The UnrollMaxUpperBound should be target dependent, since different
chips provide different register set which brings different ability of
storing more temporary values of a program. So I add a MaxUpperBound
value in UnrollingPreference which can be override by targets. All uses
of UnrollMaxUpperBound are replaced with UP.MaxUpperBound.
The default value is still 8 and the command line argument
'--unroll-max-upperbound' takes final effect if provided.
Using the LoopInfo from OMPInfoCache after the Attributor ran resulted
in a crash due to it being in an invalid state.
---------
Co-authored-by: Ivan Radanov Ivanov <ivanov2@llnl.gov>
InstCombine canonicalizes `add` to `or` when possible, but this makes
some optimizations applicable to `add` to be missed because they don't
realize that the `or` is equivalent to `add`.
In this patch we generalize `foldICmpBinOp` to handle such cases.
Functions using different constant expressions were incorrectly
merged, because a lot of state was missing from the comparison,
including the opcode, the comparison predicate, the GEP element
type, as well as the inbounds, inrange and nowrap poison flags.
If decided to resize the instruction, need to drop wrapping flags from
the resulting vector instructions to avoid incorrect
optimizations/assumptions later.
Fixes PR75995.
Otherwise we may replace undef with poison.
Note that a lot of tests regressing here already have variants
that use poison instead of undef (often in a separate
inseltpoison file), which is why I'm not adjusting them to the
new pattern.
The specialisation will not be valid when ConstantInt gains native
support for vector types.
This is largely a mechanical change but with extra attention paid to constant
folding, InstCombineVectorOps.cpp, LoopFlatten.cpp and Verifier.cpp to
remove the need to call `getIntegerType()`.
Co-authored-by: Nikita Popov <github@npopov.com>
In line with updated shufflevector semantics, this represents the
poison elements rather than undef elements now. This commit is a
pure rename, without any logic changes.
Do not poison `undef` demanded elements in `SimplifyDemandedVectorElts`.
A miscompilation issue has been addressed with refined checking.
Proofs: https://alive2.llvm.org/ce/z/WA5oD5.
This patch tries to canonicalize the pattern `Overflow | icmp pred Res,
C2` into `Overflow | icmp pred X, C2 +/- C1`, where `Overflow` and `Res`
are return values of `xxx.with.overflow X, C1`.
Alive2: https://alive2.llvm.org/ce/z/PhR_3SFixes#75360.
When ASan.MaxInlinePoisoningSize == 0 , it means that no shadow memory
operations should be made via inlined instrumentation code,
but only via calls to shadow setting functions. This change fixes one
violation of this, which happened when the function allocas count
was small, i.e. less than 5 - in the code modifying the shadow just
before ret instruction.
We now explicitly check ASan.MaxInlinePoisoningSize , and if it's 0 then
we disallow inlining. It is required for the instrumentation
emitting code suitable for handling by ABI implementation.
rdar://119513720
Co-authored-by: Mariusz Borsa <m_borsa@apple.com>
In #74514 and #74778 we marked various instrumentation-added sections as
large. This causes an extra PT_LOAD segment if using the small code
model. Since people using the small code model presumably aren't hitting
relocation limits, disable this when using the small code model to avoid
the extra segment.
This uses Module::getCodeModel() which isn't necessarily reliable since
it reads module metadata (which right now only the clang frontend sets),
but it would be nice to get to a point where we reliably put this sort
of information (e.g. PIC/code model/etc) in the IR. This requires
duplicating the existing tests since opt/llc currently don't set these
metadata. If we get to a point where they do set the code model metadata
based on command line arguments then we can deduplicate these tests.
Issue is covered by existing test
llvm/test/Transforms/SLPVectorizer/RISCV/phi-const.ll
See issue #75632 for ideas for how we could catch these more easily in
the future.
As suggested post-commit for a00227197, replace unnecessary getNumUsers
calls by boolean variable to indicate if users changed. Note that this
also requires an early exit to detect the case where a value is replaced
by itself.
An example of a "workload definition" would be "the transitive closure of functions actually called to satisfy a RPC request", i.e. a (typically significantly) smaller subset of the transitive closure (static + possible indirect call targets) of callees. This means this workload definition is a type of flat dynamic profile.
Producing one is not in scope - it can be produced offline from traces, or from sample-based profiles, etc.
This patch adds awareness to ThinLTO of such a concept. A workload is defined as a root and a list of functions. All function references are by-name (more readable than GUIDs). In the case of aliases, the expectation is the list contains all the alternative names.
The workload definitions are presented to the linker as a json file, containing a dictionary. The keys are the roots, the values are the list of functions.
The import list for a module defining a root will be the functions listed for it in the profile.
Using names this way assumes unique names for internal functions, i.e. clang's `-funique-internal-linkage-names`.
Note that the behavior affects the entire module where a root is defined (i.e. different workloads best be defined in different modules), and does not affect modules that don't define roots.
