Update libexpat to version 2.7.1.
Relevant for OpenBSD are bug fixes #980 #989, other changes #978
#986. No library bump necessary, only version defines have been
changed in public header file. This fixes a regression introduced
in libexpat 2.7.0.
OK tb@
Reuse dmesg.boot file for amd64 MDEFI check, clarify this usage pattern
install.md r1.63 assumed it to be present, r1.64 reverted to plain dmesg(8),
now hoist creation of the current boot's dmesg in a new DMESGBOOT variable
for reuse (again) and tweak comments around it.
There are different reasons and ways the installer looks at dmesg:
- Here it concerns feature detection of something that won't change across
reboots or during install/upgrade time: whether we booted via EFI/GPT.
- scan_dmesg() uses that file to lookup keyboards, displays, consoles, etc.
- diskinfo() on the other hand needs to see all disks at runtime,
so live 'dmesg | ...' is needed to find latest attach line for them.
- start_dmesg_listener() is its own mechanism and keeps a file to watch
for changes, i.e. nothing that should change whilst refining the above
mentioned.
[3 lines not shown]
Fix scope of the sa_mpls sockaddr variable.
The sa_mpls is used outside of the block via info.rti_info[RTAX_SRC] so
it needs to have the same scope as info.
Yesterday the planets and the moon were in perfect alignment to trigger
another bug in route(8) because of this and I could observe all of its
glory first hand for the first time after 7 years.
OK bluhm@ tb@
allow ice(4) to work when phy type does not map to a known media type
Map unknown PHY types to IFM_ETHER as a catch-all, as suggested by
dlg@ in https://marc.info/?l=openbsd-tech&m=173388658508179&w=2
Verified by temporarily making the phy type of my test device unknown.
Two minor tweaks to the description of -V:
1) To make it even clearer that "the first" is also odd-numbered,
say "the first and every other odd-numbered" as suggested by jmc@.
2) When introducing the example, say what exactly the non-digit substrings
are, as suggested by pascal@.
Tweaks and OK jmc@, OK pascal@ on an earlier version.
Add support for QCA2066. It is very similar to QCNFA765,
The most significant difference is that QCA2066 supports 3-antenna
configuration while QCNFA765 does not. To differentiate them,
subversion numbers are used.
See Linux commit 5dc9d1a55e953d9059ecbdd8fe6ec81e9edd349e
Also fix the regdb firmware load, code taken from qwz.c r1.11 by patrick@.
Tested by and ok stsp@
Pass "ControlMaster no" to ssh when invoked by scp & sftp.
If you have ControlMaster auto (or yes) in your config, and the
first connection you make is via scp or sftp, then you may get a
few unexpected options applied to it (eg ForwardX11 no), since sftp
and sftp explicitly disable those for reasons. These effects will
persist beyond the initial scp or sftp command.
This explicitly disables persistent session *creation* by scp and sftp.
It will not prevent them from using an existing session if one has
already been created.
From Github PR#557, ok djm@ kn@
Use the modified time of the RRDP stat file to warn if a server did not
change for more than 24h.
Right now all operational RRDP servers publish more often and this allows
to spot malfunctioning servers hopefully before the cache expires.
Based on an initial diff by job@
OK tb@ job@
There is no need to update the rrdp state file if there was no modification.
This is needed to better track how long a RRDP server shipped the same file.
The old data is still valid in that case. Especially since rpki-client allows
the sequence number to be off by 2 because CDNs are hard.
OK tb@ job@
document that power management happens in the kernel,
that some functionality is available without apmd,
and how apmd fits into the picture;
tweaks/ok jca
