 14b87e5 to 2391f24 Compare March 30, 2020 22:19

danderson approved these changes Mar 30, 2020

View reviewed changes

kroppt and others added 3 commits March 31, 2020 09:32

wgcfg: add fast CIDR.Contains implementation

33aaf34

Signed-off-by: Tyler Kropp <[email protected]>

wgcfg: clean up IP type/method signatures

786d87d

Signed-off-by: Brad Fitzpatrick <[email protected]>

wgcfg: fix bug preventing IPv6 addresses from working

9f6251f

Signed-off-by: Brad Fitzpatrick <[email protected]>

crawshaw force-pushed the crawshaw/up12 branch from 2391f24 to 9f6251f Compare March 30, 2020 22:32

crawshaw closed this Mar 30, 2020

zx2c4 reviewed Mar 30, 2020

View reviewed changes

zx2c4 reviewed Apr 5, 2020

View reviewed changes

wgcfg/ip.go

// it's not IPv4.

func (ip IP) To4() []byte {

if ip.Is4() {

return ip.Addr[12:16]

Copy link

Member

zx2c4 Apr 5, 2020

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Ditto here.

zx2c4 reviewed Apr 5, 2020

View reviewed changes

wgcfg/ip.go

c := int8(r.Mask)

i := 0

if r.IP.Is4() {

i = 12

Copy link

Member

zx2c4 Apr 5, 2020

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

If you handle v4 properly, this becomes a single register-sized and/mask.

zx2c4 reviewed Apr 5, 2020

View reviewed changes

zx2c4 reopened this Apr 5, 2020

zx2c4-bot force-pushed the master branch from 653c9de to 975e9e8 Compare May 2, 2020 07:32

zx2c4-bot force-pushed the master branch 3 times, most recently from 6b5293b to 6005c57 Compare March 9, 2021 04:32

zx2c4-bot force-pushed the master branch 6 times, most recently from 58beb0f to 54dbe24 Compare April 12, 2021 21:35

zx2c4-bot force-pushed the master branch 4 times, most recently from 1ae3898 to 4e9e5da Compare May 10, 2021 15:49

zx2c4-bot force-pushed the master branch from 9252f58 to 841756e Compare June 3, 2021 14:29

zx2c4-bot force-pushed the master branch from 0243978 to 23d4e52 Compare November 6, 2021 13:31

zx2c4-bot force-pushed the master branch 4 times, most recently from eba36c5 to ffb742d Compare November 16, 2021 20:16

zx2c4-bot force-pushed the master branch from ff73da5 to 387f7c4 Compare November 23, 2021 21:03

zx2c4-bot force-pushed the master branch 5 times, most recently from 89a9432 to b9669b7 Compare February 2, 2022 22:09

zx2c4-bot force-pushed the master branch from 3a0dfef to b51010b Compare September 4, 2022 10:57

zx2c4-bot force-pushed the master branch from c7b76d3 to 1e2c3e5 Compare February 16, 2023 15:34

zx2c4-bot force-pushed the master branch from 787da64 to f41f474 Compare March 10, 2023 13:53

zx2c4-bot force-pushed the master branch from d3cb5bd to 6f895be Compare March 24, 2023 16:05

GFW-knocker pushed a commit to GFW-knocker/wireguard that referenced this pull request May 4, 2024

Merge pull request WireGuard#11 from Ptechgithub/master

dec2446

Update script & readme

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

wgcfg: new package #11

wgcfg: new package #11

crawshaw commented Mar 30, 2020

crawshaw commented Mar 30, 2020

zx2c4 Mar 30, 2020

crawshaw Mar 31, 2020

zx2c4 Apr 5, 2020

mrmelon54 Dec 15, 2024

zx2c4 Apr 5, 2020

zx2c4 Apr 5, 2020

zx2c4 Apr 5, 2020

zx2c4 Apr 5, 2020

zx2c4 Apr 5, 2020

zx2c4 Apr 5, 2020

zx2c4 Apr 5, 2020

zx2c4 Apr 5, 2020

zx2c4 Apr 5, 2020

zx2c4 Apr 5, 2020

zx2c4 Apr 5, 2020

zx2c4 commented Apr 5, 2020

wgcfg: new package #11

Are you sure you want to change the base?

wgcfg: new package #11

Conversation

crawshaw commented Mar 30, 2020

crawshaw commented Mar 30, 2020

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

zx2c4 commented Apr 5, 2020