Benjamin Kramer [Thu, 29 Mar 2012 12:37:26 +0000 (12:37 +0000)]
Replace assert(0) with llvm_unreachable to avoid warnings about dropping off the end of a non-void function in Release builds.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153643
91177308-0d34-0410-b5e6-
96231b3b80d8
Eric Christopher [Thu, 29 Mar 2012 08:42:56 +0000 (08:42 +0000)]
Add support for objc property decls according to the page at:
http://llvm.org/docs/SourceLevelDebugging.html#objcproperty
including type and DECL. Expand the metadata needed accordingly.
rdar://
11144023
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153639
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Thu, 29 Mar 2012 08:27:32 +0000 (08:27 +0000)]
Make some headway towards compiling all of LLVM.
Module-level ASM may contain definitions of functions and globals. However, we
were not telling the linker that these globals had definitions. As far as it was
concerned, they were just declarations.
Attempt to resolve this by inserting module-level ASM functions and globals into
the '_symbol' set so that the linker will know that they have values.
This gets us further towards our goal of compiling LLVM, but it still has
problems when linking libLTO.dylib because of the `-dead_strip' flag that's
passed to the linker.
<rdar://problem/
11124216>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153638
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Thu, 29 Mar 2012 07:11:23 +0000 (07:11 +0000)]
Only allow symbolic names for (v)cmpss/sd/ps/pd encodings 8-31 to be used with 'v' version of instructions.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153636
91177308-0d34-0410-b5e6-
96231b3b80d8
Joel Jones [Thu, 29 Mar 2012 05:45:48 +0000 (05:45 +0000)]
For X86, change load/dec-or-inc/store into dec-or-inc, respectively.
This is a code change to add support for changing instruction sequences of the form:
load
inc/dec of 8/16/32/64 bits
store
into the appropriate X86 inc/dec through memory instruction:
inc[qlwb] / dec[qlwb]
The checks that were in X86DAGToDAGISel::Select(SDNode *Node)>>ISD::STORE have been extracted to isLoadIncOrDecStore and reworked to use the better
named wrappers for getOperand(unsigned) (e.g. getOffset()) and replaced Chain.getNode() with LoadNode. The comments have also been expanded.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153635
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Thu, 29 Mar 2012 04:28:00 +0000 (04:28 +0000)]
Cleanup whitespace.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153634
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Thu, 29 Mar 2012 03:34:57 +0000 (03:34 +0000)]
Cache the end() iterator.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153632
91177308-0d34-0410-b5e6-
96231b3b80d8
Joel Jones [Thu, 29 Mar 2012 01:20:56 +0000 (01:20 +0000)]
Reverted to revision 153616 to unblock build
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153623
91177308-0d34-0410-b5e6-
96231b3b80d8
Joel Jones [Thu, 29 Mar 2012 00:37:47 +0000 (00:37 +0000)]
For X86, change load/dec-or-inc/store into dec-or-inc, respectively.
This is a code change to add support for changing instruction sequences of the form:
load
inc/dec of 8/16/32/64 bits
store
into the appropriate X86 inc/dec through memory instruction:
inc[qlwb] / dec[qlwb]
The checks that were in X86DAGToDAGISel::Select(SDNode *Node)>>ISD::STORE have been extracted to isLoadIncOrDecStore and reworked to use the better
named wrappers for getOperand(unsigned) (e.g. getOffset()) and replaced Chain.getNode() with LoadNode. The comments have also been expanded.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153617
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Wed, 28 Mar 2012 23:54:28 +0000 (23:54 +0000)]
Enable machine code verification in the entire code generator.
Some targets still mess up the liveness information, but that isn't
verified after MRI->invalidateLiveness().
The verifier can still check other useful things like register classes
and CFG, so it should be enabled after all passes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153615
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Wed, 28 Mar 2012 23:31:15 +0000 (23:31 +0000)]
Enable machine code verification after PreSched2 passes.
The late scheduler depends on accurate liveness information if it is
breaking anti-dependencies, so we should be able to verify it.
