vcswatch reports that
this package seems to have new commits in its VCS but has
not yet updated debian/changelog. You should consider updating
the Debian changelog and uploading this new version into the archive.
Here are the relevant commit logs:
commit e0c2238df3c4c41b0346955a8748b72866b7b272
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Wed May 21 13:27:52 2025 -0600
Move the source copying to `execute_after_dh_install-indep`
3rd attempt to fix Debian#1105183.
I don't understand this at all. Quoting Beckmann:
> Since the override target does not call dh_install on jool-tools during
> an arch:all-only-build, it cannot record the files that it does *not*
> install and therefore dh_missing complains.
I guess "-Njool-dkms" doesn't mean "not jool-dkms", but rather,
"not jool-dkms, except on arch:all-only, which is also excluded"...
Credit to Santiago Vila and Andreas Beckmann for this solution.
commit 7d176357f717a0aef0a03a0345e6832a9cf20d8e
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Wed May 21 13:18:05 2025 -0600
Fix Lintian warning "old-fsf-address-in-copyright-file"
This license disclaimer was pointing to an outdated address.
commit 297e70e6a65e6a78c76b087ddc734172ae501b30
Merge: 56ee409 8129fd7
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Tue May 20 15:40:02 2025 -0600
Merge branch 'main' into debian
commit 8129fd716a84940bef48930dc4c4b255ee47e28a
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Tue May 20 15:39:05 2025 -0600
Update the README's license disclaimer
It was pointing to the FSF's old postal address.
Nowadays it's supposed to point to the website.
commit 5be6a6b770b0dc9d66fd98c62110ae68564559d6
Author: aiyion <git@aiyionpri.me>
Date: Sun May 18 22:39:30 2025 +0200
refactor: Alter shebangs to env bash and /bin/sh
which allows to reconfigure the shell via environment variable PATH.
In scripts where no bashisms were used, use /bin/sh.
Co-authored-by: Alberto Leiva Popper <ydahhrk@gmail.com>
commit 56ee409d50c8b621cbf51a7e4b533c9de4a10f3f
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Tue Apr 29 13:09:32 2025 -0600
Remove wildcards from installs
Guess they're not supported anymore.
commit 4071087b5f50f2c2749b1e0c04d58611a3964136
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Thu Apr 17 11:51:00 2025 -0600
Raise debhelper-compat to 13
Feedback from Phil Wyett from Debian Mentors.
The new dh_missing revealed the manpages were being installed from the
wrong source, so that's fixed too.
commit 2b2bc4bc71795d125dd1ab72b0d4419970bae197
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Wed Apr 16 22:48:43 2025 -0600
Raise standards version and list random file licenses
Feedback from Phil Wyett from Debian Mentors. (Except Lintian.)
Some of these files contain more than one license:
- aclocal.m4: 9 different Copyright statements (19 counting duplicates),
but all referring to the same license.
- configure: FSFUL, GPL-2+ with Libtool exception
- ltmain.sh: GPL-2+ with Libtool exception, Expat, GPL-2+
- m4/libtool.m4: FSFULLR, GPL-2+ with Libtool exception, FSFUL
Debian's copyright file does not allow this, so I prioritized the
headermost license in every case.
commit 3c7c4534bb5f38f7358a2db75c53be70633eff56
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Wed Apr 16 15:43:46 2025 -0600
Update debian metadata (Jool 4.1.14)
commit 4b1c5f0c2600bb560bc8340dff9d3d36fa5eece2
Merge: 4fe1c68 5afde37
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Mon Apr 14 18:26:43 2025 -0600
Merge branch 'main' into debian
commit 4fe1c68df8f081eab29e0a02b1ddd404c05ddc47
Author: Andreas Beckmann <anbe@debian.org>
Date: Thu Feb 20 11:11:37 2025 +0100
Remove -dkms suffix from dkms module name
Having -dkms in the module name is highly unusual, i.e. jool is the only
-dkms package in the Debian archive doing this.
