| Age | Commit message (Collapse) | Author |
|---|
|
Using a block inside LLVM_DEBUG allows us to simplify the code with a
single LLVM_DEBUG and remove maybe_unused.
|
|
We can use *Set::insert_range to collapse:
for (auto Elem : Range)
Set.insert(E);
down to:
Set.insert_range(Range);
In some cases, we can further fold that into the set declaration.
|
|
Identified with misc-include-cleaner.
|
|
Add `Intrinsic::getDeclarationIfExists` to lookup an existing
declaration of an intrinsic in a `Module`.
|
|
derived ones (#110575)
Performance critical core libraries could be highly-optimized for arch
or micro-arch features. For instance, the absl crc library specializes
different templated classes among different hardwares [1]. In a
practical setting, it's likely that instrumented profiles are collected
on one type of machine and used to optimize binaries that run on
multiple types of hardwares.
While this kind of specialization is rare in terms of lines of code,
compiler can do a better job to skip vtable-based ICP.
* The per-class `Extend` implementation is arch-specific as well. If an
instrumented profile is collected on one arch and applied to another
arch where `Extend` implementation is different, `Extend` might be
regarded as unlikely function in the latter case. `ABSL_ATTRIBUTE_HOT`
annotation alleviates the problem by putting all `Extend` implementation
into the hot text section [2]
This change introduces a comma-separated list to specify the mangled
vtable names, and ICP pass will skip vtable-based comparison if a vtable
variable definition is shown to be in its class hierarchy (per LLVM type
metadata).
[1]
https://github.com/abseil/abseil-cpp/blob/c6b27359c3d27438b1313dddd7598914c1274a50/absl/crc/internal/crc_x86_arm_combined.cc#L621-L650
[2]
https://github.com/abseil/abseil-cpp/blame/c6b27359c3d27438b1313dddd7598914c1274a50/absl/crc/internal/crc_x86_arm_combined.cc#L370C3-L370C21
|
|
Utility functions have been moved out to Utils. Minor opportunity to
drop the header where not needed.
|
|
This patch fixes:
llvm/lib/Transforms/Instrumentation/IndirectCallPromotion.cpp:845:12:
error: variable 'RemainingVTableCount' set but not used
[-Werror,-Wunused-but-set-variable]
llvm/lib/Transforms/Instrumentation/IndirectCallPromotion.cpp:306:23:
error: private field 'PSI' is not used
[-Werror,-Wunused-private-field]
Here are a couple of domino effects:
- Once I remove PSI, I need to update the contructor and its caller.
- Once I remove RemainingVTableCount, I don't need TotalCount, so I am
updating the caller as well.
|
|
with function count for each candidate (#106260)
The current cost-benefit analysis between vtable comparison and function
comparison require the indirect fallback branch to be cold. This is too
conservative.
This change allows vtable-comparison as long as vtable count is
comparable with function count for each function candidate and removes
the cold indirect fallback requirement.
Tested:
1. Testing this on benchmarks uplifts the measurable performance wins.
Counting the (possibly-duplicated) remarks (because of linkonce_odr
functions, cross-module import of functions) show the number of vtable
remarks increases from ~30k-ish to 50k-ish.
2. https://gcc.godbolt.org/z/sbGK7Pacn shows vtable-comparison doesn't
happen today (using the same IR input)
|
|
(#97357)
|
|
Aside from the fact that LLVM Programmer's Manual prefers ArrayRef to
const std::vector &, ArrayRef<PromotionCandidate> here makes it easier
to switch the underlying type to something like SmallVector. Note
that we typically do not have a lot of candidates.
|
|
Needed by VTableAddressPointOffsetValMap.
|
|
indirect-call-promotion with vtable profiles. (#81442)
Clang's `-fwhole-program-vtables` is required for this optimization to
take place. If `-fwhole-program-vtables` is not enabled, this change is
no-op.
* Function-comparison (before):
```
%vtable = load ptr, ptr %obj
%vfn = getelementptr inbounds ptr, ptr %vtable, i64 1
%func = load ptr, ptr %vfn
%cond = icmp eq ptr %func, @callee
br i1 %cond, label bb1, label bb2:
bb1:
call @callee
bb2:
call %func
```
* VTable-comparison (after):
```
%vtable = load ptr, ptr %obj
%cond = icmp eq ptr %vtable, @vtable-address-point
br i1 %cond, label bb1, label bb2:
bb1:
call @callee
bb2:
%vfn = getelementptr inbounds ptr, ptr %vtable, i64 1
%func = load ptr, ptr %vfn
call %func
```
Key changes:
1. Find out virtual calls and the vtables they come from.
- The ICP relies on type intrinsic `llvm.type.test` to find out virtual
calls and the
compatible vtables, and relies on type metadata to find the address
point for comparison.
