A new basic block ordering improving existing MachineBlockPlacement.
The algorithm tries to find a layout of nodes (basic blocks) of a given CFG
optimizing jump locality and thus processor I-cache utilization. This is
achieved via increasing the number of fall-through jumps and co-locating
frequently executed nodes together. The name follows the underlying
optimization problem, Extended-TSP, which is a generalization of classical
(maximum) Traveling Salesmen Problem.
The algorithm is a greedy heuristic that works with chains (ordered lists)
of basic blocks. Initially all chains are isolated basic blocks. On every
iteration, we pick a pair of chains whose merging yields the biggest increase
in the ExtTSP value, which models how i-cache "friendly" a specific chain is.
A pair of chains giving the maximum gain is merged into a new chain. The
procedure stops when there is only one chain left, or when merging does not
increase ExtTSP. In the latter case, the remaining chains are sorted by
density in decreasing order.
An important aspect is the way two chains are merged. Unlike earlier
algorithms (e.g., based on the approach of Pettis-Hansen), two
chains, X and Y, are first split into three, X1, X2, and Y. Then we
consider all possible ways of gluing the three chains (e.g., X1YX2, X1X2Y,
X2X1Y, X2YX1, YX1X2, YX2X1) and choose the one producing the largest score.
This improves the quality of the final result (the search space is larger)
while keeping the implementation sufficiently fast.
Differential Revision: https://reviews.llvm.org/D113424
MIPS I, II, and III have delay slots for floating point
comparisons and floating point register transfers (mtc1, mfc1).
Currently, these are not taken into account and thus broken code
may be generated on these targets. This patch inserts nops
as necessary, while attempting to leave the current instruction
if it is safe to stay.
The tests in this patch were updated by @sajattack
Patch by @overdrivenpotato (Marko Mijalkovic <marko.mijalkovic97@gmail.com>)
Differential Revision: https://reviews.llvm.org/D115127
The patch adds LIT tests for SULD, SUST, TEX and TLD4 instructions as
a follow up for D112232. There are a number of FIXME marks that
highlight possible bugs or missed instruction variants.
Differential Revision: https://reviews.llvm.org/D114367
Restrict duplicate FP_EXTEND/FP_TRUNC -> LOAD/STORE DAG combines to only
larger than NEON types, as these are the ones for which there is custom
lowering.
Update tests so that they go through memory to improve validation.
Differential Revision: https://reviews.llvm.org/D115166
This patch implements PAC return address signing for armv8-m. This patch roughly
accomplishes the following things:
- PAC and AUT instructions are generated.
- They're part of the stack frame setup, so that shrink-wrapping can move them
inwards to cover only part of a function
- The auth code generated by PAC is saved across subroutine calls so that AUT
can find it again to check
- PAC is emitted before stacking registers (so that the SP it signs is the one
on function entry).
- The new pseudo-register ra_auth_code is mentioned in the DWARF frame data
- With CMSE also in use: PAC is emitted before stacking FPCXTNS, and AUT
validates the corresponding value of SP
- Emit correct unwind information when PAC is replaced by PACBTI
- Handle tail calls correctly
Some notes:
We make the assembler accept the `.save {ra_auth_code}` directive that is
emitted by the compiler when it saves a register that contains a
return address authentication code.
For EHABI we need to have the `FrameSetup` flag on the instruction and
handle the `t2PACBTI` opcode (identically to `t2PAC`), so we can emit
`.save {ra_auth_code}`, instead of `.save {r12}`.
For PACBTI-M, the instruction which computes return address PAC should use SP
value before adjustment for the argument registers save are (used for variadic
functions and when a parameter is is split between stack and register), but at
the same it should be after the instruction that saves FPCXT when compiling a
CMSE entry function.
This patch moves the varargs SP adjustment after the FPCXT save (they are never
enabled at the same time), so in a following patch handling of the `PAC`
instruction can be placed between them.
Epilogue emission code adjusted in a similar manner.
PACBTI-M code generation should not emit any instructions for architectures
v6-m, v8-m.base, and for A- and R-class cores. Diagnostic message for such cases
is handled separately by a future ticket.
note on tail calls:
If the called function has four arguments that occupy registers `r0`-`r3`, the
only option for holding the function pointer itself is `r12`, but this register
is used to keep the PAC during function/prologue epilogue and clobbers the
function pointer.
