MC: Move MCSectionData::Fixup out to MCAsmFixup.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95925 91177308-0d34-0410-b5e6-96231b3b80d8

make getFixupKindInfo return a const reference, allowing
the tables to be const.  Teach MCCodeEmitter to handle
the target-indep kinds so that we don't crash on them.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95924 91177308-0d34-0410-b5e6-96231b3b80d8

Revert functional change. This broke a bunch of tests.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95921 91177308-0d34-0410-b5e6-96231b3b80d8

switch to target-indep fixups for 1/2/4/8 byte data.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95920 91177308-0d34-0410-b5e6-96231b3b80d8

revert 95903.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95918 91177308-0d34-0410-b5e6-96231b3b80d8

It is always good to do a cross-class join when the large register has a tiny interval.

Also avoid division by zero.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95917 91177308-0d34-0410-b5e6-96231b3b80d8

Added LDRT/LDRBT/STRT/STRBT for disassembly only.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95916 91177308-0d34-0410-b5e6-96231b3b80d8

unbreak the build.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95915 91177308-0d34-0410-b5e6-96231b3b80d8

llvm-db was removed.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95904 91177308-0d34-0410-b5e6-96231b3b80d8

Destroy MDNodes while destructing llvm context.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95903 91177308-0d34-0410-b5e6-96231b3b80d8

refactor x86 conditional branches to use a multipattern
that generates the 1-byte and 4-byte immediate versions
from one definition.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95902 91177308-0d34-0410-b5e6-96231b3b80d8

refactor the conditional jump instructions in the .td file to
use a multipattern that generates both the 1-byte and 4-byte
versions from the same defm

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95901 91177308-0d34-0410-b5e6-96231b3b80d8

Make Kaleidoscope not link against the interpreter, since that didn't
work anyway (Interpreter::getPointerToFunction doesn't return a
callable pointer), and improve the error message when an
ExecutionEngine can't be created.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95896 91177308-0d34-0410-b5e6-96231b3b80d8

Add an svn:ignore.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95895 91177308-0d34-0410-b5e6-96231b3b80d8

Forgot to also check in this file for vcvt (floating-point <-> fixed-point, VFP).
Sorry!

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95892 91177308-0d34-0410-b5e6-96231b3b80d8

Allow for more than one DBG_VALUE targeting the
same dead instruction.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95890 91177308-0d34-0410-b5e6-96231b3b80d8

Don't allow DBG_VALUE to affect codegen.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95889 91177308-0d34-0410-b5e6-96231b3b80d8

Added VCVT (between floating-point and fixed-point, VFP) for disassembly.
A8.6.297

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95885 91177308-0d34-0410-b5e6-96231b3b80d8

Added BKPT/tBKPT (breakpoint) to the instruction table for disassembly purpose.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95884 91177308-0d34-0410-b5e6-96231b3b80d8

Use array_pod_sort instead of std::sort for improved code size.

Use SmallVector instead of std::vector for better speed when indirectbr has
few successors.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95879 91177308-0d34-0410-b5e6-96231b3b80d8

Make sure that ConstantExpr offsets also aren't off of extern
symbols.

Thanks to Duncan Sands for the testcase!

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95877 91177308-0d34-0410-b5e6-96231b3b80d8

Add pseudo instruction TRAP for disassembly, which is encoded according to A5-21
as the "Permanently UNDEFINED" instruction.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95873 91177308-0d34-0410-b5e6-96231b3b80d8

Use .empty() instead of .size().

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95871 91177308-0d34-0410-b5e6-96231b3b80d8

dont' call getX86RegNum on X86::RIP, it doesn't like that.  This
fixes the remaining x86-64 jit failures afaik.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95867 91177308-0d34-0410-b5e6-96231b3b80d8

fix a really nasty bug I introduced in r95693: r12 (and r12d,
r12b, etc) also encodes to a R/M value of 4, which is just
as illegal as ESP/RSP for the non-sib version an address.

This fixes x86-64 jit miscompilations of a bunch of programs.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95866 91177308-0d34-0410-b5e6-96231b3b80d8

Fix (harmless) memory leak found by memcheck.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95862 91177308-0d34-0410-b5e6-96231b3b80d8

Add and commonize encoder support for all immediates.
Stub out some dummy fixups to make things work.

