]>
git.ipfire.org Git - thirdparty/kea.git/log
Marcin Siodelski [Thu, 25 Apr 2013 18:42:53 +0000 (20:42 +0200)]
[2902] Do not use the broadcast address as source in server's response.
Marcin Siodelski [Thu, 25 Apr 2013 17:32:28 +0000 (19:32 +0200)]
[2902] DHCPv4 server will try to use 'direct response' feature if possible.
Michal 'vorner' Vaner [Thu, 25 Apr 2013 13:07:35 +0000 (15:07 +0200)]
[2738] Describe behavior of clients
Michal 'vorner' Vaner [Thu, 25 Apr 2013 12:06:47 +0000 (14:06 +0200)]
[2738] Some wording & clarifications
Michal 'vorner' Vaner [Thu, 25 Apr 2013 11:37:02 +0000 (13:37 +0200)]
[2738] Clarify blocking & fast operations
Michal 'vorner' Vaner [Thu, 25 Apr 2013 11:32:39 +0000 (13:32 +0200)]
[2738] Remove ipc-high-2
It was mostly an example and is now merged to the base version. No
longer needed.
Marcin Siodelski [Thu, 25 Apr 2013 10:29:59 +0000 (12:29 +0200)]
[2902] Build LPF sockets on Linux only.
Marcin Siodelski [Thu, 25 Apr 2013 09:30:43 +0000 (11:30 +0200)]
[2902] Unit tests for linux packet filtering.
JINMEI Tatuya [Thu, 25 Apr 2013 06:42:32 +0000 (23:42 -0700)]
[master] a trivial wording fix to changelog
Naoki Kambe [Wed, 24 Apr 2013 13:21:44 +0000 (22:21 +0900)]
[2252] simplify complicated nested try-except statements
get rid of the inner try-except statement, and increment a xfrsuccess and stop
a timer, or increment a xfrfail counter in the "finally" statement
Naoki Kambe [Thu, 25 Apr 2013 02:27:54 +0000 (11:27 +0900)]
[2252] update documentation for the test cases
JINMEI Tatuya [Thu, 25 Apr 2013 01:38:15 +0000 (18:38 -0700)]
[2903] unrelated style fix: fixed too-long line
JINMEI Tatuya [Thu, 25 Apr 2013 01:30:17 +0000 (18:30 -0700)]
[2903] catch some errors from asio methods/functions explicitly
otherwise it would result in an uncatchable exception and force the
program to terminate. some exceptions are still considered super rare
and fatal, and they are still propagated (mainly from the io_service
and deadline timer). errors on close() are just ignored (there
wouldn't be much we can do to recover in this case anyway).
Further, changed the handling of errors on the result of async_accept():
instead of stop accepting on some errors, log them and keep accepting
new connections. on a closer look at the code I believe that's more
appropriate; silent stop would be rather confusing (if the error were
really fatal we should rather invoke an exception and trigger termination),
and cases like EMFILE definitely shouldn't cause this situation.
errors that can caused by stop() should still result in stopping accept(),
and the new code behaves that way.
JINMEI Tatuya [Wed, 24 Apr 2013 17:15:33 +0000 (10:15 -0700)]
[master] Merge branch 'trac2834'
Michal 'vorner' Vaner [Wed, 24 Apr 2013 15:37:22 +0000 (17:37 +0200)]
[2871] Fix copyright dates
Michal 'vorner' Vaner [Mon, 8 Apr 2013 12:17:09 +0000 (14:17 +0200)]
[2871] Tweak the times and iteration counts
So even the naive approach terminates some day.
Michal 'vorner' Vaner [Mon, 8 Apr 2013 12:15:45 +0000 (14:15 +0200)]
[2871] Don't preserve pointers to stack
It seems the boost::function created from functor takes just the
reference. This led to taking address of temporary and failing later on.
Allocate new object manually and delete it after it was used.
Michal 'vorner' Vaner [Mon, 8 Apr 2013 12:14:46 +0000 (14:14 +0200)]
[2871] Use IOService::post
Instead of abusing the timer with 0 timeout, which is rejected.
