Violating the invariants specified by attributes is undefined behavior.
Maybe we could use poison instead for some of the parameter attributes,
but I don't think it's worthwhile.
Differential Revision: https://reviews.llvm.org/D49041
llvm-svn: 337947
This fixes a warning like this:
warning: comparison of integers of different signs:
'std::__1::__libcpp_tls_key' (aka 'long') and 'DWORD'
(aka 'unsigned long') [-Wsign-compare]
if (*__key == FLS_OUT_OF_INDEXES)
~~~~~~ ^ ~~~~~~~~~~~~~~~~~~
Differential Revision: https://reviews.llvm.org/D49782
llvm-svn: 337946
Saves materializing the immediate for the "ands".
Corresponding patterns exist for lsrs+lsls, but that seems less common
in practice.
Now implemented as a DAGCombine.
Differential Revision: https://reviews.llvm.org/D49585
llvm-svn: 337945
as well as sext(C + x + ...) -> (D + sext(C-D + x + ...))<nuw><nsw>
similar to the equivalent transformation for zext's
if the top level addition in (D + (C-D + x * n)) could be proven to
not wrap, where the choice of D also maximizes the number of trailing
zeroes of (C-D + x * n), ensuring homogeneous behaviour of the
transformation and better canonicalization of such AddRec's
(indeed, there are 2^(2w) different expressions in `B1 + ext(B2 + Y)` form for
the same Y, but only 2^(2w - k) different expressions in the resulting `B3 +
ext((B4 * 2^k) + Y)` form, where w is the bit width of the integral type)
This patch generalizes sext(C1 + C2*X) --> sext(C1) + sext(C2*X) and
sext{C1,+,C2} --> sext(C1) + sext{0,+,C2} transformations added in
r209568 relaxing the requirements the following way:
1. C2 doesn't have to be a power of 2, it's enough if it's divisible by 2
a sufficient number of times;
2. C1 doesn't have to be less than C2, instead of extracting the entire
C1 we can split it into 2 terms: (00...0XXX + YY...Y000), keep the
second one that may cause wrapping within the extension operator, and
move the first one that doesn't affect wrapping out of the extension
operator, enabling further simplifications;
3. C1 and C2 don't have to be positive, splitting C1 like shown above
produces a sum that is guaranteed to not wrap, signed or unsigned;
4. in AddExpr case there could be more than 2 terms, and in case of
AddExpr the 2nd and following terms and in case of AddRecExpr the
Step component don't have to be in the C2*X form or constant
(respectively), they just need to have enough trailing zeros,
which in turn could be guaranteed by means other than arithmetics,
e.g. by a pointer alignment;
5. the extension operator doesn't have to be a sext, the same
transformation works and profitable for zext's as well.
Apparently, optimizations like SLPVectorizer currently fail to
vectorize even rather trivial cases like the following:
double bar(double *a, unsigned n) {
double x = 0.0;
double y = 0.0;
for (unsigned i = 0; i < n; i += 2) {
x += a[i];
y += a[i + 1];
}
return x * y;
}
If compiled with `clang -std=c11 -Wpedantic -Wall -O3 main.c -S -o - -emit-llvm`
(!{!"clang version 7.0.0 (trunk 337339) (llvm/trunk 337344)"})
it produces scalar code with the loop not unrolled with the unsigned `n` and
`i` (like shown above), but vectorized and unrolled loop with signed `n` and
`i`. With the changes made in this commit the unsigned version will be
vectorized (though not unrolled for unclear reasons).
How it all works:
Let say we have an AddExpr that looks like (C + x + y + ...), where C
is a constant and x, y, ... are arbitrary SCEVs. Let's compute the
minimum number of trailing zeroes guaranteed of that sum w/o the
constant term: (x + y + ...). If, for example, those terms look like
follows:
i
XXXX...X000
YYYY...YY00
...
ZZZZ...0000
then the rightmost non-guaranteed-zero bit (a potential one at i-th
position above) can change the bits of the sum to the left (and at
i-th position itself), but it can not possibly change the bits to the
right. So we can compute the number of trailing zeroes by taking a
minimum between the numbers of trailing zeroes of the terms.
