Commit Graph

109924 Commits

Author SHA1 Message Date
Bill Schmidt
6536b834f8 Change order of tablegen generated fast-isel instruction code to be
based on instruction complexity

The order that tablegen fast-isel instruction code is generated is
currently based on the text of the predicate (using string
less-than). This patch changes this to instead use the instruction
complexity. Because the complexities are not unique a C++ multimap is
used instead of a map.

This fixes the problem where code with no predicate always comes out
first (the empty string always compares as less than all other
strings) thus making the code with predicates dead code. See the FMUL
code in PPCFastISel.cpp for an example. It also more closely matches
the normal codegen ordering. Some error checking in the tablegen
fast-isel code is fixed as well.

Patch by Bill Seurer.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@222038 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-14 21:05:45 +00:00
Tom Stellard
99b3234323 R600/SI: Mark s_movk_i32 as rematerializable
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@222037 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-14 20:43:28 +00:00
Tom Stellard
6beb81daa5 R600/SI: Fix spilling of m0 register
If we have spilled the value of the m0 register, then we need to restore
it with v_readlane_b32 to a regular sgpr, because v_readlane_b32 can't
write to m0.

v_readlane_b32 can't write to m0, so

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@222036 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-14 20:43:26 +00:00
Frederic Riss
f6c436e9d7 COFF: Add support for Dwarf accelerator tables.
This allows COFF targets to emit accelerator tables
when requested by -dwarf-accel-tables=Enable instead
of aborting. The test DebugInfo/cross-cu-inlining.ll
covers this on COFF platforms.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@222034 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-14 20:33:40 +00:00
Matt Arsenault
24e874a1dd R600/SI: Combine min3/max3 instructions
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@222032 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-14 20:08:52 +00:00
David Majnemer
a9d69424f4 yaml2obj, COFF: Correctly calculate SizeOfImage and SizeOfHeaders
SizeOfHeaders must be aligned to the FileAlignment.
SizeOfImage must be at least the SizeOfHeaders aligned to the
SectionAlignment.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@222030 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-14 19:35:59 +00:00
Frederic Riss
4ad9e9b0b0 [dwarfdump] Handle relocations in Dwarf accelerator tables
ELF targets (and maybe COFF) use relocations when referring
to strings in the .debug_str section. Handle that in the
accelerator table dumper. This commit restores the
test/DebugInfo/cross-cu-inlining.ll test to its expected
platform independant form, validating that the fix works
(this test failed on linux boxes).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@222029 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-14 19:30:08 +00:00
David Blaikie
7ffaebf9da Remove redundant virtual on overriden functions.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@222023 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-14 19:06:36 +00:00
Duncan P. N. Exon Smith
6bc59dcccb IR: Make MDString inherit from Metadata
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@222022 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-14 18:45:40 +00:00
Matt Arsenault
848d9223c5 R600/SI: Fix verifier error from a branch on IMPLICIT_DEF
SIILowerI1Copies wasn't correctly handling this case.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@222020 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-14 18:43:41 +00:00
Duncan P. N. Exon Smith
e9b30c72ad IR: Take an LLVMContext in Metadata::Metadata()
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@222019 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-14 18:42:09 +00:00
Duncan P. N. Exon Smith
24593ac779 Add a blank line, NFC
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@222018 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-14 18:42:06 +00:00
Matt Arsenault
3dd7f8668b Fix unused variable warning without asserts
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@222017 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-14 18:40:49 +00:00
Matt Arsenault
01213b1132 R600/SI: Match integer min / max instructions
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@222015 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-14 18:30:06 +00:00
Matt Arsenault
8fd3b90c3f R600/SI: Use S_BFE_I64 for 64-bit sext_inreg
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@222012 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-14 18:18:16 +00:00
Chad Rosier
5c76b3d03e [Reassociate] Canonicalize the operands of all binary operators.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@222008 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-14 17:09:19 +00:00
Frederic Riss
e968287996 Tentatively appease the bots.
If this workaround gets the bots green, then we have to find out
why the -dwarf-accel-tables=Enable option doesn't work as
expected on non-darwin platforms.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@222007 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-14 17:08:18 +00:00
Chad Rosier
1298a29c64 [Reassociate] Canonicalize operands of vector binary operators.
Prior to this commit fmul and fadd binary operators were being canonicalized for
both scalar and vector versions.  We now canonicalize add, mul, and, or, and xor
vector instructions.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@222006 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-14 17:08:15 +00:00
Chad Rosier
7e61b1fb62 [Reassociate] Canonicalize constants to RHS operand.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@222005 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-14 17:05:59 +00:00
Frederic Riss
e26d79386b Reapply "[dwarfdump] Add support for dumping accelerator tables."
This reverts commit r221842 which was a revert of r221836 and of the
test parts of r221837.

