rust-lang/llvm forked from luqmana/llvm

Preserve nonnull metadata on Loads through SROA & mem2reg.

Summary:
https://llvm.org/bugs/show_bug.cgi?id=31142 :

SROA was dropping the nonnull metadata on loads from allocas that got optimized out. This patch simply preserves nonnull metadata on loads through SROA and mem2reg.

Reviewers: chandlerc, efriedma

Reviewed By: efriedma

Subscribers: hfinkel, spatel, efriedma, arielb1, davide, llvm-commits

Differential Revision: https://reviews.llvm.org/D27114

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@298540 91177308-0d34-0410-b5e6-96231b3b80d8

Differential Revision: http://reviews.llvm.org/D23619

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@279523 91177308-0d34-0410-b5e6-96231b3b80d8

Fix computeKnownBits for ARMISD::CMOV

Summary:
The true and false operands for the CMOV are operands 0 and 1.
ARMISelLowering.cpp::computeKnownBits was looking at operands 1 and 2
instead.  This can cause CMOV instructions to be incorrectly folded into
BFI if value set by the CMOV is another CMOV, whose known bits are
computed incorrectly.

This patch fixes the issue and adds a test case.

Reviewers: kristof.beyls, jmolloy

Subscribers: llvm-commits, aemerson, srhines, rengolin

Differential Revision: https://reviews.llvm.org/D31265

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@298624 91177308-0d34-0410-b5e6-96231b3b80d8

Fix inliner funclet unwind memoization

Summary:
The inliner may need to determine where a given funclet unwinds to,
and this determination may depend on other funclets throughout the
funclet tree.  The code that performs this walk in getUnwindDestToken
memoizes results to avoid redundant computations.  In the case that
a funclet's unwind destination is derived from its ancestor, there's
code to walk back down the tree from the ancestor updating the memo
map of its descendants to record the unwind destination.  This change
fixes that code to account for the case that some descendant has a
different unwind destination, which can happen if that unwind dest
is a descendant of the EHPad being queried and thus didn't determine
its unwind destination.

Also update test inline-funclets.ll, which is supposed to cover such
scenarios, to include a case that fails an assertion without this fix
but passes with it.

Fixes PR29151.

Reviewers: majnemer

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D24117

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@280610 91177308-0d34-0410-b5e6-96231b3b80d8

Archives require a symbol table on Solaris, even if empty.

On Solaris ld (and some other tools that use the underlying utility
libraries, such as elfdump) chokes on an archive library that has no
symbol table. The Solaris tools always create one, even if it's empty.

That bug has been fixed in the latest development line, and can
probably be backported to a supported release, but it would be nice if
LLVM's archiver could emit the empty symbol table, too.

Fixed in upstream LLVM in revision 297773.

[MSP430] Add SRet support to MSP430 target

This patch adds support for struct return values to the MSP430
target backend. It also reverses the order of argument and return
registers in the calling convention to bring it into closer
alignment with the published EABI from TI.

Patch by Andrew Wygle (awygle).

Differential Revision: https://reviews.llvm.org/D29069

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@296807 91177308-0d34-0410-b5e6-96231b3b80d8

x86 interrupt calling convention: only save xmm registers if the target supports SSE

…et supports SSE

The existing code always saves the xmm registers for 64-bit targets even if the
target doesn't support SSE (which is common for kernels). Thus, the compiler
inserts movaps instructions which lead to CPU exceptions when an interrupt
handler is invoked.

This commit fixes this bug by returning a register set without xmm registers
from getCalleeSavedRegs and getCallPreservedMask for such targets.

Patch by Philipp Oppermann.

Differential Revision: https://reviews.llvm.org/D29959


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295347 91177308-0d34-0410-b5e6-96231b3b80d8

use -O3 with some optimizations disabled on sparc w/ gcc

A previous commit changed it so that rust's private copy of llvm was
always built with -O0 on sparc, as initially that was the only option
that produced a working compiler.  After further investigation, it was
discovered that applying -fno-delayed-branch and -fno-omit-frame-pointer
was sufficient to avoid all of the code generation issues initially
encountered while retaining -O3.  Investigation with gcc upstream
continues.

use -O0 when building release LLVM on sparc

The bits produced by gcc above -O0 when building rust's private copy of llvm on
sparc result in either versions of llvm-tblgen that consistently segfault or
produce otherwise unuseable code.  This does not affect cross-compiles for some
reason.  Investigation with gcc upstream is in progress.

Add relocations for fixups fixup_arm_thumb_{br,bcc}

These need to be mapped through to R_ARM_THM_JUMP{11,8} respectively.

Fixes PR30279.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@280651 91177308-0d34-0410-b5e6-96231b3b80d8

Backport patches for MSP430 codegen.

