RFR: 8315767: InetAddress: constructing objects from BSD literal addresses [v2]

[Florent Guillaume]

On Wed, 10 Apr 2024 15:34:22 GMT, Sergey Chernyshev <schernyshev at openjdk.org> wrote:

>> There are two distinct approaches to parsing IPv4 literal addresses. One is the Java baseline "strict" syntax (all-decimal d.d.d.d form family), another one is the "loose" syntax of RFC 6943 section 3.1.1 [1] (POSIX `inet_addr` allowing octal and hexadecimal forms [2]). The goal of this PR is to provide interface to construct InetAddress objects from literal addresses in POSIX form, to applications that need to mimic the behavior of `inet_addr` used by standard network utilities such as netcat/curl/wget and the majority of web browsers. At present time, there's no way to construct `InetAddress` object from such literal addresses because the existing APIs such as `InetAddress.getByName()`, `InetAddress#ofLiteral()` and `Inet4Address#ofLiteral()` will consume an address and successfully parse it as decimal, regardless of the octal prefix. Hence, the resulting object will point to a different IP address.
>> 
>> Historically `InetAddress.getByName()/.getAllByName()` were the only way to convert a literal address into an InetAddress object. `getAllByName()` API relies on POSIX `getaddrinfo` / `inet_addr` which parses IP address segments with `strtoul` (accepts octal and hexadecimal bases).
>> 
>> The fallback to `getaddrinfo` is undesirable as it may end up with network queries (blocking mode), if `inet_addr` rejects the input literal address. The Java standard explicitly says that
>> 
>> "If a literal IP address is supplied, only the validity of the address format is checked."
>> 
>> @AlekseiEfimov contributed JDK-8272215 [3] that adds new factory methods `.ofLiteral()` to `InetAddress` classes. Although the new API is not affected by the `getaddrinfo` fallback issue, it is not sufficient for an application that needs to interoperate with external tooling that follows POSIX standard. In the current state, `InetAddress#ofLiteral()` and `Inet4Address#ofLiteral()` will consume the input literal address and (regardless of the octal prefix) parse it as decimal numbers. Hence, it's not possible to reliably construct an `InetAddress` object from a literal address in POSIX form that would point to the desired host.
>> 
>> It is proposed to extend the factory methods with `Inet4Address#ofPosixLiteral()` that allows parsing literal IP(v4) addresses in "loose" syntax, compatible with `inet_addr` POSIX api. The implementation is based on `.isBsdParsableV4()` method added along with JDK-8277608 [4]. The changes in the original algorithm are as follows:
>> 
>> - `IPAddressUtil#parseB...
>
> Sergey Chernyshev has updated the pull request incrementally with one additional commit since the last revision:
> 
>   addressed review comments in Javadoc

test/jdk/java/net/InetAddress/OfLiteralTest.java line 232:

> 230:                 //      with leading 0s treated as octal segment prefixes
> 231:                 Arguments.of(InetAddressClass.INET4_ADDRESS_POSIX,
> 232:                         "0177.0200401", ipv4_127_1_1_1),

Instead of 127.0.1.1 and 127.1.1.1 I would suggest instead testing 127.0.1.2 and 127.1.2.3 so that endianness is demonstrated (i.e. which of 0x102 vs 0x201 gives that result, similarly for octal).

-------------

PR Review Comment: https://git.openjdk.org/jdk/pull/18493#discussion_r1560761102