Signed-off-by: Alberto Leiva Popper <ydahhrk@gmail.com>
commit 4f39d47ab624017b75a7c0ab9934c092422fa721
Author: Andreas Beckmann <anbe@debian.org>
Date: Mon Feb 24 18:42:36 2025 +0100
dkms.conf: Set BUILD_EXCLUSIVE_KERNEL_MIN="4.12"
the 5th argument to nla_validate() was added in Linux v4.12
Signed-off-by: Alberto Leiva Popper <ydahhrk@gmail.com>
commit c9f110957b8f970a6c34ac4b8ff5edc82cc49210
Author: Andreas Beckmann <anbe@debian.org>
Date: Thu Feb 20 12:21:56 2025 +0100
dkms.conf: Fix CLEAN command: use the headers that were used for building
Signed-off-by: Alberto Leiva Popper <ydahhrk@gmail.com>
commit fc8969b48f7833c0bdcd77cccec653228246243e
Author: Andreas Beckmann <anbe@debian.org>
Date: Wed Mar 5 11:00:05 2025 +0100
release jool (4.1.13-1.1) to unstable
Signed-off-by: Alberto Leiva Popper <ydahhrk@gmail.com>
commit 494ac02cb618ed7a8196c6886cd5cf1ce0a57509
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Fri Aug 23 22:11:56 2024 -0600
Update debian metadata (Jool 4.1.13)
commit ba21c277f008e8084c2ca780ca366a802934f5d2
Merge: b246681 39ca69f
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Fri Aug 23 22:10:02 2024 -0600
Merge branch 'main' into debian
commit b246681e098f47970de0230475ee68df58a2f79a
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Sat Jun 22 22:56:58 2024 -0600
Update debian metadata (Jool 4.1.12)
commit 00fea64d31529ac28b635b14a4fb954c95b2569a
Merge: 863f91e 8d2edff
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Sat Jun 22 22:56:28 2024 -0600
Merge branch 'main' into debian
commit 863f91efce76fe753d64a10ba7dafc1f47e91dde
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Sun Dec 24 14:57:59 2023 -0600
Merge 4.1.10-2 and 4.1.11-1 in the changelog
Because 4.1.10-2 was never actually uploaded to Debian, and it's causing
some sort of conflict.
commit 33105892b91989bd52ae32da37994398ed41d9ec
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Sat Dec 23 16:37:50 2023 -0600
Update debian metadata (Jool 4.1.11)
commit 4cf7f5a8126ae1745061635048d30a554d9c910c
Merge: 586dfc8 aed966a
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Sat Dec 23 16:31:31 2023 -0600
Merge branch 'main' into debian
commit 586dfc81857bfcb02b50cdf2d11ed49bd48b5b5b
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Thu Aug 31 09:33:50 2023 -0600
Move jool service startup to After=network-pre.target
The old After=network.target causes deadlock when users attempt to start
the service during network initialization.
Fixes #409.
commit 784c6ddb275af6483c1dd3831ad04c2a18a243ba
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Mon Jul 24 17:09:56 2023 -0600
Update debian metadata (Jool 4.1.10-2)
commit d1bb57b04999bbc908d3c9d9208ab50d1dd0799b
Merge: 8718568 aa511f8
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Mon Jul 24 16:59:05 2023 -0600
Merge branch 'main' into debian
commit 87185688e7111e24784ea8245a5c3114e51295f0
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Sun Jun 11 17:34:20 2023 -0600
Update debian metadata (Jool 4.1.10)
commit eb18a45f9370680308dd9a7c83f8970e8f3e118e
Merge: 4aba5bc 47334c9
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Sun Jun 11 17:28:47 2023 -0600
Merge branch 'main' into debian
commit 4aba5bcba3753a972d7dcbd7f110ca1a74c36e53
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Fri Jan 27 10:26:02 2023 -0600
Update debian metadata (Jool 4.1.9, second attempt)
commit cf61641db3539f46b5df87baa2978bf12b6aa965
Merge: 61d64a0 490ddb0
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Fri Jan 27 10:25:43 2023 -0600
Merge branch 'main' into debian
commit 61d64a034c5860e37eccbfcb1b9d8503a42125ef
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Wed Jan 25 21:05:55 2023 -0600
Update debian metadata (Jool 4.1.9)
commit d709a326c85d52f345ba36c764ad4b5b0ad398f6
Merge: c677c51 7831ff3
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Wed Jan 25 20:02:55 2023 -0600
Merge branch 'main' into debian
commit 7831ff3fe87ce6c7a0e3688a4be37880da177dd8
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Wed Jan 25 15:19:38 2023 -0600
Protocolary updates for release 4.1.9
commit 0b5952e834da3ff1162896635776a5d7a4a79ef8
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Tue Jan 24 18:06:25 2023 -0600
F1x #378
Finally.
