Builder libpcap-solaris10-sparcv9 Build #3203

Results:

Build successful

SourceStamp:

BuildSlave:

Reason:

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

Steps and Logfiles:

1. git update ( 20 secs )
   1. stdio
2. shell Set developer mode ( 0 secs )
   1. stdio
3. shell_1 './autogen.sh' ( 20 secs )
   1. stdio
4. shell_2 './configure' ( 57 secs )
   1. stdio
   2. config.log
5. shell_3 'gmake' ( 46 secs )
   1. stdio
6. shell_4 'gmake check' ( 20 mins, 45 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, 12 secs )
   1. - no logs -
   2. tcpdump-solaris10-sparcv9 #6045

Build Properties:

Forced Build Properties:

Responsible Users:

1. Denis Ovsienko

   denisohnoyoudont@ovsienko.info

Timing:

All Changes:

:

1. Change #247943

   Category	libpcap
   Changed by	Denis Ovsienko <denisohnoyoudont@ovsienko.info>
   Changed at	Fri 07 Nov 2025 13:01:29
   Repository	https://git.tcpdump.org/libpcap
   Project	libpcap
   Branch	master
   Revision	d0d58d9333dd53e8c2a468d190885a3cf5585af8

   Comments

   Fix GENEVE- and VXLAN-encapsulated matching.
   In my recent commit 1687065 the filter programs for "vxlan [NUM]",
   "vxlan and vxlan", "geneve [NUM]" and "geneve and geneve" have become
   substantially shorter.  As it turns out, the missing instructions are
   the outputs of gen_geneve_offsets() and gen_vxlan_offsets(), so
   filtering of protocols encapsulated in GENEVE and VXLAN is no longer
   correct.  Because libpcap does not have any apply filter tests for the
   affected syntax, this was not apparent until a tcpdump test failed.
   
   The instructions went missing because gen_geneve() and gen_vxlan() need
   to use these as a block, but instead of making a new block from scratch
   the functions take the result of gen_true() and modify it to add side
   effects.  After the commit the block with the wanted side effects ended
   up incorrectly marked as a Boolean constant and susequently gen_and()
   started to discard it.
   
   In gen_cond() take an optional list of side effect instructions and make
   a Boolean constant block a specific case of a block with a degenerate
   branch statement.  Switch gen_geneve() and gen_vxlan() from gen_true()
   to gen_uncond() to reduce the violation of layers.  Restore the six
   previously affected tests exactly as they were before.

   Changed files

   • gencode.c
   • testprogs/TESTrun

2. Change #247944

   Category	libpcap
   Changed by	Denis Ovsienko <denisohnoyoudont@ovsienko.info>
   Changed at	Fri 07 Nov 2025 13:01:37
   Repository	https://git.tcpdump.org/libpcap
   Project	libpcap
   Branch	master
   Revision	2e15d1a1744ce4447c303c3603f1a1241b1ae5e8

   Comments

   TESTrun: Test GENEVE and VXLAN a bit better.
   To detect gen_geneve() and gen_vxlan() regressions that affect filtering
   of encapsulated protocols, define two new apply filter tests.  For now
   these stand for IPv4 only.

   Changed files

   • testprogs/TESTrun

3. Change #247945

   Category	libpcap
   Changed by	Denis Ovsienko <denisohnoyoudont@ovsienko.info>
   Changed at	Fri 07 Nov 2025 16:05:16
   Repository	https://git.tcpdump.org/libpcap
   Project	libpcap
   Branch	master
   Revision	9309458887afa106752e5e34eed001cc42f32181

   Comments

   TESTrun: Add detailed tests for AND/OR reductions.
   Six tests make it easy to see another regression I have introduced in
   commit 1687065.

   Changed files

   • testprogs/TESTrun

4. Change #247946

   Category	libpcap
   Changed by	Denis Ovsienko <denisohnoyoudont@ovsienko.info>
   Changed at	Fri 07 Nov 2025 16:05:16
   Repository	https://git.tcpdump.org/libpcap
   Project	libpcap
   Branch	master
   Revision	3ed0cd024e96bd3a0b016c49f0a1128e4563850a

   Comments

   Always pick the correct argument in gen_or().
   I got the comments right, but not all of the code, fix that and update
   the tests.

   Changed files

   • gencode.c
   • testprogs/TESTrun