When we do the tail call we need the five registers (`r0`-`r3` and `r12`) to
keep six values - the four function arguments, the function pointer and the PAC,
which is obviously impossible.
One option would be to authenticate the return address before all callee-saved
registers are restored, so we have a scratch register to temporarily keep the
value of `r12`. The issue with this approach is that it violates a fundamental
invariant that PAC is computed using CFA as a modifier. It would also mean using
separate instructions to pop `lr` and the rest of the callee-saved registers,
which would offset the advantages of doing a tail call.
Instead, this patch disables indirect tail calls when the called function take
four or more arguments and the return address sign and authentication is enabled
for the caller function, conservatively assuming the caller function would spill
LR.
This patch is part of a series that adds support for the PACBTI-M extension of
the Armv8.1-M architecture, as detailed here:
https://community.arm.com/arm-community-blogs/b/architectures-and-processors-blog/posts/armv8-1-m-pointer-authentication-and-branch-target-identification-extension
The PACBTI-M specification can be found in the Armv8-M Architecture Reference
Manual:
https://developer.arm.com/documentation/ddi0553/latest
The following people contributed to this patch:
- Momchil Velikov
- Ties Stuij
Reviewed By: danielkiss
Differential Revision: https://reviews.llvm.org/D112429
LLVM loops cannot represent irreducible structures in the CFG. This
change introduce the concept of cycles as a generalization of loops,
along with a CycleInfo analysis that discovers a nested
hierarchy of such cycles. This is based on Havlak (1997), Nesting of
Reducible and Irreducible Loops.
The cycle analysis is implemented as a generic template and then
instatiated for LLVM IR and Machine IR. The template relies on a new
GenericSSAContext template which must be specialized when used for
each IR.
This review is a restart of an older review request:
https://reviews.llvm.org/D83094
Original implementation by Nicolai Hähnle <nicolai.haehnle@amd.com>,
with recent refactoring by Sameer Sahasrabuddhe <sameer.sahasrabuddhe@amd.com>
Differential Revision: https://reviews.llvm.org/D112696
When compiling with indirect branch tracking and fentry (-fcf-protection=branch -mfentry -pg) the X86IndirectBranchTrackingPass will attempt to place endbr in basic blocks, checking for Calls/IsCallReturnTwice. For calling the function IsCallReturnTwice(), the pass attempts to retrieve the first operand of the respective machine instruction. Since FENTRY_CALL is considered a call, and it does not have any argument, the condition inside the pass will attempt to call IsCallReturnTwice on the machine instruction, but since it does not have operands, it will lead into a crash.
Kudos to Alyssa Milburn for helping in the issue triage. The diff brings a test, but to reproduce the problem, follow the steps below.
```
echo "int main() {};" > repro.c
clang repro.c -fcf-protection=branch -mfentry -pg
```
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D111108
Memset with a constant length was implemented with a single store followed by
a series of MVC:s. This patch changes this so that one store of the byte is
emitted for each MVC, which avoids data dependencies between the MVCs. An
MVI/STC + MVC(len-1) is done for each block.
In addition, memset with a variable length is now also handled without a
libcall. Since the byte is first stored and then MVC is used from that
address, a length of two must now be subtracted instead of one for the loop
and EXRL. This requires an extra check for the one-byte case, which is
handled in a special block with just a single MVI/STC (like GCC).
Review: Ulrich Weigand
Differential Revision: https://reviews.llvm.org/D112004
The immediate size check on StepNumerator did not take into account
that vmul.vi does not exist. It also did not account for power of 2
constants that can be done with vshl.vi.
This patch fixes this by moving the conversion from mul to shift
further up. Then we can consider the immediates separately for MUL
vs SHL. For MUL I've allowed simm12 which requires a single addi
before a vmul.vx. For SHL I've allowed any uimm5 which works with
vshl.vi. We could relax these further in the future. This is a
starting point that allows us to emit the same number of instructions
we were already using for smaller numerators.
Reviewed By: frasercrmck
Differential Revision: https://reviews.llvm.org/D115081
A new basic block ordering improving existing MachineBlockPlacement.
