Avoid editing a range of DPValues and then remapping them. This occurs
when we try to de-duplicate dbg.values, but then re-use the same
iterator range. We can instead remap them, and then erase any
duplicates.
At the same time refactor the computation of seen-intrinsic hashes, and
account for a peculiarity of loop-rotates existing behaviour: it will
only deduplicate dbg.values that are immediately before the preheaders
branch instruction, not just any dbg.value in the preheader.
Loop-rotate manually maintains dbg.value intrinsics -- it also needs to
manually maintain the replacement for dbg.value intrinsics, DPValue
objects. For the most part this patch adds parallel implementations
using the new type Some extra juggling is needed when loop-rotate hoists
loop-invariant instructions out of the loop: the DPValues attached to
such an instruction need to get rotated but not hoisted. Exercised by
the new test function invariant_hoist in dbgvalue.ll.
There's also a "don't insert duplicate debug intrinsics" facility in
LoopRotate. The value and correctness of this isn't clear, but to
continue preserving behaviour that's now tested in the "tak_dup"
function in dbgvalue.ll.
Other things in this patch include a helper DebugVariable constructor
for DPValues, a insertDebugValuesForPHIs handler for RemoveDIs
(exercised by the new tests), and beefing up the dbg.value checking in
dbgvalue.ll to ensure that each record is tested (and that there's an
implicit check-not).
BlockFrequencyInfo calculates block frequencies as Scaled64 numbers but as a last step converts them to unsigned 64bit integers (`BlockFrequency`). This improves the factors picked for this conversion so that:
* Avoid big numbers close to UINT64_MAX to avoid users overflowing/saturating when adding multiply frequencies together or when multiplying with integers. This leaves the topmost 10 bits unused to allow for some room.
* Spread the difference between hottest/coldest block as much as possible to increase precision.
* If the hot/cold spread cannot be represented loose precision at the lower end, but keep the frequencies at the upper end for hot blocks differentiable.
The formula I added to LoopRotationUtils does not produce reasonable
results if some of the branch weights are zero. Add special case
handling for this.
This fixes https://github.com/llvm/llvm-project/issues/66675
This adds code to the loop rotation transformation to ensure that the
computed block execution counts for the loop bodies are the same before
and after the transformation. This isn't always true in practice, but I
believe this is because of numeric inaccuracies in the BlockFrequency
computation.
The invariants this is modeled on and heuristic choice of 0-trip loop
amount is explained in a lenghty comment in the new
`updateBranchWeights()` function.
Differential Revision: https://reviews.llvm.org/D157462
This reverts commit 0c03f48480f69b854f86d31235425b5cb71ac921.
Going to fix forward size regression instead due to more dependent patches needing to be reverted otherwise.
Unlike every other analysis and transform, simplifyInstruction
permitted operating on instructions which are not inserted
into a function. This created an edge case no other code needs
to really worry about, and limited transforms in cases that
can make use of the context function. Only the inliner and a handful
of other utilities were making use of this, so just fix up these
edge cases. Results in some IR ordering differences since
cloned blocks are inserted eagerly now. Plus some additional
simplifications trigger (e.g. some add 0s now folded out that
previously didn't).
This is a follow-up to b71edfaa4ec3c998aadb35255ce2f60bba2940b0
since I forgot the lit.local.cfg files in that one.
Reformatting is done with `black`.
If you end up having problems merging this commit because you
have made changes to a python file, the best way to handle that
is to run git checkout --ours <yourfile> and then reformat it
with black.
If you run into any problems, post to discourse about it and
we will try to help.
RFC Thread below:
https://discourse.llvm.org/t/rfc-document-and-standardize-python-code-style
Reviewed By: barannikov88, kwk
Differential Revision: https://reviews.llvm.org/D150762
Without this, pointer IVs in loops with small constant trip counts couldn't be proven no-self-wrap. This came up in a new LSR transform, but may also benefit other SCEV consumers as well.