Michal 'vorner' Vaner [Mon, 8 Apr 2013 11:37:55 +0000 (13:37 +0200)]
[2871] IOService::post()
To execute a callback later, from the event loop. This is to be used in
the fake resolution, instead of timer with 0 timeout (which doesn't
work, we are strict with asserting invalid parameters).
Michal 'vorner' Vaner [Mon, 8 Apr 2013 11:09:59 +0000 (13:09 +0200)]
[2871] Example naive resolver benchmark
The simplest example of a resolver using the fake resolution. Processes
one query at a time.
Michal 'vorner' Vaner [Mon, 8 Apr 2013 10:33:59 +0000 (12:33 +0200)]
[2871] The fake interface implemented
Michal 'vorner' Vaner [Fri, 5 Apr 2013 11:41:19 +0000 (13:41 +0200)]
[2871] Put forgotten files to git
Michal 'vorner' Vaner [Fri, 5 Apr 2013 11:05:20 +0000 (13:05 +0200)]
[2871] Implementation of the FakeQuery
The FakeQuery class is implemented. The FakeInterface is, however, still
missing, so it doesn't work (not even link).
Michal 'vorner' Vaner [Fri, 5 Apr 2013 09:23:03 +0000 (11:23 +0200)]
[2871] Interface of the fake query
Provide interface of the FakeQuery. That one will represent the work to
be done during a benchmark for single query.
Also provide interface for FakeInterface, which will generate the
queries and handle the event loop.
Michal 'vorner' Vaner [Thu, 4 Apr 2013 09:06:26 +0000 (11:06 +0200)]
[2871] Empty benchmark binary
Create all the necessary makefiles and other stuff, to have a place to
put the benchmarks to later.
Marcin Siodelski [Wed, 24 Apr 2013 09:51:07 +0000 (11:51 +0200)]
[2902] Added unit test checking reception of the packet over INET DGRAM.
Marcin Siodelski [Wed, 24 Apr 2013 09:32:24 +0000 (11:32 +0200)]
[2902] Test sending packet over INET DGRAM socket.
JINMEI Tatuya [Wed, 24 Apr 2013 06:42:17 +0000 (23:42 -0700)]
[2712] added lettuce test to confirm 'Cmdctl shutdown' now works correctly.
also confirmed it would fail without the change of this branch. the execution
of bindctl would fail, so bind10_control.py needed to be updated so it can
ignore failure of bindctl if specified so.
JINMEI Tatuya [Wed, 24 Apr 2013 06:40:07 +0000 (23:40 -0700)]
[2712] don't directly call command_handler() from http handler thread.
this would cause a deadlock for the shutdown command. while redundant,
always sending it to msgq should work and be safer.
JINMEI Tatuya [Wed, 24 Apr 2013 03:57:58 +0000 (20:57 -0700)]
[2712] check the running thread of cmdctl's command_handler().
only one thread can call it; checking this would reveal the defect situation
more explicitly. tests are slightly adjusted to pass.
JINMEI Tatuya [Tue, 23 Apr 2013 22:35:25 +0000 (15:35 -0700)]
[2879] adjusted lettuce tests that are affected by the fix of the branch.
note: this is not a regression. the original expected value has never made
sense, so this is actually a "correction". in fact, we should have noticed
the bug of this ticket when we noticed notifyoutv6 was counted 5 times even
it should be a successful notify response.
JINMEI Tatuya [Tue, 23 Apr 2013 22:31:57 +0000 (15:31 -0700)]
[2879] more documentation
JINMEI Tatuya [Tue, 23 Apr 2013 22:26:04 +0000 (15:26 -0700)]
[2879] (unrelated) cleanup: use common address,port format for logging
JINMEI Tatuya [Tue, 23 Apr 2013 22:25:11 +0000 (15:25 -0700)]
[2879] more documentation
JINMEI Tatuya [Tue, 23 Apr 2013 22:22:00 +0000 (15:22 -0700)]
[2879] fixed the main defect with revised tests.