The algorithm tries to find a layout of nodes (basic blocks) of a given CFG
optimizing jump locality and thus processor I-cache utilization. This is
achieved via increasing the number of fall-through jumps and co-locating
frequently executed nodes together. The name follows the underlying
optimization problem, Extended-TSP, which is a generalization of classical
(maximum) Traveling Salesmen Problem.
The algorithm is a greedy heuristic that works with chains (ordered lists)
of basic blocks. Initially all chains are isolated basic blocks. On every
iteration, we pick a pair of chains whose merging yields the biggest increase
in the ExtTSP value, which models how i-cache "friendly" a specific chain is.
A pair of chains giving the maximum gain is merged into a new chain. The
procedure stops when there is only one chain left, or when merging does not
increase ExtTSP. In the latter case, the remaining chains are sorted by
density in decreasing order.
An important aspect is the way two chains are merged. Unlike earlier
algorithms (e.g., based on the approach of Pettis-Hansen), two
chains, X and Y, are first split into three, X1, X2, and Y. Then we
consider all possible ways of gluing the three chains (e.g., X1YX2, X1X2Y,
X2X1Y, X2YX1, YX1X2, YX2X1) and choose the one producing the largest score.
This improves the quality of the final result (the search space is larger)
while keeping the implementation sufficiently fast.
Differential Revision: https://reviews.llvm.org/D113424
f526c600c043 had a concern raised because of an invalid typesize request
on a scalable vector, which this patch addresses.
Prevent shouldReduceLoadWidth from attempting to query the bit size, and
add a regression test in sve-extract-fixed-vector.ll.
Differential Revision: https://reviews.llvm.org/D115156
This patch implements the intrinsic for ref.null.
In the process of implementing int_wasm_ref_null_func() and
int_wasm_ref_null_extern() intrinsics, it removes the redundant
HeapType.
This also causes the textual assembler syntax for ref.null to
change. Instead of receiving an argument: `func` or `extern`, the
instruction mnemonic is either ref.null_func or ref.null_extern,
without the need for a further operand.
Reviewed By: tlively
Differential Revision: https://reviews.llvm.org/D114979
Add all CompressPat to map instructions between 16-bit and 32-bit with using the CompressInstEmitter infra.
Although it's only used in asm printer, also enable it in asm parser to debug mapping when -enable-csky-asm-compressed-inst is on.
Differential Revision: https://reviews.llvm.org/D115026
Implement 'back-to-back' FX fusion according to Power10 User Manual
'19.1.5.4 Fusion', not enabled by default.
Reviewed By: nemanjai
Differential Revision: https://reviews.llvm.org/D114345
Expanding on D109750.
Since `DBG_VALUE` instructions have final register validity determined in
`LDVImpl::handleDebugValue`, there is no apparent reason to immediately prune
unused register operands as their defs are erased. Consequently, this renders
`MachineInstr::eraseFromParentAndMarkDBGValuesForRemoval` moot; gaining a
substantial performance improvement.
The only necessary changes involve making relevant passes consider invalid
DBG_VALUE vregs uses as valid.
Reviewed By: MatzeB
Differential Revision: https://reviews.llvm.org/D112852
This patch proposes to move emission of global variables, types,
imported entities, etc from DwarfDebug::beginModule() to DwarfDebug::endModule().
Effectively, this changes nothing but the order of debug entities which
will be as follows:
* subprograms (including related context, local variables/labels,
local imported entities; related types can be created as a part of
the emission of local entities of an abstract subprogram);
* global variables (including related context and types);
* retained types and enums;
* non-local-scoped imported entities;
* basic types;
* other types left (as a part of local variables attributes emission).
Note that the order of emitted compile units may also be changed as now we emit
units that contain subprograms first and then all other non-empty units.
The motivation behind this change is the following:
(1) DwarfDebug::beginModule() is run at the very beginning of backend's pipeline,
from this time IR can be significantly changed by target-specific passes.
If it happens for debug metadata of global entities, those changes will not
be reflected in the emitted DWARF.
(2) imported subprogram names should refer to an abstract subprogram if it exists,
but it isn't known in DwarfDebug::beginModule() (it's possible to make some
guesses based on location info, but it's not quite reliable);
(3) aforementioned entities if they are scoped within a bracketed block
(subject of D113741) couldn't be emitted in DwarfDebug::beginModule()
(they need parent emitted first). Another problem is if to try to gather
some information about local entities and defer their emission
(till subprogram's processing or DwarfDebug::endModule()) all the gathered
details might be irrelevant / invalid by the time the entities are being
emitted (because of (1)).
