Add a command line option --log-fatal-domains= and environment variable
WIRESHARK_LOG_FATAL_DOMAINS that aborts the programs if a domain in
the list is logged to.
Negative matches for fatal log domains not implemented for now,
pending a relevant use-case.
Add a LOG_LEVEL_ECHO that is always active and always non-fatal.
Use that to implement a WS_DEBUG_HERE() macro for quick print outs
during debugging sessions.
For historical reasons our logging inherited from GLib the logging of
some levels to stdout. Namely levels "info" and "debug" (to which we
added "noisy").
However this practice is discouraged because it mixes debug output
with application output for CLI tools and breaks many common usage
scenarios, like using tshark in pipes.
This change flips the logic on wslog to make logging to stderr the
default behavior.
Extcap subprocess have a hidden dependency on stdout so add that.
Some GUI users may also have a dependency on stdout. Because
GUI tools are unlikely to depend on stdout for programatic output
add another exception for wireshark GUI, to preserve backward
compatibility.
This matches the original implementation and allows displaying
logs to the console, including debug information, when running
an extcap from the CLI for testing and development purposes.
This should make extcap logging bug-for-bug compatible with the
behavior before dc7f0b88bb.
Imitate the GLib logic for selecting the console output stream
according to the log level. Levels MESSAGE and above go to
stderr. INFO and below go to stdout, unless stderr is chosen
using ws_log_console_writer_set_use_stderr().
It turns out some old extcap code was subtly dependending
on this behavior.
Extcaps require a log file when invoked in child mode. It also has
a specific flag to enable debugging, other that the wslog options.
Fix the logging to:
1. Enable debug log level if --debug is used.
2. Do not emit messages to the stderr if debug is enabled.
This brings extcap logging to the same feature level it had before
wslog replaced GLib logging.
Instead of removing extra log information in the log handler
for the default log level, do it in the ws_message() macro.
This means ws_log_full() will work as expected.
Have these functions accept a zero max length to mean "display
the whole byte array". Change the max length parameter to receive a
number of bytes to print, not the length of the output character
string.
Adjust the macros bytes_to_str() and bytes_to_string_punct() to
produce the same output. Add more tests. Rename the functions to
bytes_to_str_maxlen() and bytes_to_str_punct_maxlen() because this is
an API break.
Functions clock_gettime() and timespec_get() cover all the platforms
we support with sub-second resolution in a a portable manner. Fallback
to using time().
Pass a struct timespec to the log writer callback for maximum
flexibility.
Calling GLib functions inside the log writer is not safe,
it might infinitely recurse or abort if g_date_time_* logs
warnings because we registered our log handler for GLib itself.
Minimizing the dependencies on other wsutil and GLib functions
reduces the chance that we will have a weird recursion pattern
in wslog and makes the code easier to analyze.
Instead of receiving the program name from GLib, pass it explicitly
to ws_log_init() instead and use that to initialize the GLib program
name.
ws_log_parse_args() will now exit the program when it encounters an
argument error if exit_failure >= 0.
The --log-debug and --log-noisy now accepts a '!' to invert the
match and disable the debug (noisy respectively) log level for
the listed domains.
Note this is different from --log-domains, that option
enables/disables the entire log domain itself, regardless of log
level.
ws_log_domains.h needs to be included before wslog.h to be used
to define WS_LOG_DOMAIN. Also the definition for enum ws_log_level
needs to be exported for other APIs so move that to ws_log_domains.h
and rename the file to ws_log_defs.h to reflect the new scope.
This is intended to replace logging in dissectors that has a
debug level with #ifdef DEBUG_foo and an extra level guarded
by a #ifdef DEBUG_EXTRA_foo.
But generally it can be used as another level of granularity
for debugging output, to avoid flooding the log with too
much information with typical usage.
Rename the filter functions without the unnecessary 'str'
suffix.
Option --log-debug or WIRESHARK_LOG_DEBUG is a list
of domains that are set to a "debug" log level. This
takes precedence over the normal log level and domain
filter options.
Enviroment variable WIRESHARK_LOG_FATAL and command line
option --log-fatal set the fatal log level. Messages with
fatal or highr priority cause the program to abort. By
default the fatal level is "error", but it can be set to
"critical" or "warning" with this option.
Domain filter expressions starting with '!' invert the match.
Only domains that do not match become active. Note that '!'
must be the first character in the filter and applies to the
whole expression.
A domain filter can be given in the environment variable
'WS_LOG_DOMAINS' or in a command-line options "--log-domains".
The filter is specified as a comma separated case insensitive list,
for example:
./tshark --log-domains=main,capture
Domain data type switches from an enum to a string. There is no
constaint on adding new domains, neither in code or at runtime.
The string format is arbitrary, only positive matches will produce
output.
Experience has shown that:
1. The current logging methods are not very reliable or practical.
A logging bitmask makes little sense as the user-facing interface (who
would want debug but not crtical messages for example?); it's
computer-friendly and user-unfriendly. More importantly the console
log level preference is initialized too late in the startup process
to be used for the logging subsystem and that fact raises a number
of annoying and hard-to-fix usability issues.
2. Coding around G_MESSAGES_DEBUG to comply with our log level mask
and not clobber the user's settings or not create unexpected log misses
is unworkable and generally follows the principle of most surprise.
The fact that G_MESSAGES_DEBUG="all" can leak to other programs using
GLib is also annoying.
3. The non-structured GLib logging API is very opinionated and lacks
configurability beyond replacing the log handler.
4. Windows GUI has some special code to attach to a console,
but it would be nice to abstract away the rest under a single
interface.
5. Using this logger seems to be noticeably faster.
Deprecate the console log level preference and extend our API to
implement a log handler in wsutil/wslog.h to provide easy-to-use,
flexible and dependable logging during all execution phases.
Log levels have a hierarchy, from most verbose to least verbose
(debug to error). When a given level is set everything above that
is also enabled.
The log level can be set with an environment variable or a command
line option (parsed as soon as possible but still later than the
environment). The default log level is "message".
Dissector logging is not included because it is not clear what log
domain they should use. An explosion to thousands of domains is
not desirable and putting everything in a single domain is probably
too coarse and noisy. For now I think it makes sense to let them do
their own thing using g_log_default_handler() and continue using the
G_MESSAGES_DEBUG mechanism with specific domains for each individual
dissector.
In the future a mechanism may be added to selectively enable these
domains at runtime while trying to avoid the problems introduced
by G_MESSAGES_DEBUG.
João Valverde