BSD Commit Log Search

   cnmagic(9): minor tweaks

   Call it a "magic sequence" (not "magic key sequence") - there are no
   "keys" in the serial input and wscons(4) does not even use cnmagic(9),
   so.  Mention that latter fact in CAVEATS.

   Document CNC_BREAK.

   cpufreq_dt: communicate full voltage range to regulators

   In the last commit, fdtbus_regulator_set_voltage got changed to accept the voltage as a target, min, max triple from a min, max pair. This commit switched the cpufreq code over to sending the full voltage triple to regulators if it is available in the device tree.

   This fixes PR 60671

   fdtbus_regulator_set_voltage_triple: support target/min/max

   Some dt-bindings specify voltages as target/min/max, while the NetBSD fdt subsystem only supported min/max.

   This PR introduces fdtbus_regulator_set_voltage_triple that tries to set the voltage to an optimal value before falling back to the full range.

   See PR 60671

   calendar.computer: Add the year the first FORTRAN manual was released.

   via Martin Neitzel in PR bin/60894

   s/GX_DRAW_SOFTAWRE/GX_DRAW_SOFTWARE/ in constant definition (unused).

   s/wheter/whether/ in comments.

   linux_sys_recvfrom: translate message flags.

   sun4v: Try to use the hypervisor interrupt api version 2 if version 3 fails.

   Version 3 systems are working ok - trying to fix the setup when a version 2
   system is used.

   Not tested on yet.

   now we have enough hacks, move alpha and m68k to GCC 14.

   thanks to rin@ for making this not a regression.

   cvs(1): Fix acl/racl argument parsing.

   PR bin/60892: cvs acl/racl argument parsing is busted

   riscv: Make comapt32 compile.

   Note: I doubt it will work -- cpu_getmcontext32 and cpu_setmcontext32
   appear to be missing save/restore of most registers, so signal
   delivery will probably trash them.  Also not sure whether to use
   l->l_private or tf->tf_reg[_REG_TP] here (or _X_TP).  TBD!

   PR port-riscv/60889: riscv64: compat_netbsd32 is busticated

   sys/conf/files: riscv has no EXEC_AOUT either even for compat32.

   Do the same ifdef dance as for mips and arm.

   Fixes:

   > /home/riastradh/netbsd/current/src/sys/compat/netbsd32/netbsd32_exec_aout.c: In function 'exec_netbsd32_makecmds':
   > /home/riastradh/netbsd/current/src/sys/compat/netbsd32/netbsd32_exec_aout.c:119:15: error: 'NETBSD32_MID_MACHINE' undeclared (first use in this function)
   >   119 |         case (NETBSD32_MID_MACHINE << 16) | ZMAGIC:
   >       |               ^~~~~~~~~~~~~~~~~~~~

   PR port-riscv/60889: riscv64: compat_netbsd32 is busticated

   printf(3): try to improve description of positional arguments

   Don't use "field" to refer to positional argument specification in
   `%n$`, as the term is already used for the output field (as in "field
   width", etc).

   Reword the `*n$` paragraph that first refers to `n` as "one or more
   decimal digits" and contrasts it with the "digit string" that actually
   refers to an explicitly specified width or precision far away on the
   previous page/screen (!).  This is very unobvious, especially when you
   read the page out of order, as one often does.

   The result of this (intentionally) localized edits is still pretty
   poor overall, but at least it fixes, hopefully, the two most glaring
   problems.  Further editorial help is welcome.

   drm: Nix `highmem' memory accounting.

   This led to confusing dmesg output when the kernel map size was
   enlarged between 9 and 10, probably by this commit:

   > Module Name:    src
   > Committed By:   chs
   > Date:           Wed Aug  3 01:52:11 UTC 2022
   >
   > Modified Files:
   >         src/sys/uvm: uvm_km.c
   >
   > Log Message:
   > for platforms which define NKMEMPAGES_MAX_UNLIMITED, set nkmempages
   > high enough to allow the kernel to map all of RAM into kmem,
   > so that free physical pages rather than kernel virtual space is
   > the limiting factor in allocating kernel memory.  this gives ZFS
   > more flexibility in tuning how much memory to use for its ARC cache.
   >

    [13 lines not shown]

   Make mpl115a2 return correct values.

