Commit Graph

12757 Commits

Author SHA1 Message Date
Misha Brukman
2b3c4ca665 Make a single `NoFPElim' switch available to all targets.
llvm-svn: 14296
2004-06-21 21:07:51 +00:00
Misha Brukman
a79674d478 * Make indentation consistent at 2 chars
* Doxygenify function comments
* Wrap code at 80 cols

llvm-svn: 14295
2004-06-21 20:22:03 +00:00
Misha Brukman
4424ef9dd1 This file is no longer applicable.
llvm-svn: 14294
2004-06-21 20:17:41 +00:00
Misha Brukman
03cf1fc233 Let's be consistent: listing format `os/arch'.
llvm-svn: 14293
2004-06-21 18:43:23 +00:00
Misha Brukman
3d93b9ccca llvm/IntrinsicLowering.h => llvm/CodeGen/IntrinsicLowering.h
llvm-svn: 14292
2004-06-21 18:30:31 +00:00
Misha Brukman
095db9590a Handle shifts >= 32 bits.
llvm-svn: 14291
2004-06-21 18:01:47 +00:00
Misha Brukman
36089f7034 Intrinsic::isnan has gone away, support for it commented out.
Intrinsic::isunordered has arrived, and we just use the standard lowering
pass for it.

llvm-svn: 14290
2004-06-21 17:58:36 +00:00
Misha Brukman
1c26b2e29d Convert tabs to spaces.
llvm-svn: 14289
2004-06-21 17:41:12 +00:00
Misha Brukman
30e1885192 Type::getPrimitiveID() -> getTypeID()
llvm-svn: 14288
2004-06-21 17:25:55 +00:00
Misha Brukman
76b4796e7c Type::getPrimitiveID() -> getTypeID()
llvm-svn: 14287
2004-06-21 17:19:08 +00:00
Misha Brukman
269034c151 Initial revision
llvm-svn: 14283
2004-06-21 16:55:25 +00:00
Misha Brukman
7b1e53ba01 Mention the requirement of dlcompat for MacOS X.
llvm-svn: 14282
2004-06-21 14:00:44 +00:00
Chris Lattner
6be1c456a8 Rename Interval class to LiveInterval to avoid conflicting with the already
existing llvm::Interval class.

Patch contributed by Vladimir Prus!
http://mail.cs.uiuc.edu/pipermail/llvmbugs/2004-June/000710.html

llvm-svn: 14281
2004-06-21 13:10:56 +00:00
Chris Lattner
3a8e3bce6c Make ConstantBool act like a 1 bit ConstantInt, in order to simplify client
code.  Patch contributed by Vladimir Prus.

llvm-svn: 14280
2004-06-21 12:12:12 +00:00
Reid Spencer
116576ba01 Nullify this test as it causes infinite loop in optimized nightly test.
llvm-svn: 14279
2004-06-21 08:11:32 +00:00
Chris Lattner
97ab2cec69 *FINALLY* Fix a really nasty nondeterministic bug that has been haunting us
since May 1st.  In this code, the pred iterator was being invalidated sometimes
causing the wrong entries to be added to PHI nodes.

The fix for this is to defererence and safe the *PI value before we hack on
branch instructions, which changes use/def chains, which SOMETIMES invalidates
the iterator.

llvm-svn: 14278
2004-06-21 07:19:01 +00:00
Chris Lattner
892df61ed3 Comment out the isnan stuff until we get a proper autoconf test for it
breaking the build on sparc is not acceptable.

llvm-svn: 14277
2004-06-21 06:17:21 +00:00
Chris Lattner
8723083f3b Make order of argument addition deterministic. In particular, the layout
of ConstantInt objects in memory used to determine which order arguments
were added in in some cases.