Now let's say that our original sum with the constant is effectively
just C + X, where X = x + y + .... Let's also say that we've got 2
guaranteed trailing zeros for X:
j
CCCC...CCCC
XXXX...XX00 // this is X = (x + y + ...)
Any bit of C to the left of j may in the end cause the C + X sum to
wrap, but the rightmost 2 bits of C (at positions j and j - 1) do not
affect wrapping in any way. If the upper bits cause a wrap, it will be
a wrap regardless of the values of the 2 least significant bits of C.
If the upper bits do not cause a wrap, it won't be a wrap regardless
of the values of the 2 bits on the right (again).
So let's split C to 2 constants like follows:
0000...00CC = D
CCCC...CC00 = (C - D)
and represent the whole sum as D + (C - D + X). The second term of
this new sum looks like this:
CCCC...CC00
XXXX...XX00
----------- // let's add them up
YYYY...YY00
The sum above (let's call it Y)) may or may not wrap, we don't know,
so we need to keep it under a sext/zext. Adding D to that sum though
will never wrap, signed or unsigned, if performed on the original bit
width or the extended one, because all that that final add does is
setting the 2 least significant bits of Y to the bits of D:
YYYY...YY00 = Y
0000...00CC = D
----------- <nuw><nsw>
YYYY...YYCC
Which means we can safely move that D out of the sext or zext and
claim that the top-level sum neither sign wraps nor unsigned wraps.
Let's run an example, let's say we're working in i8's and the original
expression (zext's or sext's operand) is 21 + 12x + 8y. So it goes
like this:
0001 0101 // 21
XXXX XX00 // 12x
YYYY Y000 // 8y
0001 0101 // 21
ZZZZ ZZ00 // 12x + 8y
0000 0001 // D
0001 0100 // 21 - D = 20
ZZZZ ZZ00 // 12x + 8y
0000 0001 // D
WWWW WW00 // 21 - D + 12x + 8y = 20 + 12x + 8y
therefore zext(21 + 12x + 8y) = (1 + zext(20 + 12x + 8y))<nuw><nsw>
This approach could be improved if we move away from using trailing
zeroes and use KnownBits instead. For instance, with KnownBits we could
have the following picture:
i
10 1110...0011 // this is C
XX X1XX...XX00 // this is X = (x + y + ...)
Notice that some of the bits of X are known ones, also notice that
known bits of X are interspersed with unknown bits and not grouped on
the rigth or left.
We can see at the position i that C(i) and X(i) are both known ones,
therefore the (i + 1)th carry bit is guaranteed to be 1 regardless of
the bits of C to the right of i. For instance, the C(i - 1) bit only
affects the bits of the sum at positions i - 1 and i, and does not
influence if the sum is going to wrap or not. Therefore we could split
the constant C the following way:
i
00 0010...0011 = D
10 1100...0000 = (C - D)
Let's compute the KnownBits of (C - D) + X:
XX1 1 = carry bit, blanks stand for known zeroes
10 1100...0000 = (C - D)
XX X1XX...XX00 = X
--- -----------
XX X0XX...XX00
Will this add wrap or not essentially depends on bits of X. Adding D
to this sum, however, is guaranteed to not to wrap:
0 X
00 0010...0011 = D
sX X0XX...XX00 = (C - D) + X
--- -----------
sX XXXX XX11
As could be seen above, adding D preserves the sign bit of (C - D) +
X, if any, and has a guaranteed 0 carry out, as expected.
The more bits of (C - D) we constrain, the better the transformations
introduced here canonicalize expressions as it leaves less freedom to
what values the constant part of ((C - D) + x + y + ...) can take.
