Bug 113603 – "Empty cache..." clears the disk cache from each WebProcess

• Status: RESOLVED
• Resolution: FIXED
• Component: WebKit2

[Brady Eidson]

"Empty cache..." clears the disk cache from each WebProcess

This is silly - Just once from the network process will suffice.

[Brady Eidson]

In radar <rdar://problem/12456652>

[Brady Eidson]

Created attachment 195790 [details]
Patch v1

[EFL EWS Bot]

Comment on attachment 195790 [details]
Patch v1

Attachment 195790 [details] did not pass efl-ews (efl):
Output: http://webkit-commit-queue.appspot.com/results/17227834

[Early Warning System Bot]

Comment on attachment 195790 [details]
Patch v1

Attachment 195790 [details] did not pass qt-wk2-ews (qt):
Output: http://webkit-commit-queue.appspot.com/results/17323451

[Sam Weinig]

Comment on attachment 195790 [details]
Patch v1

View in context: https://bugs.webkit.org/attachment.cgi?id=195790&action=review

> Source/WebKit2/NetworkProcess/NetworkProcess.h:131
> +#if PLATFORM(MAC)
> +    dispatch_group_t m_clearCacheDispatchGroup;
> +#endif

Please add a FIXME and file a bug about doing this in a way that doesn't require an #ifdef.  e.g. Using a WorkQueue and BinarySemaphore.

> Source/WebKit2/UIProcess/WebResourceCacheManagerProxy.cpp:121
>      // FIXME (Multi-WebProcess): <rdar://problem/12239765> There is no need to relaunch all processes. One process to take care of persistent cache is enough.
>      context()->sendToAllProcessesRelaunchingThemIfNecessary(Messages::WebResourceCacheManager::ClearCacheForAllOrigins(cachesToClear));

Do we still want to send this if we are using the network process?

[Brady Eidson]

(In reply to comment #5)
> (From update of attachment 195790 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=195790&action=review
> 
> > Source/WebKit2/NetworkProcess/NetworkProcess.h:131
> > +#if PLATFORM(MAC)
> > +    dispatch_group_t m_clearCacheDispatchGroup;
> > +#endif
> 
> Please add a FIXME and file a bug about doing this in a way that doesn't require an #ifdef.  e.g. Using a WorkQueue and BinarySemaphore.

Due to "popular request", I'm doing that now.

> 
> > Source/WebKit2/UIProcess/WebResourceCacheManagerProxy.cpp:121
> >      // FIXME (Multi-WebProcess): <rdar://problem/12239765> There is no need to relaunch all processes. One process to take care of persistent cache is enough.
> >      context()->sendToAllProcessesRelaunchingThemIfNecessary(Messages::WebResourceCacheManager::ClearCacheForAllOrigins(cachesToClear));
> 
> Do we still want to send this if we are using the network process?

Yes - WebProcesses still need to clear their individual memory caches.

[Brady Eidson]

(In reply to comment #6)
> (In reply to comment #5)
> > (From update of attachment 195790 [details] [details])

> > 
> > Please add a FIXME and file a bug about doing this in a way that doesn't require an #ifdef.  e.g. Using a WorkQueue and BinarySemaphore.
> 
> Due to "popular request", I'm doing that now.
> 

Or, maybe not.  A binary semaphore isn't good enough.  We'd need a counting semaphore.  That's why the WebResourceCacheManager version of this code uses dispatch_group;  It has to handle multiple requests to clear and wait for all of them to complete.

I'll definitely add the FIXME about this but WorkQueue and/or Semaphore improvements are definitely needed before they can be a solution to this problem.

[Brady Eidson]

http://trac.webkit.org/changeset/147251

[Csaba Osztrogonác]

(In reply to comment #8)
> http://trac.webkit.org/changeset/147251

And buildfix for !ENABLE(NETWORK_PROCESS) landed in
http://trac.webkit.org/changeset/147268.

Otherwise are you planning to review network process patches to let
enable network process for other ports? https://bugs.webkit.org/show_bug.cgi?id=108832

Positive and/or negative review would be welcome ...

[Brady Eidson]

(In reply to comment #9)
> (In reply to comment #8)
> > http://trac.webkit.org/changeset/147251
> 
> And buildfix for !ENABLE(NETWORK_PROCESS) landed in
> http://trac.webkit.org/changeset/147268.
> 
> Otherwise are you planning to review network process patches to let
> enable network process for other ports? https://bugs.webkit.org/show_bug.cgi?id=108832
> 
> Positive and/or negative review would be welcome ...

From https://bugs.webkit.org/show_bug.cgi?id=108832 - "Let's do this in smaller steps" seems like extremely good feedback!!!

[Csaba Osztrogonác]

> From https://bugs.webkit.org/show_bug.cgi?id=108832 - "Let's do this in smaller steps" seems like extremely good feedback!!!

Balázs - the author of the patch - said it, and then filed many
bug reports for it 1.5 months before. But after it, I can't see
any feedback from the WebKit2 authors.