Commit Graph

23929 Commits

Author SHA1 Message Date
Evan Cheng 78e3d565af INSERT_VECTOR_ELT lowering bug:
store vector to $esp
  store element to $esp + sizeof(VT) * index
  load  vector from $esp
The bug is VT is the type of the vector element, not the type of the vector!

llvm-svn: 27517
2006-04-08 01:46:37 +00:00
Reid Spencer 6025241ee4 Ensure that the automatic "CVS build" detection works for both
objdir == srcdir and objdir != srcdir.

llvm-svn: 27516
2006-04-08 01:41:26 +00:00
Chris Lattner 0197e454d1 Add shufflevector support
llvm-svn: 27515
2006-04-08 01:19:47 +00:00
Chris Lattner aa3185f12e Stub out shufflevector
llvm-svn: 27514
2006-04-08 01:19:25 +00:00
Chris Lattner 17bd60588c Add supprot for shufflevector
llvm-svn: 27513
2006-04-08 01:19:12 +00:00
Chris Lattner 9ff96a70f2 regenerate
llvm-svn: 27512
2006-04-08 01:18:56 +00:00
Chris Lattner 8ec3c2e31f Parse shufflevector
llvm-svn: 27511
2006-04-08 01:18:35 +00:00
Chris Lattner bbe0a4248b Add shufflevector support, todo, implement better constant folding.
llvm-svn: 27510
2006-04-08 01:18:18 +00:00
Chris Lattner 161e7cfc17 Add shufflevector reading support
llvm-svn: 27509
2006-04-08 01:17:59 +00:00
Chris Lattner a348351da3 new testcase for shufflevector
llvm-svn: 27508
2006-04-08 01:17:42 +00:00
Chris Lattner c951f4c112 Add a new shufflevector instruction
llvm-svn: 27507
2006-04-08 01:15:18 +00:00
Evan Cheng 0df9c9f57d ldmxcsr and stmxcsr.
llvm-svn: 27506
2006-04-08 00:47:44 +00:00
Evan Cheng 8122cd5524 Fix int_x86_sse_stmxcsr and int_x86_sse_ldmxcsr. Not directly translated from
gcc builtins. They are not lowered into these intrinsics which take a ptr
argument.

llvm-svn: 27505
2006-04-08 00:47:01 +00:00
Jeff Cohen 42d5c23911 Get Visual Studio building again.
llvm-svn: 27504
2006-04-08 00:43:03 +00:00
Jim Laskey 7d4592733c Remove section change in function end, preventing override of function's real
section.

llvm-svn: 27503
2006-04-08 00:35:59 +00:00
Chris Lattner 016a0e50a5 Description for the new shufflevector instruction I'm adding.
llvm-svn: 27502
2006-04-08 00:13:41 +00:00
Evan Cheng ac847268c5 Code clean up.
llvm-svn: 27501
2006-04-07 21:53:05 +00:00
Evan Cheng c447d87424 Doh!
llvm-svn: 27500
2006-04-07 21:52:15 +00:00
Reid Spencer 4873c79e41 Invert the tests on LLVMGCC_MAJVERS to check for value 4 instead of value
3. This ensures that if llvm-gcc isn't available and consequently the value
of LLVMGCC_MAJVERS is blank, that the old (include runtime) behavior will
persist.

llvm-svn: 27499
2006-04-07 21:45:23 +00:00
Evan Cheng aa18a52545 Added patterns for MOVHPSmr and MOVLPSmr.
llvm-svn: 27497
2006-04-07 21:20:58 +00:00
Evan Cheng 748e573ce5 Keep track of an Mac OS X / x86 ABI bug.
llvm-svn: 27496
2006-04-07 21:19:53 +00:00
Evan Cheng 37b5e5aa1d Remove int_x86_sse_storeh_ps and int_x86_sse_storel_ps. These are now lowered
by the frontend.

llvm-svn: 27495
2006-04-07 21:18:40 +00:00
Jim Laskey c0d6518f27 Make sure that debug labels are defined within the same section and after the
entry point of a function.

llvm-svn: 27494
2006-04-07 20:44:42 +00:00
Chris Lattner 292574fbbc add a newline at the end of the file
llvm-svn: 27492
2006-04-07 18:41:16 +00:00
Jim Laskey 2d7298c362 Foundation for call frame information.
llvm-svn: 27491
2006-04-07 16:34:46 +00:00
Chris Lattner e690a93c64 THis -> This
llvm-svn: 27490
2006-04-07 16:21:59 +00:00
Reid Spencer 5a6537d643 For PR723:
Configure with the --enable-optimized and --enable-assertions option to
ensure that binary RPM packages are built with the options required by the
feature request.

