search for: moduletofunctionpassadapter

...get to churn all the passes that have been ported so far, removing the proxies etc. Another thing that just came to me: the current way things work with adaptors means that if a function transformation invalidates a module analysis, the module analysis won't be invalidated until after the ModuleToFunctionPassAdapter returns. So we can end up running an arbitrary number of function transformations that query that stale module analysis. This is (I think?) mostly benign given our current set of module analyses (more info in the log; search for "We may be getting by simply because we seem to not have many mo...

...orted so far, > removing the proxies etc. > > > > > > Another thing that just came to me: the current way things work with > adaptors means that if a function transformation invalidates a module > analysis, the module analysis won't be invalidated until after the > ModuleToFunctionPassAdapter returns. > > So we can end up running an arbitrary number of function transformations > that query that stale module analysis. This is (I think?) mostly benign > given our current set of module analyses (more info in the log; search for > "We may be getting by simply because we...

...ies etc. >> >> >> >> >> >> Another thing that just came to me: the current way things work with >> adaptors means that if a function transformation invalidates a module >> analysis, the module analysis won't be invalidated until after the >> ModuleToFunctionPassAdapter returns. >> >> So we can end up running an arbitrary number of function transformations >> that query that stale module analysis. This is (I think?) mostly benign >> given our current set of module analyses (more info in the log; search for >> "We may be getting b...

The more closely I look at this, the more it seems like there may be a useful incremental step in the transition to the new PM: use the new PM analysis machinery in the old PM. If this is possible, it will simplify the old PM and (hopefully) allow an incremental transition to the new PM instead of a flag day transition for the switch. I.e., AFAICT, the new PM transition is essentially about 2