Differential Revision: https://reviews.llvm.org/D146596
This is required because if there is a pure loop-invariant instruction, Loop Rotation
may decide to not clone it and just hoist it instead. If SCEV has previously cached
that it was loop-variant (not being smart enough to prove invariance), we may end
up with inconsistent cache state (which may later trigger false-negative assertion
failures checking that something was invariant).
This is a conservative fix that unconditionally drops the dispositions. We could
only drop it if the hoisting has actually happened, but it should take some time
understanding whether it's safe with all other things this function does.
Differential Revision: https://reviews.llvm.org/D134167
Reviewed By: fhahn
Summary:
The code for generating a name for loops for various reporting scenarios
created a name by serializing the loop into a string. This may result in
a very large name for a loop containing many blocks. Use the getName()
function on the loop instead.
Author: Jamie Schmeiser <schmeise@ca.ibm.com>
Reviewed By: Whitney (Whitney Tsang), aeubanks (Arthur Eubanks)
Differential Revision: https://reviews.llvm.org/D133587
Callbr is no longer an indirect terminator in the sense that is
relevant here (that it's successors cannot be updated). The primary
effect of this change is that callbr no longer prevents formation
of loop simplify form.
I decided to drop the isIndirectTerminator() method entirely and
replace it with isa<IndirectBrInst>() checks. I assume this method
was added to abstract over indirectbr and callbr, but it never
really caught on, and there is nothing left to abstract anymore
at this point.
Differential Revision: https://reviews.llvm.org/D129849
Following some recent discussions, this changes the representation
of callbrs in IR. The current blockaddress arguments are replaced
with `!` label constraints that refer directly to callbr indirect
destinations:
; Before:
%res = callbr i8* asm "", "=r,r,i"(i8* %x, i8* blockaddress(@test8, %foo))
to label %asm.fallthrough [label %foo]
; After:
%res = callbr i8* asm "", "=r,r,!i"(i8* %x)
to label %asm.fallthrough [label %foo]
The benefit of this is that we can easily update the successors of
a callbr, without having to worry about also updating blockaddress
references. This should allow us to remove some limitations:
* Allow unrolling/peeling/rotation of callbr, or any other
clone-based optimizations
(https://github.com/llvm/llvm-project/issues/41834)
* Allow duplicate successors
(https://github.com/llvm/llvm-project/issues/45248)
This is just the IR representation change though, I will follow up
with patches to remove limtations in various transformation passes
that are no longer needed.
Differential Revision: https://reviews.llvm.org/D129288
Per the documentation in Support/InstructionCost.h, the purpose of an invalid cost is so that clients can change behavior on impossible to cost inputs. CodeMetrics was instead asserting that invalid costs never occurred.
On a target with an incomplete cost model - e.g. RISCV - this means that transformations would crash on (falsely) invalid constructs - e.g. scalable vectors. While we certainly should improve the cost model - and I plan to do so in the near future - we also shouldn't be crashing. This violates the explicitly stated purpose of an invalid InstructionCost.
I updated all of the "easy" consumers where bailouts were locally obvious. I plan to follow up with loop unroll in a following change.
Differential Revision: https://reviews.llvm.org/D127131
In D115311, we're looking to modify clang to emit i constraints rather
than X constraints for callbr's indirect destinations. Prior to doing
so, update all of the existing tests in llvm/ to match.
Reviewed By: void, jyknight
Differential Revision: https://reviews.llvm.org/D115410
- CUDA cannot associate memory space with pointer types. Even though Clang could add extra attributes to specify the address space explicitly on a pointer type, it breaks the portability between Clang and NVCC.
- This change proposes to assume the address space from a pointer from the assumption built upon target-specific address space predicates, such as `__isGlobal` from CUDA. E.g.,
```
foo(float *p) {
__builtin_assume(__isGlobal(p));
// From there, we could assume p is a global pointer instead of a
// generic one.