Relax the terminator checking in the machine code verifier so it can
handle the basic blocks created by if conversion.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153614
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Wed, 28 Mar 2012 23:12:18 +0000 (23:12 +0000)]
Cleanup some whitespaces.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153612
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Wed, 28 Mar 2012 23:07:03 +0000 (23:07 +0000)]
Don't kill the base register when expanding strd.
When an strd instruction doesn't get the registers it wants, it can be
expanded into two str instructions. Make sure the first str doesn't kill
the base register in the case where the base and data registers are
identical:
t2STRi12 %R0<kill>, %R0, 4, pred:14, pred:%noreg
t2STRi12 %R2<kill>, %R0, 8, pred:14, pred:%noreg
<rdar://problem/
11101911>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153611
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Wed, 28 Mar 2012 22:50:56 +0000 (22:50 +0000)]
Preserve implicit defs in ARMLoadStoreOptimizer.
When a number of sub-register VLRDS instructions are combined into a
VLDM, preserve any super-register implicit defs. This is required to
keep the register scavenger and machine code verifier happy.
Enable machine code verification after ARMLoadStoreOptimizer.
ARM/2012-01-26-CopyPropKills.ll was failing because of this.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153610
91177308-0d34-0410-b5e6-
96231b3b80d8
Jim Grosbach [Wed, 28 Mar 2012 22:34:41 +0000 (22:34 +0000)]
Tidy up. Whitespace.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153609
91177308-0d34-0410-b5e6-
96231b3b80d8
Danil Malyshev [Wed, 28 Mar 2012 21:46:36 +0000 (21:46 +0000)]
Move getPointerToNamedFunction() from JIT/MCJIT to JITMemoryManager.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153607
91177308-0d34-0410-b5e6-
96231b3b80d8
Rafael Espindola [Wed, 28 Mar 2012 21:31:24 +0000 (21:31 +0000)]
Handle intrinsics in GlobalsModRef. Fixes pr12351.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153604
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Wed, 28 Mar 2012 21:20:32 +0000 (21:20 +0000)]
Spill DPair registers, not just QPR.
The arm_neon intrinsics can create virtual registers from the DPair
register class which allows both even-odd and odd-even D-register pairs.
This fixes PR12389.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153603
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Wed, 28 Mar 2012 20:49:30 +0000 (20:49 +0000)]
Also verify after ExpandPostRAPseudos.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153599
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Wed, 28 Mar 2012 20:48:49 +0000 (20:48 +0000)]
Inline function into its one caller.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153598
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Wed, 28 Mar 2012 20:47:37 +0000 (20:47 +0000)]
Enable machine code verification after the late machine optimization passes.
Branch folding invalidates liveness and disables liveness verification
on some targets.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153597
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Wed, 28 Mar 2012 20:47:35 +0000 (20:47 +0000)]
Skip liveness verification when MRI->tracksLiveness() is false.
Extract the liveness verification into its own method.
This makes it possible to run the machine code verifier after liveness
information is no longer required to be valid.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153596
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Wed, 28 Mar 2012 20:46:54 +0000 (20:46 +0000)]
Reformat the LTOModule code to be more inline with LLVM's coding standards. Add
a bunch of comments for the various functions. No intended functionality change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153595
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Wed, 28 Mar 2012 20:11:44 +0000 (20:11 +0000)]
Revert r153516: "Invalidate liveness in Thumb2ITBlockPass."
Revert r153519: "ARMLoadStoreOptimizer invalidates register liveness."
These patches caused miscompilations in povray by turning off branch
folding's updating of live-in lists.
It turns out the the late scheduler depends on the live-in lists, even
if it doesn't need correct kill flags.
<rdar://problem/
11139228>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153593
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Wed, 28 Mar 2012 20:11:42 +0000 (20:11 +0000)]
Allow removeLiveIn to be called with a register that isn't live-in.
This avoids the silly double search:
if (isLiveIn(Reg))
removeLiveIn(Reg);
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153592
91177308-0d34-0410-b5e6-
96231b3b80d8
Chad Rosier [Wed, 28 Mar 2012 18:42:50 +0000 (18:42 +0000)]
Revert r153521 as it's causing large regressions on the nightly testers.
