Do not call MemoryCache::prune() when a client/observer is removed

third_party/WebKit/Source/core/fetch/Resource.cpp

Index: third_party/WebKit/Source/core/fetch/Resource.cpp
diff --git a/third_party/WebKit/Source/core/fetch/Resource.cpp b/third_party/WebKit/Source/core/fetch/Resource.cpp
index 579a3667325526cc532e4db23babe049d21f3385..919993c5040d288460ec5d2ef7b53eeac0ee7fce 100644
--- a/third_party/WebKit/Source/core/fetch/Resource.cpp
+++ b/third_party/WebKit/Source/core/fetch/Resource.cpp
@@ -729,14 +729,9 @@ void Resource::didRemoveClientOrObserver()
         // "no-store: ... MUST make a best-effort attempt to remove the information from volatile storage as promptly as possible"
         // "... History buffers MAY store such responses as part of their normal operation."
         // We allow non-secure content to be reused in history, but we do not allow secure content to be reused.
-        if (hasCacheControlNoStoreHeader() && url().protocolIs("https")) {
+        if (hasCacheControlNoStoreHeader() && url().protocolIs("https"))
             memoryCache()->remove(this);

[yhirano, 2016/09/15 08:30:43]

Hm, is this statement still needed?

[hiroshige, 2016/09/15 08:45:33]

I'm not sure.
I created a CL https://codereview.chromium.org/2346483005 and seeing whether
this is covered by tests.

(Anyway it should be handled in a separate CL because it affects caching
correctness, and MemoryCache::remove() is not harmful when executed in
prefinalizers)

-            memoryCache()->prune();
-        } else {
-            memoryCache()->prune(this);

[Nate Chapin, 2016/09/16 21:28:47]

I think this is the only callsite that passes a non-null Resource* to prune().
Could we remove that unused logic, too?

[hiroshige, 2016/09/29 06:27:17]

Done.

-        }
     }
-    // This object may be dead here.
 }
 
 void Resource::allClientsAndObserversRemoved()