2. ICP pass does cost-benefit analysis and compares vtable only when the
number of vtables for a function candidate is within (option specified)
threshold.
3. Sink the function addressing and vtable load instruction to indirect
fallback.
- The sink helper functions are simplified versions of
`InstCombinerImpl::tryToSinkInstruction`. Currently debug intrinsics are
not handled. Ideally `InstCombinerImpl::tryToSinkInstructionDbgValues`
and `InstCombinerImpl::tryToSinkInstructionDbgVariableRecords` could be
moved into Transforms/Utils/Local.cpp (or another util cpp file) to
handle debug intrinsics when moving instructions across basic blocks.
4. Keep value profiles updated
1) Update vtable value profiles after inline
2) For either function-based comparison or vtable-based comparison,
update both vtable and indirect call value profiles.
|
|
|
|
Callers of getPromotionCandidatesForInstruction pass NumVals as an out
parameter for the number of value-count pairs of the value profiling
data, but nobody uses the out parameter.
This patch removes the parameter and updates the callers. Note that
the number of value-count pairs is still available via
getPromotionCandidatesForInstruction(...).size().
|
|
(#95281)
…f weights" #95136
Reverts #95060, and relands #86609, with the unintended code generation
changes addressed.
This patch implements the changes to LLVM IR discussed in
https://discourse.llvm.org/t/rfc-update-branch-weights-metadata-to-allow-tracking-branch-weight-origins/75032
In this patch, we add an optional field to MD_prof meatdata nodes for
branch weights, which can be used to distinguish weights added from
llvm.expect* intrinsics from those added via other methods, e.g. from
profiles or inserted by the compiler.
One of the major motivations, is for use with MisExpect diagnostics,
which need to know if branch_weight metadata originates from an
llvm.expect intrinsic. Without that information, we end up checking
branch weights multiple times in the case if ThinLTO + SampleProfiling,
leading to some inaccuracy in how we report MisExpect related
diagnostics to users.
Since we change the format of MD_prof metadata in a fundamental way, we
need to update code handling branch weights in a number of places.
We also update the lang ref for branch weights to reflect the change.
|
|
weights" (#95060)
Reverts llvm/llvm-project#86609
This change causes compile-time regressions for stage2 builds
(https://llvm-compile-time-tracker.com/compare.php?from=3254f31a66263ea9647c9547f1531c3123444fcd&to=c5978f1eb5eeca8610b9dfce1fcbf1f473911cd8&stat=instructions:u).
It also introduced unintended changes to `.text` which should be
addressed before relanding.
|
|
This patch implements the changes to LLVM IR discussed in
https://discourse.llvm.org/t/rfc-update-branch-weights-metadata-to-allow-tracking-branch-weight-origins/75032
In this patch, we add an optional field to MD_prof metadata nodes for
branch weights, which can be used to distinguish weights added from
`llvm.expect*` intrinsics from those added via other methods, e.g.
from profiles or inserted by the compiler.
One of the major motivations, is for use with MisExpect diagnostics,
which need to know if branch_weight metadata originates from an
llvm.expect intrinsic. Without that information, we end up checking
branch weights multiple times in the case if ThinLTO + SampleProfiling,
leading to some inaccuracy in how we report MisExpect related
diagnostics to users.
Since we change the format of MD_prof metadata in a fundamental way, we
need to update code handling branch weights in a number of places.
We also update the lang ref for branch weights to reflect the change.
|
|
and indirect call value profile update into one helper function (#80762)
* The motivation is to move indirect callee profile update inside the
function-based speculative indirect-call promotion, so that there are
fewer diffs the vtable-based transformation and profile update is
implemented in a follow-up patch.
* The Parent patch is https://github.com/llvm/llvm-project/pull/79381
|
|
Add `setBranchWeights` convenience function to ProfDataUtils.h and use
it where appropriate.
|
|
For (subjectively) more clarity; also updated the comment describing it.
