Summary:
Porting HWASan to Linux x86-64, first of the three patches, LLVM part.
The approach is similar to ARM case, trap signal is used to communicate
memory tag check failure. int3 instruction is used to generate a signal,
access parameters are stored in nop [eax + offset] instruction immediately
following the int3 one.
One notable difference is that x86-64 has to untag the pointer before use
due to the lack of feature comparable to ARM's TBI (Top Byte Ignore).
Reviewers: eugenis
Subscribers: kristof.beyls, llvm-commits
Differential Revision: https://reviews.llvm.org/D44699
llvm-svn: 328342
This patch fixes PR36658, "Constant pool entry out of range!" in Thumb1 mode.
In ARMConstantIslands::optimizeThumb2JumpTables() in Thumb1 mode,
adjustBBOffsetsAfter() is not calculating postOffset correctly by
properly accounting for the padding that is required for the constant pool
that immediately follows the jump table branch instruction.
Reviewers: t.p.northover, eli.friedman
Reviewed By: t.p.northover
Subscribers: chrib, tstellar, javed.absar, kristof.beyls, llvm-commits
Differential Revision: https://reviews.llvm.org/D44709
llvm-svn: 328341
When converting an instruction to the wider version, copy any
subregisters if the original operand has a subregister.
Patch by Brendon Cahoon.
llvm-svn: 328333
This patch adds functions to allow MachineLICM to hoist invariant stores.
Currently, MachineLICM does not hoist any store instructions, however
when storing the same value to a constant spot on the stack, the store
instruction should be considered invariant and be hoisted. The function
isInvariantStore iterates each operand of the store instruction and checks
that each register operand satisfies isCallerPreservedPhysReg. The store
may be fed by a copy, which is hoisted by isCopyFeedingInvariantStore.
This patch also adds the PowerPC changes needed to consider the stack
register as caller preserved.
Differential Revision: https://reviews.llvm.org/D40196
llvm-svn: 328326
Loads and stores can only shift the offset register by the size of the value
being loaded, but currently the DAGCombiner will reduce the width of the load
if it's followed by a trunc making it impossible to later combine the shift.
Solve this by implementing shouldReduceLoadWidth for the AArch64 backend and
make it prevent the width reduction if this is what would happen, though do
allow it if reducing the load width will let us eliminate a later sign or zero
extend.
Differential Revision: https://reviews.llvm.org/D44794
llvm-svn: 328321
This was due to a misunderstanding over what llvm calls a micro-op (retirement unit) is actually called a macro-op on the AMD/Jaguar target. Folded loads don't affect num macro ops.
llvm-svn: 328320
When building the SLP tree, we look for reuse among the vectorized tree
entries. However, each gather sequence is represented by a unique tree entry,
even though the sequence may be identical to another one. This means, for
example, that a gather sequence with two uses will be counted twice when
computing the cost of the tree. We should only count the cost of the definition
of a gather sequence rather than its uses. During code generation, the
redundant gather sequences are emitted, but we optimize them away with CSE. So
it looks like this problem just affects the cost model.
Differential Revision: https://reviews.llvm.org/D44742
llvm-svn: 328316
Summary:
Some targets does not support labels inside debug sections, but support
references in form `section+offset`. Patch adds initial support
for this.
Reviewers: echristo, probinson, jlebar
Subscribers: llvm-commits, JDevlieghere
Differential Revision: https://reviews.llvm.org/D43943
llvm-svn: 328314
When targeting execute-only and fp-armv8, float constants in a compare
resulted in instruction selection failures. This is now fixed by using
vmov.f32 where possible, otherwise the floating point constant is
lowered into a integer constant that is moved into a floating point
register.
This patch also restores using fpcmp with immediate 0 under fp-armv8.
Change-Id: Ie87229706f4ed879a0c0cf66631b6047ed6c6443
llvm-svn: 328313
For comparisons with parameters, we can use the ParamState lattice
elements which also provide constant range information. This improves
the code for PR33253 further and gets us closer to use
ValueLatticeElement for all values.
Also, as we are using the range information in the solver directly, we
do not need tryToReplaceWithConstantRange afterwards anymore.
Reviewers: dberlin, mssimpso, davide, efriedma
Reviewed By: mssimpso
Differential Revision: https://reviews.llvm.org/D43762
llvm-svn: 328307
Loop peeling also has an impact on the induction variables, so we should
benefit from induction variable simplification after peeling too.