We can now emit fixups like this:
subl $20, %esp               ## encoding: [0x83,0xec,A]
                                        ##   fixup A - offset: 2, value: 20, kind: fixup_1byte_imm

Emitting $20 as a single-byte fixup to be later resolved
by the assembler is ridiculous of course (vs just emitting
the byte) but this is a failure of the matcher, which
should be producing an imm of 20, not an MCExpr of 20.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95860 91177308-0d34-0410-b5e6-96231b3b80d8

generalize EmitDisplacementField to work with any size
and rename it to EmitImmediate.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95859 91177308-0d34-0410-b5e6-96231b3b80d8

eliminate the dead IsPCRel argument.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95858 91177308-0d34-0410-b5e6-96231b3b80d8

eliminate the dead "PCAdj" logic.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95857 91177308-0d34-0410-b5e6-96231b3b80d8

Fix some of the memcheck errors found in the JIT unittests.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95856 91177308-0d34-0410-b5e6-96231b3b80d8

Rename ValueRequiresCast to ShouldOptimizeCast, to better reflect
what it does.  Enhance it to return false to optimizing vector
sign extensions from vector comparisions, which is the idiom used
to get a splatted vector for a vector comparison.

Doing this breaks vector-casts.ll, add some compensating
transformations to handle the important case they cover without
depending on this canonicalization.

This fixes rdar://7434900 a serious pessimization of vector compares.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95855 91177308-0d34-0410-b5e6-96231b3b80d8

convert to filecheck.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95854 91177308-0d34-0410-b5e6-96231b3b80d8

Make DSE only scan blocks that are reachable from the entry
block.  Other blocks may have pointer cycles that will crash
basicaa and other alias analyses.  In any case, there is no
point wasting cycles optimizing dead blocks.  This fixes
rdar://7635088

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95852 91177308-0d34-0410-b5e6-96231b3b80d8

a testcase that doesn't crash GVN but could someday.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95851 91177308-0d34-0410-b5e6-96231b3b80d8

Make jump threading honor x|undef -> true and x&undef -> false,
instead of considering x|undef -> x, which may not be true.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95850 91177308-0d34-0410-b5e6-96231b3b80d8

Add ConstantExpr handling to Intrinsic::objectsize lowering.

Update testcase accordingly now that we can optimize another
section.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95846 91177308-0d34-0410-b5e6-96231b3b80d8

test case for r95842.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95844 91177308-0d34-0410-b5e6-96231b3b80d8

Fix to get it to compile.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95840 91177308-0d34-0410-b5e6-96231b3b80d8

Don't print out a default newline when emitting the section offset. There are
almost always comments afterwards that need printing.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95839 91177308-0d34-0410-b5e6-96231b3b80d8

Make it possible to create multiple JIT instances at the same time, by removing
the global TheJIT and TheJITResolver variables.  Lazy compilation is supported
by a global map from a stub address to the JITResolver that knows how to
compile it.

Patch by Olivier Meurant!

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95837 91177308-0d34-0410-b5e6-96231b3b80d8

Reuse operand location when updating PHI instructions.

Calling RemoveOperand is very expensive on huge PHI instructions. This makes
early tail duplication run twice as fast on the Firefox JavaScript
interpreter.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95832 91177308-0d34-0410-b5e6-96231b3b80d8

Remove duplicate successors from indirectbr instructions before building the machine CFG.

This makes early tail duplication run 60 times faster when compiling the Firefox
JavaScript interpreter, see PR6186.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95831 91177308-0d34-0410-b5e6-96231b3b80d8

Ignore dbg info intrinsics.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95828 91177308-0d34-0410-b5e6-96231b3b80d8

Remove the few # TAILCALL comments that snuck in.  As they may fail on linux.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95827 91177308-0d34-0410-b5e6-96231b3b80d8

Update the X86 assembler matcher test case now that a few more things match
with some of the recent changes that have gone into llvm-mc.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95826 91177308-0d34-0410-b5e6-96231b3b80d8

Add support to llvm-extract for extracting multiple functions and/or
multiple global variables at a time.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95825 91177308-0d34-0410-b5e6-96231b3b80d8