Differential Revision: https://reviews.llvm.org/D155888
|
|
- we can remove the `Module` field, it's obtainable from `F` and used in
only one place
- a few fields can be `const`-ed, thus enforcing compile-time
initialization checking (and we don't need to support `operator=`)
Differential Revision: https://reviews.llvm.org/D155212
|
|
Consistency with how we tend to name `ModuleAnalysisManager` parameters.
|
|
There's no caller to `promoteIndirectCalls` that would pass a nullptr
`ModuleAnalysisManager`, so passing it by reference does away with a
bunch of nullptr tests, and also removes the need for a "OwnedORE".
Differential Revision: https://reviews.llvm.org/D155027
|
|
Identified with readability-qualified-auto.
|
|
Some cl::ZeroOrMore were added to avoid the `may only occur zero or one times!`
error. More were added due to cargo cult. Since the error has been removed,
cl::ZeroOrMore is unneeded.
Also remove cl::init(false) while touching the lines.
|
|
Running iwyu-diff on LLVM codebase since fa5a4e1b95c8f37796 detected a few
regressions, fixing them.
Differential Revision: https://reviews.llvm.org/D124847
|
|
Legacy PM for optimization pipeline was deprecated in 13.0.0 and Clang dropped
legacy PM support in D123609. This change removes legacy PM passes for PGO so
that downstream projects won't be able to use it. It seems appropriate to start
removing such "add-on" features like instrumentations, before we remove more
stuff after 15.x is branched.
I have checked many LLVM users and only ldc[1] uses the legacy PGO pass.
[1]: https://github.com/ldc-developers/ldc/issues/3961
Reviewed By: davidxl
Differential Revision: https://reviews.llvm.org/D123834
|
|
Discourse thread: https://discourse.llvm.org/t/include-what-you-use-include-cleanup
Differential Revision: https://reviews.llvm.org/D121332
|
|
|
|
".llvm." suffix".
The recommit fixed a bug that symbols with "." at the beginning is not
properly handled in the last commit.
Original commit message:
Currently IndirectCallPromotion simply strip everything after the first "."
in LTO mode, in order to match the symbol name and the name with ".llvm."
suffix in the value profile. However, if -funique-internal-linkage-names
and thinlto are both enabled, the name may have both ".__uniq." suffix and
".llvm." suffix, and the current mechanism will strip them both, which is
unexpected. The patch fixes the problem.
Differential Revision: https://reviews.llvm.org/D98389
|
|
This guards against cases where the symbol was dead code eliminated in
the binary by ThinLTO, and we have a sample profile collected for one
binary but used to optimize another.
Most of the benefit from ICP comes from inlining the target, which we
can't do with only a declaration anyway. If this is in the pre-ThinLTO
link step (e.g. for instrumentation based PGO), we will attempt the
promotion again in the ThinLTO backend after importing anyway, and we
don't need the early promotion to facilitate that.
Differential Revision: https://reviews.llvm.org/D92804
|
|
IndirectCallPromotion
Reviewers: dblaikie, craig.topper, wmi
Subscribers: hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D79047
|
|
Summary:
Refactored the parameter and return type where they are too generally
typed as Instruction.
Reviewers: dblaikie, wmi, craig.topper
Subscribers: hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D79027
|
|
Summary:
Updated CallPromotionUtils and impacted sites. Parameters that are
expected to be non-null, and return values that are guranteed non-null,
were replaced with CallBase references rather than pointers.
Left FIXME in places where more changes are facilitated by CallBase, but
aren't CallSites: Instruction* parameters or return values, for example,
where the contract that they are actually CallBase values.
Reviewers: davidxl, dblaikie, wmi
Reviewed By: dblaikie
Subscribers: arsenm, jvesely, nhaehnle, eraman, hiraditya, kerbowa, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D77930
|
|
This file lists every pass in LLVM, and is included by Pass.h, which is
very popular. Every time we add, remove, or rename a pass in LLVM, it
caused lots of recompilation.
I found this fact by looking at this table, which is sorted by the
number of times a file was changed over the last 100,000 git commits
multiplied by the number of object files that depend on it in the
current checkout:
recompiles touches affected_files header
342380 95 3604 llvm/include/llvm/ADT/STLExtras.h
314730 234 1345 llvm/include/llvm/InitializePasses.h
307036 118 2602 llvm/include/llvm/ADT/APInt.h
213049 59 3611 llvm/include/llvm/Support/MathExtras.h
170422 47 3626 llvm/include/llvm/Support/Compiler.h
162225 45 3605 llvm/include/llvm/ADT/Optional.h
158319 63 2513 llvm/include/llvm/ADT/Triple.h
140322 39 3598 llvm/include/llvm/ADT/StringRef.h
137647 59 2333 llvm/include/llvm/Support/Error.h
131619 73 1803 llvm/include/llvm/Support/FileSystem.h
Before this change, touching InitializePasses.h would cause 1345 files
to recompile. After this change, touching it only causes 550 compiles in
an incremental rebuild.
