[Git][NTPsec/ntpsec][ctype] 43 commits: change type of ntpcal_get_build_date to be bool

James Browning gitlab at mg.gitlab.com
Thu Mar 19 09:49:28 UTC 2020 


James Browning pushed to branch ctype at NTPsec / ntpsec


Commits:
b0f9d525 by Hal Murray at 2020-02-08T05:48:32-08:00
change type of ntpcal_get_build_date to be bool

- - - - -
f8119747 by Hal Murray at 2020-02-08T05:48:32-08:00
Fix tests/option-tester.sh to use --disable-doc

- - - - -
acaa9832 by Hal Murray at 2020-02-08T05:48:32-08:00
Remove unused arg to leap_smear_add_offs

- - - - -
46cbce10 by Richard Laager at 2020-02-11T04:14:35+00:00
Redesign doc building options

Prior to recent changes (i.e. in the last release):
  - HTML docs were not built by default, even if you had asciidoc
    installed.  You had to explicit opt-in with --enable-doc.
  - Man pages were built by default if you had asciidoc installed.
    If asciidoc was not installed, the build continued without building
    man pages.  --disable-manpage could optionally be used (e.g. to
    speed the build).
  - CentOS 6 had too old of an asciidoc, but the build continued without
    building man pages.

After recent changes:
  - HTML docs were built by default.  If you did not have asciidoc(tor)
    installed, waf configure stopped with an error telling you to
    explicitly pass --disable-doc.
  - Man pages were built by default.  If you did not have asciidoc(tor)
    installed, waf configure stopped with an error telling you to
    explicitly pass --disable-manpage.
  - CentOS 6 had too old of an asciidoc. The build would stop with an
    error telling you to explicitly pass --disable-manpage.

  Relative to the previous state, this improved things by:
    - being consistent
    - building the useful docs by default
    - failing loudly so you can't accidentally miss building man pages
  but:
    - broke any build scripts (e.g. in distro packages) passing
      --enable-doc
    - requires CentOS 6 users to pass --disable-doc and
      --disable-manpage
    - requires users to either install asciidoc(tor) or pass
      --disable-doc and --disable-manpage

Now:
  - HTML docs are built by default if you have asciidoc(tor) installed.
    If asciidoc(tor) is not installed, the build continues without
    building HTML docs.  --disable-doc can optionally be used (e.g. to
    speed the build).  --enable-doc can optionally be used (e.g. to
    require that docs be built).
  - Man pages are built by default if you have asciidoc(tor) installed.
    If asciidoc(tor) is not installed, the build continues without
    building man pages.  --disable-manpage can optionally be used (e.g.
    to speed the build).  --enable-manpage can optionally be used (e.g.
    to require that docs be built).
  - CentOS 6 has too old of an asciidoc, but the build continues without
    building man pages.

  Relative to the original state, this keeps the following improvements:
    - being consistent
    - building the useful docs by default (but only when asciidoc(tor)
      is installed)
  allows via --enable-* (but no longer defaults to):
    - failing loudly so you can't accidentally miss building docs
      and/or man pages

  Relative to the intermediate state, it has the following advantages:
    - build scripts with --enable-doc still work
    - CentOS 6 users do not have to pass --disable-doc and
      --disable-manpage
    - users are not required to install asciidoc(tor) or pass
      --disable-doc and --disable-manpage

Additionally:
  - The configure script now outputs a "Build Docs" line to match the
    "Build Manpages" line.
  - The code for the "Build Manpages" output and the actual decision
    check the same thing, which is now a single boolean.

The CI runners are all explicitly enabling or disabling docs as desired,
so those that are supposed to be building docs cannot silently fail
(e.g. if we change the minimum asciidoc(tor) version requirements).

- - - - -
34d309c0 by Richard Laager at 2020-02-11T04:14:35+00:00
Use the full --disable-manpage for CI runners

- - - - -
27204468 by Hal Murray at 2020-02-10T20:53:20-08:00
Formatting cleanup (tabs/spaces)

- - - - -
94ded261 by Hal Murray at 2020-02-13T12:02:46-08:00
Crash if error return from RAND_bytes and RAND_priv_bytes

- - - - -
df4d37cd by Hal Murray at 2020-02-14T12:00:49-08:00
Activate attic/random

- - - - -
5b1a710a by Mark Atwood at 2020-02-14T21:49:45-08:00
New copyright declaration policy, no year needed

Most of the existing copyright statements present in the project have
been scrubbed of the year.  There is no need to specify the year in a
copyright statement.  Several large legally sophisticated companies,
include Amazon, Google, Microsoft, and Facebook, are now publishing
open source code with copyright statements without a year.  We
encourage the larger open source community to emulate this.

