Summary:
The reverse of an artbitrary bitpattern is also an arbitrary
bitpattern.
Reviewers: trentxintong, arsenm, majnemer
Reviewed By: majnemer
Subscribers: majnemer, wdng, llvm-commits
Differential Revision: https://reviews.llvm.org/D31118
llvm-svn: 298201
When not adding constraints on parameters using -polly-ignore-parameter-bounds,
the context may not necessarily list all parameter dimensions. To support code
generation in this situation, we now always iterate over the actual parameter
list, rather than relying on the context to list all parameter dimensions.
llvm-svn: 298197
After this change, enabling -polly-codegen-add-debug-printing in combination
with -polly-codegen-generate-expressions allows us to instrument the compiled
binaries to not only print the values stored and loaded to a given memory
access, but also to print the accessed location with array name and
per-dimension offset:
MemRef_A[3][2]
Store to 6299784: 5.000000
MemRef_A[3][3]
Load from 6299788: 0.000000
MemRef_A[3][3]
Store to 6299788: 6.000000
This can be very helpful for debugging.
llvm-svn: 298194
In commit r219005 lifetime markers have been introduced to mark the lifetime of
the OpenMP context data structure. However, their use seems incorrect and
recently caused a miscompile in ASC_Sequoia/CrystalMk after r298053 which was
not at all related to r298053. r298053 only caused a change in the loop order,
as this change resulted in a different isl internal representation which caused
the scheduler to derive a different schedule. This change then caused the IR to
change, which apparently created a pattern in which LLVM exploites the lifetime
markers. It seems we are using the OpenMP context outside of the lifetime
markers. Even though CrystalMk could probably be fixed by expanding the scope of
the lifetime markers, it is not clear what happens in case the OpenMP function
call is in a loop which will cause a sequence of starting and ending lifetimes.
As it is unlikely that the lifetime markers give any performance benefit, we
just drop them to remove complexity.
llvm-svn: 298192
Summary:
iterateOnFunction creates a ReversePostOrderTraversal object which does a post order traversal in its constructor and stores the results in an internal vector. Iteration over it just reads from the internal vector in reverse order.
The GVN code seems to be unaware of this and iterates over ReversePostOrderTraversal object and makes a copy of the vector into a local vector. (I think at one point in time we used a DFS here instead which would have required the local vector).
The net affect of this is that we have two vectors containing the basic block list. As I didn't want to expose the implementation detail of ReversePostOrderTraversal's constructor to GVN, I've changed the code to do an explicit post order traversal storing into the local vector and then reverse iterate over that.
I've also removed the reserve(256) since the ReversePostOrderTraversal wasn't doing that. I can add it back if we thinks it important. Though it seemed weird that it wasn't based on the size of the function.
Reviewers: davide, anemet, dberlin
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D31084
llvm-svn: 298191
The code assigned to KnownZero, but later code unconditionally assigned over it. I'm pretty sure the later code can handle the same cases and more equally well.
llvm-svn: 298190
Let targets specialize the pass with the register class so we can get a
parameterless default constructor and can put the pass into the pass
registry to enable testing with -run-pass=.
llvm-svn: 298184
Go back to behavior pre-r231309 and reduce the timeout from 8 to ~1.5
min now that we have (a) PCMCache mechanism (r298165) and (b) timeout
that doesn't cause a failure, but actually build the module (r298175).
rdar://problem/30297862
llvm-svn: 298176
Duncan's r298165 introduced the PCMCache mechanism, which guarantees
that locks aren't necessary anymore for correctness but only for
performance, by avoiding building it twice when possible.
Change the logic to avoid an error but actually build the module in case
the timeout happens. Instead of an error, still emit a remark for
debugging purposes.
rdar://problem/30297862
llvm-svn: 298175
It used to be XFAIL: *, but with the new implementation it passes in some cases
and fails in other. There are similar tests for gold and lld that are not
flaky, and a positive test for bfd that makes sure that were are not breaking
existing functionality.
llvm-svn: 298173
This is direct port of HSAILAliasAnalysis pass, just cleaned for
style and renamed.
Differential Revision: https://reviews.llvm.org/D31103
llvm-svn: 298172
When InstCombine calls into SimplifyLibCalls and it createa putChar calls, we don't infer the attributes. And since SimplifyLibCalls doesn't use InstCombine's IRBuilder the calls doesn't end up in the worklist on this iteration of InstCombine. So it gets picked up on the next iteration where it causes an IR change. This of course causes InstCombine to run another iteration.
