The e500mc does not actually support the mfocrf instruction; update the
processor definitions to reflect that fact.
Patch by Tom Rix (with some test-case cleanup by me).
llvm-svn: 254064
to a simple type when lowering a truncating store of a vector type. In this
case for an EVT we'll return Expand as we should in all of the cases anyhow.
The testcase triggered at the one in VectorLegalizer::LegalizeOp, inspection
found the rest.
llvm-svn: 254061
It was wrong order of operands (from intrinsic to DAG node).
I added more strict type specification for instruction selection.
Differential Revision: http://reviews.llvm.org/D14942
llvm-svn: 254059
1. Convert serialization methods using InstrProfRecord as source into C (impl)
interfaces using Closure.
2. Reimplement InstrProfRecord serialization method to use new C interface
as dummy wrapper.
Now it is ready to implement wrapper for runtime value profile data.
(The new code need better source location -- but not changed in this patch to
minimize diffs. )
llvm-svn: 254057
This caused PR25607 and also caused Chromium to crash on start-up.
(Also had to update test/CodeGen/X86/avx-splat.ll, which was committed
after shrink wrapping was enabled.)
llvm-svn: 254044
Add a simple initial heuristic to control importing based on the number
of instructions recorded in the function's summary. Add option to
control the limit, and test using option.
llvm-svn: 254036
When the original binary is executed and sampled, the resulting profile
contains information on the original inline stack. We currently follow
the original inline plan if we notice that the inlined callsite has more
than 0 samples to it.
A better way is to determine whether the callsite is actually worth
inlining. If the callsite accumulates a small fraction of the samples
spent in the parent function, then we don't want to bother inlining it
(as it means that the callsite is actually cold).
This patch introduces a threshold expressed in percentage of samples
in relation to the parent function. If the callsite uses less than N%
of the total samples used by its parent, the original inline decision is
not re-applied.
I've set the threshold to the very arbitrary value of 5%. I'm yet to do
any actual experiments to see what's a good value. I wanted to separate
the basic mechanism from the tuning.
llvm-svn: 254034
This patch implements a minimum spanning tree (MST) based instrumentation for
PGO. The use of MST guarantees minimum number of CFG edges getting
instrumented. An addition optimization is to instrument the less executed
edges to further reduce the instrumentation overhead. The patch contains both the
instrumentation and the use of the profile to set the branch weights.
Differential Revision: http://reviews.llvm.org/D12781
llvm-svn: 254021
r253918 had refactored expressions like "A - B.Address + C" to "A -
B.getAddressWithOffset(C)". This is incorrect, since the latter really
computes "A - B.Address - C".
None of the tests I can run locally on x86 broke due to this bug, but it
is the current suspect for breakage on the AArch64 buildbots.
llvm-svn: 254017
X86 needs to use its own FMA opcodes, preventing the standard FNEG(FMA) pattern table recognition method used by other platforms. This patch adds support for lowering FNEG(FMA(X,Y,Z)) into a single suitably negated FMA instruction.
Fix for PR24364
Differential Revision: http://reviews.llvm.org/D14906
llvm-svn: 254016
Analyze imported function bodies and add any new external calls to
the worklist for importing. Currently no controls on the importing
so this will end up importing everything possible in the call tree
below the importing module. Basic profitability checks coming next.
Update test to check for iteratively inlined functions.
llvm-svn: 254011
This patch fixes the following issues:
1. Fix the return type of X86psadbw: it should not be the same type of inputs.
For vNi8 inputs the output should be vMi64, where M = N/8.
2. Fix the return type of int_x86_avx512_psad_bw_512 accordingly.
3. Fix the definiton of PSADBW, VPSADBW, and VPSADBWY accordingly.
4. Adjust the return type when building a DAG node of X86ISD::PSADBW type.
5. Update related tests.
Differential revision: http://reviews.llvm.org/D14897
llvm-svn: 254010
The new function import pass exposed an issue when we import references
to local values on multiple importing passes. They are renamed on each
import pass, and we need to ensure that the already promoted and renamed
references existing in the dest module are correctly identified and
updated so that they aren't spuriously renamed again (due to a perceived
conflict with the newly linked reference).
llvm-svn: 254009
Summary:
Followed the guidelines in:
http://llvm.org/docs/CodingStandards.html#include-style
However, I noticed that uppercase named headers come before lowercase ones
throughout the codebase. So kept them as is.
Patch by Mandeep Singh Grang <mgrang@codeaurora.org>
Reviewers: majnemer, davide, jmolloy, atrick
Subscribers: sanjoy
Differential Revision: http://reviews.llvm.org/D14939
llvm-svn: 254005
Convert two C++ static member functions to be C APIs. This
is one of the many steps to get ready to share VP writer code
with profiler runtime.
llvm-svn: 253999
Skip imports for weak_any aliases as well. Fix the test to check
non-import of weak aliases and functions, and import of normal alias.
llvm-svn: 253991
We had duplicated definitions for the same hardware '[v]movq' instructions. For example with SSE:
def MOVZQI2PQIrr : RS2I<0x6E, MRMSrcReg, (outs VR128:$dst), (ins GR64:$src),
"mov{d|q}\t{$src, $dst|$dst, $src}", // X86-64 only
[(set VR128:$dst, (v2i64 (X86vzmovl (v2i64 (scalar_to_vector GR64:$src)))))],
IIC_SSE_MOVDQ>;
def MOV64toPQIrr : RS2I<0x6E, MRMSrcReg, (outs VR128:$dst), (ins GR64:$src),
"mov{d|q}\t{$src, $dst|$dst, $src}",
[(set VR128:$dst, (v2i64 (scalar_to_vector GR64:$src)))],
IIC_SSE_MOVDQ>, Sched<[WriteMove]>;
As shown in the test case and PR25554:
https://llvm.org/bugs/show_bug.cgi?id=25554
This causes us to miss reusing an operand because later passes don't know these 'movq' are the same instruction.