The string "(c)" is legally meaningless. Most occurances have been
removed.

- - - - -
101434f7 by Mark Atwood at 2020-02-14T23:07:31-08:00
updates to devel/hacking.adoc

- - - - -
87f5f497 by Hal Murray at 2020-02-18T22:22:38-08:00
typos in devel/tour.adoc

- - - - -
d34c4897 by Hal Murray at 2020-02-18T22:22:38-08:00
Tweak wording around getitimer

- - - - -
6410cca8 by Hal Murray at 2020-02-18T22:24:35-08:00
Hack fix for issue #642

- - - - -
59e3ed56 by Hal Murray at 2020-02-18T22:55:09-08:00
random number cleanup

remove ntp_random() - call random(3) directly.
remove ntp_random64() - no longer used
add ntp_RAND_bytes() to check return code
  and avoid check/msyslog/crash at each usage of RAND_bytes()

- - - - -
47b571e1 by James Browning at 2020-02-20T23:29:21+00:00
Config: Fix error when miscasting pointer ...

There is a warning:
'cast from pointer to integer of different size [-Wpointer-to-int-cast]'
which breaks ./waf config when it reaches the function check for
CMAC_CTX_new if CFLAGS= "-Werror" is prepended. This seems to solve it.

Single casting as (long) is unacceptable.

Returning 'p ? 1 : 0' or '!!p' breaks the macOS runners;
they detecct res_init() which is not a good thing.

Thus the double cast (int)(long)

- - - - -
d9b141a1 by Hal Murray at 2020-02-20T17:53:19-08:00
Squash warning on NetBSD 9.0

../../ntpd/ntp_control.c:1476:34: warning: '%s' directive output may be truncated writing up to 255 bytes into a region of size between 0 and 255 [-Wformat-truncation=]
                snprintf(str, sizeof(str), "%s/%s", utsnamebuf.sysname,
                         utsnamebuf.release);

It wants us to check the return code or use %.<n>s to limit
the size.

- - - - -
acd4adef by Richard Laager at 2020-02-22T17:02:50-06:00
Fix AsciiDoc 3 runner

This was recently failing, but was somewhat intermittent.  It did not
have the docbook-xsl stylesheets installed, which meant it had to
download them (repeatedly, I think) from the Internet.

- - - - -
62f23d25 by Richard Laager at 2020-02-22T17:03:05-06:00
Cleanup some .gitlab-ci shell script

This adds some quotes and spaces.

- - - - -
872d52cd by James Browning at 2020-02-22T23:15:09+00:00
ntpq peers name alignment fix


- - - - -
60c55885 by James Browning at 2020-02-23T00:04:15+00:00
Add scheduled options-tester using CI runners

- - - - -
2e0d90c3 by Richard Laager at 2020-02-22T22:57:57-06:00
Fix pipefail test

The logic on this was inadvertently flipped during one of the rewrites.

- - - - -
39a880bb by Hal Murray at 2020-02-23T14:36:41-08:00
More work on seccomp for Arch #633

I think the root of the issue is an ommision in
  /usr/include/seccomp-syscalls.h
Fedora seems fixed.  The same issue has poped up on Alpine Linux.
Editing the above header file fixed it.

- - - - -
089c4a73 by Hal Murray at 2020-02-23T14:43:29-08:00
Avoid bogus attempts to find local interface

If DNS or NTS-KE hasn't worked, the local address
hasn't been setup yet.

This was causing calls to socket() that were getting error returns.
That looked ugly on strace output.

- - - - -
cf9ad859 by Hal Murray at 2020-02-23T15:08:27-08:00
Tweaks to SIGHUP documentation.

- - - - -
63be8972 by Hal Murray at 2020-02-23T15:20:26-08:00
Cleanup findlocalinterface, from James Browning

- - - - -
924b9844 by Hal Murray at 2020-02-24T01:56:01-08:00
Eliminate more bogus find local address - pool case

- - - - -
37ab8228 by Hal Murray at 2020-02-24T01:57:52-08:00
Seccomp builds and runs on Alpine Linux 3.11.3
Added hack to dance around bug in /usr/include/seccomp-syscalls.h
Added membarrier and writev to system call list

- - - - -
0ed0d595 by Hal Murray at 2020-02-29T22:03:47-08:00
Remove unused arg from set_timer_or_die()

- - - - -
f0bd02b2 by Adam Kumiszcza at 2020-03-03T03:46:00+00:00
Update ntpheat description.

