Commit Graph

58936 Commits

Author SHA1 Message Date
Anton Korobeynikov
faa75f6e47 FP16 constfolding
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98911 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-19 00:36:35 +00:00
Chris Lattner
f65027842e change Target.getInstructionsByEnumValue to return a reference
to a vector that CGT stores instead of synthesizing it on every 
call.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98910 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-19 00:34:35 +00:00
Jeffrey Yasskin
0392080fbf Revert r98892. BSD systems may not have bash installed at all.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98909 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-19 00:32:11 +00:00
Chris Lattner
e4e8bb1541 factor copy and paste code.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98908 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-19 00:23:20 +00:00
Chris Lattner
b61e09de6d don't go through getInstructions().
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98906 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-19 00:18:23 +00:00
Jeffrey Yasskin
c3e6859d8d Bugpoint's default memory limit (100MB) was too low for valgrind, so
this patch raises the default to 800MB when valgrind's active.  800
was chosen semi-arbitrarily.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98905 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-19 00:09:28 +00:00
Chris Lattner
f30187a363 look up instructions by record, not by name.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98904 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-19 00:07:20 +00:00
Chris Lattner
e3486a4c9a Now that tblgen can handle matching implicit defs of instructions
to input patterns, we can fix X86ISD::CMP and X86ISD::BT as taking
two inputs (which have to be the same type) and *returning an i32*.
This is how the SDNodes get made in the graph, but we weren't able
to model it this way due to deficiencies in the pattern language.

Now we can change things like this:

 def UCOM_FpIr80: FpI_<(outs), (ins RFP80:$lhs, RFP80:$rhs), CompareFP,
-                  [(X86cmp RFP80:$lhs, RFP80:$rhs),
-                   (implicit EFLAGS)]>; // CC = ST(0) cmp ST(i)
+                  [(set EFLAGS, (X86cmp RFP80:$lhs, RFP80:$rhs))]>;

and fix terrible crimes like this:

-def : Pat<(parallel (X86cmp GR8:$src1, 0), (implicit EFLAGS)),
+def : Pat<(X86cmp GR8:$src1, 0),
           (TEST8rr GR8:$src1, GR8:$src1)>;

This relies on matching the result of TEST8rr (which is EFLAGS, which is
an implicit def) to the result of X86cmp, an i32.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98903 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-19 00:01:11 +00:00
Bob Wilson
50622ce8cf Update comment to refer to the right filename.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98902 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-18 23:57:57 +00:00
Chris Lattner
de3e05fe33 tidy up
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98901 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-18 23:57:57 +00:00
Chris Lattner
928795387f expand tblgen's support for instructions with implicit defs.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98900 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-18 23:57:40 +00:00
Chris Lattner
6c6ba36493 infer results of a pattern from implicit defs. This allows you to do something
like this:

 def : Pat<(add ...),
           (FOOINST)>;

When fooinst only has a single implicit def (e.g. to R1).  This will be handled
as if written as (set R1, (FOOINST ...))



