[webkit-reviews] review granted: [Bug 117842] WebProcess hangs loading eff.org (Waiting forever on a sync XHR, NetworkProcess unable to service it) : [Attachment 205108] Patch v1

[bugzilla-daemon at webkit.org]

Alexey Proskuryakov <ap at webkit.org> has granted Brady Eidson
<beidson at apple.com>'s request for review:
Bug 117842: WebProcess hangs loading eff.org (Waiting forever on a sync XHR,
NetworkProcess unable to service it)
https://bugs.webkit.org/show_bug.cgi?id=117842

Attachment 205108: Patch v1 
https://bugs.webkit.org/attachment.cgi?id=205108&action=review

------- Additional Comments from Alexey Proskuryakov <ap at webkit.org>
View in context: https://bugs.webkit.org/attachment.cgi?id=205108&action=review


> Source/WebKit2/NetworkProcess/HostRecord.cpp:198
> +    ASSERT(loader && loader->connectionToWebProcess());

Two lines.

> Source/WebKit2/NetworkProcess/HostRecord.cpp:204
> +    if (loader->connectionToWebProcess()->isSerialLoadingEnabled() &&
m_loadersInProgress.size() >= 1)
> +	   return true;

What is serial loading? Sounds like we have the exactly the same issue with it
if there is one outstanding request, and WebProcess starts a sync one.

> Source/WebKit2/NetworkProcess/HostRecord.cpp:213
> +    // If we're already past the limit of the number of loaders in flight,
we won't even serve new synchronous requests right now.
> +    if (m_loadersInProgress.size() > m_maxRequestsInFlight)
> +	   return true;

Perhaps assert that at least one active loader is synchronous?

> Source/WebKit2/NetworkProcess/NetworkResourceLoadScheduler.h:29
> +#include "HostRecord.h"

Why?