LCSSA is still the slowest pass when gccas'ing 252.eon, but now it only takes
39s instead of 289s. :)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28776 91177308-0d34-0410-b5e6-96231b3b80d8
not handling PHI nodes correctly when determining if a value was live-out.
This patch reduces the number of detected live-out variables in the testcase
from 6565 to 485.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28771 91177308-0d34-0410-b5e6-96231b3b80d8
ComputeMaskedBits. DemandedMasks and KnownZero/One masks should never have
bits set out of the range of the base datatype.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28768 91177308-0d34-0410-b5e6-96231b3b80d8
If a single exit block has multiple predecessors within the loop, it will
appear in the exit blocks list more than once. LCSSA needs to take that into
account so that it doesn't double process that exit block.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28750 91177308-0d34-0410-b5e6-96231b3b80d8
as using incoming argument registers, so the local allocator would clobber them
between their set and use. To fix this, we give the call instructions a variable
number of uses in the CALL MachineInstr itself, so live variables understands
the live ranges of these register arguments.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28744 91177308-0d34-0410-b5e6-96231b3b80d8
post-increment value, should be first cast to the appropriated type (to the
type of the common expr). Otherwise, the rewrite of a use based on (common +
iv) may end up with an incorrect type.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28735 91177308-0d34-0410-b5e6-96231b3b80d8
Change the file size field of StatusInfo to be uint64_t instead of size_t
so that we know it is always 64 bits. This prevents some overflow on
systems where size_t is 32 bits when it ought to be 64.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28726 91177308-0d34-0410-b5e6-96231b3b80d8
Provide new llvm::sys::Program facilities for converting the stdout and
stdin to binary mode. There is no standard way to do this and the available
mechanisms are platform specific. Adjust the bytecode reader and writer to
use these methods when their input is stdin or output is stdout. THis avoids
the problem with \n writing CRLF to a bytecode file on windows.
Patch Contributed by Michael Smith.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28722 91177308-0d34-0410-b5e6-96231b3b80d8
1. Add #includes to LinkAllVMCore.h to get Mangler.o and InlineAsm.o
2. Make Mangler.h and InlineAsm.h use the macros to ensure linkage
3. Make each of the tools with --load options include LinkAllVMCore.h
This should be the last set of changes for this bug and 800.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28719 91177308-0d34-0410-b5e6-96231b3b80d8
the compaction table for a function. This broke compilation of hexxagon
on darwin/x86 with recent changes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28717 91177308-0d34-0410-b5e6-96231b3b80d8
1. Fix the macros in IncludeFile.h to put everything in the llvm namespace
2. Replace the previous explicit mechanism in all the .h and .cpp files
with the macros in IncludeFile.h
This gets us a consistent mechanism throughout LLVM for ensuring linkage.
Next step is to make sure its used in enough places.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28715 91177308-0d34-0410-b5e6-96231b3b80d8
Break the "IncludeFile" mechanism into its own header file and adjust other
files accordingly. Use this facility for the IntrinsicInst problem which
was the subject of PR800.
More to follow on this.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28709 91177308-0d34-0410-b5e6-96231b3b80d8
function that is being lowered is forced to use FP. Currently this is only
true for main() / Cygwin.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28703 91177308-0d34-0410-b5e6-96231b3b80d8
when available. This speeds up hexxagon from 18.61s to 16.61s with the CBE on
PPC Mac OS (for reference, LLC is 15.48s and GCC is 23.35s).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28697 91177308-0d34-0410-b5e6-96231b3b80d8
through to gcc when its being used as a linker. This allows -L and -l
(and any other) options to be added so that non-complete bytecode files
can be processed with bugpoint. The -Xlinker option can be added as many
times as needed.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28692 91177308-0d34-0410-b5e6-96231b3b80d8
expand "ret null" (implements test/Regression/CodeGen/ARM/ret_void.ll)
note that a Flag link is missing between the copy and the branch
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28691 91177308-0d34-0410-b5e6-96231b3b80d8
built with CommandLine.h will want the --version option to report that the
tool belongs to LLVM. To override simply pass a void func() to the
cl::SetVersionPrinter() function and that void func() will be called when
it is time to print the version information.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28687 91177308-0d34-0410-b5e6-96231b3b80d8
to link in the implementation. Thanks to Anton Korobeynikov for figuring out
what was going on here.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28660 91177308-0d34-0410-b5e6-96231b3b80d8
code (while cloning) it often gets the branch/switch instructions. Since it
knows that edges of the CFG are dead, it need not clone (or even look) at
the obviously dead blocks. This should speed up the inliner substantially on
code where there are lots of inlinable calls to functions with constant
arguments. On C++ code in particular, this kicks in.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28641 91177308-0d34-0410-b5e6-96231b3b80d8
IncludeFile hack to ensure linkage of analysis passes. This works around
some -pedantic warnings about assigning an object to a function.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28621 91177308-0d34-0410-b5e6-96231b3b80d8
reimplement getValueDominatingFunction to walk the DominanceTree rather than
just searching blindly.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28618 91177308-0d34-0410-b5e6-96231b3b80d8
Turn -pedantic and -Wno-long-long compile flags on by default. In a few
places, avoid the warnings by removing these options in the local makefile.
