We were overcounting the number of arithmetic operations needed at each level before we reach a legal type. We were using the full vector type for that level, but we are going to split the input vector at that level in half. So the effective arithmetic operation cost at that level is half the width.
So for example on 8i32 on an sse target. Were were calculating the cost of an 8i32 op which is likely 2 for basic integer. Then after the loop we count 2 more v4i32 ops. For a total arith cost of 4. But if you look at the assembly there would only be 3 arithmetic ops.
There are still more bugs in this code that I'm going to work on next. The non pairwise code shouldn't count extract subvectors in the loop. There are no extracts, the types are split in registers. For pairwise we need to use 2 two src permute shuffles.
Differential Revision: https://reviews.llvm.org/D55397
llvm-svn: 348621
To make X86CondBrFoldingPass can be run with --run-pass option, this can test one wrong assertion on analyzeCompare function for SUB32ri when its operand is not imm
Patch by Jianping Chen
Differential Revision: https://reviews.llvm.org/D55412
llvm-svn: 348620
This patch attempts to improve pfm perf counter coverage for all the x86 CPUs that libpfm4 supports.
Intel/AMD CPU families tend to share names for cycle/uops counters so even if they don't have a scheduler model yet they can at least use the default values (checked against the libpfm4 source code).
The remaining CPUs (where their port/pipe resource counters are known) I've tried to add to the existing model mappings.
These are untested but don't represent a regression to current llvm-exegesis behaviour for these CPUs.
Differential Revision: https://reviews.llvm.org/D55432
llvm-svn: 348617
Summary:
We introduce a strict policy for C++ CTU. It can work across TUs only if
the C++ dialects are the same. We neither allow C vs C++ CTU. We do this
because the same constructs might be represented with different properties in
the corresponding AST nodes or even the nodes might be completely different (a
struct will be RecordDecl in C, but it will be a CXXRectordDecl in C++, thus it
may cause certain assertions during cast operations).
Reviewers: xazax.hun, a_sidorin
Subscribers: rnkovacs, dkrupp, Szelethus, gamesh411, cfe-commits
Differential Revision: https://reviews.llvm.org/D55134
llvm-svn: 348610
Summary:
Introduced special noinline function log that allows to save some
registers for optimized builds but with enabled logging. Also, it
increases the stability of the optimized builds with inlined runtime.
Reviewers: gtbercea, kkwli0
Reviewed By: gtbercea
Subscribers: caomhin, guansong, openmp-commits
Differential Revision: https://reviews.llvm.org/D55436
llvm-svn: 348606
Summary:
Adding some more CTU list tests. E.g. to check if a construct is unsupported.
We also slightly modify the handling of the return value of the `Import`
function from ASTImporter.
Reviewers: xazax.hun, balazske, a_sidorin
Subscribers: rnkovacs, dkrupp, Szelethus, gamesh411, cfe-commits
Differential Revision: https://reviews.llvm.org/D55131
llvm-svn: 348605
As discussed in the post-commit thread of r347917, this
transform is fighting with an existing transform causing
an infinite loop or out-of-memory, so this is effectively
reverting r347917 and its follow-up r348195 while we
investigate the bug.
llvm-svn: 348604
DemandedBits and BDCE currently only support scalar integers. This
patch extends them to also handle vector integer operations. In this
case bits are not tracked for individual vector elements, instead a
bit is demanded if it is demanded for any of the elements. This matches
the behavior of computeKnownBits in ValueTracking and
SimplifyDemandedBits in InstCombine.
Unlike the previous iteration of this patch, getDemandedBits() can now
again be called on arbirary (sized) instructions, even if they don't
have integer or vector of integer type. (For vector types the size of the
returned mask will now be the scalar size in bits though.)
The added LoopVectorize test case shows a case which triggered an
assertion failure with the previous attempt, because getDemandedBits()
was called on a pointer-typed instruction.
Differential Revision: https://reviews.llvm.org/D55297
llvm-svn: 348602
This change attempts to shrink scalar AND, OR and XOR instructions which take an immediate that isn't inlineable.
It performs:
AND s0, s0, ~(1 << n) -> BITSET0 s0, n
OR s0, s0, (1 << n) -> BITSET1 s0, n
AND s0, s1, x -> ANDN2 s0, s1, ~x
OR s0, s1, x -> ORN2 s0, s1, ~x
XOR s0, s1, x -> XNOR s0, s1, ~x
In particular, this catches setting and clearing the sign bit for fabs (and x, 0x7ffffffff -> bitset0 x, 31 and or x, 0x80000000 -> bitset1 x, 31).
llvm-svn: 348601
Inline cpu_specific versions referenced before the cpu_dispatch function
weren't properly emitted, since they hadn't been referred to. This
patch ensures that during resolver generation that all appropriate
versions are emitted.
