Git log: commit c677c5129a477aab2488b6a08243178ad8ce81c5
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Sun Mar 20 18:05:05 2022 -0600
Update debian metadata (Jool 4.1.8)
commit 76a5525829243f59f9e393a451b890bcdb2d75f6
Merge: cb03e887 6822bdee
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Sun Mar 20 17:52:07 2022 -0600
Merge branch 'master' into debian
commit cb03e887a774883908ff48b7a4463772c30ac55b
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Thu Jan 27 16:48:58 2022 -0600
Update debian metadata (Jool 4.1.7)
commit 0dd53806a1376003a316f8a02087d4c61772150a
Merge: 47cc9685 1c1267f2
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Thu Jan 27 16:36:46 2022 -0600
Merge branch 'master' into debian
commit 47cc968520ea0460bfbc8d5ed2c6407483705a36
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Fri Dec 10 15:39:38 2021 -0600
Update debian metadata (Jool 4.1.6)
commit 684cc0b35c76358de3704f98c0ef69c0fb50ae5f
Merge: c6c31697 e9e5c582
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Fri Dec 10 15:38:59 2021 -0600
Merge branch 'master' into debian
commit c6c31697e5adc29079bdc4da675d8a6869649574
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Fri Feb 19 17:09:21 2021 -0600
Update debian metadata (Jool 4.1.5)
commit e5d6cc7e8549b6aa3931fdd40076d2aef19a54f3
Merge: 63a61ad6 34ffbf56
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Fri Feb 19 17:05:52 2021 -0600
Merge branch 'master' into debian
commit 63a61ad6e3e108d31c59b416626a5ce055cf5baa
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Wed Oct 7 11:34:50 2020 -0500
Update debian metadata (Jool 4.1.4)
commit f27e1d5bfeb06ed682c732834708df8e6ea073cd
Merge: c283de78 b09c4903
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Wed Oct 7 11:30:48 2020 -0500
Merge branch 'master' into debian
commit b09c490342b8b7090aeb55022dd3c67bc9b89b60
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Tue Oct 6 11:02:36 2020 -0500
Protocolary updates for release 4.1.4
commit 026d45c28676757145e08afbf85abe5735f53d71
Merge: 9b2614ef 525ee6da
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Tue Oct 6 11:00:00 2020 -0500
Merge branch 'issue341'
commit 9b2614ef3333f2cad04ace86f65489383bae2075
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Mon Oct 5 11:24:33 2020 -0500
documentation: Current agreed changes for #345
commit 525ee6da6041121e392e2eb39fe6e5ed0ee199a5
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Thu Oct 1 16:49:10 2020 -0500
denylist4: Rename from blacklist4
Adds denylist, deprecates "blacklist."
First half of #341.
commit 79bd450d31da73de9adeb4aadd199e518c670cfb
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Thu Oct 1 11:22:55 2020 -0500
documentation: Update the intro to xlat
Adds basic explanations of SIIT-DC, SIIT-DC-2xlat, 464XLAT and MAP-T
to the xlat introduction. It also slightly improves the old basic
explanations.
Also the MAP-T dedicated page, which is still a WIP.
commit 972c2de00a09d9a817e7aa30d985bbac931507c5
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Wed Sep 30 14:28:14 2020 -0500
documentation: Update instance flag's pool4 examples
Fixes #343.
commit c3620137c5510047ab9e458dc70a1b12043e8098
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Fri Sep 18 09:55:23 2020 -0500
generic denylist: allow /32 addresses
Needed by L2 networks. (loopback is assigned /32, Linux responds
ARP requests, packet is translated despite the minimal addressing
and routing.)
Currently untested. Attempts to fix #342.
commit c283de787a82c5efafcc380d68c612f053e42143
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Tue Sep 1 16:02:16 2020 -0500
Update debian metadata (Jool 4.1.3)
commit 12288309089d8638c7e56bbb948410743485061b
Merge: a719338e 9b628265
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Tue Sep 1 15:57:33 2020 -0500
Merge branch 'master' into debian
commit 9b628265dc046bc61e1ca3a05b9d4cc9f0f34eee
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Tue Sep 1 13:35:51 2020 -0500
Protocolary updates for release 4.1.3
commit 35267a3886b1a84045ff53eb6cc75e39dd27b47b
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Sun Aug 30 11:56:18 2020 -0500
Issue #338: Bump version number
Just for testing purposes.
commit bd974032990670676c64feeaa301c19e4755d580
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Fri Aug 28 10:58:55 2020 -0500
siit: apply generic blacklist to dst address
So apparently, SIIT wasn't applying the generic blacklist to the
destination address of incoming IPv4 packets. Strangest is that, even
though this should cause untold mayhem immediately, I was somehow unable
to reproduce it for the longest time.
