yv01p/Umbraco-CMS
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

NuCache: fix panic exception

Browse Source
This commit is contained in:
Stephan
2019-03-14 18:56:23 +01:00
parent 7e413ed406
commit 851c844c8b
3 changed files with 87 additions and 13 deletions
Download Patch File Download Diff File Expand all files Collapse all files

15
src/Umbraco.Web/PublishedCache/NuCache/SnapDictionary.cs
View File

@@ -139,18 +139,12 @@ namespace Umbraco.Web.PublishedCache.NuCache
_wlocked++;
lockInfo.Count = true;
// fixme - this comment: "ok to have holes in generation objects" is annoying
// 'cos if you have a hole, then doing _liveGen-1 in some places would be wrong
// besides, forceGen is used only when getting a scoped write lock, which is not reentrant?
//
// but... if _wlocked ==1 and we just incremented it, it means it was 0, so it wasnt locked, so wtf?
// this is the only place where _nextGen would turn true so ... this makes no sense at all!
if (_nextGen == false || (forceGen && _wlocked == 1)) // if true already... ok to have "holes" in generation objects
if (_nextGen == false || (forceGen && _wlocked == 1))
{
// because we are changing things, a new generation
// is created, which will trigger a new snapshot
_nextGen = true; // this is the ONLY place where _nextGen becomes true
_genObjs.Enqueue(_genObj = new GenObj(_liveGen));
_liveGen += 1;
}
}
@@ -379,9 +373,6 @@ namespace Umbraco.Web.PublishedCache.NuCache
if (_genObj == null)
_genObjs.Enqueue(_genObj = new GenObj(snapGen));
// fixme - getting a panic exception here
// means _genObj != null, means _nextGen == true
// if we have one already, ensure it's consistent
else if (_genObj.Gen != snapGen)
throw new Exception("panic");
@@ -551,6 +542,8 @@ namespace Umbraco.Web.PublishedCache.NuCache
set => _dict._collectAuto = value;
}
public GenObj GenObj => _dict._genObj;
public ConcurrentQueue<GenObj> GenObjs => _dict._genObjs;
public Snapshot LiveSnapshot => new Snapshot(_dict, _dict._liveGen);