devel/cbang: databases/postgresql-plv8js: Chase v8 upgrade
V8 15.4 headers refuse to compile ("Non-CppGC MicrotaskQueue definition
is not supported") unless the embedder defines V8_CPPGC_MICROTASK_QUEUE,
matching how lang/v8 itself is built.
databases/postgresql-plv8js: pass the define in CXXFLAGS.
devel/cbang: define it in the v8 scons config before the v8.h header
check, which otherwise fails and silently disables V8 support, breaking
packaging. Also record it in cbang/config.h for consumers.
lang/v8: update to 15.4.80.20 (Chrome 154 stable)
- Update to 15.4.80.20 (Chrome 154.0.8037.97 stable pin)
- Update sub-dependency revisions from DEPS
- Add patch-.gn: use python3 from PATH instead of the hermetic
CPython CIPD package, which is not fetched; this also avoids a
host_cpu reference unsupported by devel/gn in the dotfile
- Add USES=llvm:min=20,build for proper LLVM toolchain selection
- Add 3 new sub-dependencies: cpu_features (chromium wrapper),
disarm, fadec
- Add ObtainCurrentThreadStackReservedLimit() for FreeBSD
- Remove upstreamed wasm-shuffle-reducer.cc patch
- Remove libcppgc.so and libthird_party_protobuf_protobuf_lite.so
(merged/removed upstream)
- Remove stale v8_use_libm_trig_functions GN arg (removed upstream)
- Specify explicit ALL_TARGET to avoid building test targets
- Regenerate all patches for new version
- Use USES=elfctl to mark d8, v8_shell, mksnapshot and mkgrokdump
as wxneeded: V8's JIT maps its code range read+write+execute and
[8 lines not shown]
www/remark42: stop baking REMARK_URL at build time
Since 1.17.0 the server fills the {% REMARK_URL %} placeholder into
the embedded frontend at serve time, using the REMARK_URL it was
started with. The port was still substituting the placeholder with
http://127.0.0.1:8080 before embedding, which left nothing for the
server to fill in, so the widget pointed at an unreachable host
whenever a page omitted remark_config.host.
- Drop the build-time substitution and the REMARK_URL make variable;
remark_url in rc.conf now reaches the frontend as well.
- Document remark_remark_web_root in the rc script.
www/redmine61: update to 6.1.5
Security and bug fix release: fixes two private data leaks via
the REST API.
Upstream now pins Rails 7.2.4; keep the dependency floor at
7.2.3.2, the version currently in the tree.
www/redmine70: update to 7.0.2
Security and bug fix release: fixes two private data leaks via
the REST API and a DOM-based XSS via clipboard HTML paste.
Upstream now pins Rails 8.1.4; keep the dependency floor at
8.1.3.1, the version currently in the tree.
MFC: fsck_msdosfs: fix head bitmap over-allocation on 32-bit platforms
bitmap_ctor() computed the allocation size as
roundup2(bits, LONG_BIT) / (LONG_BIT / 8)
The dividend is a count of bits, so converting it to bytes requires
dividing by 8 (bits per byte), not by LONG_BIT / 8 (bytes per long).
The two divisors happen to coincide on LP64, but on ILP32 platforms
the head bitmap was allocated at twice the required size; for a
FAT32 file system with close to 2^28 clusters, that is 64 MiB instead
of 32 MiB.
The extra half of the allocation was never accessed, so there is no
functional change other than the reduced memory footprint.
Pull Request: https://github.com/freebsd/freebsd-src/pull/2440
(cherry picked from commit 269002da453895fc935101afefeb02d2c7d10360)
MFC: fsck_msdosfs: fix head bitmap over-allocation on 32-bit platforms
bitmap_ctor() computed the allocation size as
roundup2(bits, LONG_BIT) / (LONG_BIT / 8)
The dividend is a count of bits, so converting it to bytes requires
dividing by 8 (bits per byte), not by LONG_BIT / 8 (bytes per long).
The two divisors happen to coincide on LP64, but on ILP32 platforms
the head bitmap was allocated at twice the required size; for a
FAT32 file system with close to 2^28 clusters, that is 64 MiB instead
of 32 MiB.
The extra half of the allocation was never accessed, so there is no
functional change other than the reduced memory footprint.
Pull Request: https://github.com/freebsd/freebsd-src/pull/2440
(cherry picked from commit 269002da453895fc935101afefeb02d2c7d10360)
www/caddy-custom: Fix build with BUILD_AS_NON_ROOT=yes
Two fixes:
1. Establish a GOCACHE and GOPATH for BUILD_AS_NON_ROOT so it doesn't
try to use /nonexistant/go (patch by Brad Ackerman).
2. Reset PORTREVISION (Vick Khera). I added a note in caddy/Makefile
so maybe I'll remember in the future.
PR: 295182
www/caddy: Update to 2.11.6
Changes: https://github.com/caddyserver/caddy/releases/tag/v2.11.6
There are some breaking changes in this release, so you should read
that list. They are primarily security-tightening changes and should
be invisible to most users. If you were relying on uploading files in
headers rather than the body, you'll want to make some changes.
sendfile: keep the lowat hack more conservative
First, record the original lowat value and later in the wait/EAGAIN loop
look at the original value, rather than on the value that we just faked.
This eliminates some blank syscalls, where socket was reported as writable
and immediate write instantly fails. In my testing the ratio of such
syscalls was really small, under 2%, however in a different scenario this
negative effect can be more profound.
Second, cap the lowat growth to 1/2 of original socket buffer size, rather
than to current size. The problem was there before, but it became more
profound after 587c6c121504.
Note: we are considering to evaluate if the lowat hack is needed at all.
Reviewed by: tuexen
Differential Revision: https://reviews.freebsd.org/D60106
tcp: use SB_AUTOSIZE flag to tell if socket buffer was set
The check against V_tcp_sendspace is not a correct one, as a buffer may
grow larger than the initial value. The conjunction was always false up
until 587c6c121504, and only after it the bug surfaced.
If we already grow our buffer past the value stored in the hostcache,
prefer our value.
Reviewed by: tuexen
Differential Revision: https://reviews.freebsd.org/D60105
