sys/atomic.h: Avoid needless stack spillage with typeof_unqual.
Using typeof_unqual strips the volatile qualifier on the local
temporaries, so the compiler doesn't issue store/load cycles for
access to them.
PR kern/60272: sys/atomic.h: unnecessary stack spillage
psci: ensure cores are in a known state before SYSTEM_OFF
The PSCI spec says that the caller must place all cores in a known state
prior to calling SYSTEM_OFF.
It doesn't really specify what a "known state" is, but it does suggest
that one way to do this is to call CPU_OFF on all other cores. So let's
do that.
This has the nice side effect of workaround around a heartbeat timeout
when attempting to power down a system with firmware that doesn't return
from a SYSTEM_OFF call but does not actually know how to power off the
system (like Rockchip's secure firmware on RK356x based boards).
lib/libc/README: Note that jemalloc definition has moved.
Should maybe find a better example than malloc/free for this but
let's at least fix this one point of confusion.
Pull up following revision(s) (requested by hgutch in ticket #287):
tools/xz-include/Makefile: revision 1.5
tools/xz-include/Makefile: revision 1.6
Avoid "sandbox" features in tools xz; fixes tools build failure on Linux
tools/xz: fix tools build errors on NetBSD/i386 hosts
Pull up following revision(s) (requested by roy in ticket #286):
sys/net/if.c: revision 1.538
if: abort early if no queue and the link state is the same
Restores a minor optimisation the prior commit accidently removed.
gcc/vax: adjust the constraints on the output operands of the
'*extzv_aligned' instruction patterns, removing the 'earlyclobber'
modifier, to be consistent with the constraints on the output operands
of the related 'extv_aligned' instruction patterns.
elftoolchain/netbsd-base-system-import.sh: Update to upstream [r4372].
This change implements a '-rREV' option for importing a specific
upstream revision.
Restore line whose content vanished in previous
Somehow in the previous version, the content of one line was
removed, leaving just the indentation tabs... restore that line.
While here, and inspired by that line remnant, check for trailing
whitespace, and obliterate all of it that is unintentional (there
is one space in an EDIT_ME in a here-doc which is intended).
Also, in the rarely used "zones added"/"zones removed" commit message,
for the set lists, add some vertical white space before the listings
of any zones added or removed (happens so rarely, I'm not sure that
code has ever been used).
Pull up following revision(s) (requested by christos in ticket #285):
external/ibm-public/postfix/dist/conf/main.cf: revision 1.13
fix so that it works out of the box (from RVP)
Pull up following revision(s) (requested by skrll in ticket #284):
sys/arch/riscv/include/vmparam.h: revision 1.16
risc-v: bump some parameter values on riscv64
Match all other 64bit platforms for
- PAGER_MAP_DEFAULT_SIZE
- UBC_WINSHIFT
- UBC_NWINS
Spotted by thorpej
PR misc/60253 (conditionally) remove CRYPTO using example
This completes the previous (2026-03-03) change, by removing an
example from the EXAMPLES section, which would only work if nc
was built with the CRYPTO option.
Like the previous, changing the definition of the number register C
in the man page source from 0 to 1 will reinstate the example, along
with all of the CRYPTO options, but makes no sense unless someone
does the required work to be able to build nc with CRYPTO defined.
No pullups required, the previous changes weren't pulled up either.
nist_hash_drbg: Fix citation to standard.
- Note that it's Rev. 1.
- Provide a URL.
- Link also to the test vectors.
- And link to an archive of the test vectors, just in case.
virtio(4): Allow virtio 0.9 BAR0 type to be memory rather than I/O.
This matches virtio>=1.0, and can't break working `hardware': any
existing virtio devices that worked must have reported I/O-type BAR0,
so they will continue to work; this will only enable previously
unusable virtio devices, reporting memory-type BAR0, to work.
Patch from Petri Koistinen.
PR kern/60247: virtio(4): legacy attach fails when BAR0 is MMIO
aarch64: fix mod/ref emulation in the MI pmap support code
Correct the initial PTE settings as per the table below
ref | mod | AF | AP | OS_MODEMUL
emul | emul | | |
-----+------+-----+---------+------------
no | no | set | RO/RW | not set
yes | no | clr | RO/RW | not set
no | yes | set | RO | set
yes | yes | clr | RO | set
[*] a write fault should be fixed up, and allowed.
and detect the modified emulation using OS_MODEMUL in pmap_fault fixup.
TODO:
- MI PMAP pmap_clear_reference needs to look more like pmap_clear_modify
and actually change PTE entries.
- MI PMAP needs support for HW updated modified and referenced bits.