[CodeGen] Fix result type for SMULO/UMULO legalization

On some platforms (like MSP430) the second element of the result
structure for SMULO/UMULO may have a shorter type than the one
returned by SetCC. We need to truncate it to the right type, or
else some incorrect code may be generated later on.

This fixes issue rust-lang/rust#37829

Patch by Vadzim Dambrouski!

Differential Revision: https://reviews.llvm.org/D27154

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@288857 91177308-0d34-0410-b5e6-96231b3b80d8

Fix PR27500: on MSP430 the branch destination offset is measured in words, not bytes.

Summary: In addition, the branch instructions will have proper BB destinations, not offsets, like before.

Reviewers: asl

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D23718

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@286252 91177308-0d34-0410-b5e6-96231b3b80d8

# Conflicts:
#	lib/Target/MSP430/MSP430BranchSelector.cpp

Add support for DW_AT_main_subprogram

I'd like the Rust compiler to emit DW_AT_main_subprogram in its DWARF
output, so that debuggers can find the user-provided main entry point.
See rust-lang/rust#32620

This patch adds support for DW_AT_main_subprogram to LLVM to support
this effort.

A variant of this patch was accepted by upstream LLVM; there,
DebugInfoFlags.def already has a flag with value 1<<20, which is why
this patch uses 1<<21 -- this will avoid the need to change the Rust
compiler when a new LLVM is imported.

Backport rL277331

UMAAL is a DSP instruction and it is not available on thumbv7m
(Cortex-M3) and thumbv6m (Cortex-M0+1) targets. Also fix wrong
CHECK prefix in longMAC.ll test.

Patch by Vadzim Dambrouski.

Differential Revision: https://reviews.llvm.org/D25890

# Conflicts:
#	lib/Target/ARM/ARMISelLowering.cpp
#	test/CodeGen/ARM/longMAC.ll

Fuchsia is a new operating system.

Change-Id: Id2cd63b5a98c71be555d64e4c4228b0d273610af

When combining an integer load with !range metadata that does not include 0 to a pointer load, make sure emit !nonnull metadata on the newly-created pointer load. This prevents the !nonnull metadata from being dropped during a ptrtoint/inttoptr pair.

This fixes PR30597.

Patch by Ariel Ben-Yehuda!

Differential Revision: https://reviews.llvm.org/D25215

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@283836 91177308-0d34-0410-b5e6-96231b3b80d8

This addresses PR26055 LiveDebugValues is very slow.

Contrary to the old LiveDebugVariables pass LiveDebugValues currently
doesn't look at the lexical scopes before inserting a DBG_VALUE
intrinsic. This means that we often propagate DBG_VALUEs much further
down than necessary. This is especially noticeable in large C++
functions with many inlined method calls that all use the same
"this"-pointer.

For example, in the following code it makes no sense to propagate the
inlined variable a from the first inlined call to f() into any of the
subsequent basic blocks, because the variable will always be out of
scope:

void sink(int a);
void __attribute((always_inline)) f(int a) { sink(a); }
void foo(int i) {
   f(i);
   if (i)
     f(i);
   f(i);
}

This patch reuses the LexicalScopes infrastructure we have for
LiveDebugVariables to take this into account.

The effect on compile time and memory consumption is quite noticeable:
I tested a benchmark that is a large C++ source with an enormous
amount of inlined "this"-pointers that would previously eat >24GiB
(most of them for DBG_VALUE intrinsics) and whose compile time was
dominated by LiveDebugValues. With this patch applied the memory
consumption is 1GiB and 1.7% of the time is spent in LiveDebugValues.

https://reviews.llvm.org/D24994
Thanks to Daniel Berlin and Keith Walker for reviewing!

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@282611 91177308-0d34-0410-b5e6-96231b3b80d8

Variables are sometimes missing their debug location information in
blocks in which the variables should be available. This would occur
when one or more predecessor blocks had not yet been visited by the
routine which propagated the information from predecessor blocks.

This is addressed by only considering predecessor blocks which have
already been visited.

The solution to this problem was suggested by Daniel Berlin on the
LLVM developer mailing list.

Differential Revision: https://reviews.llvm.org/D24927


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@282506 91177308-0d34-0410-b5e6-96231b3b80d8

…ults

GetCaseResults assumed that a terminator with one successor was an
unconditional branch.  This is not necessarily the case, it could be a
cleanupret.

Strengthen the check by querying whether or not the terminator is
exceptional.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@283517 91177308-0d34-0410-b5e6-96231b3b80d8

This is a minimal import of the emscripten "fastcomp" LLVM patchset.
All it contains is the target definitions necessary to create a
TargetMachine with the correct data layout. With this rustc can
emit LLVM IR that emcc will run through the PNaCl lagalizer and
the JS backend to generate asm.js.