BSD Commit Log Search

   sun4v: Improve register printout in print_winregs()

   menu(3x): Constification and voidification of pointer arguments.

   libmenu has several functions that take a non-const char * argument
   and then (effectively) strdup(3) it. I've changed it so that these
   functions take a const char * argument. Modern GCC objects to string
   literals being passed to these functions otherwise.

   In other cases, a userptr allows some data to be associated with a
   data structure, but its type is char *. I've changed it to void *.

   Both of these changes are consistent with similar changes in ncurses,
   although are inconsistent with SVR4 libmenu versions. libmenu has
   unfortunately never been standardized.

   "that is fine" - blymn@

   sun4v: Remove incorrect mem_address_not_aligned trap printout.

   Need to figure out how to print the actual address causing the trap.

   Improve layout in print_trapframe() - NFC

   change failure and success path in testing bpf in npf

   There are two tests here, standlone BPF byte code and JIT-compiled code

   we don't need if/else block, we need an assertion to trigger
   on any if there is unsuccessful compilation.

   this avoids tests failures being unnoticed because we only see failures when
   running in verbose mode

   atto(1): Sprinkle __dead on functions that never return.

   Pacifies -Werror=missing-noreturn in clang:

   /home/source/ab/HEAD-llvm/src/usr.bin/atto/main.c:24:1: error: function 'usage' could be declared with attribute 'noreturn' [-Werror,-Wmissing-noreturn]
   {
   ^
   /home/source/ab/HEAD-llvm/src/usr.bin/atto/main.c:134:1: error: function 'fatal' could be declared with attribute 'noreturn' [-Werror,-Wmissing-noreturn]
   {
   ^
   2 errors generated.

   g/c __NEWINTR

   typo in comment.

   aarch64: no need to DSB after every DC instruction in ln_dcache_*_all

   The ln_dcache_{wb,wbinv,inv}_all helper do not need to DSB after each
   DC instruction. Instead DSB after completing the cache level.

   In-tree GCC has been fixed to not generate overlowing relocations
   when profiling on 68010, so no need to force NOPROFILE anymore.

   FUNCTION_PROFILER(): On 68010, use the same two step profile counter
   reference strategy as ColdFire does.  This avoids overflowing R_68K_PC16
   relocations for pretty much every reasonably sized program build for
   profiling.

   Fix the mcount entry name in the 68010 case.

   nix whitespace

   Welcome to 11.1!

   Disable 11.0 dedication

   Decorate next68k_bootargs() with no_instrument_function.

   new expat 2.10.0 out

   Add -fno-unwind-tables to CFLAGS on all m68k platforms.

   m68k/conf/Makefile.kern.m68k.inc now sets -fno-omit-frame-pointer
   automagically when it's needed, so don't bother doing it in
   individual kernel configs.

   Also set -fno-omit-frame-pointer if DDB or KGDB is configured.

   just avoid hardcoding the interface key.

   use the name variable and adjust for it in kernel.

   needs both userland and kernel update to be in sync

   revert last diff

   passed wrong nvlist key to the kernel

   fixes broken tests

   Profiling requires the frame pointer, so if PROF is set in the Makefile,
   add -fno-omit-frame-pointer.

   Problem reported by Romain Dolbeau on port-m68k.

   i915: Fix execlist submission register indexing harder.

   This time, with any luck, I transcribed the indexing correctly.

   PR port-amd64/58657: i915drmkms blank screen

   atto(1): Errors relating to the menu generally aren't fatal,
   so don't treat them as such.

   atto(1): fatal cannot take special format string arguments other than
   the initial %s for the program name. Gain some safety by not having
   callers pass a format string in the first place.

   With this, we can raise WARNS to 5.

   atto(1): don't cast malloc return value

   Revert the reclaimd changes until I've had a chance to look into the
   ramifications more closely.

   PR lib/60949 pthread*_getname_np(3) - avoid UB if len == 0

   Just make len == 0 be an EINVAL error.

   Update pthread_attr_getname_np(3) man page, so it no longer says
   there are no errors from pthread_attr_getname_np() - which was wrong
   in any case, as there was already one (if attr is invalid), and
   while there, one minor grammar fix.

   No pullups currently planned.

   Correct mapping of KS_Power for the "Keyboard Power" Usage ID.

   It's USB Keyboard/Keypad Page usage 102 (not 109 which is F18).