[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[platform-testers] please test inetutils 1.9.5 rc1
[platform-testers] please test inetutils 1.9.5 rc1
Wed, 01 Jan 2020 14:45:39 +0100
Gnus/5.13 (Gnus v5.13) Emacs/26.1 (gnu/linux)
The first release candidate of GNU InetUtils 1.9.5 is out:
Please test and send reports to: address@hidden
My focus is to avoid regressions since 1.9.4 so please compare any
issues with the previous release:
Reports of new problems are definitely encouraged, but it would help me
to understand if a problem is newly introduced or exists in the last
release as well.
Mats, you are responsible for most of the changes (thank you!) since the
last release, so a confirmation from you that everything is in order
would be appreciated.
** Implement subprivileged echo method.
Add fallback code usable by GNU/Linux to create a datagram socket for
sending ICMP ECHOREQUEST. The change is related to issue #37252.
** Support changing of hardware address.
Implement the setting of link level addresses for systems running
** Statistics for BSD systems.
** A work-around specific to GNU/Hurd.
This system uses a mixture of BSD definitions and GNU/Linux
interfaces. In particular, some care is needed to check interface
** Improve the support for GNU/Hurd.
Add system specific abilities, including parsing of command line and
** More info on tunnel interfaces.
** Implement a hardware list for BSD systems.
Patterned on the hardware recognition already in use by GNU/Linux, use
a similar mechanism for all of BSD.
** Sporadically appearing regression.
Device flags are sometimes calculated before their query using
SIOCGIFFLAGS. First observed on GNU/Hurd.
** Apply flags correctly.
When finalizing actions, some flags could be forgotten. In
particular, the flag 'up' was ignored whenever an address was assigned
at the same time. This is related to the fact that an invocation only
effecting interface flags, still resulted in a full printout if
interface status. This latter behaviour differs from all other
implementations and is rather annoying. The issue was detected and
reported by Hans-Peter Budek in
** Verbose report on changed flags.
** Symbolic name as peer address.
** Allow invocation, as well as command `open', to accept an explicit
** remote user name as extended host argument: `user@host'.
** Portability to Android.
Without a check for HAVE_DECL_GETPASS and without making a local
function declaration, the implicit declaration causes crashes on
64-bit systems lacking declared getpass(), such as Android. See
** Mention mandatory argument for <max-timeout>.
** Integer overflow.
During transmission of large files, size exceeding 2GB, a long integer
variable overflows. Issue reported and identified by Wenlin Kang in
** Avoid a trailing space.
Suppress the space character after the last presented IP number.
Problem reported by Terje Strand in
** Make -F handle comment-only input files without trailing newline.
Problem reported by Omer Anson <address@hidden> in
** Redefine faulty macro LOG_MAKEPRI whenever needed.
Systems with Glibc header file <syslog.h> prior to 2.17, are not
correctly defining the helper macro LOG_MAKEPRI. This effects our
syslog service whenever it receives kernel messages from a remote
host. The issue was mentioned in Debian's BTS as report #729666.
** AIX portability
** Add LOG_NDELAY to openlog(). The file descriptor for logging must
** be active before chrooting happens.
** Subprivileged use case.
A fallback for ICMP tracing relevant to GNU/Linux is implemented,
allowing a rudimentary but suid-less use case. The ability to
identify intermediary hosts is missing, due to the crippled capability
of receiving ICMP packets other than ICMP_ECHOREPLY.
Telnet -E(no escape) is treating _POSIX_VDISABLE char as escape.
Causes problems when sending binary data through telnet connections.
** Use tty, not pty on Solaris.
Setting of terminal attributes as well setting of window size must be
done via the slave descriptor, not the master descriptor.
** Scrub USER from environment.
Discard the environment variable USER. It will later be set properly
for autologin, but at least one BSD system passes a preset value when
telnetd starts, a value which will cause rejected login when autologin
is not in effect.
** Portability of TTY termcap to Solaris systems.
** Portable option debugging.
BSD systems assign IAC and _POSIX_VDISABLE the common decimal value
255. Hence the NVT enforces value duplication more often during their
transmission, than GNU and Solaris systems do.
** Premature connection closure.
When many connections are attempted in quick succession, a substantial
number of them are cancelled. This does not appear for manual use
cases, but for contrived automated set-ups. The cause seems to be a
change in the evaluation of pty_read(), which was done to coincide
with the condition in use by the original BSD implementation. Issue
reported and suggested by Chris Severance in
** Update Canadian TLD server.
Old host name no longer exists. Reported by Neil Mayhem:
* Improved documentation, self-tests and build environment.
Description: PGP signature
- [platform-testers] please test inetutils 1.9.5 rc1,
Simon Josefsson <=