Refines the gather/scatter cost model, but also changes the TTI
function getIntrinsicInstrCost to accept an additional parameter
which is needed for the gather/scatter cost evaluation.
This did require trivial changes in some non-ARM backends to
adopt the new parameter.
Extending gathers and truncating scatters are now priced cheaper.
Differential Revision: https://reviews.llvm.org/D75525
Summary:
Instead of generating two i32 instructions for each load or store of a volatile
i64 value (two LDRs or STRs), now emit LDRD/STRD.
These improvements cover architectures implementing ARMv5TE or Thumb-2.
The code generation explicitly deviates from using the register-offset
variant of LDRD/STRD. In this variant, the register allocated to the
register-offset cannot be reused in any of the remaining operands. Such
restriction seems to be non-trivial to implement in LLVM, thus it is
left as a to-do.
Reviewers: dmgreen, efriedma, john.brawn, nickdesaulniers
Reviewed By: efriedma, nickdesaulniers
Subscribers: danielkiss, alanphipps, hans, nathanchance, nickdesaulniers, vvereschaka, kristof.beyls, hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D70072
Behavior of IEEEFloat::roundToIntegral is aligned with IEEE-754
operation roundToIntegralExact. In partucular this function now:
- returns opInvalid for signaling NaNs,
- returns opInexact if the result of rounding differs from argument.
Differential Revision: https://reviews.llvm.org/D75246
If the loaded memory size was smaller than the result size, this would
produce out of bounds memory accesses. I'm wondering if we need a
distinct narrow memory legalize action type, since a case I care about
is decomposing a 4-byte unaligned access into 4 extending loads, which
would leave the original result register type. I'm currently awkwardly
using narrowScalar to handle unaligned accesses that need to be split.
Summary:
Enables JIT-linking by RuntimeDyld of COFF objects that contain references to
dllimport symbols. This is done by recognizing symbols that start with the
reserved "__imp_" prefix and building a pointer entry to the target symbol in
the stubs area of the section. References to the "__imp_" symbol are updated to
point to this pointer.
Work in progress: The generic code is in place, but only RuntimeDyldCOFFX86_64
and RuntimeDyldCOFFI386 have been updated to look for and update references to
dllimport symbols.
Reviewers: compnerd
Subscribers: hiraditya, ributzka, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D75884
This patch allows rtdyld-check / jitlink-check expressions to be extended over
multiple lines by terminating each line with a '\'. E.g.
# llvm-rtdyld: *{8}X = \
# llvm-rtdyld: Y
X:
.quad Y
This will be used to break up some long lines in upcoming test cases.
This avoids regressions in a future patch. I'm confused by the use of
the gfx9 usage legacy_mad. Was this a pointless instruction rename, or
uses fmul_legacy handling? Why is regular mac avilable in that case?
Summary:
As far as I can tell on gfx10 conversions to/from f32 (that are not
converting f32 to/from f64) are full rate instructions, but they were
marked as quarter rate instructions.
I have fixed this for gfx10 only. I assume the scheduling model was
correct for older architectures, though I don't have any documentation
handy to confirm that.
Reviewers: rampitec, arsenm
Subscribers: kzhuravl, jvesely, wdng, nhaehnle, yaxunl, dstuttard, tpr, t-tye, hiraditya, kerbowa, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D75392
This essentially reverts some of the SimplifyLibcalls part changes of D45736 [SimplifyLibcalls] Replace locked IO with unlocked IO.
C11 7.21.5.2 The fflush function
> If stream is a null pointer, the fflush function performs this flushing action on all streams for which the behavior is defined above.
i.e. fopen'ed FILE* is inherently captured.
POSIX.1-2017 getc_unlocked, getchar_unlocked, putc_unlocked, putchar_unlocked - stdio with explicit client locking
> These functions can safely be used in a multi-threaded program if and only if they are called while the invoking thread owns the ( FILE *) object, as is the case after a successful call to the flockfile() or ftrylockfile() functions.
After a thread fopen'ed a FILE*, when it is calling foobar() which is now replaced by foobar_unlocked(),
if another thread is concurrently calling fflush(0), the behavior is undefined.
C11 7.22.4.4 The exit function
> Next, all open streams with unwritten buffered data are flushed, all open streams are closed, and all files created by the tmpfile function are removed.
