)]}'
{
  "commit": "a9cd14bfb0c113f1cd77be4349784663bec71bb4",
  "tree": "4828dcacf8ee6cfcf26c81359eec46f5f5a29268",
  "parents": [
    "b52ef450139f5632d7e53e6fd5f5ec46c7029dac"
  ],
  "author": {
    "name": "Karl Mehltretter",
    "email": "kmehltretter@gmail.com",
    "time": "Sat Sep 26 21:13:25 2026 +0200"
  },
  "committer": {
    "name": "Will Deacon",
    "email": "will@kernel.org",
    "time": "Wed Oct 07 09:39:11 2026 +0000"
  },
  "message": "arm64: irq: exclude the softirq stack switch from KCOV\n\nOn IRQ exit, __irq_exit_rcu() drops HARDIRQ_OFFSET before calling\ninvoke_softirq(). The arm64 do_softirq_own_stack() wrapper and its\n____do_softirq() trampoline run before __do_softirq() establishes\nsoftirq context, so KCOV records their PCs in the interrupted task\u0027s\ncoverage buffer. They also run outside softirq accounting when\nlocal_bh_enable() reaches do_softirq().\n\nThe IRQ-exit coverage makes the KCOV boot selftest fail even after\nthe scheduler and timer coverage leaks are suppressed. The generic\nsoftirq.o is already excluded from KCOV, but that exclusion does not\ncover the separately compiled arm64 wrappers.\n\nExclude irq.o from KCOV instrumentation, as is already done for the\narm64 entry code. This covers both wrappers even with compilers that\nlack the no_sanitize_coverage attribute. The softirq action functions\nremain instrumented for explicit remote coverage.\n\nFixes: 8eb858c44b98 (\"arm64: run softirqs on the per-CPU IRQ stack\")\nCc: stable@vger.kernel.org\nAssisted-by: LLM\nSigned-off-by: Karl Mehltretter \u003ckmehltretter@gmail.com\u003e\nSigned-off-by: Will Deacon \u003cwill@kernel.org\u003e\n",
  "tree_diff": [
    {
      "type": "modify",
      "old_id": "d2690c3ec52885eaadbc1fd753fc3ff4afd1773e",
      "old_mode": 33188,
      "old_path": "arch/arm64/kernel/Makefile",
      "new_id": "ea66339435bf0b373eb5a0bf92106367280a83d6",
      "new_mode": 33188,
      "new_path": "arch/arm64/kernel/Makefile"
    }
  ]
}
