The previously implemented approach is to follow value definitions and
create write accesses ("push defs") while searching for uses. This
requires the same relatively validity- and requirement conditions to be
replicated at multiple locations (PHI instructions, other instructions,
uses by PHIs).
We replace this by iterating over the uses in a SCoP ("pull in
requirements"), and add writes only when at least one read has been
added. It turns out to be simpler code because each use is only iterated
over once and writes are added for the first access that reads it. We
need another iteration to identify escaping values (uses not in the
SCoP), which also makes the difference between such accesses more
obvious. As a side-effect, the order of scalar MemoryAccess can change.
Differential Revision: http://reviews.llvm.org/D15706
llvm-svn: 259987
Summary:
When alias analysis is uncertain about the aliasing between any two accesses,
it will return MayAlias. This uncertainty from alias analysis restricts LICM
from proceeding further. In cases where alias analysis is uncertain we might
use loop versioning as an alternative.
Loop Versioning will create a version of the loop with aggressive aliasing
assumptions in addition to the original with conservative (default) aliasing
assumptions. The version of the loop making aggressive aliasing assumptions
will have all the memory accesses marked as no-alias. These two versions of
loop will be preceded by a memory runtime check. This runtime check consists
of bound checks for all unique memory accessed in loop, and it ensures the
lack of memory aliasing. The result of the runtime check determines which of
the loop versions is executed: If the runtime check detects any memory
aliasing, then the original loop is executed. Otherwise, the version with
aggressive aliasing assumptions is used.
The pass is off by default and can be enabled with command line option
-enable-loop-versioning-licm.
Reviewers: hfinkel, anemet, chatur01, reames
Subscribers: MatzeB, grosser, joker.eph, sanjoy, javed.absar, sbaranga,
llvm-commits
Differential Revision: http://reviews.llvm.org/D9151
llvm-svn: 259986
user process dyld binary and/or a mach kernel binary image. By
default, it prefers the kernel if it finds both.
But if it finds two kernel binary images (which can happen when
random things are mapped into memory), it may pick the wrong
kernel image.
DynamicLoaderDarwinKernel has heuristics to find a kernel in memory;
once we've established that there is a kernel binary in memory,
call over to that class to see if it can find a kernel address via
its search methods. If it does, use that.
Some minor cleanups to DynamicLoaderDarwinKernel while I was at it.
<rdar://problem/24446112>
llvm-svn: 259983
Summary:
Different devices may in some cases require different code generation schemes in order to implement OpenMP. This is required not only for performance reasons, but also because it may not be possible to have the current (default) implementation working for these devices. E.g. GPU's cannot implement the same scheme a target such as powerpc or x86b would use, in the sense that it does not have the ability to fork threads, instead all the threads are always executing and need to be managed by the implementation.
This patch proposes a reorganization of the code in the OpenMP code generation to pave the way to have specialized implementation of OpenMP support. More than a "real" patch this is more a request for comments in order to understand if what is proposed is acceptable or if there are better/easier ways to do it.
In this patch part of the common OpenMP codegen infrastructure is moved to a new file under a new namespace (CGOpenMPCommon) so it can be shared between the default implementation and the specialized one. When CGOpenMPRuntime is created, an attempt to select a specialized implementation is done.
In the patch a specialization for nvptx targets is done which currently checks if the target is an OpenMP device and trap if it is not.
Let me know comments suggestions you may have.
Reviewers: hfinkel, carlo.bertolli, arpith-jacob, kkwli0, ABataev
Subscribers: Hahnfeld, cfe-commits, fraggamuffin, caomhin, jholewinski
Differential Revision: http://reviews.llvm.org/D16784
llvm-svn: 259977
It is possible for enums to be created as part of their own
declcontext. We need to cache a placeholder to avoid the type being
created twice before hitting the cache.
<rdar://problem/24493203>
llvm-svn: 259975
There is a legitimate use-case in clang where we need to replace a
temporary placeholder node with the temporary node that may be a
forward declaration.
<rdar://problem/24493203>
llvm-svn: 259973
The initial segment protection was also being used to set the maximum
segment protection level. Instead, the maximum should be set according
to the architecture we are linking. For example on Mac OS it should be
RWX on most pages, but on iOS is often on R_X.
rdar://problem/24515136
llvm-svn: 259966
a significant fraction of the file size (for files that otherwise have few
records). Also include an average size per record in the summary information.
llvm-svn: 259965
Obviously, if the original Debugger goes away, those commands are holding on to now stale memory, which has the potential to cause crashes
Fixes rdar://24460882
llvm-svn: 259964
We currently tag on a "__LINKEDIT" when we are emitting the segments.
However, an upcoming patch aims to set the initprot and maxprot segment members
to their correct values, and in order to share code, its better to create this
segment for real and handle it in buildFileOffsets the same way ld64 does.
The commit for segment protections will add a test for this all being correct so
no test here until that code is committed.
llvm-svn: 259960
Temporarily relax check in test to avoid
breakage for format change in LLVM side. Once that is
done, the test case will be retightened.
llvm-svn: 259955
I removed "CIE/FIE size is too large" error because that was not
checking for correct error conditions. [UINT_MAX - 4, UINT_MAX) is
a correct range as a size of a CIE/FDE record. It's just that the
size cannot be larger than the section size.
llvm-svn: 259951
Sometimes, char arrays are used as bit storage, with no difference made between
signed and unsigned char. Thus, it is reasonable to use 0 to 255 instead of
-128 to 127 and not trigger this warning.
llvm-svn: 259947
check wrong when inheriting a member through two levels of private inheritance,
where the middle one is a class template specialization.
llvm-svn: 259943
-fsized-deallocation. Disable sized deallocation for all objects derived from
TrailingObjects, as we expect the storage allocated for these objects to be
larger than the size of their dynamic type.
llvm-svn: 259942
Simon Pilgrim