shm: eliminate race/deadlock in region subscriptions

This commit is contained in:
Alexey Rybalchenko
2021-04-12 09:36:17 +02:00
parent c6a6a5f21b
commit 2c89b24857
2 changed files with 23 additions and 17 deletions

View File

@@ -384,10 +384,10 @@ class Manager
void RemoveRegion(const uint16_t id)
{
fRegions.erase(id);
{
boost::interprocess::scoped_lock<boost::interprocess::named_mutex> lock(fShmMtx);
fShmRegions->at(id).fDestroyed = true;
fRegions.erase(id);
(fEventCounter->fCount)++;
}
fRegionsGen += 1; // signal TL cache invalidation
@@ -483,19 +483,26 @@ class Manager
auto infos = GetRegionInfoUnsafe();
for (const auto& i : infos) {
auto el = fObservedRegionEvents.find({i.id, i.managed});
if (el == fObservedRegionEvents.end()) {
fRegionEventCallback(i);
if (el == fObservedRegionEvents.end()) { // if event id has not been observed
fObservedRegionEvents.emplace(std::make_pair(i.id, i.managed), i.event);
++fNumObservedEvents;
} else {
// if a region has been created and destroyed rapidly, we could see 'destroyed' without ever seeing 'created'
// TODO: do we care to show 'created' events if we know region is already destroyed?
if (i.event == RegionEvent::created) {
fRegionEventCallback(i);
++fNumObservedEvents;
} else {
fNumObservedEvents += 2;
}
} else { // if event id has been observed (expected - there are two events per id - created & destroyed)
// fire a callback if we have observed 'created' event and incoming is 'destroyed'
if (el->second == RegionEvent::created && i.event == RegionEvent::destroyed) {
fRegionEventCallback(i);
el->second = i.event;
++fNumObservedEvents;
} else {
// LOG(debug) << "ignoring event for id" << i.id << ":";
// LOG(debug) << "incoming event: " << i.event;
// LOG(debug) << "stored event: " << el->second;
// LOG(debug) << "ignoring event for id " << i.id << ":"
// << " incoming: " << i.event << ","
// << " stored: " << el->second;
}
}
}
@@ -688,7 +695,7 @@ class Manager
std::thread fRegionEventThread;
bool fRegionEventsSubscriptionActive;
std::function<void(fair::mq::RegionInfo)> fRegionEventCallback;
std::map<std::pair<uint16_t, bool>, RegionEvent> fObservedRegionEvents;
std::map<std::pair<uint16_t, bool>, RegionEvent> fObservedRegionEvents; // pair: <region id, managed>
uint64_t fNumObservedEvents;
DeviceCounter* fDeviceCounter;