The previous fix of widening divides that trap was too fragile as it depends on custom
lowering and requires that certain types exist in ValueTypes.h.  Modified widening to
check if an op can trap and if so, the widening algorithm will apply only the op on
the defined elements.  It is safer to do this in widening because the optimizer can't
guarantee removing unused ops in some cases.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95823 91177308-0d34-0410-b5e6-96231b3b80d8

Ignore debug info one more place during coalescing.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95819 91177308-0d34-0410-b5e6-96231b3b80d8

Allow isDebug inquiry on any MO.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95818 91177308-0d34-0410-b5e6-96231b3b80d8

Delete dead PHI machine instructions.  These can be created due to type
legalization even when the IR-level optimizer has removed dead phis, such
as when the high half of an i64 value is unused on a 32-bit target.
I had to adjust a few test cases that had dead phis.
This is a partial fix for Radar 7627077.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95816 91177308-0d34-0410-b5e6-96231b3b80d8

Skip debug info in a couple of places.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95814 91177308-0d34-0410-b5e6-96231b3b80d8

Use an index instead of pointers into the vector. If the vector resizes, then
the pointer values could be invalid.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95813 91177308-0d34-0410-b5e6-96231b3b80d8

When I rewrote this loop per Chris' preference I
changed its behavior.  Oops.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95811 91177308-0d34-0410-b5e6-96231b3b80d8

add a virtual dtor to MCTargetExpr, hopefully silencing some warnings.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95810 91177308-0d34-0410-b5e6-96231b3b80d8

A few missed optimizations; the last one could have a significant impact on
code with lots of bitfields.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95809 91177308-0d34-0410-b5e6-96231b3b80d8

work around a gcc bug with -Wuninitialized.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95808 91177308-0d34-0410-b5e6-96231b3b80d8

Strip new llvm.dbg.value intrinsic.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95807 91177308-0d34-0410-b5e6-96231b3b80d8

MC/X86 AsmMatcher: Fix a use after free spotted by d0k, and de-XFAIL
x86_32-encoding.s in on expectation of it passing.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95806 91177308-0d34-0410-b5e6-96231b3b80d8

XFAIL this on linux until I figure out what is happening.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95804 91177308-0d34-0410-b5e6-96231b3b80d8

lit: Ignore dot files when scanning for tests (e.g., editor temprary files,
etc.)

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95803 91177308-0d34-0410-b5e6-96231b3b80d8

MC/AsmMatcher: Tweak conversion function name.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95802 91177308-0d34-0410-b5e6-96231b3b80d8

Minor whitespace cleanups.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95801 91177308-0d34-0410-b5e6-96231b3b80d8

Use an AssemblyAnnotatorWriter to clean up IVUsers' debug output.
The "uses=" comments are just clutter in this context.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95799 91177308-0d34-0410-b5e6-96231b3b80d8

Add a hook to AssemblyAnnotationWriter to allow custom info comments
to be printed, in place of the familiar "uses=" comments.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95798 91177308-0d34-0410-b5e6-96231b3b80d8

Use doxygen comment syntax.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95797 91177308-0d34-0410-b5e6-96231b3b80d8

Fix several comments which had previously been "the the" where a
different word was intended.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95795 91177308-0d34-0410-b5e6-96231b3b80d8

Replace this file containing 4 tests of x86 32-bit encodings with a file
containing the subset of the full auto generated test case that currently
encodes correctly.  Again it is useful as we bring up the the new encoder
to make sure currently working stuff stays working.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95791 91177308-0d34-0410-b5e6-96231b3b80d8

Added NOP, DBG, SVC to the instruction table for disassembly purpose.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95784 91177308-0d34-0410-b5e6-96231b3b80d8

Fix "the the" and similar typos.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95781 91177308-0d34-0410-b5e6-96231b3b80d8

Minor code simplification.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95780 91177308-0d34-0410-b5e6-96231b3b80d8

Silence GCC warnings.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95779 91177308-0d34-0410-b5e6-96231b3b80d8

MC/AsmMatcher: Add support for creating tied operands when constructing MCInsts.
 - Pretty messy, but we need to rework how we handle tied operands in MCInst
   anyway.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95774 91177308-0d34-0410-b5e6-96231b3b80d8

emit some simple (and probably incorrect) fixups for symbolic
displacement values.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95773 91177308-0d34-0410-b5e6-96231b3b80d8

keep track of what the current byte being emitted is
throughout the X86 encoder.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95771 91177308-0d34-0410-b5e6-96231b3b80d8

simplify displacement handling, emit displacements by-operand
even for the immediate case.  No functionality change.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95770 91177308-0d34-0410-b5e6-96231b3b80d8

