)]}'
{
  "commit": "b791d1bdf9212d944d749a5c7ff6febdba241771",
  "tree": "c207137a4d4f6b5dae3b1ecdf0ffaa357852fa7c",
  "parents": [
    "9716e57a0195dae356ae1425df121988abd27131",
    "1f44328ea24c9de368a3cfe5cc0e110b949afb2e"
  ],
  "author": {
    "name": "Linus Torvalds",
    "email": "torvalds@linux-foundation.org",
    "time": "Thu Jun 11 18:55:43 2020 -0700"
  },
  "committer": {
    "name": "Linus Torvalds",
    "email": "torvalds@linux-foundation.org",
    "time": "Thu Jun 11 18:55:43 2020 -0700"
  },
  "message": "Merge tag \u0027locking-kcsan-2020-06-11\u0027 of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip\n\nPull the Kernel Concurrency Sanitizer from Thomas Gleixner:\n \"The Kernel Concurrency Sanitizer (KCSAN) is a dynamic race detector,\n  which relies on compile-time instrumentation, and uses a\n  watchpoint-based sampling approach to detect races.\n\n  The feature was under development for quite some time and has already\n  found legitimate bugs.\n\n  Unfortunately it comes with a limitation, which was only understood\n  late in the development cycle:\n\n     It requires an up to date CLANG-11 compiler\n\n  CLANG-11 is not yet released (scheduled for June), but it\u0027s the only\n  compiler today which handles the kernel requirements and especially\n  the annotations of functions to exclude them from KCSAN\n  instrumentation correctly.\n\n  These annotations really need to work so that low level entry code and\n  especially int3 text poke handling can be completely isolated.\n\n  A detailed discussion of the requirements and compiler issues can be\n  found here:\n\n    https://lore.kernel.org/lkml/CANpmjNMTsY_8241bS7\u003dXAfqvZHFLrVEkv_uM4aDUWE_kh3Rvbw@mail.gmail.com/\n\n  We came to the conclusion that trying to work around compiler\n  limitations and bugs again would end up in a major trainwreck, so\n  requiring a working compiler seemed to be the best choice.\n\n  For Continous Integration purposes the compiler restriction is\n  manageable and that\u0027s where most xxSAN reports come from.\n\n  For a change this limitation might make GCC people actually look at\n  their bugs. Some issues with CSAN in GCC are 7 years old and one has\n  been \u0027fixed\u0027 3 years ago with a half baken solution which \u0027solved\u0027 the\n  reported issue but not the underlying problem.\n\n  The KCSAN developers also ponder to use a GCC plugin to become\n  independent, but that\u0027s not something which will show up in a few\n  days.\n\n  Blocking KCSAN until wide spread compiler support is available is not\n  a really good alternative because the continuous growth of lockless\n  optimizations in the kernel demands proper tooling support\"\n\n* tag \u0027locking-kcsan-2020-06-11\u0027 of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (76 commits)\n  compiler_types.h, kasan: Use __SANITIZE_ADDRESS__ instead of CONFIG_KASAN to decide inlining\n  compiler.h: Move function attributes to compiler_types.h\n  compiler.h: Avoid nested statement expression in data_race()\n  compiler.h: Remove data_race() and unnecessary checks from {READ,WRITE}_ONCE()\n  kcsan: Update Documentation to change supported compilers\n  kcsan: Remove \u0027noinline\u0027 from __no_kcsan_or_inline\n  kcsan: Pass option tsan-instrument-read-before-write to Clang\n  kcsan: Support distinguishing volatile accesses\n  kcsan: Restrict supported compilers\n  kcsan: Avoid inserting __tsan_func_entry/exit if possible\n  ubsan, kcsan: Don\u0027t combine sanitizer with kcov on clang\n  objtool, kcsan: Add kcsan_disable_current() and kcsan_enable_current_nowarn()\n  kcsan: Add __kcsan_{enable,disable}_current() variants\n  checkpatch: Warn about data_race() without comment\n  kcsan: Use GFP_ATOMIC under spin lock\n  Improve KCSAN documentation a bit\n  kcsan: Make reporting aware of KCSAN tests\n  kcsan: Fix function matching in report\n  kcsan: Change data_race() to no longer require marking racing accesses\n  kcsan: Move kcsan_{disable,enable}_current() to kcsan-checks.h\n  ...\n",
  "tree_diff": []
}