---
- branch: MAIN
  date: Sat Apr  9 12:07:01 UTC 2022
  files:
  - new: '1.31'
    old: '1.30'
    path: src/common/lib/libc/arch/i386/atomic/atomic.S
    pathrev: src/common/lib/libc/arch/i386/atomic/atomic.S@1.31
    type: modified
  - new: '1.24'
    old: '1.23'
    path: src/common/lib/libc/arch/x86_64/atomic/atomic.S
    pathrev: src/common/lib/libc/arch/x86_64/atomic/atomic.S@1.24
    type: modified
  - new: '1.54'
    old: '1.53'
    path: src/sys/arch/amd64/include/frameasm.h
    pathrev: src/sys/arch/amd64/include/frameasm.h@1.54
    type: modified
  - new: '1.34'
    old: '1.33'
    path: src/sys/arch/i386/include/frameasm.h
    pathrev: src/sys/arch/i386/include/frameasm.h@1.34
    type: modified
  - new: '1.50'
    old: '1.49'
    path: src/sys/arch/x86/x86/patch.c
    pathrev: src/sys/arch/x86/x86/patch.c@1.50
    type: modified
  id: 20220409T120701Z.bbf3347e611cc331dbe4dc7d31032717db603588
  log: |
    x86: Every load is a load-acquire, so membar_consumer is a noop.

    lfence is only needed for MD logic, such as operations on I/O memory
    rather than normal cacheable memory, or special instructions like
    RDTSC -- never for MI synchronization between threads/CPUs.  No need
    for hot-patching to do lfence here.

    (The x86_lfence function might reasonably be patched on i386 to do
    lfence for MD logic, but it isn't now and this doesn't change that.)
  module: src
  subject: 'CVS commit: src'
  unixtime: '1649506021'
  user: riastradh