From my reading of the history, this bug first appeared during the
2019-10-30 RFC7915 review. Before that, the generic blacklist behavior
used to be
- Source address: Always enabled
- Destination address: Always enabled
RFC7915 wanted me to disable the generic blacklist for the source
address for ICMP errors, but for some reason what I actually did was
- Source address: Disabled on ICMP errors
- Destination address: Disabled on ICMP errors
Then, during the 7915 graybox testing of 2020-02-14, I ran into some
problem with this and nesciently reverted back to
- Source address: Always enabled
- Destination address: Always enabled
Then, during another graybox batch test on 2020-05-20, it became
- Source address: Disabled on ICMP errors
- Destination address: Always disabled
This commit changes into what I believe is the correct behavior (and
which is consistent with RFC 7915):
- Source address: Disabled on ICMP errors
- Destination address: Always enabled
This commit fixes the translator-router ping of #338, and also probably
the entirety of #339.
commit a719338eaa42179bcaff7fbae523c3278de10e63
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Tue Jul 21 16:00:51 2020 -0500
Update debian metadata (Jool 4.1.2)
commit ce35e46ce8e0b504829ea2a8162955aa897c6640
Merge: 308481e1 163b9f4d
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Tue Jul 21 15:57:01 2020 -0500
Merge branch 'master' into debian
commit 163b9f4d553d1d1ae7d3b319234791ea61df3db3
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Mon Jul 20 18:34:27 2020 -0500
Protocolary updates for release 4.1.2
commit 07279bfdaa4b24c7ce60dad231150497b50b1c7d
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Mon Jul 20 18:04:24 2020 -0500
iptables: Remove `static` from `_init()`.
F1xes #337. Solution provided by @yaoli-zheng-axcient.
commit 661e4960eed13719e4e7c641c893e3dec208f265
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Wed Jul 15 19:21:44 2020 -0500
Merge debug and trace configuration options
They had pretty much the same purpose, so I fused them into debug.
Debug always prints instance ID now, as well as the packet data
trace gave away.
commit c9a3d2d2b25036ad81b287fcfc93d7ae62b65b22
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Wed Jul 15 15:42:28 2020 -0500
Patch unit tests
The API changed during the debug refactors 2 commits ago
commit f647e14f5b311f6179cee105cb5989caf82afe94
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Tue Jul 14 15:33:03 2020 -0500
Patch compilation on CentOS 8
Tested on CentOS Linux release 8.2.2004 (Core).
F1xes #334.
commit c4f38e82cf48b135f509e0f31d0591b7200059e4
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Mon Jul 13 18:09:48 2020 -0500
Turn debug into a runtime switch
Got fed up with asking everyone to recompile.
commit bdab9612ffb2b548dd0d7d97a43101af9c205e66
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Wed Jul 8 11:13:47 2020 -0500
Patch a bunch of compile bugs reported by Fatih USTA
commit 308481e1c1345eed351e7493445a6cd9e22c2ba7
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Thu Jul 2 12:23:25 2020 -0500
Update debian metadata (Jool 4.1.1)
I'm also changing the Vcs URLs in hopes of getting rid of that
"X new commits since last upload, is it time to release?" bug.
Not sure if it's going to be successful, honestly. Nothing I've
done so far has worked.
commit 68a0553750c1bb4a15d2c374e8255ded339febb0
Merge: 470c256b fc1a6877
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Thu Jul 2 12:17:33 2020 -0500
Merge branch 'master' into debian
commit fc1a6877b4bbdb4b7e67c1c5d47c9a775eb92771
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Wed Jul 1 11:21:53 2020 -0500
Protocolary updates for release 4.1.1
commit 212acb49c48ec5f155ccc9cf6f629c7fc45146c4
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Mon Jun 29 15:03:24 2020 -0500
GRO: Patch GRO in the IPv4 to IPv6 direction
Second half of #331.
commit eba6df146b0168b4a2e30c160261a2496d2d4a49
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Sat Jun 27 01:34:58 2020 -0500
pool4: patch port management in stateful NAT64.
