* Remove fullSize return from compressionLenSearch
This wasn't used anywhere but TestCompressionLenSearch, and was very
wrong.
* Add generated compressedLen functions and use them
This replaces the confusing and complicated compressionLenSlice
function.
* Use compressedLenWithCompressionMap even for uncompressed
This leaves the len() functions unused and they'll soon be removed.
This also fixes the off-by-one error of compressedLen when a (Q)NAME
is ".".
* Use Len helper instead of RR.len private method
* Merge len and compressedLen functions
* Merge compressedLen helper into Msg.Len
* Remove compress bool from compressedLenWithCompressionMap
* Merge map insertion into compressionLenSearch
This eliminates the need to loop over the domain name twice when we're
compressing the name.
* Use compressedNameLen for NSEC.NextDomain
This was a mistake.
* Remove compress from RR.len
* Add test case for multiple questions length
* Add test case for MINFO and SOA compression
These are the only RRs with multiple compressible names within the same
RR, and they were previously broken.
* Rename compressedNameLen to domainNameLen
It also handles the length of uncompressed domain names.
* Use off directly instead of len(s[:off])
* Move initial maxCompressionOffset check out of compressionLenMapInsert
This should allow us to avoid the call overhead of
compressionLenMapInsert in certain limited cases and may result in a
slight performance increase.
compressionLenMapInsert still has a maxCompressionOffset check inside
the for loop.
* Rename compressedLenWithCompressionMap to msgLenWithCompressionMap
This better reflects that it also calculates the uncompressed length.
* Merge TestMsgCompressMINFO with TestMsgCompressSOA
They're both testing the same thing.
* Remove compressionLenMapInsert
compressionLenSearch does everything compressionLenMapInsert did anyway.
* Only call compressionLenSearch in one place in domainNameLen
* Split if statement in domainNameLen
The last two commits worsened the performance of domainNameLen
noticably, this change restores it's original performance.
name old time/op new time/op delta
MsgLength-12 550ns ±13% 510ns ±21% ~ (p=0.050 n=10+10)
MsgLengthNoCompression-12 26.9ns ± 2% 27.0ns ± 1% ~ (p=0.198 n=9+10)
MsgLengthPack-12 2.30µs ±12% 2.26µs ±16% ~ (p=0.739 n=10+10)
MsgLengthMassive-12 32.9µs ± 7% 32.0µs ±10% ~ (p=0.243 n=9+10)
MsgLengthOnlyQuestion-12 9.60ns ± 1% 9.20ns ± 1% -4.16% (p=0.000 n=9+9)
* Remove stray newline from TestMsgCompressionMultipleQuestions
* Remove stray newline in length_test.go
This was introduced when resolving merge conflicts.
* Remove redundant parenthesis
These were caught with:
gofmt -r '(a) -> a' -w *.go
This commit only includes the changes where the formatting makes the
ordering of operations clear.
* Remove more redundant parenthesis
These were caught with:
gofmt -r '(a) -> a' -w *.go
This commit includes the remaining changes where the formatting does not
make the ordering of operations as clear as the previous commit.
This offset max was not taking into account leading Len() to emit a
smaller value that could not be matched by Pack(), i.e all names can
be fully compressed or used as a target for compression.
Split length tests off in seperate file length_test.go to clean up
dns_test.og a bit.
This was missing and generated the wrong code for TKEY; it adds a +1 to
the amount. This should happen (technically).
I think the fallout is not super bad (of the +1) as we allocate a byte
more for when pack a message.
* Add support for TKEY RRs
- make sure Key and Data fields are variable length hex fields
- checkin output from 'go generate'
- add a TKEY specific test to ensure this stays working
* go format changes
* address review comments
* add ability to parse TKEY via string
* handle review comments - change TKEY string output
Use :0 for loopback testing. This is more portable between testing environments.
Add testRR that calls NewRR and throws error away - apply it everywhere where needed.
It seems only Go 1.9 can deal with :0 being used. Disable 1.8 in travis.
Move some of them to Errorf and friends, but most of them are just
gone: This make go test -v actually readable.
