Builder libpcap-solaris10-sparcv9 Build #3194

Results:

Build successful

SourceStamp:

BuildSlave:

Reason:

The SingleBranchScheduler scheduler named 'schedule-libpcap-solaris10-sparcv9' triggered this build

Steps and Logfiles:

1. git update ( 23 secs )
   1. stdio
2. shell Set developer mode ( 1 secs )
   1. stdio
3. shell_1 './autogen.sh' ( 35 secs )
   1. stdio
4. shell_2 './configure' ( 1 mins, 26 secs )
   1. stdio
   2. config.log
5. shell_3 'gmake' ( 1 mins, 36 secs )
   1. stdio
6. shell_4 'gmake check' ( 18 mins, 0 secs )
   1. stdio
7. shell_5 'rm -rf ...' ( 0 secs )
   1. stdio
8. shell_6 'gmake install ...' ( 3 secs )
   1. stdio
9. trigger triggered trigger-tcpdump-solaris10-sparcv9 ( 17 mins, 19 secs )
   1. - no logs -
   2. tcpdump-solaris10-sparcv9 #6036

Build Properties:

Forced Build Properties:

Responsible Users:

1. Denis Ovsienko

   denisohnoyoudont@ovsienko.info

Timing:

All Changes:

:

1. Change #247149

   Category	libpcap
   Changed by	Denis Ovsienko <denisohnoyoudont@ovsienko.info>
   Changed at	Thu 30 Oct 2025 14:20:03
   Repository	https://git.tcpdump.org/libpcap
   Project	libpcap
   Branch	master
   Revision	e7651403901b5fbed6c54e1ac0a938408f27613b

   Comments

   Use inet_pton() directly to validate HID6 in scanner.l.
   To validate a string to be a well-formed IPv6 address, an invocation of
   getaddrinfo(3) with hints.ai_family set to AF_INET6 and hints.ai_flags
   set to AI_NUMERICHOST effectively is the same as an invocation of
   inet_pton(3), but involves an indirection with setup and teardown, so do
   not over-engineer it.  See also commit 5acbaa8.

   Changed files

   • gencode.c
   • scanner.l

2. Change #247150

   Category	libpcap
   Changed by	Denis Ovsienko <denisohnoyoudont@ovsienko.info>
   Changed at	Thu 30 Oct 2025 14:21:27
   Repository	https://git.tcpdump.org/libpcap
   Project	libpcap
   Branch	master
   Revision	2641704f6925cb1dc742139c3ddae1bae03e7142

   Comments

   Decipher a few loads in gen_protochain().

   Changed files

   • gencode.c

3. Change #247151

   Category	libpcap
   Changed by	Denis Ovsienko <denisohnoyoudont@ovsienko.info>
   Changed at	Thu 30 Oct 2025 15:36:33
   Repository	https://git.tcpdump.org/libpcap
   Project	libpcap
   Branch	master
   Revision	c34f4a98fbf08c1dac474e279faa913bc9e9b72a

   Comments

   For proto-unqualified "port" test for SCTP after TCP and UDP.
   Point 3 of GH #1022 is that the current implementation of "port" in
   gen_port_common() and gen_port6_common() tests for SCTP first, then for
   TCP, then for UDP, but for a random packet SCTP is the least likely
   protocol in this set.  Therefore make the SCTP match the last one.
   
   As to which of UDP and TCP is more likely than the other in a random
   packet, [1] discusses that in 1998-2008 TCP was substantially more
   likely with no tendency for change, so until proven otherwise let's
   suppose this still holds.  Therefore keep the UDP match after the TCP
   match.
   
   1: "Media Streaming Observations: Trends in UDP to TCP Ratio",
      pp. 147-162, vol. 3, No 3 & 4, year 2010, International Journal on
      Advances in Systems and Measurements
      https://www.iariajournals.org/systems_and_measurements/sysmea_v3_n34_2010_paged.pdf

   Changed files

   • gencode.c
   • testprogs/TESTrun

4. Change #247152

   Category	libpcap
   Changed by	Denis Ovsienko <denisohnoyoudont@ovsienko.info>
   Changed at	Thu 30 Oct 2025 19:45:02
   Repository	https://git.tcpdump.org/libpcap
   Project	libpcap
   Branch	master
   Revision	1f3648a649c4cd12b5d864ef30ac9f35ff283053

   Comments

   TESTrun: Add reject tests for invalid dir qualifiers.

   Changed files

   • testprogs/TESTrun

5. Change #247153

   Category	libpcap
   Changed by	Denis Ovsienko <denisohnoyoudont@ovsienko.info>
   Changed at	Thu 30 Oct 2025 20:13:32
   Repository	https://git.tcpdump.org/libpcap
   Project	libpcap
   Branch	master
   Revision	a887dd3044a44cb5057442db7cea5bf51ae79cbf

   Comments

   Use bpf_error() to handle all unexpected dir qualifiers.
   If a direction qualifier has been validated by the parser and a switch
   block on the result has a case for every valid value (as is the case
   now), the default case with abort() is dead code, otherwise it is a bug
   in libpcap (as was the case before commit 4d6e12c and potentially could
   be again in future) that is specific to the filter string.
   
   The latter could or could not be a user input, but in any case instead
   of terminating the process it would be more proper to report an ordinary
   failure to compile the expression and to let the application handle it.
   Thus in all switch blocks that take a direction qualifier handle both
   the expected invalid values and any unexpected invalid values using
   bpf_error().

   Changed files

   • gencode.c