Header ports are supposed to be inferred from the outgoing tuple,
not from the flowi. When pool4 premature routing kicks in (ie.
when it's not populated), flowi ports don't have much meaning.
Fixes #332.
commit b8c20a35e2fedfb823a62e18ea998cb8475a4dfc
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Fri Jun 26 19:04:17 2020 -0500
GRO: Patch GRO in the IPv6 to IPv4 direction
I still don't know if IPv4 to IPv6 also needs a patch.
In case it does, this is only the first half of the #331 fix.
commit b00265dfa0b4a09b62c32be034e9e727fc7dcee5
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Wed Jun 24 15:25:12 2020 -0500
Documentation: Patch release leftover noise
commit 470c256bba93213204f37e92ad16e0959c3211f4
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Fri Jun 19 18:04:52 2020 -0500
Update debian metadata (Jool 4.1.0)
commit 70631f8bfffe4b0ba272d451f7070ae2d7a8bd6e
Merge: 6272cf33 b5f4cb55
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Fri Jun 19 17:58:43 2020 -0500
Merge branch 'master' into debian
commit b5f4cb55f933929085e6daccac7b2e852cb4fb42
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Fri Jun 19 12:12:06 2020 -0500
graybox: Test addendums inspired by release protocol
commit c3a482b69595214c4ca6b71cf752771aef0e0b66
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Thu Jun 18 17:56:07 2020 -0500
Patch empty pool4 bug
commit 1118be57d9d3e89dab6c90103fe5c69fdd699f54
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Tue Jun 16 16:31:13 2020 -0500
Protocolary updates for release 4.1.0 (unfinished)
Added support for the latest stable kernel (5.7) as part of the
paperwork. (Fixes #329.) Tested on 5.7.2-050702-generic.
Testing revealed that empty pool4 is broken. Still working...
commit 539a65e11b92dd550d87795e38d812e7b87cf562
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Mon Jun 15 19:56:28 2020 -0500
Purge `force-slow-path-46`
Turns out it doesn't work as intended. allocate_slow() assumes the
resulting packets will have fragment headers, which is not always
true when the user wants to force slow path.
But removing the assumption would require further modifications and
testing. I don't think the help we might get from the flag is worth
the effort. Users can downgrade to 4.0.9 if things go south.
commit b17353d7ac183ecb9a0084a2498526ac853ba51b
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Mon Jun 15 17:09:46 2020 -0500
Add `lowest-ipv6-mtu` documentation
commit 1db50109a3360b68eff13cca37a2a66b687802bf
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Fri Jun 12 16:07:55 2020 -0500
graybox: NAT64 test suite review, part 3
- Review, restore and/or document all the remaining broken NAT64
tests
- Patch the refcount bug from the previous commit
I'm finally done. Jool 4.1.0 is ready for release.
commit 04076bac2edc9128937735adcc7d4a2bc8820d91
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Thu Jun 11 11:44:28 2020 -0500
graybox: NAT64 test suite review, part 2
- review, patch and restore several commented broken tests
- move clients to dedicated namespaces so the global routing table
doesn't interfere with some tests
- the graybox kernel module is no longer married to the global network
namespace
- remove bashisms (tested with dash)
There's still a broken reference count somewhere. Still looking.
commit c439fc4c3aaa4e6eaad2c2883bd850f46f5a51e9
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Tue May 26 14:18:57 2020 -0500
graybox: NAT64 test suite review, part 1
- Patch endian bug affecting TCP/UDP ports on NAT64
- Patch diff bug on graybox; fields that were not byte-sized were not
being compared correctly
- Update and document several tests
The existing NAT64 test suite no longer reports errors (assuming the
global network namespace lacks default routes), but there are still
several commented tests worth reviewing.
commit 7848ad4f6d13322fbb83f3f4c68656139ce73f88
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Wed May 20 18:16:18 2020 -0500
graybox: SIIT test suite review
All the SIIT tests have been reviewed, updated and documented.
