This patch fixes a bug with the updating of DBG_VALUE's in
BreakAntiDependencies. Previously, it would only attempt to update the first
DBG_VALUE following the instruction whose register is being changed,
potentially leaving DBG_VALUE's referring to the wrong register. Now the code
will update all DBG_VALUE's that immediately follow the instruction.
This issue was detected as a result of an optimized codegen difference with
"-g" where an X86 byte/word fixup was not performed due to a DBG_VALUE
referencing the wrong register.
Differential Revision: https://reviews.llvm.org/D31755
llvm-svn: 301309
Summary:
The toolchain file has been deprecated in favor of the "official"
toolchain file present in the Android NDK. Also update the web
build instructions to reflect this.
Reviewers: eugene
Subscribers: srhines, mgorny, dgross, tberghammer, lldb-commits
Differential Revision: https://reviews.llvm.org/D32441
llvm-svn: 301306
One of the fast-math optimizations is to replace calls to standard double
functions with their float equivalents, e.g. exp -> expf. However, this can
cause infinite loops for the following:
float expf(float val) { return (float) exp((double) val); }
A similar inline declaration exists in the MinGW-w64 math.h header file which
when compiled with -O2/3 and fast-math generates infinite loops.
So this fix checks that the calling function to the standard double function
that is being replaced does not match the float equivalent.
Differential Revision: https://reviews.llvm.org/D31806
llvm-svn: 301304
Do not conflict if the value of Existing and Proposed are the same.
This change only affects unit tests, but no functional changes are
expected on LLVM-IR, as no Known information is yet extracted and
consequently this functionality is only triggered through unit tests.
Differential Revision: https://reviews.llvm.org/D32025
llvm-svn: 301301
Added a small change to the way pointer arguments are set in the kernel
code generation. The way the pointer is retrieved now, specifically requests
global address space to be annotated. This is necessary, if the IR should be
run through NVPTX to generate OpenCL compatible PTX.
The changes do not affect the PTX Strings generated for the CUDA target
(nvptx64-nvidia-cuda), but are necessary for OpenCL (nvptx64-nvidia-nvcl).
Additionally, the data layout has been updated to what the NVPTX Backend requests/recommends.
Contributed-by: Philipp Schaad
Reviewers: Meinersbur, grosser, bollu
Reviewed By: grosser, bollu
Subscribers: jlebar, pollydev, llvm-commits, nemanjai, yaxunl, Anastasia
Tags: #polly
Differential Revision: https://reviews.llvm.org/D32215
llvm-svn: 301299
Summary:
In a previous change I changed SCEV's normalization / denormalization
to work with non-affine add recs. So the bailout in IVUsers can be
removed.
Reviewers: atrick, efriedma
Reviewed By: atrick
Subscribers: davide, mcrosier, llvm-commits
Differential Revision: https://reviews.llvm.org/D32105
llvm-svn: 301298
This patch is part of D28975's breakdown.
Genreating the control-flow to guard predicated instructions modified to
only use SplitBlockAndInsertIfThen() for producing the if-then construct.
Differential Revision: https://reviews.llvm.org/D32224
llvm-svn: 301293
addIgnored defines a given symbol even if there is no existing
symbol with the same name. So, even if libc provides __tls_get_addr,
we should still be able to call addIgnored.
Differential Revision: https://reviews.llvm.org/D32053
llvm-svn: 301290
We need to do this to prevent a miscompile which sinks an objc_retain
past an objc_release that releases the object objc_retain retains. This
happens because the top-down and bottom-up traversals each determines
the insert point for retain or release individually without knowing
where the other instruction is moved.
For example, when the following IR is fed to the ARC optimizer, the
top-down traversal decides to insert objc_retain right before
objc_release and the bottom-up traversal decides to insert objc_release
right after clang.arc.use.
(IR before ARC optimizer)
%11 = call i8* @objc_retain(i8* %10)
call void (...) @clang.arc.use(%0* %5)
call void @llvm.dbg.value(...)
call void @objc_release(i8* %6)
This reverses the order of objc_release and objc_retain, which causes
the object to be destructed prematurely.
(IR after ARC optimizer)
call void (...) @clang.arc.use(%0* %5)
call void @objc_release(i8* %6)
call void @llvm.dbg.value(...)
%11 = call i8* @objc_retain(i8* %10)
rdar://problem/30530580
llvm-svn: 301289
When both, OccupiedAndKnown and Unused are given, use the former only
for the Known values. The relation Unused \union Occupied must always
hold.
This allows us to specify Known independently of Occupied. It is needed
for an artificial test case in https://reviews.llvm.org/D32025.
llvm-svn: 301284
BSD's __progname symbol is defined in crt1.o and linked against main
executables. The libc expects that main executables export __progname
symbol via .dynsym sections. In order to handle this case, we scan
undefined symbols in DSOs and exported them by setting Sym->ExportDynamic
to true.
But it turned out that setting that variable is not enough to make sure
that symbols are exported in all use cases. If a -dynamic-list option is
given, all symbols not explicitly mentioned in a version script are
hidden by default. That hides __progname symbol. This patch fixes the issue.
Fixes https://bugs.llvm.org/show_bug.cgi?id=32703
llvm-svn: 301282
Summary:
Before this change, SCEV Normalization would incorrectly normalize
non-affine add recurrences. To work around this there was (still is)
a check in place to make sure we only tried to normalize affine add
recurrences.