Canonicalize sizeof and alignof on pointer types to a canonical
pointer type.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95769 91177308-0d34-0410-b5e6-96231b3b80d8

Implement operators |=, &=, and ^= for SmallBitVector, and remove the
restriction in BitVector for |= and ^= that the operand must be the
same length.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95768 91177308-0d34-0410-b5e6-96231b3b80d8

MC: Switch MCFixup to just hold an MCExpr pointer instead of index into the
MCInst it came from.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95767 91177308-0d34-0410-b5e6-96231b3b80d8

Fix a signed comparison warning.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95766 91177308-0d34-0410-b5e6-96231b3b80d8

Remove stray DOS newline.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95765 91177308-0d34-0410-b5e6-96231b3b80d8

Add a ReleaseNotes FIXME.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95764 91177308-0d34-0410-b5e6-96231b3b80d8

Prevented build on WINDOWS using default make system. Stopped WINDOWS build
at eh llvm/examples level using if check on LLVM_ON_UNIX.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95763 91177308-0d34-0410-b5e6-96231b3b80d8

Updated the enhanced disassembly library's TableGen
backend to not use exceptions at all except in cases
of actual error.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95762 91177308-0d34-0410-b5e6-96231b3b80d8

Prevented ExceptionDemo example being built on WINDOWS via if( NOT WIN32 )
check in examples cmake list file. This has NOT been tested.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95761 91177308-0d34-0410-b5e6-96231b3b80d8

Updated the TableGen emitter for the Enhanced
Disassembler to take advantage of the refactored
AsmWriterInst.h.  Note removed parser code.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95760 91177308-0d34-0410-b5e6-96231b3b80d8

Changed AsmWriterOperand to also include the index of the
operand into the CodeGenInstruction's list of operands,
which is useful for EDEmitter.  (Still working on PR6219)

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95759 91177308-0d34-0410-b5e6-96231b3b80d8

Now that ShrinkDemandedOps() is separated out from DAG combine. It sometimes leave some obvious nops which dag combine used to clean up afterwards e.g. (trunk (ext n)) -> n. Look for them and squash them.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95757 91177308-0d34-0410-b5e6-96231b3b80d8

"fixup" a comment.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95754 91177308-0d34-0410-b5e6-96231b3b80d8

Introduce a new CodeGenInstruction::ConstraintInfo class
for representing constraint info semantically instead of
as a c expression that will be blatted out to the .inc
file.  Fix X86RecognizableInstr to use this instead of
parsing C code :).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95753 91177308-0d34-0410-b5e6-96231b3b80d8

llvm-mc: Remove --show-fixups and always show as part of --show-encoding.

Also, fix a silly memory leak.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95752 91177308-0d34-0410-b5e6-96231b3b80d8

Rewrite loop to suit Chris' preference.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95749 91177308-0d34-0410-b5e6-96231b3b80d8

fix a layering violation: VirtRegRewriter.cpp shouldn't use AsmPrinter.h.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95748 91177308-0d34-0410-b5e6-96231b3b80d8

Remove duplicated #include.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95747 91177308-0d34-0410-b5e6-96231b3b80d8

Emit an error for illegal inline asm constraint (which uses illegal type) rather than asserting.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95746 91177308-0d34-0410-b5e6-96231b3b80d8

fix missing #includes.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95745 91177308-0d34-0410-b5e6-96231b3b80d8

daniel *really* likes fixups!

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95742 91177308-0d34-0410-b5e6-96231b3b80d8

Stop MachineInstr.h from #including AsmPrinter.h

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95741 91177308-0d34-0410-b5e6-96231b3b80d8

Improve comments a even more.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95740 91177308-0d34-0410-b5e6-96231b3b80d8

Skip DBG_VALUE many places in live intervals and
register coalescing.  This fixes many crashes and
places where debug info affects codegen (when
dbg.value is lowered to machine instructions, which
it isn't yet in TOT).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95739 91177308-0d34-0410-b5e6-96231b3b80d8