JINMEI Tatuya [Tue, 23 Apr 2013 21:06:02 +0000 (14:06 -0700)]
[2879] added more test cases for notify_handler (timeout case)
with some cleanups:
- removed unnecessary setup
- use 192.0.2.1 instead of 1.1.1.1; former should be a bit better
- removed meaninglessly repeated test cases
- use faked time.time() to test timeout value more reliably
JINMEI Tatuya [Tue, 23 Apr 2013 19:20:00 +0000 (12:20 -0700)]
[2521] constify
JINMEI Tatuya [Tue, 23 Apr 2013 18:52:23 +0000 (11:52 -0700)]
[2521] style cleanups: folded a long line; constify variables
Marcin Siodelski [Tue, 23 Apr 2013 17:50:28 +0000 (19:50 +0200)]
[2902] More comprehensive comments in the packet receiving function.
Marcin Siodelski [Tue, 23 Apr 2013 17:42:00 +0000 (19:42 +0200)]
[2902] Parse incoming packet's headers on reception.
JINMEI Tatuya [Tue, 23 Apr 2013 17:27:29 +0000 (10:27 -0700)]
[2834] untabified
Marcin Siodelski [Tue, 23 Apr 2013 16:54:24 +0000 (18:54 +0200)]
[2902] Implemented a function decoding IP and UDP header.
Marcin Siodelski [Tue, 23 Apr 2013 15:50:08 +0000 (17:50 +0200)]
[2902] Implemented function which decodes Ethernet frame header.
Marcin Siodelski [Tue, 23 Apr 2013 12:37:21 +0000 (14:37 +0200)]
[2902] Implemented utility functions encoding UDP packets.
JINMEI Tatuya [Mon, 22 Apr 2013 22:00:28 +0000 (15:00 -0700)]
[2879] simplification: removed EVENT_NONE as it's simply unused
Marcin Siodelski [Tue, 16 Apr 2013 22:24:01 +0000 (15:24 -0700)]
[2902] Moved protocol utility functions to a separate file.
Marcin Siodelski [Tue, 16 Apr 2013 22:23:31 +0000 (15:23 -0700)]
[2902] Basic implementation to send packet over raw socket.
JINMEI Tatuya [Sat, 13 Apr 2013 06:50:19 +0000 (23:50 -0700)]
[2831] added a test (compile time in fact) for def of INITIAL_SIZE.
JINMEI Tatuya [Sat, 13 Apr 2013 06:32:02 +0000 (23:32 -0700)]
[2831] make sure resetting a segment before opening one
JINMEI Tatuya [Sat, 13 Apr 2013 06:31:29 +0000 (23:31 -0700)]
[2831] more doc about reader-writer conflicts
JINMEI Tatuya [Sat, 13 Apr 2013 06:05:24 +0000 (23:05 -0700)]
[2831] use boost file lock to detect any violation of read-write conflicts
this is done best-effort basis. update doc accordingly.
Paul Selkirk [Sat, 13 Apr 2013 03:40:15 +0000 (23:40 -0400)]
Merge branch 'trac2832'
Paul Selkirk [Sat, 13 Apr 2013 03:37:52 +0000 (23:37 -0400)]
[2832] roll back all of
946b1df except typo fix
JINMEI Tatuya [Fri, 12 Apr 2013 23:52:33 +0000 (16:52 -0700)]
[2831] clarified the check after repeated shrinkToFit.
JINMEI Tatuya [Fri, 12 Apr 2013 23:38:23 +0000 (16:38 -0700)]
[2831] corrected assert condition for overflow check
JINMEI Tatuya [Fri, 12 Apr 2013 23:31:24 +0000 (16:31 -0700)]
[2831] corrected a typo
JINMEI Tatuya [Fri, 12 Apr 2013 23:31:01 +0000 (16:31 -0700)]
[2831] more clarified when mapped setNamedAddress returns true.
JINMEI Tatuya [Fri, 12 Apr 2013 23:22:07 +0000 (16:22 -0700)]
[2831] in multiProcess test make sure two processes open segments in parallel.
JINMEI Tatuya [Fri, 12 Apr 2013 23:13:22 +0000 (16:13 -0700)]
[2831] reverted the change for deleting checkNamedData().
the order of deletion was intentional; BOOST_FOREACH for map wouldn't work
that way.
JINMEI Tatuya [Fri, 12 Apr 2013 23:09:21 +0000 (16:09 -0700)]
[2831] removed test for segment size after shrink.
I don't think the underlying boost shrink_to_fit() ensures this.
