livie
/
phoenix-firestorm
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

DRTVWR-209 Additional merge of viewer-development with SH-3316 drano-http code. 

Restore original deleteRequest/removeRequest implementation removing
a small race.  Remove a short-lived additional timeout scheme on requests
which really isn't appropriate as originally implemented as we can have
very long-lived requests on big regions.
meow-7.2.2
Monty Brandenberg 2012-09-07 15:08:12 -04:00
parent 04e8d074b2
commit 09cd2a4b1a
1 changed files with 28 additions and 33 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

61
indra/newview/lltexturefetch.cpp
View File

@ -1068,8 +1068,6 @@ void LLTextureFetchWorker::startWork(S32 param)
// Threads: Ttf
bool LLTextureFetchWorker::doWork(S32 param)
{
static const F32 FETCHING_TIMEOUT = 120.f;//seconds
static const LLCore::HttpStatus http_not_found(HTTP_NOT_FOUND); // 404
static const LLCore::HttpStatus http_service_unavail(HTTP_SERVICE_UNAVAILABLE); // 503
static const LLCore::HttpStatus http_not_sat(HTTP_REQUESTED_RANGE_NOT_SATISFIABLE); // 416;
@ -1637,32 +1635,12 @@ bool LLTextureFetchWorker::doWork(S32 param)
}
else
{
// *FIXME: This auxiliary timeout logic appeared recently and then
// quickly disappeared. While I haven't seen it invoked, I'm leaving
// it active for now.
if(FETCHING_TIMEOUT < mRequestedTimer.getElapsedTimeF32())
{
//timeout, abort.
LL_WARNS("Texture") << "Fetch of texture " << mID << " timed out after "
<< mRequestedTimer.getElapsedTimeF32()
<< " seconds. Canceling request." << LL_ENDL;
if (LLCORE_HTTP_HANDLE_INVALID != mHttpHandle)
{
// Issue cancel on any outstanding request. Asynchronous
// so cancel may not actually take effect if operation is
// complete & queued. Either way, notification will
// complete and the request can be transitioned.
mFetcher->mHttpRequest->requestCancel(mHttpHandle, NULL);
}
else
{
// Shouldn't happen but if it does, cancel quickly.
mState = DONE;
releaseHttpSemaphore();
return true;
}
}
// *HISTORY: There was a texture timeout test here originally that
// would cancel a request that was over 120 seconds old. That's
// probably not a good idea. Particularly rich regions can take
// an enormous amount of time to load textures. We'll revisit the
// various possible timeout components (total request time, connection
// time, I/O time, with and without retries, etc.) in the future.
setPriority(LLWorkerThread::PRIORITY_LOW | mWorkPriority);
return false;
@ -2571,19 +2549,36 @@ void LLTextureFetch::removeFromHTTPQueue(const LLUUID& id, S32 received_size)
mHTTPTextureBits += received_size * 8; // Approximate - does not include header bits
} // -Mfnq
// NB: If you change deleteRequest() you should probably make
// parallel changes in removeRequest(). They're functionally
// identical with only argument variations.
//
// Threads: T*
void LLTextureFetch::deleteRequest(const LLUUID& id, bool cancel)
{
lockQueue(); // +Mfq
LLTextureFetchWorker* worker = getWorkerAfterLock(id);
unlockQueue(); // -Mfq
if (worker)
{
size_t erased_1 = mRequestMap.erase(worker->mID);
unlockQueue(); // -Mfq
// *TODO: Refactoring this code may have introduced a thread race
// here where other code can run between the lookup above and the
// removeRequest() below.
removeRequest(worker, cancel);
llassert_always(erased_1 > 0) ;
removeFromNetworkQueue(worker, cancel);
llassert_always(!(worker->getFlags(LLWorkerClass::WCF_DELETE_REQUESTED))) ;
worker->scheduleDelete();
}
else
{
unlockQueue(); // -Mfq
}
}
// NB: If you change removeRequest() you should probably make
// parallel changes in deleteRequest(). They're functionally
// identical with only argument variations.
//
// Threads: T*
void LLTextureFetch::removeRequest(LLTextureFetchWorker* worker, bool cancel)
{