It was just uninitialized memory showing up as a nonzero initial
counter. Presumably, it only happens when packaged because of the
optimization flags Debian defaults.
And, because the bug normally only appears when packaged, this bug
should not be closed until the release is out.
READY FOR RELEASE!!111!!1!!!11!!!!1oneone
Ugh
commit c2962a736b7678377cb8952d1faa26f6256e715a
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Mon Jan 23 10:43:46 2023 -0600
Netlink: Allow some backward compatibility
New implementation: Major and Minor version numbers must match, but
Revision and Development can be anything.
Rationale:
- Because it pretty much by definition involves new tables, Major
changes involve tweaks that pretty much guarantee communication
incompatibility. Therefore, binaries that do not share Major should
refuse to talk to each other.
- Minor is a gray area, and should probably be evaluated on a
case-by-case basis. But that's outside the scope of this commit, and
the default behavior should be conservative.
- Revision only signals bugfixes that rarely tweak the Netlink code, so
it's probably safe to ignore it.
- Development doesn't matter because I'm the only one who uses it.
Proposed fix for #347.
commit 6dfa2f6957b330584782309d488134ca8855cebc
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Thu Dec 29 13:37:56 2022 -0600
Allow < 128 ICMP extension'd packets
Some broken routers generate ICMP errors containing ICMP extensions,
while also carrying internal packets that length < 128 bytes.
Though these packets are clearly illegal according to RFC 4884, we
shouldn't break traceroutes over it. Jool will fix them during
translation, adding the mandated padding.
Fixes #396.
commit c95e211ff3ff8550bdcbe8952907a75ced3de0bf
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Wed Dec 28 11:52:37 2022 -0600
Remove traces of the master branch from the documentation
Branch "master" was renamed to "main" a while ago; update remaining
links and mentions.
Also patches some old/broken links.
commit c1b64e9eae3a86afc6e3dfee547859008d894892
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Wed Dec 28 10:54:34 2022 -0600
Move content of the LICENSE file to the README
The LICENSE file (which is just the GPLv2 minidisclaimer) is confusing Github,
because it thinks it's a separate license:
Licenses found
Unknown: LICENSE
GPL-2.0: COPYING
So, moving the minidisclaimer to the README, deleting LICENSE.
commit aeedd34080ee11a39674b4dee6e15551eddd6b8d
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Tue Dec 27 17:49:56 2022 -0600
Add support for RHEL 8.6, 8.7, 9.0 and 9.1
Cross-compiles fine using the kernel-devel packages of
- 4.18.0-348.2.1.el8_5.x86_64 (RHEL 8.5)
- 4.18.0-372.9.1.el8.x86_64 (RHEL 8.6)
- 4.18.0-408.el8.x86_64 (CentOS 8 Stream, I think)
- 4.18.0-425.3.1.el8.x86_64 (RHEL 8.7)
- 5.14.0-70.13.1.el9_0.x86_64 (RHEL 9.0)
- 5.14.0-162.6.1.el9_1.x86_64 (RHEL 9.1)
(https://access.redhat.com/articles/3078)
Also cross-compiles fine on new stock kernels 5.18-6.1,
but I didn't have to do anything to make that work.
Fixes #395
commit 226b37e34129748abfa64e5b7a2158512536342e
Author: Sly Gryphon <sly@gamertheory.net>
Date: Mon Oct 24 08:38:34 2022 +1000
Update openwrt.md
Update documentation for latest version of OpenWRT, as per issue #391
commit 48c3d44fab53387fc2337e94e73502fff3608f3b
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Wed Oct 5 12:02:17 2022 -0500
Add pool4 validation to atomic config BIBs
Should prevent #388 from ever happening again.
commit bb49c60b9aed4c5694ef28157604e9fdee88b31c
Author: Yuji Hagiwara <yuuzi41@gmail.com>
Date: Tue Oct 4 21:57:29 2022 +0900
Fix the example configuration for NAT64
commit 78c503953e83229c4a2552b7ffcc871c0583c91c
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Tue Jun 14 10:20:07 2022 -0500
Doc: Highlight that xlat ping doesn't work from xlator namespace
Feedback from uh... someone identified as "Axel".
commit 61e6764854daee826aa3d78f3641671295938f07
Author: DasSkelett <dasskelett@gmail.com>
Date: Wed May 11 13:52:33 2022 +0200
DKMS: Fix RHEL kernel version checks on 9.99 / 5.17
Since at least kernel 5.16 these changes are also present in the Fedora 35 & 36 kernels.