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98897 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-18 23:15:10 +00:00
Jeffrey Yasskin
1ff781fb50 Work around a valgrind oddity where it doesn't pass the full path of a
script to the #! command by using bash instead of /bin/sh.  Bash
searches $PATH for its script argument, but dash, which /bin/sh
resolves to on some systems, does not.
https://bugs.kde.org/show_bug.cgi?id=231257 tracks the valgrind
problem.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98892 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-18 22:56:02 +00:00
Anton Korobeynikov
341ab138fb Get rid of target-specific fp <-> int nodes when still I'm here.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98889 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-18 22:35:45 +00:00
Anton Korobeynikov
f0d500768d Get rid of target-specific nodes for fp16 <-> fp32 conversion.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98888 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-18 22:35:37 +00:00
Anton Korobeynikov
6755d97a62 Refactor Reg-Reg copy emission routine for ARM. This makes cross-regclass copies weirdness more straightforward. Also, add GPR <-> SPR copy support.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98887 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-18 22:35:02 +00:00
Eric Christopher
88dee30a6e Couple of changes that Dan mentioned for llvm.stackprotector fast-isel.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98881 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-18 21:58:33 +00:00
Daniel Dunbar
9fdac902d4 X86MCCodeEmitter: Fix two minor issues with reloc_riprel_4byte_movq_load, we
were missing it on some movq instructions and were not including the appropriate
PCrel bias.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98880 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-18 21:53:54 +00:00
Chris Lattner
f506b6b471 capture implicit uses and defs in CodeGenInstruction
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98879 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-18 21:42:03 +00:00
Chris Lattner
ef988984d7 simplify this code, the # of sets in the pattern for an instruction
shouldn't change this.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98872 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-18 21:15:05 +00:00
Chris Lattner
0b6c926856 disallow concatenation of two dags with different operators.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98871 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-18 21:07:51 +00:00
Chris Lattner
f55eed299b rewrite this to not artificially force concat the ins/outs list.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98870 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-18 21:07:39 +00:00
Chris Lattner
b7d5226a4e fix some buggy ops concatentation
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98869 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-18 21:06:54 +00:00
Chris Lattner
b0be4d261b eliminate support for "ops" in the input/output list of an
instruction.  Instructions must use 'ins' and 'outs' now.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98868 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-18 20:56:35 +00:00
Chris Lattner
09a2769a7f use ins/outs
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98867 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-18 20:55:31 +00:00
Chris Lattner
6784bdf863 use ins/outs.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98866 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-18 20:55:18 +00:00
Chris Lattner
cedef1ccf0 remove some code that was working around old sparc v9 backend bugs.
Add checking that the input/output operand list in spelled right.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98865 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-18 20:50:52 +00:00
Chris Lattner
34b8a8857c outs come before ins.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98864 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-18 20:50:06 +00:00
Eric Christopher
07754c2a1b Make fast-isel understand llvm.stackprotector.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98862 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-18 20:27:26 +00:00
Bob Wilson
95808328b4 Refactor NEON ld/st instructions to hardcode class arguments that are constants.
No functional changes.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98860 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-18 20:18:39 +00:00
Daniel Dunbar
1814cf0f10 lit: Tweak example tests config.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98859 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-18 20:06:05 +00:00
Daniel Dunbar
f98bc6320b MC/X86/AsmMatcher: Use the new instruction cleanup routine to implement a
temporary workaround for matching inc/dec on x86_64 to the correct instruction.
 - This hack will eventually be replaced with a robust mechanism for handling
   matching instructions based on the available target features.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98858 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-18 20:06:02 +00:00
Daniel Dunbar
8cc9c0c487 MC/AsmMatcher: Add support for target specific "instruction cleanup" functions,
to allow custom post-processing of matched instructions.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98857 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-18 20:05:56 +00:00
Chris Lattner
618d0ed4bc fix an x86-64 encoding bug Daniel found.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98855 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-18 20:04:36 +00:00
Dan Gohman
debbef20ee offsetof always has type size_t.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98854 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-18 19:37:35 +00:00
Dan Gohman
40b037ef61 Simplify this code.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98853 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-18 19:34:33 +00:00
Benjamin Kramer
d720712f67 Fix a name clash that was breaking MSVC builds.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98852 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-18 19:34:06 +00:00
Dan Gohman
aea1b501ce Explicitly cast std::min's arguments to avoid type mismatches.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98851 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-18 19:01:12 +00:00
Gabor Greif
fe7bb05ad1 another one
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98850 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-18 18:59:08 +00:00
Dan Gohman
9553188fcc Define placement new wrappers for BumpPtrAllocator and
RecyclingAllocator to allow client code to be simpler, and
simplify several clients.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98847 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-18 18:49:47 +00:00
Dan Gohman
5ffc72e7ce Make this test more lenient; with SmallVector now using actually
aligned storage, the capacity may be more than what is explicitly
requested.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98846 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-18 18:47:50 +00:00
Bob Wilson
d1ec31dca5 Fix pr6543: svn r88806 changed MachineJumpTableInfo::getJumpTableIndex() to
always create a new jump table.  The intention was to avoid merging jump
tables in SelectionDAGBuilder, and to wait for the branch folding pass to
merge tables.  Unfortunately, the same getJumpTableIndex() method is also
used to merge tables in branch folding, so as a result of this change
branch tables are never merged.  Worse, the branch folding code is expecting
getJumpTableIndex to always return the index of an existing table, but with
this change, it never does so.  In at least some cases, e.g., pr6543, this
creates references to non-existent tables.

I've fixed the problem by adding a new createJumpTableIndex function, which
will always create a new table, and I've changed getJumpTableIndex to only
look at existing tables.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98845 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-18 18:42:41 +00:00
Dan Gohman
74cdc75313 Fix a bug found by inspection; in the __GNUC__ code, the alignment
doesn't apply to the type, only to the variable, so subsequent uses
of U which expect it to be aligned weren't actually aligned.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98843 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-18 18:40:47 +00:00
Andrew Lenharth
8bba4dd9bc Nick wants these alphabetized
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98840 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-18 18:18:00 +00:00
Chris Lattner
0f53cf2236 add a special relocation type for movq loads for object
files that produce special relocation types where the 
linker changes movq's into lea's.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98839 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-18 18:10:56 +00:00
Dan Gohman
1584390e55 Add explicit braces to avoid ambiguous ‘else’.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98838 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-18 18:07:13 +00:00
Chris Lattner
e10038e54e callq is pcrelative
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98835 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-18 17:52:22 +00:00
Bob Wilson
35ee7d6e04 Check if function names start with "llvm." before trying to lookup them up as
intrinsics.  The intrinsic lookup code assumes that this check has been done
and assumes the names are at least 6 characters long.  Valgrind complained
about this.  pr6638.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98831 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-18 16:52:15 +00:00
Devang Patel
1ac2429d1f Fix comment.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98830 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-18 16:41:16 +00:00
Dan Gohman
c93b4cff89 Add the ability to "intern" FoldingSetNodeID data into a
BumpPtrAllocator-allocated region to allow it to be stored in a more
compact form and to avoid the need for a non-trivial destructor call.

Use this new mechanism in ScalarEvolution instead of
FastFoldingSetNode to avoid leaking memory in the case where a
FoldingSetNodeID uses heap storage, and to reduce overall memory
usage.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98829 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-18 16:16:38 +00:00