This utility function is useful outside of epan. Move it to wsutil
and export the interface.
The move isn't completely clean as it requires duplicating two small
inline functions but that was necessary to avoiding moving too much at
once.
We have two format_size()s, with and without wmem scoped memory.
Move the wmem version to wsutil and add a convenience macro to
use g_malloc()ed memory.
This allows wmem to be used from other libraries, namely wsutil.
It is often the case that a funtion exists in wsutil and cannot
be used with a wmem scope, requiring some code duplication or
extra memory allocations, or vice-versa, code in epan cannot be
moved to wsutil because it has a wmem dependency.
To this end wmem is moved to wsutil. Scope management remains part
of epan because those scope semantics are specific to dissection.
fcntl.h appears to be available on all of our supported platforms,
including Windows. We've also been including it without HAVE_FCNTL_H
guards in a few places (e.g. sshdump.c) without any issues for some
time.
floorl is part of C99.
Mark wsutil's includes SYSTEM PRIVATE. This exposed a lot of targets
that were indirectly picking up include paths via the wsutil target, so
add direct includes where needed. The G.722 and G.726 codecs were
implicilty including tiffio.h; find it explicitly instead.
Mark some of wsutil's libraries PRIVATE, but leave commonly-used ones
PUBLIC.
Ping #17477.
Revert change to format_size() added in
f509a83381. This commit broke formatting
with spaces and introduced some dead code.
Also replace unnecessary call to format_size_wmem() and remove
unnecessary casts (since our warning settings were fixed in the
mean time).
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.
At least according to the Single UNIX Standard, it merely has to be big
enough to hold a value in the range [-1, 1000000], and there must be
*an* environment in which it's no *larger* than a long.
Just cast it to long, and continue to print the result of dividing it by
1000 with %03ld.
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.
This includes as little as possible in the assertion header, so
that it can be included globally in every file without pulling
any unwanted definitions. In particular pulling stdlib.h is
avoided because that can have side effects if it wants to
include non-portable extensions.
It is possible to have side-effects from include glib.h too, for
example because of G_LOG_DOMAIN.
These side-effects are usually avoidable with careful ordering
of pre-processor directives but with multiple levels of indirections
it can be hard to track. Better to make it robust to these kinds
of failures in the first place.
Also integrate with our logger for a cohesive experience (but
keep it a private dependency).
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.
This avoids having to manage two different implementations.
For example with this change GLib functions will terminate
if Wireshark's fatal log level is set to a matching level
and the --log-file option will also output messages from
GLib itself.
This changes color use to be the very similar with GLib to
maintain familiarity. The only difference is that Message
and Info use a different color than Debug.
Also use the more familiar format of <domain> <level> instead
of <level> <domain>.
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.
Currently we are not filtering the unset (NULL) domain, on
the assumption that every log call should belong to a defined
domain.
However there are still many places in the codebase where this isn't
true and the fact that the null/default domain name is omitted from
the output and never filtered is probably surprising and user-unfriendly.
Users might understandably assume the filtering is buggy.
Give an indication, such as (none)-MESSAGE, to make this more
obvious.
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.
Alexis La Goutte