Commit Graph

28970 Commits

Author SHA1 Message Date
Chris Lattner
7560c5f913 add options to view the dags before the first or second pass of dag combine.
llvm-svn: 33249
2007-01-16 04:55:25 +00:00
Bill Wendling
661f651728 Added a -mtriple flag to force darwin backend.
llvm-svn: 33248
2007-01-16 04:15:20 +00:00
Bill Wendling
08272a2fa6 Make inline ASM the INTEL one if it's in that emission mode.
llvm-svn: 33247
2007-01-16 04:13:03 +00:00
Bill Wendling
cfdd717db5 Fix for PR1095:
LLVM would miscompile ASM dialects when compiling for PPC. Added dialects for
the X86 and PPC backends. It defaults to "0", the first variant of a compound
inline asm expression.

llvm-svn: 33246
2007-01-16 03:42:04 +00:00
Devang Patel
2d421d58d6 Code refactoring.
llvm-svn: 33245
2007-01-16 02:00:38 +00:00
Devang Patel
b61a40bc5a Remove extra white spaces. Fix comments.
llvm-svn: 33244
2007-01-15 23:06:56 +00:00
Devang Patel
67c975490d Do not record last users of Pass Manager.
llvm-svn: 33243
2007-01-15 20:31:54 +00:00
Jeff Cohen
c4ab4ae3e1 Unbreak VC++ build.
llvm-svn: 33242
2007-01-15 20:27:18 +00:00
Chris Lattner
c9024c6200 test darwin too! :)
llvm-svn: 33241
2007-01-15 18:32:24 +00:00
Reid Spencer
4212cc8c4e One more @ to remove.
llvm-svn: 33240
2007-01-15 18:28:34 +00:00
Chris Lattner
b0a126d212 make this more efficient in release builds (time and space)
llvm-svn: 33239
2007-01-15 18:28:18 +00:00
Reid Spencer
a3435cea23 Update documentation for parameter attributes for the syntax change.
llvm-svn: 33238
2007-01-15 18:27:39 +00:00
Chris Lattner
946ceae164 Fix a regression in my isIntegral patch that broke 471.omnetpp. This is
because TargetData::getTypeSize() returns the same for i1 and i8.  This fix
is not right for the full generality of bitwise types, but it fixes the
regression.

llvm-svn: 33237
2007-01-15 17:55:20 +00:00
Nick Lewycky
36110b8456 Don't print address of ETNode. Print the DFSNumIn which uniquely identifies
the basic block and is stable across runs in gdb or valgrind.

Make Node::update handle edges which dominate and are tighter than
existing edges.

Replace makeEqual's "squeeze theorem" code. Fixes miscompilation.

Gate the calls to defToOps and opsToDef. Before this, we were getting IG
edges about values which weren't even defined in the dominated area. This
reduces the size of the IG by about half.

llvm-svn: 33236
2007-01-15 14:30:07 +00:00
Chris Lattner
a89a450f1c Run an instcombine pass after inlining but before scalarrepl. This allows
instcombine to clean up the code, which makes more code suitable for SRoA.
This helps C++ code in particular, e.g. speeding up tramp3d by 31%.

llvm-svn: 33235
2007-01-15 07:41:51 +00:00
Chris Lattner
1a96e01743 Implement InstCombine/phi.ll:test7, deletion of trivial value loops for
induction variables.

llvm-svn: 33234
2007-01-15 07:30:06 +00:00
Chris Lattner
1140c5c398 new testcase
llvm-svn: 33233
2007-01-15 07:29:29 +00:00
Chris Lattner
d4a0c800f4 simplify some code now that types are signless
llvm-svn: 33232
2007-01-15 07:02:54 +00:00
Chris Lattner
bc687c27d7 delete stores to allocas with one use. This is a trivial form of DSE which
often kicks in for ?: expressions.

llvm-svn: 33231
2007-01-15 06:51:56 +00:00
Chris Lattner
4b84530d59 add a simple case where instcombine can detect and remove a dead alloca
llvm-svn: 33230
2007-01-15 06:51:25 +00:00
Chris Lattner
64205203a5 Constant fold llvm.powi.*. This speeds up tramp3d--v4 by 9.5%
llvm-svn: 33229
2007-01-15 06:27:37 +00:00
Chris Lattner
851b9e93a2 add some notes
llvm-svn: 33228
2007-01-15 06:25:39 +00:00
Reid Spencer
a322dfce5f Regenerate.
llvm-svn: 33227
2007-01-15 02:41:46 +00:00
Reid Spencer
c32e0fedab For PR1113:
Increment the counter after the second use, not after the first use. This
fixes PR1113.

Also, rename some classes for simplicity and to more naturally be
reminscient of LLVM 1.9. This in preparation for additional classes that
will provide a scaled down model of the LLVM 1.9 IR.

llvm-svn: 33226
2007-01-15 02:40:33 +00:00
Chris Lattner
c5e1611848 rename Type::isIntegral to Type::isInteger, eliminating the old Type::isInteger.
rename Type::getIntegralTypeMask to Type::getIntegerTypeMask.

