commit 3e349ef42f
Author: Gerald Combs <gerald@wireshark.org>
Date:   Tue Nov 18 10:04:23 2025 -0800

    Transifex: Set our resource names in the config file

commit 4288052baa
Author: Gerald Combs <gerald@wireshark.org>
Date:   Mon Nov 17 13:28:04 2025 -0800

    Prep for 4.6.1

commit 2a744048e9
Author: Stig Bjørlykke <stig@bjorlykke.org>
Date:   Mon Nov 17 21:32:10 2025 +0000

    sgp32: Register CoAP URI-Path

    Register CoAP URI-Path "/gsma/rsp2/asn1".

    (cherry picked from commit e885b5b29d6ba4af5eabf545ade37835765ed361)

    Co-authored-by: Stig Bjørlykke <stig@bjorlykke.org>

commit 6186a8e647
Author: Stig Bjørlykke <stig@bjorlykke.org>
Date:   Mon Nov 17 19:20:22 2025 +0000

    coap: Add URI-Path dissector table

    Split Uri-Host and Uri-Path into separate strings. Add a dissector
    table using URI-Path. Some protocols are using the Uri-Path instead
    of Content-Format to specify the payload type.

    Use Message ID as transaction key if Token is not present.

    (cherry picked from commit 1867d2e6b600298c97b05e03ce8b13501fffc7fd)

    Co-authored-by: Stig Bjørlykke <stig@bjorlykke.org>

commit 45034f9a70
Author: Gerald Combs <gerald@wireshark.org>
Date:   Sun Nov 16 10:18:39 2025 +0000

    [Automatic update for 2025-11-16]

    Update manuf, services enterprise numbers, translations, and other items.

commit 38df909ca8
Author: John Thacker <johnthacker@gmail.com>
Date:   Sat Nov 15 23:52:53 2025 +0000

    epan: Use a GSequence for tvb_composite and make it non-recursive

    To speed up tvb_composites consisting of many child tvbuffs,
    use a GSequence. This is a data structure that internally has
    a balanced binary tree. It differs from GTree in that one can
    do initial insertions without sorting. In our case, the ordering
    at the time of tvb_composite_finalize is *the* ordering, and is
    the same ordering that comparing according to the calculated
    end offsets will have, so this is faster on insert than a GTree.

    Like a GTree, it is much faster O(log N) vs O(N) on lookup to find
    the location of the TVB with end offset greater than or equal to
    the desired offset than using a GQueue.

    Also, make composite_memcpy iterative instead of recursive, which
    removes the current hard limit of 500 without causing stack overflow
    with very large numbers of child TVBs.

    The file in #20860 works fine now but there could still be limits
    imposed (at least on the total number of bytes?) for a compression
    bomb more similar to the file found in #12077.

    (cherry picked from commit 9139917bd8e2c80a5db7079993d5528db74e3519)

    Co-authored-by: John Thacker <johnthacker@gmail.com>

commit 6f1411406a
Author: John Thacker <johnthacker@gmail.com>
Date:   Sat Nov 15 19:12:51 2025 +0000

    HTTP3: Fix reporting error codes

    NGHTTP3 error codes are negative. Either store them in a signed integer
    (a int16_t is large enough for all actually used values) or, if storing
    them in an unsigned integer, take the unary negative when storing and
    then using them. It seems easier to store them in a signed integer.

    Ping #20860

    (cherry picked from commit 8ca207b84c0e309d61c021f32a089db4683e1f1d)

    Co-authored-by: John Thacker <johnthacker@gmail.com>

commit 68228be0b6
Author: John Thacker <johnthacker@gmail.com>
Date:   Sat Nov 15 17:41:40 2025 +0000

    HTTP3: Fix individual compressed header length

    The compressed header offset is the number of bytes advanced in
    the input compressed buffer, not the total number of bytes in
    the compressed buffer (which doesn't seem to be correctly updated.)

    This fixes highlighting the compressed header bytes when selecting
    it in the proto tree.

    (cherry picked from commit de623560eb60ae80610e58d58cbe3377ca24a8b0)

    Co-authored-by: John Thacker <johnthacker@gmail.com>

commit ac6578bd80
Author: Stig Bjørlykke <stig@bjorlykke.org>
Date:   Fri Nov 14 22:29:38 2025 +0000

    NAS EPS: do not append "Ciphered message" to info column if deciphered

    (cherry picked from commit cf03d3860d1c849952b68b666ca33264b46ed8aa)

    Co-authored-by: Pascal Quantin <pascal@wireshark.org>

commit 73257a759b
Author: Jaap Keuter <jaap.keuter@xs4all.nl>
Date:   Fri Nov 14 13:49:56 2025 +0000

    DTLS: Add recent DTLS-SRTP Protection Profiles

    (cherry picked from commit 0bc6580f9ec4677221374f22cb73ffd831ed1bb8)

    Co-authored-by: Jaap Keuter <jaap.keuter@xs4all.nl>

commit a6fed16bc2
Author: Jaap Keuter <jaap.keuter@xs4all.nl>
Date:   Fri Nov 14 13:50:38 2025 +0000

    DNS: disable experimental DNS over DTLS conflicting with DNS over QUIC

    Allow DNS over QUIC preference over DNS over DTLS, while maintaining
    the option for using "Decode as" for DNS over DTLS.

    Fixes #20856

    (cherry picked from commit 4b948df23102d648ad315d9aeaf335cc6126c3fa)

    Co-authored-by: Jaap Keuter <jaap.keuter@xs4all.nl>

commit 641855c2a4
Author: Anders Broman <a.broman58@gmail.com>
Date:   Thu Nov 13 09:38:58 2025 +0100

    asn2wrs: Respect VAL_PTR for BER UTCTime

    Two dissectors set the VAL_PTR directive for UTCTime, but asn2wrs.py
    ignores it so the intended dissection has had no effect. Implement it.
    Instead of returning the tvb length, return the string, since the
    dissectors want the string.

    Instead of not adding the string to the tree if the string is retrieved,
    only don't add the string if the passed in HF_INDEX is -1, so that
    dissectors can add the string to the tree normally but also retrieve
    the string for, e.g., adding to the INFO column.

    Use the default body with directives for X.509, replacing the use
    of tvblen with saving the old offset.

    (cherry picked from commit 461566a97aa23973854956923b1ac6e64ac1c10d)

    Co-authored-by: John Thacker <johnthacker@gmail.com>

commit 7d45c8dbec
Author: Anders Broman <a.broman58@gmail.com>
Date:   Thu Nov 13 09:33:09 2025 +0100

    BT L2CAP: Don't add incomplete reassemblies

    If mfp->cur_off != mfp->tot_len there was missing data or other
    reassembly issues.

