Change the Prefix field of the Route Information Option to be variable-length.

[y-kzm]

This PR enables prefixes to be sent in smaller chunks within the Route Information Options.

According to RFC 4191, Section 2.3:
The Length field is 1, 2, or 3 depending on the Prefix Length. If Prefix Length is greater than 64, then Length must be 3. If Prefix Length is greater than 0, then Length must be 2 or 3. If Prefix Length is zero, then Length must be 1, 2, or 3.

• Before change

          route info option (24), length 24 (3):  2001:db8:1::1/128, pref=medium, lifetime=300s
            0x0000:  8000 0000 012c 2001 0db8 0001 0000 0000
            0x0010:  0000 0000 0001
          route info option (24), length 24 (3):  2001:db8:2::/64, pref=medium, lifetime=300s
            0x0000:  4000 0000 012c 2001 0db8 0002 0000 0000
            0x0010:  0000 0000 0000

• After change

          route info option (24), length 24 (3):  2001:db8:1::1/128, pref=medium, lifetime=300s
            0x0000:  8000 0000 012c 2001 0db8 0001 0000 0000
            0x0010:  0000 0000 0001
          route info option (24), length 16 (2):  2001:db8:2::/64, pref=medium, lifetime=300s
            0x0000:  4000 0000 012c 2001 0db8 0002 0000

[robbat2]

silently dropping the route here seems weird. At the very least it should drop; but also it should be IMPOSSIBLE to reach this if we validate routes earlier.

[robbat2]

@y-kzm bump

[Neustradamus]

@y-kzm: Have you seen @robbat2 review?

[y-kzm]

The test failed because the expected data for the prefix field of the RIO had a fixed length.
Before (expected data): test/send.c:164

	unsigned char expected[] = {
		0x18, 0x03, 0x30, 0x18, 0x00, 0x00, 0x27, 0x10, 0xfe, 0x80, 0x00, 0x0f, 0x00, 0x01, 0x00, 0x00, 0x00, 0x00,
		0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x18, 0x03, 0x28, 0x08, 0xff, 0xff, 0xff, 0xff, 0xfe, 0x80, 0x00, 0x0f,
		0x00, 0x02, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x18, 0x03, 0x20, 0x00, 0x00, 0x00,
		0x0b, 0xb8, 0xfe, 0x80, 0x00, 0x0f, 0x00, 0x02, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
	};

type: 0x18 (24)
length: 0x03 (3*8=24byte)
prefix length: 0x30 (/48)
Resvd|Prf|Resvd : 0x18 (0b000 11 000) low=0b11
Route Lifetime: 0x00002710 (=10000)
Prefix : 0xfe, 0x80, 0x00, 0x0f, 0x00, 0x01, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00 (fe80:000f:0001:0000:0000:0000:0000:0000) // fixed
…

After (expected data): test/send.c:164

	unsigned char expected[] = {
		0x18, 0x02, 0x30, 0x18, 0x00, 0x00, 0x27, 0x10, 0xfe, 0x80, 0x00, 0x0f, 0x00, 0x01, 0x00, 0x00,
		0x18, 0x02, 0x28, 0x08, 0xff, 0xff, 0xff, 0xff, 0xfe, 0x80, 0x00, 0x0f, 0x00, 0x02, 0x00, 0x00,
		0x18, 0x02, 0x20, 0x00, 0x00, 0x00, 0x0b, 0xb8, 0xfe, 0x80, 0x00, 0x0f, 0x00, 0x02, 0x00, 0x00,
		0x18, 0x03, 0x80, 0x00, 0x00, 0x00, 0x0b, 0xb8, 0xfe, 0x80, 0x00, 0x0f, 0x00, 0x02, 0x00, 0x00,
		0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
	};

type: 0x18 (24)
length: 0x02 (2*8=16byte)
prefix length: 0x30 (/48)
Resvd|Prf|Resvd : 0x18 (0b000 11 000) low=0b11
Route Lifetime: 0x00002710 (=10000)
Prefix : 0xfe, 0x80, 0x00, 0x0f, 0x00, 0x01, 0x00, 0x00 (fe80:000f:0001:0000:) // variable
...

With this change, I have added a test case for when the prefix length is /128 in test/test1.conf.
Please review and approve the workflow again.
Thank you.

[stappersg]

On Mon, Nov 25, 2024 at 02:32:41AM -0800, yykzm wrote: With this change, I have added a test case for when the prefix length is /128 in `test/test1.conf`. Please review and approve the workflow again.

After approval of the workflow, there was: All checks have passed, 40 successful checks

Thank you.

Thank you for your contribution to radvd. Groeten Geert Stappers -- Silence is hard to parse

[stappersg]

Oh, #193 (comment) reminded me
on add Sign-off-by trailer

[y-kzm]

I added Signed-off-by trailer to the relevant commits.