Original commit message for r153521 (aka r153423):
Use the new range metadata in computeMaskedBits and add a new optimization to
instruction simplify that lets us remove an and when loding a boolean value.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153587
91177308-0d34-0410-b5e6-
96231b3b80d8
Pete Cooper [Wed, 28 Mar 2012 17:02:22 +0000 (17:02 +0000)]
Fixed commuteInstructions bug where if its called pre-regalloc the subreg indices weren't commuted
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153579
91177308-0d34-0410-b5e6-
96231b3b80d8
Benjamin Kramer [Wed, 28 Mar 2012 14:50:09 +0000 (14:50 +0000)]
GlobalOpt: If we have an inbounds GEP from a ConstantAggregateZero global that we just determined to be constant, replace all loads from it with a zero value.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153576
91177308-0d34-0410-b5e6-
96231b3b80d8
Benjamin Kramer [Wed, 28 Mar 2012 10:50:18 +0000 (10:50 +0000)]
Add another note about a missed compare with nsw arithmetic instcombine.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153574
91177308-0d34-0410-b5e6-
96231b3b80d8
Richard Barton [Wed, 28 Mar 2012 10:18:11 +0000 (10:18 +0000)]
Fixup VST1.32 with writeback instruction. Also re-factor non-writeback version.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153573
91177308-0d34-0410-b5e6-
96231b3b80d8
Chandler Carruth [Wed, 28 Mar 2012 08:38:27 +0000 (08:38 +0000)]
Switch to WeakVHs in the value mapper, and aggressively prune dead basic
blocks in the function cloner. This removes the last case of trivially
dead code that I've been seeing in the wild getting inlined, analyzed,
re-inlined, optimized, only to be deleted. Nukes a FIXME from the
cleanup tests.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153572
91177308-0d34-0410-b5e6-
96231b3b80d8
Eric Christopher [Wed, 28 Mar 2012 07:34:36 +0000 (07:34 +0000)]
More debug output.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153571
91177308-0d34-0410-b5e6-
96231b3b80d8
Eric Christopher [Wed, 28 Mar 2012 07:34:31 +0000 (07:34 +0000)]
Fix the output of the DW_TAG_friend tag to include DW_AT_friend
and not the rest of the member tag.
Fixes PR11695
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153570
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Wed, 28 Mar 2012 04:17:34 +0000 (04:17 +0000)]
Some whitespace cleanup.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153567
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Wed, 28 Mar 2012 02:39:06 +0000 (02:39 +0000)]
Use the correct filename for the error message.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153564
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Wed, 28 Mar 2012 01:30:51 +0000 (01:30 +0000)]
Use Nakamura's suggestion of bypassing using 'filename' and just the pointers directly.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153558
91177308-0d34-0410-b5e6-
96231b3b80d8
Akira Hatanaka [Wed, 28 Mar 2012 00:52:23 +0000 (00:52 +0000)]
Turn off post-RA scheduler by default.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153557
91177308-0d34-0410-b5e6-
96231b3b80d8
Chad Rosier [Wed, 28 Mar 2012 00:35:33 +0000 (00:35 +0000)]
Fix 80-column violation.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153556
91177308-0d34-0410-b5e6-
96231b3b80d8
Akira Hatanaka [Wed, 28 Mar 2012 00:25:01 +0000 (00:25 +0000)]
Fix test case.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153555
91177308-0d34-0410-b5e6-
96231b3b80d8
Akira Hatanaka [Wed, 28 Mar 2012 00:24:17 +0000 (00:24 +0000)]
Turn on post register allocation scheduler.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153554
91177308-0d34-0410-b5e6-
96231b3b80d8
Akira Hatanaka [Wed, 28 Mar 2012 00:23:33 +0000 (00:23 +0000)]
Sort relocation entries before they are written out to a file. MIPS ABI
imposes a constraint that GOT16 referring to a local symbol or HI16 has to be
followed immediately by a matching LO16 relocation.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153553
91177308-0d34-0410-b5e6-
96231b3b80d8
Akira Hatanaka [Wed, 28 Mar 2012 00:22:50 +0000 (00:22 +0000)]
Emit all directives except for ".cprestore" during asm printing rather than emit
them as machine instructions. Directives ".set noat" and ".set at" are now
emitted only at the beginning and end of a function except in the case where
they are emitted to enclose .cpload with an immediate operand that doesn't fit
in 16-bit field or unaligned load/stores.