Reviewed By: dblaikie
Differential Revision: https://reviews.llvm.org/D114705
As an extension to D111976, this converts clamp fptosi, clamped between
0 and (2^n)-1 to a fptoui.sat. This can greatly help on targets with
conversions that naturally saturate, such as Arm.
X86 disables the transform as some of the test cases increases in size.
A fptoui.sat necessitates a fp clamp without native support, so there is
little use in converting if the instruction is just going to be
expanded.
Differential Revision: https://reviews.llvm.org/D112428
This reverts commits
* ee691970a9a85470948ada623c31f0ab8773617c (D113741),
* 79d3132998b2828be8f7d2ec411f91fb11b3e01f (D114705)
due to lldb and dexter test failures.
Code using indirect calls is broken without this, and there isn't
really much value in supporting the old attempt to vary the argument
placement based on uses. This resulted in more argument shuffling code
anyway.
Also have the option stop implying all inputs need to be passed. This
will no rely on the amdgpu-no-* attributes to avoid passing
unnecessary values.
Previously we would require adding an attribute to kernels to enable
the inputs passed in the kernarg segment, accessed by
llvm.amdgcn.implicitarg.ptr. This violates the principle of being
correct by default. Some OpenMP testcases were broken recently since
it wasn't correctly setting this attribute, and no known frontends are
setting this to anything other than the maximum.
Most of the test changes are from load widening of argument loads
since there now more implied dereferenceable bytes.
This patch proposes to move emission of global variables, types,
imported entities, etc from DwarfDebug::beginModule() to DwarfDebug::endModule().
Effectively, this changes nothing but the order of debug entities which
will be as follows:
* subprograms (including related context, local variables/labels,
local imported entities; related types can be created as a part of
the emission of local entities of an abstract subprogram);
* global variables (including related context and types);
* retained types and enums;
* non-local-scoped imported entities;
* basic types;
* other types left (as a part of local variables attributes emission).
Note that the order of emitted compile units may also be changed as now we emit
units that contain subprograms first and then all other non-empty units.
The motivation behind this change is the following:
(1) DwarfDebug::beginModule() is run at the very beginning of backend's pipeline,
from this time IR can be significantly changed by target-specific passes.
If it happens for debug metadata of global entities, those changes will not
be reflected in the emitted DWARF.
(2) imported subprogram names should refer to an abstract subprogram if it exists,
but it isn't known in DwarfDebug::beginModule() (it's possible to make some
guesses based on location info, but it's not quite reliable);
(3) aforementioned entities if they are scoped within a bracketed block
(subject of D113741) couldn't be emitted in DwarfDebug::beginModule()
(they need parent emitted first). Another problem is if to try to gather
some information about local entities and defer their emission
(till subprogram's processing or DwarfDebug::endModule()) all the gathered
details might be irrelevant / invalid by the time the entities are being
emitted (because of (1)).
Reviewed By: dblaikie
Differential Revision: https://reviews.llvm.org/D114705
The ray_origin, ray_dir and ray_inv_dir arguments should all be vec3 to
match how the hardware instruction works.
Don't change the API of the corresponding OpenCL builtins.
Differential Revision: https://reviews.llvm.org/D115032
When Control Flow Guard Check is inserted, funclet bundle was not checked. Therefore, it didn't generate code correctly when a target function has "funclet" bundle.
Reviewed By: rnk
Differential Revision: https://reviews.llvm.org/D114914
Two-address pass works right before RA and if an immediate
was folded into an instruction there is nothing to remove
the dead def. We end up with something like:
v_mov_b32_e32 v14, 0xc1700000
v_mov_b32_e32 v14, 0x41200000
v_fmaak_f32 v51, s67, v19, 0xc1700000
v_fmaak_f32 v38, v51, v19, 0x4120000
The patch kills the dead move instruction right in the folding.
Differential Revision: https://reviews.llvm.org/D114999
Prior to this patch, tail duplication handled debug info poorly -
specifically, debug instructions would be dropped instead of being set
undef, potentially extending the lifetimes of prior debug values that
should be killed. The pass was also very aggressive with dropping debug
info, dropping debug info even when the SSA value it referred to was
still present. This patch attempts to handle debug info more carefully,
checking to see whether each affected debug value can still be live,
setting it undef if not.