Remove a bunch of test that used ipv6 on localhost as this does not work
on Travis.
* Generate the compressionHelper functions and fix compression.
This was a long standing TODO: generate the compression helper
functions. This now automatically picks up new names that can be
used for compression.
When packing add names to compression map:
When packing a message we should only compress when compress is true.
But whenever the compression map is not nil we should still add names
to it that can be *used* for future compression. The packing
inadvertently only added those names when compress would be true.
* Removed unused functions
Update the size-xxx-member tags to point to another field in the struct
that should be used for the length in that field. Fix NSEC3/HIP and TSIG
to use to this and generate the correct pack/unpack functions for them.
Remove IPSECKEY from the lib and handle it as an unknown record - it is
such a horrible RR, needed kludges before - now just handle it as an
unknown RR.
All types now use generated pack and unpack functions. The blacklist is
removed.
Remove the use of reflection when packing and unpacking, instead
generate all the pack and unpack functions using msg_generate.
This will generate zmsg.go which in turn calls the helper functions from
msg_helper.go.
This increases the speed by about ~30% while cutting back on memory
usage. Not all RRs are using it, but that will be rectified in upcoming
PR.
Most of the speed increase is in the header/question section parsing.
These functions *are* not generated, but straight forward enough. The
implementation can be found in msg.go.
The new code has been fuzzed by go-fuzz, which turned up some issues.
All files that started with 'z', and not autogenerated were renamed,
i.e. zscan.go is now scan.go.
Reflection is still used, in subsequent PRs it will be removed entirely.
Remove trailing \n from t.Log and t.Error messages as it's unnecessary.
In some instances, combine multiple t.Error()s into one
To provide more consistency across the tests, rename e to err and use %v
as the format arg for errors.
Replace Logf and Errorf with Log and Error when it made sense. For
example t.Errorf("%v", err) to t.Error(err)
If you have a system with large amounts of copies, these slice
allocations start stacking up. Use a shared slice and then subslice
them with a cap limit so that append works properly.
Also, add a benchmark and test for Msg.Copy
Benchcmp:
benchmark old ns/op new ns/op delta
BenchmarkCopy 1880 1672 -11.06%
benchmark old allocs new allocs delta
BenchmarkCopy 13 11 -15.38%
benchmark old bytes new bytes delta
BenchmarkCopy 528 528 +0.00%
IPSECKEY is kinda strange because it has a type selector which tells
what type a later rdata field has. The type can be a domainname, address
or v6 address. You sort of wish Go would have a union type for this, but
alas.
Currently this is implemented as:
GatewayA net.IP `dns:"a"`
GatewayAAAA net.IP `dns:"aaaa"`
GatewayName string `dns:"domain-name"`
In the IPSECKEY. Only one of these is active at any one time. When
parsing/packing and unpacking the value of GatewayType is checked
to see what to do.
Parsing from strings is also implemented properly and tested. The Unpack
function still needs work.
Instead of going through the fmt package, we can use append int,
which saves an allocation.
benchmark old ns/op new ns/op delta
BenchmarkUnpackDomainNameUnprintable 2147 506 -76.43%
When printing unknown records it is best to print the entire thing
as unknown, instead of relying on the internal defined type. An
example A record, printed as an unknown one:
miek.nl. 3600 CLASS1 TYPE1 \# 4 0a000101
Currently msg.Len() overshoots a _lot_ when compression is
enabled. The main problem is that once a part of a domain is found to
be compressed, the whole domain is _not_ added back to compression
dictionary.
That means if we once cache ".microsoft.com" we won't cache
"blah.microsoft.com".
Ammended a bit: put the test in dns_test.go
packLen() was a featureless mirror of Len(). Remove it, and just use
Len() internally too.
Fix bug in Len() too, where the length of the additional section was
not counted.
Tom Thorogood
Miek Gieben
spsholleman
Lars Lehtonen
Filippo Valsorda
Roland Shoemaker
Michael Haro
Daniel Morsing
Andrew Tunnell-Jones
Alex Ciuba