llvm-svn: 27489
2006-04-07 16:07:37 +00:00
Reid Spencer 056f022569 For PR723:
1. Don't force debug builds to have assertion checking turned on always.
   Let the default (on) be taken, or overridden by the command line
2. Create two new BuildModes based on assertion checking: Release+Assert
   and Debug-Assert.
3. Ensure that when building a distribution we get a release build with
   assertions enabled, regardless of the tree's configuration.
4. (unrelated) Fix library name generation for llvm-config usage.

llvm-svn: 27488
2006-04-07 16:06:18 +00:00
Reid Spencer b65ade8bd9 For PR723:
Support detection of a "CVS" directory at configure time to distinguish
whether this is a release build or a "from tree" build. This knowledge is
used to set the defaults for --enable-optimzied and --enable-assertions
options.

llvm-svn: 27487
2006-04-07 16:01:51 +00:00
Reid Spencer 080eb0b3ff For PR723:
- Provide for the include of ENABLE_ASSERTIONS flag from the configuration.
- Enhance the documentation for ENABLE_OPTIMIZED

llvm-svn: 27486
2006-04-07 15:59:41 +00:00
Reid Spencer 754cfecd59 For PR723:
Print out something at the end of the build that indicates what kind of
build has just completed.

llvm-svn: 27485
2006-04-07 15:58:18 +00:00
Reid Spencer 1a63f48c52 Ignore stuff built by "make dist".
llvm-svn: 27484
2006-04-07 15:55:18 +00:00
Evan Cheng bb66ec263c A saner workaround. I hope.
llvm-svn: 27483
2006-04-07 08:31:56 +00:00
Evan Cheng 1e86e679a5 Temporary workaround for a Mac OSX specific issue.
llvm-svn: 27482
2006-04-07 08:10:09 +00:00
Evan Cheng 085c61d825 Added more shuffle tests
llvm-svn: 27481
2006-04-07 05:35:45 +00:00
Chris Lattner 92b1a57010 We have an assertion that checks that we do not encode null values into the
.bc file if they are supposed to be implicit.  This is cool, except that it
checked *after* constant expr folding: improving constant expr folding could
cause the .bc reader to assert out on old .bc files.  Move the check so that
it checks all simple constants, but no constantexprs.

llvm-svn: 27480
2006-04-07 05:00:02 +00:00
Chris Lattner e4f9d7b23c Constant fold extractelement(zero, x) -> zero
llvm-svn: 27479
2006-04-07 04:44:06 +00:00
Chris Lattner 8ec0205de4 Fix inlining of insert/extract element constantexprs
llvm-svn: 27478
2006-04-07 04:41:03 +00:00
Evan Cheng 5c349b8b7d Add code to RemapOperand() to handle Instruction::ExtractElement and
Instruction::InsertElement.

llvm-svn: 27477
2006-04-07 01:27:42 +00:00
Evan Cheng d8e1a01be6 A MOVPS2SSmr, i.e. _mm_store_ss, encoding bug.
Also MOVPDI2DIrr.

llvm-svn: 27476
2006-04-06 23:53:29 +00:00
Evan Cheng bc658994f6 Remove int_x86_sse_loadh_ps and int_x86_sse_loadl_ps. These are now lowered
by the frontend to shuffles.

llvm-svn: 27475
2006-04-06 23:25:41 +00:00
Evan Cheng c995b45f67 - movlp{s|d} and movhp{s|d} support.
- Normalize shuffle nodes so result vector lower half elements come from the
  first vector, the rest come from the second vector. (Except for the
  exceptions :-).
- Other minor fixes.

llvm-svn: 27474
2006-04-06 23:23:56 +00:00
Evan Cheng acf8b3c828 New entries.
llvm-svn: 27473
2006-04-06 23:21:24 +00:00
Evan Cheng 613996c55e 1. If both vector operands of a vector_shuffle are undef, turn it into an undef.
2. A shuffle mask element can also be an undef.

llvm-svn: 27472
2006-04-06 23:20:43 +00:00
Andrew Lenharth 1596a1b276 This may be overconservative, but it lets the new cfe compile
llvm-svn: 27471
2006-04-06 23:18:45 +00:00
Chris Lattner e61cfad815 Add an item
llvm-svn: 27470
2006-04-06 23:16:19 +00:00
Chris Lattner 466841ddc7 Make sure to return the result in the right type.
llvm-svn: 27469
2006-04-06 23:12:19 +00:00
Andrew Lenharth 477169e36a Move this to lib/Analysis.
This reduces Core size, yay.

llvm-svn: 27468
2006-04-06 22:58:58 +00:00
Chris Lattner a4bbfaed5c Match vpku[hw]um(x,x).
Convert vsldoi(x,x) to work the same way other (x,x) cases work.

llvm-svn: 27467
2006-04-06 22:28:36 +00:00
Chris Lattner b8fc537634 Add testcases for vpku[hw]um(x,x)
llvm-svn: 27466
2006-04-06 22:27:59 +00:00