Also, make the following changes:
- Remove function isUnalignedLoadStore and use a switch-case statement to
determine whether an instruction is an unaligned load or store.
- Define helper function CreateMCInst which generates an instance of an MCInst
from an opcode and a list of operands.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153552
91177308-0d34-0410-b5e6-
96231b3b80d8
Akira Hatanaka [Wed, 28 Mar 2012 00:21:37 +0000 (00:21 +0000)]
Mark flag neverHasSideEffects of pattern-less instructions that do not have
any side effects.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153551
91177308-0d34-0410-b5e6-
96231b3b80d8
Francois Pichet [Tue, 27 Mar 2012 23:52:22 +0000 (23:52 +0000)]
MSVC doesn't like the mixing of declarations and statements in a .c file.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153549
91177308-0d34-0410-b5e6-
96231b3b80d8
Benjamin Kramer [Tue, 27 Mar 2012 22:42:42 +0000 (22:42 +0000)]
Add a note about a cute little fabs optimization.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153543
91177308-0d34-0410-b5e6-
96231b3b80d8
Benjamin Kramer [Tue, 27 Mar 2012 22:03:19 +0000 (22:03 +0000)]
Add two missed instcombines related to compares with nsw arithmetic.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153542
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Tue, 27 Mar 2012 21:17:04 +0000 (21:17 +0000)]
Try to use the CWD if the path to the GCDA output is not available (e.g., the
executable has been moved to another machine). If that's not available
(read-only or something), then exit gracefully.
<rdar://problem/
11111686>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153538
91177308-0d34-0410-b5e6-
96231b3b80d8
Akira Hatanaka [Tue, 27 Mar 2012 20:35:51 +0000 (20:35 +0000)]
Remove trailing white space.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153536
91177308-0d34-0410-b5e6-
96231b3b80d8
Lang Hames [Tue, 27 Mar 2012 19:10:45 +0000 (19:10 +0000)]
Use a SmallVector and linear lookup instead of a DenseSet - SourceMap values
will always be tiny sets, so DenseSet is overkill (SmallSet won't work as we
need iteration support).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153529
91177308-0d34-0410-b5e6-
96231b3b80d8
Akira Hatanaka [Tue, 27 Mar 2012 19:08:42 +0000 (19:08 +0000)]
Add member EmitNOAT and its setter and getter functions to class MipsFunctionInfo.
If EmitNOAT is true, directives ".set noat" and ".set at" are emitted at the
beginning and end of a function.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153528
91177308-0d34-0410-b5e6-
96231b3b80d8
Eric Christopher [Tue, 27 Mar 2012 18:35:57 +0000 (18:35 +0000)]
Add a test for the previous commit. Also, remove two tests that were
testing a) the wrong behavior or b) something that I'm already testing
in the new test.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153525
91177308-0d34-0410-b5e6-
96231b3b80d8
Eric Christopher [Tue, 27 Mar 2012 18:35:54 +0000 (18:35 +0000)]
Use DW_AT_low_pc for a single entry point into a routine.
Fixes PR10105
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153524
91177308-0d34-0410-b5e6-
96231b3b80d8
Chad Rosier [Tue, 27 Mar 2012 17:44:52 +0000 (17:44 +0000)]
Reapply r153423; the original commit was fine. The failing test, distray, had
undefined behavior, which Rafael was kind enough to fix.
Original commit message for r153423:
Use the new range metadata in computeMaskedBits and add a new optimization to
instruction simplify that lets us remove an and when loding a boolean value.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153521
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Tue, 27 Mar 2012 17:33:52 +0000 (17:33 +0000)]
ARMLoadStoreOptimizer invalidates register liveness.
This pass tries to update kill flags, but there are still many bugs.
Passes after the load/store optimizer don't need accurate liveness, so
don't even try.
<rdar://problem/
11101911>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153519
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Tue, 27 Mar 2012 17:17:16 +0000 (17:17 +0000)]
Print SSA and liveness tracking flags in MF::print().
