Commit Graph

53011 Commits

Author SHA1 Message Date
Matt Beaumont-Gay
0310038d5a Sink variable into assert
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150841 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-17 21:40:48 +00:00
Lang Hames
fbc8dd306a Add support for regmask slots to HMEditor. Also fixes a comment error.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150840 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-17 21:29:41 +00:00
Kevin Enderby
d9165eb02f Fix typo in comment ldopen() -> dlopen().
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150836 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-17 19:26:00 +00:00
Jakob Stoklund Olesen
2420b558de Handle regmask operands in ARMInstrInfo.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150833 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-17 19:23:15 +00:00
Kevin Enderby
fc0d7400bc Put back the initializing the targets in the disassembler API with a comment as
to why this is needed.  This broke the darwin's otool(1) program.  This change
was made in r144385.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150832 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-17 19:18:29 +00:00
Jakob Stoklund Olesen
8c3b87cf19 Fix ARMBaseInstrInfo::getInstrLatency for calls.
Calls always clobber CPSR.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150831 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-17 19:07:59 +00:00
Jakob Stoklund Olesen
d9f0ff56a1 Transfer regmasks to MRI.
MRI keeps track of which physregs have been used. Make sure it gets
updated with all the regmask-clobbered registers.

Delete the closePhysRegsUsed() function which isn't necessary.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150830 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-17 19:07:56 +00:00
Dan Gohman
dbe266be35 Calls and invokes with the new clang.arc.no_objc_arc_exceptions
metadata may still unwind, but only in ways that the ARC
optimizer doesn't need to consider. This permits more
aggressive optimization.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150829 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-17 18:59:53 +00:00
Lang Hames
3dc7c5138d Refactor 'handleMove' code in live intervals. Clients of LiveIntervals won't see
any changes.

Internally this adds a private inner class HMEditor, to LiveIntervals. HMEditor provides
an API for updating live intervals when code is moved or bundled.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150826 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-17 18:44:18 +00:00
Dan Gohman
ebad58dc58 Remove a comment about an alternative approach that wouldn't
actually work, at least as described. LLVM Metadata is not
intended to suppress LLVM IR rules, as it can be stripped at
any time.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150821 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-17 18:33:38 +00:00
Jim Grosbach
7842a741eb Tidy up.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150820 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-17 17:35:10 +00:00
David Chisnall
4cbcee1618 Generate the correct EH frame section types on Solaris, this time without breaking other platforms...
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150819 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-17 17:31:15 +00:00
David Chisnall
8bb51ef6d2 Revert r150814. It turns out that there is a good reason for this after all...
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150818 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-17 16:51:02 +00:00
Jakob Stoklund Olesen
0f7de3542f Revert r150288, "Allow Post-RA LICM to hoist reserved register reads."
This caused miscompilations on out-of-tree targets, and possibly i386 as
well.

I'll find some other way of hoisting %rip-relative loads from loops
containing calls.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150816 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-17 16:40:44 +00:00
David Chisnall
637b25a6a0 Don't lazily allocate eh_frame. We're not lazily allocating things like the LSDA, which are only used when the eh frame is used, so this lazy allocation doesn't really make sense.
Fix the type of eh_frame on Solaris so that Sun ld doesn't fail to combine them (thus making it impossible for the unwind library to find them and breaking exceptions).



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150814 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-17 16:32:07 +00:00
David Chisnall
50f603fff6 ... and it's probably best to use the correct alignment, rather than just guessing that it's the same as the size.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150813 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-17 16:30:39 +00:00
David Chisnall
ca5b752d18 It turns out that putting an 8-byte symbol in a 4-byte section makes Solaris ld sulk. GNU ld is perfectly happy with it, which is worrying for a whole other set of reasons...
Thanks to Anton, Duncan and Rafael for helping me track this down.
Pointy hat to Rafael for introducing the bug in the first place.




