To store a byval parameter the existing code would store as many 8 byte elements
as was required to store the full size of the byval parameter.
For example, a paramter of size 16 would store two element of 8 bytes.
A paramter of size 12 would also store two elements of 8 bytes.
This would sometimes store too many bytes as the size of the paramter is not
always a factor of 8.
This patch fixes that issue and now byval paramters are stored with the correct
number of bytes.
Reviewed By: nemanjai, #powerpc, quinnp, amyk
Differential Revision: https://reviews.llvm.org/D121430
Add a compiler option and the instructions required to set the
special Data Stream Control Register (DSCR). The special register will
not be set by default.
Original patch by: Muhammad Usman
Reviewed By: nemanjai, #powerpc
Differential Revision: https://reviews.llvm.org/D117013
The BL8_NOTOC_RM instruction was incorrectly producing a relocation that reqired
a TOC restore after the call. This patch fixes that issue and the notoc
relocation is now used.
Reviewed By: jsji
Differential Revision: https://reviews.llvm.org/D122012
After D108936, @llvm.smul.with.overflow.i64 was lowered to __multi3
instead of __mulodi4, which also doesn't exist on PowerPC 32-bit, not
even with compiler-rt. Block it as well so that we get inline code.
Because libgcc doesn't have __muloti4, we block that as well.
Fixes#54460.
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D122090
This is the first patch to enable the XCOFF64 object writer.
Currently only fileHeader and sectionHeaders are supported.
Reviewed By: jhenderson, DiggerLin
Differential Revision: https://reviews.llvm.org/D120861
When generating a all-one mask value whose bitwidth is larger than 64, signed extension should be used rather then zero extension.
Reviewed By: jsji
Differential Revision: https://reviews.llvm.org/D120865
Add the calling convention for the vector pair registers.
These registers overlap with the vector registers.
Part of an original patch by: Lei Huang
Reviewed By: nemanjai, #powerpc
Differential Revision: https://reviews.llvm.org/D117225
We are going to remove the old 'perfect shuffle' optimization since it
brings performance penalty in hot loop around vectors. For example, in
following loop sharing the same mask:
%v.1 = shufflevector ... <0,1,2,3,8,9,10,11,16,17,18,19,24,25,26,27>
%v.2 = shufflevector ... <0,1,2,3,8,9,10,11,16,17,18,19,24,25,26,27>
The generated instructions will be `vmrglw-vmrghw-vmrglw-vmrghw` instead
of `vperm-vperm`. In some large loop cases, this causes 20%+ performance
penalty.
The original attempt to resolve this is to pre-record masks of every
shufflevector operation in DAG, but that is somewhat complex and brings
unnecessary computation (to scan all nodes) in optimization. Here we
disable it by default. There're indeed some cases becoming worse after
this, which will be fixed in a more careful way in future patches.
Reviewed By: jsji
Differential Revision: https://reviews.llvm.org/D121082
VSX introduced some permute instructions that are direct
replacements for Altivec ones except they can target all
the VSX registers. We have added code generation for most
of these but somehow missed the low/hi word merges (XXMRG[LH]W).
This caused some additional spills on some large
computationally intensive code.
This patch simply adds the missed patterns.
Currently in Clang, we have two types of builtins for fnmsub operation:
one for float/double vector, they'll be transformed into IR operations;
one for float/double scalar, they'll generate corresponding intrinsics.
But for the vector version of builtin, the 3 op chain may be recognized
as expensive by some passes (like early cse). We need some way to keep
the fnmsub form until code generation.
This patch introduces ppc.fnmsub.* intrinsic to unify four fnmsub
intrinsics.
Reviewed By: shchenz
Differential Revision: https://reviews.llvm.org/D116015
When inserting undef into buildvectors created from shuffles of
buildvectors, we convert elements to the largest needed type. This had
the effect of converting undef into 0, which isn't needed as the
buildvector implicitly truncates and trunc(zext(undef)) == undef.