Reviewed By: mzolotukhin, efriedma
Differential Revision: https://reviews.llvm.org/D48853
llvm-svn: 337943
Summary:
The VS compiler (on Windows) has a bug which results in fieldFromInstruction being optimized out in some circumstances. This only happens in *release no debug info* builds that have assertions *turned off* - in all other situations the function is not inlined, so the functionality is correct. All of the bots have assertions turned on, so this path is not regularly tested. The workaround is to not inline the function on Windows - if the bug is fixed in a later release of the VS compiler, the noinline specification can be removed.
The test that consistently reproduces this is Lanai v11.txt test.
Reviewers: asmith, labath, zturner
Subscribers: dblaikie, stella.stamenova, aprantl, JDevlieghere, llvm-commits
Differential Revision: https://reviews.llvm.org/D49753
llvm-svn: 337942
the children.
Special internal helper expressions/statements for the OpenMP directives
should not be exposed as children, only the main substatement must be
represented as the child.
llvm-svn: 337941
When VectorLegalizer::LegalizeOp creates a new SDValue after iterating
over its arguments, we need to refer to the same result number of the
new node that the original value used.
Reviewed by: cameron.mcinally
Differential Revision: https://reviews.llvm.org/D49805
llvm-svn: 337939
Currently ComputeNumSignBits does early exit while processing some
of the operations (add, sub, mul, and select). This prevents the
function from using AssumptionCacheTracker if passed.
Differential Revision: https://reviews.llvm.org/D49759
llvm-svn: 337936
For example v = <2 x i1> is represented as bbbbaaaa in a predicate register,
where b = v[1], a = v[0]. Extracting v[1] is equivalent to extracting bit 4
from the predicate register.
llvm-svn: 337934
This recommits r337910 after fixing an "ambiguous call to addAttribute"
error with some compilers (gcc circa 4.9 and MSVC). It seems that these
compilers will consider a "false -> pointer" conversion during overload
resolution. This creates ambiguity because one I added an overload which
takes a MCExpr * as an argument.
I fix this by making the new overload take MCExpr&, which avoids the
conversion. It also documents the fact that we expect a valid MCExpr
object.
Original commit message follows:
The motivation for this is D49493, where we'd like to test details of
debug_str_offsets behavior which is difficult to trigger from a
traditional test.
This adds the plubming necessary for dwarfgen to generate this section.
The more interesting changes are:
- I've moved emitStringOffsetsTableHeader function from DwarfFile to
DwarfStringPool, so I can generate the section header more easily from
the unit test.
- added a new addAttribute overload taking an MCExpr*. This is used to
generate the DW_AT_str_offsets_base, which links a compile unit to the
offset table.
I've also added a basic test for reading and writing DW_form_strx forms.
Reviewers: dblaikie, JDevlieghere, probinson
Subscribers: llvm-commits, aprantl
Differential Revision: https://reviews.llvm.org/D49670
llvm-svn: 337933
Summary:
Replace the existing combination of FastDemangle and the fallback to llvm::itaniumDemangle() with LLVM's new ItaniumPartialDemangler. It slightly reduces complexity and slightly improves performance, but doesn't introduce conceptual changes. This patch is preparing for more fundamental improvements on LLDB's demangling approach.
Reviewers: friss, jingham, erik.pilkington, labath, clayborg, mgorny, davide, JDevlieghere
Reviewed By: JDevlieghere
Subscribers: teemperor, JDevlieghere, labath, clayborg, davide, lldb-commits, mgorny, erik.pilkington
Differential Revision: https://reviews.llvm.org/D49612
llvm-svn: 337931
Initially, in https://reviews.llvm.org/D44890, I had these defined as
empty functions inside the header when the respective event listener
was not built in. As done in that commit, that wasn't correct, because
it was a ODR violation. Krasimir hot-fixed that in r333265, but that
wasn't quite right either, because it'd lead to the symbol not being
available.
Instead just move the fallbacksto ExecutionEngineBindings.cpp. Could
define them as static inlines in the header too, but I don't think it
matters.
Reviewers: whitequark
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D49654
llvm-svn: 337930
By default, xray filters events that takes less than 5uS from its log.
In this existing test, should printf complete very quickly this will
lead to test-critical function calls being filtered (i.e. print_parent_tid).