The replacement is only feasible if the program is single threaded, or exit or fflush(0) is never called.
See also http://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20180528/556615.html
for how the replacement makes libc interceptors difficult to implement.
dalias: in a worst case, it's unbounded data corruption because of concurrent access to pointers
without synchronization. f->wpos or rpos could get outside of the buffer, thread A could do
f->wpos += j after knowing j is in bounds, while thread B also changes it concurrently.
This can produce exploitable conditions depending on libc internals.
Revert the SimplifyLibcalls part change because the cons obviously
overweigh the pros. Even when the replacement is feasible, the benefit
is indemonstrable, more so in an application instead of an artificial
glibc benchmark. Theoretically the replacement could be beneficial when
calling getc_unlocked/putc_unlocked in a loop, but then it is better
using a blocked IO operation and the user is likely aware of that.
The function attribute inference is still useful and thus kept.
Reviewed By: xbolva00
Differential Revision: https://reviews.llvm.org/D75933
Summary:
This patch extends the TargetMachine to let targets specify the integer size
used by the sjljehprepare pass. This is 64bit for the VE target and otherwise
defaults to 32bit for all targets, which was hard-wired before.
Reviewed By: arsenm
Differential Revision: https://reviews.llvm.org/D71337
Summary: Rewrite the fsub-0.0 idiom to fneg and always emit fneg for fp
negation. This also extends the scalarization cost in instcombine for unary
operators to result in the same IR rewrites for fneg as for the idiom.
Reviewed By: cameron.mcinally
Differential Revision: https://reviews.llvm.org/D75467
Summary:
Follow up on D75283.
Also remove the test code that was moved to another test and was to be removed.
Reviewers: davidxl
Subscribers: eraman, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D75630
In case the source value ends up in a VGPR, insert a readfirstlane to
avoid producing an illegal copy later. If it turns out to be
unnecessary, it can be folded out.
Swap the compare operands if LHS is spilled while updating the CCMask:s of
the CC users. This is relatively straight forward since the live-in lists for
the CC register can be assumed to be correct during register allocation
(thanks to 659efa2).
Also fold a spilled operand of an LOCR/SELR into an LOC(G).
Review: Ulrich Weigand
Differential Revision: https://reviews.llvm.org/D67437
SimplifyAddWithRemainder currently also matches for vector types, but
tries to create an integer constant, which causes a crash.
By using Constant::getIntegerValue() we can support both the scalar and
vector cases.
The 2 added test cases crash without the fix.
Reviewers: spatel, lebedev.ri
Reviewed By: spatel, lebedev.ri
Differential Revision: https://reviews.llvm.org/D75906
Summary:
This patch generalizes the existing code to support CDE intrinsics
which will share some properties with existing MVE intrinsics
(some of the intrinsics will be polymorphic and accept/return values
of MVE vector types).
Specifically the patch:
* Adds new tablegen backends -gen-arm-cde-builtin-def,
-gen-arm-cde-builtin-codegen, -gen-arm-cde-builtin-sema,
-gen-arm-cde-builtin-aliases, -gen-arm-cde-builtin-header based on
existing MVE backends.
* Renames the '__clang_arm_mve_alias' attribute into
'__clang_arm_builtin_alias' (it will be used with CDE intrinsics as
well as MVE intrinsics)
* Implements semantic checks for the coprocessor argument of the CDE
intrinsics as well as the existing coprocessor intrinsics.
* Adds one CDE intrinsic __arm_cx1 to test the above changes
Reviewers: simon_tatham, MarkMurrayARM, ostannard, dmgreen
Reviewed By: simon_tatham
Subscribers: sdesmalen, mgorny, kristof.beyls, danielkiss, cfe-commits, llvm-commits
Tags: #clang, #llvm
Differential Revision: https://reviews.llvm.org/D75850
SimplifyCFG should not merge empty return blocks and leave a CallBr behind
with a duplicated destination since the verifier will then trigger an
assert. This patch checks for this case and avoids the transformation.
CodeGenPrepare has a similar check which also has a FIXME comment about why
this is needed. It seems perhaps better if these two passes would eventually
instead update the CallBr instruction instead of just checking and avoiding.