Reviewers: sanjoy, bogner, mzolotukhin, efriedma
Reviewed By: efriedma
Differential Revision: https://reviews.llvm.org/D43878
llvm-svn: 328301
The VMOVMSKBrr was in a separate InstRW with a lower latency, but I assume they should be the same and the higher latency matches Agners table so I'm going with that.
llvm-svn: 328291
Summary:
Revert r325687 workaround for PR36032 since
a fix was committed in r326154.
Reviewers: sbaranga
Differential Revision: http://reviews.llvm.org/D44768
From: Evgeny Stupachenko <evstupac@gmail.com>
<evgeny.v.stupachenko@intel.com>
llvm-svn: 328257
This makes the Y position consistent with other instructions.
This should have been NFC, but while refactoring the multiclass I noticed that VROUNDPD memory forms were using the register itinerary.
llvm-svn: 328254
In our real world application, we found the following optimization is missed in DAGCombiner
(zext (and/or/xor (shl/shr (load x), cst), cst)) -> (and/or/xor (shl/shr (zextload x), (zext cst)), (zext cst))
If the user of original zext is an add, it may enable further lea optimization on x86.
This patch add a new function CombineZExtLogicopShiftLoad to do this optimization.
Differential Revision: https://reviews.llvm.org/D44402
llvm-svn: 328252
Transforms/Scalar/SCCP.cpp implemented both the Scalar and IPO SCCP, but
this meant Transforms/Scalar including Transfroms/IPO headers, creating
a circular dependency. (IPO depends on Scalar already) - so move the IPO
SCCP shims out into IPO and the basic library implementation accessible
from Scalar/SCCP.h to be used from the IPO/SCCP.cpp implementation.
llvm-svn: 328250
Value::printAsOperand has been scanning the entire module just to
print a single value as an operand, regardless being asked to print a
type or not at all, and regardless really needing to scan the module
to print a type.
It made some of the users of the method exceptionally slow on large
IR-modules (or large MIR-files with large IR-modules embedded).
This patch defers scanning a module looking for struct types, mostly
numbered struct types, as much as possible, speeding up those users
w/o changing any APIs at all.
See speedup examples below:
Release Build:
# 83 seconds -> 5.5 seconds
time ./bin/llc -start-before=irtranslator -stop-after=irtranslator \
-global-isel -global-isel-abort=2 -simplify-mir sqlite3.O0.ll -o \
sqlite3.O0.ll.regbankselected.mir
# 133 seconds -> 6.2 seconds
time ./bin/opt sqlite3.O0.ll -dot-cfg -disable-output
Release + Asserts Build:
# 95 seconds -> 5.5 seconds
time ./bin/llc -start-before=irtranslator -stop-after=irtranslator \
-global-isel -global-isel-abort=2 -simplify-mir sqlite3.O0.ll -o \
sqlite3.O0.ll.regbankselected.mir
# 146 seconds -> 6.2 seconds
time ./bin/opt sqlite3.O0.ll -dot-cfg -disable-output
# 1096 seconds -> 553 seconds
time ./bin/llc -debug-only=isel -fast-isel=false -stop-after=isel \
sqlite3.O0.ll -o /dev/null 2> err
where sqlite3.O0.ll is non-optimized IR produced from
sqlite-amalgamation (http://sqlite.org/download.html), which is entire
SQLite3 implementation in a single C-file.
Benchmarked on 4-cores / 8 threads PCI-E SSD iMac running macOS
Reviewers: dexonsmith, bkramer, void, chandlerc, aditya_nandakumar, dsanders, qcolombet,
Reviewed By: bogner
Subscribers: thegameg, llvm-commits
Differential Revision: https://reviews.llvm.org/D44132
llvm-svn: 328246
Summary:
This reverts commit 364eb09576a7667bc6d3ff80c52a83014ccac976 and separates out
the portion that was fixing binary reader error propagation - turns out, there
are production cases where that causes a regression.
Will re-introduce the error propagation fix separately.
The fix to the text reader error propagation is still "in".
Reviewers: bkramer
Reviewed By: bkramer
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D44807
llvm-svn: 328244
Krzysztof Parzyszek