However the 5.16 series did not have RHEL_RELEASE_CODE and other RHEL_ veriables defined for whatever reason
and the LINUX_VERSION_AT_LEAST/LINUX_VERSION_LOWER_THAN conditional helpers handled them as "stock" kernels.
This made DKMS compilation still work.
The 5.17 kernels however do have RHEL_RELEASE_CODE defined again, thus the build failed.
Now the version checks are updated to AT_LEAST / LOWER_THAN (a, b, c, 9, 99).
This has only been tested with Fedora 35 5.16 and 5.17 kernels. I assume that the RHEL and CENTOS distros
have the same set of backports/patches when marked as 9.99+.
x.99 seems to be the version code of the development tree before a stable RHEL x+1 drops,
so I assume that 9.0-98 did not have these patches yet.
commit 5604a61f0519230bc91660c5e16509b211ce545a
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Mon May 9 09:35:12 2022 -0500
Remove all references to jool.mx
I've lost contact with the people in charge of the website, and my
patience is exhausted. The mirror has often been problematic, and I'm
ready to let it go.
jool.mx WILL NOT BE UPDATED ANYMORE. USE nicmx.github.io/Jool instead.
PLEASE UPDATE YOUR BOOKMARKS.
commit abe9a9ec4c00a38c84c6220fb065fce7ea55fe28
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Wed Apr 27 18:48:57 2022 -0500
Add sustainability survey to the site
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: cb03e88 6822bde
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Sun Mar 20 17:52:07 2022 -0600
Merge branch 'master' into debian
commit 6822bdee4ec63467e82d723a3381b3116c1853d9
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Sun Mar 20 14:04:49 2022 -0600
Protocolary updates for release 4.1.8
Includes removing support for kernels 4.4-4.8, because it looks like
they've been EOL'd.
Also, CNAME removed. Looks like jool.mx is going to stay a mirror for a
while.
commit 4af409cff6636cdb559c85efc5ef74a10b4cc622
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Sat Mar 19 19:51:46 2022 -0600
Revert 21b97b1e4fd29676ae3dd6800dc3286c70dd0690
Bug #247 is not bound to show up again.
commit 344b058eff13fb3537a34a386b5497a62db1625b
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Sat Mar 19 18:20:37 2022 -0600
4.1.8 release review
Fixed a few small errors:
- Wasn't rejecting double fragment header
- Wasn't rejecting non-reassembled fragmented ICMP errors (even though,
just like ICMP infos, their checksums are not translatable by SIIT)
- When routing a non-reassembled fragment, code was including garbage
ports in TCP and UDP routing
- validate_size() wasn't checking ICMP protocol before extracting ICMP
header
commit c48cf90ab21111d4b32a166851370a4bb339459f
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Sat Mar 19 13:49:31 2022 -0600
RFC7915: Update the GRO comments I could find
commit e95308fe606f3bf02678437b633091bb52c5b739
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Sun Mar 13 15:59:01 2022 -0600
6to4: Remove ptk_len()
The kernel's complications weren't being taken into account while
setting the IPv4 header's DF flag.
Just review; not associated to any bugs.
commit b87b6c66a35ecc83d5c7c96ab08ecb1487d86321
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Sat Mar 5 19:29:03 2022 -0600
7915: Fix checksums on Slow Path
If the theory is correct, this is what's causing #366 and #375.
If input was type CHECKSUM_PARTIAL, Slow Path yielded CHECKSUM_NONE and
the checksum itself was a weird mush between a negative IPv4
pseudoheader and a positive IPv6 pseudoheader. Therefore, the receiver
was probably dropping the packets.
If input is CHECKSUM_PARTIAL now, Slow Path computes the outgoing
checksum from scratch. It's slow but unavoidable, because Slow Path
generates separate fragments, and tx-checksumming doesn't know how to
checksum them. I also don't want to use frag_list because there's little
to no API, and the code is complicated enough as it is.