llvm-svn: 14276
2004-06-21 00:07:58 +00:00
Chris Lattner
fea46da8de REALLY fix PR378: crash in scalar evolution analysis
llvm-svn: 14275
2004-06-20 20:32:16 +00:00
Reid Spencer
8187cd46dc Added a note about requiring QMTest 2.0.3 instead of any other version.
llvm-svn: 14274
2004-06-20 19:21:11 +00:00
Chris Lattner
7ae7999106 Bug fixed
llvm-svn: 14273
2004-06-20 19:08:50 +00:00
Chris Lattner
59d13ac701 Fix a bug in my change last night that caused a few test failures.
llvm-svn: 14270
2004-06-20 17:01:44 +00:00
Chris Lattner
fb800c656c IntrinsicLowering.cpp now lives in lib/CodeGen/
llvm-svn: 14269
2004-06-20 07:59:27 +00:00
Chris Lattner
a7f1c89ab2 Update path
llvm-svn: 14268
2004-06-20 07:53:22 +00:00
Chris Lattner
6f6b95abe2 Header moved into the CodeGen directory
llvm-svn: 14267
2004-06-20 07:50:32 +00:00
Chris Lattner
cc465361d9 Move the IntrinsicLowering header into the CodeGen directory, as per PR346
llvm-svn: 14266
2004-06-20 07:49:54 +00:00
Chris Lattner
8a9f1ebd22 Move the IntrinsicLowering header into the CodeGen directory
llvm-svn: 14265
2004-06-20 07:46:18 +00:00
Chris Lattner
9da4181217 Start moving IntrinsicLowering out of VMCore into libcodegen, as per PR346
llvm-svn: 14264
2004-06-20 07:40:46 +00:00
Chris Lattner
8c90177e5f Do not sort SCEV objects by address: instead sort by complexity and group
by address.  This prevents the resultant SCEV objects from depending on
where in memory other scev objects happen to live.

llvm-svn: 14263
2004-06-20 06:23:15 +00:00
Chris Lattner
27578fd2f5 Make use of BinaryOperator::create* methods to shrinkify code.
llvm-svn: 14262
2004-06-20 05:04:01 +00:00
Chris Lattner
ba933051e3 Add methods like BinaryOperator::createAdd that take an instruction to insert
before.

llvm-svn: 14261
2004-06-20 05:02:56 +00:00
Chris Lattner
6324d6d00b Fix the inliner to be deterministic, not letting its output depend on the
relative location of Function objects in memory.

llvm-svn: 14260
2004-06-20 04:11:48 +00:00
Chris Lattner
084fef46e4 Make fpcmp handle running off of the beginning or end of the file correctly.
llvm-svn: 14259
2004-06-20 03:12:18 +00:00
Chris Lattner
58b3f16797 Add some DEBUG output to the simplifycfg routines
Fix another non-deterministic behavior, this one should actually speed up the
code though as it was doing silly things.

llvm-svn: 14258
2004-06-20 01:13:18 +00:00
Reid Spencer
06aff16337 A utility to search the LLVM source tree for a grep pattern. This is a
replacement for getsrcs.sh which now generates too much text to put on a
Linux command line. The approach taken with llvmgrep is to execute a find
command and execute a grep on each file that matches the name pattern. The
arguments to this script are the same as those of egrep. Note that the -H
and -n options to egrep will always be passed so that you always get the
file and line number of matches.

llvm-svn: 14255
2004-06-19 20:32:55 +00:00
Chris Lattner
39a31b56ac Now that dominator tree children are built in determinstic order, this horrible code
can go away

llvm-svn: 14254
2004-06-19 20:23:35 +00:00
Chris Lattner
a1d29e1b15 compute dominator tree children in a deterministic order that does not depend
on the address of BasicBlock objects in memory.  This eliminates stuff like this:

 Inorder Dominator Tree:
   [1]  %entry
     [2]  %loopentry
-      [3]  %loopexit
       [3]  %no_exit
-        [4]  %endif
         [4]  %then
+        [4]  %endif
+      [3]  %loopexit
       [3]  %return

llvm-svn: 14253
2004-06-19 20:13:48 +00:00
Chris Lattner
ba131c993e Print out immediate dominators in program order, not in random order based on the address
of BasicBlock objects

llvm-svn: 14252
2004-06-19 20:04:47 +00:00
Chris Lattner
ef8a52f263 This will hopefully fix a heisenbug that Vladimir Merzliakov is running
into valiantly trying to compile stuff on freebsd.

