I have some questions regarding how globals are represented in DS graph.
Specifically, I wrote the following simple program:
List *g;
void alloc_func(){
g = ( List* ) malloc( sizeof( List ) );
}
void free_func(){
free( g );
}
int main(){
alloc_func();
free_func();
}
I noticed that the DSnode for g in alloc_func is different from that of
free_func and NEITHER of them had GlobalNode bit set in their types. Only
the malloc bit and the incomplete bit have been set. I am completely
clueless how to figure out if it is a global node or not. I used
getGlobals and found that the returned map had size 0.
How can I get the "common" global node of g?
I've been trying to merge DSGraphs from different Functions, but i'm
not going anywhere. For a Function F and DSGraph G, I tried this
(assumming no SCC in the callgraph):
vector<DSCallSite> CSV = G->getAuxFunctionCalls();
for (vector<DSCallSite>::iterator I = CSV.begin(), E = CSV.end();
I != E; I++) {
Value *V = I->getCallInst().getOperand(0);
Function &SF = cast<Function>(*V);
if (!SF.isExternal()) {
DSGraph *SFG = new DSGraph(SF, GG);
G->mergeInGraph(*I, *SFG, 0);
}
}
but the resulting graph has no new merged DSNodes, even though in my
test case there is a DSNode pointed by the CallInst in F and another
DSNode in G that is returned to F.
Any ideas on what I'm doing wrong?
nicolas
> Specifically, I wrote the following simple program: > List *g; > void alloc_func(){ > g = ( List* ) malloc( sizeof( List ) ); > } > void free_func(){ > free( g ); > }I don't have an LLVM tree available to me right now (as tank/llvm.cs.uiuc.edu) is down, so take this with a grain of salt...> I noticed that the DSnode for g in alloc_func is different from that of > free_func and NEITHER of them had GlobalNode bit set in their types. Only > the malloc bit and the incomplete bit have been set. I am completely > clueless how to figure out if it is a global node or not. I used > getGlobals and found that the returned map had size 0.There may be a few different things going on here. First of all, the "globals" graph is only partially implemented currently. Right now, nodes eligable to be moved to the globals graph are properly removed from the individual function graphs, but they are never reinserted into the globals graph. This can cause some funny disappearances that don't make a lot of sense, and...> How can I get the "common" global node of g?... the common node in this case never gets unified. This is one of the top priorities for me when I get back to Illinois (after thanksgiving break), but it is unlikely to make much progress in the mean time. If you would like to disable this behavior, look in the DSGraph::removeDeadNodes (or similar) in DataStructure.cpp. In it you will see the code that marks nodes as being "alive" in the current graph. All you have to do is add code to mark all globals as alive, and they will be propogated up to main. Alternatively, take a look at the CVS revisions a few checkins back to see when the globals graph stuff was added... and revert it. I hope this helps some :) -Chris -- http://llvm.cs.uiuc.edu/ http://www.nondot.org/~sabre/Projects/
On Sun, 24 Nov 2002, Juan Nicolas Ruiz wrote:> I've been trying to merge DSGraphs from different Functions, but i'm > not going anywhere. For a Function F and DSGraph G, I tried this > (assumming no SCC in the callgraph):> vector<DSCallSite> CSV = G->getAuxFunctionCalls();Are you sure this is returning any call site entries? Depending on what _type_ of graph G is, it may be empty. If G is a local graph, it's empty. If G is a BU graph, CSV will only contain "unresolvable" call sites. You probably want just the getFunctionCalls() list, which is the set of function calls originally found by the Local's pass. If that doesn't help, you'll have to give me some more details about what is not happening: is the mergeInGraph function even being called? -Chris> for (vector<DSCallSite>::iterator I = CSV.begin(), E = CSV.end(); > I != E; I++) { > Value *V = I->getCallInst().getOperand(0); > Function &SF = cast<Function>(*V); > if (!SF.isExternal()) { > DSGraph *SFG = new DSGraph(SF, GG); > G->mergeInGraph(*I, *SFG, 0); > } > } > > but the resulting graph has no new merged DSNodes, even though in my > test case there is a DSNode pointed by the CallInst in F and another > DSNode in G that is returned to F. > > Any ideas on what I'm doing wrong? > > nicolas > > _______________________________________________ > LLVM Developers mailing list > LLVMdev at cs.uiuc.edu http://llvm.cs.uiuc.edu > http://mail.cs.uiuc.edu/mailman/listinfo/llvmdev >-Chris -- http://llvm.cs.uiuc.edu/ http://www.nondot.org/~sabre/Projects/
Ganesh, I modified DataStructure.cpp so that global nodes are no longer removed from any graphs. Only that file has changed. Chris, if you get a chance to do this, it would be nice if you could take a quick look at the change I made and make sure that's all that's needed. Thanks, --Vikram http://www.cs.uiuc.edu/~vadve> From: Chris Lattner <sabre at nondot.org> > Subject: Re: [LLVMdev] globals in DS graph > Sender: llvmdev-admin at cs.uiuc.edu > Date: Mon, 25 Nov 2002 01:34:14 -0600 (CST) > > > Specifically, I wrote the following simple program: > > List *g; > > void alloc_func(){ > > g = ( List* ) malloc( sizeof( List ) ); > > } > > void free_func(){ > > free( g ); > > } > > I don't have an LLVM tree available to me right now (as > tank/llvm.cs.uiuc.edu) is down, so take this with a grain of salt... > > > I noticed that the DSnode for g in alloc_func is different from that of > > free_func and NEITHER of them had GlobalNode bit set in their types. Only > > the malloc bit and the incomplete bit have been set. I am completely > > clueless how to figure out if it is a global node or not. I used > > getGlobals and found that the returned map had size 0. > > There may be a few different things going on here. First of all, the > "globals" graph is only partially implemented currently. Right now, nodes > eligable to be moved to the globals graph are properly removed from the > individual function graphs, but they are never reinserted into the globals > graph. This can cause some funny disappearances that don't make a lot of > sense, and... > > > How can I get the "common" global node of g? > > ... the common node in this case never gets unified. This is one of the > top priorities for me when I get back to Illinois (after thanksgiving > break), but it is unlikely to make much progress in the mean time. If you > would like to disable this behavior, look in the DSGraph::removeDeadNodes > (or similar) in DataStructure.cpp. In it you will see the code that marks > nodes as being "alive" in the current graph. All you have to do is add > code to mark all globals as alive, and they will be propogated up to main. > Alternatively, take a look at the CVS revisions a few checkins back to see > when the globals graph stuff was added... and revert it. > > I hope this helps some :) > > -Chris > > -- > http://llvm.cs.uiuc.edu/ > http://www.nondot.org/~sabre/Projects/ > > _______________________________________________ > LLVM Developers mailing list > LLVMdev at cs.uiuc.edu http://llvm.cs.uiuc.edu > http://mail.cs.uiuc.edu/mailman/listinfo/llvmdev >