We suspect a shutdown race of some sort for which the full teardown is not
noticed during the close but should be during the free.
For that, we flag the conflux set as in full teardown (if so) in the free path
in case the close path didn't caught it.
Fixes#40870
Signed-off-by: David Goulet <dgoulet@torproject.org>
This has been misspelled when using consensus method 31 or later
since 0.4.6.1-alpha. Fixes bug 40869.
This commit is a backport of b9b0abd6c2 to 0.4.8.
Now that we never use an earlier consensus method, our tests actually
hit this, and we find that we have misspelled "maxunmeasurdbw" (sic) in
dirvote.c.
I have opened ticket #40869 to track this misspelling.
This patch causes `tor_compress_is_compression_bomb()` to emit a
warning-level log message that lets us learn the potential ratio of the
input to output buffer sizes. Hopefully, this will give us a bit of a
better idea whether the compression bomb ratio needs some tuning.
See: tpo/core/tor#40739.
This patch removes a call to `tor_assert_nonfatal_unreached()` in
`relay_key_is_unavailable_()` that is only called when Tor is compiled
without relay support.
Unfortunately, the non-fatal assertion causes a BUG log
message to appear for clients when they start up without relay support
for each CPU worker we spawn. This makes it spotting issues during
bootstrap harder particularly for our iOS developers.
Since the call sites to `get_master_identity_key()` handles `NULL`
values already, we do not think this will be an issue later on.
Reported by Benjamin Erhart (@tla) from Guardian Project.
Fixes tpo/core/tor#40848.
When we implemented prop275 in 0.4.8.1-alpha, we changed the
behavior of networkstatus_getinfo_helper_single to omit meaningful
published_on times, replacing them with "2038-01-01". This is
necessary when we're formatting a routerstatus with no additional
info, since routerstatus objects no longer include a published_on.
But in networkstatus_getinfo_by_purpose, we do have a routerinfo
that does have a published_on. This patch uses that information
to report published_on times in our output when we're making a
"virtual" networkstatus for a big file of routerinfo_t objects.
This is mostly important for bridge authorities, since when
they dump a secret list of the bridges, they want to include
published_on times.
Closes#40855. Bugfix on 0.4.8.1-alpha.
With LibreSSL-3.8.1 these engines are no long available causing a build
failure, but LibreSSL correctly defines OPENSSL_NO_ENGINE as part of its
opensslfeatures.h. However Tor includes crypto_openssl_mgt.h before any
of the openssl includes which would define OPENSSL_NO_ENGINE and then
fails to define DISABLE_ENGINES.
As the define is used in only a single .c file it is best to move it
there.
Signed-off-by: orbea <orbea@riseup.net>
Equi-X supports optionally allocating its solver memory using huge
pages, to reduce the virtual memory subsystem overhead required to make
the entire solver buffer live.
Tor doesn't use this feature, since it seems to have no noticeable
performance benefit at this time, but we still included code for it at
compile time. To improve portability, this patch disables huge page
support by default and enables it only in the cmake build system used
for equix benchmarks.
With this patch equix-bench still supports huge pages. Verified using
strace that we're making the hugepage allocation.
There's no fallback for huge pages, so Equi-X initialization will fail
if they are requested and we don't support them for any runtime or
compile-time reason.
Addresses #40843 (NetBSD) but also prevents future porting issues
related to huge pages.
NetBSD includes the idea of a 'maximum protection' per-region,
and an mprotect which exceeds the max protection will be denied.
If we explicitly ask for a maximum which includes execute permission, we
can successfully swap our code buffer's permissions between read-write
and read-execute when each hash program is compiled.
With this patch, the crypto/hashx tests pass on NetBSD 9.
This addresses bug #40844
As suggested by @wiz on #40843, let's add an explicit check to
hashx_vm_alloc_huge() that avoids using a Linux-style default
on NetBSD targets.
This doesn't change the questionable Linux-style default,
but a future patch will disable this code by default so it's not a
portability liability.
(This code is in hashx's VM layer but it's actually only relevant to
equix.)
This addresses bug #40843. Another patch will disable huge pages
by default entirely, but this patch is sufficient to fix the NetBSD
build.
I saw this test fail intermittently due to what seemed like a filesystem
race in docker? The cleanup task was failing with a 'directory not
empty' error, despite trying to do a recursive 'rm'. This patch adds an
'ls' to the same directory, hoping the output might be useful to
diagnose future intermittent failures.
This commit adds a new consensus method which, when present, causes
authorities not to consider package fingerprints when computing a
consensus. It builds on earlier work which dropped support for putting
these lines into the votes.
Considering a compression bomb before looking for errors led to false negative
log warnings. Instead, it is possible the work failed for whatever reasons
which is not indicative of a compression bomb.
Fixes#40739
Signed-off-by: David Goulet <dgoulet@torproject.org>
This was causing CI failures that didn't reproduce on my local machine.
The DoS subsystem now has a new assert() which triggers a BUG on some
nonzero memory contents (or_conn->tracked_for_dos_mitigation), and
uninitialized stack memory might be nonzero.
