Add support for changing the mode on PCF8583 RTC chips.
Add mode 2 (counter) support and support devices at I2C address 0x51.
From Dave Tyson in PR kern/60700.
Pull up following revision(s) (requested by gdt in ticket #1407):
sys/net/if_tun.c: revision 1.179
sys/net/if_tun.c: Drop spurious EAFNOSUPPORT
When an outgoing packet is put in the queue, that might fail, and if
so the error will almost certainly be ENOBUFS. Let that error be
returned, rather than inexplicably forcing it to be EAFNOSUPPORT.
(EAFNOSUPPORT is returned nearby in the code for other cases, such as
when sending a packet that isn't IPv4, without specifying one of the
ways to encode the address family. The change in this commit is
entirely unrelated to that use of EAFNOSUPPORT.)
The specific problem was identified by Bill Green. Tested in NetBSD
11 context by Matti Andreas K=C3=A4h=C3=A4ri as reported on netbsd-users@.
Pull up following revision(s) (requested by gdt in ticket #564):
sys/net/if_tun.c: revision 1.179
sys/net/if_tun.c: Drop spurious EAFNOSUPPORT
When an outgoing packet is put in the queue, that might fail, and if
so the error will almost certainly be ENOBUFS. Let that error be
returned, rather than inexplicably forcing it to be EAFNOSUPPORT.
(EAFNOSUPPORT is returned nearby in the code for other cases, such as
when sending a packet that isn't IPv4, without specifying one of the
ways to encode the address family. The change in this commit is
entirely unrelated to that use of EAFNOSUPPORT.)
The specific problem was identified by Bill Green. Tested in NetBSD
11 context by Matti Andreas K=C3=A4h=C3=A4ri as reported on netbsd-users@.
Pull up following revision(s) (requested by riastradh in ticket #1406):
usr.sbin/postinstall/postinstall.in: revision 1.56
usr.sbin/postinstall/postinstall.in: revision 1.57
usr.sbin/postinstall/postinstall.in: revision 1.73
postinstall(8): Modify default certs.conf.
When manually configured /etc/openssl/certs is detected, just
uncomment the `#manual' line in the default certs.conf rather than
writing a new one. That way, you can switch to certctl-managed and
still get the default path by just deleting /etc/openssl/certs and
re-commenting the `manual' line.
postinstall(8): Use /usr/sbin/certctl.
Obviates need to have /usr/sbin in PATH when running this.
[7 lines not shown]
Pull up following revision(s) (requested by riastradh in ticket #563):
usr.sbin/postinstall/postinstall.in: revision 1.73
postinstall(8): Do obsolete _then_ opensslcertsrehash.
Otherwise any certificates removed by an update will remain in
/etc/openssl/certs/ca-certificates.crt, which is very bad!
PR bin/60932: postinstall obsolete + opensslcertsrehash interact badly
Import tzdata2026f from https://github.com/JodaOrg/global-tz/releases/download/2026fgtz/tzdata2026fgtz.tar.gz
Summary of changes in tzdata2026f (2026-10-27 23:16:49 -0700):
* Cuba will not fall back to -05 on 2026-11-01 and instead plans to
maintain -04 through 2027-11-07.
* In 1965 Brazil ended DST on 03-31 at 24:00 not 00:00, and in 1966
it ended DST on 03-01 at 01:00 not 00:00.
Support systems where the aux control block isn't decoded by the
ATA device (e.g. dedicated CompactFlash interfaces): if the device
tree doesn't have the second reg entry for the aux control block,
set WDC_CAPABILITY_NO_AUXCTL.
From Bob Green.
sys/net/if_tun.c: Drop spurious EAFNOSUPPORT
When an outgoing packet is put in the queue, that might fail, and if
so the error will almost certainly be ENOBUFS. Let that error be
returned, rather than inexplicably forcing it to be EAFNOSUPPORT.
(EAFNOSUPPORT is returned nearby in the code for other cases, such as
when sending a packet that isn't IPv4, without specifying one of the
ways to encode the address family. The change in this commit is
entirely unrelated to that use of EAFNOSUPPORT.)
The specific problem was identified by Bill Green. Tested in NetBSD
11 context by Matti Andreas Kähäri as reported on netbsd-users@.
Should be pulled up to 11 and 10.
