bind9: fix broken build

[tc-agent]

Summary

Fix the broken OSS-Fuzz build for bind9. The build has been failing since 2026-04-06 (build status JSON, last successful build 6808bda2 on 2026-04-05). All daily builds since then fail at:

+ git apply --ignore-space-change --ignore-whitespace /src/patch.diff
error: patch failed: meson.build:1584
error: meson.build: patch does not apply

(see latest build log)

Root cause

Upstream bind9 commit e16a3d7a8e ("embed default sanitizer flags in executables", merged via 7c60b7da on 2026-04-05) added a new sources: default_sanitize_options, line directly after link_args: in the fuzz_binaries executable() block. The block also moved from meson.build:1584 to meson.build:1882, so the existing patch.diff hunk no longer matches its surrounding context.

The bind9 meson.options file already provides a proper fuzzing-backend combo option (none/afl/libfuzzer/oss-fuzz) and an oss-fuzz-args string option, replacing the need to manually inject -fsanitize=fuzzer via patch.

Changes

build.sh: Pass -Dfuzzing-backend=oss-fuzz -Doss-fuzz-args="$LIB_FUZZING_ENGINE" -Db_lundef=false to meson setup. The first two activate the upstream-supported integration path; the b_lundef=false flag is required by an explicit assertion in meson.build when any fuzzing backend is enabled with clang.

patch.diff: Drop the obsolete meson.build hunk (it tried to manually append -fsanitize=fuzzer, which is now handled by the oss-fuzz backend). Keep only the tests/meson.build hunk converting libtest = shared_library(...) → static_library(...). The fuzz binaries depend on libtest_dep; keeping libtest static avoids a runtime libbindtest.so dependency that the OSS-Fuzz runner container cannot resolve.

Coverage

Baseline: Fuzz Introspector currently shows N/A because both the fuzzer and coverage builds have been failing. The last successful coverage report (2026-04-05) reported:

Metric	Coverage
Line	21.23% (26334 / 124031)
Function	24.17% (1433 / 5928)
Region	16.57% (36638 / 221064)

After this fix (5 minutes of fuzzing per harness, then coverage build over the resulting corpus):

Aggregate (all harnesses combined): line 17.53% (21687/123741), function 21.61% (1280/5922), region 13.46% (29666/220328).

Harness	Line	Function	Region
dns_master_load_fuzzer	4.53%	7.08%	3.33%
dns_message_checksig_fuzzer	7.68%	11.91%	5.65%
dns_message_parse_fuzzer	7.32%	9.72%	5.41%
dns_name_fromtext_target_fuzzer	0.72%	1.29%	0.48%
dns_name_fromwire_fuzzer	0.76%	1.40%	0.51%
dns_qp_fuzzer	0.54%	1.17%	0.38%
dns_qpkey_name_fuzzer	0.84%	1.53%	0.56%
dns_rdata_fromtext_fuzzer	5.07%	5.57%	4.24%
dns_rdata_fromwire_text_fuzzer	3.97%	5.67%	3.28%
isc_lex_getmastertoken_fuzzer	0.99%	1.62%	0.67%
isc_lex_gettoken_fuzzer	0.75%	1.38%	0.53%

The aggregate gap to the 2026-04-05 baseline (~3.7 line-coverage points) reflects the ~5 minute corpus generation budget vs. the long-accumulated production corpus; once OSS-Fuzz resumes daily fuzzing, the corpus will grow back to its previous reach.

[github-actions]

Fuzzing Coverage Report

Tested: project bind9 · base b4ad4f8 → head 3be7869 · 300s total fuzz budget · updated 2026-05-28 12:21 UTC · workflow run

Metric	Before	After	Delta
Static reachability	12.9% (1248/9683)	12.9% (1248/9683)	+0.0%
Line coverage	0%	17.6% (21894/124127)	new
Branch coverage	0%	11.7% (13056/111765)	new
Function coverage	0%	21.5% (1279/5936)	new

Per-harness

Harness	Lines before	Lines after	Δ
dns_master_load_fuzzer	0%	4.3% (5302/123058)	new
dns_message_checksig_fuzzer	0%	7.7% (9431/123266)	new
dns_message_parse_fuzzer	0%	7.6% (9297/123107)	new
dns_name_fromtext_target_fuzzer	0%	0.7% (876/123035)	new
dns_name_fromwire_fuzzer	0%	0.8% (928/123171)	new
dns_qp_fuzzer	0%	0.5% (664/123397)	new
dns_qpkey_name_fuzzer	0%	0.8% (1030/123053)	new
dns_rdata_fromtext_fuzzer	0%	5.7% (7018/123093)	new
dns_rdata_fromwire_text_fuzzer	0%	4.4% (5371/123128)	new
isc_lex_getmastertoken_fuzzer	0%	0.9% (1142/123047)	new
isc_lex_gettoken_fuzzer	0%	0.7% (917/123036)	new

_{Δ = (after − before) / before, to accommodate that denominators may change. "new" when before is 0; "deleted" when after is 0.}