This patch deletes one pair of these defs.
Sadly, this won't fix the original test case in the bug report. Something else is still broken.
Differential Revision: http://reviews.llvm.org/D14941
llvm-svn: 253988
The one regression in the builtin tests is in the read2 test which now
(again) has many extra copies, but this should be solved once the pass
is replaced with a DAG combine.
llvm-svn: 253974
The patch in http://reviews.llvm.org/D13745 is broken into four parts:
1. New interfaces without functional changes.
2. Use new interfaces in SelectionDAG, while in other passes treat probabilities
as weights.
3. Use new interfaces in all other passes.
4. Remove old interfaces.
This the second patch above. In this patch SelectionDAG starts to use
probability-based interfaces in MBB to add successors but other MC passes are
still using weight-based interfaces. Therefore, we need to maintain correct
weight list in MBB even when probability-based interfaces are used. This is
done by updating weight list in probability-based interfaces by treating the
numerator of probabilities as weights. This change affects many test cases
that check successor weight values. I will update those test cases once this
patch looks good to you.
Differential revision: http://reviews.llvm.org/D14361
llvm-svn: 253965
Summary:
This is a helper to perform cross-module import for ThinLTO. Right now
it is importing naively every possible called functions.
Reviewers: tejohnson
Subscribers: dexonsmith, llvm-commits
Differential Revision: http://reviews.llvm.org/D14914
From: Mehdi Amini <mehdi.amini@apple.com>
llvm-svn: 253954
This patch detects the AVG pattern in vectorized code, which is simply
c = (a + b + 1) / 2, where a, b, and c have the same type which are vectors of
either unsigned i8 or unsigned i16. In the IR, i8/i16 will be promoted to
i32 before any arithmetic operations. The following IR shows such an example:
%1 = zext <N x i8> %a to <N x i32>
%2 = zext <N x i8> %b to <N x i32>
%3 = add nuw nsw <N x i32> %1, <i32 1 x N>
%4 = add nuw nsw <N x i32> %3, %2
%5 = lshr <N x i32> %N, <i32 1 x N>
%6 = trunc <N x i32> %5 to <N x i8>
and with this patch it will be converted to a X86ISD::AVG instruction.
The pattern recognition is done when combining instructions just before type
legalization during instruction selection. We do it here because after type
legalization, it is much more difficult to do pattern recognition based
on many instructions that are doing type conversions. Therefore, for
target-specific instructions (like X86ISD::AVG), we need to take care of type
legalization by ourselves. However, as X86ISD::AVG behaves similarly to
ISD::ADD, I am wondering if there is a way to legalize operands and result
types of X86ISD::AVG together with ISD::ADD. It seems that the current design
doesn't support this idea.
Tests are added for SSE2, AVX2, and AVX512BW and both i8 and i16 types of
variant vector sizes.
Differential revision: http://reviews.llvm.org/D14761
llvm-svn: 253952
Caller saved regs differ between SysV and Win64. Use the tail call available set to scavenge from.
Refactor register info to create new helper to get at tail call GPRs. Added a new test case for windows. Fixed up a number of X64 tests since now RCX is preferred over RDX on SysV.
Differential Revision: http://reviews.llvm.org/D14878
llvm-svn: 253927
With the '=' suffix now indicating which operands are output operands, it's
no longer as important to distinguish between a call's inputs and its outputs
using operand ordering, so we can go back to printing them in the normal order.
llvm-svn: 253925
This distinguishes input operands from output operands. This is something of
a syntactic experiment to see whether the mild amount of clutter this adds is
outweighed by the extra information it conveys to the reader.
llvm-svn: 253922
Summary:
For relocation types that are known to not require stub functions, there
is no need to allocate extra space for the stub functions.
Reviewers: lhames, reames, maksfb
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D14676
llvm-svn: 253920
Summary:
Change SectionEntry to keep track of the size of its underlying
allocation, and use that to bounds check advanceStubOffset.
Reviewers: lhames, andrew.w.kaylor, reames
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D14675
llvm-svn: 253919
Summary:
Remove naked access to the data members in `SectionEntry` and route
accesses through accessor functions. This makes it obvious how the
instances of the class are used, and will also facilitate adding bounds
checking to `advanceStubOffset` in a later change.
Reviewers: lhames, loladiro, andrew.w.kaylor
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D14674
llvm-svn: 253918
The new option is similar to the SampleProfile dump option.
- dump raw/indexed format into text profile format
- merge the profile and output into text profile format.
Note that Value Profiling data text format is not yet designed.
That functionality will be added later.
Differential Revision: http://reviews.llvm.org/D14894
llvm-svn: 253913
The existing coverage tracker counts the number of records that were used
from the input profile. An alternative view of coverage is to check how
many available samples were applied.