------------------------------------------------------------------------
I was able to reuse the code to also fix a long-standing TODO: If
IPv4/UDP packet contains zero checksum, instead of computing the
IPv6/UDP checksum manually, Jool now offloads it to the NIC. Gains a
slight performance boost.
commit 454b83f859ed43cbca075282b38f8a1a7078beac
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Wed Mar 2 03:16:49 2022 -0600
GRO: Slow Path review; remove pkt_len() usage
It does very much look like pkt_len() was being used incorrectly, but I
don't see how it could be causing #375.
commit ebb5f7e1fb872abf3f59629039d5e72fb3403d02
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Wed Feb 9 11:45:40 2022 -0600
Documentation: Update sample atomic files
Sample documentation was outdated and fairly random.
I'm afraid this might have created bogus configurations out there
in the wild.
commit 89f3218e8902562a53a751743ee82155d425c6ee
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Thu Jan 27 17:55:13 2022 -0600
Create CNAME
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: 47cc968 1c1267f
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Thu Jan 27 16:36:46 2022 -0600
Merge branch 'master' into debian
commit 1c1267f2ec67323bfd1206ff2930cc941c6969d2
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Thu Jan 27 16:13:51 2022 -0600
Protocolary updates for release 4.1.7
commit 04ef98dd6c7528fc25f68bef304fb74c59cb7083
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Wed Jan 19 15:45:58 2022 -0600
xtables: print enabled status on startup
Meant to test #372.
commit 81d6ad1d57236d458d6ce41ed4f970ea42ac89fb
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Mon Jan 17 10:27:54 2022 -0600
Autotools: Add --with-xtables
Turns out including a dependency depending on installedness is not
standard practice.
Manually includes and excludes xtables from the userspace binaries:
./configure # xtables included
./configure --with-xtables # xtables included
./configure --with-xtables=yes # xtables included
./configure --with-xtables=no # xtables excluded
Took a while, but I think I finally landed optional iptables
properly.
Progress on #273.
commit a036f08ad41a8bdf99dc828da08f3a57bcd805c8
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Mon Jan 17 09:04:10 2022 -0600
Makefile: Remove JOOL_FLAGS
CFLAGS_MODULE is the new JOOL_FLAGS. It's more standard.
Instead of
make JOOL_FLAGS=-D<flag>
do
make CFLAGS_MODULE=-D<flag>
Progress on #273.
commit 23fcde6729d15ad32c602a6cbaf81ccc6daf9fc6
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Sun Jan 16 20:59:08 2022 -0600
iptables: Make optional
Userspace iptables depends on whether the configure script detects
libxtables-dev installed.
Kernelspace iptables needs to be removed manually:
make JOOL_FLAGS=-DXTABLES_DISABLED
This feature was requested in #273.
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: c6c3169 e9e5c58
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Fri Dec 10 15:38:59 2021 -0600
Merge branch 'master' into debian
commit e9e5c582c0d782ea3c1dbb402ed35481486ab734
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Thu Dec 9 17:45:58 2021 -0600
Protocolary updates for release 4.1.6
commit 8d23b996f6086967ce62d31e7a308156a6b4da27
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Thu Dec 9 11:23:43 2021 -0600
Add support for kernels 5.12-5.15, RHEL 7.9, 8.4 and 8.5
commit 555d61a04b6955ccf8d75a6143c89334c90c6e07
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Thu Nov 25 17:52:38 2021 -0600
Netfilter: Remove hardcoded array length
Makes it easier to add and remove Netfilter hooks.
commit 8833946518c3f2a84cd1387148636de9f02825bc
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Tue Nov 16 15:58:52 2021 -0600
Documentation: Update OpenWRT install instructions
Fixes #360.
commit c6897c57d579d2a98b8b2e225c46fb5872a9a6e1
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Mon Nov 15 12:48:32 2021 -0600
Documentation: Add contact pointer to Matrix
commit 458d7887b67dca442dab3867e9d9abfd7fba64b7
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Fri Nov 12 16:03:31 2021 -0600
rbtree: Replace obsolete foreach function with kernel's macro
Fixes #368.
commit a1d2d1a5d6bc35dc39489a56808d233c13863fa3
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Tue Oct 19 09:20:12 2021 -0500
Documentation: Add map-t-type to the MAP-T tutorial
Fixes #367.
commit a72e9554edb5d4bb1ced8f7c14b0e0a86cbae953
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Tue Oct 5 10:31:14 2021 -0500
README: Update Github IPv6 availability status
My logic is,
this has already happened before, and IPv6 on Github didn't last for
very long last time.
Therefore, I'll let jool.mx stick around for a little while.