    Fix #20852

    (cherry picked from commit 2497f40704a3c26b0fcd9282a3b39f55ec84c78b)

    Co-authored-by: John Thacker <johnthacker@gmail.com>

commit 4faa767b16
Author: Anders Broman <a.broman58@gmail.com>
Date:   Thu Nov 13 09:31:55 2025 +0100

    SNMP: Restore BulkPDU request-id abbrev

    Remove the FIELD_RENAME from the BulkPDU version of request-id
    so that it has the same field abbreviation as the one in PDU.

    Fix #20849

    (cherry picked from commit 33b154e8cb701223250cc3969ea87d371edf3ab3)

    Co-authored-by: John Thacker <johnthacker@gmail.com>

commit f6136f4121
Author: Pascal Quantin <pascal@wireshark.org>
Date:   Wed Nov 12 19:34:24 2025 +0100

    ACDR: Fix crash when filtering acdr && !rtp

    (cherry picked from commit 66f50b4ff1c31191f8fcc34637a0973a95b97c9c)

    Co-authored-by: Orgad Shaneh <orgad.shaneh@audiocodes.com>

commit 2e897e10ac
Author: John Thacker <johnthacker@gmail.com>
Date:   Wed Nov 12 14:54:27 2025 +0000

    BT L2CAP: Fix reassembly of I-Frames

    For the first fragment, [try to] copy only what is in the fragment,
    not the entire reassembled SDU length, which will fail. (It will
    fail on only the first pass, creating a mysterious expert info that
    doesn't show up in the Expert Info dialog.)

    Fixes reassembly of the I-frame SDUs beginning in frames 76 and 79
    on the capture file in issue #2241; they are ultimately reassembled
    into one OBEX packet in frame 81 which is correctly dissected now.

    Fixup 7950fc1c941b92beb23fd29f7fed90eac9ba8c89

    (cherry picked from commit 36bd1936adde3696cd168ef80e3da9bedecf01b2)

    Co-authored-by: John Thacker <johnthacker@gmail.com>

commit 2da8eababc
Author: Alexis La Goutte <alexis.lagoutte@gmail.com>
Date:   Wed Nov 12 07:35:05 2025 +0100

    LTP: Fix SDA offset for each sub-ID

    (cherry picked from commit a12019300764c07e629db818a983103216583365)

    Co-authored-by: Brian Sipos <brian.sipos@gmail.com>

commit e452c5fb88
Author: John Thacker <johnthacker@gmail.com>
Date:   Sun Nov 9 12:37:21 2025 +0000

    ISOBUS VT: Fix UTF-16 endianness

    If the first two bytes of a string buffer, retrieved in Little Endian
    order, are 0xFEFF (BOM), then the string is UCS-2 or UTF-16 in Little
    Endian order, not Big Endian.

    FIx #20845

    (cherry picked from commit 5b77cfb3392345b0491eae0dba25cc2416918ca3)

    Co-authored-by: John Thacker <johnthacker@gmail.com>

commit bdf14ffa53
Author: Alexis La Goutte <alexis.lagoutte@gmail.com>
Date:   Sun Nov 9 12:45:33 2025 +0100

    Fix Ethernet CFM BNM PDU dissection

    (cherry picked from commit 48c1203fd01c62628dcb2853e96cff131f6f477a)

    Co-authored-by: mikkle <mikkel@troest.dk>

commit e2120612d7
Author: Gerald Combs <gerald@wireshark.org>
Date:   Sun Nov 9 10:18:09 2025 +0000

    [Automatic update for 2025-11-09]

    Update manuf, services enterprise numbers, translations, and other items.

commit 8883af2d02
Author: John Thacker <johnthacker@gmail.com>
Date:   Fri Nov 7 15:49:36 2025 +0000

    peektagged: Increase maximum section size

    Fix #20842

    (cherry picked from commit b50cafe59e765a637c565956f57bbd8d9c9a5fae)

    Co-authored-by: John Thacker <johnthacker@gmail.com>

commit 7f6519be81
Author: Joakim Karlsson <oakimk@gmail.com>
Date:   Wed Nov 5 13:43:47 2025 +0000

    CMake: Update brotli to 1.2.0

    (cherry picked from commit cd9cf8f537875fa5e19be9b2960308e72c5fb762)

    1931cc75 CMake: Update brotli to 1.2.0
    425f273f Edit FetchArtifacts.cmake
    8f2f2f7d Edit FetchArtifacts.cmake

    Co-authored-by: Gerald Combs <gerald@zing.org>

commit 183de4149b
Author: John Thacker <johnthacker@gmail.com>
Date:   Mon Nov 3 21:32:35 2025 +0000

    EAPOL MKA dissector: Fix SAK unwrapping length

    (cherry picked from commit af29e6c2f95814d50367f314096bc399d6643dce)

    Co-authored-by: Karel Hevessy <karel.hevessy@machsystems.cz>

commit 8be7847dc9
Author: John Thacker <johnthacker@gmail.com>
Date:   Mon Nov 3 19:12:12 2025 +0000

    MKA: Use proto_data, not pinfo->private_table

    pinfo->private_table can be used by (and is intended for) Lua
    dissectors, and destroying the GHashTable it points to unconditionally
    at the end of MKA dissection can possibly interfere with Lua dissectors
    or anything else that has put something in pinfo->private_table.

    Use pinfo->pool scoped proto data to achieve the same thing, which also
    eliminates having to worry about feeing the hash table if there is an
    exception while dissecting.

    See also #17156.

    (cherry picked from commit bdeaabea725b0824a04d86edd9326bff9b327f6a)

    Co-authored-by: John Thacker <johnthacker@gmail.com>

commit ff0a5da6e2
Author: John Thacker <johnthacker@gmail.com>
Date:   Sun Nov 2 21:51:58 2025 +0000

    ZigBee Direct: Don't use a packed struct

    zbee_sec_ccm_decrypt() expects the nonce to be a array of
    uint8_t of size ZBEE_SEC_CONST_NONCE_LEN. Just use that, as
    packet-zbee-security and packet-zbee-nwk-gp do, instead of
    creating a packed struct whose members are never really used.
    (Among other things, the "frame_counter" member of that struct
    is unnecessarily a uint32_t, as it needs to be copied in the
    same little-endian order as in the back, and thus trying to
    access the struct member as a uint32_t would require byte swapping
    on Big-Endian systems anyway.)

    The same dissection works on the sample files attached to !12052.

    Fix #20776

    (cherry picked from commit 81f95ab1fd2ef70ea82d7bcd781d044de5adb09e)

    Co-authored-by: John Thacker <johnthacker@gmail.com>

commit c63d8859c5
Author: John Thacker <johnthacker@gmail.com>
Date:   Sun Nov 2 21:53:38 2025 +0000

    VLAN: Restore the dissection depth when leaving the dissector

    The VLAN dissection depth should only count nested tags, it doesn't
    need to keep increasing when a frame contains multiple VLAN tags that
    are not nested, whether from "View as File Format", or from protocols
    that can contain multiple Ethernet frames in a large frame, like
    TECMP.