Given that we're not testing the filtering feature, disable it for this
test.
llvm-svn: 337929
This reverts commit r337910 as it's generating "ambiguous call to
addAttribute" errors on some bots.
Will resubmit once I get a chance to look into the problem.
llvm-svn: 337924
Summary:
The macro was manually expanded in the Z3 backend and this patch adds it back.
Adding the expanded code is dangerous as the macro may change in the future and the expanded code might be left outdated.
Reviewers: NoQ, george.karpenkov
Reviewed By: george.karpenkov
Subscribers: xazax.hun, szepet, a.sidorin
Differential Revision: https://reviews.llvm.org/D49769
llvm-svn: 337923
Summary:
Third patch in the refactoring series, to decouple the SMT Solver from the Refutation Manager (1st: D49668, 2nd: D49767).
The refutation API in the `SMTConstraintManager` was a hack to allow us to create an SMT solver and verify the constraints; it was conceptually wrong from the start. Now, we don't actually need to use the `SMTConstraintManager` and can create an SMT object directly, add the constraints and check them.
While updating the Falsification visitor, I inlined the two functions that were used to collect the constraints and add them to the solver.
As a result of this patch, we could move the SMT API elsewhere and as it's not really dependent on the CSA anymore. Maybe we can create a new dir (utils/smt) for Z3 and future solvers?
Reviewers: NoQ, george.karpenkov
Reviewed By: george.karpenkov
Subscribers: xazax.hun, szepet, a.sidorin
Differential Revision: https://reviews.llvm.org/D49768
llvm-svn: 337922
Summary:
This is the second part of D49668, and moves all the code that's not specific to a ConstraintManager to SMTSolver.
No functional change intended.
Reviewers: NoQ, george.karpenkov
Reviewed By: george.karpenkov
Subscribers: xazax.hun, szepet, a.sidorin
Differential Revision: https://reviews.llvm.org/D49767
llvm-svn: 337921
Summary:
This patch changes how the SMT bug refutation runs in an equivalent bug report class.
Now, all other visitor are executed until they find a valid bug or mark all bugs as invalid. When the one valid bug is found (and crosscheck is enabled), the SMT refutation checks the satisfiability of this single bug.
If the bug is still valid after checking with Z3, it is returned and a bug report is created. If the bug is found to be invalid, the next bug report in the equivalent class goes through the same process, until we find a valid bug or all bugs are marked as invalid.
Massive speedups when verifying redis/src/rax.c, from 1500s to 10s.
Reviewers: NoQ, george.karpenkov
Reviewed By: george.karpenkov
Subscribers: xazax.hun, szepet, a.sidorin
Differential Revision: https://reviews.llvm.org/D49693
llvm-svn: 337920
Summary:
This patch moves a lot of code from `Z3ConstraintManager` to `SMTConstraintManager`, leaving only the necessary:
* `canReasonAbout` which returns if a Solver can handle a given `SVal` (should be moved to `SMTSolver` in the future).
* `removeDeadBindings`, `assumeExpr` and `print`: methods that need to use `ConstraintZ3Ty`, can probably be moved to `SMTConstraintManager` in the future.
The patch creates a new file, `SMTConstraintManager.cpp` with the moved code. Conceptually, this is move in the right direction and needs further improvements: `SMTConstraintManager` still does a lot of things that are not required by a `ConstraintManager`.
We ought to move the unrelated to `SMTSolver` and remove everything that's not related to a `ConstraintManager`. In particular, we could remove `addRangeConstraints` and `isModelFeasible`, and make the refutation manager create an Z3Solver directly.
Reviewers: NoQ, george.karpenkov
Reviewed By: george.karpenkov
Subscribers: mgorny, xazax.hun, szepet, a.sidorin
Differential Revision: https://reviews.llvm.org/D49668
llvm-svn: 337919
Summary:
Created new SMT generic API.
Small changes to `Z3ConstraintManager` because of the new generic objects (`SMTSort` and `SMTExpr`) returned by `SMTSolver`.