Change-Id: I94c3766aaf9c75ca07a0ad8258efdbb834654ff8
llvm-svn: 348600
This reverts commit 65df29f9318ac13a633c0ce13b2b0bccf06e79ca.
AS suggested by @rsmith here: https://reviews.llvm.org/rL345839
I'm reverting this and solving the initial problem in a different way.
llvm-svn: 348595
Summary:
With a new switch we may be able to print to stderr if a new TU is being loaded
during CTU. This is very important for higher level scripts (like CodeChecker)
to be able to parse this output so they can create e.g. a zip file in case of
a Clang crash which contains all the related TU files.
Reviewers: xazax.hun, Szelethus, a_sidorin, george.karpenkov
Subscribers: whisperity, baloghadamsoftware, szepet, rnkovacs, a.sidorin, mikhail.ramalho, donat.nagy, dkrupp,
Differential Revision: https://reviews.llvm.org/D55135
llvm-svn: 348594
This patch introduces a new instinsic `@llvm.experimental.widenable_condition`
that allows explicit representation for guards. It is an alternative to using
`@llvm.experimental.guard` intrinsic that does not contain implicit control flow.
We keep finding places where `@llvm.experimental.guard` is not supported or
treated too conservatively, and there are 2 reasons to that:
- `@llvm.experimental.guard` has memory write side effect to model implicit control flow,
and this sometimes confuses passes and analyzes that work with memory;
- Not all passes and analysis are aware of the semantics of guards. These passes treat them
as regular throwing call and have no idea that the condition of guard may be used to prove
something. One well-known place which had caused us troubles in the past is explicit loop
iteration count calculation in SCEV. Another example is new loop unswitching which is not
aware of guards. Whenever a new pass appears, we potentially have this problem there.
Rather than go and fix all these places (and commit to keep track of them and add support
in future), it seems more reasonable to leverage the existing optimizer's logic as much as possible.
The only significant difference between guards and regular explicit branches is that guard's condition
can be widened. It means that a guard contains (explicitly or implicitly) a `deopt` block successor,
and it is always legal to go there no matter what the guard condition is. The other successor is
a guarded block, and it is only legal to go there if the condition is true.
This patch introduces a new explicit form of guards alternative to `@llvm.experimental.guard`
intrinsic. Now a widenable guard can be represented in the CFG explicitly like this:
%widenable_condition = call i1 @llvm.experimental.widenable.condition()
%new_condition = and i1 %cond, %widenable_condition
br i1 %new_condition, label %guarded, label %deopt
guarded:
; Guarded instructions
deopt:
call type @llvm.experimental.deoptimize(<args...>) [ "deopt"(<deopt_args...>) ]
The new intrinsic `@llvm.experimental.widenable.condition` has semantics of an
`undef`, but the intrinsic prevents the optimizer from folding it early. This form
should exploit all optimization boons provided to `br` instuction, and it still can be
widened by replacing the result of `@llvm.experimental.widenable.condition()`
with `and` with any arbitrary boolean value (as long as the branch that is taken when
it is `false` has a deopt and has no side-effects).
For more motivation, please check llvm-dev discussion "[llvm-dev] Giving up using
implicit control flow in guards".
This patch introduces this new intrinsic with respective LangRef changes and a pass
that converts old-style guards (expressed as intrinsics) into the new form.
The naming discussion is still ungoing. Merging this to unblock further items. We can
later change the name of this intrinsic.
Reviewed By: reames, fedor.sergeev, sanjoy
Differential Revision: https://reviews.llvm.org/D51207
llvm-svn: 348593
Summary:
This patch adds the scaffolding necessary for lldb to recognise symbol
files generated by breakpad. These (textual) files contain just enough
information to be able to produce a backtrace from a crash
dump. This information includes:
- UUID, architecture and name of the module
- line tables
- list of symbols
- unwind information
A minimal breakpad file could look like this:
MODULE Linux x86_64 0000000024B5D199F0F766FFFFFF5DC30 a.out
INFO CODE_ID 00000000B52499D1F0F766FFFFFF5DC3
FILE 0 /tmp/a.c
FUNC 1010 10 0 _start
1010 4 4 0
1014 5 5 0
1019 5 6 0
101e 2 7 0
PUBLIC 1010 0 _start
STACK CFI INIT 1010 10 .cfa: $rsp 8 + .ra: .cfa -8 + ^
STACK CFI 1011 $rbp: .cfa -16 + ^ .cfa: $rsp 16 +
STACK CFI 1014 .cfa: $rbp 16 +
Even though this data would normally be considered "symbol" information,
in the current lldb infrastructure it is assumed every SymbolFile object
is backed by an ObjectFile instance. So, in order to better interoperate
with the rest of the code (particularly symbol vendors).
