BSD Commit Log Search

   prevent invalid sequences being written out, instead print '?' (Kilian Feess)

   prevent negative offsets hiding bugs (Kilian Feess)

   widen types to prevent truncation (Kilian Feess)

   Add checks to unchecked allocations (Kilian Feess)

   avoid buffer overreads (Kilian Feess)

   set nscrap to 0 on malloc failure (Kilian Feess)

   Handle the BACKSPACE_KEY explicitly and ignore c > 255 (Kilian Feess)

   fix missing initialization of b_bname (Kilian Feess)

   use a temporary file and rename to save a file atomically (Kilian Feess)

   Avoid filename trucation of paths > NAME_MAX, change it to PATH_MAX and
   detect truncation and beep (completion) or print an error message (save)
   (Kilian Feess)

   protect against the hiliter reading past EOF (Kilian Feess)

   check for pp < end_p in both forward and reverse search (Kilian Feess)

   make sure c is initialized when mbtowc fails and guard against reading
   incomplete utf8 sequences past the end of buffer {Kilian Feess)

   atto: in filename completion, don't insert letters for function keys

   m68881_{save,restore} -> fpu_{save,restore}

   m68k_cached_fpu_idle_frame -> fpu_idle_frame
   m68k_make_fpu_idle_frame -> fpu_make_idle_frame

   Fix cpu_{get,set}mcontext() vis a vis FP state in the presence of
   FPU_NONE+FPU_EMULATE.  Previously, the emulated 68881 registers,
   which are present in the PCB, were completely ignored becaue fputype
   is FPU_NONE in that case.

   This fix entails a few things:
   - In fpu_emulate(), when we've reached the point where we know an
     FPU instruction is going to be performed, set the PCB's copy of
     the FP frame to indicate an "idle" state.  This sets the stage
     for normal FPU handling code to use the saved emulator state
     just as it would with a real FPU.
   - In fpu_init(), if we set FPU_NONE, also initialize m68k_cached_fpu_idle_frame
     to mimic an idle 68881 idle frame so that the _UC_FPU+_UC_M68K_UC_USER
     case in cpu_setmcontext() works correctly.
   - In cpu_{get,set}mcontext(), always use the saved FP registers in the
     PCB, and only interact with the FPU if fputype != FPU_NONE and curlwp.

   While here, simplify the construction of the FPFRAME_IS_NULL() expression.

   Problem reported by Romain Dolbeau on port-m68k.

   powerpc: oea: downgrade "failed to allocate DEAD ZONE" log to debug

   0xDEA[CDE]xxxx may not exist on the platform, so no need to map them away.
   Downgrade the log message to a debug message since it's not fatal.

   texinfo: build makedoc as a host tool in src/tools

   makedoc is necessary to generate files for info(1) and infokey(1)
   and it was built in src/external/gpl2/texinfo/makedoc with config.h
   prepared for the target NetBSD hosts, but it's really confusing.

   Additional install rules is based on tools/Makefile.host and
   afterinstall ordering follows tools/groff/Makefile.

   Tested 'build.sh -U -m amd64 release' on ubuntu 26.04 with GCC 15.

   npf.conf.5: minor spelling and grammar fixes

   (While here, also bump the date for changes made in previous commit.
   Follow-on for PR misc/60946.)

   texinfo: add HAVE_DECL_STRCASECMP and HAVE_DECL_STRNCASECMP in config.h

   Should have been added on configure.ac fixes:
    https://mail-index.netbsd.org/source-changes/2026/09/25/msg164265.html

   PR misc/60946

   man for random block procedure extension in NPF

   Need to include opt_fpu_emulate.h here.

   natto.debug isn't installed, just atto.debug

   Remove natto.debug for now, until someone works out the correct
   atto Makefile magic to install it as a link, but only when MKDEBUG is
   set.

   Ticket #1408

   Pull up the following, requested by kre in ticket #1408:

        external/public-domain/tz/dist/Makefile         up to 1.8
        external/public-domain/tz/dist/NEWS             up to 1.11
        external/public-domain/tz/dist/TZDATA_VERSION   up to 1.49
        external/public-domain/tz/dist/africa           up to 1.5
        external/public-domain/tz/dist/backzone         up to 1.7
        external/public-domain/tz/dist/europe           up to 1.7
        external/public-domain/tz/dist/leapseconds      up to 1.13
        external/public-domain/tz/dist/leapseconds.awk  up to 1.1.1.16
        external/public-domain/tz/dist/northamerica     up to 1.10
        external/public-domain/tz/dist/southamerica     up to 1.7
        external/public-domain/tz/dist/theory.html      up to 1.9
        external/public-domain/tz/dist/version          up to 1.22
        external/public-domain/tz/dist/zone.tab         up to 1.8
        external/public-domain/tz/dist/zone1970.tab     up to 1.9
        external/public-domain/tz/dist/zonenow.tab      up to 1.10
        doc/3RDPARTY                                    (apply patch)

   Updated tzdata to version 2026f (using 2026fgtz).

   Fix editor mishap in previous

   PR bin/60862


   rump tests for NPF rule procedures

   for logging, we check if it loads successfuly and packet traverses it
   for routing, we also check if packet treversal returns successfuly
   for normalization, we check the packet for the fragment bit, the new maxmss
   and the ttl set.

   in reload, and fini, we "init" the module and also "fini" the module.

   atf-run and you see the reults.

   simplify the nvlist using the name variable

   atto(1): Rename to natto via symlink/alias.

   This is different enough from the spirit of the original editor now
   that it has libmenu support.