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153518
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Tue, 27 Mar 2012 17:06:09 +0000 (17:06 +0000)]
Branch folding may invalidate liveness.
Branch folding can use a register scavenger to update liveness
information when required. Don't do that if liveness information is
already invalid.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153517
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Tue, 27 Mar 2012 17:06:06 +0000 (17:06 +0000)]
Invalidate liveness in Thumb2ITBlockPass.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153516
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Tue, 27 Mar 2012 16:27:21 +0000 (16:27 +0000)]
fix what looks like a real logic bug, found by PVS-Studio (part of PR12357)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153513
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Tue, 27 Mar 2012 15:13:58 +0000 (15:13 +0000)]
Add an MRI::tracksLiveness() flag.
Late optimization passes like branch folding and tail duplication can
transform the machine code in a way that makes it expensive to keep the
register liveness information up to date. There is a fuzzy line between
register allocation and late scheduling where the liveness information
degrades.
The MRI::tracksLiveness() flag makes the line clear: While true,
liveness information is accurate, and can be used for register
scavenging. Once the flag is false, liveness information is not
accurate, and can only be used as a hint.
Late passes generally don't need the liveness information, but they will
sometimes use the register scavenger to help update it. The scavenger
enforces strict correctness, and we have to spend a lot of code to
update register liveness that may never be used.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153511
91177308-0d34-0410-b5e6-
96231b3b80d8
NAKAMURA Takumi [Tue, 27 Mar 2012 11:25:16 +0000 (11:25 +0000)]
llvm/docs/*.html: Fix markups.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153508
91177308-0d34-0410-b5e6-
96231b3b80d8
Chandler Carruth [Tue, 27 Mar 2012 10:48:28 +0000 (10:48 +0000)]
Make a seemingly tiny change to the inliner and fix the generated code
size bloat. Unfortunately, I expect this to disable the majority of the
benefit from r152737. I'm hopeful at least that it will fix PR12345. To
explain this requires... quite a bit of backstory I'm afraid.
TL;DR: The change in r152737 actually did The Wrong Thing for
linkonce-odr functions. This change makes it do the right thing. The
benefits we saw were simple luck, not any actual strategy. Benchmark
numbers after a mini-blog-post so that I've written down my thoughts on
why all of this works and doesn't work...
To understand what's going on here, you have to understand how the
"bottom-up" inliner actually works. There are two fundamental modes to
the inliner:
1) Standard fixed-cost bottom-up inlining. This is the mode we usually
think about. It walks from the bottom of the CFG up to the top,
looking at callsites, taking information about the callsite and the
called function and computing th expected cost of inlining into that
callsite. If the cost is under a fixed threshold, it inlines. It's
a touch more complicated than that due to all the bonuses, weights,
etc. Inlining the last callsite to an internal function gets higher
weighth, etc. But essentially, this is the mode of operation.
2) Deferred bottom-up inlining (a term I just made up). This is the
interesting mode for this patch an r152737. Initially, this works
just like mode #1, but once we have the cost of inlining into the
callsite, we don't just compare it with a fixed threshold. First, we
check something else. Let's give some names to the entities at this
point, or we'll end up hopelessly confused. We're considering
inlining a function 'A' into its callsite within a function 'B'. We
want to check whether 'B' has any callers, and whether it might be
inlined into those callers. If so, we also check whether inlining 'A'
into 'B' would block any of the opportunities for inlining 'B' into
its callers. We take the sum of the costs of inlining 'B' into its
callers where that inlining would be blocked by inlining 'A' into
'B', and if that cost is less than the cost of inlining 'A' into 'B',
then we skip inlining 'A' into 'B'.
Now, in order for #2 to make sense, we have to have some confidence that
we will actually have the opportunity to inline 'B' into its callers
when cheaper, *and* that we'll be able to revisit the decision and
inline 'A' into 'B' if that ever becomes the correct tradeoff. This
often isn't true for external functions -- we can see very few of their
callers, and we won't be able to re-consider inlining 'A' into 'B' if
'B' is external when we finally see more callers of 'B'. There are two
cases where we believe this to be true for C/C++ code: functions local
to a translation unit, and functions with an inline definition in every
translation unit which uses them. These are represented as internal
linkage and linkonce-odr (resp.) in LLVM. I enabled this logic for
linkonce-odr in r152737.
