Current algorithm to update branch weights of latch block and its copies is
based on the assumption that number of peeling iterations is approximately equal
to trip count.
However it is not correct. According to profitability check in one case we can decide to peel
in case it helps to reduce the number of phi nodes. In this case the number of peeled iteration
can be less then estimated trip count.
This patch introduces another way to set the branch weights to peeled of branches.
Let F is a weight of the edge from latch to header.
Let E is a weight of the edge from latch to exit.
F/(F+E) is a probability to go to loop and E/(F+E) is a probability to go to exit.
Then, Estimated TripCount = F / E.
For I-th (counting from 0) peeled off iteration we set the the weights for
the peeled latch as (TC - I, 1). It gives us reasonable distribution,
The probability to go to exit 1/(TC-I) increases. At the same time
the estimated trip count of remaining loop reduces by I.
As a result after peeling off N iteration the weights will be
(F - N * E, E) and trip count of loop becomes
F / E - N or TC - N.
The idea is taken from the review of the patch D63918 proposed by Philip.
Reviewers: reames, mkuper, iajbar, fhahn
Reviewed By: reames
Subscribers: hiraditya, zzheng, llvm-commits
Differential Revision: https://reviews.llvm.org/D64235
llvm-svn: 366665
icmp ne %x, INT_MIN can be treated similarly to icmp sgt %x, INT_MIN.
icmp ne %x, INT_MAX can be treated similarly to icmp slt %x, INT_MAX.
icmp ne %x, UINT_MAX can be treated similarly to icmp ult %x, UINT_MAX.
We already treat icmp ne %x, 0 similarly to icmp ugt %x, 0
llvm-svn: 366662
We were creating a bitmask from a shift of unsigned instead of uintptr_t, meaning we couldn't create masks for indices above 31.
Noticed due to a MSVC analyzer warning.
llvm-svn: 366657
Summary:
We assume in LLDB that every type comes from an ASTContext with an associated ClangASTContext.
However the types inside the ClangModuleDeclVendor don't have a ClangASTContext so we end up
crashing whenever we create a CompilerType for one of these types.
Simplest way to trigger this bug is to just look up NSObject from a module:
(lldb) expr @import Foundation
(lldb) type lookup NSObject
Assertion failed: (m_type_system != nullptr), function CompilerType, file /Users/teemperor/llvm1/llvm-project/lldb/source/Symbol/CompilerType.cpp, line 39.
This patch just creates a ClangASTContext for the ASTContext used by ClangModuleDeclVendor.
Reviewers: davide, shafik
Reviewed By: davide
Subscribers: lldb-commits
Tags: #lldb
Differential Revision: https://reviews.llvm.org/D64989
llvm-svn: 366653
We intend to make PdbAstBuilder abstract and implement
PdbAstBuilderClang along with any other languages that wish to use
PDBs. Thus, change GetOrCreateDeclForUid from returning a clang decl
to a lldb_private::CompilerDecl.
llvm-svn: 366650
For equality, the function called getTrue/getFalse with the VT
of the comparison input. But getTrue/getFalse need the boolean VT.
So if this code ever executed, it would assert.
I believe these cases are removed by InstSimplify so we don't get here.
So this patch just fixes up an assert to exclude the equality
possibility and removes the broken code.
llvm-svn: 366649
AddOne/SubOne create new Constant objects. That seems heavy for
comparing ConstantInts which wrap APInts. Just do the math on
on the APInts and compare them.
llvm-svn: 366648
with '-mframe-pointer'
After D56351 and D64294, frame pointer handling is migrated to tri-state
(all, non-leaf, none) in clang driver and on the function attribute.
This patch makes the frame pointer handling cc1 option tri-state.
Reviewers: chandlerc, rnk, t.p.northover, MaskRay
Reviewed By: MaskRay
Differential Revision: https://reviews.llvm.org/D56353
llvm-svn: 366645
lld currently selects the relocation model automatically depending on
the link flags specified, but in some cases it'd be useful to allow
explicitly overriding the relocation model using a flag.
llvm-svn: 366644
Much like with `urem`, the same optimization (albeit with slightly
different algorithm) applies for the signed case, too.
I'm simply copying the test coverage from `urem` case for now,
i believe it should be (close to?) sufficient.
llvm-svn: 366640
This follows the same pattern as Clang and permits the user to specify
the tablegen to use via `-DLLVM_TABLEGEN=`. This allows for
cross-compiling LLDB for a foreign target (e.g. Windows ARM64 on Windows
X64). The LLVM dependency for LLDB in that case must be a Windows ARM64
build which cannot cross-compile llvm-tblgen due to the way that Visual
Studio works. Instead, permit the user to have a separate tablegen
build which can be used during the build.
llvm-svn: 366639
Summary:
Four things here:
1. Generalize the fold to handle non-splat divisors. Reasonably trivial.
2. Unban power-of-two divisors. I don't see any reason why they should
be illegal.