llvm-svn: 14251
2004-06-19 19:01:26 +00:00
Reid Spencer
c864bb89bb Add a note about GCC 3.3.2 optimization bug that causes llc to spin.
llvm-svn: 14250
2004-06-19 18:24:05 +00:00
Chris Lattner
3d68124555 Fix a nasty bug, noticed by Reid
llvm-svn: 14249
2004-06-19 18:15:50 +00:00
Chris Lattner
185f2b060b Fix one source of nondeterminism in the -licm pass: the hoist pass
was processing blocks in whatever order they happened to end up in the
dominator tree data structure.  Force an ordering.

llvm-svn: 14248
2004-06-19 08:56:43 +00:00
Chris Lattner
d2720878ec Change to use the StableBasicBlockNumbering class
llvm-svn: 14247
2004-06-19 08:42:40 +00:00
Chris Lattner
49af533d57 Initial checkin of the StableBasicBlockNumbering, a little helper class for computing
(strangely enough) a stable (determinstic) numbering for basic blocks.

llvm-svn: 14246
2004-06-19 08:41:59 +00:00
Chris Lattner
9b19f34fe2 Fix a tiny bug in the -no-aa pass, in which it did not ever get a target data.
This is a regression from 1.2, though noone uses -no-aa anyway

llvm-svn: 14245
2004-06-19 08:05:58 +00:00
Chris Lattner
3aed3a8bf2 Do not let the numbering of PHI nodes placed in the function depend on
non-deterministic things like the ordering of blocks in the dominance
frontier of a BB.  Unfortunately, I don't know of a better way to solve
this problem than to explicitly sort the BB's in function-order before
processing them.  This is guaranteed to slow the pass down a bit, but
is absolutely necessary to get usable diffs between two different tools
executing the mem2reg or scalarrepl pass.

Before this, bazillions of spurious diff failures occurred all over the
place due to the different order of processing PHIs:

-       %tmp.111 = getelementptr %struct.Connector_struct* %upcon.0.0, uint 0, uint 0
+       %tmp.111 = getelementptr %struct.Connector_struct* %upcon.0.1, uint 0, uint 0

Now, the diffs match.

llvm-svn: 14244
2004-06-19 07:40:14 +00:00
Chris Lattner
6e7b76ce94 Do not sort by the address of LLVM ConstantInt* objects. This produces
nondeterministic results that depend on where these objects land in memory.
Instead, sort by the value of the constant, which is stable.

Before this patch, the -simplifycfg pass run from two different compilers
could cause different code to be generated, though it was semantically the
same:

@@ -12258,8 +12258,8 @@
        %s_addr.1 = phi sbyte* [ %s, %entry ], [ %inc.0, %no_exit ]             ; <sbyte*> [#uses=5]
        %tmp.1 = load sbyte* %s_addr.1          ; <sbyte> [#uses=1]
        switch sbyte %tmp.1, label %no_exit [
-                sbyte 0, label %loopexit
                 sbyte 46, label %loopexit
+                sbyte 0, label %loopexit
        ]

We need to stomp all of this stuff out.

llvm-svn: 14243
2004-06-19 07:02:14 +00:00
Chris Lattner
5b4da6dd16 Do not loop over uses as we delete them. This causes iterators to be
invalidated out from under us.  This bug goes back to revision 1.1: scary.

llvm-svn: 14242
2004-06-19 02:02:22 +00:00
Chris Lattner
e42a9cfbb5 Bug fixed
llvm-svn: 14241
2004-06-19 01:11:02 +00:00
Chris Lattner
f006210fe9 Test for PR374
llvm-svn: 14240
2004-06-19 01:08:12 +00:00