This new version fixes an UB bug pointed out by David (along with
addressing some other review comments), makes some dumping more
resilient to broken input data and forces the accelerator tables
to be dumped in the tests where we use them (this decision is
platform specific otherwise).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@222003 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-14 16:15:53 +00:00
Cameron McInally
b3625eb445 [AVX512] Add 512b masked integer shift by immediate patterns.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@222002 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-14 15:43:00 +00:00
Chad Rosier
1945dcff86 [Reassociate] Improve rank debug information. NFC.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@221999 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-14 15:01:38 +00:00
Tom Stellard
239d5231f3 R600/SI: Fix assembly names for exec_hi and exec_lo
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@221995 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-14 14:08:04 +00:00
Tom Stellard
19cb35b4bc R600/SI: Start implementing an assembler
This was done using the Sparc and PowerPC AsmParsers as guides.  So far it
is very simple and only supports sopp instructions.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@221994 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-14 14:08:00 +00:00
Bill Schmidt
40b0f5d6ce [PowerPC] Add VSX builtins for vec_div
This patch adds builtin support for xvdivdp and xvdivsp, along with a
test case.  Straightforward stuff.

There's a companion patch for Clang.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@221983 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-14 12:10:40 +00:00
David Majnemer
080c5fe70e Calm down build bots
r221975 seemed to trigger an ambiguous conversion that only irritated
clang, not gcc.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@221977 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-14 08:38:17 +00:00
David Majnemer
237544b16d obj2yaml, yaml2obj: Add support for COFF executables
In support of serializing executables, obj2yaml now records the virtual address
and size of sections.  It also serializes whatever we strictly need from
the PE header, it expects that it can reconstitute everything else via
inference.

yaml2obj can reconstitute a fully linked executable.

In order to get executables correctly serialized/deserialized, other
bugs were fixed as a circumstance.  We now properly respect file and
section alignments.  We also avoid writing out string tables unless they
are strictly necessary.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@221975 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-14 08:15:42 +00:00
Rafael Espindola
dada992be7 Use size_type for operator[].
This matches std::vector and is more efficient as it avoids
truncations.

With this the text segment of opt goes from 19705442 bytes
to 19703930 bytes.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@221973 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-14 07:02:38 +00:00
NAKAMURA Takumi
5d94c2a6ad SearchForAddressOfSymbol(): Disable 3 symbols, copysignf, fminf, and fmaxf, on msc17. *These were added in VS 2013*
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@221971 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-14 04:53:55 +00:00
Matt Arsenault
d9cd6cfb7d R600/SI: Make constant array static
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@221965 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-14 02:21:58 +00:00
Justin Bogner
d24e90a5de llvm-cov: Sink some reporting logic into CoverageMapping
This teaches CoverageMapping::getCoveredFunctions to filter to a
particular file and uses that to replace most of the logic found in
llvm-cov report.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@221962 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-14 01:50:32 +00:00
Tim Northover
4a7bbf4c29 X86: use getConstant rather than getTargetConstant behind BUILD_VECTOR.
getTargetConstant should only be used when you can guarantee the instruction
selected will be able to cope with the raw value. BUILD_VECTOR is rather too
generic for this so we should use getConstant instead. In that case, an
instruction can still consume the constant, but if it doesn't it'll be
materialised through its own round of ISel.

Should fix PR21352.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@221961 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-14 01:30:14 +00:00
Duncan P. N. Exon Smith
ed7bdcaf03 IR: Rewrite uniquing and creation of MDString
Stop using `Value::getName()` to get the string behind an `MDString`.
Switch to `StringMapEntry<MDString>` so that we can find the string by
its coallocation.

This is part of PR21532.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@221960 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-14 01:17:09 +00:00
David Blaikie
7987683c39 StringMap: Test and finish off supporting perfectly forwarded values in StringMap operations.
Followup to r221946.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@221958 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-14 00:41:46 +00:00
David Blaikie
677506434b Fix 80 cols caught by the linter...
We have a linter running in our build now?

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@221957 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-14 00:41:42 +00:00
Reid Kleckner
4f3c9858e0 Fix build of Mips code with MSVC by using our macro instead of __attribute__((unused)) directly
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@221956 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-14 00:39:33 +00:00
Reid Kleckner
98c86d76df Allow the use of functions as typeinfo in landingpad clauses
This is one step towards supporting SEH filter functions in LLVM.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@221954 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-14 00:35:50 +00:00
Tim Northover
0dfb75c92a CodeGen: assert an instruction is being inserted with the correct iterator.
When "MBB->Insert(It, ...)" is called, we want It to be pointing inside the
correct basic block. No actual failures at the moment, but it's caused problems
before.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@221953 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-14 00:34:59 +00:00
Duncan P. N. Exon Smith
f4742f4c84 IR: Make MDString::getName() private
Hide the fact that `MDString`'s string is stored in `Value::Name` --
that's going to change soon.  Update the only in-tree client that was
using it instead of `Value::getString()`.

