It is no longer legal to initialize arrays and structs with single

scalar values (that ispc used to smear across the array/struct
elements).  Now, initializers in variable declarations must be
{ }-delimited lists, with one element per struct member or array
element, respectively.

There were a few problems with the previous implementation of the
functionality to initialize from scalars.  First, the expression
would be evaluated once per value initialized, so if it had side-effects,
the wrong thing would happen.  Next, for large multidimensional arrays,
the generated code would be a long series of move instructions, rather
than loops (and this in turn made LLVM take a long time.)

While both of these problems are fixable, it's a non-trivial
amount of re-plumbing for a questionable feature anyway.

Fixes issue #50.
This commit is contained in:
Matt Pharr
2011-07-01 13:45:58 +01:00
parent a2940d63b4
commit d2d5858be1
40 changed files with 114 additions and 57 deletions

View File

@@ -217,21 +217,10 @@ lInitSymbol(llvm::Value *lvalue, const char *symName, const Type *type,
exprList->exprs[i], ctx, pos);
}
}
else if (initExpr->GetType()->IsNumericType() ||
initExpr->GetType()->IsBoolType()) {
// Otherwise initialize all of the elements in turn with the
// initExpr.
for (int i = 0; i < collectionType->GetElementCount(); ++i) {
llvm::Value *ep = ctx->GetElementPtrInst(lvalue, 0, i, "element");
lInitSymbol(ep, symName, collectionType->GetElementType(i),
initExpr, ctx, pos);
}
}
else {
else
Error(initExpr->pos, "Can't assign type \"%s\" to \"%s\".",
initExpr->GetType()->GetString().c_str(),
collectionType->GetString().c_str());
}
return;
}