Differential Revision: https://reviews.llvm.org/D121002
When parsing MachineMemOperands, MIRParser treated the "align" keyword
the same as "basealign". Really "basealign" should specify the
alignment of the MachinePointerInfo base value, and "align" should
specify the alignment of that base value plus the offset.
This worked OK when the specified alignment was no larger than the
alignment of the offset, but in cases like this it just caused
confusion:
STW killed %18, 4, %stack.1.ap2.i.i :: (store (s32) into %stack.1.ap2.i.i + 4, align 8)
MIRPrinter would never have printed this, with an offset of 4 but an
align of 8, so it must have been written by hand. MIRParser would
interpret "align 8" as "basealign 8", but I think it is better to give
an error and force the user to write "basealign 8" if that is what they
really meant.
Differential Revision: https://reviews.llvm.org/D120400
Change-Id: I7eeeefc55c2df3554ba8d89f8809a2f45ada32d8
This test doesn't work because the CHECK-NOT line is actually checking
something that only exists on stderr and not stdout.
Changed the test so that we now check both stderr and stdout.
Changed the test so that we check pwr9, pwr10, and future. The cpu names of
power9 or power10 are not supported in the llc backend.
Reviewed By: nemanjai, #powerpc
Differential Revision: https://reviews.llvm.org/D120349
Previous we used sra (X, size(X)-1); xor (add (X, Y), Y).
By placing sub at the end, we allow RISCV to combine sign_extend_inreg
with it to form subw.
Some X86 tests for Z - abs(X) seem to have improved as well.
Other targets look to be a wash.
I had to modify ARM's abs matching code to match from sub instead of
xor. Maybe instead ISD::ABS should be made legal. I'll try that in
parallel to this patch.
This is an alternative to D119099 which was focused on RISCV only.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D119171
the same address by symbol types.
Summary: In XCOFF, each section comes with a default symbol
with the same name as the section. It doesn't bind
to code locations and it may cause incorrect display
of symbol names under `llvm-objdump -d`.
This patch changes the priority of symbols with the
same address by symbol type.
Reviewed By: jhenderson, shchenz
Differential Revision: https://reviews.llvm.org/D117642
The `__builtin_pdepd` and `__builtin_pextd` are P10 builtins that are meant to
be used under 64-bit only. For instance, when the builtins are compiled under
32-bit mode:
```
$ cat t.c
unsigned long long foo(unsigned long long a, unsigned long long b) {
return __builtin_pextd(a,b);
}
$ clang -c t.c -mcpu=pwr10 -m32
ExpandIntegerResult #0: t31: i64 = llvm.ppc.pextd TargetConstant:i32<6928>, t28, t29
fatal error: error in backend: Do not know how to expand the result of this operator!
```
This patch adds sema checking for these builtins to compile under 64-bit
mode only and on P10. The builtins will emit a diagnostic when they are compiled on
non-P10 compilations and on 32-bit mode.
Differential Revision: https://reviews.llvm.org/D118753
This patch updates the handling of vectors in getPreferredVectorAction():
For single-element and scalable vectors, fall back to default vector legalization
handling. For vNi1 vectors, add handling to either split or promote them in
order to prevent the production of wide v256i1/v512i1 types.
The following assertion is fixed by this patch, as we ended up producing the
wide vector types (that are used for MMA) in the backend prior to this fix.
```
Assertion failed: VT.getSizeInBits() == Operand.getValueSizeInBits() &&
"Cannot BITCAST between types of different sizes!"
```
Differential Revision: https://reviews.llvm.org/D119521
Power ISA 3.1 adds xsmaxcqp/xsmincqp for quad-precision type-c max/min selection,
and this opens the opportunity to improve instruction selection on: llvm.maxnum.f128,
llvm.minnum.f128, and select_cc ordered gt/lt and (don't care) gt/lt.
Reviewed By: nemanjai, shchenz, amyk
Differential Revision: https://reviews.llvm.org/D117006
For PGO on AIX, when we switch to the linux-style PGO variable access
(via _start and _stop labels), we need the compiler to generate a .ref
assembly for each of the three csects:
- __llvm_prf_data[RW]
- __llvm_prf_names[RO]
- __llvm_prf_vnds[RW]
We insert the .ref inside the __llvm_prf_cnts[RW] csect so that if it's
live then the 3 csects are live.
