JF Bastien [Tue, 16 Dec 2014 20:15:45 +0000 (20:15 +0000)]
x86-32: PUSHF/POPF use/def EFLAGS
Summary: As a side-quest for D6629 jvoung pointed out that I should use -verify-machineinstrs and this found a bug in x86-32's handling of EFLAGS for PUSHF/POPF. This patch fixes the use/def, and adds -verify-machineinstrs to all x86 tests which contain 'EFLAGS'. One exception: this patch leaves inline-asm-fpstack.ll as-is because it fails -verify-machineinstrs in a way unrelated to EFLAGS. This patch also modifies cmpxchg-clobber-flags.ll along the lines of what D6629 already does by also testing i386.
Quentin Colombet [Tue, 16 Dec 2014 19:09:03 +0000 (19:09 +0000)]
[CodeGenPrepare] Move sign/zero extensions near loads using type promotion.
This patch extends the optimization in CodeGenPrepare that moves a sign/zero
extension near a load when the target can combine them. The optimization may
promote any operations between the extension and the load to make that possible.
Although this optimization may be beneficial for all targets, in particular
AArch64, this is enabled for X86 only as I have not benchmarked it for other
targets yet.
** Context **
Most targets feature extended loads, i.e., loads that perform a zero or sign
extension for free. In that context it is interesting to expose such pattern in
CodeGenPrepare so that the instruction selection pass can form such loads.
Sometimes, this pattern is blocked because of instructions between the load and
the extension. When those instructions are promotable to the extended type, we
can expose this pattern.
As it is, this IR generates the following assembly on x86_64:
[...]
movzbl (%rdi), %eax # zero-extended load
movl (%rsi), %es # plain load
addl %eax, %esi # 32-bit add
movslq %esi, %rdi # sign extend the result of add
movzbl %dl, %edx # zero extend the first argument
addl %eax, %edx # 32-bit add
movslq %edx, %rsi # sign extend the result of add
addl %eax, %ecx # 32-bit add
movslq %ecx, %rdx # sign extend the result of add
[...]
The throughput of this sequence is 7.45 cycles on Ivy Bridge according to IACA.
Now, by promoting the additions to form more extended loads we would generate:
[...]
movzbl (%rdi), %eax # zero-extended load
movslq (%rsi), %rdi # sign-extended load
addq %rax, %rdi # 64-bit add
movzbl %dl, %esi # zero extend the first argument
addq %rax, %rsi # 64-bit add
movslq %ecx, %rdx # sign extend the second argument
addq %rax, %rdx # 64-bit add
[...]
The throughput of this sequence is 6.15 cycles on Ivy Bridge according to IACA.
This kind of sequences happen a lot on code using 32-bit indexes on 64-bit
architectures.
Note: The throughput numbers are similar on Sandy Bridge and Haswell.
** Proposed Solution **
To avoid the penalty of all these sign/zero extensions, we merge them in the
loads at the beginning of the chain of computation by promoting all the chain of
computation on the extended type. The promotion is done if and only if we do not
introduce new extensions, i.e., if we do not degrade the code quality.
To achieve this, we extend the existing “move ext to load” optimization with the
promotion mechanism introduced to match larger patterns for addressing mode
(r200947).
The idea of this extension is to perform the following transformation:
ext(promotableInst1(...(promotableInstN(load))))
=>
promotedInst1(...(promotedInstN(ext(load))))
The promotion mechanism in that optimization is enabled by a new TargetLowering
switch, which is off by default. In other words, by default, the optimization
performs the “move ext to load” optimization as it was before this patch.
** Performance **
Configuration: x86_64: Ivy Bridge fixed at 2900MHz running OS X 10.10.
Tested Optimization Levels: O3/Os
Tests: llvm-testsuite + externals.
Results:
- No regression beside noise.
- Improvements:
CINT2006/473.astar: ~2%
Benchmarks/PAQ8p: ~2%
Misc/perlin: ~3%
Kevin Enderby [Tue, 16 Dec 2014 18:58:11 +0000 (18:58 +0000)]
Fix the arm build bots for a test that was added. A printing routine was incorrectly using PRIx32
when it should have been using PRIx64 for the value that was passed as uint64_t .