}
```
This makes the code portable without introducing the implementation-specific features.
Note that NVCC starts to support __builtin_assume from version 11.
Reviewed By: arsenm
Differential Revision: https://reviews.llvm.org/D112041
This patch fixes problems reported in PR51981.
When rotating a loop it isn't enough to just forget SCEV for that
loop nest. When rotating we might clone some instructions from the
old header into the preheader, and insert new PHI nodes to merge
values together. There could be users of the original value that are
updated to use the PHI result. And those users were not necessarily
depending on a PHI node earlier, so they weren't cleaned up when just
forgetting all SCEV:s for the loop nest. So we need to explicitly
forget those values to avoid invalid cached SCEV expressions.
Reviewed By: fhahn, mkazantsev
Differential Revision: https://reviews.llvm.org/D110813
This is enabled by default. Drop explicit uses in preparation for
removing the option.
Also drop RUN lines that are now the same (typically modulo a
-verify-memoryssa option).
Since d6de1e1a71406c75a4ea4d5a2fe84289f07ea3a1, no attributes is quivalent to
setting attribute to false.
This is a preliminary commit for https://reviews.llvm.org/D99080
Printing pass manager invocations is fairly verbose and not super
useful.
This allows us to remove DebugLogging from pass managers and PassBuilder
since all logging (aside from analysis managers) goes through
instrumentation now.
This has the downside of never being able to print the top level pass
manager via instrumentation, but that seems like a minor downside.
Reviewed By: ychen
Differential Revision: https://reviews.llvm.org/D101797
Change several pass sequence sensitive tests to be indifferent
to the PreserveCFGChecker by explicitly settting the option
-verify-cfg-preserved=0. It is a preparation step that allows
a redesign of PreserveCFGChecker.
Reviewed By: skatkov
Differential Revision: https://reviews.llvm.org/D99878
Similar to D92887, LoopRotation also needs duplicate the noalias scopes when rotating a `@llvm.experimental.noalias.scope.decl` across a block boundary.
This is based on the version from the Full Restrict paches (D68511).
The problem it fixes also showed up in Transforms/Coroutines/ex5.ll after D93040 (when enabling strict checking with -verify-noalias-scope-decl-dom).
Reviewed By: nikic
Differential Revision: https://reviews.llvm.org/D94306
We tend to assume that the AA pipeline is by default the default AA
pipeline and it's confusing when it's empty instead.
PR48779
Initially reverted due to BasicAA running analyses in an unspecified
order (multiple function calls as parameters), fixed by fetching
analyses before the call to construct BasicAA.
Reviewed By: asbirlea
Differential Revision: https://reviews.llvm.org/D95117
We tend to assume that the AA pipeline is by default the default AA
pipeline and it's confusing when it's empty instead.
PR48779
Reviewed By: asbirlea
Differential Revision: https://reviews.llvm.org/D95117
83daa49758a1 made loop-rotate more conservative in the presence of
function calls in the prepare-for-lto stage. The code did not properly
account for calls that are no actual function calls, like calls to
intrinsics. This patch updates the code to ensure only calls that are
lowered to actual calls are considered inline candidates.
D84108 exposed a bad interaction between inlining and loop-rotation
during regular LTO, which is causing notable regressions in at least
CINT2006/473.astar.
The problem boils down to: we now rotate a loop just before the vectorizer
which requires duplicating a function call in the preheader when compiling
the individual files ('prepare for LTO'). But this then prevents further
inlining of the function during LTO.
This patch tries to resolve this issue by making LoopRotate more
conservative with respect to rotating loops that have inline-able calls
during the 'prepare for LTO' stage.
I think this change intuitively improves the current situation in
general. Loop-rotate tries hard to avoid creating headers that are 'too
big'. At the moment, it assumes all inlining already happened and the
cost of duplicating a call is equal to just doing the call. But with LTO,
inlining also happens during full LTO and it is possible that a previously
duplicated call is actually a huge function which gets inlined
during LTO.
