macOS /w Xcode 15: force the old linker when targeting macOS 11 or lower
[libreoffice.git] / tsan-suppress.txt
blobe1eb43c0a7728588e69098a4a97c2e6ebec42556
2 # This file is part of the LibreOffice project.
4 # This Source Code Form is subject to the terms of the Mozilla Public
5 # License, v. 2.0. If a copy of the MPL was not distributed with this
6 # file, You can obtain one at http://mozilla.org/MPL/2.0/.
9 # Use via environment variable TSAN_OPTIONS=suppressions=.../tsan-suppress.txt
11 # There looks to be a lock ordering problem here, but I can't see how it could
12 # actually be hit in practice.
13 deadlock:cppuhelper::ServiceManager::disposing()
14 deadlock:cppuhelper::ServiceManager::loadImplementation
15 deadlock:AffineBridge::v_callInto_v
17 # Ignore stuff in external DBUS library.
18 # Some kind of dbus lock issue when we call it from psp::CUPSManager.
19 deadlock:_dbus_lock
21 # inside an assert so I don't care.
22 race:cppu::OWeakConnectionPoint::acquire
23 race:AffineBridge::v_enter
24 race:__vsnprintf_chk
26 # right now, I'm not interested in deadlocks at all, too many false+
27 deadlock:
29 # This is checking SAL_STRING_IS_STATIC, which is safe because that is written at compile time.
30 # Depending on the optimisation/debug level, the call stack may have different leaves, hence
31 # specifying this in more than one way.
32 race:rtl::str::acquire<_rtl_uString>
33 race:rtl::str::release<_rtl_uString>
34 race:rtl_uString_acquire
35 race:rtl_uString_release
36 race:rtl_uString_assign
38 # I've convinced myself this is a false+, caused by ping-ponging the buffer between two
39 # threads, but I might be wrong.
40 race:XBufferedThreadedStream::getNextBlock
42 # I am not interested in stuff the embedded JVM does.
43 race:libjvm.so
45 # I think this is OK, because at this point we are doing
46 #  if (nRefCount > 1)
47 # and we know from our callers that the refcount must be at least one
48 # so there is no failure mode.
49 race:ireallocSequence
51 # TODO There appears to be a race here, initialising the
52 #   ::com::sun::star::uno::Sequence< T >::s_pType
53 # field. But no idea at all how to fix it.
54 race:cppu::getTypeFavourUnsigned
57 # This is all inside GIO/Glib, no idea what it is doing.
59 race:slab_allocator_alloc_chunk
60 race:g_source_destroy_internal
61 race:g_source_unref_internal
62 race:g_task_finalize
63 race:g_socket_send_message_with_timeout
64 race:g_idle_source_new
65 race:g_slice_alloc0
67 # Python build, lock order inversion
68 deadlock:take_gil
70 # Not interested in deadlock issues when they involve the shutdown path 
71 # - the shutdown path is special, so if we see a lock-ordering here that conflicts with
72 # a "normal" lock-ordering path, that is highly unlikely to result in a real issue.
73 deadlock:DeInitVCL
75 # Sometimes tsan will report data-races in these, who knows why, they should be thread-safe
76 race:malloc
77 race:free
78 race:close
80 # There is a lock-cycle here, together with the mutexes acquired in
81 #    SvtSysLocaleOptions::SvtSysLocaleOptions
82 #    SvtSysLocale::SvtSysLocale
83 # but I can't see how it could lead to a problem in practice
84 deadlock:ItemHolder1::impl_addItem
86 # ignore warning about nRefCount in sal_Sequence.
87 # <sberg> noelgrandin, we generally assume that sal_Int32 is small enough, and hardware sane enough,
88 #      that reading it will always give a sane value, i.e., if the read happens in parallel with a write,
89 #      the read produces either the old, original value, or the newly written one.
90 race:cppu::createEmptySequence
91 race:uno_type_sequence_reference2One