BSD Commit Log Search

   net/libslirp: update to 4.9.5

   Addressed CVE-2026-95507 and CVE-2026-95508

   OK: sthen@

   add missing bdep on graphics/vulkan-headers

   databases/pgbouncer: security update to 1.26.0

   see https://www.pgbouncer.org/2026/09/pgbouncer-1-26-0, fixes:

   * CVE-2026-19888: DoS due to crash, triggerable by unauthenticated
     clients. Caused by a SCRAM client-final-message without a nonce.
   * CVE-2026-6668: DoS due to infinite loop, triggerable by unauthenticated
     clients. Caused by an integer overflow in the packet buffer growth
     logic.
   *  CVE-2026-6669: DoS due to unbounded work during login, triggerable by a
     malicious PostgreSQL server. Caused by an unbounded SCRAM iteration
     count.

   ok naddy@

   Bump version to 9.3

   process GuC responses while waiting

   Avoids GuC initialisation errors daniel@ saw on Alder Lake.
   from daniel@

   revert changing some busy waits to tsleep

   landry's ThinkPad T470s (Kaby Lake) with external HDMI monitor could
   no longer run X, stuck in a loop of:
   'modeset(0): hotplug event: connector 106's link-state is BAD'

   x509_verify: free current_chain on extension cache failure

   Plug the leak reported by CID 656293.

   ok tb

   fix builds with ccache, ok rsadowski naddy

   qt5/qtbase ports changes, ok rsadowski naddy

   - add getexecpath support to qt5/qtbase

   - fix builds with ccache; the way this was handled in qtbase could be
   simplified as it's already done by ports infrastructure, and the way
   it was done here previously isn't compatible with recent bsd.port.mk
   changes

   print/texlive: unbreak the minted latex package.

   Reported by espie@. OK sthen@.

   Thanks!

   On Qualcomm X2 SoCs we need to explicitly configure the DBI
   and ATU register windows in the PARF wrapper, otherwise they
   are silently ignored.

   This fixes a timeout in dwpcie_atu_config and makes the NVME
   attach on my HP Elitebook X G2q.

   ok kettenis@ patrick@

   add missing dep to fix random build error; ok naddy@

   Do not try to decode UTF-8 if it is too long to match the pattern.

   Do not allocate for repeat inside loop but allocate up front and just
   copy in the loop.

   Copy all #s at once when multiple are seen before [ rather than only
   copying one at a time which can be very slow.

   Check time every iteration for expensive repeat loop.

   qwx: fall back to world channels if the regulatory event fails

   If the first regulatory channel list event cannot be processed, qwx
   attaches with an empty channel list and cannot scan.  Populate the
   ath11k world channel set in that case.  Unlike qwz, keep channels an
   earlier event configured, and leave ieee80211_channel_init() to
   qwx_init() as before.

   From qwz 6f139a5bd.

   OK stsp@

   qwx: parse the regulatory ext channel list event

   Replace the stub with the 2 GHz/5 GHz part of ath11k's parser.  6 GHz
   rules, including any the firmware appends to the 5 GHz list, are
   ignored.  Firmware only sends this event if the host advertises
   WMI_CFG_HOST_SERVICE_FLAG_REG_CC_EXT, which qwx does not do yet.

   From qwz 901b347bb.

   OK stsp@

   qwx: fix scan BSSID by using the etheranyaddr of all zeros

   From qwz 6f139a5bd.

   OK stsp@

   in sigreturn if fpustate copyin fails, repair the potential partial copy.
   from daniel, ok guenther

   Update to a recent commit to unbreak several printing scenarii.

   Use mode screen for cursor not base which may be wrong during
   synchronized updates. GitHub issue 5648 from bugreporter23 at gmail dot
   com.

   Do not use subtraction for comparators, this doesn't really matter for
   IDs but it breaks key sorting. GitHub issue 5646 from Alexandre Fiori.

   Fix dragging a floating pane by its top border over a tiled pane's
   status line, GitHub issue 5645 from Michael Grant.

   Fix floating panes not being detected as obscuring the pane behind them,
   GitHub issue 5641 from Michael Grant.

   Exit 128 + SIGHUP like popups did if waiting for a pane that was killed,
   from Alexandre Fiori.

   sys/qwz: avoid waiting for recovery locks

   Backport of sys/dev/ic/qwx.c,v 1.129

   OK: stsp@

   sys/qwz: skip recovery from stopped tasks

   Backport of sys/dev/ic/qwx.c,v 1.124

   OK: stsp@

   sys/qwz: close the stop task scheduling window

   Backport of sys/dev/ic/qwx.c,v 1.109 by only moving ~IFF_RUNNING.

   OK: stsp@

   Add a second comparison for key bindings to avoid ties, GitHub issue
   5647 from Alexandre Fiori.