Reviewers: george.karpenkov, NoQ
Reviewed By: george.karpenkov
Subscribers: mgorny, xazax.hun, szepet, a.sidorin
Differential Revision: https://reviews.llvm.org/D49495
llvm-svn: 337918
Summary:
New base class for all future SMT Exprs.
No major changes except moving `areEquivalent` and `getFloatSemantics` outside of `Z3Expr` to keep the class minimal.
Reviewers: NoQ, george.karpenkov
Reviewed By: george.karpenkov
Subscribers: xazax.hun, szepet, a.sidorin
Differential Revision: https://reviews.llvm.org/D49551
llvm-svn: 337917
Summary:
New base class for all future SMT sorts.
The only change is that the class implements methods `isBooleanSort()`, `isBitvectorSort()` and `isFloatSort()` so it doesn't rely on `Z3`'s enum.
Reviewers: NoQ, george.karpenkov
Reviewed By: george.karpenkov
Subscribers: xazax.hun, szepet, a.sidorin
Differential Revision: https://reviews.llvm.org/D49550
llvm-svn: 337916
Summary:
Although it is a big patch, the changes are simple:
1. There is one `Z3_Context` now, member of the `SMTConstraintManager` class.
2. `Z3Expr`, `Z3Sort`, `Z3Model` and `Z3Solver` are constructed with a reference to the `Z3_Context` in `SMTConstraintManager`.
3. All static functions are now members of `Z3Solver`, e.g, the `SMTConstraintManager` now calls `Solver.fromBoolean(false)` instead of `Z3Expr::fromBoolean(false)`.
Most of the patch only move stuff around except:
1. New method `Z3Sort MkSort(const QualType &Ty, unsigned BitWidth)`, that creates a sort based on the `QualType` and its width. Used to simplify the `fromData` method.
Unfortunate consequence of this patch:
1. `getInterpretation` was moved from `Z3Model` class to `Z3Solver`, because it needs to create a `Z3Sort` before returning the interpretation. This can be fixed by changing both `toAPFloat` and `toAPSInt` by removing the dependency of `Z3Sort` (it's only used to check which Sort was created and to retrieve the type width).
Reviewers: NoQ, george.karpenkov, ddcc
Reviewed By: george.karpenkov
Subscribers: xazax.hun, szepet, a.sidorin
Differential Revision: https://reviews.llvm.org/D49236
llvm-svn: 337915
Summary:
This patch creates `SMTContext` which will wrap a specific SMT context, through `SMTSolverContext`.
The templated `SMTSolverContext` class it's a simple wrapper around a SMT specific context (currently only used in the Z3 backend), while `Z3Context` inherits `SMTSolverContext<Z3_context>` and implements solver specific operations like initialization and destruction of the context.
This separation was done because:
1. We might want to keep one single context, shared across different `SMTConstraintManager`s. It can be achieved by constructing a `SMTContext`, through a function like `CreateSMTContext(Z3)`, `CreateSMTContext(BOOLECTOR)`, etc. The rest of the CSA only need to know about `SMTContext`, so maybe it's a good idea moving `SMTSolverContext` to a separate header in the future.
2. Any generic SMT operation will only require one `SMTSolverContext`object, which can access the specific context by calling `getContext()`.
Reviewers: NoQ, george.karpenkov
Reviewed By: george.karpenkov
Subscribers: xazax.hun, szepet, a.sidorin
Differential Revision: https://reviews.llvm.org/D49233
llvm-svn: 337914
Add support for lowering pointer arguments.
Changing type from pointer to integer is already done in
MipsTargetLowering::getRegisterTypeForCallingConv.
Patch by Petar Avramovic.
Differential Revision: https://reviews.llvm.org/D49419
llvm-svn: 337912
Summary:
The motivation for this is D49493, where we'd like to test details of
debug_str_offsets behavior which is difficult to trigger from a
traditional test.
This adds the plubming necessary for dwarfgen to generate this section.
