Defend against selectCache() being called multiple times....

Defend against selectCache() being called multiple times. There is a bug filed upstream to fix the HTML parser such that
it won't generate multiple calls to this, but given the complexity of the parser I'm putting this defense directly in the
appcache system.

BUG=72986, 73118
TEST=manually run Poppit


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=75481

[ericu, 2011-02-18 01:27:34]

http://codereview.chromium.org/6546004/diff/1/webkit/appcache/appcache_host.cc
File webkit/appcache/appcache_host.cc (right):

http://codereview.chromium.org/6546004/diff/1/webkit/appcache/appcache_host.c...
webkit/appcache/appcache_host.cc:70: 
Given that this condition is now prevented in the renderer, I suggest that we
either CHECK here, or just not handle it here at all, given that it should now
be impossible.  Or you could put your addition behind NDEBUG flags and DCHECK,
if you feel that's better.  But let's only solve it in one place.

[michaeln, 2011-02-18 01:41:46]

New snapshot.

http://codereview.chromium.org/6546004/diff/1/webkit/appcache/appcache_host.cc
File webkit/appcache/appcache_host.cc (right):

http://codereview.chromium.org/6546004/diff/1/webkit/appcache/appcache_host.c...
webkit/appcache/appcache_host.cc:70: 
On 2011/02/18 01:27:34, ericu wrote:
> Given that this condition is now prevented in the renderer, I suggest that we
> either CHECK here, or just not handle it here at all, given that it should now
> be impossible.  Or you could put your addition behind NDEBUG flags and DCHECK,
> if you feel that's better.  But let's only solve it in one place.

Right, I'm being overly paranoid after seeing a CRASHING bug report :) I should
be able to depend on the logic in the renderer to defend against it for real.

[ericu, 2011-02-19 00:20:31]

LGTM