Summary: This prevents a bug when passing nullptr, StringRef ctor would call strlen(nullptr).

Reviewers: vlad.tsyrklevich

Subscribers: cfe-commits

Tags: #clang

Differential Revision: https://reviews.llvm.org/D69569

llvm/test/DebugInfo/MIR/X86/live-debug-values-reg-copy.mir failed with
EXPENSIVE_CHECKS enabled, causing the patch to be reverted in
rG2c496bb5.

This patch relands the patch with a proper fix to the
live-debug-values-reg-copy.mir tests, by ensuring the MIR encodes the
callee-saves correctly so that the CalleeSaved info is taken from MIR
directly, rather than letting it be recalculated by the PEI pass. I've
done this by running `llc -stop-before=prologepilog` on the LLVM
IR as captured in the test files, adding the extra MOV instructions
that were manually added in the original test file, then running `llc
-run-pass=prologepilog` and finally re-added the comments for the MOV
instructions.

Stores are vectorized with maximum vectorization factor of 16. Patch
tries to improve the situation and use maximal vectorization factor.

Reviewers: spatel, RKSimon, mkuper, hfinkel

Differential Revision: https://reviews.llvm.org/D43582

See D68028

Apple's greendragon bot uses a standalone build of lldb which would
fail to build after a recent change to LLVM as it relied on LLVM setting
global flags for its build.  Attempt to repair the standalone build for
greendragon bot.

`LLVM_DEFAULT_TARGET_TRIPLE` is a cached variable, which means that it
may actually be unset.  Furthermore, in standalone builds, the variable
may be fully undefined.  Apply the regular expression over the empty
string in such a case.  This should improve the state of the green
dragon bot.

This is a fix for:
https://bugs.llvm.org/show_bug.cgi?id=43730
...and as shown there, we have existing test cases that show potential miscompiles.

We could just bail out for vector constants that contain any undef elements, or we can do as shown here:
allow the transform, but replace the undefs with a safe value.

For most of the tests shown, this results in a full splat constant (no undefs) which is probably a win
for further IR analysis because we conservatively don't match undefs in most cases. Codegen can probably
recover these kinds of undef lanes via demanded elements analysis if that's profitable.

Differential Revision: https://reviews.llvm.org/D69519

Summary:
This is a follow up on https://reviews.llvm.org/D61634
This patch is simpler and only adds the no_builtin attribute.

Reviewers: tejohnson, courbet, theraven, t.p.northover, jdoerfert

Subscribers: mgrang, cfe-commits

Tags: #clang

Differential Revision: https://reviews.llvm.org/D68028

This is a re-submit after it got reverted in https://reviews.llvm.org/rGbd8791610948 since the breakage doesn't seem to come from this patch.

The giant if-else isn't conforming to LLVM code style.

Need to analyze inner target regions in case of implicit mapping of the
data members when target region is created in one of the class member
functions.

In some cases, we fail to reduce the pass list earlier because of
complex pass dependencies, but we can reduce it after we simplified the
reproducer.

An example of that is PR43474, which can limit the crash to
-loop-interchange. Adding a test case would require at least 2
interacting Loop passes I think.

Reviewers: davide, reames, modocache

Reviewed By: reames

Differential Revision: https://reviews.llvm.org/D69236

Increase coverage for D69519.

This fixes https://llvm.org/PR43829.

Define BitVector::BitWord as uintptr_t instead of unsigned long, as long does not necessarily translates to a pointer size (especially on 64-bit Visual Studio).

Committed on behalf of @ekatz (Ehud Katz)

Differential Revision: https://reviews.llvm.org/D69336

This should trigger a dereference before null-check warning,
but I don't see it when building with clang. In any case, the
current and known future users of this helper require non-null
args, so I'm converting the 'if' to an assert.

Associate the integer literals with BitWord, instead of the embedded UL suffix.

Required for https://reviews.llvm.org/D69336.

Committed on behalf of @ekatz (Ehud Katz)

Differential Revision: https://reviews.llvm.org/D69357

This is the NFC part of D69519.
We had this functionality locally in instcombine, but it can be used
elsewhere, so hoisting it to Constant class.

The virtual container/header section caused the section list to be
offset by one, but by using FindSectionByID, the layout of the
section list shouldn't matter.

Differential Revision: https://reviews.llvm.org/D69366

(I'm currently trying to debug a strange error message I get when
pushing to github, despite the pushes being successful).

