RPCS3/llvm-mirror
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-12-19 09:36:44 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity
68,336 Commits 4 Branches 5 Tags 634 MiB
e92f8121d4
Commit Graph

44204 Commits

Author SHA1 Message Date
Chris Lattner
e92f8121d4 improve switch formation to handle small range 
comparisons formed by comparisons.  For example,
this:

void foo(unsigned x) {
  if (x == 0 || x == 1 || x == 3 || x == 4 || x == 6) 
    bar();
}

compiles into:

_foo:                                   ## @foo
## BB#0:                                ## %entry
	cmpl	$6, %edi
	ja	LBB0_2
## BB#1:                                ## %entry
	movl	%edi, %eax
	movl	$91, %ecx
	btq	%rax, %rcx
	jb	LBB0_3

instead of:

_foo:                                   ## @foo
## BB#0:                                ## %entry
	cmpl	$2, %edi
	jb	LBB0_4
## BB#1:                                ## %switch.early.test
	cmpl	$6, %edi
	ja	LBB0_3
## BB#2:                                ## %switch.early.test
	movl	%edi, %eax
	movl	$88, %ecx
	btq	%rax, %rcx
	jb	LBB0_4

This catches a bunch of cases in GCC, which look like this:

 %804 = load i32* @which_alternative, align 4, !tbaa !0
 %805 = icmp ult i32 %804, 2
 %806 = icmp eq i32 %804, 3
 %or.cond121 = or i1 %805, %806
 %807 = icmp eq i32 %804, 4
 %or.cond124 = or i1 %or.cond121, %807
 br i1 %or.cond124, label %.thread, label %808

turning this into a range comparison.

llvm-svn: 122045
 2010-12-17 06:20:15 +00:00
Daniel Dunbar
eea7fd1c3e MC/ARM: Use aggressive symbol folding (important for jump tables, for example). 
llvm-svn: 122044
 2010-12-17 06:00:24 +00:00
Daniel Dunbar
1f9fd0b79b MC/Expr: Implemnt more aggressive folding during symbol evaluation using 
IsSymbolRefDifferenceFullyResolved(). For example, we will now fold away
something like:
--
_a:
...
L0:
...
L1:
...
.long (L1 - L0) / 2
--

llvm-svn: 122043
 2010-12-17 05:50:33 +00:00
Daniel Dunbar
6979fc1057 MC/Mach-O: On second thought, use a custom hook for enabling aggressive 
IsSymbolRefDifferenceFullyResolved, it turns out this does change behavior on
enough cases for x86-32 that I would rather wait a bit on it.
 - In practice, we will want to change this eventually because it only means we
   generate less relocations (it also eliminates the need for the horrible
   '.set' hack that Darwin requires in some places).

llvm-svn: 122042
 2010-12-17 05:50:29 +00:00
Daniel Dunbar
14208868a4 MC/Mach-O: Implement IsSymbolRefDifferenceFullyResolved. 
- Unlike for fixups, we always do the "reliable" thing (not just for x86_64).
 - Since Darwin 'as' would typically reject things that using this will allow,
   we don't need to worry about compatibility.

llvm-svn: 122038
 2010-12-17 04:54:58 +00:00
Daniel Dunbar
2caf23b5a9 MC/ObjectWriter: Add a new IsSymbolRefDifferenceFullyResolved target format specific hook. 
- Currently just has stub implementations for Mach-O, ELF, and COFF.

llvm-svn: 122037
 2010-12-17 04:54:54 +00:00
Jakob Stoklund Olesen
40f23cd5ca Provide LiveIntervalUnion::Query::checkLoopInterference. 
This is a three-way interval list intersection between a virtual register, a
live interval union, and a loop. It will be used to identify interference-free
loops for live range splitting.

llvm-svn: 122034
 2010-12-17 04:09:47 +00:00
Daniel Dunbar
f77af712b7 MC/Assembler: Strip out object writer arguments, now that it is always available 
-- and remove FIXME asking for the same!

llvm-svn: 122032
 2010-12-17 02:45:59 +00:00
Daniel Dunbar
54b3f3e5db MC/Assembler: Make the MCObjectWriter available through the lifetime of the 
assembler.

llvm-svn: 122031
 2010-12-17 02:45:41 +00:00
Jim Grosbach
3bb926adff Trailing whitespace and 80 column fixups. 
llvm-svn: 122026
 2010-12-17 02:10:59 +00:00
Daniel Dunbar
6d4a3e98f3 MC/Target: Remove HasScatteredSymbols target hook variable, which has been 
superceded and was effectively dead.