Part of PR21532.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@221951 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-13 23:59:16 +00:00
Reid Kleckner
53c7d53c50 Fix the VS 2012 build
VS 2012 doesn't have fminf or fmaxf.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@221949 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-13 23:45:50 +00:00
Reed Kotler
198bb22754 First stage of call lowering for Mips fast-isel
Summary:
This has most of what is needed for mips fast-isel call lowering for O32.
What is missing I will add on the next patch because this patch is already too large.
It should not be doing anything wrong but it will punt on some cases that it is basically
capable of doing.

The mechanism is there for parameters to be passed on the stack but I have not enabled it because it serves as a way for now to prevent some of the strange cases of O32 register passing that I have not fully checked yet and have some issues.

The Mips O32 abi rules are very complicated as far how data is passed in floating and integer registers.

However there is a way to think about this all very simply and this implementation reflects that.

Basically, the ABI rules are written as if everything is passed on the stack and aligned as such.
Once that is conceptually done, it is nearly trivial to reassign those locations to registers and
then all the complexity disappears.

So I have told tablegen that all the data is passed on the stack and during the lowering I fix
this by assigning to registers as per the ABI doc.

This has been my approach and you can line up what I did with the ABI document and see 1 to 1 what
is going on.



Test Plan: callabi.ll

Reviewers: dsanders

Reviewed By: dsanders

Subscribers: jholewinski, echristo, ahatanak, llvm-commits, rfuhler

Differential Revision: http://reviews.llvm.org/D5714

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@221948 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-13 23:37:45 +00:00
Reid Kleckner
016f651f8d Fix symbol resolution of floating point libc builtins in MCJIT
Fix for LLI failure on Windows\X86: http://llvm.org/PR5053

LLI.exe crashes on Windows\X86 when single precession floating point
intrinsics like the following are used: acos, asin, atan, atan2, ceil,
copysign, cos, cosh, exp, floor, fmin, fmax, fmod, log, pow, sin, sinh,
sqrt, tan, tanh

The above intrinsics are defined as inline-expansions in math.h, and are
not exported by msvcr120.dll (Win32 API GetProcAddress returns null).

For an FREM instruction, the JIT compiler generates a call to a stub for
the fmodf() intrinsic, and adds a relocation to fixup at load time. The
loader searches the libraries for the function, but fails because the
symbol is not exported. So, the call target remains NULL and the
execution crashes.

Since the math functions are loaded at JIT/runtime, the JIT can patch
CALL instruction directly instead of the searching the libraries'
exported symbols.  However, this fix caused build failures due to
unresolved symbols like _fmodf at link time.

Therefore, the current fix defines helper functions in the Runtime
link/load library to perform the above operations.  The address of these
helper functions are used to patch up the CALL instruction at load time.

Reviewers: lhames, rnk

Reviewed By: rnk

Differential Revision: http://reviews.llvm.org/D5387

Patch by Swaroop Sridhar!

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@221947 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-13 23:32:52 +00:00
Duncan P. N. Exon Smith
4dc6aa2c17 ADT: Use perfect forwarding in StringMapEntry::Create()
Now you can pass references into constructors.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@221946 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-13 23:23:02 +00:00
Reid Kleckner
fb59c03af9 Relax the gcov version.ll test to check '.' instead of '\*'
The escaping of the '\*' doesn't work with my combination of testing
tools.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@221944 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-13 23:07:55 +00:00
Reid Kleckner
b036e71e41 Silence MSVC warning on missing return after fully covered switch
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@221943 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-13 23:07:22 +00:00
Matt Arsenault
6f485c0bc5 R600/SI: Fix fmin_legacy / fmax_legacy matching for SI
select_cc is expanded on SI, so this was never matched.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@221941 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-13 23:03:09 +00:00
Reid Kleckner
e094cca7f0 Use nullptr instead of NULL for variadic sentinels
Windows defines NULL to 0, which when used as an argument to a variadic
function, is not a null pointer constant. As a result, Clang's
-Wsentinel fires on this code. Using '0' would be wrong on most 64-bit
platforms, but both MSVC and Clang make it work on Windows. Sidestep the
issue with nullptr.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@221940 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-13 22:55:19 +00:00
Chad Rosier
7984fde2dc Revert "[GVN] Perform Scalar PRE on gep indices that feed loads before doing Load PRE."
This reverts commit r221924.  It appears the commit was a bit premature and is causing
bot failures that need further investigation.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@221939 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-13 22:54:59 +00:00
Chandler Carruth
a5408b9c7c [x86] Add some tests for specific patterns of lane-flips combined with
in-lane shuffles that aren't always handled well by the current vector
shuffle lowering.

No functionality change yet, that will follow in a subsequent commit.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@221938 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-13 22:49:44 +00:00
Rafael Espindola
83e56d6b47 Move calls to push_back out of readAbbreviated(Literal|Field).
These functions always return a single value and not all callers want to
push them into a SmallVector.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@221934 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-13 22:29:02 +00:00