With ThreadSanitizer support added to the Userspace RCU, we no longer
need to wrap the call_rcu and caa_container_of with
__tsan_{acquire,release} hints. Remove the direct calls to
__tsan_{acquire,release} and the isc_urcu_{container,cleanup} macros.
The cds_lfht_for_each_entry and cds_lfht_for_each_entry_duplicate macros
had a code that operated on the NULL pointer, at the end of the list it
was calling caa_container_of() on the NULL pointer in the init-clause
and iteration-expression, but the result wasn't actually used anywhere
because the cond-expression in the for loop has prevented executing
loop-statement. This made AddressSanitizer notice the invalid operation
and rightfully complain.
This was reported to the upstream and fixed there. Pull the upstream
fix into our <isc/urcu.h> header, so our CI checks pass.
The isc_stats_create() can no longer return anything else than
ISC_R_SUCCESS. Refactor isc_stats_create() and its variants in libdns,
libns and named to just return void.
to reduce the amount of common code that will need to be shared
between the separated cache and zone database implementations,
clean up unused portions of dns_db.
the methods dns_db_dump(), dns_db_isdnssec(), dns_db_printnode(),
dns_db_resigned(), dns_db_expirenode() and dns_db_overmem() were
either never called or were only implemented as nonoperational stub
functions: they have now been removed.
dns_db_nodefullname() was only used in one place, which turned out
to be unnecessary, so it has also been removed.
dns_db_ispersistent() and dns_db_transfernode() are used, but only
the default implementation in db.c was ever actually called. since
they were never overridden by database methods, there's no need to
retain methods for them.
in rbtdb.c, beginload() and endload() methods are no longer defined for
the cache database, because that was never used (except in a few unit
tests which can easily be modified to use the zone implementation
instead). issecure() is also no longer defined for the cache database,
as the cache is always insecure and the default implementation of
dns_db_issecure() returns false.
for similar reasons, hashsize() is no longer defined for zone databases.
implementation functions that are shared between zone and cache are now
prepended with 'dns__rbtdb_' so they can become nonstatic.
serve_stale_ttl is now a common member of dns_db.
When compiled using a malloc that lacks an equivalent to sallocx(),
the jemalloc_shim adds a size prefix to each allocation. We must check
that this does not overflow.
Closes#4121
As well as clearing the fresh memory, `calloc()`-like functions must
ensure that the count and size do not overflow when multiplied.
Use `isc_mem_callocate()` in `isc__uv_calloc()`.
The `ISC_OVERFLOW_XXX()` macros are usually wrappers around
`__builtin_xxx_overflow()`, with alternative implementations
for compilers that lack the builtins.
Replace the overflow checks in `isc/time.c` with the new macros.
The free_all_cpu_call_rcu_data() call can consume hundreds of
milliseconds on shutdown. Don't try to be smart and let the RCU library
handle this internally.
In HTTP/1.0 and HTTP/1.1, RFC 9112 section 9.6 says the last response
in a connection should include a `Connection: close` header, but the
statschannel server omitted it.
In an HTTP/1.0 response, the statschannel server can sometimes send a
`Connection: keep-alive` header when it is about to close the
connection. There are two ways:
If the first request on a connection is keep-alive and the second
request is not, then _both_ responses have `Connection: keep-alive`
but the connection is (correctly) closed after the second response.
If a single request contains
Connection: close
Connection: keep-alive
then RFC 9112 section 9.3 says the keep-alive header is ignored, but
the statschannel sends a spurious keep-alive in its response, though
it correctly closes the connection.
To fix these bugs, make it more clear that the `httpd->flags` are part
of the per-request-response state. The Connection: flags are now
described in terms of the effect they have instead of what causes them
to be set.
The isc_result_t enum was to sparse when each library code would skip to
next << 16 as a base. Remove the huge holes in the isc_result_t enum to
make the isc_result tables more compact.
This change required a rewrite how we map dns_rcode_t to isc_result_t
and back, so we don't ever return neither isc_result_t value nor
dns_rcode_t out of defined range.
In some cases, the inlined version rcu_dereference() would not compile
when working on pointer to opaque struct (namely Ubuntu Jammy). Detect
such condition in the autoconf and disable the inlining of the small
functions if it breaks the build.
isc_mem_put NULL's the pointer to the memory being freed. The
equality test 'parent->r == node' was accidentally being turned
into a test against NULL.
RUNTIME_CHECK on the "wrap" variable avoids possible NULL dereference:
thread.c: In function 'thread_wrap':
thread.c:60:15: error: dereference of possibly-NULL 'wrap' [CWE-690] [-Werror=analyzer-possible-null-dereference]
60 | *wrap = (struct thread_wrap){
The RUNTIME_CHECK was there before
7d1ceaf35d.