This fixes https://bugs.llvm.org/show_bug.cgi?id=45062.
Review: Craig Topper
Differential Revision: https://reviews.llvm.org/D75620
It seems like the SLPVectorizer is currently not aware of vector
versions of functions provided by libraries like Accelerate [1].
This patch updates SLPVectorizer to use the same infrastructure
the LoopVectorizer uses to detect vectorizable library functions.
For calls, it computes the cost of an intrinsic call (existing behavior)
and the cost of a vector function library call, if available. Like
LoopVectorizer, it assumes the cost of the vector function is simply the
cost of a call to a vector function.
[1] https://developer.apple.com/documentation/accelerate
Reviewers: ABataev, RKSimon, spatel
Reviewed By: ABataev
Differential Revision: https://reviews.llvm.org/D75878
According to the Twine.h comment, the Twines should only
be used as const references in arguments.
Differential Revision: https://reviews.llvm.org/D75727
Summary:
When SI_INDIRECT_DST_V* pseudos has indexes in VGPR, they get expanded into the self-looped basic block that modifies EXEC in a loop.
To keep EXEC consistent it is stored before and then re-stored after the pseudo expansion result.
%95:vreg_512 = SI_INDIRECT_DST_V16 %93:vreg_512(tied-def 0), %94:sreg_32, 0, killed %1500:vgpr_32
results to
s_mov_b64 s[6:7], exec
BB0_16:
v_readfirstlane_b32 s8, v28
v_cmp_eq_u32_e32 vcc, s8, v28
s_and_saveexec_b64 vcc, vcc
s_set_gpr_idx_on s8, gpr_idx(DST)
v_mov_b32_e32 v6, v25
s_set_gpr_idx_off
s_xor_b64 exec, exec, vcc
s_cbranch_execnz BB0_16
; %bb.17:
s_mov_b64 exec, s[6:7]
The bug appeared in case this expansion occurs in the ELSE block of the CF.
Originally
%110:vreg_512 = SI_INDIRECT_DST_V16 %103:vreg_512(tied-def 0), %85:vgpr_32, 0, %107:vgpr_32,
%112:sreg_64 = SI_ELSE %108:sreg_64, %bb.19, 0, implicit-def dead $exec, implicit-def dead $scc, implicit $exec
expanded to
****************** <== here exec has "THEN" context
s_mov_b64 s[6:7], exec
BB0_16:
v_readfirstlane_b32 s8, v28
v_cmp_eq_u32_e32 vcc, s8, v28
s_and_saveexec_b64 vcc, vcc
s_set_gpr_idx_on s8, gpr_idx(DST)
v_mov_b32_e32 v6, v25
s_set_gpr_idx_off
s_xor_b64 exec, exec, vcc
s_cbranch_execnz BB0_16
; %bb.17:
s_or_saveexec_b64 s[4:5], s[4:5] <-- exec mask is restored for "ELSE" but immediately overwritten.
s_mov_b64 exec, s[6:7]
The rest of the "ELSE" block is executed not by the workitems which constitute the "else mask" but by those which constitute "then mask"
SILowerControlFlow::emitElse always considers the basic block begin() as an insertion point for s_or_saveexec.
Proposed fix: The SI_INDIRECT_DST_V* procedure should split the reminder block to create landing pad for the EXEC restoration.
Reviewers: rampitec, vpykhtin, nhaehnle
Reviewed By: vpykhtin
Subscribers: arsenm, kzhuravl, jvesely, wdng, yaxunl, dstuttard, tpr, t-tye, hiraditya, kerbowa, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D75472
On some Arm cores there is a performance penalty when forwarding from an
S register to a D register. Calculating VMAX in a D register creates
false forwarding hazards, so don't do that unless we're on a core which
specifically asks for it.
Patch by James Greenhalgh
Differential Revision: https://reviews.llvm.org/D75248
For pre-AVX512 targets, combine binary shuffles to X86ISD::VPERM2X128 if possible. This mainly helps optimize the blend(extract_subvector(x,1),y) pattern.
At some point soon we're going to have make a decision about when to combine AVX512 shuffles more aggressively - we bail out if there is any change in element size (to protect predicate mask merging) which means we miss out on a lot of optimizations.