This way, if the profile contains several records with few samples, it
doesn't really matter much that they were not applied. The more
interesting records to apply are the ones that contribute many samples.
llvm-svn: 253912
The current approach to using get_local and set_local is to use them
implicitly, as register uses and defs. Introduce new copy instructions
which are themselves no-ops except for the get_local and set_local
that they imply, so that we use get_local and set_local consistently.
llvm-svn: 253905
Add a shared helper routine to read the function index from a file
and create/return the function index object. Use it in llvm-link and
llvm-lto.
llvm-svn: 253903
WebAssembly is currently using labels to end scopes, so for example a
loop scope looks like this:
BB0_0:
loop BB0_1
...
BB0_1:
with BB0_0 being the label of the first block not in the loop. This
requires that the label be printed even when it's only reachable via
fallthrough. To arrange this, insert a no-op LOOP_END instruction in
such cases at the end of the loop.
llvm-svn: 253901
Always starting blocks at the top of their containing loops works, but creates
unnecessarily deep nesting because it makes all blocks in a loop overlap.
Refine the BLOCK placement algorithm to start blocks at nearest common
dominating points instead, which significantly shrinks them and reduces
overlapping.
llvm-svn: 253876
Disable custom handling of signed 32-bit and 64-bit integer divide.
Add test cases for both 32-bit and 64-bit integer overflow crashes.
llvm-svn: 253865
Duplicate a few common definitions between DFAPacketizer.cpp and
DFAPacketizerEmitter.cpp to avoid including files from CodeGen
in TableGen.
llvm-svn: 253820
ISERT_SUBVECTOR for i1 vectors may be done with shifts, when we insert into the lower part, or into the upper part, on into all-zero vector.
CONCAT_VECTORS uses ISERT_SUBVECTOR.
Differential Revision: http://reviews.llvm.org/D14815
llvm-svn: 253819
In profile runtime implementation for Darwin, Linux and FreeBSD, the
names of sections holding profile control/counter/naming data need
to be known by the runtime in order to locate the start/end of the
data. Moving the name definitions to the common file to specify the
connection.
llvm-svn: 253814
We had two code paths. One would create names like "foo.1" and the other
names like "foo1".
For globals it is important to use "foo.1" to help C++ name demangling.
For locals there is no strong reason to go one way or the other so I
kept the most common mangling (foo1).
llvm-svn: 253804
Summary:
Several fixes to the handling of bitcode files without function summary
sections so that they are skipped during ThinLTO processing in llvm-lto
and the gold plugin when appropriate instead of aborting.
1 Don't assert when trying to add a FunctionInfo that doesn't have
a summary attached.
2 Skip FunctionInfo structures that don't have attached function summary
sections when trying to create the combined function summary.
3 In both llvm-lto and gold-plugin, check whether a bitcode file has
a function summary section before trying to parse the index, and skip
the bitcode file if it does not.
4 Fix hasFunctionSummaryInMemBuffer in BitcodeReader, which had a bug
where we returned to early while looking for the summary section.
Also added llvm-lto and gold-plugin based tests for cases where we
don't have function summaries in the bitcode file. I verified that
either the first couple fixes described above are enough to avoid the
crashes, or fixes 1,3,4. But have combined them all here for added
robustness.
Reviewers: joker.eph
Subscribers: llvm-commits, joker.eph
Differential Revision: http://reviews.llvm.org/D14903
llvm-svn: 253796
Extended DFA tablegen to:
- added "-debug-only dfa-emitter" support to llvm-tblgen
- defined CVI_PIPE* resources for the V60 vector coprocessor
- allow specification of multiple required resources
- supports ANDs of ORs
- e.g. [SLOT2, SLOT3], [CVI_MPY0, CVI_MPY1] means:
(SLOT2 OR SLOT3) AND (CVI_MPY0 OR CVI_MPY1)
- added support for combo resources
- allows specifying ORs of ANDs
- e.g. [CVI_XLSHF, CVI_MPY01] means:
(CVI_XLANE AND CVI_SHIFT) OR (CVI_MPY0 AND CVI_MPY1)
- increased DFA input size from 32-bit to 64-bit
- allows for a maximum of 4 AND'ed terms of 16 resources
- supported expressions now include:
expression => term [AND term] [AND term] [AND term]
term => resource [OR resource]*
resource => one_resource | combo_resource
combo_resource => (one_resource [AND one_resource]*)
Author: Dan Palermo <dpalermo@codeaurora.org>
kparzysz: Verified AMDGPU codegen to be unchanged on all llc
tests, except those dealing with instruction encodings.
Reapply the previous patch, this time without circular dependencies.
llvm-svn: 253793
Extended DFA tablegen to:
- added "-debug-only dfa-emitter" support to llvm-tblgen
- defined CVI_PIPE* resources for the V60 vector coprocessor
- allow specification of multiple required resources
- supports ANDs of ORs
- e.g. [SLOT2, SLOT3], [CVI_MPY0, CVI_MPY1] means:
(SLOT2 OR SLOT3) AND (CVI_MPY0 OR CVI_MPY1)
- added support for combo resources
- allows specifying ORs of ANDs
- e.g. [CVI_XLSHF, CVI_MPY01] means:
(CVI_XLANE AND CVI_SHIFT) OR (CVI_MPY0 AND CVI_MPY1)
- increased DFA input size from 32-bit to 64-bit
- allows for a maximum of 4 AND'ed terms of 16 resources
- supported expressions now include:
expression => term [AND term] [AND term] [AND term]
term => resource [OR resource]*
resource => one_resource | combo_resource
combo_resource => (one_resource [AND one_resource]*)
Author: Dan Palermo <dpalermo@codeaurora.org>
kparzysz: Verified AMDGPU codegen to be unchanged on all llc
tests, except those dealing with instruction encodings.
llvm-svn: 253790
When MergeConsecutiveStores() combines two loads and two stores into
wider loads and stores, the chain users of both of the original loads
must be transfered to the new load, because it may be that a chain
user only depends on one of the loads.