    Remove the special workaround for TECMP (#16685) as it's not needed
    anymore. That's a little cleaner because the TECMP dissector doesn't
    have to know about the VLAN dissector.

    Fix #20831

    (cherry picked from commit 4a9370c276312f82905f8affd2289ae49c6185b3)

    Co-authored-by: John Thacker <johnthacker@gmail.com>

commit ed7a2fdbfc
Author: Gerald Combs <gerald@wireshark.org>
Date:   Sun Nov 2 10:19:12 2025 +0000

    [Automatic update for 2025-11-02]

    Update manuf, services enterprise numbers, translations, and other items.

commit df4895c707
Author: John Thacker <johnthacker@gmail.com>
Date:   Thu Oct 30 14:13:29 2025 +0000

    packaging: Windows version checks

    While Qt has officially only supported Windows 10 1809 or later since
    Qt 6.2[1], in practice our releases with Qt 6.5 have run on Windows
    10 version 1607 and Windows Server 2016 (but not Windows 10 1507[2]).
    However, Qt 6.8 and later doesn't run on those versions, as qt6core.dll
    calls SetThreadDescription, which isn't available on Windows 10 version 1607
    and Windows Server 2016 without Run Time Dynamic Linking that Qt doesn't
    do by default.[3,4] It is possible to compile Qt 6.9.1 and later with
    a newly added flag to re-enable Windows Server 2016 (and Windows 10
    version 1607) support, but I'm not sure we want to do that even on 4.6;
    certainly not for the 5.0 release.[5]

    Because some of the Windows version variables were frozen with the
    transition to Windows as a Service, and with the difference in the
    Win32 API available at various build versions (all called Windows
    10 and with the same VersionNT[6]), we have to check the build number
    instead.

    Update NSIS, WIX, and various other places for the minimum of
    Windows 10 1809, aka Redstone 5 aka build number 17763. (Which
    also corresponds to Windows Server 2019.)

    [1] - https://doc.qt.io/archives/qt-6.2/supported-platforms.html
    [2] - https://ask.wireshark.org/question/37410/qt6-entrypoint-could-not-be-found/
    [3] - https://ask.wireshark.org/question/37552/wireshark-460-qt6coredll-setthreaddescription-error-on-server-2016/
    [4] - https://learn.microsoft.com/en-us/windows/win32/api/processthreadsapi/nf-processthreadsapi-setthreaddescription
    [5] - https://bugreports.qt.io/browse/QTBUG-134075
    [6] - https://learn.microsoft.com/en-us/troubleshoot/windows-client/application-management/versionnt-value-for-windows-10-server

    (cherry picked from commit 1c7ffb59daa8917940cf35f79563af7a38346f32)

    Co-authored-by: John Thacker <johnthacker@gmail.com>

commit 6d1b145f07
Author: Anders Broman <a.broman58@gmail.com>
Date:   Thu Oct 30 09:09:50 2025 +0100

    wsutil: Define JSMN_PARENT_LINKS to speed up JSON parsing

    (cherry picked from commit e7dc16780267b19cbd572f5e0ec2ca3b38c3d38f)

    Co-authored-by: Moshe Kaplan <mosheekaplan@gmail.com>

commit 0dc131b802
Author: Darius Davis <darius-wireshark@free-range.com.au>
Date:   Wed Oct 29 20:03:39 2025 +1000

    LZ4: Ensure "lz4_prefs" is configured before use.

    lz4wfile_fdopen was calling into LZ4F_compressBound -- which used the contents
    of state->lz4_prefs -- before that structure had been initialized.  Move it
    down to after state->lz4_prefs has been set up.

    Fixes #20779.

    (Hopefully.)

    (cherry picked from commit e8aa4ba3cb0a1cc2218e780ee1261b5d9921d1a4)

    (Manually resolved code movement from wiretap/file_wrappers.c to
    wsutil/file_compressed.c.)

commit 50ebaf05b3
Author: Stig Bjørlykke <stig@bjorlykke.org>
Date:   Tue Oct 28 08:49:01 2025 +0000

    sgp32: Fix a CMake path to cnf file

    (cherry picked from commit 288600ad3fbf5753976d4805e5b6dd41d5887100)

    Co-authored-by: Stig Bjørlykke <stig@bjorlykke.org>

commit 82808d434b
Author: Stig Bjørlykke <stig@bjorlykke.org>
Date:   Mon Oct 27 16:16:24 2025 +0000

    nmea0183: Fix entry for lat/lon [empty]

    The latitude and longitude entries are FT_NONE and does not add
    a trailing ": ".

    (cherry picked from commit 12e4744b3b59fae72bda5be49664aff1e6c227ef)

    Co-authored-by: Stig Bjørlykke <stig@bjorlykke.org>

commit 6c30f55070
Author: Anders Broman <a.broman58@gmail.com>
Date:   Sun Oct 26 21:40:15 2025 +0100

    PTPV2: Malformed Packet error when parsing IEEE_C37_238 TLV

    (cherry picked from commit 8df832ec564ce66c61f4223bc6300b7ff0b9e339)

    b31a2b26 PTPV2: Malformed Packet error when parsing IEEE_C37_238 TLV

    Co-authored-by: Prashant Tripathi <prashant_tripathi@selinc.com>

commit 0f3ff92bf9
Author: John Thacker <johnthacker@gmail.com>
Date:   Sun Oct 26 14:07:46 2025 +0000

    WINREG: Align winreg_String to 5 bytes

    Judging from the file in #20813, winreg_String is aligned to
    64 bit on NDR64, update the conformance file implementation.
    The files in #3709 still work.

    Fix #20813

    (cherry picked from commit ed94b447a2e03d33ecea2b76ea6f35df4c1236e2)

    Co-authored-by: John Thacker <johnthacker@gmail.com>

commit 711e4b5222
Author: Gerald Combs <gerald@wireshark.org>
Date:   Sun Oct 26 10:18:11 2025 +0000

    [Automatic update for 2025-10-26]

    Update manuf, services enterprise numbers, translations, and other items.

commit 158ed1e98b
Author: Darius Davis <darius-wireshark@free-range.com.au>
Date:   Sat Oct 25 15:01:34 2025 +1000

    Kafka: Fix decompress_snappy with no xerial chunks.

    Instead of returning true without setting outputs, report a failure to
    decompress and return false to the caller.

