a nice shiny B+ Tree variant. This fixes the last of the known algorithmic
issues with the rewriter, allowing a significant speedup. For example,
-emit-html on Ted's 500K .i file speeds up from 26.8s -> 0.64s in a
debug build (41x!) and 5.475s -> 0.132s (41x!) in an optimized build.
This code is functional but needs to be cleaned up, ifdefs removed, better
commented, and moved to a .cpp file. I plan to do this tomorrow.
llvm-svn: 49635
much simpler than in LegalizeDAG because calls are
not yet expanded into call sequences: that happens
after type legalization has finished.
llvm-svn: 49634
inheritance in C++. It'll parse the base-specifier list, e.g.,
class D : public B1, virtual public B2 { };
and do some of the simpler semantic checks (B1 and B2 are classes;
they aren't unions or incomplete types, etc).
llvm-svn: 49623
allows the parsing of "class" in addition to "struct" and "union" to
declare a record. So this patch allows:
class C { };
class C c1;
But it does not contain the lookup bits, so this won't work yet:
C c2;
Patch by Doug Gregor!
llvm-svn: 49613
in its maps. Add some sanity checks that catch
this kind of thing. Hopefully these can be
removed one day (once all problems are fixed!)
but for the moment it seems wise to have them in.
llvm-svn: 49612
This results in an (IMO) simpler algorithm, results in fewer splits, and
is more amenable to delta handling (there is no reason to mutate the tree
at all when adding a delta to a position that already exists in the tree).
llvm-svn: 49609
(but simple!) datastructures in the rewriter with a more complex but
more efficient one.
This replaces the Deltas vector with a specialized BTree that makes
delta lookups much more efficient. This speeds up -emit-html on a 500K
.i file from 157.154 to 27.127 seconds on my machine (5.8x).
While this code is functional, it isn't very pretty, I have much
refactoring planned for it, and will remove the USE_VECTOR ifdef.
Stay tuned.
llvm-svn: 49586
optimized x86-64 (and x86) calls so that they work (... at least for
my test cases).
Should fix the following problems:
Problem 1: When i introduced the optimized handling of arguments for
tail called functions (using a sequence of copyto/copyfrom virtual
registers instead of always lowering to top of the stack) i did not
handle byval arguments correctly e.g they did not work at all :).
Problem 2: On x86-64 after the arguments of the tail called function
are moved to their registers (which include ESI/RSI etc), tail call
optimization performs byval lowering which causes xSI,xDI, xCX
registers to be overwritten. This is handled in this patch by moving
the arguments to virtual registers first and after the byval lowering
the arguments are moved from those virtual registers back to
RSI/RDI/RCX.
llvm-svn: 49584
on any current target and aren't optimized in DAGCombiner. Instead
of using intermediate nodes, expand the operations, choosing between
simple loads/stores, target-specific code, and library calls,
immediately.
Previously, the code to emit optimized code for these operations
was only used at initial SelectionDAG construction time; now it is
used at all times. This fixes some cases where rep;movs was being
used for small copies where simple loads/stores would be better.
This also cleans up code that checks for alignments less than 4;
let the targets make that decision instead of doing it in
target-independent code. This allows x86 to use rep;movs in
low-alignment cases.
Also, this fixes a bug that resulted in the use of rep;stos for
memsets of 0 with non-constant memory size when the alignment was
at least 4. It's better to use the library in this case, which
can be significantly faster when the size is large.
This also preserves more SourceValue information when memory
intrinsics are lowered into simple loads/stores.
llvm-svn: 49572
Ted Kremenek