1
0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-10-21 03:53:04 +02:00
Commit Graph

73122 Commits

Author SHA1 Message Date
Nick Lewycky
bf55e4b776 Emit trailing padding on constant vectors when TargetData says that the vector
is larger than the sum of the elements (including per-element padding).

llvm-svn: 133631
2011-06-22 18:55:03 +00:00
Justin Holewinski
da4d209996 PTX: Fix FrameIndex mapping bug
llvm-svn: 133619
2011-06-22 16:07:03 +00:00
Jay Foad
888cd746b8 Replace the existing forms of ConstantArray::get() with a single form
that takes an ArrayRef.

llvm-svn: 133615
2011-06-22 09:24:39 +00:00
Jay Foad
7ef9c9ad1e Make ConstantVector::get() always take an ArrayRef, never a std::vector.
llvm-svn: 133614
2011-06-22 09:10:19 +00:00
Dan Bailey
2d3b8e126a Test Commit.
llvm-svn: 133613
2011-06-22 09:04:30 +00:00
Jay Foad
f653a9e838 Eliminate a temporary std::vector in ConstantStruct::get().
llvm-svn: 133612
2011-06-22 08:55:11 +00:00
Jay Foad
f921ffd79e Extend ConstantUniqueMap with a new template parameter ValRefType,
representing a constant reference to ValType. Normally this is just
"const ValType &", but when ValType is a std::vector we want to use
ArrayRef as the reference type.

llvm-svn: 133611
2011-06-22 08:50:06 +00:00
Andrew Trick
887ccaa3dc Only do config-time substitution of LLVM_BUILD_MODE in
test/lit.site.cfg, not Unit/test/lit.site.cfg.

llvm-svn: 133608
2011-06-22 05:43:36 +00:00
Rafael Espindola
9cc87c6608 Reenable the optimization added in 133415, but change the definition of a "simple" bb to
be one with only one unconditional branch and no phis. Duplicating the phis in this case
is possible, but requeres liveness analysis or breaking edges.

llvm-svn: 133607
2011-06-22 04:01:58 +00:00
Justin Holewinski
376f1d46d4 PTX: Add signed integer comparisons
llvm-svn: 133599
2011-06-22 02:09:50 +00:00
Justin Holewinski
0844ac41b6 PTX: Add .address_size directive if PTX version >= 2.3
Patch by Wei-Ren Chen

llvm-svn: 133589
2011-06-22 00:43:56 +00:00
Devang Patel
f610afdefb Test case for r133560.
llvm-svn: 133585
2011-06-22 00:03:42 +00:00
Francois Pichet
3a4cc9df06 Unbreak the CMake build
llvm-svn: 133574
2011-06-21 23:19:23 +00:00
Devang Patel
8b96f42a1e After register is spilled there should not be any DBG_VALUE referring the same register.
llvm-svn: 133569
2011-06-21 23:02:36 +00:00
Jim Grosbach
cfcba2d22b Consolidate some TableGen diagnostic helper functions.
TableGen had diagnostic printers sprinkled about in a few places. Pull them
together into a single location in Error.cpp.

llvm-svn: 133568
2011-06-21 22:55:50 +00:00
Owen Anderson
deeadc6b55 Fix some trailing issues from my introduction of MVT::untyped and its use for REGISTER_SEQUENCE.
llvm-svn: 133567
2011-06-21 22:54:23 +00:00
Nick Lewycky
8e5c09b7dc Add support for assembling "movq" when it's correct to do so, while continuing
to emit "movd" across the board to continue supporting a Darwin assembler bug.
This is the reincarnation of r133452.

llvm-svn: 133565
2011-06-21 22:45:41 +00:00
Bill Wendling
94ed847adb Add verbose EH table printing to SjLj exception tables.
llvm-svn: 133561
2011-06-21 22:40:24 +00:00
Devang Patel
42bd3da91f There could be more than one DBG_VALUE instructions for variables where all of them have offset based on one register.
llvm-svn: 133560
2011-06-21 22:36:03 +00:00
Bill Wendling
16b9bd4460 Improve the comment printing for the EH table. This gives a much more detailed
explanation of what the EH table describes.

llvm-svn: 133559
2011-06-21 22:30:20 +00:00
Eric Christopher
5939b74818 Handle the memory-ness of all U+ ARM constraints.
Noticed on inspection.

llvm-svn: 133553
2011-06-21 22:10:57 +00:00
Devang Patel
b1e608f669 Remove r130409, as requested by Chris.
llvm-svn: 133536
2011-06-21 19:46:09 +00:00
Evan Cheng
dc39e02b91 Reorg. No functionality change.
llvm-svn: 133533
2011-06-21 19:00:54 +00:00
Bob Wilson
5b04895bb8 Revert r133452: "Emit movq for 64-bit register to XMM register moves..."
This is breaking compiler-rt and llvm-gcc builds on MacOSX when not using
the integrated assembler.

llvm-svn: 133524
2011-06-21 17:35:13 +00:00
Anna Zaks
488fc45c84 Add support for sadd.with.overflow and uadd.with.overflow intrinsics to the CBackend by emitting definitions for each intrinsic that occurs in the module.
llvm-svn: 133522
2011-06-21 17:18:15 +00:00
Andrew Trick
04c3bb47ce IVUsers no longer needs to record the phis.
llvm-svn: 133518
2011-06-21 15:43:52 +00:00
Jay Foad
6015684974 Remove deprecated forms of StringMap::GetOrCreateValue().
llvm-svn: 133517
2011-06-21 15:37:05 +00:00
Jay Foad
ea4f714a73 Remove some unnecessary uses of c_str().
llvm-svn: 133516
2011-06-21 15:36:24 +00:00
Benjamin Kramer
02bc17d36d Remove unused variables.
llvm-svn: 133514
2011-06-21 14:58:30 +00:00
Jay Foad
2691fd9891 Reinstate r133435 and r133449 (reverted in r133499) now that the clang
self-hosted build failure has been fixed (r133512).