From the perspective of LV, not much should change overall. Most loops
calling user-provided functions won't get vectorized to start with
(unless we can infer that the function does not touch memory, has no
other side effects). If we do not inline the 'inline-able' call during
the LTO stage, we merely delayed loop-rotation & vectorization. If we
inline during LTO, chances should be very high that the inlined code is
itself vectorizable or the user call was not vectorizable to start with.
There could of course be scenarios where we inline a sufficiently large
function with code not profitable to vectorize, which would have be
vectorized earlier (by scalarzing the call). But even in that case,
there probably is no big performance impact, because it should be mostly
down to the cost-model to reject vectorization in that case. And then
the version with scalarized calls should also not be beneficial. In a way,
LV should have strictly more information after inlining and make more
accurate decisions (barring cost-model issues).
There is of course plenty of room for things to go wrong unexpectedly,
so we need to keep a close look at actual performance and address any
follow-up issues.
I took a look at the impact on statistics for
MultiSource/SPEC2000/SPEC2006. There are a few benchmarks with fewer
loops rotated, but no change to the number of loops vectorized.
Reviewed By: sanwou01
Differential Revision: https://reviews.llvm.org/D94232
phi-empty.ll does not pass under the new PM because the NPM runs
-loop-simplify. Running -loop-simplify ends up not reproing
https://llvm.org/PR48296.
Verified that this test fails when 9eb2c011 is reverted.
Reviewed By: spatel
Differential Revision: https://reviews.llvm.org/D92807
This was suggested in D92247 - I initially committed an alternate
fix ( bfd2c216ea ) to avoid the crash/assert shown in
https://llvm.org/PR48296 ,
but that was reverted because it caused msan failures on other
tests. We can try to revive that patch using the test included
here, but I do not have an immediate plan to isolate that problem.
https://llvm.org/PR48296 shows an example where we delete all of the operands
of a phi without actually deleting the phi, and that is currently considered
invalid IR. The reduced test included here would crash for that reason.
A suggested follow-up is to loosen the assert to allow 0-operand phis
in unreachable blocks.
Differential Revision: https://reviews.llvm.org/D92247
It was already disabled under -Oz in
buildFunctionSimplificationPipeline(), but not in
buildModuleOptimizationPipeline()/addPGOInstrPasses().
Reviewed By: fhahn
Differential Revision: https://reviews.llvm.org/D89927
If an analysis is actually invalidated, there's already a log statement
for that: 'Invalidating analysis: FooAnalysis'.
Otherwise the statement is not very useful.
Reviewed By: asbirlea, ychen
Differential Revision: https://reviews.llvm.org/D84981
Problem:
Right now, our "Running pass" is not accurate when passes are wrapped in adaptor because adaptor is never skipped and a pass could be skipped. The other problem is that "Running pass" for a adaptor is before any "Running pass" of passes/analyses it depends on. (for example, FunctionToLoopPassAdaptor). So the order of printing is not the actual order.
Solution:
Doing things like PassManager::Debuglogging is very intrusive because we need to specify Debuglogging whenever adaptor is created. (Actually, right now we're not specifying Debuglogging for some sub-PassManagers. Check PassBuilder)
This patch move debug logging for pass as a PassInstrument callback. We could be sure that all running passes are logged and in the correct order.
This could also be used to implement hierarchy pass logging in legacy PM. We could also move logging of pass manager to this if we want.
The test fixes looks messy. It includes changes:
- Remove PassInstrumentationAnalysis
- Remove PassAdaptor
- If a PassAdaptor is for a real pass, the pass is added
- Pass reorder (to the correct order), related to PassAdaptor
- Add missing passes (due to Debuglogging not passed down)
Reviewed By: asbirlea, aeubanks
Differential Revision: https://reviews.llvm.org/D84774
