Builder libpcap-solaris11-i386 Build #13

Results:

Build successful

SourceStamp:

BuildSlave:

Reason:

The SingleBranchScheduler scheduler named 'schedule-libpcap-solaris11-i386' triggered this build

Steps and Logfiles:

1. git update ( 4 secs )
   1. stdio
2. shell './build_matrix.sh' ( 3 hrs, 23 mins, 8 secs )
   1. stdio
   2. config.log

Build Properties:

Forced Build Properties:

Responsible Users:

1. John Thacker

   johnthackerohnoyoudont@gmail.com

Timing:

All Changes:

:

1. Change #230822

   Category	libpcap
   Changed by	John Thacker <johnthackerohnoyoudont@gmail.com>
   Changed at	Sat 24 May 2025 03:28:18
   Repository	https://git.tcpdump.org/libpcap
   Project	libpcap
   Branch	master
   Revision	8862ca3a4ca8d56917833994721f6a5c2152b8f8

   Comments

   Linux: Simplify libnl handling.
   Use the constants NL_AUTO_PORT and NL_AUTO_SEQ instead of their value
   0, for clarity.  Use nl_send_auto() instead of nl_send_auto_complete(),
   which is deprecated.  Factor out some common nlmsg_free() calls.  Use
   nl_send_sync() in add_mon_if() and del_mon_if(), which combines
   nl_send_auto() and nl_wait_for_ack() when using the standard callback.
   
   Signed-off-by: John Thacker <johnthacker@gmail.com>

   Changed files

   • pcap-linux.c

2. Change #230823

   Category	libpcap
   Changed by	John Thacker <johnthackerohnoyoudont@gmail.com>
   Changed at	Sat 24 May 2025 03:29:17
   Repository	https://git.tcpdump.org/libpcap
   Project	libpcap
   Branch	master
   Revision	575768e4dbf4d932cd59f12912a1111d570efa22

   Comments

   Linux: Fix a memory leak in get_if_type().
   The function does not deallocate the struct nl_cb, no matter if it exits
   normally (wlan1) or due to an error (wlan0).  To address the former code
   path, add a missing nl_cb_put() call and make it as early as possible;
   to address the latter code path, move the nl_cb_alloc() call to make it
   as late as possible.
   
   Before:
    # valgrind --quiet --leak-check=full ./testprogs/opentest -i wlan0 -I
   opentest: wlan0: Generic error
   (SIOCGIFINDEX: No such device)
   ==1748== 224 bytes in 1 blocks are definitely lost in loss record 2 of 3
   ==1748==    at 0x4889F94: calloc (vg_replace_malloc.c:1328)
   ==1748==    by 0x48FF9BB: nl_cb_alloc (in /usr/lib/aarch64-linux-gnu/libnl-3.so.200.26.0)
   ==1748==    by 0x12C6CB: get_if_type (pcap-linux.c:647)
   ==1748==    by 0x12C6CB: enter_rfmon_mode (pcap-linux.c:4934)
   ==1748==    by 0x12D157: setup_socket (pcap-linux.c:2505)
   ==1748==    by 0x12D157: pcap_activate_linux (pcap-linux.c:1249)
   ==1748==    by 0x110E2F: pcap_activate (pcap.c:2708)
   ==1748==    by 0x10DF07: main (opentest.c:174)
   ==1748==
    # valgrind --quiet --leak-check=full ./testprogs/opentest -i wlan1 -I
   wlan1 opened successfully
   ==1754== 224 bytes in 1 blocks are definitely lost in loss record 1 of 1
   ==1754==    at 0x4889F94: calloc (vg_replace_malloc.c:1328)
   ==1754==    by 0x48FF9BB: nl_cb_alloc (in /usr/lib/aarch64-linux-gnu/libnl-3.so.200.26.0)
   ==1754==    by 0x12C6CB: get_if_type (pcap-linux.c:647)
   ==1754==    by 0x12C6CB: enter_rfmon_mode (pcap-linux.c:4934)
   ==1754==    by 0x12D157: setup_socket (pcap-linux.c:2505)
   ==1754==    by 0x12D157: pcap_activate_linux (pcap-linux.c:1249)
   ==1754==    by 0x110E2F: pcap_activate (pcap.c:2708)
   ==1754==    by 0x10DF07: main (opentest.c:174)
   ==1754==
   
   After:
    # valgrind --quiet --leak-check=full ./testprogs/opentest -i wlan0 -I
   opentest: wlan0: Generic error
   (SIOCGIFINDEX: No such device)
    # valgrind --quiet --leak-check=full ./testprogs/opentest -i wlan1 -I
   wlan1 opened successfully
   
   Signed-off-by: John Thacker <johnthacker@gmail.com>

   Changed files

   • pcap-linux.c

3. Change #230824

   Category	libpcap
   Changed by	John Thacker <johnthackerohnoyoudont@gmail.com>
   Changed at	Sat 24 May 2025 03:44:34
   Repository	https://git.tcpdump.org/libpcap
   Project	libpcap
   Branch	master
   Revision	258e6ca3a537c02454546d9640274d3bbbac8eaf

   Comments

   Linux: Always wait for an ACK from nl80211
   get_if_type and add_mon_if share the same generic netlink socket.
   By default ACKs are sent for every transaction. If get_if_type
   does not wait for the ACK, then add_mon_if will receive the ACK
   and wrongly interpret its NL80211_CMD_NEW_INTERFACE to add the
   monitor device as a success even if it failed. This can cause
   libpcap to assume that a monitor device was added even when it
   failed, later causing a ENODEV error when trying to open the
   non-existent device.
   
   Before:
    # ./testprogs/opentest -i wlan0 -I
   opentest: wlan0: Generic error
   (SIOCGIFINDEX: No such device)
   
   After:
    # ./testprogs/opentest -i wlan0 -I
   opentest: wlan0: Generic error
   (wlan0: nl_send_sync failed adding mon0 interface: Operation not supported)
   
   Signed-off-by: John Thacker <johnthacker@gmail.com>

   Changed files

   • pcap-linux.c

4. Change #230825

   Category	libpcap
   Changed by	John Thacker <johnthackerohnoyoudont@gmail.com>
   Changed at	Sat 24 May 2025 03:44:48
   Repository	https://git.tcpdump.org/libpcap
   Project	libpcap
   Branch	master
   Revision	186f90842bcf4c14372e228c9af161f59efd12f9

   Comments

   Linux: Fix returning PCAP_ERROR_RFMON_NOTSUP with libnl.
   If Netlink reports that an interface is a mac80211 device (not in
   monitor mode), but then reports NLE_OPNOTSUPP when trying to add
   that interface as a monitor mode device, report that as the
   PCAP_ERROR_RFMON_NOTSUP so the libpcap error message is used instead
   of the generic Netlink error message.  For other error types, continue
   to report a generic error.
   
   Before:
    # ./testprogs/opentest -i wlan0 -I
   opentest: wlan0: Generic error
   (wlan0: nl_send_sync failed adding mon0 interface: Operation not supported)
   
   After:
    # ./testprogs/opentest -i wlan0 -I
   opentest: wlan0: That device doesn't support monitor mode
   (That device doesn't support monitor mode)
   
   Signed-off-by: John Thacker <johnthacker@gmail.com>

   Changed files

   • CHANGES
   • pcap-linux.c