    Fix #20823

    (cherry picked from commit 49137f8ce93c9f7ac55b69c8e089ba6a422f633e)

commit c10b365a00
Author: John Thacker <johnthacker@gmail.com>
Date:   Sat Oct 25 20:42:07 2025 +0000

    pidl: Check for undefined SwitchType first

    If the switch type is not defined (e.g. in a union with no
    discriminant) check for that first instead of doing a bunch of
    comparisons with an uninitialized scalar variable. Doesn't change
    the dissector results, but prevents a bunch of Perl warnings:

    Use of uninitialized value $t in hash element at /wireshark/tools/pidl/lib/Parse/Pidl/Typelist.pm line 194.
    Use of uninitialized value in string eq at /wireshark/tools/pidl/lib/Parse/Pidl/Wireshark/NDR.pm line 480.
    Use of uninitialized value in string eq at /wireshark/tools/pidl/lib/Parse/Pidl/Wireshark/NDR.pm line 482.
    Use of uninitialized value in string eq at /wireshark/tools/pidl/lib/Parse/Pidl/Wireshark/NDR.pm line 484.

    (cherry picked from commit edc98c1b537cbd7b8fb35cdb8b61b7dd000a1544)

    Co-authored-by: John Thacker <johnthacker@gmail.com>

commit 1dbf6f0bb3
Author: John Thacker <johnthacker@gmail.com>
Date:   Sat Oct 25 20:02:22 2025 +0000

    winreg: Fix dissector param / string prettification

    Three of the strings are at a different element level, so the PARAM
    directive was not being applied. Fix that, and eliminate some warnings:

    winreg.cnf:111: warning: dissector param never used
    winreg.cnf:112: warning: dissector param never used
    winreg.cnf:109: warning: dissector param never used

    (cherry picked from commit 49d0c1445f6eac85f6a9456a0fd24170d182a03c)

    Co-authored-by: John Thacker <johnthacker@gmail.com>

commit 3a21f2941a
Author: John Thacker <johnthacker@gmail.com>
Date:   Sat Oct 25 19:48:35 2025 +0000

    pidl: Warn appropriately on unsupported switch_is discriminants

    Microsoft's MIDL supports switch_is discriminants which have limited
    C-language expressions including conditionals, logical, relational,
    and arithmetic expressions. [1] Some of the distributed IDL files include
    such expressions. The current handling in PIDL's Wireshark NDR.pm only
    supports a single identifier (possibly a pointer), which appears to be
    the way it is defined in the original DCE IDL. [2,3]

    In addition, the switch_is discriminant may simply be misspelled or
    otherwise not found when parsing, leading to an empty array reference.
    Test for the empty array in order to produce a helpful warning message
    with the filename and line number, instead of unhelpful Perl warnings.

    This does not provide support for the construct not affect the generated
    dissectors at all, only reduces 4 unclear warnings into a single useful
    warning.

    Before:
    Use of uninitialized value $name in string ne at /home/johnthacker/wireshark/tools/pidl/lib/Parse/Pidl/Wireshark/NDR.pm line 516.
    Use of uninitialized value in string eq at /home/johnthacker/wireshark/tools/pidl/lib/Parse/Pidl/Wireshark/NDR.pm line 525.
    Use of uninitialized value $name in string ne at /home/johnthacker/wireshark/tools/pidl/lib/Parse/Pidl/Wireshark/NDR.pm line 527.
    Use of uninitialized value $name in string ne at /home/johnthacker/wireshark/tools/pidl/lib/Parse/Pidl/Wireshark/NDR.pm line 531.

    After:
    drsuapi.idl:828: warning: ctr switch_is discriminant `level|(type<<16)' not found. (Only single identifiers are supported, not expressions as in MIDL.)

    [1] - https://learn.microsoft.com/en-us/windows/win32/midl/switch-is
    [2] - https://pubs.opengroup.org/onlinepubs/9629399/chap4.htm#tagcjh_08_02_12
    [3] - https://pubs.opengroup.org/onlinepubs/9629399/chap4.htm#tagcjh_08_04

    (cherry picked from commit 448a0d6a239f3f274d3a3a3a39ea9c6ce7fd2611)

    Co-authored-by: John Thacker <johnthacker@gmail.com>

commit da48c1aeaa
Author: John Thacker <johnthacker@gmail.com>
Date:   Sat Oct 25 13:34:51 2025 +0000

    pidl: Don't warn about params that are used

    The "USED" indicator is never set for params, which results in many
    bogus "warning: dissector param never used" messages when compiling
    the pidl dissectors. Set it when used, to reduce the number of messages.

    (cherry picked from commit 919f15ab2ea313d69cc2578426430f22f65a51b5)

    Co-authored-by: John Thacker <johnthacker@gmail.com>

commit 92941818c5
Author: Jaap Keuter <jaap.keuter@xs4all.nl>
Date:   Sat Oct 25 09:16:00 2025 +0000

    EPAN: Have proto_tree_add_bits functions set correct FI bit offset

    Proto_tree_add_bits functions have their offset in bits, rather than
    bytes. During tree item creation a field_info is created with start at
    the first byte of the bit field. Therefore the bit offset to set in the
    field info should be the remaining bits in the byte, rather than the
    full bit_offset. Therefore pass bit_offset mod 8 to FI_BITS_OFFSET.

    Followup on b0d9cba7
    Fixes #20820

    (cherry picked from commit 4748d8c21ebcd20431a03973981ae16873a08ac1)

    Co-authored-by: Jaap Keuter <jaap.keuter@xs4all.nl>

commit 12309f1860
Author: John Thacker <johnthacker@gmail.com>
Date:   Fri Oct 24 15:46:47 2025 +0000

    Qt: Don't make the DataSourceTab visible if it wasn't

    When making sure that the DataSourceTab is not visible in order
    to add new tabs, restore it to its previous hidden state instead
    of always making it visible.

    Fixup 2a09c1a676ee10ae6bcb9a9beab749b5e15a5636
    Fix #20819

    (cherry picked from commit 515e5fe288207883fed5e49a9683510fcfca6673)

    Co-authored-by: John Thacker <johnthacker@gmail.com>

commit 4b4f9dec24
Author: Eugène Adell <eugene.adell@gmail.com>
Date:   Thu Oct 23 17:50:22 2025 +0200

    TCP: Fix the Flow Counter statistics

commit 986a0ef8a4
Author: John Thacker <johnthacker@gmail.com>
Date:   Thu Oct 23 14:14:58 2025 +0000

    pidl: Use has_property to avoid comparing undef with the empty string

    Prevents a bunch of (harmless) warnings when compiling:

    Use of uninitialized value in string eq at /builds/wireshark/wireshark/tools/pidl/lib/Parse/Pidl/Wireshark/NDR.pm line 521.