Unfortunately, when I did that, I also introduced a subtle bug. There
was an implicit assumption that the last caller of the function within
the TU was the last caller of the function in the program. We want to
bonus the last caller of the function in the program by a huge amount
for inlining because inlining that callsite has very little cost.
Unfortunately, the last caller in the TU of a linkonce-odr function is
*not* the last caller in the program, and so we don't want to apply this
bonus. If we do, we can apply it to one callsite *per-TU*. Because of
the way deferred inlining works, when it sees this bonus applied to one
callsite in the TU for 'B', it decides that inlining 'B' is of the
*utmost* importance just so we can get that final bonus. It then
proceeds to essentially force deferred inlining regardless of the actual
cost tradeoff.
The result? PR12345: code bloat, code bloat, code bloat. Another result
is getting *damn* lucky on a few benchmarks, and the over-inlining
exposing critically important optimizations. I would very much like
a list of benchmarks that regress after this change goes in, with
bitcode before and after. This will help me greatly understand what
opportunities the current cost analysis is missing.
Initial benchmark numbers look very good. WebKit files that exhibited
the worst of PR12345 went from growing to shrinking compared to Clang
with r152737 reverted.
- Bootstrapped Clang is 3% smaller with this change.
- Bootstrapped Clang -O0 over a single-source-file of lib/Lex is 4%
faster with this change.
Please let me know about any other performance impact you see. Thanks to
Nico for reporting and urging me to actually fix, Richard Smith, Duncan
Sands, Manuel Klimek, and Benjamin Kramer for talking through the issues
today.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153506
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Tue, 27 Mar 2012 07:54:11 +0000 (07:54 +0000)]
Prune some includes
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153502
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Tue, 27 Mar 2012 07:21:54 +0000 (07:21 +0000)]
Remove unnecessary llvm:: qualifications
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153500
91177308-0d34-0410-b5e6-
96231b3b80d8
Akira Hatanaka [Tue, 27 Mar 2012 03:13:56 +0000 (03:13 +0000)]
Pass the llvm IR pointer value and offset to the constructor of
MachinePointerInfo when getStore is called to create a node that stores an
argument passed in register to the stack. Without this change, the post RA
scheduler will fail to discover the dependencies between the stores
instructions and the instructions that load from a structure passed by value.
The link to the related discussion is here:
http://lists.cs.uiuc.edu/pipermail/llvmdev/2012-March/048055.html
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153499
91177308-0d34-0410-b5e6-
96231b3b80d8
Akira Hatanaka [Tue, 27 Mar 2012 02:55:31 +0000 (02:55 +0000)]
Fix bug in LowerConstantPool.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153498
91177308-0d34-0410-b5e6-
96231b3b80d8
Akira Hatanaka [Tue, 27 Mar 2012 02:46:25 +0000 (02:46 +0000)]
Add T9 to the list of live-in registers of the entry basic block.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153497
91177308-0d34-0410-b5e6-
96231b3b80d8
Akira Hatanaka [Tue, 27 Mar 2012 02:33:05 +0000 (02:33 +0000)]
Retrieve and add the offset of a symbol in applyFixup rather than retrieve and
set it in MipsMCCodeEmitter::getMachineOpValue. Assert in getMachineOpValue if
MachineOperand MO is of an unexpected type.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153494
91177308-0d34-0410-b5e6-
96231b3b80d8
Akira Hatanaka [Tue, 27 Mar 2012 02:04:18 +0000 (02:04 +0000)]
Define function MipsGetSymAndOffset which returns a fixup's symbol and the
offset applied to it.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153493
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Tue, 27 Mar 2012 01:50:58 +0000 (01:50 +0000)]
Post-ra LICM should take care not to hoist an instruction that would clobber a
register that's read by the preheader terminator.
rdar://
11095580
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153492
91177308-0d34-0410-b5e6-
96231b3b80d8
Akira Hatanaka [Tue, 27 Mar 2012 01:50:08 +0000 (01:50 +0000)]
Rewrite computation of Value in adjustFixupValue so that the upper 48-bits are
cleared. No functionality change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153491
91177308-0d34-0410-b5e6-
96231b3b80d8
Lang Hames [Tue, 27 Mar 2012 00:44:47 +0000 (00:44 +0000)]
During MachineCopyPropagation a register may be the source operand of multiple
copies being considered for removal. Make sure to track all of the copies,
rather than just the most recent encountered, by holding a DenseSet instead of
an unsigned in SrcMap.
