BSD Commit Log Search

   delete rport_input cos it's the same as p2p_input provided by the stack.

   typo in comment

   correctly check subsystem command is not the empty string (was
   repeatedly checking the subsystem name)
   spotted by Coverity (CID 898836)

   detect rport interfaces using the if_enqueue function pointer

   this is how rport decides if it should partner with another interface.
   at the moment it's using if_input point at rport_input to check,
   but i want to replace rport_input with p2p_input because they're
   largely identical.

   rport_start doesnt need to use ifnet refcnts cos it runs in softnet

   softnet threads get special treatment, basically. if they can get
   hold of a valid ifnet pointer they can hold it for as long as they
   want without accounting for it because the interface destroy code
   uses barriers to wait for softnets to finish before freeing them.

   regression test for bz3906: sshd crashing at connection time if the
   config lacks a subsystem directive but one is defined in a match
   block.

   don't crash at connection time if the main sshd_config lacks any
   subsystem directive but one is defined in a Match block

   bz#3906; ok dtucker

   check that invalid subsystem directives inside Match blocks are
   noticed at startup; bz#3906

   detect invalid sshd_config Subsystem directives inside Match
   blocks at startup rather than failing later at runtime;

   noticed via bz#3906; ok dtucker

   don't mess around with m->m_pkthdr.pf.inp and inp links in pf_test_state.

   pf_test_state copies the inp pointer from the mbuf, and then calls
   pf_mbuf_link_inpcb to set it again. nothing between these statements
   clears the mbuf inp pointer. this only works if the mbuf inp pointer
   is always NULL because pf_mbuf_link_inpcb asserts that the mbuf inp
   pointer is NULL before setting it.

   this relies on the inp pointer not being set for incoming packets,
   and the inp pointer getting cleared by pf_find_state for outgoing
   packets.

   at worst it's a hazard for the future, at best it's a waste of time.
   let's just get rid of it.

   ok sashan@

   move the tcp code in the pf_test pd.virtual_proto switch statement up

   there's a switch statement in pf_test() that handles pd.virtual_proto.
   however, the very first line of the code in the default case is
   "if (pd.virtual_proto == IPPROTO_TCP)", so we can hoist that up
   above the default case and then have it fall through.

   no functional change.

   ok sashan@

   new sentence, new line

   fix markup, .CM -> .Cm

   new sentence, new line

   update jupyter-notebook to 7.5.1

   drm/amdkfd: Use huge page size to check split svm range alignment

   From Xiaogang Chen
   c5bb128353e127da8beb07b0a8612f05caa559b9 in linux-6.12.y/6.12.63
   bf2084a7b1d75d093b6a79df4c10142d49fbaa0e in mainline linux

   drm/plane: Fix IS_ERR() vs NULL check in drm_plane_create_hotspot_properties()

   From Dan Carpenter
   3371a55632b8430a0740d95896d6af69269b32b3 in linux-6.12.y/6.12.63
   479acb9db3199cdb70e5478a6f633b5f20c7d8df in mainline linux

   drm/amd/display: Fix logical vs bitwise bug in get_embedded_panel_info_v2_1()

   From Dan Carpenter
   332825d72dcb3fcd57c9b7b06d902eda02932998 in linux-6.12.y/6.12.63
   1a79482699b4d1e43948d14f0c7193dc1dcad858 in mainline linux

   update spyder to 6.1.2

   Update to 19439 and jdk 17+. feedback and ok sthen@

   net/flare-messenger: add devel/protobuf to BDEP

   ok tb@

   Security update to exim 4.99.1 from maintainer

   1. Incomplete SQL injection fix - CVE-2025-26794's patch doesn't escape single quotes
   2. Heap buffer overflow - Unvalidated database field used as array bound (NEW)
   https://code.exim.org/exim/exim/src/commit/d46a6727798fc48d1756190a6d46d19216348c25/doc/doc-txt/exim-security-2025-12-09.1/report.txt

   Is it finally time to take this behind the barn?

   audio/opus: update to 1.6

   Major changes:
   * A new wideband-to-fullband bandwidth extension (BWE) module
   * Support for 96 kHz audio with Opus HD
   * Significant improvement to Deep Redundancy (DRED)
   * A new 24-bit encoder/decoder API

   Update to cargo-insta 1.45.0

   https://github.com/mitsuhiko/insta/releases/tag/1.45.0

   audio/libopusenc: Avoid __opus_check_* macros

   opus 1.6 removed internal macros starting with an __opus prefix.
   From upstream.

   Update MODJAVA_VER=11+ OK kirill@

   Update for Nextcloud 32.0.3

   OK landry@

   Update for Nextcloud 31.0.12

   OK landry@

   More assertions to track down the wired vs managed inconsistency.

   . a page should not be wired when being inserted to a LRU.
   . a page should not be on a LRU when incrementing its wired counter
   . a page should no longer be wired when returned to the per-CPU cache

   Only dequeue pages if they are known to be on a LRU.

   Allow for more sanity checks inside uvm_pagedequeue().