    (cherry picked from commit 8dee8601776175a6132bef7ab19f36ee3d589d06)

    Co-authored-by: John Thacker <johnthacker@gmail.com>

commit e34318bb5d
Author: John Thacker <johnthacker@gmail.com>
Date:   Thu Oct 23 14:04:57 2025 +0000

    Netlogon: add dissection of dsraddresstositenames[ex]w

    Signed-off-by: Ronnie Sahlberg <ronniesahlberg@gmail.com>

    (cherry picked from commit 653c35ddaa357d29b890c5c36bd7cd50d3384408)

    Co-authored-by: Ronnie Sahlberg <ronniesahlberg@gmail.com>

commit 4b8c27acc3
Author: John Thacker <johnthacker@gmail.com>
Date:   Thu Oct 23 13:31:59 2025 +0000

    epan: Generate the filter for a FT_NONE frame bytes fallback better

    Use the internal ftypes function bytes_to_dfilter_repr to generate
    the dfilter representation of the frame bytes fallback for a
    FT_NONE field instead of re-implementing it. This eliminates a
    snprintf call, and fixes the case where the byte string is exactly
    length 1 (see 881dec9b90aec4e4c4b7ad25e86ee6b61eac4913 and
    f37c7c4062db513bd4dc6fffa36a0c90c74a4339) by making it consistent
    with other dfilter uses and automatically update if FT_BYTES has
    any future changes.

    Fix #20818. (The filter documentation still needs to be updated,
    though.)

    (cherry picked from commit 2215b6862a10c1c3d06c75b0f403b3df9438bd3f)

    Co-authored-by: John Thacker <johnthacker@gmail.com>

commit 4636bdbc7e
Author: John Thacker <johnthacker@gmail.com>
Date:   Thu Oct 23 07:51:26 2025 -0400

    GitLab CI: Fix FALCO_PLUGIN_DIR on Windows for 4.6

    Fixup the backport e46dbbe5cf8b58d899d38453f86f1e712676b960
    by appending the version to get the correct path.

commit e46dbbe5cf
Author: John Thacker <johnthacker@gmail.com>
Date:   Wed Oct 22 09:12:24 2025 -0400

    GitLab CI: Set FALCO_PLUGIN_DIR relative to WIRESHARK_BASE_DIR on Win

    WIRESHARK_BASE_DIR determines _PROJECT_LIB_DIR / ARTIFACTS_DIR.

    Fixup 18568a92d63d0507dfac2c6c64eec161baecd651

    (backported from commit b50c773a9bd9bbdffc5d67c1a2698825ef113290)

commit 78ff334d37
Author: Gerald Combs <gerald@wireshark.org>
Date:   Wed Sep 3 13:19:55 2025 -0700

    GitLab CI+packaging: Reenable Falco plugins on Windows

    Reenable the gcpaudit and k8saudit plugins. Make the plugin path a
    variable.

    (backported from commit 62fb70b7403b9418a95a0465a59016454aad36b8
    and b50c773a9bd9bbdffc5d67c1a2698825ef113290)

commit f690ad48b7
Author: John Thacker <johnthacker@gmail.com>
Date:   Wed Oct 22 12:57:35 2025 +0000

    DFS: Fix VolumeState bitmap

    pidl, or at least the Wireshark module, doesn't handle a bitmap using
    values from a enum defined elsewhere in the IDL. It's been incorrectly
    interpreted "DFS_..." as the hexstring "DF" and then bailing out with
    errors at the "S":
    Illegal hexadecimal digit 'S' ignored at /builds/wireshark/wireshark/tools/pidl/lib/Parse/Pidl/Wireshark/NDR.pm line 253.

    Fix that by using the actual value.

    (cherry picked from commit ba0da02a1c30dd7a13414a494f411fd419489272)

    Co-authored-by: John Thacker <johnthacker@gmail.com>

commit 98758039eb
Author: Stig Bjørlykke <stig@bjorlykke.org>
Date:   Wed Oct 22 10:07:37 2025 +0000

    wslua: Setup packet record in FileHandler

    Call wtap_setup_packet_rec() in FileHandler:read_packet() and
    FileHandler:seek_read() to avoid a crash because of missing
    rec_type_name and not set packet encap. This happens when Lua
    does not call FrameInfo:setup_packet_rec(encap).

    Fixes #20817

    (cherry picked from commit 498570c8b47dc990451f8497d48d4dd01144efc1)

    Co-authored-by: Stig Bjørlykke <stig@bjorlykke.org>

commit 0f8d5bdc6d
Author: John Thacker <johnthacker@gmail.com>
Date:   Tue Oct 21 22:02:46 2025 +0000

    GitLab CI: Fix the Debian Stable APT Test job

    Make sure we can install dependencies.

    (cherry picked from commit b2759ef7a13dcbe93cc123f6b6ce9c121118a2bd)

    Co-authored-by: Gerald Combs <gerald@wireshark.org>

commit 955e336554
Author: John Thacker <johnthacker@gmail.com>
Date:   Tue Oct 21 21:31:27 2025 +0000

    Qt: Don't try to update the context menu if it doesn't exist

    Note that update is always called when creating, so ignoring updates
    until then is fine.

    Fixup 6aaf9baf8d350dff79f8e12da82ea24bde075e0f

    (cherry picked from commit ede7bd8c906d0c518d613998faa16fac5406c532)

    Co-authored-by: John Thacker <johnthacker@gmail.com>

commit 15b5a71c3a
Author: John Thacker <johnthacker@gmail.com>
Date:   Tue Oct 21 18:38:09 2025 +0000

    pidl: NULL is not a valid value for TYPE alignment in cnf files

    A TYPE directive in a conformance file has to include an alignment
    value, but it must be numeric. "NULL" isn't a valid value; with Perl
    warnings on, it will still be treated as 0 but issues a warning:

    Argument "NULL" isn't numeric in numeric lt (<) at /home/johnthacker/wireshark/tools/pidl/lib/Parse/Pidl/NDR.pm line 477.

    Replace NULL with 0 where it appears in the cnf files. This doesn't
    change the generated dissectors, but quiets warnings.

    (cherry picked from commit 72634a64c95738ddb5b84e610e8b229fbf91fff8)

    Co-authored-by: John Thacker <johnthacker@gmail.com>

commit cdfd94607c
Author: John Thacker <johnthacker@gmail.com>
Date:   Tue Oct 21 16:06:45 2025 +0000

    Qt: Create the hex data source context menu on demand

    Creating the context menu is moderately expensive, and only one
    of the data source tabs is visible and can have its context menu
    selected at any one time. Create it the first time it is requested,
    rather than creating possibly thousands of context menus at the
    beginning.