Move registration and deregistration of the main thread from
`isc_loopmgr_run()` into `isc__initialize()` / `isc__shutdown()`:
liburcu-qsbr fails an assertion if we try to use it from an
unregistered thread, and we need to be able to use it when the
event loops are not running.
Use `rcu_assign_pointer()` and `rcu_dereference()` in qp-trie
transactions so that they properly mark threads as online. The
RCU-protected pointer is no longer declared atomic because
liburcu does not (yet) use standard C atomics.
Fix the definition of `isc_qsbr_rcu_dereference()` to return
the referenced value, and to call the right function inside
liburcu.
Change the thread sanitizer suppressions to match any variant of
`rcu_*_barrier()`
An omission pointed out by the following report from Coverity:
/lib/isc/loop.c: 483 in isc_loopmgr_pause()
>>> CID 455002: Error handling issues (CHECKED_RETURN)
>>> Calling "uv_async_send" without checking return value (as is done elsewhere 5 out of 6 times).
483 uv_async_send(&loop->pause_trigger);
when reading on a streamdns socket failed due to timeout, but
the dispatch was still waiting for other responses, it would
resume reading by calling isc_nm_read() again. this caused
an assertion because the socket was already reading.
we now check that either the socket is reading, or that it was
already reading on the same handle.
Create and free per-CPU helper threads from the main thread and tell
thread sanitizer to suppress leaking threads. (We are not leaking
threads ourselves and we can safely ignore the Userspace-RCU thread
leaks.)
All the places the qp-trie code was using `call_rcu()` needed
`__tsan_release()` and `__tsan_acquire()` annotations, so
add a couple of wrappers to encapsulate this pattern.
With these wrappers, the tests run almost clean under thread
sanitizer. The remaining problems are due to `rcu_barrier()`
which can be suppressed using `.tsan-suppress`. It does not
suppress the whole of `liburcu`, because we would like thread
sanitizer to detect problems in `call_rcu()` callbacks, which
are called from `liburcu`.
The CI jobs have been updated to use `.tsan-suppress` by
default, except for a special-case job that needs the
additional suppressions in `.tsan-suppress-extra`.
We might be able to get rid of some of this after liburcu gains
support for thread sanitizer.
Note: the `rcu_barrier()` suppression is not entirely effective:
tsan sometimes reports races that originate inside `rcu_barrier()`
but tsan has discarded the stack so it does not have the
information required to suppress the report. These "races" can
be made much easier to reproduce by adding `atexit_sleep_ms=1000`
to `TSAN_OPTIONS`. The problem with tsan's short memory can be
addressed by increasing `history_size`: when it is large enough
(6 or 7) the `rcu_barrier()` stack usually survives long enough
for suppression to work.
Memory reclamation by `call_rcu()` is asynchronous, so during shutdown
it can lose a race with the destruction of its memory context. When we
defer memory reclamation, we need to attach to the memory context to
indicate that it is still in use, but that is not enough to delay its
destruction. So, call `rcu_barrier()` in `isc_mem_destroy()` to wait
for pending RCU work to finish before proceeding to destroy the memory
context.
It can be fairly long-winded to allocate space for a struct with a
flexible array member: in general we need the size of the struct, the
size of the member, and the number of elements. Wrap them all up in a
STRUCT_FLEX_SIZE() macro, and use the new macro for the flexible
arrays in isc_ht and dns_qp.
The Userspace-RCU headers are now needed for more parts of the libisc
and libdns, thus we need to add it globally to prevent compilation
failures on systems with non-standard Userspace-RCU installation path.
The isc_quota API was using locked list of isc_job_t objects to keep the
waiting TCP accepts. Change the isc_quota implementation to use
cds_wfcqueue internally - the enqueue is wait-free and only dequeue
needs to be locked.
The isc_async API was using lock-free stack (where enqueue operation was
not wait-free). Change the isc_async to use cds_wfcqueue internally -
enqueue and splice (move the queue members from one list to another) is
nonblocking and wait-free.
Instead of having a global hashtable with a global rwlock for the GLUE
cache, move the glue_list directly into rdatasetheader and use
Userspace-RCU to update the pointer when the glue_list is empty.
Additionally, the cached glue_lists needs to be stored in the RBTDB
version for early cleaning, otherwise the circular dependencies between
nodes and glue_lists will prevent nodes to be ever cleaned up.
Clang 16 LeakSanitizer reports a memory leak when dns_request_create()
returned a TLS error in the nsupdate system test. While technically a
memory leak on error handling, it's not a problem because the program is
immediately terminated; nsupdate is not expected to run for a prolonged
time.
Stop deliberately breaking const rules by copying file->name into
dirbuf and truncating it there. Handle files located in the root
directory properly. Use unlinkat() from POSIX 200809.
Removing old timestamp or increment versions of log backup files did
not work when the file is an absolute path: only the entry name was
provided to the file remove function.