New test case: test/CodeGen/SystemZ/dag-combine-01.ll
Reviewed by James Y Knight.
Bugzilla: https://llvm.org/bugs/show_bug.cgi?id=25310#c6
llvm-svn: 253779
It turns out we have a number of places that just grab the first type attached to a register class for various reasons. This is fine unless for some reason that type isn't legal on the current target, such as for SSE1 which doesn't support v16i8/v8i16/v4i32/v2i64 - all of which were included before 4f32 in the class.
Given that this is such a rare situation I've just re-ordered the types and placed the float types first.
Fix for PR16133
Differential Revision: http://reviews.llvm.org/D14787
llvm-svn: 253773
This assert was meant to execute at the end of parseMetadata, but
we return early and never reach the end of the function. Caught
by a compile-time warning since the function doesn't return a value
from that location.
llvm-svn: 253762
Summary:
Add and instructions immediately after loads that only have their low
bits used, assuming that the (and (load x) c) will be matched as a
extload and the ands/truncs fed by the extload will be removed by isel.
Reviewers: mcrosier, qcolombet, ab
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D14584
llvm-svn: 253722
The included test only checks for a compiler crash for now. Several people are
facing this issue, so we first resolve the crash, and will increase shrinkwrap's
coverage later in a follow-up patch.
llvm-svn: 253718
If a function was originally inlined but not actually hot at runtime,
its samples will not be counted inside the parent function. This throws
off the coverage calculation because it expects to find more used
records than it should.
Fixed by ignoring functions that will not be inlined into the parent.
Currently, this is inlined functions with 0 samples. In subsequent
patches, I'll change this to mean "cold" functions.
llvm-svn: 253716
This change merges adjacent zero stores into a wider single store.
For example :
strh wzr, [x0]
strh wzr, [x0, #2]
becomes
str wzr, [x0]
This will fix PR25410.
llvm-svn: 253711
incorrect, as the chosen representative of the weak symbol may not live
with the code in question. Always indirect the access through the TOC
instead.
Patch by Kyle Butt!
llvm-svn: 253708
This reverts r253661.
Turns out that the assignment is not redundant (despite the Clang static analyzer claiming the opposite).
The variable is being used by the lambda function AddUsersToWorklistIfCapturing().
llvm-svn: 253696
Summary:
This change refactors two aspects of InstrProfRecord:
1) Add a merge() method to InstrProfRecord (previously InstrProfWriter combineInstrProfRecords()) in order to better encapsulate this functionality and to make the InstrProfRecord and SampleRecord APIs more consistent.
2) Make InstrProfRecord mergeValueProfData() a private method since it is only ever called internally by merge().
Reviewers: dnovillo, bogner, davidxl
Subscribers: silvas, vsk, llvm-commits
Differential Revision: http://reviews.llvm.org/D14786
llvm-svn: 253695
Summary:
This follows D14577 to treat ARMv6-J as an alias for ARMv6,
instead of an architecture in its own right.
The functional change is that the default CPU when targeting ARMv6-J
changes from arm1136j-s to arm1136jf-s, which is currently used as
the default CPU for ARMv6; both are, in fact, ARMv6-J CPUs.
The J-bit (Jazelle support) is irrelevant to LLVM, and it doesn't
affect code generation, attributes, optimizations, or anything else,
apart from selecting the default CPU.
Reviewers: rengolin, logan, compnerd
Subscribers: aemerson, llvm-commits, rengolin
Differential Revision: http://reviews.llvm.org/D14755
llvm-svn: 253675
While debugging some sampling coverage problems, I found this useful:
When applying samples from a profile, it helps to also know what line
offset and discriminator the sample belongs to. This makes it easy to
correlate against the input profile.
llvm-svn: 253670
Summary:
This is split out from the ThinLTO metadata mapping patch
http://reviews.llvm.org/D14752.
To avoid needing to parse the module level metadata during function
importing, a new module-level record is added which holds the
number of module-level metadata values. This is required because
metadata value ids are assigned implicitly during parsing, and the
function-level metadata ids start after the module-level metadata ids.
I made a change to this version of the code compared to D14752
in order to add more consistent and thorough assertion checking of the
new record value. We now unconditionally use the record value to
initialize the MDValueList size, and handle it the same in parseMetadata
for all module level metadata cases (lazy loading or not).
Reviewers: dexonsmith, joker.eph
Subscribers: davidxl, llvm-commits, joker.eph
Differential Revision: http://reviews.llvm.org/D14825
llvm-svn: 253668
Terrifyingly, one of them is a mishandling of floating point vectors
in Constant::isZero(). How exactly this issue survived this long
is beyond me.
llvm-svn: 253655
WebAssembly does not have physical registers, so even if LLVM uses physical
registers like SP, they'll need to be lowered to virtual registers before
AsmPrinter time.
llvm-svn: 253644
The nuw constraint will not be satisfied unless <expr> == 0.
This bug has been around since r102234 (in 2010!), but was uncovered by
r251052, which introduced more aggressive optimization of nuw scev expressions.