Fixes #354.
commit 40022698e0d4db0727572c617ad181001fc44a2f
Author: Laura Zelenku <laura.zelenku@wandera.com>
Date: Fri Aug 27 09:16:11 2021 +0200
Update of configuration flag: --handle-rst-during-fin-rcv
Flag now handles RST packets from both sides (IPv4 and IPv6).
commit 725b0e125c911ad91bfe232967dfe455f8091c31
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Sat Aug 14 18:37:44 2021 -0500
ICMP: Do not rely on flowi to translate internal type and error
The flowi type and code refer to the outer packet, but were being used
to translate the inner packet as well.
This was breaking translation of ICMP errors which were responses to
echo requests and echo replies.
commit cd1c03f9087dca591fb3744264087ec25dc1ea3e
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Tue Jul 27 11:02:52 2021 -0500
Address xlat: Change ::1 verdict from DROP to ACCEPT
SIIT Jool was dropping all traffic sourced ::1.
Which was a huge loopback problem in Netfilter mode.
New behavior is ACCEPT; ie. "return to the kernel."
commit 2323a135237efb9b8fe6fc479ec35b8b31101c4a
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Fri Jun 4 12:33:05 2021 -0500
EAMT: Fix #363 properly
Eliminates RCU synchronizations, but only on atomic configuration
(`file handle`) mode.
When adding EAMT entries through atomic configuration, the translator
is inactive. Therefore, there are no readers, and therefore, the RCU
syncs are redundant. So they can be thrown away at no cost.
`eamt add`, `eamt remove` and `eamt flush` were broken in the previous
commit, and now they're back in working order. (But those haven't been
optimized.)
- This optimization cannot be applied to `eamt add`, because `eamt add`
operates on running translators.
- `eamt remove` and `eamt flush` never needed optimizations. Turns out
I broke them for no reason.
I don't really know why synchronize_rcu() is so slow in some systems,
but presumably, it is implemented differently depending on architecture.
commit 3a0da31ccc3361df21f3d5a6835d6280a7d87489
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Thu Jun 3 19:52:40 2021 -0500
Radix Trie: Purge RCU waits
Temporal commit; recklessly nullifies locking in the EAMT write
operations. Might yield a performance boost for #363.
Do not execute eamt add, eamt remove nor eamt display while
running this commit.
commit 6f3ad879fe567713e092e6b024349b519cb247bf
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Fri May 28 15:32:39 2021 -0500
joold: Add NLA_F_NESTED to joold add requests
Apparently, Linux 5.10+ now validates the flag. Maybe 5.9 too.
This bug only affects joold because of its unique multicast API.
Fixes #362.
commit 5dc6ae4fbc1620d3c6cefb8b98f054bb81c47a2a
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Tue Apr 6 19:25:39 2021 -0500
Documentation: OpenWRT review
Work on #360.
commit 3c4c71cca4ae3c1430191e9a3cbf1b73eb7ad703
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Thu Mar 11 10:44:28 2021 -0600
Documentation: Update Privacy Notice
The one from two commits before was transitional.
This is the permanent privacy notice for now.
Don't quote me on this, but it seems the only thing that changed
is the address of NIC Mexico.
commit 673f89a741d9d1cb55bf8bf089da7714cc48bf7f
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Wed Mar 10 19:58:12 2021 -0600
Documentation: Add "does't follow Jool's protocol" FAQ entry
Fixes #358.
commit a08c1784bab74be97d61f326c5daf9a90c0f0048
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Wed Mar 3 19:38:10 2021 -0600
Documentation: Update Privacy Notice
commit 3a873911ec3f3e7d23f3fb570397949aeb9cfb67
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Mon Feb 22 11:08:38 2021 -0600
Documentation: Downgrade recommendation priority of 4.0.9
Jool 4.0.9 is almost obsolete now.
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: 63a61ad 34ffbf5
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Fri Feb 19 17:05:52 2021 -0600
Merge branch 'master' into debian
commit 34ffbf56a0c8bd53c4cb554e102a9084b301170e
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Thu Feb 18 16:30:32 2021 -0600
Protocolary updates for release 4.1.5
commit ff8f8bdbe3198961a70d15de4984955a9984de19
Merge: 978190a dc7b39b
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Thu Feb 18 13:25:33 2021 -0600
Merge branch 'issue352'
commit 978190a96538d6fbb749587e66639db547b91611
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Mon Jan 11 13:12:47 2021 -0600
Documentation: Update year
commit ebe4abbb9f2fe0e8ecbbf5bbae20e44f156646d3
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Mon Jan 11 13:01:36 2021 -0600
Documentation: Fix DNS64 tutorial 2
- The "Joined" tabs are now the default ones.