CVS: ----------------------------------------------------------------------
CVS: CVSROOT cvs.NetBSD.org:/cvsroot
CVS: please use "PR category/123" to have the commitmsg appended to PR 123
[13 lines not shown]
PR lib/60885 - fix wscanf(3) conversion of multi-byte decimal point
This alters the way that the decimal point character in the locale
data is converted before processing float (all sizes) data fields.
This is a much simpler change to that for scanf(3), as the decimal
point (radix character) remains a single (wchar_t) character, and is
processed as such. All that is needed for wscanf() is to properly
convert the value from the locale data to a wchar_t.
No pullups planned, no NetBSD release contains any locales with
multi-byte LC_NUMERIC decimal points (or LC_MONETARY either, though
that would be irrelevant to scanf) so there is no point adding the
code to deal with something that doesn't exist there.
PR lib/60885 - fix scanf(3) handling of multi-byte decimal point
This alters the way that the decimal point character is scanned
when processing float (all sizes) data fields.
A new state is added to the state machine to handle moving through
the bytes that make up the decimal point - entered only when there
is more than one such byte.
No pullups planned, no NetBSD release contains any locales with
multi-byte LC_NUMERIC decimal points (or LC_MONETARY either, though
that would be irrelevant to scanf) so there is no point adding the
code to deal with something that doesn't exist there.
PR lib/60886 - fix weird printf(3) grouping for specific integer values
See the PR for the details - the bug is so bizarre, and impossible to
trigger with any real locale, that new ones needed to be invented to
expose it. This fixes that bug.
Since this issue cannot be observed with any real world locale, and
no locale in any released version of NetBSD has settings which would
allow the bug to manifest, no pullups for this are planned.
Fix PR kern/60919 (mysterious new RAID sets after a component is spared out)
Note that this fix will only prevent _new_ kernels with the fix from creating
these "rogue" sets from spared out components of mirrors. It will not ignore
nor delete any sets that may have been created in this way by an old kernel
without the fix; to fix that, the user should raidctl -A no the rogue set and
then delete it. This is left as it is now from an abundance of caution; being
less conservative could cause a lot of trouble for any users who may have,
unfortunately, initialized multiple RAID sets with the same serial number.
Other related fixes:
1. Sparing out a component at runtime could leak its vnode and also
disappear it from raidctl -s output if another spare was subsequently
added.
2. Make raidctl -r, -a, -R work with components that have been spared out.
Fixes some leaks and zeroes existing labels.
[10 lines not shown]
postinstall(8): Do obsolete _then_ opensslcertsrehash.
Otherwise any certificates removed by an update will remain in
/etc/openssl/certs/ca-certificates.crt, which is very bad!
PR bin/60932: postinstall obsolete + opensslcertsrehash interact badly
PR lib/60885 and PR lib/60886 - add test cases to demonstrate bugs
The changes in this test (amongst others, including adding tests of
the nb_NO.UTF-8 domain for PR bin/60496 which should already be fixed)
should expose the problems in printf(3) and scanf(3) noted in the PRs.
Note that we do not currently have tests for wscanf(3) which has a similar
problem to PR lib/60885, and consequently remains untested.
These new tests are currently expected to fail - but are not going to
be marked as expected fails, I want to see them actually fail.
Fixes for the issues already exist, and will be committed to the tree
soon after test failures have been observed. Similarly, a fix for the
wscanf(3) issue will be made at the same time (the bug is glaringly obvious,
and its fix is considerably simpler than that for scanf(3) - not that that
one was difficult) without that one ever having been observed - unless
the proverbial someone quickly implements a test case for that.
None of this is intended to be pulled up to any already released version
[11 lines not shown]
Add 3 new (bizarre) locales for ATF test use only.
These locales are inspired by issues reported in PR bin/60496 , PR lib/60677 ,
PR lib/60885 and PR lib/60886 (though while 60496 inspired all of this, it
does not need these locales to be reproduced).
Add the x1_X1 x1_X2 and X1_X3 locales (all .UTF-8) to be used to stress test
implementations with bizarre locale features that no real world locale would
ever use. Use of these locales for purposes other than implementation
testing is not intended, and getting weird results from any other use is to
be expected.
In the very near future, a test of the LC_NUMERIC categories of these locales
will appear (for 60496, 60885 & 60886) - that will depend upon the settings
of these locales' LC_NUMERIC parameters, so those values must not be altered
without also altering the tests.
In the not too distant future, a test of the LC_MONETARY categories of these
locales will also appear (for 60677) - and likewise will depend upon the
[37 lines not shown]
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