Differential Revision: http://reviews.llvm.org/D14850
llvm-svn: 253627
coloring pass. Turn the logic into "look for an insert point and
then move things past the insert point".
No functional change intended.
llvm-svn: 253626
This adds a new API, LTOCodeGenerator::setFileType, to choose the output file
format for LTO CodeGen. A corresponding change to use this new API from
llvm-lto and a test case is coming in a separate commit.
Differential Revision: http://reviews.llvm.org/D14554
llvm-svn: 253622
Now that the register allocator knows about the barriers on funclet
entry and exit, testing has shown that this is unnecessary.
We still demote PHIs on unsplittable blocks due to the differences
between the IR CFG and the Machine CFG.
llvm-svn: 253619
The change exposed a bug in IndVarSimplify (PR25578), which led to a
failure (PR25538). When the bug is fixed, this patch can be reapplied.
The tests are kept in tree, as they're useful anyway, and will not break
with this revert.
llvm-svn: 253596
Summary: The new algorithm is more efficient (O(n), n is number of basic blocks). And it is guaranteed to cover all cases of multiple BB mapped to same line.
Reviewers: dblaikie, davidxl, dnovillo
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D14738
llvm-svn: 253594
Summary :
* Rename isSmallTypeLdMerge() to isNarrowLoad().
* Rename NumSmallTypeMerged to NumNarrowTypePromoted.
* Use Subtarget defined as a member variable.
llvm-svn: 253587
We currently bail out of global localization if the global has non-instruction users. However, often these can be simple bitcasts or constant-GEPs, which we can easily turn into instructions before localizing. Be a bit more aggressive.
llvm-svn: 253584
This change extends r251438 to handle more narrow load promotions
including byte type, unscaled, and signed. For example, this change will
convert :
ldursh w1, [x0, #-2]
ldurh w2, [x0, #-4]
into
ldur w2, [x0, #-4]
asr w1, w2, #16
and w2, w2, #0xffff
llvm-svn: 253577
This is another step towards allowing SimplifyCFG to speculate harder, but then have
CGP clean things up if the target doesn't like it.
Previous patches in this series:
http://reviews.llvm.org/D12882http://reviews.llvm.org/D13297
D13297 should catch most expensive ops, but speculation of cttz/ctlz requires special
handling because of weirdness in the intrinsic definition for handling a zero input
(that definition can probably be blamed on x86).
For example, if we have the usual speculated-by-select expensive op pattern like this:
%tobool = icmp eq i64 %A, 0
%0 = tail call i64 @llvm.cttz.i64(i64 %A, i1 true) ; is_zero_undef == true
%cond = select i1 %tobool, i64 64, i64 %0
ret i64 %cond
There's an instcombine that will turn it into:
%0 = tail call i64 @llvm.cttz.i64(i64 %A, i1 false) ; is_zero_undef == false
This CGP patch is looking for that case and despeculating it back into:
entry:
%tobool = icmp eq i64 %A, 0
br i1 %tobool, label %cond.end, label %cond.true
cond.true:
%0 = tail call i64 @llvm.cttz.i64(i64 %A, i1 true) ; is_zero_undef == true
br label %cond.end
cond.end:
%cond = phi i64 [ %0, %cond.true ], [ 64, %entry ]
ret i64 %cond
This unfortunately may lead to poorer codegen (see the changes in the existing x86 test),
but if we increase speculation in SimplifyCFG (the next step in this patch series), then
we should avoid those kinds of cases in the first place.
The need for this patch was originally mentioned here:
http://reviews.llvm.org/D7506
with follow-up here:
http://reviews.llvm.org/D7554
Differential Revision: http://reviews.llvm.org/D14630
llvm-svn: 253573
When dumping function samples or writing them out as text format, it
helps if the samples are emitted sorted by source location. The sorting
of the maps is a bit slow, so we only do it on demand.
llvm-svn: 253568
Copying one mask register to another under BW should be done with kmovq instruction, otherwise we can loose some bits.
Copying 8 bits under DQ may be done with kmovb.
Differential Revision: http://reviews.llvm.org/D14812
llvm-svn: 253563
The lowering patterns for X86ISD::VZEXT_MOVL for 128-bit to 256-bit vectors were just copying the lower xmm instead of actually masking off the first scalar using a blend.
Fix for PR25320.
Differential Revision: http://reviews.llvm.org/D14151
llvm-svn: 253561
Make X86AsmBackend generate smarter nops instead of a bunch of 0x90 for code alignment for CPUs which don't support long nop instructions.
Differential Revision: http://reviews.llvm.org/D14178
llvm-svn: 253557
This provides a way to force a function to have certain attributes from the command line. This can be useful when debugging or doing workload exploration, where manually editing IR is tedious or not possible (due to build systems etc).
The syntax is -force-attribute=function_name:attribute_name
All function attributes are parsed except alignstack as it requires an argument.
llvm-svn: 253550
The masked intrinsics support all integer and floating point data types. I added the pointer type to this list.
Added tests for CodeGen and for Loop Vectorizer.
Updated the Language Reference.
Differential Revision: http://reviews.llvm.org/D14150
llvm-svn: 253544
The LLVMContext was only used for Diagnostic. Pass a DiagnosticHandler
instead.
Differential Revision: http://reviews.llvm.org/D14794
From: Mehdi Amini <mehdi.amini@apple.com>
llvm-svn: 253540
Optimizations like LoadPRE in GVN will insert new instructions.