Thomas Markwalder [Fri, 12 Apr 2013 18:42:39 +0000 (14:42 -0400)]
[master] Added entry 604.
Thomas Markwalder [Fri, 12 Apr 2013 18:40:12 +0000 (14:40 -0400)]
[2785] Merge branch 'trac2785'. Corrects packet time statistics in
perfdhcp.
Mukund Sivaraman [Fri, 12 Apr 2013 14:44:21 +0000 (20:14 +0530)]
[2831] Also clear the named address before shrinkToFit()
Mukund Sivaraman [Fri, 12 Apr 2013 14:32:21 +0000 (20:02 +0530)]
[2831] Write the status to the waiting parent anyway
We should write it even if the test failed.
Mukund Sivaraman [Fri, 12 Apr 2013 14:29:23 +0000 (19:59 +0530)]
[2831] Reset the parent's segment before forking
Though we won't have two readers at the same time, it's best we
reset the segment so that it's not carried over for the child
to use accidentally.
Mukund Sivaraman [Fri, 12 Apr 2013 14:27:53 +0000 (19:57 +0530)]
[2831] Fix the segment that is used from the child
Mukund Sivaraman [Fri, 12 Apr 2013 14:23:04 +0000 (19:53 +0530)]
[2831] Close remaining pipe fd after using it
Mukund Sivaraman [Fri, 12 Apr 2013 14:19:46 +0000 (19:49 +0530)]
[2831] Use ASSERT_NE with pointer value
Boolean testing also works, but in the test it's easier to remember
that it's a pointer.
Mukund Sivaraman [Fri, 12 Apr 2013 14:11:04 +0000 (19:41 +0530)]
[2831] Check that the segment actually shrunk
Mukund Sivaraman [Fri, 12 Apr 2013 14:05:57 +0000 (19:35 +0530)]
[2831] Iterate over the source data instead of using redundant data
The order is now non-reverse (i.e., sorted order), but it is anyway
better we delete in the same order we insert so the first in is first
out.
Mukund Sivaraman [Fri, 12 Apr 2013 14:03:04 +0000 (19:33 +0530)]
[2831] Make minor comment updates
Paul Selkirk [Fri, 12 Apr 2013 01:27:25 +0000 (21:27 -0400)]
[2832] add value checking for 'cache-type'
Mukund Sivaraman [Thu, 11 Apr 2013 13:58:52 +0000 (19:28 +0530)]
[2381] Change more occurences of 0 to NULL
Tomek Mrugalski [Thu, 11 Apr 2013 12:52:21 +0000 (14:52 +0200)]
[2898] Documentation for v6 relays written
Stephen Morris [Thu, 11 Apr 2013 12:30:32 +0000 (13:30 +0100)]
[2467] Removed "raw" pointers from b10-dhcp4/6 tests
Tomek Mrugalski [Thu, 11 Apr 2013 12:05:25 +0000 (14:05 +0200)]
[2898] Support for v6 relayed traffic added.
Stephen Morris [Thu, 11 Apr 2013 10:55:01 +0000 (11:55 +0100)]
[2467] Removed "raw" pointers from dhcpsrv tests
Also tidied up a few comments.
Marcin Siodelski [Thu, 11 Apr 2013 10:45:10 +0000 (12:45 +0200)]
[master] Removed memory leak in dhcp4_srv_unittest.
Michal 'vorner' Vaner [Thu, 11 Apr 2013 09:32:36 +0000 (11:32 +0200)]
Merge #2877 (the rest)
Clean-ups so one doesn't have to pass the hash twice when deleting an NSEC3
record.
Michal 'vorner' Vaner [Thu, 11 Apr 2013 09:19:07 +0000 (11:19 +0200)]
[2738] Note about libraries
Michal 'vorner' Vaner [Thu, 11 Apr 2013 08:56:02 +0000 (10:56 +0200)]
[2738] Restructure the document
Restructure it so the levels are more separated, terms are defined, etc.
Marcin Siodelski [Thu, 11 Apr 2013 08:22:24 +0000 (10:22 +0200)]
[master] Added Changelog entry for #991.
