This reverts 8cb7876cb366b5803ca35d92313ea00eadf29b78 and follow-ups.
GNU ld/gold/ld.lld -O has nothing to do with any code related linker optimizations.
It has very small benefit (save 144Ki (.hash, .gnu_hash) with GNU ld, save 0.7%
.debug_str with gold/ld.lld) while it makes gold/ld.lld significantly slower
when linking RelWithDebInfo clang (gold: 16.437 vs 19.488; ld.lld: 1.882 vs 4.881).
It is fairly easy to forget SectionBase::repl after ICF.
Let ICF rewrite a Defined symbol's `section` field to avoid references to
SectionBase::repl in subsequent passes. This slightly improves the --icf=none
performance due to less indirection (maybe for --icf={safe,all} as well if most
symbols are Defined).
With this change, there is only one reference to `repl` (--gdb-index D89751).
We can undo f4fb5fd7523f8e3c3b3966d43c0a28457b59d1d8 (`Move Repl to SectionBase.`)
but move `repl` to `InputSection` instead.
Reviewed By: ikudrin
Differential Revision: https://reviews.llvm.org/D116093
Single-variant enums were still getting placed on a single line
even when AllowShortEnumsOnASingleLine was false. This fixes that
by checking that setting when looking to merge lines.
Differential Revision: https://reviews.llvm.org/D116188
This patch moves some static functions from AffineStructures.cpp to
Presburger/Utils.cpp and some to be private members of FlatAffineConstraints
(which will later be moved to IntegerPolyhedron) to allow for a smoother
transition for moving FlatAffineConstraints math functionality to
Presburger/IntegerPolyhedron.
This patch is part of a series of patches for moving math functionality to
Presburger directory.
Reviewed By: arjunp, bondhugula
Differential Revision: https://reviews.llvm.org/D115869
Otherwise, it is possible that the state defined in the determinator
block defines the state for the next iteration of the loop, rather than
for the current one.
Fixes llvm-test-suite's
SingleSource/Regression/C/gcc-c-torture/execute/pr80421.c
Differential Revision: https://reviews.llvm.org/D115832
In 'trunc' i16/32/64 to i1 pattern the 'and $src, 1' node supply operand to 'setcc'.
The latter is selected to S_CMP_EQ/V_CMP_EQ dependent on the divergence. In case the 'and' is scalar
and 'setcc' is divergent, we need VGPR to SGPR copy to adjust input operand for V_CMP_EQ.
This patch changes the S_AND_B32 to V_AND_B32_e64 in the 'trunc to i1' divergent patterns.
Reviewed By: rampitec
Differential Revision: https://reviews.llvm.org/D116241
Officially this is currently required to always use the datalayout's
alloca address space. This may change in the future, and it's cleaner
to propagate the existing alloca's addrspace anyway.
This is a triple fix. Initially the change in simplifyAllocaArraySize
would drop the address space, but produce output. Fixing this hit an
assertion in the cast combine.
This patch also makes the changes to handle this situation from
a33e12801279a947c74fdee2655b24480941fb39 dead, so eliminate
it. InstCombine should not take it upon itself to introduce
addrspacecasts, and preserve the original address space instead.
This patch adds the support for `atomic compare` in parser. The support
in Sema and CodeGen will come soon. For now, it simply eimits an error when it
is encountered.
Reviewed By: ABataev
Differential Revision: https://reviews.llvm.org/D115561
Converts RSHRN/RSHRN2 to RADDHN/RADDHN2 when the shift amount is half
the width of the vector element. The latter has twice the throughput
and half the latency on Arm out-of-order cores. Setting up the zero
register adds no latency.
Differential Revision: https://reviews.llvm.org/D116166
This fixes a typo in 81d69e1bda9e4b6a83f29ba1f614e43ab4700972.
Of course we should only skip the particular store if it isn't
removable, not bail out of the whole loop. Add a test to cover
this case.