If the insertion point is in a already processed BB, they should
get a value number explicitly. If the insertion point is after
current instruction, then just leave it. However, current GVN framework
has no support for it.
In this patch, we just bail out if a VN can't be found.
Dfferential Revision: http://reviews.llvm.org/D14670
A test/Transforms/GVN/pr25440.ll
M lib/Transforms/Scalar/GVN.cpp
llvm-svn: 253536
This bug would manifest in some very specific cases where all the following
conditions are fullfilled:
- GVN didn't remove block
- The regular GVN iteration didn't change the IR
- PRE is enabled
- PRE will not split critical edge
- The last instruction processed by PRE didn't change the IR
Because the CallGraph PassManager relies on this returned value to decide
if it needs to recompute a node after the execution of Function passes,
not returning the right value can lead to unexpected results.
Fix for: https://llvm.org/bugs/show_bug.cgi?id=24715
Patch by Wenxiang Qiu <vincentqiuuu@gmail.com>
From: Mehdi Amini <mehdi.amini@apple.com>
llvm-svn: 253518
Note, this was reviewed (and more details are in) http://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20151109/312083.html
These intrinsics currently have an explicit alignment argument which is
required to be a constant integer. It represents the alignment of the
source and dest, and so must be the minimum of those.
This change allows source and dest to each have their own alignments
by using the alignment attribute on their arguments. The alignment
argument itself is removed.
There are a few places in the code for which the code needs to be
checked by an expert as to whether using only src/dest alignment is
safe. For those places, they currently take the minimum of src/dest
alignments which matches the current behaviour.
For example, code which used to read:
call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 500, i32 8, i1 false)
will now read:
call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 8 %dest, i8* align 8 %src, i32 500, i1 false)
For out of tree owners, I was able to strip alignment from calls using sed by replacing:
(call.*llvm\.memset.*)i32\ [0-9]*\,\ i1 false\)
with:
$1i1 false)
and similarly for memmove and memcpy.
I then added back in alignment to test cases which needed it.
A similar commit will be made to clang which actually has many differences in alignment as now
IRBuilder can generate different source/dest alignments on calls.
In IRBuilder itself, a new argument was added. Instead of calling:
CreateMemCpy(Dst, Src, getInt64(Size), DstAlign, /* isVolatile */ false)
you now call
CreateMemCpy(Dst, Src, getInt64(Size), DstAlign, SrcAlign, /* isVolatile */ false)
There is a temporary class (IntegerAlignment) which takes the source alignment and rejects
implicit conversion from bool. This is to prevent isVolatile here from passing its default
parameter to the source alignment.
Note, changes in future can now be made to codegen. I didn't change anything here, but this
change should enable better memcpy code sequences.
Reviewed by Hal Finkel.
llvm-svn: 253511
This patch adds support for vector constant folding of integer/float comparisons.
This requires FoldConstantVectorArithmetic to support scalar constant operands (in this case ISD::CONDCASE). In future we should be able to support other scalar constant types as necessary (and possibly start calling FoldConstantVectorArithmetic for all node creations)
Differential Revision: http://reviews.llvm.org/D14683
llvm-svn: 253504
It turns out we decide whether to use SjLj exceptions or some alternative in
two separate places in the backend, and they disagreed with each other. This
led to inconsistent code and is generally a terrible idea.
So make them consistent and add an assert that they *do* match (unfortunately
MCAsmInfo isn't available in opt, so it can't be used to initialise the CodeGen
version directly).
llvm-svn: 253502
This change introduces an instrumentation intrinsic instruction for
value profiling purposes, the lowering of the instrumentation intrinsic
and raw reader updates. The raw profile data files for llvm-profdata
testing are updated.
llvm-svn: 253484
This patch adds a cost estimate for some missing sign and zero extensions. The
costs were determined by counting the number of shift instructions generated
without context for each new extension.
Differential Revision: http://reviews.llvm.org/D14730
llvm-svn: 253482
This also takes the push/pop syntax another step forward, introducing stack
slot numbers to make it easier to see how expressions are connected. For
example, the value pushed in $push7 is popped in $pop7.
And, this begins an experiment with making get_local and set_local implicit
when an operation directly uses or defines a register. This greatly reduces
clutter. If this experiment succeeds, it may make sense to do this for
const instructions as well.
And, this introduces more special code for ARGUMENTS; hopefully this code
will soon be obviated by proper support for live-in virtual registers.
llvm-svn: 253465
Starting on an input stream that is not at offset 0 would trigger the
assert in WinCOFFObjectWriter.cpp:1065:
assert(getStream().tell() <= (*i)->Header.PointerToRawData &&
"Section::PointerToRawData is insane!");
llvm-svn: 253464
The virtual register containing the address for returned value on
stack should in the DAG be represented with a CopyFromReg node and not
a Register node. Otherwise, InstrEmitter will not make sure that it
ends up in the right register class for the target instruction.
SystemZ needs this, becuause the reg class for address registers is a
subset of the general 64 bit register class.
test/SystemZ/CodeGen/args-07.ll and args-04.ll updated to run with
-verify-machineinstrs.
Reviewed by Hal Finkel.
llvm-svn: 253461
We use to have an odd difference among MapVector and SetVector. The map
used a DenseMop, but the set used a SmallSet, which in turn uses a
std::set.
I have changed SetVector to use a DenseSet. If you were depending on the
old behaviour you can pass an explicit set type or use SmallSetVector.