An instruction alias defined with InstAlias and an optional operand in the
middle of the AsmString field, "..${a} <operands>", would get the final
"}" printed in the instruction disassembly. This wouldn't happen if the optional
operand appeared as the last item in the AsmString which is how the current
backends avoided the problem.
There don't appear to be any tests for this part of Tablegen but it passes the
pre-commit tests. Manually tested the change by enabling the generic alias
printer in the ARM backend and checking the output.
Ahmed Bougacha [Tue, 16 Dec 2014 18:05:28 +0000 (18:05 +0000)]
[MC] Reset the MCInst in the matcher function before adding opcode/operands.
On X86, the Intel asm parser tries to match all memory operand sizes when
none is explicitly specified. For LEA, which doesn't really have a memory
operand (just a pointer one), this results in multiple successful matches,
one for each memory size. There's no error because it's same opcode, so
really, it's just one match. However, the tablegen'd matcher function
adds opcode/operands to the passed MCInst, and this results in multiple
duplicated operands.
This commit clears the MCInst in the tablegen'd matcher function.
We sometimes clear it when the match failed, so there's no expectation of
keeping the previous content anyway.
Sanjay Patel [Tue, 16 Dec 2014 16:30:01 +0000 (16:30 +0000)]
combine consecutive subvector 16-byte loads into one 32-byte load
This is a fix for PR21709 ( http://llvm.org/bugs/show_bug.cgi?id=21709 ).
When we have 2 consecutive 16-byte loads that are merged into one 32-byte vector,
we can use a single 32-byte load instead.
But we don't do this for SandyBridge / IvyBridge because they have slower 32-byte memops.
We also don't bother using 32-byte *integer* loads on a machine that only has AVX1 (btver2)
because those operands would have to be split in half anyway since there is no support for
32-byte integer math ops.
Vladimir Medic [Tue, 16 Dec 2014 15:29:12 +0000 (15:29 +0000)]
The single check for N64 inside MipsDisassemblerBase's subclasses is actually wrong. It should be testing for FeatureGP64bit.There are no functional changes.
Masked Load and Store Intrinsics in loop vectorizer.
The loop vectorizer optimizes loops containing conditional memory
accesses by generating masked load and store intrinsics.
This decision is target dependent.
X86: Added FeatureVectorUAMem for all AVX architectures.
According to AVX specification:
"Most arithmetic and data processing instructions encoded using the VEX prefix and
performing memory accesses have more flexible memory alignment requirements
than instructions that are encoded without the VEX prefix. Specifically,
With the exception of explicitly aligned 16 or 32 byte SIMD load/store instructions,
most VEX-encoded, arithmetic and data processing instructions operate in
a flexible environment regarding memory address alignment, i.e. VEX-encoded
instruction with 32-byte or 16-byte load semantics will support unaligned load
operation by default. Memory arguments for most instructions with VEX prefix
operate normally without causing #GP(0) on any byte-granularity alignment
(unlike Legacy SSE instructions)."
The same for AVX-512.
This change does not affect anything right now, because only the "memop pattern fragment"
depends on FeatureVectorUAMem and it is not used in AVX patterns.
All AVX patterns are based on the "unaligned load" anyway.
IR: Make MDNode::dump() useful by adding addresses
It's horrible to inspect `MDNode`s in a debugger. All of their operands
that are `MDNode`s get dumped as `<badref>`, since we can't assign
metadata slots in the context of a `Metadata::dump()`. (Why not? Why
not assign numbers lazily? Because then each time you called `dump()`,
a given `MDNode` could have a different lazily assigned number.)
