diff options
Diffstat (limited to 'tsan-suppress.txt')
| -rw-r--r-- | tsan-suppress.txt | 62 |
1 files changed, 62 insertions, 0 deletions
diff --git a/tsan-suppress.txt b/tsan-suppress.txt new file mode 100644 index 000000000..ca6d17dc3 --- /dev/null +++ b/tsan-suppress.txt @@ -0,0 +1,62 @@ +# +# This file is part of the LibreOffice project. +# +# This Source Code Form is subject to the terms of the Mozilla Public +# License, v. 2.0. If a copy of the MPL was not distributed with this +# file, You can obtain one at http://mozilla.org/MPL/2.0/. +# + +# Use via environment variable TSAN_OPTIONS=suppressions=.../tsan-suppress.txt + +# There looks to be a lock ordering problem here, but I can't see how it could +# actually be hit in practice. +deadlock:cppuhelper::ServiceManager::disposing() +deadlock:cppuhelper::ServiceManager::loadImplementation + +# Ignore stuff in external DBUS library +# Some kind of dbus lock issue when we call it from psp::CUPSManager +deadlock:_dbus_lock + +# inside an assert so I don't care +race:cppu::OWeakConnectionPoint::acquire +race:AffineBridge::v_enter +race:__vsnprintf_chk + +# right now, I'm not interested in deadlocks at all, too many false+ +deadlock: + +# This is checking SAL_STRING_IS_STATIC, which is safe because that is written at compile time. +# Depending on the optimisation/debug level, the call stack may have different leaves, hence +# specifying this in more than one way. +race:rtl::str::acquire<_rtl_uString> +race:rtl::str::release<_rtl_uString> +race:rtl_uString_acquire +race:rtl_uString_release +race:rtl_uString_assign + +# I've convinced myself this is a false+, caused by ping-ponging the buffer between two +# threads, but I might be wrong +race:XBufferedThreadedStream::getNextBlock + +# not introduced in stuff the embedded JVM does +race:libjvm.so + +# I think this is OK, because at this point we are doing +# if (nRefCount > 1) +# and we know from our callers that the refcount must be at least one +# so there is no failure mode +race:ireallocSequence + +# TODO There appears to be a race here, initialising the +# ::com::sun::star::uno::Sequence< T >::s_pType +# field. But no idea at all how to fix it. +race:cppu::getTypeFavourUnsigned + + +# This is all inside GIO/Glib, no idea what it is doing +# +race:slab_allocator_alloc_chunk +race:g_source_destroy_internal +race:g_source_unref_internal +race:g_task_finalize +race:g_socket_send_message_with_timeout |