llvm-svn: 133513
2011-06-21 10:33:19 +00:00
Jay Foad
d7276856c6 Don't use PN->replaceUsesOfWith() to change a PHINode's incoming blocks,
because it won't work after my phi operand changes, because the incoming
blocks will no longer be Uses.

llvm-svn: 133512
2011-06-21 10:02:43 +00:00
Jay Foad
7bf84923b1 Add a reduced test case for the buildbot failure (clang self-hosted
build) caused by r133435.

llvm-svn: 133509
2011-06-21 08:33:49 +00:00
Chris Lattner
a9ee8489df use the MapEntryTy typedef instead of writing it out long form,
add some fixme's about methods that should be removed.

Merged from type-system-rewrite.

llvm-svn: 133504
2011-06-21 06:22:33 +00:00
Evan Cheng
40adfc21f6 Teach dag combine to match halfword byteswap patterns.
1. (((x) & 0xFF00) >> 8) | (((x) & 0x00FF) << 8)
   => (bswap x) >> 16
2. ((x&0xff)<<8)|((x&0xff00)>>8)|((x&0xff000000)>>8)|((x&0x00ff0000)<<8))
   => (rotl (bswap x) 16)

This allows us to eliminate most of the def : Pat patterns for ARM rev16
revsh instructions. It catches many more cases for ARM and x86.

rdar://9609108

llvm-svn: 133503
2011-06-21 06:01:08 +00:00
Andrew Trick
0f76c6afe8 indvars -disable-iv-rewrite: Adds support for eliminating identity
ops.

This is a rewrite of the IV simplification algorithm used by
-disable-iv-rewrite. To avoid perturbing the default mode, I
temporarily split the driver and created SimplifyIVUsersNoRewrite. The
idea is to avoid doing opcode/pattern matching inside
IndVarSimplify. SCEV already does it. We want to optimize with the
full generality of SCEV, but optimize def-use chains top down on-demand rather
than rewriting the entire expression bottom-up. This was easy to do
for operations that SCEV can prove are identity function. So we're now
eliminating bitmasks and zero extends this way.

A result of this rewrite is that indvars -disable-iv-rewrite no longer
requires IVUsers.

llvm-svn: 133502
2011-06-21 03:22:38 +00:00
Chad Rosier
16a86e04cf Revert r133435 and r133449 to appease buildbots.
llvm-svn: 133499
2011-06-21 02:09:03 +00:00
Akira Hatanaka
288c049012 Add A0 and A1 to the list of registers used for returning a value in order to
handle functions with return type Complex long long.

llvm-svn: 133497
2011-06-21 01:28:11 +00:00
Akira Hatanaka
1197db1f9b Coding style fixes.
llvm-svn: 133496
2011-06-21 01:02:03 +00:00
Akira Hatanaka
1e08980a21 Re-apply 132758 and 132768 which were speculatively reverted in 132777.
llvm-svn: 133494
2011-06-21 00:40:49 +00:00
Bill Wendling
54d7d72c94 Don't mark the eh.dispatch.setup with a memory access marker. We want this to
stick around even during fast isel.
<rdar://problem/9637156>

llvm-svn: 133493
2011-06-21 00:35:15 +00:00
Dan Gohman
d8baa3f8ce Completely short-circuit out ARC optimization if the ARC runtime
functions do not appear in the module.

llvm-svn: 133478
2011-06-20 23:20:43 +00:00
Bill Wendling
c04086de5c Remove the subclassing. This will be moved to the ASM printer.
llvm-svn: 133473
2011-06-20 22:12:24 +00:00
Jakob Stoklund Olesen
22d12568aa Skip fields that don't exist in the Register class.
llvm-svn: 133470
2011-06-20 20:56:05 +00:00
Justin Holewinski
e62da847fa PTX: Fix conversion between predicates and value types
llvm-svn: 133454
2011-06-20 18:42:48 +00:00
Nick Lewycky
831fb8200d Emit movq for 64-bit register to XMM register moves, but continue to accept
movd when assembling.

llvm-svn: 133452
2011-06-20 18:33:26 +00:00
Jay Foad
e834069c4d Fix a check for PHINodes with two incoming values.
llvm-svn: 133449
2011-06-20 17:46:19 +00:00
Justin Holewinski
c4253975bd PTX: Fix if-then-else formatting and add missing asserts
llvm-svn: 133447
2011-06-20 17:08:56 +00:00
Rafael Espindola
1c53a39ff9 Disable again.
llvm-svn: 133446
2011-06-20 17:04:08 +00:00
Justin Holewinski
a5d3db3bd2 PTX: Add basic register spilling code
The current implementation generates stack loads/stores, which are
really just mov instructions from/to "special" registers.  This may
not be the most efficient implementation, compared to an approach where
the stack registers are directly folded into instructions, but this is
easier to implement and I have yet to see a case where ptxas is unable
to see through this kind of register usage and know what is really
going on.

llvm-svn: 133443
2011-06-20 15:56:20 +00:00
Roman Divacky
79578394f5 Don't apply on PPC64 the 32bit ADDIC optimizations as there's no overflow
with 32bit values.

llvm-svn: 133439
2011-06-20 15:28:39 +00:00