Reviewers: bkramer, asbirlea, bollu, jdoerfert
Differential Revision: https://reviews.llvm.org/D70211
|
|
Now that we've moved to C++14, we no longer need the llvm::make_unique
implementation from STLExtras.h. This patch is a mechanical replacement
of (hopefully) all the llvm::make_unique instances across the monorepo.
llvm-svn: 369013
|
|
llvm-svn: 357773
|
|
Rename the functions that query the optimization kind attributes.
Differential revision: https://reviews.llvm.org/D60287
llvm-svn: 357731
|
|
Create method `optForNone()` testing for the function level equivalent of
`-O0` and refactor appropriately.
Differential revision: https://reviews.llvm.org/D59852
llvm-svn: 357638
|
|
Summary:
When attaching prof metadata to promoted direct calls in SamplePGO
mode, no need to construct and use a SmallVector to pass a single count
to the ArrayRef parameter, we can simply use a brace-enclosed init list.
This made a small but consistent improvement for a ThinLTO backend
compile I was measuring.
Reviewers: wmi
Subscribers: mehdi_amini, dexonsmith, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D57706
llvm-svn: 353123
|
|
to reflect the new license.
We understand that people may be surprised that we're moving the header
entirely to discuss the new license. We checked this carefully with the
Foundation's lawyer and we believe this is the correct approach.
Essentially, all code in the project is now made available by the LLVM
project under our new license, so you will see that the license headers
include that license only. Some of our contributors have contributed
code under our old license, and accordingly, we have retained a copy of
our old license notice in the top-level files in each project and
repository.
llvm-svn: 351636
|
|
update client code.
Also rename it to use the more generic term `call` instead of something
that could be confused with a praticular type.
Differential Revision: https://reviews.llvm.org/D56183
llvm-svn: 350508
|
|
Every Analysis pass has a get method that returns a reference of the Result of
the Analysis, for example, BlockFrequencyInfo
&BlockFrequencyInfoWrapperPass::getBFI(). I believe that
ProfileSummaryInfo::getPSI() is the only exception to that, as it was returning
a pointer.
Another change is renaming isHotBB and isColdBB to isHotBlock and isColdBlock,
respectively. Most methods use BB as the argument of variable names while
methods usually refer to Basic Blocks as Blocks, instead of BB. For example,
Function::getEntryBlock, Loop:getExitBlock, etc.
I also fixed one of the comments.
Patch by Rodrigo Caetano Rocha!
Differential Revision: https://reviews.llvm.org/D54669
llvm-svn: 347182
|
|
Differential revision: http://reviews.llvm.org/D51193
llvm-svn: 340657
|
|
The DEBUG() macro is very generic so it might clash with other projects.
The renaming was done as follows:
- git grep -l 'DEBUG' | xargs sed -i 's/\bDEBUG\s\?(/LLVM_DEBUG(/g'
- git diff -U0 master | ../clang/tools/clang-format/clang-format-diff.py -i -p1 -style LLVM
- Manual change to APInt
- Manually chage DOCS as regex doesn't match it.
In the transition period the DEBUG() macro is still present and aliased
to the LLVM_DEBUG() one.
Differential Revision: https://reviews.llvm.org/D43624
llvm-svn: 332240
|
|
subdirectory
llvm-svn: 328379
|
|
This patch factors out the main code transformation utilities in the pgo-driven
indirect call promotion pass and places them in Transforms/Utils. The change is
intended to be a non-functional change, letting non-pgo-driven passes share a
common implementation with the existing pgo-driven pass.
The common utilities are used to conditionally promote indirect call sites to
direct call sites. They perform the underlying transformation, and do not
consider profile information. The pgo-specific details (e.g., the computation
of branch weight metadata) have been left in the indirect call promotion pass.
Differential Revision: https://reviews.llvm.org/D40658
llvm-svn: 319963
|
|
warnings; other minor fixes (NFC).
llvm-svn: 316253
|
|
parameterized emit() calls
Summary: This is not functional change to adopt new emit() API added in r313691.
Reviewed By: anemet
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D38285
llvm-svn: 315476
|
|
Sync it up with the name of the class actually defined here. This has been
bothering me for a while...
llvm-svn: 315249
|