This makes naming much more consistent.  For example, there are now no longer any
instances of IntegerType that are not considered isInteger! :)

llvm-svn: 33225
2007-01-15 02:27:26 +00:00
Chris Lattner
ad283250fa regenerate
llvm-svn: 33224
2007-01-15 02:12:07 +00:00
Chris Lattner
f1797fe8bf teach VMCore to accept i1 add's and shifts
llvm-svn: 33223
2007-01-15 02:05:34 +00:00
Chris Lattner
f48900e322 eliminate calls to Type::isInteger, preferring isIntegral instead.
llvm-svn: 33222
2007-01-15 02:03:16 +00:00
Chris Lattner
0903f7f832 allow i1 to operators like shift and add.
llvm-svn: 33221
2007-01-15 02:00:29 +00:00
Chris Lattner
4d7eed35ae Update code to eliminate calls to isInteger, calling isIntegral instead.
llvm-svn: 33220
2007-01-15 01:58:56 +00:00
Chris Lattner
0a18f2a05f Change to match API changes.
llvm-svn: 33219
2007-01-15 01:55:32 +00:00
Chris Lattner
8fbcb083aa Eliminate calls to isInteger, generalizing code and tightening checks as needed.
llvm-svn: 33218
2007-01-15 01:55:30 +00:00
Chris Lattner
6026cbf165 Terminology change: drop 'integral' completely. Now we just have integer,
which includes i1.  Change the semantics of the various instructions to
support i1 where they didn't before: for example, it is now legal [though
not terribly useful, as it is the same as xor] to add two i1 values.

llvm-svn: 33217
2007-01-15 01:54:13 +00:00
Chris Lattner
bf132d8342 Make use of isInteger vs isIntegral more explicit
llvm-svn: 33216
2007-01-15 01:48:11 +00:00
Chris Lattner
01b3134b12 ensure these don't xpass if opt crashes
llvm-svn: 33215
2007-01-15 01:04:21 +00:00
Chris Lattner
32e75c19b5 Fix correlated exprs regressions
llvm-svn: 33214
2007-01-15 01:02:34 +00:00
Chris Lattner
9bda331dc3 don't discriminate against i1
llvm-svn: 33213
2007-01-15 00:45:50 +00:00
Reid Spencer
0b130d9f4d Regenerate
llvm-svn: 33212
2007-01-15 00:26:18 +00:00
Reid Spencer
a5cbe5f4e5 Reorganize things a bit in preparation for rewrite. Although this looks
like a lot, its really only two changes:

1. Move stuff that should be private to .y out of the .h file.
2. Make all semantic values pointers.

This cleans up the silly destroy methods and changes them to delete. It
also moves the TypeInfo and ValueInfo classes into the .y where we can
modify them more readily because they have no clients any more.

This shouldn't result in any functional changes in llvm-upgrade.

llvm-svn: 33211
2007-01-15 00:25:53 +00:00
Chris Lattner
e7c5d04b53 Fix Analysis/Dominators/2006-10-02-BreakCritEdges.ll
llvm-svn: 33210
2007-01-15 00:15:09 +00:00
Reid Spencer
884f048f42 Primitive types don't include integers any more, adjust.
llvm-svn: 33209
2007-01-14 21:46:42 +00:00
Chris Lattner
34dd9c63b6 instcombine has always been miscompiling fcmp x, x, disregarding possible
NANs.  This fixes PR1111 and Transforms/InstCombine/2007-01-14-FcmpSelf.ll

llvm-svn: 33208
2007-01-14 19:42:17 +00:00
Chris Lattner
018d873ce8 add accessors
llvm-svn: 33207
2007-01-14 19:41:24 +00:00
Chris Lattner
de35355e62 new testcase
llvm-svn: 33206
2007-01-14 19:40:48 +00:00
Chris Lattner
fd98d4e56d verify that hidden visibility support never goes away
llvm-svn: 33205
2007-01-14 18:57:37 +00:00
Chris Lattner
9801f36d97 Fix PR1110 and Analysis/Dominators/2007-01-14-BreakCritEdges.ll by being
more careful about unreachable code when updating dominator info.

llvm-svn: 33204
2007-01-14 18:33:35 +00:00
Chris Lattner
4d1a38e365 testcase for pr1110
llvm-svn: 33203
2007-01-14 18:33:11 +00:00
Anton Korobeynikov
9398ef2167 No hidden visiblity on Mingw32/Cygwin
llvm-svn: 33202
2007-01-14 11:49:39 +00:00
Chris Lattner
ad0711b2eb clarify linkonce/weak linkage
llvm-svn: 33201
2007-01-14 06:51:48 +00:00
Chris Lattner
69b1c31416 add support for hidden visibility to darwin/ppc and linux/ppc targets
llvm-svn: 33200
2007-01-14 06:37:54 +00:00