    Ping #20797

    (cherry picked from commit 6aaf9baf8d350dff79f8e12da82ea24bde075e0f)

    Co-authored-by: John Thacker <johnthacker@gmail.com>

commit fc37e7bea3
Author: Gerald Combs <gerald@wireshark.org>
Date:   Sun Oct 19 10:05:58 2025 -0700

    GitLab CI: Update WIRESHARK_BASE_DIR in the Windows jobs

    (cherry picked from commit 18568a92d63d0507dfac2c6c64eec161baecd651)

commit 116508f178
Author: John Thacker <johnthacker@gmail.com>
Date:   Tue Oct 21 16:07:09 2025 +0000

    Qt: Defer updating data tab layout and scrollbars when not visible

    Defer updating the hex or JSON data source tabs when they aren't
    visible, instead marking the layout as dirty and then update them
    when shown if the layout is dirty. When creating many tabs at once,
    the tabs aren't visible, and only one is shown at once. These
    layout updates are moderately expensive; unnoticeable when updating
    one tab when it's shown, but noticeable if adding thousands of tabs
    at once.

    Ping #20797

    (cherry picked from commit 6dac146bfd4b5dcd2332ff0e7374457b1d470c7a)

    Co-authored-by: John Thacker <johnthacker@gmail.com>

commit 4b5e46c29e
Author: John Thacker <johnthacker@gmail.com>
Date:   Tue Oct 21 00:05:42 2025 +0000

    Qt: QTabWidget::clear() performance improvement from Qt 6.8.2

    It's much faster to clear all tabs from a QTabWidget from the back
    than from the front. That became the default in Qt 6.8.2. Do it
    ourselves before that.

    https://github.com/qt/qtbase/commit/8717c1752f9b72ac7c028b722f0a068e84e64eca
    https://github.com/qt/qtbase/commit/828ece4743a0d44f7f37f1a980dec076783a8abe

    Ping #20797

    (cherry picked from commit 4ffdebd13b0ef518c8589f4837374950a599fa0c)

    Co-authored-by: John Thacker <johnthacker@gmail.com>

commit 7b05b863ca
Author: John Thacker <johnthacker@gmail.com>
Date:   Mon Oct 20 20:33:56 2025 +0000

    GitLab CI: Build the Rocky 9 packages with Qt 6

    Qt 6 is available in EPEL for Rocky 9, so that's what we're
    installing in the container now.

    (cherry picked from commit b1c31489a64d6a4f8f9090a73af8cb339dfdb85d)

    Co-authored-by: John Thacker <johnthacker@gmail.com>

commit 560b2e38cc
Author: John Thacker <johnthacker@gmail.com>
Date:   Mon Oct 20 01:10:26 2025 +0000

    PER: Don't add data sources that aren't new

    When dissecting a ASN.1 BIT STRING in PER, if the bit string is
    both byte aligned and an even number of bytes, then the aligned
    tvb produced is merely a subset tvb of the one passed in and does
    not have a new data source tvb. In such a case, don't add it as
    a new data source.

    Adding a new data source for a tvb that shares ds_tvb with another
    data source is not helpful in the GUI, as clicking on the field in
    the proto tree will select the earlier tab with the same ds_tvb.
    (Clicking on the bytes in the extra tab also does not appear to work.)

    On frame 16 in the capture in #20797 this reduces the number of data
    source tabs from 4594 to 4027, as there are a large number of 16 bit
    BIT STRINGs.

    (cherry picked from commit 4d8b93339a8438875307ccf629d16ca23111d064)

    Co-authored-by: John Thacker <johnthacker@gmail.com>

commit e47c1619b4
Author: John Thacker <johnthacker@gmail.com>
Date:   Mon Oct 20 00:58:02 2025 +0000

    smb: do not attempt to dissect wfw3.11 negotiate protocol blob

    We don't know what this 8 byte blob contains but at least do
    not flag it as malformed.

    Signed-off-by: Ronnie Sahlberg <ronniesahlberg@gmail.com>

    (cherry picked from commit 595e4df20fa7b5c7836d3e13540713b060dba1b7)

    Co-authored-by: Ronnie Sahlberg <ronniesahlberg@gmail.com>

commit 0d3ad053af
Author: John Thacker <johnthacker@gmail.com>
Date:   Mon Oct 20 00:56:22 2025 +0000

    SMB: Remove proto_tree_add_string from dissect_create_file_request

    The proto_tree_add_string caused a Malformed error.

    (cherry picked from commit 8bd6c2a3b66d8eb176d480f443a3a3e861fc17d9)

    Co-authored-by: Cal Turney <cturney111@gmail.com>

commit a1d16bfb1d
Author: Gerald Combs <gerald@wireshark.org>
Date:   Sat Oct 18 17:18:56 2025 -0700

    CMake: Properly remove our artifacts

    Unquoted list variables are lists. Quoted list variables are
    semicolon-separated strings.

    (cherry picked from commit 0e3c952d40a5fe4ff013cdd472b8c141ca6097b6)

commit 9140c83916
Author: John Thacker <johnthacker@gmail.com>
Date:   Sun Oct 19 14:34:14 2025 +0000

    Qt: Workaround QTabBar O(N^2) bug

    Workaround https://bugreports.qt.io/browse/QTBUG-141187 on
    Qt 6.3 and higher.

    Create a class derived from QTabBar to short circuit tabSizeHint
    and minimumTabSizeHint when the QTabBar is not visible. This isn't
    the perfect solution, as the functions that really need to be
    short-circuited aren't virtual and can't be overridden unlike these
    two. However, while adding lots of tabs is still O(N^2) due to
    all the layoutTabs() calls, it is about ten times faster and for
    the case in #20797 reduces the time to select the packet from a
    close to unusable 20 seconds to a merely painful 2 s.

    (cherry picked from commit 362dd5aa9fbe05a82d80d5f46748e8b270290e10)

    Co-authored-by: John Thacker <johnthacker@gmail.com>

commit 8e25bb64ae
Author: Anders Broman <a.broman58@gmail.com>
Date:   Sun Oct 19 16:11:08 2025 +0200

    Windows: Update GnuTLS to 3.8.10-2

    Build GnuTLS with `--with-system-priority-file=`.

    (cherry picked from commit 1f753ba4d1d3055ca1f30f0afe23fddf2db0cce0)

    Co-authored-by: Gerald Combs <gerald@wireshark.org>

commit 603e3b722c
Author: John Thacker <johnthacker@gmail.com>
Date:   Sun Oct 19 11:23:18 2025 +0000

    Qt: Fix compilation with clang-cl

    Apparently a forward declaration isn't good enough for clang-cl here.

    (cherry picked from commit 07f5ddf67510f773ad9f1f9ea7f9b54d13d328a4)

    Co-authored-by: John Thacker <johnthacker@gmail.com>

commit b3e781f03a
Author: Gerald Combs <gerald@wireshark.org>
Date:   Sun Oct 19 10:19:06 2025 +0000

    [Automatic update for 2025-10-19]

    Update manuf, services enterprise numbers, translations, and other items.

commit e4ed255efd
Author: Stig Bjørlykke <stig@bjorlykke.org>
Date:   Sat Oct 18 09:49:15 2025 +0000

    tls: Abbreviated Handshake Using New Session Ticket

    When the ServerHello contains a NewSessionTicket the stored
    master secret for the SessionTicket used in ClientHello must be
    restored to be able to save it with the NewSessionTicket.

