Magnus Hagander wrote:
> If we are uncertain about delivering signals on an APC (with longjmp
> considerations etc), we could even just do our QueueUserAPC() on a
<snip>
This started to get me really concerned, so I did a fair bit of googling
on the subject. AFIK, we are probably ok longjmp-ing out of a callback
issued by QueueUserAPC. There are three mitigating factors wrt longjmp
& QueueUserAPC:
1. The callback runs on the same thread as the backend/signal routine
2. the signal is called/raised at the 'end' of the callback
3. there is very little stack involved.
Here is a link which has some technical info about windows messaging and
process control (courtesy of HP):
http://www.cag.lcs.mit.edu/dynamorio/doc/DynamoRIO.htm
Merlin