Bugs found and fixed:
- Hairpinning + empty pool6791 used to cause packet drop
- Redesigned the ha test (RFC requirement was misunderstood)
The NAT64 tests are next.
commit 2a8d0adcce40c25f2e5883c572f4b0664f080c4b
Merge: c188c1b7 3e1e8fd1
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Mon May 18 16:51:41 2020 -0500
Merge branch 'master' into issue136
commit c188c1b77ff7c804293afb7a587dbc2901f333ce
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Fri May 15 15:13:09 2020 -0500
graybox: Big cleanup and refactor
- Improve the packet diff:
- Instead of showing differing bytes, the output now prints
differing header fields with proper formatting. This leads to
faster debugging.
- The diff is also smarter in that it now validates IPv4 header
checksums manually, so random identifications do not prevent
their checking.
- Print sending and expected packet names on errors.
- Document the old "pktgen" tests. (See pktgen.md)
- Update several still relevant old tests.
- Delete some old and obsolete SIIT tests. (See below)
- Finish the SIIT network description file, add NAT64 network
description file.
I'm not done; the tests still need some cleaning, particularly the NAT64
ones.
Each deleted "old and obsolete SIIT test" is listed below.
Since they were never documented, and for lack of a better place, I will
describe them in addition to stating why I'm deleting them:
-- igmp64 --
Sends an IPv6 packet containing unknown layer 4 data.
Expects an IPv4 packet containing the same unknown layer 4 data.
The "unknown layer 4 data" uses the IGMP nexthdr, but it's just a small
monotonically-increasing payload. (Jool doesn't care.)
Made redundant by 7915.f*.
-- igmp46 --
Same as igmp46, except in the IPv4 -> IPv6 direction.
Also made redundant by 7915.f*.
-- frag-icmp6 --
Sends an ICMPv6 error sized 1396 (40+8+40+8+1300).
Expects an ICMPv4 error sized 576 (20+8+20+8+520).
Tests ICMPv4 error truncation. (Name is misleading.)
Made redundant by 7915.be.
-- frag-icmp4 --
Sends an ICMPv4 error sized 1356 (20+8+20+8+1300).
Expects an ICMPv6 error sized 1280 (40+8+40+8+1184).
Tests ICMPv6 error truncation. (Name is misleading.)
Made redundant by 7915.bb and 7915.bc.
-- frag-minmtu6-big --
Sends an IPv4/UDP packet sized 1428 (20+8+1400). DF is disabled.
Expects two IPv6 fragments sized 1280 (40+8+8+1224) and 224 (40+8+176).
Tests `lowest-ipv6-mtu`. (Which is why it's been disabled all this
time.)
Now that `lowest-ipv6-mtu` is implemented, test succeeds.
Made redundant by 7915.j*.
commit 1849b01e026cd1c760f206c3430a243cd1fb1cc4
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Fri May 15 13:32:09 2020 -0500
Design and implement tests jb* and jc*
No errors found.
commit 7f72f85d45f2d7b1c0698a4eac2f62a7421036f1
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Wed Apr 22 12:43:19 2020 -0500
Design and implement graybox test ja
commit 6272cf337ffda356936d635e5d574506d0d96a96
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Wed May 6 15:49:12 2020 -0500
Update debian metadata (Jool 4.0.9)
commit 1875ba88088041b8315998c34bb6b65758adf333
Merge: 85d81a15 3e1e8fd1
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Wed May 6 15:44:23 2020 -0500
Merge branch 'master' into debian
commit 3e1e8fd1e2230a11e938d5bd04ec32d2c9b1ceeb
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Tue May 5 13:28:32 2020 -0500
Protocolary updates for release 4.0.9
commit b846a1941d1ac8c25e9a7609bf42b5795553baa6
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Tue May 5 10:22:59 2020 -0500
Testing
- Patch graybox module compilation
- Return optionalness of pool6 during SIIT instance add
Old topics:
From fad9dc986a185581d6c316d823037f8537557816:
> - Add NLA_F_NESTED to all nested attributes from userspace.
> Might need to be mirrored in kernelspace for the sake of
> future-proofness.