The more interesting changes are:
- I've moved emitStringOffsetsTableHeader function from DwarfFile to
DwarfStringPool, so I can generate the section header more easily from
the unit test.
- added a new addAttribute overload taking an MCExpr*. This is used to
generate the DW_AT_str_offsets_base, which links a compile unit to the
offset table.
I've also added a basic test for reading and writing DW_form_strx forms.
Reviewers: dblaikie, JDevlieghere, probinson
Subscribers: llvm-commits, aprantl
Differential Revision: https://reviews.llvm.org/D49670
llvm-svn: 337910
NFC changes to make scheduler TableGen files more readable, by using loops
instead of a lot of similar defs with just e.g. a latency value that changes.
https://reviews.llvm.org/D49598
Review: Ulrich Weigand, Javed Abshar
llvm-svn: 337909
Summary:
This has a shape to similar logarithm function but grows much slower for
large #usages.
Metrics: https://reviews.llvm.org/P8096
Reviewers: ilya-biryukov
Reviewed By: ilya-biryukov
Subscribers: MaskRay, jkorous, arphaman, cfe-commits, sammccall
Differential Revision: https://reviews.llvm.org/D49780
llvm-svn: 337907
Tuple has tests that ensure we diagnose non-lifetime extended
reference bindings inside tuples constructors. As of yesterday,
Clang now does this for us.
Adjust the test to tolerate the new diagnostics, while still
testing that we emit diagnostics of our own. Maybe after this
version of Clang has been adopted by most users we should
remove our diagnostics; but for now more error detection is
better!
llvm-svn: 337905
r337828 resolves a PredicateInfo issue with unnamed types.
Original message:
This patch updates IPSCCP to use PredicateInfo to propagate
facts to true branches predicated by EQ and to false branches
predicated by NE.
As a follow up, we should be able to extend it to also propagate additional
facts about nonnull.
Reviewers: davide, mssimpso, dberlin, efriedma
Reviewed By: davide, dberlin
llvm-svn: 337904
Add support for inline assembly with output operand that do not
naturally go in the register class it is constrained to (eg. double in a
32-bit GPR as in the PR).
llvm-svn: 337903
This patch introduces the core building block of the next-generation
Clangd symbol index - Dex. Search tokens are the keys in the inverted
index and represent a characteristic of a specific symbol: examples of
search token types (Token Namespaces) are
* Trigrams - these are essential for unqualified symbol name fuzzy
search * Scopes for filtering the symbols by the namespace * Paths, e.g.
these can be used to uprank symbols defined close to the edited file
This patch outlines the generic for such token namespaces, but only
implements trigram generation.
The intuition behind trigram generation algorithm is that each extracted
trigram is a valid sequence for Fuzzy Matcher jumps, proposed
implementation utilize existing FuzzyMatcher API for segmentation and
trigram extraction.
However, trigrams generation algorithm for the query string is different
from the previous one: it simply yields sequences of 3 consecutive
lowercased valid characters (letters, digits).
Dex RFC in the mailing list:
http://lists.llvm.org/pipermail/clangd-dev/2018-July/000022.html
The trigram generation techniques are described in detail in the
proposal:
https://docs.google.com/document/d/1C-A6PGT6TynyaX4PXyExNMiGmJ2jL1UwV91Kyx11gOI/edit#heading=h.903u1zon9nkj
Reviewers: sammccall, ioeric, ilya-biryukovA
Subscribers: cfe-commits, klimek, mgorny, MaskRay, jkorous, arphaman
Differential Revision: https://reviews.llvm.org/D49591
llvm-svn: 337901
Summary:
Extend the Clang-Format IncludeCategories documentation by adding a link to the supported regular expression standard (POSIX).
And extenting the example with a system header regex.
[[ https://bugs.llvm.org/show_bug.cgi?id=35041 | bug 35041]]
Contributed by WimLeflere!
Reviewers: krasimir, Typz
Reviewed By: krasimir
Subscribers: cfe-commits
Differential Revision: https://reviews.llvm.org/D48827
llvm-svn: 337899
Mikhail R. Gadelha