BSD Commit Log Search

   micphy: Apply errata workaround for KSZ9031

   This fixes ethernet on an olimex a64 board, and should be pulled up
   to -11, -10, -9...

   Thanks jmcneill for pointing out the problem.

   PR port-arm/60193: olimex a64 - no carrier with emac0 ethernet

   3RDPARTY: note newer Lua releases

   Add cast to fix the build on some platforms

   sunxidrm: suppress misleading error message

   Do not print "failed to activate endpoint" if the device tree leaves a
   CRTC unused.

   sunxi: Inject PMF display on/off events in CRTC-0 atomic_{enable,disable}

   Fixes "dpms" on Pinebook after update to drm atomic interface.

   catch up with gftfb and hyperfb:
   - support bold mono characters
   - use glyphcache_underline() where appropriate
   - set & support RI_FULLCLEAR

   use glyphcache_underline(), add & support RI_FULLCLEAR

   some minor fixes:
   - underling anti-aliased characters properly
   - set & support RI_FULLCLEAR

   update for xorg-server 21.1.22.

   Handle the new common m68k genassym.cf

   Add missing archs

   re-instate the namespace protection from 5.3.0

   tests/lint: add test cases for possibly zero-extended sign bit

   lint: simplify deriving pointer types in expressions

   lint: add more details to message about snprintf combinations

   put back jemalloc to 5.3.0, got enough build info

   lint: warn about 'const' or 'volatile' in function return type

   tests/lint: remove outdated comment, add volatile cast

   try moving everyone to the new jemalloc

   switch x86 to jemalloc 5.3.1

   Re-factor genassym.cf into common m68k definitions and platform-specific
   ones.

   This machine can't have an HP MMU, so don't default mmutype to that value.

   Fix overlapping strcpy usages

   Using overlapping src & dest witgh strcpy is undefined, so anything
   like
        strcpy(str + N, str + M);
   is automatically invalid, regardless of the valuse of N and M.

   Change offending usages here to use memmove() instead.

   Note: 2 issues:  First, there is no (in sntp/compat) anything to deal
   with the possibility that memmove() is not available - this should not
   be a problem building nbsd, but it might be if any attempt were to be
   made to upstream this change (the autoconf machinery to detect memmove()
   would be needed, along wth a memmove() implementation).

   Second, and more serious here, I haven't been able to get a build to
   actually compile this code, so while I believe it is correct, I am unable
   to test it.   This is intended to fix the sh3 builds on in the releng
   daily builds, so we will see what happens.  As best I can tell my build

    [2 lines not shown]

   Fix previous for GENERIC64 and restore the call to pmap_tlb_asid_acquire
   in pmap_activate

   Note jemalloc switch requires cleaning of libc and external/bsd/jemalloc

   rgephy.c: drop a long-obsolete comment

   In fact, FreeBSD borrowed NetBSD's approach here twenty years ago.

   merge changes between jemalloc-5.3.0 and 5.3.1

   riscv/byte_swap: remove unneeded LINTED comment

   The cast to uint16_t is enough to convince lint that the code truncates
   the value on purpose.

   sysctl: fix lint warnings

   libusbhid: remove unneeded LINTED comment in do-while macro