![intel gma 4500 md intel gma 4500 md](https://www.notebookcheck.net/uploads/tx_nbc2/9172655_ra.jpg)
I pulled down 167 dumps from the past four weeks.
![intel gma 4500 md intel gma 4500 md](https://www.notebookcheck.net/uploads/tx_nbc2/ivy_bridge_graphics.png)
![intel gma 4500 md intel gma 4500 md](https://www.notebookcheck.org/fileadmin/_processed_/4/3/csm_Intel-Pentium-Silver-and-Celeron-chip_84771d137b.jpg)
Hitting font fallback on facebook (due to the other-language links) at startup is likely very common.
#Intel gma 4500 md Patch#
Looking at the possible pushlog range suggested in comment #0, I'm concerned there might be a link with the system-font-fallback telemetry that was added in bug 705590 (though I don't actually see a problem in the patch there). Most of the crashes are happening at startup, and one has a comment "facebook". The dwrite-related addresses on the stack seem to imply involvement of font/text-rendering code, even though I don't actually believe the backtraces are accurate. if someone figures out the specific scenario where this happens, they might be able to control what gets corrupted, and hence arrive at some kind of exploit.
#Intel gma 4500 md code#
For the rest of them, I'm guessing that someone is writing beyond an allocated buffer, perhaps corrupting something else that follows, but crashing when they reach the end of the writable page.Īs a precaution, I'm marking this as security-sensitive, given that it looks like there's code writing to memory it shouldn't be touching. the crashes I looked at are all EXCEPTION_ACCESS_VIOLATION_WRITE, and virtually all of them occured at a page boundary (the only exception I saw was at 0x20, which sounds like it's related to an unexpectedly-NULL pointer). The stacks here look fairly bogus to me, unfortunately.