Fix a typo (share/locale/Makefile.locale)
While in an actual Makefile variable assigmnent, this value seems
not to be used (the LANGUAGE_XX variables appear to be never used),
so the typo was completely benign. Note that there is no "au" in
the LANGUAGE list.
Fix it for consistency, and so no-one else ever needs to wonder
what is going on here (for which the answer is "nothing").
pchtemp(4): Don't block suspend.
Don't think this keeps any interesting state or operations on it that
need to be quiesced while suspended.
PR port-amd64/60441: pchtemp(4) blocks suspend
Revert previous (x86/ALL: Enable nvmm(4)).
- i386 has no nvmm support (HAVE_NVMM=no) and fails with incomplete type.
- amd64 cannot have both nvmm and KASAN and fails pmap_ept_* functions
as KASAN clears __HAVE_DIRECT_MAP and therefore pmap_ept_* gets excluded.
Both kernels compile again.
Ok: Taylor R Campbell
atto(1): Initial import of tiny text editor, as discussed on
tech-userlevel@.
This is Hugh Barney's atto text editor modified for inclusion in
NetBSD. Since the upstream repository is inactive and the original
author wishes to keep the code base below 2000 lines of C, this is
effectively a fork that lives in usr.bin/atto. It serves as an
alternative to vi(1) for users more accustomed to modeless text
editors.
Notable changes include reworked keyboard input code to handle a wider
range of terminals (via libcurses cursor key translation), more useful
help for unrecognized key presses, and standardized option handling
using getopt (with syntax highlighting disabled by default).
Thanks martin and other folks for feedback and testing.
ATTO(1) General Commands Manual ATTO(1)
[115 lines not shown]
Additionally pull up following revision(s) (requested by riastradh in ticket #1380):
lib/libc/gen/pthread_atfork.c: revision 1.20
Specify MAP_ANON, the default (MAP_FILE) with a fd == -1 makes no sense.
Introduce debug function print_winregs() and clear_winregs().
These debug functions can be used to print out snapshots of the register
window.
The SAVE_WINREGS() macro in locore.s creates the snapshot.
Pull up following revision(s) (requested by yurix in ticket #561):
sys/arch/arm/ti/am3_prcm.c: revision 1.15
am335x: fix timer clock offset
Closes PR 60910
Fix by Elias Howell.
Pull up following revision(s) (requested by joe in ticket #1405):
sys/dev/pci/if_vmx.c: revision 1.19
fix wrong range from computing the MIN of tx and rx queues.
Pull up following revision(s) (requested by joe in ticket #560):
sys/dev/pci/if_vmx.c: revision 1.19
fix wrong range from computing the MIN of tx and rx queues.
Pull up following revision(s) (requested by rin in ticket #1402):
sys/arch/alpha/alpha/vm_machdep.c: revision 1.124
sys/arch/alpha/alpha/machdep.c: revision 1.386
alpha: Align kernel stack pointer to 16-byte boundaries
as ABI requires. With this fix, full ATF run successfully completed
without any regression.
PR port-alpha/60865 NetBSD-current/alpha panics running getcwd test
OK thorpej@
Pull up following revision(s) (requested by rin in ticket #557):
sys/arch/alpha/alpha/vm_machdep.c: revision 1.124
sys/arch/alpha/alpha/machdep.c: revision 1.386
alpha: Align kernel stack pointer to 16-byte boundaries
as ABI requires. With this fix, full ATF run successfully completed
without any regression.
PR port-alpha/60865 NetBSD-current/alpha panics running getcwd test
OK thorpej@
Pull up following revision(s) (requested by riastradh in ticket #1401):
libexec/ld.elf_so/expand.c: revision 1.8
libexec/ld.elf_so/paths.c: revision 1.43
lib/libc/misc/stack_protector.c: revision 1.10
lib/libc/include/__sysctl.h: revision 1.1
ld.elf_so: Use __sysctl() syscall stub rather than sysctl() wrapper.
This avoids pulling in some unnecessary junk from libc, saving on a
couple kilobytes of space and some relative relocations at runtime.
PR lib/59387: ld.elf_so: needlessly includes libc/gen/sysctl.c user.*
Pull up following revision(s) (requested by riastradh in ticket #554):
sys/kern/vfs_vnode.c: revision 1.159
vdrain: Recompute target after each iteration.
This way if sysctl -w kern.maxvnodes=N fails, the vdrain task doesn't
perpetually keep trying to attain the unattainable.
Doesn't fix everything about racy access to desiredvnodes, but it
should help with the major symptom.
PR kern/60603: Sisyphean vdrain task