git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150811 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-17 16:05:50 +00:00
Jia Liu
c5707112e7 remove Emacs-tag form .cpp files in Mips Backend, and fix some typo.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150805 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-17 08:55:11 +00:00
Eric Christopher
ed993de642 Typo in variable name.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150796 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-17 07:08:46 +00:00
Craig Topper
6e94e68b75 Remove the last of the old vector_shuffle patterns from X86 isel.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150795 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-17 07:02:34 +00:00
Nick Lewycky
81266c5c93 Add support for invariant.start inside the static constructor evaluator. This is
useful to represent a variable that is const in the source but can't be constant
in the IR because of a non-trivial constructor. If globalopt evaluates the
constructor, and there was an invariant.start with no matching invariant.end
possible, it will mark the global constant afterwards.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150794 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-17 06:59:21 +00:00
Akira Hatanaka
38bdc5762f Do not promote i32 arguments to i64. This was causing unnecessary sign extension
instructions to be emitted.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150782 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-17 02:20:26 +00:00
Bill Wendling
3106b8b18b Remove redundant comment. Use a more efficient datatype.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150780 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-17 02:12:54 +00:00
Bill Wendling
1bc147b892 Fix some grammar-os and formatting.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150779 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-17 02:09:28 +00:00
Lang Hames
00195d828b Reverse iterator - should be incrementing rather than decrementing.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150778 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-17 01:54:11 +00:00
Jia Liu
8f5e8c1cd6 add Emacs tag and fix some comment error in file headers
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150775 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-17 01:23:50 +00:00
Chad Rosier
b8703fe265 [fast-isel] Add support for returning non-legal types with no sign- or zero-
entend flag.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150774 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-17 01:21:28 +00:00
Lang Hames
29879d7f86 MachineScheduler shouldn't use/preserve LiveDebugVariables.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150773 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-17 01:11:37 +00:00
Lang Hames
5d0827582f Oops - isRegLiveIntoSuccessor is used in non-assert builds now. Remove NDEBUG guards.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150771 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-17 00:51:32 +00:00
Lang Hames
c2e08db4e5 Re-enable 150652 and 150654 - Make FPSCR non-reserved, and make MachineCSE bail on reserved registers. This *should* be safe as of r150786.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150769 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-17 00:27:16 +00:00
Lang Hames
af8b34dae9 Turn off assertion, conservatively compute liveness for live-in un-allocatable registers.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150768 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-17 00:18:18 +00:00
Akira Hatanaka
178d870795 Remove comment.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150739 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-16 22:52:29 +00:00
Chad Rosier
a69feb0f33 Remove unnecessary assignment to temporary, ResultReg.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150737 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-16 22:45:33 +00:00
Jakob Stoklund Olesen
8b0a3f2df9 Remove the YMM_HI_6_15 hack.
Call clobbers are now represented with register mask operands.  The
regmask can easily represent the fact that xmm6 is call-preserved while
ymm6 isn't.  This is automatically computed by TableGen from the
CalleeSavedRegs containing xmm6.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150709 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-16 17:56:06 +00:00
Jakob Stoklund Olesen
527a08b253 Use the same CALL instructions for Windows as for everything else.
The different calling conventions and call-preserved registers are
represented with regmask operands that are added dynamically.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150708 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-16 17:56:02 +00:00
Akira Hatanaka
e3186774f3 Remove trailing whitespace. Add newline.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150706 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-16 17:48:20 +00:00
Benjamin Kramer
1a7ea9dd04 Disable machine copy propagation for now. It's known to be buggy (PR11940) and introduces subtle miscompiles in many places.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150703 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-16 17:29:50 +00:00
Benjamin Kramer
ea79b8e03c Revert "InstSimplify: Strip pointer casts early."
Turns out this isn't safe, because the code below depends on LHS and RHS having
the same type.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150695 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-16 15:19:59 +00:00
Benjamin Kramer
475ebf5a74 InstSimplify: Strip pointer casts early.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150694 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-16 15:03:04 +00:00
Benjamin Kramer
fd8779a94b InstSimplify: Ignore pointer casts when constant folding compares between pointers.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150690 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-16 13:49:39 +00:00
Bill Wendling
e6bd7a8058 s/ModAttrBehavior/ModFlagBehavior/g to be consistent with how module flags are named elsewhere.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150679 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-16 10:28:10 +00:00
James Molloy
6afa3f744a Remove extraneous #include and spelling mistake introduced in r150669.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150670 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-16 09:48:07 +00:00
James Molloy
6660c05da3 Modify the algorithm when traversing the DAGCombiner's worklist to be O(log N) for all operations. This fixes a horrible worst case with lots of nodes where 99% of the time was being spent in std::remove.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150669 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-16 09:17:04 +00:00
NAKAMURA Takumi
22bed5db2f VMCore/AsmWriter.cpp: Tweak to check #INF and #NAN earlier.
With MSVCRT, prior checker missed emission of #INF and #NAN.

FIXME: Checking should be simpler.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150667 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-16 08:12:24 +00:00
NAKAMURA Takumi
c8782a1a53 VMCore/AsmWriter.cpp: Use APFloat instead of atof(3).
atof(3) might behave differently among platforms.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150661 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-16 04:19:15 +00:00
Lang Hames
1834df8384 Oop - r150653 + r150654 broke one of my test cases. Backing out for now...
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150655 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-16 02:32:10 +00:00
Lang Hames
afae28b1c6 FPSCR shouldn't be reserved.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150654 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-16 02:28:14 +00:00
Lang Hames
f7e31b9e81 MachineCSE shouldn't extend the live ranges of reserved or allocatable registers.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150653 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-16 02:19:35 +00:00
Eli Friedman
5e6162e756 loop-rotate shouldn't hoist alloca instructions out of a loop. Patch by Patrik Hägglund, with slightly modified test. Issue reported by Patrik Hägglund on llvmdev.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150642 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-16 00:41:10 +00:00
Jakob Stoklund Olesen
8bcde2aa66 Enable register mask operands for x86 calls.
Call instructions no longer have a list of 43 call-clobbered registers.
Instead, they get a single register mask operand with a bit vector of
call-preserved registers.

This saves a lot of memory, 42 x 32 bytes = 1344 bytes per call
instruction, and it speeds up building call instructions because those
43 imp-def operands no longer need to be added to use-def lists. (And
removed and shifted and re-added for every explicit call operand).

Passes like LiveVariables, LiveIntervals, RAGreedy, PEI, and
BranchFolding are significantly faster because they can deal with call
clobbers in bulk.

Overall, clang -O2 is between 0% and 8% faster, uniformly distributed
depending on call density in the compiled code.  Debug builds using
clang -O0 are 0% - 3% faster.

I have verified that this patch doesn't change the assembly generated
for the LLVM nightly test suite when building with -disable-copyprop
and -disable-branch-fold.

Branch folding behaves slightly differently in a few cases because call
instructions have different hash values now.

Copy propagation flushes its data structures when it crosses a register
mask operand. This causes it to leave a few dead copies behind, on the
order of 20 instruction across the entire nightly test suite, including
SPEC. Fixing this properly would require the pass to use different data
structures.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@150638 91177308-0d34-0410-b5e6-96231b3b80d8
2012-02-16 00:02:50 +00:00