- - - - -
dd6a88fd by James Browning at 2020-03-07T03:03:27-08:00
Less dumb mru  kv list assembly


- - - - -
6d648e44 by James Browning at 2020-03-07T03:05:35-08:00
rework test to work w/ new reassembly


- - - - -
7075f4ea by James Browning at 2020-03-07T08:00:42-08:00
should sort mru kv(s) metadata then numbered records


- - - - -
51f41df5 by James Browning at 2020-03-07T08:07:53-08:00
Ignore a macOS specific file


- - - - -
4ac74571 by Hal Murray at 2020-03-09T21:02:45-07:00
Rate limiting cleanup

The old rate limiting scheme was carefully tuned to work with
a single copy of the same code.  That doesn't work if you have
several clients behind a NAT box.

That chunk of code now uses a simple exponential decay (aka
leaky bucket) to keep track of recent traffic.  The time constant
is 20 seconds and the rate limit is 1 packet per second.  There is
no config option to change them.  (yet?)  The "score" is scaled
such that the units are packets/second.

That pair of parameters means that a client can start with a burst
of 20 packets but over the long term, but can only sustain 1 packet
per second.

ntpq/mru now displays the score and the number of packets
from that IP Address that were dropped.

- - - - -
4cbab792 by Hal Murray at 2020-03-10T19:19:44-07:00
seccomp: Add comment about systemd to catchTrap

- - - - -
6fa76a4d by Hal Murray at 2020-03-12T16:31:20-07:00
Add KoD limiting

- - - - -
0093b02b by Hal Murray at 2020-03-12T18:20:41-07:00
Add sorting by score and drop to ntpq/mrulist

- - - - -
0e0e4db4 by Hal Murray at 2020-03-12T23:36:19-07:00
Add filtering by mindrop and minscore to ntpq/mrulist

- - - - -
f67f8bf1 by Hal Murray at 2020-03-13T02:04:25-07:00
Add minlstint and doc for maxlstint for ntpq/mrulist

- - - - -
9c9fab07 by Richard Laager at 2020-03-16T23:12:33-05:00
Fix Python 2 options tests on Debian/Ubuntu

- - - - -
854e9947 by James Browning at 2020-03-19T01:58:39-07:00
Atomic blob phase 1


- - - - -
d0cd803d by James Browning at 2020-03-19T02:04:23-07:00
Atomic blob phase 2r6

- - - - -
1694b963 by James Browning at 2020-03-19T02:39:17-07:00
Atomic blob phase 3r1


- - - - -


30 changed files:

- .gitignore
- .gitlab-ci.yml
- + .gitlab-opttest-ci.yml
- INSTALL.adoc
- NEWS.adoc
- attic/README
- attic/ntpver
- attic/sht.c
- attic/wscript
- contrib/ntpheat
- devel/hacking.adoc
- devel/tour.adoc
- devel/trace/ntpsec-ntp-proto.ini
- devel/trace/ntpsec-ntp-worker.ini
- devel/trace/ntpsec-trace-log.ini
- devel/trace/ntpsec-trace.ini
- devel/trace/ntpsec-work-thread.ini
- devel/y2k.adoc
- docs/copyright.adoc
- docs/includes/mrufmt.adoc
- docs/includes/ntpd-body.adoc
- docs/includes/ntpq-body.adoc
- docs/mode6.adoc
- include/ascii.h
- include/binio.h
- include/ieee754io.h
- include/isc_interfaceiter.h
- include/isc_netaddr.h
- include/isc_result.h
- include/mbg_gps166.h


The diff was not included because it is too large.


View it on GitLab: https://gitlab.com/NTPsec/ntpsec/-/compare/f0108abd20197f86186a1dbd15c86ca1f2a52b20...1694b963fcc6799d704d1c8b5af215c9720eaecf

-- 
View it on GitLab: https://gitlab.com/NTPsec/ntpsec/-/compare/f0108abd20197f86186a1dbd15c86ca1f2a52b20...1694b963fcc6799d704d1c8b5af215c9720eaecf
You're receiving this email because of your account on gitlab.com.


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.ntpsec.org/pipermail/vc/attachments/20200319/b715b0c8/attachment-0001.htm>

More information about the vc mailing list