Pull up following revision(s) (requested by mrg in ticket #562):
external/mit/xorg/lib/brotli/brotli.mk: revision 1.2
external/mit/xorg/lib/brotli/brotli-rename.mk: revision 1.2
add new decoder sources and symbols for the renamer.
Reclaim empty segments in a separate thread, rather than doing so at
segment unlock. This allows segment reclaim to be done without the segment
lock. Should address PR 60820.
use the size of the buffer instead of NAME_MAX and strlcpy since there is
no format. fixes:
__builtin___snprintf_chk' specified bound 511 exceeds destination size 16
[-Werror=stringop-overflow=]
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.
