Factor gen_host46_byname() out.#1519
Merged
infrastation merged 7 commits intothe-tcpdump-group:masterfrom Oct 9, 2025
Merged
Conversation
Show gen_mac48host_byname() in the code path. The comments about pcap_ether_hostton() do not apply since commit ad56fa2.
For all MAC-48 DLTs generalise two existing EN10MB tests and introduce three more tests to cover all six varieties of ID that "gateway" rejects. For all other DLTs introduce five more tests to cover all eight varieties of ID that "gateway" rejects and run all tests in one foreach loop.
Add three new reject tests to cover more cases when a proto-qualified "host NAME" encounters a hostname that either does not resolve at all or only resolves to an address family that the protocol cannot use. Add four new reject tests to cover the current [incorrect] behaviour of "arp host NAME" and "rarp host NAME" when the hostname resolves in IPv6, regardless of whether it resolves in IPv4 (GH bug the-tcpdump-group#1512). Update three existing tests for consistency and place all the tests in one place.
For these two forms of the predicate, when a hostname resolves in both IPv4 and IPv6, the IPv6 part is supposed to be irrelevant, but it causes an unexpected failure (GH bug the-tcpdump-group#1512). Also, when a hostname does not resolve in IPv4 and resolves in IPv6, the error message is supposed to be correct, but it is not: $ filtertest EN10MB 'arp host eth-ipv4-ipv6.host123.libpcap.test' filtertest: 'arp' is not a valid qualifier for 'ip6 host' $ filtertest EN10MB 'rarp host eth-ipv4-ipv6.host123.libpcap.test' filtertest: 'rarp' is not a valid qualifier for 'ip6 host' $ filtertest EN10MB 'arp host eth-noipv4-ipv6.host123.libpcap.test' filtertest: 'arp' is not a valid qualifier for 'ip6 host' $ filtertest EN10MB 'rarp host eth-noipv4-ipv6.host123.libpcap.test' filtertest: 'rarp' is not a valid qualifier for 'ip6 host' In gen_scode() case Q_HOST get the IPv6 address skip condition right to fix this: $ filtertest EN10MB 'arp host eth-ipv4-ipv6.host123.libpcap.test' (000) ldh [12] (001) jeq #0x806 jt 2 jf 7 (002) ld [28] (003) jeq #0xa141e28 jt 6 jf 4 (004) ld [38] (005) jeq #0xa141e28 jt 6 jf 7 (006) ret #262144 (007) ret #0 $ filtertest EN10MB 'rarp host eth-ipv4-ipv6.host123.libpcap.test' (000) ldh [12] (001) jeq #0x8035 jt 2 jf 7 (002) ld [28] (003) jeq #0xa141e28 jt 6 jf 4 (004) ld [38] (005) jeq #0xa141e28 jt 6 jf 7 (006) ret #262144 (007) ret #0 $ filtertest EN10MB 'arp host eth-noipv4-ipv6.host123.libpcap.test' filtertest: unknown host 'eth-noipv4-ipv6.host123.libpcap.test' for specified address family $ filtertest EN10MB 'rarp host eth-noipv4-ipv6.host123.libpcap.test' filtertest: unknown host 'eth-noipv4-ipv6.host123.libpcap.test' for specified address family In pcap-filter(7) explain the address family semantics better in the description of "host NAME". Update the tests.
infrastation
force-pushed
the
gen_host46_byname
branch
from
3ae4822 to
6245c96
Compare
Member
Author
|
The previous revision failed 80 tests, this revision passes all tests, including the new tests it adds. It also includes a bug fix for the bug #1512. It passes a build matrix with Valgrind on Linux. |
infrastation
changed the title
In gen_scode() case Q_GATEWAY delegate the remaining memory management and error handling downstream; now if gen_gateway() returns, it always returns a non-NULL pointer. In gen_gateway() handle all errors as early as possible; instead of testing whether ai_addr is NULL test whether ai_family is set to the AF of interest: when that's the case, ai_addr is not NULL (this is how gen_scode() implements case Q_HOST).
infrastation
force-pushed
the
gen_host46_byname
branch
from
6245c96 to
350351f
Compare
Member
Author
|
This revision handles integer types a bit better, let's see if that clears the Windows CI failure. |
After the previous commit much of what gen_gateway() does is the IPv4 subset of case Q_HOST in gen_scode(). Factor the common code to a new function and update various comments.
infrastation
force-pushed
the
gen_host46_byname
branch
from
350351f to
c64aba1
Compare
Member
Author
|
This revision spells an assignment in a conditional expression in a way that hopefully won't upset MSVC, let's see if that worked. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
This is the more difficult part of a prototype I started, but could not quickly complete earlier. The intent is simple: since
gateway Xnominally includesnot host X, and since bothgen_gateway()andgen_scode()callpcap_addrinfo()and process the result(s) in a very similar manner, having a slightly different code block in each function makes the code harder to maintain. However, the complexity turned out to be in the detail (including #1512), so this work-in-progress revision fails many tests. Also before this can be progressed any further, the reject filter tests must cover all code paths that go throughpcap_addrinfo()and fail to resolve the hostname.