fix KX in Cadet for revocation-style sequencing

resume curl download if we paused it before

follow twister change e01669953d7a8e82a505fca2308e57e971d43c67: check return value on upload_cb registration with curl

code cleanup, fix strlen

update gitignore

fix handling of chunked encoding by proxy, and handling of connection termination

typo

use real GNS record types in tests to avoid failures in new extra serialization/deserialization checking

testcase cleanup

fix a few disting issues

script was removed

fix typo in Makefile.am

use DEBUG_KX for new core log messages

merge benchmark changes

add function for getopt uint16_t arguments

more logging

more crypto KX logging

more core KX logging

fix logic

proper datacache expiration by proximity first

more logging

add more logging for core KX to improve diagnostics

Niibe writes:

Sorry, I was not reading the code of GNUnet well.  I overlooked how the
eddsa_d_to_a function was written and its intention.  I read it again.

Indeed, the eddsa_d_to_a function tries to handle the case where
gcry_mpi_print returns rawmpilen < 32, putting "left pad" by DIGEST.

The problem is:

DIGEST is not cleared (although comment says so).

I think that the stack had zero-byte for some reason on your 32-bit
machine.

Here is the correction.  Clear DIGEST, as comment says.

diff --git a/src/util/crypto_ecc.c b/src/util/crypto_ecc.c
index 8d9091b23..280603234 100644
--- a/src/util/crypto_ecc.c
+++ b/src/util/crypto_ecc.c
@@ -1273,24 +1273,15 @@ eddsa_d_to_a (gcry_mpi_t d)

   b = 256 / 8; /* number of bytes in `d` */

+  memset (hvec, 0, sizeof hvec);
   /* Note that we clear DIGEST so we can use it as input to left pad
      the key with zeroes for hashing.  */
-  memset (hvec, 0, sizeof hvec);
+  memset (digest, 0, sizeof digest);
   rawmpilen = sizeof (rawmpi);
   GNUNET_assert (0 ==
                  gcry_mpi_print (GCRYMPI_FMT_USG,
   rawmpi, rawmpilen, &rawmpilen,
                                  d));
-  if (rawmpilen < 32)
-  {
-    memmove (rawmpi + 32 - rawmpilen,
-             rawmpi,
-             rawmpilen);
-    memset (rawmpi,
-            0,
-            32 - rawmpilen);
-    rawmpilen = 32;
-  }
   hvec[0].data = digest;
   hvec[0].off = 0;
   hvec[0].len = b > rawmpilen ? (b - rawmpilen) : 0;
--

add -2 flag to gnunet-gns-benchmark

add proximity considerations to datacache

disable xu transport, header missing

reduce loop counters to more practical levels

style fix

Merge branch 'master' of ssh://gnunet.org/gnunet

adding xu transport

add patch from Niibe Yutaka for
#5328

skip validation logic

skip validation logic

track hello passing

gitignore: Add pytest specific

Signed-off-by: Nils Gillmann <ng0@n0.is>

add TODO notes in doc/documentation

Signed-off-by: Nils Gillmann <ng0@n0.is>

Merge branch 'h2020' of git+ssh://gnunet.org/gnunet into h2020

disable LAN/loopback for AC

build xt

exclude SOA-only zones

xt transport for experiments:

Add gnunetgnsrecord to Libs

Merge branch 'master' of gnunet.org:gnunet

c-tutorial: grammar fixes and general improvements

Signed-off-by: Nils Gillmann <ng0@n0.is>

stop DNS lookup immediately instead of async to avoid use-after free if we get BOTH an IPv4 and IPv6 response

fix bug where we might fail the resolution of the same DHT heap node twice due to async cleanup

more flakes.

Signed-off-by: Nils Gillmann <ng0@n0.is>

test_integration_boostrap_and_connect: flake8

Signed-off-by: Nils Gillmann <ng0@n0.is>

integration-tests: clique: flake8

Signed-off-by: Nils Gillmann <ng0@n0.is>

integration-tests: gnunet-testing: flake8

Signed-off-by: Nils Gillmann <ng0@n0.is>

integration-tests reconnect_nat: flake8

Signed-off-by: Nils Gillmann <ng0@n0.is>

Merge branch 'master' of gnunet.org:gnunet

integration-tests disconnect_nat flake8

Signed-off-by: Nils Gillmann <ng0@n0.is>

revocation: test_local_revocation: Adjust to flake8

Signed-off-by: Nils Gillmann <ng0@n0.is>

starting with another CADET testcase (unfinished)

no need to dup cfg

fix indentation

log on success, too

+

Signed-off-by: Nils Gillmann <ng0@n0.is>

fix test locations

Signed-off-by: Nils Gillmann <ng0@n0.is>

test_gnunet_statistics: flake8

Signed-off-by: Nils Gillmann <ng0@n0.is>

testbed/buildvars.py: flake8

Signed-off-by: Nils Gillmann <ng0@n0.is>

+

Signed-off-by: Nils Gillmann <ng0@n0.is>

janitor: flake8ism

Signed-off-by: Nils Gillmann <ng0@n0.is>

pyexpect: flake8ism

Signed-off-by: Nils Gillmann <ng0@n0.is>

+

Signed-off-by: Nils Gillmann <ng0@n0.is>

text_gnunet_fs_psd.py.in: Fix loading pxpect + flake8'ism

Signed-off-by: Nils Gillmann <ng0@n0.is>

find_typedefs.py: Comment about variable 'li'

Signed-off-by: Nils Gillmann <ng0@n0.is>

contrib/scripts/terminate.py.in: bump the year

Signed-off-by: Nils Gillmann <ng0@n0.is>

contrib/scripts/terminate.py.in: indent fixes + comments

Signed-off-by: Nils Gillmann <ng0@n0.is>

++

Signed-off-by: Nils Gillmann <ng0@n0.is>

More flakes

Signed-off-by: Nils Gillmann <ng0@n0.is>

gitignore flake8.log

Signed-off-by: Nils Gillmann <ng0@n0.is>

Merge branch 'master' of gnunet.org:gnunet

peerinfo-tool: flake8 adjustments

Signed-off-by: Nils Gillmann <ng0@n0.is>

sanitize log levels

fix crash on unexpected client disconnect on incoming message, remove bogus assertion

disable debug logic

update ignore

update ignore

update ignore

fix serialization/deserialization error in namestore deferred monitor logic

check return values from GNSRECORD_record_serialize/size always

off by 0 in log msg

fix crash on namestore disconnect

guard more carefully against integer overflows

add test for gns2dns records pointing to DNS name

fix warning

add logic to count concurrently active GNS requests

allow multiple GNS2DNS records to co-exist

remove dead shorten logic

fix partitioning

do not crash on multiple CNAMEs

fix another off by one

fix off-by-one error in cadet connection construction, also enforce better timeouts for retransmissions of handshake

flake8: add .py.in to tox.ini and realize the process is far from finished.

Signed-off-by: Nils Gillmann <ng0@n0.is>

Fix failing peerinfo test

Signed-off-by: Nils Gillmann <ng0@n0.is>

disable crypto bug workaround

undo hostkey patch

src/consensus/consensus-simulation: Use past.builtins for xrange

Signed-off-by: Nils Gillmann <ng0@n0.is>