Commit Graph

5094 Commits (67078018dbef52af6f6f7710a223b2731466c501)
 

Author SHA1 Message Date
Jeff Becker 6fd714d193
contrib/testnet: fix up testnet config generator to make super centralized topology
llarp/context.cpp, llarp/nodedb.{h,c}pp: load netdb AFTER whitelist
llarp/router/router.cpp: explore always
llarp/router/{i,}rc_lookup_handler.{h,c}pp explore with whitelist, update routers with lookup before stale
5 years ago
Jeff Becker 9322df013b
check with rc handler before insert 5 years ago
Jeff Becker 2af64671bf
make it compile 5 years ago
Jeff Becker 0526fef37b
store RCs if valid 5 years ago
Jeff Becker b56a3528db
periodic nodedb flush 5 years ago
Jeff Becker 495ad6d566
testnet patch 5 years ago
Jeff Becker 09f5f6dfa7
disable iterative lookup 5 years ago
Jeff Becker c7b94f32be
reduce disk io 5 years ago
Jeff 0e2c5736a8
Merge pull request #1034 from majestrate/dht-timeout-fixes-2020-01-10
dht timeout tweak
5 years ago
Jeff 0ad2809ac7
Merge pull request #1033 from majestrate/prevent-segfault-2020-01-10
prevent segfault
5 years ago
Jeff Becker 78cdf8f96b
variable lookup timeout on participants 5 years ago
Jeff Becker 8bb492da6e
dht timeout tweak 5 years ago
Jeff Becker 77f0944d6d
use shared_from_this 5 years ago
Jeff a3bd44ae4f
Merge pull request #1032 from majestrate/dht-explore-faster-2020-01-09
explore dht faster
5 years ago
Jeff 7b36f45f26
Merge pull request #1031 from jagerman/avx2-detection-again
Fix avx2 cpu detection on amd64 (again)
5 years ago
Jeff Becker 78e325c4c5
always lookup introset from nearest router 5 years ago
Jeff Becker 2d82e931da
try doing lookup from close router, randomize when above lookup threshold 5 years ago
Jeff Becker b2784dd5ec
explore faster 5 years ago
Jason Rhinelander aef1ac9eed 0.6.2 release 5 years ago
Jason Rhinelander ead2461335 Enable cpu detection on amd64 5 years ago
Jason Rhinelander 955c475dbe Enable cpu detection on amd64
The optional runtime detection shouldn't rely on being compiled in avx2
mode.
5 years ago
Jason Rhinelander 759972004c 0.6.2 release 5 years ago
Jason Rhinelander 40842fdf6e Merge branch 'master' into debian/sid 5 years ago
Jason Rhinelander f66c6826da Rediff patches 5 years ago
Jason Rhinelander 24eba5d842 0.6.2 release 5 years ago
Jason Rhinelander 2936dad666 v0.6.2 patch release
-----BEGIN PGP SIGNATURE-----
 
 iHUEABYIAB0WIQRn72umjnsLDW6099TzV7O0L2+bBQUCXhTsOgAKCRDzV7O0L2+b
 BR2UAP9mC3a0fsHUciqnnFWNASkDLfsx3Eplh6zgxS+/Nd+DDQEAx0Z/XtAOKWSk
 TEj4CMmnKh2oZmsTBQGsj/jXBXvqjAg=
 =0cU0
 -----END PGP SIGNATURE-----

Merge tag 'v0.6.2' into debian/sid

v0.6.2 patch release

# gpg: Signature made Tue 07 Jan 2020 04:38:18 PM AST
# gpg:                using EDDSA key 67EF6BA68E7B0B0D6EB4F7D4F357B3B42F6F9B05
# gpg: Can't check signature: No public key
5 years ago
Jason Rhinelander 220fabb6b6 Lower compiler settings to -march=x86-64
This should allow it to work pretty much anywhere, and in 0.6.2 we still
get the run-time AVX2 support for crypto.
5 years ago
Jeff 71bed2e52d
Merge pull request #1030 from loki-project/dev
Dev into master for 0.6.2 release
5 years ago
Jeff ac46d99107
Merge pull request #1029 from majestrate/expire-snode-sessions-based-off-activity-2020-01-08
Expire snode sessions based off activity
5 years ago
Jeff Becker eb174923ad
m_LastUse shouldn't be 0 ever now 5 years ago
Jeff Becker 1a80153352
mark used on costructed so it doesn't go away 5 years ago
Jeff Becker 9e50920dad
dont update timestamp of last use using when we get exit only when we use it 5 years ago
Jeff cefdaeadee
Merge pull request #1028 from majestrate/remove-assert-with-sideffect-2020-01-08
remove side effect from address detection assert
5 years ago
Jeff Becker a96fd645c5
remove side effect from address detection assert 5 years ago
Jeff 71d5fd27d0
Merge pull request #1005 from majestrate/staitc-link-2019-12-26
make static linking actually static
5 years ago
Jeff 2e15310afb
Merge pull request #1025 from jagerman/download-sodium-avx2-fix
Fix ntru avx2 compilation when DOWNLOAD_SODIUM
5 years ago
Jason Rhinelander 6069b726e1 Fix ntru avx2 compilation when DOWNLOAD_SODIUM
We don't explicitly have to link against it, but it doesn't hurt *and*
this also pulls in the proper sodium include path, which we do need.
5 years ago
Jeff Becker da5eb913e9
use proper cmake call 5 years ago
Jeff Becker fde0ca374f
i hate curl 5 years ago
Jeff Becker 336a748695
make it build in normal mode too 5 years ago
Jeff Becker 77213d7fa4
add hashpinned curl tarball for static build 5 years ago
Jeff Becker 80cbddde7b
fixups for review 5 years ago
Jeff Becker 7ffc786bce
fix up docs 5 years ago
Jeff Becker 2ce592e4af
static link with docs 5 years ago
Jeff de819d3876
Merge pull request #1024 from jagerman/ntru-avx2-compilation
Enable AVX2 compilation whenever possible
5 years ago
Jason Rhinelander 7cec0db371 Simplify CRYPTO_FLAGS logic, remove RYZEN_HACK
We currently spend a bunch of time setting CRYPTO_FLAGS and then just
completely wipe it out if NATIVE_BUILD is specified.

The AMD_RYZEN_HACK is really not needed; it was only some early CPU
microcode and compiler combinations that had trouble detecting Ryzen's
FMA3 support (and Ryzen *doesn't* properly support FMA4--it can be
forced on and runs, but apparently it can give wrong results).
5 years ago
Jason Rhinelander 6f3757e933 Modernize non-AVX2 tuning to haswell
Tuning to an ancient architecture doesn't make a lot of sense; we want
to support the ancient architecture, but don't want to optimize for it.

Also change the AVX2 tuning to use haswell so that optimizations don't
depend on the CPU in the system the build runs on.
5 years ago
Jason Rhinelander 210cb837c6 Enable -avx2 -fma for ntru even when compiling without AVX2 support
This code is runtime guarded with CPU feature tests, so even if we are
making a non-AVX2 build we still want to compile the avx code path with
-mavx2 -mfma so that it can be opportunistically used when run on
AVX2-supporting CPUs (particularly for the debs and release binaries).
5 years ago
Jeff e223f5d023
Merge pull request #1016 from loki-project/dev
propagate dev into master
5 years ago
Jeff cceb531e58
version bump 5 years ago