Newer kernels already add NLA_F_NESTED automatically. There's no need to
mirror the userspace patch.
From 388e2dda81deb5e1c6e4a8891ac58bbaaa798f4c:
> I'm still investigating whether this is the right way to fix #325
The #325 issue only seems to show up when "pkglib" libraries depend on
other "pkglib" libraries. Maybe there's something about the "pkglib"
library definition that warrants this, and it's simply going over my
head.
For now, leaving them all as "noinst" libraries is fine.
commit 388e2dda81deb5e1c6e4a8891ac58bbaaa798f4c
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Wed Apr 29 15:24:34 2020 -0500
autotools: Convert all shared libs to convenience libraries
Fixes #325.
I'm still investigating whether this is the right way to fix #325,
but it does have an interesting side effect: it also fixes the library
install warning ("libtool: warning: relinking 'XXX'") which has been
haunting me clueless for months.
What I don't like is that a "convenience library" is really not a
library at all, but rather a bunch of objects that get merged directly
into the programs at compile time. So I don't get to reuse functionality
at runtime. But maybe that's what GNU wants.
This commit also makes all LDADD/LIBADD references relative, and all
CFLAGS include references absolute. I don't really have a good reason
for imposing this, other than it being the textbook way. (I'm reading
Chapter 7 of "Autotools, 2nd Edition: A Practitioner's Guide to GNU
Autoconf, Automake, and Libtool" by John Calcote).
See https://stackoverflow.com/questions/61533589
commit fad9dc986a185581d6c316d823037f8537557816
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Mon Apr 27 18:01:20 2020 -0500
netlink: Update API usage again
- Add NLA_F_NESTED to all nested attributes from userspace.
Might need to be mirrored in kernelspace for the sake of
future-proofness.
- Remove NLA_UNSPEC, since apparently it now means "murder me
cold-blooded" instead of "don't worry about me."
- Reimplemented optional prefix serialization and deserialization
because the kernel doesn't like empty nested attributes.
I don't understand why the client tests didn't detect these issues
earlier. Need to review the protocol.
Probably broke the API usage on lower kernels, so won't collapse
to master yet.
Fixes #326.
commit a16b5274b18c478d206d9a12b66bd1f0b166cf20
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Wed Apr 22 12:20:36 2020 -0500
graybox: Redesign and implement all id tests
commit caf6285dd5afcb4fe0b56eae290c97ff438bc427
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Mon Apr 20 17:59:14 2020 -0500
Implement graybox tests ic5, ic6, id1, id2 and id3
(ie. more ICMP extension tests)
Bugfix: Old code wasn't adding padding when outgoing internal
packet length < 128 and ICMP extension was meant to be removed.
commit 2deb5efbe11fc6d07d64b4a7e80f1eec840e4704
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Thu Apr 9 18:40:38 2020 -0500
Implement graybox tests ic1, ic2, ic3 and ic4
(ie. the first four tests of ICMP extensions)
Bugfix: Old code wasn't adding padding when outgoing internal
packet length < 128.
commit 088bbff45ddea851e43bb6b33575c66f3c2896e9
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Tue Apr 7 15:39:59 2020 -0500
Patch graybox tests
There was only one error.
commit a0cfa6f284e6080df0f0545dce581ef2485c42e0
Merge: 1a4eaac7 6cd661f7
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Thu Apr 2 12:47:18 2020 -0600
Merge branch 'issue136'
commit 85d81a157618ec2b68fac9a44697b95ca47f11b7
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Tue Mar 31 14:11:53 2020 -0600
Update debian metadata
commit f14829a3c52264fac304b4bcf723e28694d777ed
Merge: b858801a 1a4eaac7
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Tue Mar 31 12:24:17 2020 -0600
Merge branch 'master' into debian
commit 1a4eaac769e2c82ed0b98f6ec1a8be7ad0f2bef8
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Fri Mar 27 15:21:26 2020 -0600
Protocolary updates for release 4.0.8
commit 2a3e09910e2db9ac34932a9828bbdafeb525089a
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Thu Mar 26 14:42:06 2020 -0600
Code review and testing
Approved for release.
commit d41e3dc4852435eca20b8abbae59b9b6a61d567a
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Tue Mar 24 16:47:54 2020 -0600
Code review and testing
I've been patching so many small things, I can't keep track of
them. Sorry these commit messages aren't very helpful, but this is
all just a gargantuan grind.
commit b541c4cc16d7b05be9796ceca6d0b04d1a54581a
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Wed Mar 18 20:29:10 2020 -0600
Code review
Mostly just moving things around.