Marcin Siodelski [Thu, 11 Apr 2013 06:53:09 +0000 (08:53 +0200)]
[master] Merge branch 'trac991'
Naoki Kambe [Thu, 11 Apr 2013 01:42:24 +0000 (10:42 +0900)]
[2252] treat an xfr request as success with an XfrinZoneUptodate raised
Naoki Kambe [Wed, 10 Apr 2013 10:36:22 +0000 (19:36 +0900)]
[2252] check an xfrfail incremented with multiple exceptions
check an xfrfail counter incremented exactly by raising intentionally
multiple expected failure exceptions in a row
Naoki Kambe [Tue, 9 Apr 2013 12:16:24 +0000 (21:16 +0900)]
[2252] update a fake class name to camel convention including a typo fix
Naoki Kambe [Wed, 10 Apr 2013 10:09:24 +0000 (19:09 +0900)]
[2252] make the references class-local: _orig_datetime, _orig_start_timer
Naoki Kambe [Wed, 10 Apr 2013 10:04:19 +0000 (19:04 +0900)]
[2252] remove exceptional cases when checking initial statistics of Xfrout
On my environment socket.unixdomain.open happened to count to 1
before, but now it no longer happen. If it happens in future again, we
need to investigate why it happens, and then fix a issue or add a
workaround.
Naoki Kambe [Wed, 10 Apr 2013 09:47:39 +0000 (18:47 +0900)]
[2252] remove some constant definitions from the class scope
- remove some constant definitions from the class scope:
master_addrinfo, name_to_counter, zones, ipver
- add getter properties and instance objects instead of this removal
- define these variables as private or class-local by adding one or
two underscores into their prefixes
Naoki Kambe [Tue, 9 Apr 2013 11:27:52 +0000 (20:27 +0900)]
[2252] revise description of zonename (add 'e.g.' before the zone classes)
Naoki Kambe [Wed, 10 Apr 2013 07:33:15 +0000 (16:33 +0900)]
[2252] update descriptions of last_ixfr_duration and last_axfr_duration
- revise to understand more easily
- express a failure case instead of using an implement detail as "timer"
- switch the order of last_ixfr_duration and last_axfr_duration
Naoki Kambe [Mon, 8 Apr 2013 05:53:20 +0000 (14:53 +0900)]
[2252] use 127.0.0.1:47809 as address and port of the primary server in lettuce
127.0.0.1:47807 seems to be used by another application in Mac OS X.
47808 is currently registered by IANA. As a result 47809 is here chose as a
new number. For details about the reason, please see a note:
tests/lettuce/configurations/DO_NOT_USE_127.0.0.1:47807.
JINMEI Tatuya [Wed, 10 Apr 2013 22:01:02 +0000 (15:01 -0700)]
[master] editorial: folded long lines.
JINMEI Tatuya [Wed, 10 Apr 2013 21:31:07 +0000 (14:31 -0700)]
[2831] make sure BOOST_NUMERIC_CAST_WOULDFAIL is set in all cases
also, corrected incorrect reset of CXXFLAGS
JINMEI Tatuya [Wed, 10 Apr 2013 21:12:58 +0000 (14:12 -0700)]
[2831] corrected a typo
JINMEI Tatuya [Wed, 13 Mar 2013 21:56:11 +0000 (14:56 -0700)]
[2831] corrected the position of BOOST_MAPPED_FILE_CXXFLAG
JINMEI Tatuya [Wed, 13 Mar 2013 21:55:30 +0000 (21:55 +0000)]
[2831] set -Wall and -Wextra to detect possible warnings more accurately.
JINMEI Tatuya [Wed, 13 Mar 2013 21:26:20 +0000 (14:26 -0700)]
[2831] allow skipping build sh-mem support. set -Wno-error if necessary.
JINMEI Tatuya [Wed, 13 Mar 2013 21:24:31 +0000 (14:24 -0700)]
[2831] detect buildability of managed_mapped_file, w/ or w/o -Werror.
Thomas Markwalder [Wed, 10 Apr 2013 21:02:10 +0000 (17:02 -0400)]
[2355] Initial commit of refactoring bin/dchp4 and bin/dchp6
configuration parsing classes into a common set of classes
in lib/dhcpsrv. Moved most of the parsers classes into
new files dhcp_parsers.h/cc. Eliminated use of global
variables within common parsers.