So this patch just gets the attributes right the first time. We already did this for puts and some other libcalls.
Differential Revision: https://reviews.llvm.org/D31094
llvm-svn: 298171
We had a few Config member functions that returns configuration values.
For example, we had is64() which returns true if the target is 64-bit.
The return values of these functions are constant and never change.
This patch is to compute them only once to make it clear that they'll
never change.
llvm-svn: 298168
Clang's internal build system for implicit modules uses lock files to
ensure that after a process writes a PCM it will read the same one back
in (without contention from other -cc1 commands). Since PCMs are read
from disk repeatedly while invalidating, building, and importing, the
lock is not released quickly. Furthermore, the LockFileManager is not
robust in every environment. Other -cc1 commands can stall until
timeout (after about eight minutes).
This commit changes the lock file from being necessary for correctness
to a (possibly dubious) performance hack. The remaining benefit is to
reduce duplicate work in competing -cc1 commands which depend on the
same module. Follow-up commits will change the internal build system to
continue after a timeout, and reduce the timeout. Perhaps we should
reconsider blocking at all.
This also fixes a use-after-free, when one part of a compilation
validates a PCM and starts using it, and another tries to swap out the
PCM for something new.
The PCMCache is a new type called MemoryBufferCache, which saves memory
buffers based on their filename. Its ownership is shared by the
CompilerInstance and ModuleManager.
- The ModuleManager stores PCMs there that it loads from disk, never
touching the disk if the cache is hot.
- When modules fail to validate, they're removed from the cache.
- When a CompilerInstance is spawned to build a new module, each
already-loaded PCM is assumed to be valid, and is frozen to avoid
the use-after-free.
- Any newly-built module is written directly to the cache to avoid the
round-trip to the filesystem, making lock files unnecessary for
correctness.
Original patch by Manman Ren; most testcases by Adrian Prantl!
llvm-svn: 298165
This commit adds the necessary target hooks for outlining in AArch64. It also
refactors the switch statement used in `getMemOpBaseRegImmOfsWidth` into a
more general function, `getMemOpInfo`. This allows the outliner to share that
code without copying and pasting it.
The AArch64 outliner can be run using -mllvm -enable-machine-outliner, as with
the X86-64 outliner.
The test for this pass verifies that the outliner does, in fact outline
functions, fixes up the stack accesses properly, and can correctly generate a
tail call. In the future, this test should be replaced with a MIR test, so that
we can properly test immediate offset overflows in fixed-up instructions.
llvm-svn: 298162
Runtime support for the new instrumentation of globals based on !associated, and a bunch of tests.
Differential Revision: https://reviews.llvm.org/D30120
llvm-svn: 298159
Use a combination of !associated, comdat, @llvm.compiler.used and
custom sections to allow dead stripping of globals and their asan
metadata. Sometimes.
Currently this works on LLD, which supports SHF_LINK_ORDER with
sh_link pointing to the associated section.
This also works on BFD, which seems to treat comdats as
all-or-nothing with respect to linker GC. There is a weird quirk
where the "first" global in each link is never GC-ed because of the
section symbols.
At this moment it does not work on Gold (as in the globals are never
stripped).
Differential Revision: https://reviews.llvm.org/D30121
llvm-svn: 298158
This is an ELF-specific thing that adds SHF_LINK_ORDER to the global's section
pointing to the metadata argument's section. The effect of that is a reverse dependency
between sections for the linker GC.
!associated does not change the behavior of global-dce. The global
may also need to be added to llvm.compiler.used.
Since SHF_LINK_ORDER is per-section, !associated effectively enables
fdata-sections for the affected globals, the same as comdats do.
Differential Revision: https://reviews.llvm.org/D29104
llvm-svn: 298157
Handle TokenFactors more aggressively in
SDValue::reachesChainWithoutSideEffects. This isn't really a
very effective change anymore because of other changes to
chain handling, but it's a cheap check, and the expanded
comments are still useful.
It might be possible to loosen the hasOneUse() requirement with a
deeper analysis, but a naive implementation of that check would be
expensive.
Differential Revision: https://reviews.llvm.org/D29845
llvm-svn: 298156
__start_xxx symbol keeps section xxx alive only if it is not
SHF_LINK_ORDER. Such sections can be used for user metadata, when
__start_xxx is used to iterate over section contents at runtime, and
the liveness is determined solely by the linked (associated) section.
This was earlier implemented in r294592, and broken in r296723.
Differential Revision: https://reviews.llvm.org/D30964
llvm-svn: 298154
Zachary Turner