This reverts commit 8af5ada0.

As Bjorn pointed out in D68816, the iteration over `UserVals` may not be safe.

Reverting on behalf of Orlando.

Summary:
This provides a convenient way to see the SymbolID/USR of the symbol, mainly
for debugging purpose.

Reviewers: ilya-biryukov

Subscribers: MaskRay, jkorous, arphaman, kadircet, usaxena95, cfe-commits

Tags: #clang

Differential Revision: https://reviews.llvm.org/D69517

Summary:
Currently we only forget the loop we added LCSSA phis for. But SCEV
expressions in other loops could also depend on the instruction we added
a PHI for and currently we do not invalidate those expressions. This can
happen when we use ScalarEvolution before converting a function to LCSSA
form. The SCEV expressions will refer to the non-LCSSA value. If this
SCEV expression is then used with the expander, we do not preserve LCSSA
form.

This patch properly forgets the values we created PHIs for. Those need
to be recomputed again. This patch fixes PR43458.

Currently SCEV::verify does not catch this mismatch and any test would
need to run multiple passes to trigger the error (e.g. -loop-reduce
-loop-unroll). I will also look into catching this kind of mismatch in
the verifier. Also, we currently forget the whole loop in LCSSA and I'll
check if we can be more surgical.

Reviewers: efriedma, sanjoy.google, reames

Reviewed By: efriedma

Subscribers: zzheng, hiraditya, javed.absar, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D68194

Revert rG70f5aece - "Reland [AArch64][DebugInfo] Do not recompute CalleeSavedStackSize (Take 2)"

This fails on EXPENSIVE_CHECKS builds

Use the existing helper function in BranchFolding, "countsAsInstruction",
to skip over non-instructions. Otherwise debug instructions can be
identified as the last real instruction in a block, leading to different
codegen decisions when debug is enabled as demonstrated by the test case.

Patch by: yechunliang (Chris Ye)!

Differential Revision: https://reviews.llvm.org/D66467

This fixes a minor oversight mentioned in the review of D69379:
we should push extractelement into the operands of getelementptr
regardless of whether that enables further folding.

Addresses post-commit comments for D69160.

To address post commit review comment for D69167.

Reviewers: ilya-biryukov

Subscribers: MaskRay, jkorous, arphaman, kadircet, usaxena95, cfe-commits

Tags: #clang

Differential Revision: https://reviews.llvm.org/D69506

Before this change .symtab section was required for SHT_REL[A] section
declarations. yaml2obj automatically defined it in case when YAML document
did not have it.

With this change it is now possible to produce an object that
has a relocation section, but has no symbol table.

It simplifies the code and also it is inline with how we handle Link fields
for another special sections.

Differential revision: https://reviews.llvm.org/D69260

Currently, when we do not specify "Info" field in a YAML description
for SHT_GROUP section, yaml2obj reports an error:
"error: unknown symbol referenced: '' by YAML section '.group1'"

Also, we do not link it with a symbol table by default,
though it is what we do for AddrsigSection, HashSection, RelocationSection.
(http://www.sco.com/developers/gabi/latest/ch4.sheader.html#sh_link)

The patch fixes missings mentioned.

Differential revision: https://reviews.llvm.org/D69299

Sections may have zero size and zero-sized sections may share a start address
with other zero-sized sections. For the section overlap test to function
correctly zero-sized sections must be ordered before any non-zero sized ones.

This should fix the intermittent failures in the
test/ExecutionEngine/JITLink/X86/MachO_zero_fill_alignment.s test case that
have been observed on some builders.

When TableGen is inferring register classes from contexts, it uses a
sorting function based on the number of registers in the class. Since
this was being treated as an alias of VGPR_32, they had exactly the
same size. The sort used wasn't a stable sort, and even if it were, I
believe the tie breaker would effectively end up being the
alphabetical ordering of the class name. There appear to be issues
trying to use an empty set of registers, so add only one so this will
always sort to the end.

Also add a comment explaining how VReg_1 is a dirty hack for
SelectionDAG.

This does end up changing the behavior of i1 with inline asm and VGPR
constraints, but the existing behavior was was already nonsensical and
inconsistent. It should probably be disallowed anyway.

Fixes bug 43699

Remove RA from reserved register list, so we could use it as callee saved register

Differential Revision: https://reviews.llvm.org/D67698

The callback provides no benefits since `run.execute()` does not take
any arguments anymore.