Use WebKit::initializeWithoutV8() in SandboxIPCProcess.

Use WebKit::initializeWithoutV8() in SandboxIPCProcess.

This code only needs to access a few static fontconfig-related
methods, so use the minimal WebKit::initializeWithoutV8()
instead of the full initialize() (the same call is used by the
browser's in_process_webkit).

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=112490

[agl, 2011-11-30 19:35:26]

lgtm

[adamk, 2011-11-30 20:40:07]

Hmm, looks like we _do_ run into problems in browser_tests:


ASSERTION FAILED: m_key != PTHREAD_KEYS_MAX
third_party/WebKit/Source/JavaScriptCore/wtf/ThreadIdentifierDataPthreads.cpp(65)
: static WTF::ThreadIdentifier WTF::ThreadIdentifierData::identifier()

Partial stack trace:

[8881:8881:1130/115110:3221455658:ERROR:process_util_posix.cc(134)] Received
signal 11
	base::debug::StackTrace::StackTrace() [0x1571226]
	base::(anonymous namespace)::StackDumpSignalHandler() [0x15c1755]
	0x7fbeb8af7af0
	WTF::ThreadIdentifierData::identifier() [0x24fd12c]
	WTF::currentThread() [0x24d4eaa]
	WTF::ThreadRestrictionVerifier::setShared() [0x24213ec]
	WTF::RefCountedBase::ref() [0x2421524]
	WTF::refIfNotNull<>() [0x2425ccd]
	WTF::RefPtr<>::RefPtr() [0x2425c91]
	WebKit::WebCString::assign() [0x242584d]
	WebKit::WebCString::WebCString() [0x24c815a]
	WebKit::WebFontInfo::familyForChars() [0x24c7d2c]
	SandboxIPCProcess::HandleGetFontFamilyForChars() [0x387043b]
	SandboxIPCProcess::HandleRequestFromRenderer() [0x386fc3c]
	SandboxIPCProcess::Run() [0x386fa03]
	RenderSandboxHostLinux::Init() [0x386f2df]

So the use of RefPtr is the trigger here.

[levin, 2011-11-30 20:51:02]

On 2011/11/30 20:40:07, adamk wrote:
> Hmm, looks like we _do_ run into problems in browser_tests:
> 
> 
> ASSERTION FAILED: m_key != PTHREAD_KEYS_MAX
>
third_party/WebKit/Source/JavaScriptCore/wtf/ThreadIdentifierDataPthreads.cpp(65)
> : static WTF::ThreadIdentifier WTF::ThreadIdentifierData::identifier()
> 
> Partial stack trace:
> 
> [8881:8881:1130/115110:3221455658:ERROR:process_util_posix.cc(134)] Received
> signal 11
> 	base::debug::StackTrace::StackTrace() [0x1571226]
> 	base::(anonymous namespace)::StackDumpSignalHandler() [0x15c1755]
> 	0x7fbeb8af7af0
> 	WTF::ThreadIdentifierData::identifier() [0x24fd12c]
> 	WTF::currentThread() [0x24d4eaa]
> 	WTF::ThreadRestrictionVerifier::setShared() [0x24213ec]
> 	WTF::RefCountedBase::ref() [0x2421524]
> 	WTF::refIfNotNull<>() [0x2425ccd]
> 	WTF::RefPtr<>::RefPtr() [0x2425c91]
> 	WebKit::WebCString::assign() [0x242584d]
> 	WebKit::WebCString::WebCString() [0x24c815a]
> 	WebKit::WebFontInfo::familyForChars() [0x24c7d2c]
> 	SandboxIPCProcess::HandleGetFontFamilyForChars() [0x387043b]
> 	SandboxIPCProcess::HandleRequestFromRenderer() [0x386fc3c]
> 	SandboxIPCProcess::Run() [0x386fa03]
> 	RenderSandboxHostLinux::Init() [0x386f2df]
> 
> So the use of RefPtr is the trigger here.

Yep. As I indicated the other day, this is fine with me (LGTM) as long as it
passes all tests. I guess it fails in browser tests and not in DRT as I
indicated in my check in comment (sorry about that).

[adamk, 2011-11-30 21:22:14]

Okay, new approach: just use a more minimal WebKit::initializeWithoutV8(), which
still sets up WTF appropriately but leaves out any more-complex initialization
(initializeWithoutV8 is also called for in_process_webkit).

[agl, 2011-11-30 21:44:28]

Rubber-stamp LGTM if tests pass.

[adamk, 2011-11-30 23:07:19]

+darin for an OWNERS stamp.

Darin, this is needed before we can safely re-land http://webkit.org/b/71242.

[darin (slow to review), 2011-12-01 17:20:45]

LGTM

[commit-bot: I haz the power, 2011-12-01 17:24:13]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/adamk@chromium.org/8748015/7001

[commit-bot: I haz the power, 2011-12-01 18:27:19]

Change committed as 112490