The common cases for needing to do it are:
* Optimizing for small sets.
* Sets for types not supported by DenseSet.
llvm-svn: 253439
Summary:
This change teaches LLVM's inliner to track and suitably adjust
deoptimization state (tracked via deoptimization operand bundles) as it
inlines through call sites. The operation is described in more detail
in the LangRef changes.
Reviewers: reames, majnemer, chandlerc, dexonsmith
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D14552
llvm-svn: 253438
If a section is rw, it is irrelevant if the dynamic linker will write to
it or not.
It looks like llvm implemented this because gcc was doing it. It looks
like gcc implemented this in the hope that it would put all the
relocated items close together and speed up the dynamic linker.
There are two problem with this:
* It doesn't work. Both bfd and gold will map .data.rel to .data and
concatenate the input sections in the order they are seen.
* If we want a feature like that, it can be implemented directly in the
linker since it knowns where the dynamic relocations are.
llvm-svn: 253436
When looking for the best successor from the outer loop for a block
belonging to an inner loop, the edge probability computation can be
improved so that edges in the inner loop are ignored. For example,
suppose we are building chains for the non-loop part of the following
code, and looking for B1's best successor. Assume the true body is very
hot, then B3 should be the best candidate. However, because of the
existence of the back edge from B1 to B0, the probability from B1 to B3
can be very small, preventing B3 to be its successor. In this patch, when
computing the probability of the edge from B1 to B3, the weight on the
back edge B1->B0 is ignored, so that B1->B3 will have 100% probability.
if (...)
do {
B0;
... // some branches
B1;
} while(...);
else
B2;
B3;
Differential revision: http://reviews.llvm.org/D10825
llvm-svn: 253414
While still allowing CodeGen/AsmPrinter in llvm to own them using a bump
ptr allocator. (might be nice to replace the pointers there with
something that at least automatically calls their dtors, if that's
necessary/useful, rather than having it done explicitly (I think a typed
BumpPtrAllocator already does this, or maybe a unique_ptr with a custom
deleter, etc))
llvm-svn: 253409
The logic for handling the pattern without a shift is identical
to the logic for handling the pattern with a shift if you set
the shift amount to zero for the former.
This should make it easier to see that we probably don't even need
optimizeIntToFloatBitCast().
If we call something like foldVecTruncToExtElt() from visitTrunc(),
we'll solve PR25543:
https://llvm.org/bugs/show_bug.cgi?id=25543
llvm-svn: 253403
Summary:
This change tries to make the root cause of instrumented profile data merge failures clearer.
Previous:
$ llvm-profdata merge test_0.profraw test_1.profraw -o test_merged.profdata
test_1.profraw: foo: Function count mismatch
test_1.profraw: bar: Function count mismatch
test_1.profraw: baz: Function count mismatch
...
Changed:
$ llvm-profdata merge test_0.profraw test_1.profraw -o test_merged.profdata
test_1.profraw: foo: Function basic block count change detected (counter mismatch)
Make sure that all profile data to be merged is generated from the same binary.
test_1.profraw: bar: Function basic block count change detected (counter mismatch)
test_1.profraw: baz: Function basic block count change detected (counter mismatch)
...
Reviewers: dnovillo, davidxl, bogner
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D14739
llvm-svn: 253384
Summary:
Now that there is a one-to-one mapping from MachineFunction to
WinEHFuncInfo, we don't need to use a DenseMap to select the right
WinEHFuncInfo for the current funclet.
The main challenge here is that X86WinEHStatePass is an IR pass that
doesn't have access to the MachineFunction. I gave it its own
WinEHFuncInfo object that it uses to calculate state numbers, which it
then throws away. As long as nobody creates or removes EH pads between
this pass and SDAG construction, we will get the same state numbers.
The other thing X86WinEHStatePass does is to mark the EH registration
node. Instead of communicating which alloca was the registration through
WinEHFuncInfo, I added the llvm.x86.seh.ehregnode intrinsic. This
intrinsic generates no code and simply marks the alloca in use.
Reviewers: JCTremoulet
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D14668
llvm-svn: 253378
The instruction combiner previously removed types from filter clauses in Landing Pad instructions if the type had previously been seen in a catch clause. This is incorrect and prevents unexpected exception handlers from rethrowing the caught type.
Differential Revision: http://reviews.llvm.org/D14669
llvm-svn: 253370
When resolving R_PPC64_REL24, code used to check for an address delta
that fits in 24 bits, while the instructions that take this relocation
actually can process address deltas that fit into *26* bits (as those
instructions have a 24 bit field, but implicitly append two zero bits
at the end since all instruction addresses are a multiple of 4).
This means that code would signal overflow once a single object's text
section exceeds 8 MB, while we can actually support up to 32 MB.
Partially fixes PR25540.
llvm-svn: 253369
This patch removes the std::string& argument from a number of C++ LTO API calls
and instead makes them use the installed diagnostic handler. This would also
improve consistency of diagnostic handling infrastructure: if an LTO client used
lto_codegen_set_diagnostic_handler() to install a custom error handler, we do
not want some error messages to go through the custom error handler, and some
other error messages to go into sLastErrorString.
llvm-svn: 253367
While setting function attributes we check all instructions that may access memory. For a call instruction we check all arguments. The special check is required for pointers.
I added vector-of-pointers to the call arguments types that should be checked.