One notable exception: lib/Target/CBackend/Writer.cpp. These warnings are
left on as a reminder to developers to clean them up.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28614 91177308-0d34-0410-b5e6-96231b3b80d8
BCWriter, and bzip2 libraries. Adjust the various makefiles to accommodate
these changes. This was done to speed up link times.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28610 91177308-0d34-0410-b5e6-96231b3b80d8
/bzlib.c:126: warning: string length `1056' is greater than the length `509' ISO
C89 compilers are required to support
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28602 91177308-0d34-0410-b5e6-96231b3b80d8
string. Instead of specifying the length, just specify whether the user
wants a terminating null or not. The default is "true" to retain the same
behavior as previously provided by this function.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28562 91177308-0d34-0410-b5e6-96231b3b80d8
Tcl to claim that the program had an error and thus produces errors in
the dejagnu testing when its really just a warning.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28556 91177308-0d34-0410-b5e6-96231b3b80d8
implementation always added a null byte to the end of the string. It turns
out that this is not always wanted. By adding a length parameter we preserve
this behavior when length==0 (default value) but also allow other lengths
(not null terminated) to be created.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28552 91177308-0d34-0410-b5e6-96231b3b80d8
Fold c2 in (x << c1) | c2 where (c2 < c1)
e.g.
int test(int x) {
return (x << 3) + 7;
}
This can be codegen'd as:
leal 7(,%eax,8), %eax
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28550 91177308-0d34-0410-b5e6-96231b3b80d8
uses DSA to make find targets of calls. It provides a very convinient
interface to DSA results to do things with indirect calls, such as
write a devirtualizer (which I have and may commit one of these days).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28545 91177308-0d34-0410-b5e6-96231b3b80d8
is now theoretically feature-complete. It has not, however, been thoroughly
test, and is still considered experimental.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28529 91177308-0d34-0410-b5e6-96231b3b80d8
the iterated Dominance Frontier of the loop-closure Phi's. This is the
second phase of the LCSSA pass. The third phase (coming soon) will be to
update all uses of loop variables to use the loop-closure Phi's instead.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28524 91177308-0d34-0410-b5e6-96231b3b80d8
makes it so that it constant folds instructions on the fly. This is good
for several reasons:
0. Many instructions are constant foldable after inlining, particularly if
inlining a call with constant arguments.
1. Without this, the inliner has to allocate memory for all of the instructions
that can be constant folded, then a subsequent pass has to delete them. This
gets the job done without this extra work.
2. This makes the inliner *pass* a bit more aggressive: in particular, it
partially solves a phase order issue where the inliner would inline lots
of code that folds away to nothing, but think that the resultant function
is big because of this code that will be gone. Now the code never exists.
This is the first part of a 2-step process. The second part will be smart
enough to see when this implicit constant folding propagates a constant into
a branch or switch instruction, making CFG edges dead.
This implements Transforms/Inline/inline_constprop.ll
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28521 91177308-0d34-0410-b5e6-96231b3b80d8
SimplifySelectOps would eliminate a Select, delete it, then return true.
The clients would see that it did something and return null.
The top level would see a null return, and decide that nothing happened,
proceeding to process the node in other ways: boom.
The fix is simple: clients of SimplifySelectOps should return the select
node itself.
In order to catch really obnoxious boogs like this in the future, add an
assert that nodes are not deleted. We do this by checking for a sentry node
type that the SDNode dtor sets when a node is destroyed.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28514 91177308-0d34-0410-b5e6-96231b3b80d8
instead of MVT::i1. Either is fine except MVT::i32 is probably a legal type
for most (if not all) platforms while MVT::i1 is not.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28511 91177308-0d34-0410-b5e6-96231b3b80d8
nonccc calls (we were dropping the CC and tail flag). This broke several
FORTRAN programs.
Testcase here: Regression/Assembler/2006-05-26-VarargsCallEncode.ll
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28501 91177308-0d34-0410-b5e6-96231b3b80d8