llvm-svn: 122024
 2010-12-17 02:06:08 +00:00
Daniel Dunbar
0d8fdf42c7 MC/Expr: Simplify. 
llvm-svn: 122023
 2010-12-17 02:05:45 +00:00
Bob Wilson
e06f6eabe7 Fix crash compiling a QQQQ REG_SEQUENCE for a Neon vld3_lane operation. 
Radar 8776599

llvm-svn: 122018
 2010-12-17 01:21:12 +00:00
Bob Wilson
125f1a26e0 Use PairDRegs to implement ConcatVectors. No functionality change. 
llvm-svn: 122017
 2010-12-17 01:21:08 +00:00
Bob Wilson
f6fe49f7d2 Fix a comment typo. 
llvm-svn: 122016
 2010-12-17 01:21:05 +00:00
Dan Gohman
f8949c3d1a Revert r64460. strtol and friends cannot be marked readonly, even with 
a null endptr argument, because they may write to errno.

This fixes a seflhost miscompile observed on Linux targets when TBAA
was enabled.

llvm-svn: 122014
 2010-12-17 01:09:43 +00:00
Daniel Dunbar
565c90b316 MC: Remove another dead MCAssembler argument, and update clients. 
llvm-svn: 122013
 2010-12-17 01:07:35 +00:00
Daniel Dunbar
9f83b4d605 MC: Remove dead MCAssembler argument -- Rafael, can you check the FIXME I added 
here?

llvm-svn: 122012
 2010-12-17 01:07:31 +00:00
Daniel Dunbar
17e37880d3 MC: Simplify (remove unnecessary MCAssembler argument, obsoleted by containment 
in MCAsmLayout).

llvm-svn: 122011
 2010-12-17 01:07:28 +00:00
Daniel Dunbar
ed4c226521 Write => in a more normal form. 
llvm-svn: 122009
 2010-12-17 01:07:22 +00:00
Daniel Dunbar
2a0a64029a MC/Expr: Simplify (and add a FIXME). 
llvm-svn: 122008
 2010-12-17 01:07:20 +00:00
Rafael Espindola
bd13ceed72 "Fix" FDE alignment to match what gas does. 
llvm-svn: 122006
 2010-12-17 00:28:02 +00:00
Rafael Espindola
3ee4530406 Make pushq produce signed relocations. 
llvm-svn: 122005
 2010-12-16 22:50:01 +00:00
Wesley Peck
069e991659 Fix MBlaze backend call instructions so that arguments passed through registers 
are correctly marked as used. This removes a hack where the call instructions
marked all possible argument registers as used in the tablegen description.

llvm-svn: 121994
 2010-12-16 19:41:31 +00:00
Jim Grosbach
ce773319e4 Pseudo-ize the Thumb1 tBfar pattern. rdar://8777974 
llvm-svn: 121990
 2010-12-16 19:11:16 +00:00
Daniel Dunbar
471378a4ab MC/Expr: Add a doxyment. 
llvm-svn: 121988
 2010-12-16 18:36:25 +00:00
Daniel Dunbar
2277f06687 MC/Mach-O: Lift some MachObjectWriter arguments into the target specific 
interface.

llvm-svn: 121981
 2010-12-16 17:21:02 +00:00
Daniel Dunbar
5fee4f6862 MC/Mach-O: Stub out explicit MCMachObjectTargetWriter interface. 
llvm-svn: 121973
 2010-12-16 16:09:19 +00:00
Daniel Dunbar
93dc10ffcb Fix indentation (per style guide). 
llvm-svn: 121972
 2010-12-16 16:08:43 +00:00
Daniel Dunbar
636e690f58 MC/Mach-O: Move createMachObjectWriter into MCMachObjectWriter.h. 
llvm-svn: 121971
 2010-12-16 16:08:33 +00:00
Daniel Dunbar
af109884b9 MC/Mach-O: Use fixup info instead of hard coded list. 
llvm-svn: 121970
 2010-12-16 15:42:31 +00:00
Frits van Bommel
56dc4fc838 Fix a bug in the loop in JumpThreading::ProcessThreadableEdges() where it could falsely produce a MultipleDestSentinel value if the first predecessor ended with an 'indirectbr'. If that happened, it caused an unnecessary FindMostPopularDest() call. 
This wasn't a correctness problem, but it broke the fast path for single-predecessor blocks.

llvm-svn: 121966
 2010-12-16 12:16:00 +00:00
Duncan Sands
22de496ae3 Speculatively revert commit 121905 since it looks like it might have broken the 
dragonegg self-host buildbot.  Original commit message:

Add an InstCombine transform to recognize instances of manual overflow-safe addition
(performing the addition in a wider type and explicitly checking for overflow), and
fold them down to intrinsics.  This currently only supports signed-addition, but could
be generalized if someone works out the magic constant formulas for other operations.

llvm-svn: 121965
 2010-12-16 09:40:54 +00:00
Michael J. Spencer
86f6a9ac6e MemoryBuffer now return an error_code and returns a OwningPtr<MemoryBuffer> via an out parm. 
llvm-svn: 121958
 2010-12-16 03:29:14 +00:00
Daniel Dunbar
fc14f1e5f6 MC: Move target specific fixup info descriptors to TargetAsmBackend instead of 
the MCCodeEmitter, which seems like a better organization.
 - Also, cleaned up some magic constants while in the area.

llvm-svn: 121953
 2010-12-16 03:20:06 +00:00
Jason W Kim
2f4a8d7553 1. ARM/MC/ELF: A few more ELF relocs for .o 
2. Fixed EmitLocalCommonSymbol for ELF (Yes, they exist. :)
   Test added.

llvm-svn: 121951
 2010-12-16 03:12:17 +00:00
Daniel Dunbar
8ab9be2005 MC: Make TargetAsmBackend available to the AsmStreamer. 
- Treaty talks on the non-proliferation of MC objects broke down.

llvm-svn: 121949
 2010-12-16 03:05:59 +00:00
Dan Gohman
8540450279 Update a comment. 
llvm-svn: 121946
 2010-12-16 02:55:10 +00:00
Dan Gohman
e106936414 Make memcpyopt TBAA-aware. 
llvm-svn: 121944
 2010-12-16 02:51:19 +00:00
NAKAMURA Takumi
b542a59c37 lib/Support/regexec.c: Let Regex LLP64-aware. 
On LLP64 Win64, 'states1' (for small version) was expanded to 'char *' for large version. Thus small version would be mischosen when nstates > 32 regardless of sizeof(long) on Win64.

llvm-svn: 121942
 2010-12-16 01:48:15 +00:00
Matt Beaumont-Gay
9b5939da47 Delete an extra "Imm5 = ", caught by GCC's -Wsequence-point but not by Clang 
(see PR4579).

llvm-svn: 121939
 2010-12-16 01:34:26 +00:00
Bill Wendling
c22ad4d7ae Remove fixup_arm_thumb_ldst. The code was never calling the "fixup" stuff for 
it. I.e., it was always an immediate value.

llvm-svn: 121932
 2010-12-16 00:50:33 +00:00
Bill Wendling
6ce0807502 Add tSpill and tRestore to the opcodes to replace with tSTRi and tLDRi 
respectively.

It may be a bug that these opcodes are getting this far into machine code
generation.

llvm-svn: 121931
 2010-12-16 00:49:54 +00:00
Bill Wendling
448cd7f1bb Add encodings for Thumb1 Spill and Restore pseudos. 
llvm-svn: 121929
 2010-12-16 00:38:41 +00:00
Dan Gohman
b9583fff47 Enable TBAA by default. 
llvm-svn: 121923
 2010-12-15 23:58:44 +00:00
Dan Gohman
a2fd4f2e22 Preserve TBAA tags when doing load PRE. 
llvm-svn: 121921
 2010-12-15 23:53:55 +00:00
Jim Grosbach
84c2b29b58 Thumb1 had two patterns for the same load-from-constant-pool instruction. 
Canonicalize on tLDRpci and remove tLDRcp.

llvm-svn: 121920
 2010-12-15 23:52:36 +00:00
Eric Christopher
339499f8f3 Don't handle -arm-long-calls in fast isel for now. 
llvm-svn: 121919
 2010-12-15 23:47:29 +00:00
Jakob Stoklund Olesen
1811e4cb20 Start using SplitKit and MachineLoopRanges in RegAllocGreedy in preparation of 
live range splitting around loops guided by register pressure.

So far, trySplit() simply prints a lot of debug output.

llvm-svn: 121918
 2010-12-15 23:46:13 +00:00
Jakob Stoklund Olesen
d40af5ffbd Add MachineLoopRanges analysis. 
A MachineLoopRange contains the intervals of slot indexes covered by the blocks
in a loop. This representation of the loop blocks is more efficient to compare
against interfering registers during register coalescing.

llvm-svn: 121917
 2010-12-15 23:41:23 +00:00