Summary for 1.8.1 libpcap release
Add a target in Makefile.in for Exuberant Ctags use: 'extags'.
Rename configure.in to configure.ac: autoconf 2.59
Clean up the name-to-DLT mapping table.
Add some newer DLT_ values: IPMI_HPM_2,ZWAVE_R1_R2,ZWAVE_R3,WATTSTOPPER_DLM,ISO_14443,RDS
Clarify what the return values are for both success and failure.
Many changes to build on windows
Check for the "break the loop" condition in the inner loop for TPACKET_V3.
Fix handling of packet count in the TPACKET_V3 inner loop: GitHub issue #493.
Filter out duplicate looped back CAN frames.
Fix the handling of loopback filters for IPv6 packets.
Add a link-layer header type for RDS (IEC 62106) groups.
Use different intermediate folders for x86 and x64 builds on Windows.
On Linux, handle all CAN captures with pcap-linux.c, in cooked mode.
Removes the need for the "host-endian" link-layer header type.
Compile with '-Wused-but-marked-unused' in devel mode if supported
Have separate DLTs for big-endian and host-endian SocketCAN headers.
Reflect version.h being renamed to pcap_version.h.
Require that version.h be generated: all build procedures we support generate version.h (autoconf, CMake, MSVC)!
Properly check for sock_recv() errors.
Re-impose some of Winsock's limitations on sock_recv().
Replace sprintf() with pcap_snprintf().
Fix signature of pcap_stats_ex_remote().
Initial cmake support for remote packet capture.
Have rpcap_remoteact_getsock() return a SOCKET and supply an "is active" flag.
Clean up {DAG, Septel, Myricom SNF}-only builds.
Do UTF-16-to-ASCII conversion into the right place.
pcap_create_interface() needs the interface name on Linux.
Clean up hardware time stamp support: the "any" device does not support any time stamp types.
Add support for capturing on FreeBSD usbusN interfaces.
Add a LINKTYPE/DLT_ value for FreeBSD USB.
Go back to using PCAP_API on Windows.
CMake support
Add TurboCap support from WinPcap.
Recognize 802.1ad nested VLAN tag in vlan filter.
Thursday Sep. 3, 2015 guy@alum.mit.edu
Summary for 1.7.5 libpcap release
Man page cleanups.
Add some allocation failure checks.
Fix a number of Linux/ucLinux configure/build issues.
Fix some memory leaks.
Recognize 802.1ad nested VLAN tag in vlan filter.
Fix building Bluetooth Linux Monitor support with BlueZ 5.1+
Saturday Jun. 27, 2015 mcr@sandelman.ca
Summary for 1.7.4 libpcap release
Include fix for GitHub issue #424 -- out of tree builds.
Friday Apr. 10, 2015 guy@alum.mit.edu
Summary for 1.7.3 libpcap release
Work around a Linux bonding driver bug.
Summary for 4.8.1 tcpdump release
Fix "-x" for Apple PKTAP and PPI packets
Use PRIx64 to print a 64-bit number in hex.
Printer for HNCP (RFCs 7787 and 7788).
dagid is always an IPv6 address, not an opaque 128-bit string, and other fixes to RPL printer.
RSVP: Add bounds and length checks
OSPF: Do more bounds checking
Handle OpenSSL 1.1.x.
Initial support for the REdis Serialization Protocol known as RESP.
Add printing function for Generic Protocol Extension for VXLAN
draft-ietf-nvo3-vxlan-gpe-01
Network Service Header: draft-ietf-sfc-nsh-01
Don't recompile the filter if the new file has the same DLT.
Pass an adjusted struct pcap_pkthdr to the sub-printer.
Add three test cases for already fixed CVEs
CVE-2014-8767: OLSR
CVE-2014-8768: Geonet
CVE-2014-8769: AODV
Don't do the DDP-over-UDP heuristic first: GitHub issue #499.
Use the new debugging routines in libpcap.
Harmonize TCP source or destination ports tests with UDP ones
Introduce data types to use for integral values in packet structures.
RSVP: Fix an infinite loop
Support of Type 3 and Type 4 LISP packets.
Don't require IPv6 library support in order to support IPv6 addresses.
Many many changes to support libnetdissect usage.
Add a test that makes unaligned accesses: GitHub issue #478.
add a DNSSEC test case: GH #445 and GH #467.
BGP: add decoding of ADD-PATH capability
fixes to LLC header printing, and RFC948-style IP packets
Friday April 10, 2015 guy@alum.mit.edu
Summary for 4.7.4 tcpdump release
RPKI to Router Protocol: Fix Segmentation Faults and other problems
RPKI to Router Protocol: print strings with fn_printn()
wb: fix some bounds checks
bpf_mtap of some drivers is still called in hardware interrupt context.
We want to run them in softint as well as bpf_mtap of most drivers
(see if_percpuq_softint and if_input).
To this end, bpf_mtap_softint mechanism is implemented; it defers
bpf_mtap processing to a dedicated softint for a target driver.
By using the machanism, we can move bpf_mtap processing to softint
without changing target drivers much while it adds some overhead
on CPU and memory. Once target drivers are changed to softint-based,
we should return to normal bpf_mtap.
Proposed on tech-kern and tech-net
- Don't hold softnet_lock in some functions if NET_MPSAFE
- Add softnet_lock to sysctl_net_inet_icmp_redirtimeout
- Add softnet_lock to expire_upcalls of ip_mroute.c
- Restore softnet_lock for in{,6}_pcbpurgeif{,0} if NET_MPSAFE
- Mark some softnet_lock for future work
It is necessary for example when we use tun(4). Without it the following
panic occurs:
panic: kernel diagnostic assertion "(kpreempt_disabled() || cpu_softintr_p() || ISSET(curlwp->l_pflag, LP_BOUND))" failed: file "/usr/src/sys/kern/subr_psref.c", line 291 passive references are CPU-local, but preemption is enabled and the caller is not in a softint or CPU-bound LWP
Backtrace:
vpanic()
ch_voltag_convert_in()
psref_release()
pfil_run_arg.isra.0()
if_initialize()
if_attach()
tun_clone_create()
tunopen()
cdev_open()
spec_open()
VOP_OPEN()
vn_open()
do_open()
do_sys_openat()
sys_open()
syscall()