Fortunately, the C memory model gives us perfectly good identifiers for
`MDNode`. Add pointer addresses to the dumps, transforming this:
(lldb) e N->dump()
!{i32 662302, i32 26, <badref>, null}
(lldb) e ((MDNode*)N->getOperand(2))->dump()
!{i32 4, !"foo"}
into:
(lldb) e N->dump()
!{i32 662302, i32 26, <0x100706ee0>, null}
(lldb) e ((MDNode*)0x100706ee0)->dump()
!{i32 4, !"foo"}
and this:
(lldb) e N->dump()
0x101200248 = !{<badref>, <badref>, <badref>, <badref>, <badref>}
DebugInfo: Update testcase to actually check something
This test was missing a `Debug Info Version` so it's `not grep` was
passing vacuously. Update it to CHECK for something useful at the same
time so it doesn't bitrot quite so easily in the future.
The PowerPC backend, somewhat embarrassingly, did not generate an
optimal-length sequence of instructions for a 32-bit bswap. While adding a
pattern for the bswap intrinsic to fix this would not have been terribly
difficult, doing so would not have addressed the real problem: we had been
generating poor code for many bit-permuting operations (by which I mean things
like byte-swap that permute the bits of one or more inputs around in various
ways). Here are some initial steps toward solving this deficiency.
Bit-permuting operations are represented, at the SDAG level, using ISD::ROTL,
SHL, SRL, AND and OR (mostly with constant second operands). Looking back
through these operations, we can build up a description of the bits in the
resulting value in terms of bits of one or more input values (and constant
zeros). For each bit, we compute the rotation amount from the original value,
and then group consecutive (value, rotation factor) bits into groups. Groups
sharing these attributes are then collected and sorted, and we can then
instruction select the entire permutation using a combination of masked
rotations (rlwinm), imm ands (andi/andis), and masked rotation inserts
(rlwimi).
The result is that instead of lowering an i32 bswap as:
Matthias Braun [Tue, 16 Dec 2014 04:03:38 +0000 (04:03 +0000)]
LiveRangeCalc: Rewrite subrange calculation
This changes subrange calculation to calculate subranges sequentially
instead of in parallel. The code is easier to understand that way and
addresses the code review issues raised about LiveOutData being
hard to understand/needing more comments by removing them :)
Adrian Prantl [Tue, 16 Dec 2014 00:20:49 +0000 (00:20 +0000)]
ARM/AArch64: Attach the FrameSetup MIFlag to CFI instructions.
Debug info marks the first instruction without the FrameSetup flag
as being the end of the function prologue. Any CFI instructions in the
middle of the function prologue would cause debug info to end the prologue
too early and worse, attach the line number of the CFI instruction, which
incidentally is often 0.
JF Bastien [Mon, 15 Dec 2014 22:34:58 +0000 (22:34 +0000)]
x86: Emit LOCK prefix after DATA16
Summary: x86 allows either ordering for the LOCK and DATA16 prefixes, but using GCC+GAS leads to different code generation than using LLVM. This change matches the order that GAS emits the x86 prefixes when a semicolon isn't used in inline assembly (see tc-i386.c comment before define LOCK_PREFIX), and helps simplify tooling that operates on the instruction's byte sequence (such as NaCl's validator). This change shouldn't have any performance impact.
Michael Ilseman [Mon, 15 Dec 2014 21:36:29 +0000 (21:36 +0000)]
Revert of r223763, in spirit.
r223763 was made to work around a temporary issue where a user of the
JIT was passing down a declaration (incorrectly). This shouldn't
occur, so assert rather than silently continue.
Mark Heffernan [Mon, 15 Dec 2014 21:19:53 +0000 (21:19 +0000)]
Clarify HowFarToZero computation when the step is a positive power of two. Functionally this should be identical to the existing code except for the case where Step is maximally negative (eg, INT_MIN). We now punt in that one corner case to make reasoning about the code easier.
Matthias Braun [Mon, 15 Dec 2014 21:16:21 +0000 (21:16 +0000)]
LiveRangeCalc: Rewrite subrange calculation
This changes subrange calculation to calculate subranges sequentially
instead of in parallel. The code is easier to understand that way and
addresses the code review issues raised about LiveOutData being
hard to understand/needing more comments by removing them :)
I wrote an upgrade script that handled almost all of the tests in llvm
and many of the tests in cfe (even handling many `CHECK` lines). I've
attached it (or will attach it in a moment if you're speedy) to PR21532
to help everyone update their out-of-tree testcases.