The dirname was also bogus, since the file separater was put back too
soon.
Fix these issues to make log file rotation work when the file is
configured to be an absolute path.
All the per-loop `libuv` setup remains in `isc_loop`, but the per-thread
RCU setup is moved to `isc_thread` alongside the other per-thread setup.
This avoids repeating the per-thread setup for `call_rcu()` helpers,
and explains a little better why some parts of the per-thread setup
is missing for `call_rcu()` helpers.
This also removes the per-loop `call_rcu()` helpers as we refactored the
isc__random_initialize() in the previous commit.
Instead of writing complicated wrappers for every thread, move the
initialization back to isc_random unit and check whether the random seed
was initialized with a thread_local variable.
Ensure that isc_entropy_get() returns a non-zero seed.
This avoids problems with thread sanitizer tests getting stuck in an
infinite loop.
Remove the `isc_threadarg_t` and `isc_threadresult_t`
typedefs which were unhelpful disguises for `void *`,
and free the dummy jemalloc allocation sooner.
When liburcu is not installed from a system package, its headers are
not treated as system headers by the compiler, so BIND's -Werror and
other warning options take effect. The liburcu headers have a lot
of inline functions, some of which do not use all their arguments,
which BIND's build treats as an error.
This commit allows BIND 9 to be compiled with different flavours of
Userspace RCU, and improves the integration between Userspace RCU and
our event loop:
- In the RCU QSBR, the thread is put offline when polling and online
when rcu_dereference, rcu_assign_pointer (or friends) are called.
- In other RCU modes, we check that we are not reading when reaching the
quiescent callback in the event loop.
- We register the thread before uv_work_run() callback is called and
after it has finished. The rcu_(un)register_thread() has a large
overhead, but that's fine in this case.
There's a recurring pattern walking the ISC_LISTs that just repeats over
and over. Add two macros:
* ISC_LIST_FOREACH(list, elt, link) - walk the static list
* ISC_LIST_FOREACH_SAFE(list, elt, link, next) - walk the list in
a manner that's safe against list member deletions
The spinlock is small (atomic_uint_fast32_t at most), lightweight
synchronization primitive and should only be used for short-lived and
most of the time a isc_mutex should be used.
Add a isc_spinlock unit which is either (most of the time) a think
wrapper around pthread_spin API or an efficient shim implementation of
the simple spinlock.
When shutting down TCP sockets, the read callback calling logic was
flawed, it would call either one less callback or one extra. Fix the
logic in the way:
1. When isc_nm_read() has been called but isc_nm_read_stop() hasn't on
the handle, the read callback will be called with ISC_R_CANCELED to
cancel active reading from the socket/handle.
2. When isc_nm_read() has been called and isc_nm_read_stop() has been
called on the on the handle, the read callback will be called with
ISC_R_SHUTTINGDOWN to signal that the dormant (not-reading) socket
is being shut down.
3. The .reading and .recv_read flags are little bit tricky. The
.reading flag indicates if the outer layer is reading the data (that
would be uv_tcp_t for TCP and isc_nmsocket_t (TCP) for TLSStream),
the .recv_read flag indicates whether somebody is interested in the
data read from the socket.
Usually, you would expect that the .reading should be false when
.recv_read is false, but it gets even more tricky with TLSStream as
the TLS protocol might need to read from the socket even when sending
data.
Fix the usage of the .recv_read and .reading flags in the TLSStream
to their true meaning - which mostly consist of using .recv_read
everywhere and then wrapping isc_nm_read() and isc_nm_read_stop()
with the .reading flag.
4. The TLS failed read helper has been modified to resemble the TCP code
as much as possible, clearing and re-setting the .recv_read flag in
the TCP timeout code has been fixed and .recv_read is now cleared
when isc_nm_read_stop() has been called on the streaming socket.
5. The use of Network Manager in the named_controlconf, isccc_ccmsg, and
isc_httpd units have been greatly simplified due to the improved design.
6. More unit tests for TCP and TLS testing the shutdown conditions have
been added.
Co-authored-by: Ondřej Surý <ondrej@isc.org>
Co-authored-by: Artem Boldariev <artem@isc.org>
By inspecting the code, it was discovered that .sendbuf member of the
isc__nm_networker_t was unused and just consuming ~64k per worker.
Remove the member and the association allocation/deallocation.
In e185412872, the TCP accept quota code
became broken in a subtle way - the quota would get initialized on the
first accept for the server socket and then deleted from the server
socket, so it would never get applied again.
Properly fixing this required a bigger refactoring of the isc_quota API
code to make it much simpler. The new code decouples the ownership of
the quota and acquiring/releasing the quota limit.
After (during) the refactoring it became more clear that we need to use
the callback from the child side of the accepted connection, and not the
server side.