    Fixes #20802

    (cherry picked from commit 8ded5ea9e4b53d751b4ca1814e0bbd2a5c896329)

    Co-authored-by: Stig Bjørlykke <stig@bjorlykke.org>

commit a038fb9ab8
Author: John Thacker <johnthacker@gmail.com>
Date:   Thu Oct 16 23:18:15 2025 +0000

    HTTP: Fix upgrade info server port

    HTTP upgrades are acknowledged by the server, and so the server port
    is the *source* port in those messages from server, as is done in the
    previous line.

    Fixup f82ed9a537379aeb29a7cd1dbe3b9e024c97ecd8
    Fix #20803

    (cherry picked from commit 6bb3628b816240d85736c2daaa481f52e3172b5e)

    Co-authored-by: John Thacker <johnthacker@gmail.com>

commit 53a275911a
Author: John Thacker <johnthacker@gmail.com>
Date:   Thu Oct 16 12:00:55 2025 +0000

    Qt: Put back the clear call in DataSourceTab

    It's not needed on Linux, but apparently is on macOS and maybe Windows.
    On my Linux and Windows tests with Qt 6.9 the delete operation is still O(N)
    instead of O(N^2) even with the clear call; hopefully that's true
    in general.

    Fixup 2a09c1a676ee10ae6bcb9a9beab749b5e15a5636

    (cherry picked from commit d86f462f939c87b150f476f17a679d95e73b4b10)

    Co-authored-by: John Thacker <johnthacker@gmail.com>

commit 54ac6bcc36
Author: Anders Broman <a.broman58@gmail.com>
Date:   Thu Oct 16 08:23:54 2025 +0200

    Qt: Hide the DataSourceTab before adding or removing tabs

    When a tab is added or removed, the QTabBar can re-layout all the tabs,
    which calculates the sizeHint and minimumSizeHint for every tab. If
    it is called every time a tab is added or removed, it becomes an O(N^2)
    operation to add or remove N tabs.

    Making the DataSourceTab not visible before adding or removing tabs
    and then visible again at the end can in some cases remove this to
    a single layoutTabs call, reducing things to O(N).

    Unfortunately a Qt commit in 6.3.0 means that layoutTabs is always
    called when inserting a new tab, even if the QTabWidget is hidden:
    https://bugreports.qt.io/browse/QTBUG-141187

    However, on all versions of Qt this makes removing when switching to
    a new frame much faster, so this addresses part of #20797

    (cherry picked from commit 2a09c1a676ee10ae6bcb9a9beab749b5e15a5636)

    Co-authored-by: John Thacker <johnthacker@gmail.com>

commit 80464b3a49
Author: Paul Armstrong <psa@users.noreply.github.com>
Date:   Wed Oct 15 18:08:09 2025 +1000

    packet-kerberos.c: Fix extraneous ) in wmem_strbuf_append_printf call

commit 6c96f346ec
Author: John Thacker <johnthacker@gmail.com>
Date:   Wed Oct 15 12:33:42 2025 +0000

    expert: Clear same_name_expinfo before insert

    Insert missing line
    Fixup 1e16fb62741042fcb9f5a313b2f61ed80f289e09

    (cherry picked from commit 92ae88e49280f6f483c4cb424cc96e535c7f9e5c)

    Co-authored-by: John Thacker <johnthacker@gmail.com>

commit 5d49f6ca52
Author: Alexis La Goutte <alexis.lagoutte@gmail.com>
Date:   Wed Oct 15 10:33:30 2025 +0200

    expert: Track expert info with the same abbrev

    It's possible to register multiple expert info with the same abbrev.
    In order for freeing and deregistering to work properly, they have
    to be tracked somehow. (This apparently didn't cause a problem before
    because nothing was actually freed or had something that needed to
    be set to NULL to avoid trying to free a static string.)

    Fixup 0e64c8871d48c3f8dfbe0a92561c7fe002f2031b
    Fix #20794

    (cherry picked from commit 58341860d5aa0ef9fc6682e30edb68b9e2d26005)

    Co-authored-by: John Thacker <johnthacker@gmail.com>

commit 15f71b6b47
Author: John Thacker <johnthacker@gmail.com>
Date:   Mon Oct 13 21:13:58 2025 +0000

    CMake: Preserve include directory structure

    Install the include directory preserving its directory structure,
    so that c128 include files are installed into
    ${PROJECT_INSTALL_INCLUDEDIR}/include/c128
    and not in just in
    ${PROJECT_INSTALL_INCLUDEDIR}/include

    To avoid including these files when attempting to include
    glibc's endian.h header.

    This means including them with #include<c128/...> instead, but as
    they aren't currently used yet there's no changes elsewhere in the tree.

    Fixup eaf3bc2d7d5c446851ea4a75e1ff7a5570eac6c7
    Fix #20786

    (cherry picked from commit 1ff90f5d49be74449da0f44a8a88da795ef31c34)

    Co-authored-by: John Thacker <johnthacker@gmail.com>

commit c46074964f
Author: Stig Bjørlykke <stig@bjorlykke.org>
Date:   Mon Oct 13 16:04:24 2025 +0000

    sgp22: Use the generated exports

    Remove the manually created sgp22_exp.cnf and use the generated one.

    (cherry picked from commit 3e85fceae941c745d92ad8eeb664e8c1d6d0787f)

    Co-authored-by: Stig Bjørlykke <stig@bjorlykke.org>

commit 3e704631aa
Author: Stig Bjørlykke <stig@bjorlykke.org>
Date:   Mon Oct 13 11:57:05 2025 +0000

    sgp22: More type export

    It should have been possible to do this in asn2wrs.py.

    (cherry picked from commit 219a927b9e91e23fb603b3c66b33aebdea44dacb)

    Co-authored-by: Stig Bjørlykke <stig@bjorlykke.org>

commit 1b0ca4c499
Author: Anders Broman <a.broman58@gmail.com>
Date:   Mon Oct 13 08:33:05 2025 +0200

    win32-utils: fetch exception descriptions from ntdll.dll.

    Instead of supplying our own descriptions, and not having a description
    for every exception, load ntdll.dll and get the exception string from
    there with FormatMessageW() and convert it to UTF-8.

    This also means we don't use a static buffer, so win32strexception() is
    thread-safe.

