[Webkit-unassigned] [Bug 168699] [GTK] Hangs when showing Google search results

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=168699

--- Comment #9 from Carlos Garcia Campos <cgarcia at igalia.com> ---
(In reply to comment #8)
> Comment on attachment 302509 [details]
> Patch
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=302509&action=review

Thanks for the review Fuji!

> > Source/WebKit2/ChangeLog:13
> > +        socket monitor as readSocketMonitor and adds another one for polling output. When sendsmg fails with
> 
> s/sendsmg/sendmsg/

Oops

> > Source/WebKit2/Platform/IPC/unix/ConnectionUnix.cpp:365
> > +bool Connection::sendOutputMessage(UnixMessage& outputMessage)
> 
> Mac port uses unique_ptr. You don't use unique_ptr<UnixMessage> for
> outputMessage for performance?

Yes, the case of EWOULDBLOCK hardly ever happens, so only in that case I heap allocate the message to saved it.

> > Source/WebKit2/Platform/IPC/unix/ConnectionUnix.cpp:458
> > +                            sendOutputMessage(*message);
> 
> You don't need to check the return value of sendOutputMessage?

I'm assuming it won't fail because the poll said we can write without blocking. In any case, what we do is calling sendOutgoingMessages() that will return early if sendOutputMessage failed.

> > Source/WebKit2/Platform/IPC/unix/UnixMessage.h:66
> > +class AttachmentInfo {
> 
> You don't need to move AttachmentInfo, too.

Right.

> > Source/WebKit2/Platform/IPC/unix/UnixMessage.h:114
> > +            memcpy(m_body, other.m_body, m_messageInfo.bodySize());
> 
> You don't need to copy the body if m_messageInfo.isMessageBodyIsOutOfLine().

Good point!

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.webkit.org/pipermail/webkit-unassigned/attachments/20170224/5f84feb9/attachment.html>