In case of error, a negative value will be returned or NULL written into
first supplied argument.
This patch uses both cases to comply with style in the specific files.
A tor_vasprintf error in process_vprintf would lead to a NULL dereference
later on in buf_add, because the return value -1 casted to size_t would
pass an assertion check inside of buf_add.
On the other hand, common systems will fail on such an operation, so it
is not a huge difference to a simple assertion. Yet it is better to
properly fail instead of relying on such behaviour on all systems.
Signed-off-by: Tobias Stoeckmann <tobias@stoeckmann.org>
We have a getaddrinfo() implementation that we prefer, and a
gethostbyname*() implementation that we fall back on. Give them
both the same interface, and let them be called by the same name.
This is a preparatory step for making them both mockable.
The documentation for this function says that the smartlist can
contain NULLs, but the code only handled NULLs if they were at the
start of the list.
We didn't notice this for a long time, because when Tor is run
normally, the sequence of msg_id_t is densely packed, and so this
list (mapping msg_id_t to channel_id_t) contains no NULL elements.
We could only run into this bug:
* when Tor was running in embedded mode, and starting more than once.
* when Tor ran first with more pubsub messages enabled, and then
later with fewer.
* When the second run (the one with fewer enabled pubsub messages)
had at least some messages enabled, and those messages were not
the ones with numerically highest msg_id_t values.
Fixes bug 31898; bugfix on 47de9c7b0a
in 0.4.1.1-alpha.
There is a bad design choice in two of our configuration types,
where the empty string encodes a value that is not the same as the
default value. This design choice, plus an implementation mistake,
meant that config_dup() did not preserve the value of routerset_t,
and thereby caused bug #31495.
This comment-only patch documents the two types with the problem,
and suggests that implementors try to avoid it in the future.
Closes ticket 31907.
The log mutex is dynamically initialized, guarded by log_mutex_initialized.
We don't want to destroy it, because after it is destroyed, we won't see
any more logs.
If tor is re-initialized, log_mutex_initialized will still be 1. So we
won't trigger any undefined behaviour by trying to re-initialize the
log mutex.
Part of 31736, but committed in this branch to avoid merge conflicts.
cb_buf_mutex is statically initialised, so we can not destroy it when
we are shutting down the err subsystem. If we destroy it, and then
re-initialise tor, all our backtraces will fail.
Part of 31736, but committed in this branch to avoid merge conflicts.
Move SEVERITY_MASK_IDX() to log.h private/unit tests section, so that
we can use it in log.c, the unit tests, and the fuzzers.
(The test and fuzzer code changes are in a subsequent commit.)
Preparation for bug 31334.
When processing a %included folder, a bug caused the pointer to
the last element of the options list to be set to NULL when
processing a file with only comments or whitepace. This could
cause options from other files on the same folder to be
discarded depending on the lines after the affected %include.
Since the flags are now stored with compatible numbering, we can
just OR them together and see whether the flag we want is in the
result.
(Net code removal!)
Using a bitfield here will enable us to unify the var_type_def_t flags
with the config_var_t flags.
(This commit does not yet do that unification, and does not yet
rename or refactor any flags. It only changes booleans into bits.)
Previously they checked the individual flags inside var_type_def_t;
now they call the appropriate var_type_is_*() functions.
(These functions will be removed entirely by the end of this branch.)
These errors can occur if we are built on a system with support for
madvise(MADV_NOFORK) but then we are run on a system whose kernel
does not support that flag.
If the error is something that we don't tolerate at all, we now log
it before crashing.
Fixes bug 31696. I am calling this a bugfix on 0.4.1.1-alpha, where
we actually started using the map_anon code.
This is similar to, but not the same as, the fix for #31570.
We used to do this on Windows only, but it appears to affect
multiple platforms when building with certain versions of GCC, and a
common pattern for defining the floating-point classifier functions.
Fixes part of 31687. I'm calling this a bugfux on 31687, when we
started suppressing these warnings on Windows.
Fix levels for subsystems that depend on log/err
* winprocess (security) doesn't use err:
* call windows process security APIs as early as possible
* init err after winprocess
* move wallclock so it's still after err
* network and time depend on log:
* make sure that network and time can use logging.
* init network and time after log
Add comments explaining the module init order.
Fixes bug 31615; bugfix on 0.4.0.1-alpha.
Now that the variants of these functions that took config_line_t are
gone, there is no longer any reason for the remaining variants to
have "ex" at the end of their names.
This commit was made by running this perl script over all the files
in src/:
#!/usr/bin/perl -w -i -p
s{typed_var_(assign|free|encode|copy|eq|ok|kvassign|kvencode|mark_fragile)_ex}
{typed_var_$1}g;
Previously we used int in some places and off_t for others. Neither
is correct: ptrdiff_t is right for differences between pointers.
(off_t is only for offsets and sizes on the filesystem.)
Also add an explanation of a possible future refactoring where we
might remove the config_type_t enumeration entierly.
Fixes ticket 31624.
No changes file, since this is a comment-only change.
Previously we used int here, but it is more correct to use
ptrdiff_t. (This never actually matters for our code in practice,
since the structure we are managing here never exceed INT_MAX in
size.)