For example, for a testcase with at least one function definition, when
compiled with -fprofile-generate we should generate:
.csect __llvm_prf_cnts[RW],3
.ref __llvm_prf_data[RW] <<============ needs to be inserted
.ref __llvm_prf_names[RO] <<===========
the __llvm_prf_vnds is not always present, so we reference it only when
it's present.
Reviewed By: sfertile, daltenty
Differential Revision: https://reviews.llvm.org/D116607
This patch introduces the conversions from math function calls
to MASS library calls. To resolves calls generated with these conversions, one
need to link libxlopt.a library. This patch is tested on PowerPC Linux and AIX.
Differential: https://reviews.llvm.org/D101759
Reviewer: bmahjour
This patch updates the P10 patterns with a load feeding into an insertelt to
utilize the refactored load and store infrastructure, as well as updating any
tests that exhibit any codegen changes.
Furthermore, custom legalization is added for v4f32 on Power9 and above to not
only assist with adjusting the refactored load/stores for P10 vector insert,
but also it enables the utilization of direct moves.
Differential Revision: https://reviews.llvm.org/D115691
This patch updates how splat loads handled and is an extension of D106555.
Particularly, for v2i64/v4f32/v4i32 types, they are updated to handle only
non-extending loads. For v8i16/v16i8 types, they are updated to handle extending
loads only if the memory VT is the same vector element VT type.
A test case has been added to illustrate a scenario where a PPCISD::LD_SPLAT
node should not be produced. In this test, it depicts the following f64
extending load used in a v2f64 build vector, but the extending load is actually
used in more places other than the build vector (such as in t12 and t16).
```
Type-legalized selection DAG: %bb.0 'test:entry'
SelectionDAG has 20 nodes:
t0: ch = EntryToken
t4: i64,ch = CopyFromReg t0, Register:i64 %1
t6: i64,ch = CopyFromReg t0, Register:i64 %2
t11: f64,ch = load<(load (s64) from %ir.b, !tbaa !7)> t0, t4, undef:i64
t16: f64 = fadd t31, t37
t34: ch = store<(store (s64) into %ir.c, !tbaa !7)> t31:1, t16, t6, undef:i64
t36: ch = TokenFactor t34, t37:1
t27: v2f64 = BUILD_VECTOR t37, t37
t22: ch,glue = CopyToReg t36, Register:v2f64 $v2, t27
t12: f64 = fadd t11, t37
t28: ch = store<(store (s64) into %ir.b, !tbaa !7)> t11:1, t12, t4, undef:i64
t31: f64,ch = load<(load (s64) from %ir.c, !tbaa !7)> t28, t6, undef:i64
t2: i64,ch = CopyFromReg t0, Register:i64 %0
t37: f64,ch = load<(load (s32) from %ir.a, !tbaa !3), anyext from f32> t0, t2, undef:i64
t23: ch = PPCISD::RET_FLAG t22, Register:v2f64 $v2, t22:1
```
Differential Revision: https://reviews.llvm.org/D117803
In commit 1674d9b6b2, I fixed the bug where we didn't consider
both words of the result of the comparison. However, the logic
needs to be different for eq and ne.
Namely for eq, we need both words of the doubleword to equal so it
is an AND. OTOH for ne, we need either word to be unequal so it
is an OR.
According to GNU as documentation, PowerPC supports some .gnu_attribute
tags to represent the vector and float ABI type in the object file.
Some linkers like GNU ld respects the attribute and will prevent objects
with conflicting ABIs being linked.
This patch emits gnu_attribute value in assembly printer according to
the float-abi metadata. More attributes for soft-fp, hard single/double
and even vector ABI need to be supported in the future.