* There is no ban in Hacker's Delight
* I think the ban came from the same bug that caused the miscompile
in the base patch - in `floor((2^W - 1) / D)` we were dividing by
`D0` instead of `D`, and we **were** ensuring that `D0` is not `1`,
which made sense.
3. Unban `1` divisors. I no longer believe Hacker's Delight actually says
that the fold is invalid for `D = 0`. Further considerations:
* We know that
* `(X u% 1) == 0` can be constant-folded to `1`,
* `(X u% 1) != 0` can be constant-folded to `0`,
* Also, we know that
* `X u<= -1` can be constant-folded to `1`,
* `X u> -1` can be constant-folded to `0`,
* https://godbolt.org/z/7jnZJXhttps://rise4fun.com/Alive/oF6p
* We know will end up with the following:
`(setule/setugt (rotr (mul N, P), K), Q)`
* Therefore, for given new DAG nodes and comparison predicates
(`ule`/`ugt`), we will still produce the correct answer if:
`Q` is a all-ones constant; and both `P` and `K` are *anything*
other than `undef`.
* The fold will indeed produce `Q = all-ones`.
4. Try to re-splat the `P` and `K` vectors - we don't care about
their values for the lanes where divisor was `1`.
Reviewers: RKSimon, hermord, craig.topper, spatel, xbolva00
Reviewed By: RKSimon
Subscribers: hiraditya, javed.absar, dexonsmith, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D63963
llvm-svn: 366637
As detailed on PR42674, we can reduce a vXi8 down until we have the final <8 x i8>, and then use PSADBW with zero, to sum those values. We then extract the bottom i8, discarding any overflow from the upper bits of the i16 result.
llvm-svn: 366636
If the blockaddress is not destoryed, the destination block will still
be marked as having its address taken, limiting further transformations.
I think there are other places where the dead blockaddress constants are kept
around, I'll look into that as follow up.
Reviewers: craig.topper, brzycki, davide
Reviewed By: brzycki, davide
Differential Revision: https://reviews.llvm.org/D64936
llvm-svn: 366633
Summary: Make debugserver a tool like lldb-server, so it can be included/excluded via `LLDB_TOOL_DEBUGSERVER_BUILD`. This replaces the old `LLDB_NO_DEBUGSERVER` flag. Doing the same for darwin-debug while I am here.
Reviewers: xiaobai, JDevlieghere, davide
Reviewed By: xiaobai, JDevlieghere
Subscribers: mgorny, lldb-commits, #lldb
Tags: #lldb
Differential Revision: https://reviews.llvm.org/D64994
llvm-svn: 366631
Sometimes, you can end up with cross-bank copies between same-sized GPRs and
FPRs, which feed into G_STOREs. When these copies feed only into stores, they
aren't necessary; we can just store using the original register bank.
This provides some minor code size savings for some floating point SPEC
benchmarks. (Around 0.2% for 453.povray and 450.soplex)
This issue doesn't seem to show up due to regbankselect or anything similar. So,
this patch introduces an early select function, `contractCrossBankCopyIntoStore`
which performs the contraction when possible. The selector then continues
normally and selects the correct store opcode, eliminating needless copies
along the way.
Differential Revision: https://reviews.llvm.org/D65024
llvm-svn: 366625
Summary:
Add immutable WASM global `__tls_align` which stores the alignment
requirements of the TLS segment.
Add `__builtin_wasm_tls_align()` intrinsic to get this alignment in Clang.
The expected usage has now changed to:
__wasm_init_tls(memalign(__builtin_wasm_tls_align(),
__builtin_wasm_tls_size()));
Reviewers: tlively, aheejin, sbc100, sunfish, alexcrichton
Reviewed By: tlively
Subscribers: dschuff, jgravelle-google, hiraditya, cfe-commits, llvm-commits
Tags: #clang, #llvm
Differential Revision: https://reviews.llvm.org/D65028
llvm-svn: 366624
Summary:
Regular LTO modules do not need LTO Unit splitting, only ThinLTO does
(they must be consistently split into regular and Thin units for
optimizations such as whole program devirtualization and lower type
tests). In order to avoid spurious errors from LTO when combining with
split ThinLTO modules, always set this flag for regular LTO modules.
Reviewers: pcc
Subscribers: mehdi_amini, Prazek, inglorion, steven_wu, dexonsmith, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D65009
llvm-svn: 366623
There were two main problems:
* The 'nativecodegen' pseudo-component was unconditionally adding
${native_tgt}CodeGen even though it conditionally added ${native_tgt}Info and
${native_tgt}Desc. This has been fixed by making ${native_tgt}CodeGen
conditional too
* The 'all' pseudo-component was causing library names like LLVMLLVMDemangle as
the expansion was to a library name and not a component. There doesn't seem to
be a list of available components anywhere so this has been fixed by moving the
expansion of 'all' back where it was before. This manifested in different ways
on different builders but it was the same root cause
llvm-svn: 366622
Some targets are missing LLVMDemangle, one is adding the LLVM prefix twice, and two
are hitting the very error this patch fixes for my target. Reverting while I work
through the reports.
llvm-svn: 366615
Kai Luo