- Add tabs to the BIND configuration block, so I don't have to
overexplain the "translator" ACL.
F1xes #354.
commit e7141a10cd3b393bde09f3d093046c9f29faf4ff
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Fri Jan 8 16:47:22 2021 -0600
Documentation: Fix DNS64 tutorial
Because Jool cannot translate its own namespace's traffic,
sending it DNS64'd records tends to break its own attempts to
communicate with the IPv4 world.
The new DNS64 documentation avoids this pitfall.
F1xes #354.
commit dc7b39b77a2a5c0397668c3d05d771f44b45ff59
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Sun Jan 3 16:38:50 2021 -0600
issue352: Add debugging information
The most likely suspect is one of the page fragments pointed by
the outgoing packet's page array.
This code catches the invalid packet and prints a bunch of
relevant data, and leaks the packet instead of crashing the
kernel.
commit 8dd0b9fba3f3ee673f6860ae012a5769b3b9df2f
Author: Molly Miller <sysvinit@users.noreply.github.com>
Date: Sat Jan 2 02:09:58 2021 +0000
Define the netlink header magic string in the preprocessor.
The magic string was introduced to the netlink header struct in 58bf14e as part
of the fix for NICMx/Jool#340, initially as a hard-coded byte sequence. This
commit moves the magic string and its length into a preprocessor definition,
and reads and writes this field using memcmp() and memmove() -- if the string
ever needs to be changed in the future, then the change will be automatically
picked up by all the code which reads or writes this header field.
commit 58bf14e0f9e13ae1bc2ab3e3584f0b9fecf1a634
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Fri Jan 1 17:11:18 2021 -0600
joold: Properly initalize the Jool header
This is actually fixing two bugs:
1. The kernel module was not initializing the Jool header on joold
packets. Ever. At all.
2. joold wasn't validating the Jool header.
The two bugs were working in beautiful concert, cancelling each other
in the unit tests. FML
Further progress on #340.
commit ba85afbfa40fa4c42b55d43541c804f267b52590
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Thu Dec 31 15:11:02 2020 -0600
joold: Patch Netlink callback mess
Tentatively fixes #340.
commit d3208d8c14dad4abc896ca6adc3da1f1831ec1ef
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Thu Dec 24 20:50:14 2020 -0600
Documentation: Update with 4.2.0-rc1
commit 73dc38c21f9eed680c4a895d995d1e498905c6b8
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Tue Dec 22 14:30:49 2020 -0600
Hairpinning: Reinforce state object initialization on SIIT
SIIT Jool was not zeroing the hairpinning translation state object on
initialization.
I don't actually know if this yielded misbehavior, but it was
inconsistent with the way the normal packet pipeline, as well as the
hairpinning NAT64 pipeline initialized their state objects.
So I don't know if this is a bugfix, but the code is certainly more
bug-resistant now.
commit 5ad9b702b5b989414493526cfc481a3396229278
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Tue Dec 22 13:59:49 2020 -0600
Netlink: Remove CAP_NET_ADMIN validation on stats display handler
Seeing that anyone can seemingly query `nstat`, Jool is likely
being overly paranoid by requiring privileges on showing stats.
Fixes #350.
commit aec9f25f5f7bb1a6c7a39652a176118b1152fa65
Author: Cooper Lees <me@cooperlees.com>
Date: Mon Dec 21 16:49:55 2020 -0800
Add a timeseries stats section
- Explain about the new jool-exporter being available
- Keep generic + link to projects README
- State that it's not maintained by Jool core developers
commit fb731d99bfbec6b0cc09e8367e366248d8198267
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Fri Dec 18 18:59:24 2020 -0600
Documentation: Update MAP-T usage, add prototype Cheat Sheet
MAP-T usage: Reflects latest changes in the MAP-T branch.
Cheat Sheet: For #345
(I just noticed it's missing 464XLAT, ugh)
commit 705c5adf5ea885191fe0c1c58ce2e25cc28a9e99
Author: Alberto Leiva Popper <ydahhrk@gmail.com>
Date: Mon Dec 14 17:16:30 2020 -0600
MAP-T: Documentation review
![[bug history graph]](https://qa.debian.org/data/bts/graphs/j/jool.png){kind=link}