Reviewed By: jsji
Differential Revision: https://reviews.llvm.org/D117193
During fast-isel calling 'markFunctionEnd' in the base class will call
tidyLandingPads. This can cause an issue where we have determined that
we need ehinfo and emitted a traceback table with the bits set to
indicate that we will be emitting the ehinfo, but the tidying deletes
all landing pads. In this case we end up emitting a reference to
__ehinfo.N symbol, but not emitting a definition to said symbol and the
resulting file fails to assemble.
Differential Revision: https://reviews.llvm.org/D117040
An sra is basically sign-extending a narrower value. Fold away the
shift by doing a sextload of a narrower value, when it is legal to
reduce the load width accordingly.
Differential Revision: https://reviews.llvm.org/D116930
This patch emits a warning when the stack pointer register (`R1`) is found in
the clobber list of an inline asm statement. Clobbering the stack pointer is
not supported.
Reviewed By: #powerpc, nemanjai
Differential Revision: https://reviews.llvm.org/D112073
SelectionDAG::getNode() canonicalises constants to the RHS if the
operation is commutative, but it doesn't do so for constant splat
vectors. Doing this early helps making certain folds on vector types,
simplifying the code required for target DAGCombines that are enabled
before Type legalization.
Somewhat to my surprise, DAGCombine doesn't seem to traverse the
DAG in a post-order DFS, so at the time of doing some custom fold where
the input is a MUL, DAGCombiner::visitMUL hasn't yet reordered the
constant splat to the RHS.
This patch leads to a few improvements, but also a few minor regressions,
which I traced down to D46492. When I tried reverting this change to see
if the changes were still necessary, I ran into some segfaults. Not sure
if there is some latent bug there.
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D117794
The bulk of the implementation is common between 'release' mode (==AOT-ed
model) and 'development' mode (for training), the main difference is
that in development mode, we may also log features (for training logs),
inject scoring information (currently after the Virtual Register
Rewriter) and then produce the log file.
This patch also introduces the score injection pass, 'Register
Allocation Pass Scoring', which is trivially just logging the score in
development mode.
Differential Revision: https://reviews.llvm.org/D117147
When doing the float to int conversion the strict conversion also needs to
retun a chain. This patch fixes that.
Reviewed By: nemanjai, #powerpc, qiucf
Differential Revision: https://reviews.llvm.org/D117464
FAST-ISEL should fall back to DAG-ISEL when a global variable has the
toc-data attribute. A number of the checks were duplicated in the lit
test becuase of
1) Slightly different output between -O0 and -O2 due to FAST-ISEL vs
DAG-ISEL codegen.
2) In preperation of a peephole optimization that will run when
optimizations are enabled.
Differential Revision: https://reviews.llvm.org/D115373
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
As pointed out in https://reviews.llvm.org/D115688#inline-1108193, we
don't want to sink the save point past an INLINEASM_BR, otherwise
prologepilog may incorrectly sink a prolog past the MBB containing an
INLINEASM_BR and into the wrong MBB.
ShrinkWrap is getting this wrong because LR is not in the list of callee
saved registers. Specifically, ShrinkWrap::useOrDefCSROrFI calls
RegisterClassInfo::getLastCalleeSavedAlias which reads
CalleeSavedAliases which was populated by
RegisterClassInfo::runOnMachineFunction by iterating the list of
MCPhysReg returned from MachineRegisterInfo::getCalleeSavedRegs.
Because PPC's LR is non-allocatable, it's NOT considered callee saved.
Add an interface to TargetRegisterInfo for such a case and use it in
Shrinkwrap to ensure we don't sink a prolog past an INLINEASM or
INLINEASM_BR that clobbers LR.
Reviewed By: jyknight, efriedma, nemanjai, #powerpc
Differential Revision: https://reviews.llvm.org/D116424
This commit fixes a missed opportunity in merging consecutive stores.
The code that searches for stores skipped the case of stores that
directly connect to the root. The comment above the implementation lists
this case but the code did not handle it. I found this pattern when
looking into the shared_ptr destructor. GCC generates the right
sequence. Here is a small repo:
int foo(int* buff) {
buff[0] = 0;
int x = buff[1];
buff[1] = 0;
return x;
}
Differential Revision: https://reviews.llvm.org/D116895
Stefan Pintilie