In this patch I just parse the breakpad header, which is enough to
populate the UUID and architecture fields of the ObjectFile interface.
The rough plan for followup patches is to expose the individual parts of
the breakpad file as ObjectFile "sections", which can then be used by
other parts of the codebase (SymbolFileBreakpad ?) to vend the necessary
information.
Reviewers: clayborg, zturner, lemo, amccarth
Subscribers: mgorny, fedor.sergeev, markmentovai, lldb-commits
Differential Revision: https://reviews.llvm.org/D55214
llvm-svn: 348592
When we had dynamic call frames (i.e. sp adjustment around each call) we
were including that adjustment into offsets calculated based on r6, even
though it's only sp that changes. This led to incorrect stack slot
accesses.
llvm-svn: 348591
Summary:
...that fires when running completion inside an argument of
UnresolvedMemberExpr (see the added test).
The assertion that fires is from Sema::TryObjectArgumentInitialization:
assert(FromClassification.isLValue());
This happens because Sema::AddFunctionCandidates does not account for
object types which are pointers. It ends up classifying them incorrectly.
All usages of the function outside code completion are used to run
overload resolution for operators. In those cases the object type being
passed is always a non-pointer type, so it's not surprising the function
did not expect a pointer in the object argument.
However, code completion reuses the same function and calls it with the
object argument coming from UnresolvedMemberExpr, which can be a pointer
if the member expr is an arrow ('->') access.
Extending AddFunctionCandidates to allow pointer object types does not
seem too crazy since all the functions down the call chain can properly
handle pointer object types if we properly classify the object argument
as an l-value, i.e. the classification of the implicitly dereferenced
pointer.
Reviewers: kadircet
Reviewed By: kadircet
Subscribers: cfe-commits
Differential Revision: https://reviews.llvm.org/D55331
llvm-svn: 348590
Summary:
We plan to introduce additional CTU related lit test. Since lit may run the
tests in parallel, it is not safe to use the same directory (%T) for these
tests. It is safe to use however test case specific directories (%t).
Reviewers: xazax.hun, a_sidorin
Subscribers: rnkovacs, dkrupp, Szelethus, gamesh411, cfe-commits
Differential Revision: https://reviews.llvm.org/D55129
llvm-svn: 348587
Adds fatal errors for any target that does not support the Tiny or Kernel
codemodels by rejigging the getEffectiveCodeModel calls.
Differential Revision: https://reviews.llvm.org/D50141
llvm-svn: 348585
Summary: This line is longer than 80 characters.
Subscribers: llvm-commits, jakehehrlich
Differential Revision: https://reviews.llvm.org/D55419
llvm-svn: 348580
Summary: This line is longer than 80 characters.
Subscribers: llvm-commits, jakehehrlich
Differential Revision: https://reviews.llvm.org/D55419
llvm-svn: 348578
It was failing as below. Adding a triple seems to help.
--
: 'RUN: at line 2'; /work/llvm.combined/build.release/bin/llvm-mca -march=aarch64 -mcpu=exynos-m1 -resource-pressure=false < /work/llvm.combined/llvm/test/tools/llvm-mca/AArch64/Exynos/direct-branch.s | /work/llvm.combined/build.release/bin/FileCheck /work/llvm.combined/llvm/test/tools/llvm-mca/AArch64/Exynos/direct-branch.s -check-prefixes=ALL,M1
: 'RUN: at line 3'; /work/llvm.combined/build.release/bin/llvm-mca -march=aarch64 -mcpu=exynos-m3 -resource-pressure=false < /work/llvm.combined/llvm/test/tools/llvm-mca/AArch64/Exynos/direct-branch.s | /work/llvm.combined/build.release/bin/FileCheck /work/llvm.combined/llvm/test/tools/llvm-mca/AArch64/Exynos/direct-branch.s -check-prefixes=ALL,M3
--
Exit Code: 1
Command Output (stderr):
--
/work/llvm.combined/llvm/test/tools/llvm-mca/AArch64/Exynos/direct-branch.s:36:12: error: M1-NEXT: expected string not found in input
^
<stdin>:21:2: note: scanning from here
1 0 0.25 b Ltmp0
^
--
llvm-svn: 348577
Sanjay Patel