But also, the previous commit had what appears to be a fatal
compilation "warning," which wasn't caught by any of the tests...
It's patched now.
commit 9e9a636b6ce8398f72e3b195819ed00963843c72
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Sat Mar 14 12:25:44 2020 -0600
Restore successful compilation on all supported kernels
commit a687860b0466bc51e37abd85758bf1c70b160c16
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Thu Mar 12 15:10:19 2020 -0600
Code review
- Hardening up (adding validations)
- Improving the joolnl library's API since there's apparently a new
user under development:
https://mail-lists.nic.mx/pipermail/jool-list/2020-March/000336.html
commit f144ae519f01956077b1e0ccf0d31dc50a333864
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Wed Mar 11 10:27:09 2020 -0600
netlink: add joold
Issue #322: All of the Netlink code has been migrated to the
Netlink attributes model.
Could use some testing still.
commit 277e7ed39ace4562406a93ed3f3e2263ab413f7e
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Fri Mar 6 22:28:06 2020 -0600
Patch compilation on all supported kernels
Except for RHEL-7.0
commit 286fe5a9003282579520a41d18861962cfbf0c59
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Fri Mar 6 20:01:40 2020 -0600
netlink: add atomic configuration
Complements the previous commit.
Still missing joold.
commit c3dd25a571c33b61ebe924c641d3fbfe19c2edba
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Wed Mar 4 19:46:03 2020 -0600
Complete rewrite of the Netlink code
Replaces the old model (which exchanged potentially padded
structures between kernelspace and userland) with the Netlink
Attributes one.
Fixes #322, though it's missing atomic configuration and joold.
commit 6cd661f78308f66dc75183c605e3114380582004
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Mon Feb 17 13:39:31 2020 -0600
Implement translation of ICMP Extensions (IE)
Where "translation" of an IE simply means "treated as an opaque
but indivisible bit string meant to be either copied to a rightful
place or chopped off completely." The ICMP header length is also
updated accordingly.
Previously, Jool used be unbeknownst of IEs, which means it
treated them like the rest of the ICMP error payload (trimming it
arbitrarily sometimes), and zeroizing the ICMP header length.
I keep forgetting that these requirements exist because they are
positioned akwardly in RFC 7915. This is work that should have
been implemented as part of #221.
This code is untested still.
commit b970c7cd86c699a95d3403c13c0ed1d77d380d18
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Fri Feb 14 15:46:04 2020 -0600
Complete RFC 7915 graybox tests
(Taking into account that lowest-ipv6-mtu is now implemented.)
Swatted a few bugs in the meantime.
commit de9f370149afe7eaf08239e9b1c9b4b358f60b12
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Wed Feb 12 12:13:49 2020 -0600
Look at all these bugfixes
The very old API of __ip_select_ident() has become overly
cumbersome, so I decided to drop support for kernels 3.13, 3.14,
3.15, RHEL 7.0, RHEL 7.1 and RHEL 7.2.
I think this is fine. 3.16 is the oldest longterm kernel right now.
commit 4727147974827768ef045748c9c10fa7328b6e19
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Tue Feb 11 18:10:47 2020 -0600
Optimized further
Code checkpoint; still needs more testing.
commit 240270717af047d9983459cd4997cac135ebaabf
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Fri Feb 7 13:39:45 2020 -0600
Patch submodule deinitialization order
Also officially adds support for kernel 5.5. Tested on
5.5.0-050500-generic.