No test case - couldn't reduce something with a sane size.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153487
91177308-0d34-0410-b5e6-
96231b3b80d8
Akira Hatanaka [Tue, 27 Mar 2012 00:40:56 +0000 (00:40 +0000)]
Reserve hardware registers.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153486
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Mon, 26 Mar 2012 23:31:00 +0000 (23:31 +0000)]
ARM has a peephole optimization which looks for a def / use pair. The def
produces a 32-bit immediate which is consumed by the use. It tries to
fold the immediate by breaking it into two parts and fold them into the
immmediate fields of two uses. e.g
movw r2, #40885
movt r3, #46540
add r0, r0, r3
=>
add.w r0, r0, #
3019898880
add.w r0, r0, #
30146560
;
However, this transformation is incorrect if the user produces a flag. e.g.
movw r2, #40885
movt r3, #46540
adds r0, r0, r3
=>
add.w r0, r0, #
3019898880
adds.w r0, r0, #
30146560
Note the adds.w may not set the carry flag even if the original sequence
would.
rdar://
11116189
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153484
91177308-0d34-0410-b5e6-
96231b3b80d8
Lang Hames [Mon, 26 Mar 2012 23:07:23 +0000 (23:07 +0000)]
Add a debug option to dump PBQP graphs during register allocation.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153483
91177308-0d34-0410-b5e6-
96231b3b80d8
Andrew Trick [Mon, 26 Mar 2012 22:33:59 +0000 (22:33 +0000)]
SCEV fix: Handle loop invariant loads.
Fixes PR11882: NULL dereference in ComputeLoadConstantCompareExitLimit.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153480
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Mon, 26 Mar 2012 22:15:12 +0000 (22:15 +0000)]
Add 'undef's to make SWIG happier. Patch by Baozeng Ding.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153479
91177308-0d34-0410-b5e6-
96231b3b80d8
Eric Christopher [Mon, 26 Mar 2012 21:56:56 +0000 (21:56 +0000)]
Add InitializeNativeTargetDisassembler function.
Patch by Ojab.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153476
91177308-0d34-0410-b5e6-
96231b3b80d8
Andrew Trick [Mon, 26 Mar 2012 21:45:37 +0000 (21:45 +0000)]
Unit test for PR11950: LSR crash.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153472
91177308-0d34-0410-b5e6-
96231b3b80d8
Eric Christopher [Mon, 26 Mar 2012 21:38:38 +0000 (21:38 +0000)]
Use the file in the inlined die rather than the compile unit for
backtrace locations.
Testcase forthcoming, but I wanted to get some testing here.
Should fix:
PR12323
PR12314
rdar://
11091100
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153471
91177308-0d34-0410-b5e6-
96231b3b80d8
Nadav Rotem [Mon, 26 Mar 2012 21:00:53 +0000 (21:00 +0000)]
153465 was incorrect. In this code we wanted to check that the pointer operand is of pointer type (and not vector type).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153468
91177308-0d34-0410-b5e6-
96231b3b80d8
Sean Callanan [Mon, 26 Mar 2012 20:45:52 +0000 (20:45 +0000)]
Made RuntimeDyldMachO support vanilla i386
relocations. The algorithm is the same as
that for x86_64. Scattered relocations, a
feature present in i386 but not on x86_64,
are not yet supported.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153466
91177308-0d34-0410-b5e6-
96231b3b80d8
Nadav Rotem [Mon, 26 Mar 2012 20:39:18 +0000 (20:39 +0000)]
PR12357: The pointer was used before it was checked.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153465
91177308-0d34-0410-b5e6-
96231b3b80d8
Andrew Trick [Mon, 26 Mar 2012 20:28:37 +0000 (20:28 +0000)]
LSR ivchain bug fix: corner case with ConstantExpr.