This reverts commit a954558e878ed9e97e99036229e99af8c6b6c881.
Thanks Yuanfang's help. I think I found the root cause of the buildbot
fail.
The failed test has both Memory and Immediate X86Operand. All data of
different operand kinds share the same memory space by a union
definition. So it has chance we get the wrong result if we don't check
the operand kind.
It's probably it happen to be the correct value in my local environment
so that I can't reproduce the fail.
Differential Revision: https://reviews.llvm.org/D116090
At this point the instruction may either have an analyzable
write or be a terminator. For terminators, isRemovable() is not
necessarily well-defined. Move the check until after we have ensured
that it is not a terminator.
The only non-trivial change here is that the isReadClobber()
check for redundant stores is now on the DefLoc, not the
UpperLoc. This is semantically the right location to use, though
in practice it makes no difference (the locations are either the
same, or the def inst does not read).
We have Value->stripInBoundsConstantOffsets() which does what we
want here, but the inbounds requirement isn't actually necessary.
We should probably add Value->stripConstantOffsets() as well.
Remove the special casing for intrinsics in MemoryLocation::getForDest()
and handle them through the general attribute based code. On the DSE
side, this means that isRemovable() now needs to handle more than a
hardcoded list of intrinsics. We consider everything apart from
volatile memory intrinsics and lifetime markers to be removable.
This allows us to perform DSE on intrinsics that DSE has not been
specially taught about, using a matrix store as an example here.
There is an interesting test change for invariant.start, but I
believe that optimization is correct. It only looks a bit odd
because the code is immediate UB anyway.
Differential Revision: https://reviews.llvm.org/D116210
Instead of using the ArgumentPromotion implementation, we now walk
call sites using checkForAllCallSites() and directly call
areTypesABICompatible() using the replacement types. I believe
that resolves the TODO in the code.
Differential Revision: https://reviews.llvm.org/D116033
The reduction initialization code creates a "naturally aligned null
pointer to void lvalue", which I found somewhat odd, even though it
works out in the end because it is not actually used. It doesn't
look like this code actually needs an LValue for anything though,
and we can use an invalid Address to represent this case instead.
Differential Revision: https://reviews.llvm.org/D116214
According to [dcl.fct.def.coroutine]p6, the promise_type is allowed to
not define return_void nor return_value:
> If searches for the names return_void and return_value in the scope
> of the promise type each find any declarations, the program is
> ill-formed.
> [Note 1: If return_void is found, flowing off the end of a coroutine is
> equivalent to a co_return with no operand. Otherwise, flowing off the
> end of a coroutine results in
> undefined behavior ([stmt.return.coroutine]). — end note]
So the program isn't ill-formed if the promise_type doesn't define
return_void nor return_value. It is just a potential UB. So the program
should be allowed to compile.
Reviewed By: urnathan
Differential Revision: https://reviews.llvm.org/D116204
Causes invalid debug_gnu_pubnames (& I think non-gnu pubnames too) -
visible as 0 values for the offset in gnu pubnames. More details on the
original review in D115325.
This reverts commit 78d15a112cbd545fbb6e1aa37c221ef5aeffb3f2.
This reverts commit 54586582d3e17c0bba76258d2930486a6dfaaf2a.
The 'r' constraint uses the GPR class. There is generic support
for bitcasting and extending/truncating non-integer VTs to the
required integer VT. This doesn't work for scalable vectors and
instead crashes.
To prevent this, explicitly reject vectors. Fixed vectors might
work without crashing, but it doesn't seem worthwhile to allow.
While there remove an unnecessary level of indentation in the
"vr" and "vm" constraint handling.
Differential Revision: https://reviews.llvm.org/D115810
Querying threads directly from the thread pool fails if there is no thread pool or if multithreading is not enabled. Returns 1 by default.
Reviewed By: mehdi_amini
Differential Revision: https://reviews.llvm.org/D116259