   Do not byteswap calibration coefficients - apparently, this was always
   wrong. Properly wait before reading. Calculate in 64-bit arithmetic.
   Use ENVSYS_PRESSURE.

   tmux 3.8 is out

   vr(4): set or clear VR_RXCFG_RX_PROMISC bit in vr_setmulti().

   Previously the promiscuous bit was only handled in vr_init().
   When an interface is added to or removed from bridge(4), its
   IFF_PROMISC flag is changed through the interface flags (SIOCSIFFLAGS).
   On a running interface it does not reinitialize device, but calls
   vr_setmulti() directly, which did not update the promiscuous bit in RX filter.

   PR port-i386/60884

   move null check on top to avoid null pointer dereference

   que pointer is not likely to be null though( stated by tls@)

   printf(3): use emphasis when a term is used for the first time

   Use it for "conversion specification", "flag", "field width,
   "precision", "length modifier", "conversion specifier".

   printf(3): tweak formatting of n$

   XXX: The way the description of positional arguments is worded is
   pretty awkward.  It [ab]uses "field" to refer to the conversion
   specification itself, when the standard already uses it to refer to
   the output field.

   printf(3): use plain Cm for flags, use consistent -width Cm

   Ticket #1364

   Pull up following revision(s) (requested by kre in ticket #1364):

        common/lib/libc/string/memset2.c: revision 1.12

   PR port-mips/60780 UBSan: long int << fix (src/common/lib/libc/string/memset2.c)

   The comments say that memword_t is intended to be unsigned, but
   presume that __register_t will provide that attribute.   It usually
   doesn't, so we need an unsigned type for building the fill value.

   Thanks to Alexander Schreiber for running the UBSan built system
   and reporting this (and others), and hopefully, in advance for
   testing that this one is now fixed.

   printf(3) - add missing ',' (comma).

   In the description of the ' ' (space) flag, a comma was missing.

   avoid shifting 32 bits in in a 32 bits integer in IPv6 flow hashing

   avoid shifting 32 bits in in a 32 bits integer in IP flow hashing

   compat32: Advertise true stack base to getcontext(2).

   Needs adjustment for PaX ASLR.

   This was missed as part of:

   > Module Name:    src
   > Committed By:   christos
   > Date:           Wed Dec 26 22:11:54 UTC 2007
   >
   > Modified Files:
   >         src/sys/conf: files
   >         src/sys/kern: exec_elf32.c exec_subr.c init_sysctl.c kern_exec.c
   >             kern_pax.c kern_sig.c
   >         src/sys/sys: exec.h exec_elf.h pax.h proc.h
   >         src/sys/uvm: uvm_extern.h uvm_mmap.c
   >
   > Log Message:
   > Add PaX ASLR (Address Space Layout Randomization) [from elad and myself]

    [5 lines not shown]

   Ticket #1363

   Pull up following revision(s) (requested by kre in ticket #1363):

        lib/libc/stdio/vfwprintf.c: revision 1.42,1.43

   PR lib/60708 - include 'F' format when fetching %n$ args.

   PR bin/60496 - fix printf(3) with non-single-byte decimal point & thousands sep

   These fixes are relatively simple, and should be safe -- I have had them
   in my source tree for ages now.

   The fix for the decimal point case is taken more or less directly from
   an old FreeBSD fix for the same issue.

   The thousands separator case is different - our code and FreeBSD's have
   diverged so much in this area that simply using their method would have
   been more trouble than it was worth, so I reimplemented it differently
   to better fit our code model.


    [3 lines not shown]

   Ticet #531