Vladimir Medic [Mon, 15 Dec 2014 15:22:33 +0000 (15:22 +0000)]
This is the first in a series of patches that add missing disassembler tests for mips platform. The patches are divided per version of mips CPU to keep the patches smaller and ease the review. There are no functional changes, code is changed only if new tests reveal a bug.This patch adds disassembler tests for mips1 CPU.
Sink store based on alias analysis
- by Ella Bolshinsky
The alias analysis is used define whether the given instruction
is a barrier for store sinking. For 2 identical stores, following
instructions are checked in the both basic blocks, to determine
whether they are sinking barriers.
[X86] Break false dependencies before partial register updates when the source operand is in memory
Adds the various "rm" instruction variants into the list of instructions that have a partial register update. Also adds all variants of SQRTSD that were missing in the original list.
Alexey Bataev [Mon, 15 Dec 2014 04:45:43 +0000 (04:45 +0000)]
Fix line mapping information in LLVM JIT profiling with Vtune
The line mapping information for dynamic code is reported incorrectly. It causes VTune to map LLVM generated code to source lines incorrectly. This patch fix this issue.
Patch by Denis Pravdin.
Differential Revision: http://reviews.llvm.org/D6603
David Majnemer [Sun, 14 Dec 2014 09:41:56 +0000 (09:41 +0000)]
APInt: udivrem should use machine instructions for single-word APInts
This mirrors the behavior of APInt::udiv and APInt::urem. Some
architectures, like X86, have a single instruction which can compute
both division and remainder.
Hal Finkel [Sun, 14 Dec 2014 05:53:19 +0000 (05:53 +0000)]
[PowerPC] Handle cmp op promotion for SELECT[_CC] nodes in PPCTL::DAGCombineExtBoolTrunc
PPCTargetLowering::DAGCombineExtBoolTrunc contains logic to remove unwanted
truncations and extensions when dealing with nodes of the form:
zext(binary-ops(binary-ops(trunc(x), trunc(y)), ...)
There was a FIXME in the implementation (now removed) regarding the fact that
the function would abort the transformations if any of the non-output operands
of a SELECT or SELECT_CC node would need to be promoted (because they were
also output operands, for example). As a result, we continued to generate
unnecessary zero-extends for code such as this:
unsigned foo(unsigned a, unsigned b) {
return (a <= b) ? a : b;
}
r223862 tried to also combine base-updating load/stores.
r224198 reverted it, as "it created a regression on the test-suite
on test MultiSource/Benchmarks/Ptrdist/anagram by scrambling the order
in which the words are shown."
Reapply, with a fix to ignore non-normal load/stores.
Truncstores are handled elsewhere (you can actually write a pattern for
those, whereas for postinc loads you can't, since they return two values),
but it should be possible to also combine extloads base updates, by checking
that the memory (rather than result) type is of the same size as the addend.
Original commit message:
We used to only combine intrinsics, and turn them into VLD1_UPD/VST1_UPD
when the base pointer is incremented after the load/store.
We can do the same thing for generic load/stores.
Note that we can only combine the first load/store+adds pair in
a sequence (as might be generated for a v16f32 load for instance),
because other combines turn the base pointer addition chain (each
computing the address of the next load, from the address of the last
load) into independent additions (common base pointer + this load's
offset).
This reverts commit r223862, as it created a regression on the test-suite
on test MultiSource/Benchmarks/Ptrdist/anagram by scrambling the order
in which the words are shown. We'll investigate the issue and re-apply
when safe.
Akira Hatanaka [Sat, 13 Dec 2014 04:52:04 +0000 (04:52 +0000)]
Rename argument strings of codegen passes to avoid collisions with command line
options.
This commit changes the command line arguments (PassInfo::PassArgument) of two
passes, MachineFunctionPrinter and MachineScheduler, to avoid collisions with
command line options that have the same argument strings.
This bug manifests when the PassList construct (defined in opt.cpp) is used
in a tool that links with codegen passes. To reproduce the bug, paste the
following lines into llc.cpp and run llc.
projects / llvm / log