On Fri, Apr 08, 2005 at 07:57:31PM -0400, Daniel Berlin wrote: > Some of these global properties probably belong in cgraph_var node's > instead of shoving them into the tree structure.
I tend to agree. This would allow us to do things like find out that the variable is read-write while compiling the function that handles all the static initialization. Later, while looking at functions that are not connected with initialization, they'll appear to be read-only. The important part is that cgraph will have to know about which functions are called from where, and whether or not a function can possibly be invoked at a time in which a variable hasn't completed its initialization. And, possibly, whether we ought to generate multiple copies of a function -- one with the variable read-only and one read-write. r~