We recently found a bug in aforementioned check to bail out of
normalizing non-affine add recurrences. However, instead of fixing
the bailout, I have decided to teach SCEV normalization to work
correctly with non-affine add recurrences, making the bailout
unnecessary (I'll remove it in a subsequent change).
I've also added some unit tests (which would have failed before this
change).
Reviewers: atrick, sunfish, efriedma
Reviewed By: atrick
Subscribers: mcrosier, mzolotukhin, llvm-commits
Differential Revision: https://reviews.llvm.org/D32104
llvm-svn: 301281
This broke the self-host build on Windows (PR32777).
Original commit message:
> [builtins] Implement emulated TLS on Windows.
>
> Summary:
> LLVM JIT needs to be able to use emulated TLS on all platforms, and this provides a reference one can compile to enable emutls for Linux/Mac/Windows.
>
> Reviewers: chh, howard.hinnant
>
> Reviewed By: chh
>
> Subscribers: mgorny, llvm-commits
>
> Differential Revision: https://reviews.llvm.org/D30787
llvm-svn: 301274
Many times a user wants to access a type when there's a variable of
the same name, or a variable when there's a type of the same name.
Depending on the precise context, currently the expression parser
can fail to resolve one or the other.
This is because ClangExpressionDeclMap has logic to limit the
amount of information it searches, and that logic sometimes cuts
down the search prematurely. This patch removes some of those early
exits.
In that sense, this patch trades performance (early exit is faster)
for correctness.
I've also included two new test cases showing examples of this
behavior – as well as modifying an existing test case that gets it
wrong.
llvm-svn: 301273
I'm proposing a fold for increment-of-sexted-bool in:
https://reviews.llvm.org/D31944
...so we need to know what happens in more cases like these.
llvm-svn: 301269
Remove the temporary, poorly named getSlotSet method which did the same
thing. Also remove getSlotNode, which is a hold-over from when we were
dealing with AttributeSetNode* instead of AttributeSet.
llvm-svn: 301267
Earlier, the call to buildFlow was:
WAR = buildFlow(Write, Read, MustWrite, Schedule).
This meant that Read could block another Read, since must-sources can
block each other.
Fixed the call to buildFlow to correctly compute Read. The resulting
code needs to do some ISL juggling to get the output we want.
Bug report: https://bugs.llvm.org/show_bug.cgi?id=32623
Reviewers: Meinersbur
Tags: #polly
Differential Revision: https://reviews.llvm.org/D32011
llvm-svn: 301266
LLDB uses clang::DeclContexts for lookups, and variables get put into
the DeclContext for their abstract origin. (The abstract origin is a
DWARF pointer that indicates the unique definition of inlined code.)
When the expression parser is looking for variables, it locates the
DeclContext for the current context. This needs to be done carefully,
though, e.g.:
__attribute__ ((always_inline)) void f(int a) {
{
int b = a * 2;
}
}
void g() {
f(3);
}
Here, if we're stopped in the inlined copy of f, we have to find the
DeclContext corresponding to the definition of f – its abstract
origin. Clang doesn't allow multiple functions with the same name and
arguments to exist. It also means that any variables we see must be
placed in the appropriate DeclContext.
[Bug 1]: When stopped in an inline block, the function
GetDeclContextDIEContainingDIE for that block doesn't properly
construct a DeclContext for the abstract origin for inlined
subroutines. That means we get duplicated function DeclContexts, but
function arguments only get put in the abstract origin's DeclContext,
and as a result when we try to look for them in nested contexts they
aren't found.
[Bug 2]: When stopped in an inline block, the DWARF (for space
reasons) doesn't explicitly point to the abstract origin for that
block. This means that the function GetClangDeclContextForDIE returns
a different DeclContext for each place the block is inlined. However,
any variables defined in the block have abstract origins, so they
will only get placed in the DeclContext for their abstract origin.
In this fix, I've introduced a test covering both of these issues,
and fixed them.
Bug 1 could be resolved simply by making sure we look up the abstract
origin for inlined functions when looking up their DeclContexts on
behalf of nested blocks.
For Bug 2, I've implemented an algorithm that makes the DeclContext
for a block be the containing DeclContext for the closest entity we
would find during lookup that has an abstract origin pointer. That
means that in the following situation:
{ // block 1
int a;
{ // block 2
int b;
}
}
if we looked up the DeclContext for block 2, we'd find the block
containing the abstract origin of b, and lookup would proceed
correctly because we'd see b and a. However, in the situation
{ // block 1
int a;
{ // block 2
}
}
since there isn't anything to look up in block 2, we can't determine
its abstract origin (and there is no such pointer in the DWARF for
blocks). However, we can walk up the parent chain and find a, and its
abstract origin lives in the abstract origin of block 1. So we simply
say that the DeclContext for block 2 is the same as the DeclContext
for block 1, which contains a. Lookups will return the same results.
Thanks to Jim Ingham for review and suggestions.
Differential revision: https://reviews.llvm.org/D32375
llvm-svn: 301263
Summary:
`git apply` on Windows doesn't work for files that SVN checks out as
CRLF. There is no way to force SVN to check everything out with Unix
line endings on Windows. Files with svn:eol-style=native will always
come out with CRLF, breaking `git apply`, which wants Unix line endings.
My workaround is to list all files with this property set in the change,
and run `dos2unix` on them. SVN doesn't commit a massive line ending
change because the svn:eol-style property indicates that these are text
files.
Tested on r301245.
Reviewers: zturner, jlebar
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D32452
llvm-svn: 301262
Alex Lorenz