    (cherry picked from commit 560fa04c9022db5ae2ac29c315159ee4236104b8)

    Co-authored-by: Guy Harris <gharris@sonic.net>

commit 025ef4920c
Author: Guy Harris <gharris@sonic.net>
Date:   Sun Oct 12 17:37:12 2025 +0000

    capture_sync: fix a check for a read error.

    There's an error if pipe_read_bytes() returns -1, not if it *doesn't*
    return -1.

    (cherry picked from commit 33d60c5153e34f909d590d41f52d3660ad7bff95)

    Co-authored-by: Guy Harris <gharris@sonic.net>

commit 470574c1ff
Author: Stig Bjørlykke <stig@bjorlykke.org>
Date:   Sun Oct 12 13:43:25 2025 +0000

    sgp22: Improve type export

    Add correct type attributes to exported types.
    Dissect Iccid and VersionType according to spec.
    Use the asn2wrs option to add constraint checks.
    Register to decode-as "media_type".

    Export SubjectKeyIdentifier type from pkix1implicit.

    (cherry picked from commit 7f37406d807ac05b72118a5075a405a30de90bb4)

    Co-authored-by: Stig Bjørlykke <stig@bjorlykke.org>

commit 66195ba8eb
Author: Gerald Combs <gerald@wireshark.org>
Date:   Sun Oct 12 10:18:07 2025 +0000

    [Automatic update for 2025-10-12]

    Update manuf, services enterprise numbers, translations, and other items.

commit 73c4df9260
Author: Guy Harris <gharris@sonic.net>
Date:   Sun Oct 12 10:41:23 2025 +0000

    tshark: clean up call to sync_pipe_start().

    No need to save its return value, just check it when the call is made.

    (cherry picked from commit 3399d2082c343ba6d412c6f195501571f9778345)

    Co-authored-by: Guy Harris <gharris@sonic.net>

commit 239e58a640
Author: Stig Bjørlykke <stig@bjorlykke.org>
Date:   Sat Oct 11 18:33:09 2025 +0000

    ber: Remove unknown octetstring item for hf_id -1

    The whole idea of providing hf_id = -1 (HF_INDEX = -1 in cnf) is
    to avoid adding the item to the tree.

    (cherry picked from commit 78a0a51d77d7877a55eadae151a60a51c5758ade)

    Co-authored-by: Stig Bjørlykke <stig@bjorlykke.org>

commit b3295ac628
Author: John Thacker <johnthacker@gmail.com>
Date:   Sat Oct 11 14:15:06 2025 +0000

    bpv7: Fix pinfo conv_elements address lifetime

    The Bundle protocol dissector uses a different scope lifetime
    for its eid elements and frees them slightly earlier. For
    putting addresses in the pinfo->conv_elements structure we need
    to use something with pinfo->pool lifetime. Use the addresses
    already copied to pinfo->src and pinfo->dst in pinfo->pool scope.

    Fix #20770

    (cherry picked from commit 8d0b2e248a84da60f8df01a2f7fdd5ae3770be1d)

    Co-authored-by: John Thacker <johnthacker@gmail.com>

commit bfd75715c7
Author: John Thacker <johnthacker@gmail.com>
Date:   Sat Oct 11 12:41:20 2025 +0000

    TFTP: Fix conversation lookups

    Both find_conversation_strat calls are using the same direction; we
    need to check both (we don't know which side is the original client
    port).

    Always create a new conversation for a read or write request (which
    is always the first TFTP opcode according to RFC 1350.) The
    tftp_conv_info_t struct only handles data for a single transfer;
    we can't reuse it for another transfer and have things work on the
    second pass. (Maybe the conversation data should be a wmem_tree_t
    of these?)

    Do so even if the port doesn't match, as we could have gotten here
    from, e.g., a conversation dissector being sent (if ports are reused
    in the same capture), in which case match_uint might be 17 for UDP
    from the IP dissector, say.

    Fixes the handling of the files in #16767 again.

    Fixup a37b6c166929141a9044abe1520f13bc4894a893

    (cherry picked from commit 401a10460d07f280e4ad797092c05fa4d6338767)

    Co-authored-by: John Thacker <johnthacker@gmail.com>

commit db8cd8cd92
Author: Alexis La Goutte <alexis.lagoutte@gmail.com>
Date:   Fri Oct 10 20:27:51 2025 +0200

    radiotap: fix unknown TLV item parent

    These items should be parented to the radiotap tree, not the
    overall packet tree. Fix that for TLVs, the else branch for
    non-TLV fields already does this.

    (cherry picked from commit 28c737d987867ab30d22971903588b3ac3659ea8)

    Co-authored-by: Johannes Berg <johannes.berg@intel.com>

commit 7cefee1288
Author: Alexis La Goutte <alexis.lagoutte@gmail.com>
Date:   Fri Oct 10 07:52:32 2025 +0200

    packet-darwin: fixed typo that causes the BK_SYS to be marked as "Unknown(100)"

    (cherry picked from commit e1751bc600909f9bc59b088cf777afdc1066ee2f)

    Co-authored-by: Omer Shapira <omer_shapira@apple.com>

commit 57c5223bc5
Author: Gerald Combs <gerald@wireshark.org>
Date:   Mon Sep 29 11:19:22 2025 -0700

    CMake+docs: Deprecate Qt5

    Print a deprecation warning if we're building with Qt5, and recommend
    against it in the documentation and tools help.

    (cherry picked from commit eb20d6e3b361cc81141c0265f0b759277478f249)

commit c12b8279fd
Author: Uli Heilmeier <uh@heilmeier.eu>
Date:   Thu Oct 9 21:34:15 2025 +0200

    DNS: fix HIP PK algo field name

    Related: #20768

    (cherry picked from commit 65fb8a629b76f955d88e5e8495e21641d8945bd3)

    Co-authored-by: Uli Heilmeier <uh@heilmeier.eu>

commit be6111df60
Author: Uli Heilmeier <uh@heilmeier.eu>
Date:   Wed Oct 8 21:50:31 2025 +0200

    DNS: fix field name for HIP HIT algo

    Fixes: #20768

    (cherry picked from commit 07bba19c48e6d8a899badd4b083835351abe3806)

    Co-authored-by: Uli Heilmeier <uh@heilmeier.eu>

commit 5ac0045c5a
Author: Gerald Combs <gerald@wireshark.org>
Date:   Thu Oct 9 09:02:45 2025 -0700

    Version: 4.6.0 → 4.6.1 [skip ci]

commit e0cf07cfc4
Author: Anders Broman <a.broman58@gmail.com>
Date:   Thu Oct 9 09:04:29 2025 +0200

    wslua: Add length check in TreeItem_set_len

    Add a check in TreeItem_set_len() for illegal length values.

    (cherry picked from commit 63dc61bf810c9ab37b86eab3cb503932c0edabde)

    Co-authored-by: Stig Bjørlykke <stig@bjorlykke.org>