Differential Revision: http://reviews.llvm.org/D14693
llvm-svn: 253363
The underlying issues surrounding codegen for 32-bit vselects have been resolved. The pessimistic costs for 64-bit vselects remain due to the bad
scalarization that is still happening there.
I tested this on A57 in T32, A32 and A64 modes. I saw no regressions, and some improvements.
From my benchmarks, I saw these improvements in A57 (T32)
spec.cpu2000.ref.177_mesa 5.95%
lnt.SingleSource/Benchmarks/Shootout/strcat 12.93%
lnt.MultiSource/Benchmarks/MiBench/telecomm-CRC32/telecomm-CRC32 11.89%
I also measured A57 A32, A53 T32 and A9 T32 and found no performance regressions. I see much bigger wins in third-party benchmarks with this change
Differential Revision: http://reviews.llvm.org/D14743
llvm-svn: 253349
Summary:
This patch changes the behavior of path::system_temp_directory() on Windows to be closer to GetTempPath Windows API call. Enforces path separator to be the native one, makes path absolute, etc. GetTempPath is not used directly because of limitations/implementation bugs on Windows 7.
Windows specific unit tests are added. Most of them runs in separated process with modified environment variables.
This change fixes FileSystemTest.CreateDir unittest that had been failing when run from Unix-like shell on Windows (Unix-like path separator (/) used in env variables).
Reviewers: chapuni, rafael, aaron.ballman
Subscribers: rafael, llvm-commits
Differential Revision: http://reviews.llvm.org/D14231
llvm-svn: 253345
SELECT_CC has the nasty property of having operands with unrelated
types. So if you do something like:
f32 = select_cc f16, f16, f32, f32, cc
You'd only look for the action for <select_cc, f32>, but never f16.
If the types are all legal, but the op isn't (as for f16 on AArch64,
or for f128 on x86_64/AArch64?), then you get into trouble.
For f128, we have softenSetCCOperands to handle this case.
Similarly, for f16, we can directly promote the CC operands.
llvm-svn: 253344
setExecutable() should do everything that's needed to make the memory
executable on host, i.e. unconditionally set permissions + invalidate
instruction cache. llvm-rtdyld will be updated in my next commit.
Discusseed with: Lang Hames (as part of D13631).
llvm-svn: 253341
Statepoint lowering currently expects that the target method of a
statepoint only defines a single value. This precludes using
statepoints with ABIs that return values in multiple registers
(e.g. the SysV AMD64 ABI). This change adds support for lowering
statepoints with mutli-def targets.
llvm-svn: 253339
Several places in AsmPrinter.cpp print comments describing MachineOperand
registers using MCRegisterInfo, which uses MCOperand-oriented names. This
doesn't work for targets that use virtual registers exclusively, as
WebAssembly does, since virtual registers are represented and printed
differently.
This patch preserves what seems to be the spirit of r229978, avoiding the
use of TM.getSubtargetImpl(), while still using MachineOperand-oriented
printing for MachineOperands.
Differential Revision: http://reviews.llvm.org/D14709
llvm-svn: 253338
Currently, if the assembler encounters an error after parsing (such as an
out-of-range fixup), it reports this as a fatal error, and so stops after the
first error. However, for most of these there is an obvious way to recover
after emitting the error, such as emitting the fixup with a value of zero. This
means that we can report on all of the errors in a file, not just the first
one. MCContext::reportError records the fact that an error was encountered, so
we won't actually emit an object file with the incorrect contents.
Differential Revision: http://reviews.llvm.org/D14717
llvm-svn: 253328
This adds reportError to MCContext, which can be used as an alternative to
reportFatalError when the assembler wants to try to continue processing the
rest of the file after the error is reported, so that all of the errors ina
file can be reported. It records the fact that an error was encountered, so we
can avoid emitting an object file if any errors occurred.
This patch doesn't add any uses of this function (a later patch will convert
most uses of reportFatalError to use it), but there is a small functional
change: we use the SourceManager to print the error message, even if we have a
null SMLoc. This means that we get a SourceManager-style message, with the file
and line information shown as <unknown>, rather than the "LLVM ERROR" style
used by report_fatal_error.
llvm-svn: 253327
Indexed profile data as designed today does not guarantee
counter data to be well aligned, so reading needs to use
the slower form (with memcpy). This is less than ideal and
should be improved in the future (i.e., with fixed length
function key instead of variable length name key).
llvm-svn: 253309
The way prelink used to work was
* The compiler decides if a given section only has relocations that
are know to point to the same DSO. If so, it names it
.data.rel.ro.local<something>.
* The static linker puts all of these together.
* The prelinker program assigns addresses to each library and resolves
the local relocations.
There are many problems with this:
* It is incompatible with address space randomization.
* The information passed by the compiler is redundant. The linker
knows if a given relocation is in the same DSO or not. If could sort
by that if so desired.
* There are newer ways of speeding up DSO (gnu hash for example).
* Even if we want to implement this again in the compiler, the previous
implementation is pretty broken. It talks about relocations that are
"resolved by the static linker". If they are resolved, there are none
left for the prelinker. What one needs to track is if an expression
will require only dynamic relocations that point to the same DSO.
At this point it looks like the prelinker is an historical curiosity.
For example, fedora has retired it because it failed to build for two
releases
(http://pkgs.fedoraproject.org/cgit/prelink.git/commit/?id=eb43100a8331d91c801ee3dcdb0a0bb9babfdc1f)
This patch removes support for it. That is, it stops printing the
".local" sections.
llvm-svn: 253280
Sanjay Patel