2015-02-26 18:48 GMT+03:00 Adam Wolk <[email protected]>:
> WebCore::SocketStreamHandlePrivate::send

Looking at the body of this (rather small) function, the only way
itself could cause SIGSEGV is when m_socket points to garbage (not
NULL). This could happen only if SocketStreamHandlePrivate object is
created using second constructor that inherits QTcpSocket* from
outside. Thus I'm thinking about use-after-free, rather than NULL
deref. I'm not a GDB expert, though. :(

This idea could be somewhat proved by AFJR options set in
MALLOC_OPTIONS environment variable prior starting otter-browser.
Those options could trigger other problems as well, though.

I'll take a closer look at the weekend. Thank you for the report.

--
  WBR,
  Vadim Zhukov

Reply via email to