Fixes PR11950.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153463
91177308-0d34-0410-b5e6-
96231b3b80d8
Andrew Trick [Mon, 26 Mar 2012 20:28:35 +0000 (20:28 +0000)]
comment typo
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153462
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 26 Mar 2012 19:13:57 +0000 (19:13 +0000)]
eliminate an unneeded branch, part of PR12357
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153458
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 26 Mar 2012 19:11:51 +0000 (19:11 +0000)]
fix a failure path to print the right thing, part of PR12357
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153457
91177308-0d34-0410-b5e6-
96231b3b80d8
Eric Christopher [Mon, 26 Mar 2012 19:09:40 +0000 (19:09 +0000)]
Tidy.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153456
91177308-0d34-0410-b5e6-
96231b3b80d8
Eric Christopher [Mon, 26 Mar 2012 19:09:38 +0000 (19:09 +0000)]
Tidy.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153455
91177308-0d34-0410-b5e6-
96231b3b80d8
Chad Rosier [Mon, 26 Mar 2012 18:07:14 +0000 (18:07 +0000)]
Revert r153423 as this is causing failures on our internal nightly testers.
Original commit message:
Use the new range metadata in computeMaskedBits and add a new optimization to
instruction simplify that lets us remove an and when loading a boolean value.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153452
91177308-0d34-0410-b5e6-
96231b3b80d8
Andrew Trick [Mon, 26 Mar 2012 18:03:16 +0000 (18:03 +0000)]
LSR cleanup: potential bug caught by PVS-Studio.
Thanks Andrey.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153451
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Dunbar [Mon, 26 Mar 2012 18:01:14 +0000 (18:01 +0000)]
docs/lit: Add some notes on the lit test run output format.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153450
91177308-0d34-0410-b5e6-
96231b3b80d8
Kostya Serebryany [Mon, 26 Mar 2012 17:35:03 +0000 (17:35 +0000)]
[tsan] treat vtable pointer updates in a special way (requires tbaa); fix a bug (forgot to return true after instrumenting); make sure the tsan tests are run
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153448
91177308-0d34-0410-b5e6-
96231b3b80d8
Benjamin Kramer [Mon, 26 Mar 2012 14:17:26 +0000 (14:17 +0000)]
No need to do an expensive stable sort for a bunch of integers.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153438
91177308-0d34-0410-b5e6-
96231b3b80d8
Douglas Gregor [Mon, 26 Mar 2012 14:04:17 +0000 (14:04 +0000)]
Add missing include of <new>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153436
91177308-0d34-0410-b5e6-
96231b3b80d8
Anton Korobeynikov [Mon, 26 Mar 2012 12:05:51 +0000 (12:05 +0000)]
Fix GetMainExecutable on kFreeBSD.
Patch by Sylvestre Ledru!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153435
91177308-0d34-0410-b5e6-
96231b3b80d8
Benjamin Kramer [Mon, 26 Mar 2012 11:16:50 +0000 (11:16 +0000)]
Remove stale CBackend tests.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153433
91177308-0d34-0410-b5e6-
96231b3b80d8
Benjamin Kramer [Mon, 26 Mar 2012 11:08:03 +0000 (11:08 +0000)]
TableGen: Don't emit the llvm intrinsic -> gcc builtin table, its only user was the c backend.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153432
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Mon, 26 Mar 2012 06:58:25 +0000 (06:58 +0000)]
Prune some includes and forward declarations.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153429
91177308-0d34-0410-b5e6-
96231b3b80d8
Eric Christopher [Mon, 26 Mar 2012 06:10:32 +0000 (06:10 +0000)]
Add a debug statement.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153428
91177308-0d34-0410-b5e6-
96231b3b80d8
Eric Christopher [Mon, 26 Mar 2012 02:09:01 +0000 (02:09 +0000)]
Add some fixes to the configure script for isInf and add
--enable-libcpp to projects/sample.
Patch by Dmitri Shubin with additional fixes by me.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@153425
91177308-0d34-0410-b5e6-
96231b3b80d8