Fixes #320.
commit 54b9aaa78a1bdd15fd95c836b38480eb1172ebeb
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Wed Jan 22 17:11:42 2020 -0600
Still testing and thinking
I think I can optimize it more, but I need to analyze the kernel
code further.
commit fde105c102f331f6e1bc9e30e3277ee87bdec755
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Tue Jan 21 15:20:50 2020 -0600
Bunch of testing
commit 07a9c95130e09eacefbdc71e945dfe1fbb7ae0da
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Wed Jan 8 17:02:38 2020 -0600
Documentation: Add Arch Linux dependencies
The Raspberri Pi dependencies are the same as Debian's (and it
looks like they fixed that old weird headers package bug), so I've
decided to obviate it.
commit 50a4920a47222ab185eee0fdab1101de8bd02dfb
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Mon Jan 6 17:43:24 2020 -0600
Improve documentation based on latest feedback
- Always warn that package binaries may conflict with compiled
binaries (fixes [0])
- Add pkg-config dependency to Ubuntu's from-source installation
recipe (fixes #316)
- Review dependencies of from-source installations for Debian,
Ubuntu, CentOS and openSUSE. (TODO: Arch, Raspberri)
- Shrink survey window (hardcoded height was wasting space)
[0] https://mail-lists.nic.mx/pipermail/jool-list/2020-January/000312.html
commit be53fe41d950e2e0b0c1bc73832587d1abce7dc5
Merge: da67695e 76c12343
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Mon Jan 6 11:29:13 2020 -0600
Merge branch 'master' into issue136
commit 76c123433f501b0cef348d98ca0aa612a9046fa8
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Sun Dec 29 21:45:26 2019 -0600
Improve trace based on latest feedback
Add instance stateness and namespace, as well as ICMP ID.
commit 0fada0fc17a39f3417dc821b0cbbd4cb79a1c987
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Fri Dec 20 17:29:24 2019 -0600
Add the `trace` global configuration option
commit 1aaf79e447e30de8bb15671470f3d52f8a286880
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Fri Dec 20 12:55:51 2019 -0600
Update the logging documentation
It was still using Kbuild instead of DKMS.
commit 866779a2d1b7dc7897768db9ad590e32965b2de7
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Wed Dec 18 15:28:31 2019 -0600
Add uninstallation steps for source-compiled binaries
More or less fixes #312.
commit b858801a4f13f5b5be0adb495c90999cea20ee26
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Tue Dec 17 15:42:17 2019 -0600
Initialize debian branch properly
Patches an error I'm getting from the Debian Package Tracker.
commit 4ff45fc13aef6779cb207847b76177133246b555
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Tue Dec 17 13:15:55 2019 -0600
Protocolary updates for release 4.0.7
commit 78d5424a2fe04351c4c040d3db6d3fe0e55469b4
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Mon Dec 16 12:23:50 2019 -0600
Kbuild: s/SUBDIRS/M/
M was always a modernized version of SUBDIRS, and SUBDIRS was
apparently dropped recently.
commit ef48ec3b09879f48f8004ea994ec9c3cab3a56af
Merge: cfa601aa 0343b851
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Mon Dec 16 11:37:23 2019 -0600
Merge branch 'issue221'
commit da67695eb4d4560d70206b754e111ae11e28f554
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Mon Dec 16 11:36:33 2019 -0600
WIP
commit 535d83d1d0c82274e5d549dc380c00fd2b3ffff1
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Fri Dec 13 17:40:55 2019 -0600
Add --lowest-ipv6-mtu
First development version; untested.
Fixes #136.
commit cfa601aa012940326dc6632f11ee67836312d204
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Tue Dec 10 09:55:00 2019 -0600
Print iptables-save properly
Patch courtesy of @fatihusta.
Fixes #311.
commit 61a8bb49eca2a4f303f6c44cec758ec045163922
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Mon Dec 9 13:47:44 2019 -0600
Fix kern/usr communication protocol structure sizes
Redos the previous commit, because those structs were made to work
properly for the wrong reasons.
This time, the fix also includes the patch to the struct producing
the iptables mayhem.
Fixes #310.
commit d39e1d53d3b22fcdff9e54f95d2341c26d65c5c2
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Fri Dec 6 15:30:53 2019 -0600
Fix kern/usr communication protocol structure sizes
Some fields and structures were not aligning on 32-bit
architectures the same way as they did on 64-bit architectures.
This caused communication mayhem when kernel and userspace
bittage was different.
Fixes the instance display half of #310. The iptables problem
probably persists.
commit 553d16f98237553f8b422b5256f397fefeb68f21
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Mon Dec 2 13:16:27 2019 -0600
Add survey for 2020 development