Reviewed By: jmorse
Differential Revision: https://reviews.llvm.org/D106875
This adjusts all the MVE and CDE intrinsics now that v2i1 is a legal
type, to use a <2 x i1> as opposed to emulating the predicate with a
<4 x i1>. The v4i1 workarounds have been removed leaving the natural
v2i1 types, notably in vctp64 which now generates a v2i1 type.
AutoUpgrade code has been added to upgrade old IR, which needs to
convert the old v4i1 to a v2i1 be converting it back and forth to an
integer with arm.mve.v2i and arm.mve.i2v intrinsics. These should be
optimized away in the final assembly.
Differential Revision: https://reviews.llvm.org/D114455
The Power ISA defined l[bhwdq]arx as both base and
extended mnemonics. The base mnemonic takes the EH
bit as an operand and the extended mnemonic omits
it, making it implicitly zero. The existing
implementation only handles the base mnemonic when
EH is 1 and internally produces a different
instruction. There are historical reasons for this.
This patch simply removes the limitation introduced
by this implementation that disallows the base
mnemonic with EH = 0 in the ASM parser.
This resolves an issue that prevented some files
in the Linux kernel from being built with
-fintegrated-as.
Also fix a crash if the value is not an integer immediate.
MVE can treat v16i1, v8i1, v4i1 and v2i1 as different views onto the
same 16bit VPR.P0 register, with v2i1 holding two 8 bit values for the
two halves. This was never treated as a legal type in llvm in the past
as there are not many 64bit instructions and no 64bit compares. There
are a few instructions that could use it though, notably a VSELECT (as
it can handle any size using the underlying v16i8 VPSEL), AND/OR/XOR for
similar reasons, some gathers/scatter and long multiplies and VCTP64
instructions.
This patch goes through and makes v2i1 a legal type, handling all the
cases that fall out of that. It also makes VSELECT legal for v2i64 as a
side benefit. A lot of the codegen changes as a result - usually in way
that is a little better or a little worse, but still expensive. Costs
can change a little too in the process, again in a way that expensive
things remain expensive. A lot of the tests that changed are mainly to
ensure correctness - the code can hopefully be improved in the future
where it comes up in practice.
The intrinsics currently remain using the v4i1 they previously did to
emulate a v2i1. This will be changed in a followup patch but this one
was already large enough.
Differential Revision: https://reviews.llvm.org/D114449
As discussed on D114589 as the constant case gets affected by SimplifyDemandedBits a lot - the non-constant case currently falls back to copysignl libcalls
Add clamp combine. Source is fminnum(fmaxnum(Val, 0.0), 1.0) or
fmaxnum(fminnum(Val, 1.0), 0.0) or fmed3 intrinsic with 0.0 and
1.0 as two out of three operands.
Differential Revision: https://reviews.llvm.org/D90052
Add floating point version of med3 combine.
Source is fminnum(fmaxnum(Val, K0), K1) or fmaxnum(fminnum(Val, K1), K0)
where K0 and K1 are constants and K0 <= K1.
Differential Revision: https://reviews.llvm.org/D90051
Recognize constant splat padded with undef in isCanonicalized.
Fcanonicalize will be removed by RemoveFcanonicalize in post-legalizer
combiner. We will treat undef as value that will result in a splat
in clamp combine after regbankselect.
Differential Revision: https://reviews.llvm.org/D104408
Changed registers to R10 and R11 because PLT resolution clobbers them. Also changed the implementation to use R11 instead of RCX, which saves a push/pop.
Reviewed By: vitalybuka
Differential Revision: https://reviews.llvm.org/D115002
Do not infer no-amdgpu-implicitarg-ptr for sanitized functions. If a
function is explicitly marked amdgpu-no-implicitarg-ptr and
sanitize_address, infer that it is required.
The load/store infrastructure previously made an incorrect assumption that
whenever it is used with a load/store intrinsic on Power10 - those intrinsics
would automatically be the lxvp/stxvp intrinsics introduced in Power10.
However, this is obviously not the case as there are multiple instances of
pre-P10 intrinsics that use the refactored load/store implementation.
This patch corrects this assumption, and produces the expected intrinsic on pre-P10.
Differential Revision: https://reviews.llvm.org/D114978
