The function 'getTargetShuffleMask' already knows how to deal with PSHUFB nodes
where the mask node is a load from constant pool, and the constant pool node
is wrapped by a X86ISD::Wrapper node. This patch extends that logic by teaching
it how to also look through X86ISD::WrapperRIP.
This helps function combineX86ShufflesRecusively to combine more shuffle
sequences containing PSHUFB nodes if we are in RIPRel PIC mode.
Before this change, llc (with -relocation-model=pic -march=x86-64) was unable
to decode a pshufb where the mask was loaded from a constant pool. For example,
the no-op shuffle from test 'x86-fold-pshufb.ll' was not folded into its
operand, so instead of generating a single 'movaps' the backend always
generated a sub-optimal 'movdqa + pshufb' sequence.
Added test x86-fold-pshufb.ll.
llvm-svn: 236863
GCC allows case-insensitive values for -mcpu, -march and -mtune options.
This patch implements the same behaviour for the -mcpu option.
Patch by Gabor Ballabas.
llvm-svn: 236859
I'm not yet sure what's wrong with pexpect and I cannot reproduce this
problem locally. XFAIL these tests for now to reduce buildbot noise so
I can track down the remaining buildbot failures.
llvm.org/pr22784
llvm-svn: 236857
Propagate the 'incomplete-format' state back through clang-format's command
line interace and adapt the emacs integration to show a better result.
llvm-svn: 236854
It's reported to pass consistently on Linux now, and it passed for me
in local testing on FreeBSD. Remove the expectedFailureFreeBSD decorator
for now.
llvm.org/pr20274
llvm-svn: 236853
In particular:
* If the difference between the longest and shortest element, we copped
out of column format completely. Now, we instead allow to arrange
these in a single column, essentially enforcing a one-per-line format.
* Allow column layout even if there are braced lists. Especially, if
there are many short lists, this can be beneficial. The bad case,
where there is a long nested init list is usually caught as we now
limit the length difference of the longest and shortest element.
llvm-svn: 236851
1) check whether the alignment of the memory is sufficient for the
*merged* store or load to be efficient.
Not doing so can result in some ridiculously poor code generation, if
merging creates a vector operation which must be aligned but isn't.
2) DON'T check that the alignment of each load/store is equal. If
you're merging 2 4-byte stores, the first *might* have 8-byte
alignment, but the second certainly will have 4-byte alignment. We do
want to allow those to be merged.
llvm-svn: 236850
Mark longjmp tests as XFAIL because longjmp assembly for mips is not yet implemented.
Reviewers: dsanders, dvyukov, samsonov
Subscribers: llvm-commits, mohit.bhakkad, jaydeep
Differential Revision: http://reviews.llvm.org/D9526
llvm-svn: 236847
llvm::Triple::getARMCPUForArch now returns nullptr for invalid -march
values, instead of silently translating it to arm7tdmi. Use this to
give an error message, which is consistent with how gcc behaves.
Differential Revision: http://reviews.llvm.org/D9602
llvm-svn: 236846
Restructure Triple::getARMCPUForArch so that invalid values will
return nullptr, while retaining the behaviour that an argument
specifying no particular architecture version will give a default
CPU. This will be used by clang to give an error on invalid -march
values.
Also restructure the extraction of the architecture version from
the MArch string a little to hopefully make what it's doing clearer.
Differential Revision: http://reviews.llvm.org/D9599
llvm-svn: 236845
This test was failing on mips because mips addresses are 40-bit long.
Using regex for address solves this issue.
Reviewers: dsanders, kcc, samsonov
Subscribers: llvm-commits, mohit.bhakkad, jaydeep
Differential Revision: http://reviews.llvm.org/D9516
llvm-svn: 236844
If no temp directory specified by the user on android then fall back
to /data/local/tmp what is always present on the device. It removes
the dependency of specifying TMPDIR for executing platform commands
on android.
Differential revision: http://reviews.llvm.org/D9569
llvm-svn: 236843
This patch includes the following changes:
# Add CMIUtilString::ConvertToASCII to convert unicode to ASCII (MI)
# Rework CMICmnLLDBUtilSBValue::GetSimpleValue to print wide/unicode char (MI)
## Add CMICmnLLDBUtilSBValue::GetSimpleValueChar
## Extend CMICmnLLDBUtilSBValue::IsCharType to accept char16_t/char32_t
## Don't ignore the fail_value in SBValue::GetValueAsUnsigned
# Rework CMIUtilString::ConvertToASCII (MI)
## Don't use templates
## Change the function signature to convert signle characters (was std::basic_string)
## Rename CMIUtilString::ConvertToASCII to CMIUtilString::ConvertToPrintableASCII
# Add CMIUtilString::ConvertToPrintableASCII for char (MI)
## Refactor CMIUtilString::Escape
## Simplify CMICmnLLDBUtilSBValue::GetSimpleValueChar for char
# Add char16_t* and char32_t* support in CMICmnLLDBUtilSBValue::GetSimpleValue (MI)
## Add CMICmnLLDBUtilSBValue::GetSimpleValueCStringPointer
## Add CMICmnLLDBUtilSBValue::ReadCStringFromHostMemory2
(it's extended version of CMICmnLLDBUtilSBValue::ReadCStringFromHostMemory)
# Improve error checking in CMICmnLLDBUtilSBValue::GetSimpleValueChar (MI)
# Refactor CMICmnLLDBUtilSBValue (MI)
## Remove CMICmnLLDBUtilSBValue::ReadCStringFromHostMemory
(use CMICmnLLDBUtilSBValue::ReadCStringFromHostMemory2 instead)
## Rename CMICmnLLDBUtilSBValue::ReadCStringFromHostMemory2 to
CMICmnLLDBUtilSBValue::ReadCStringFromHostMemory
## Move CMICmnLLDBUtilSBValue::GetValueCString to private methods
## Rename CMICmnLLDBUtilSBValue::GetValueCString to
CMICmnLLDBUtilSBValue::GetSimpleValueCStringArray
## Remove CMICmnLLDBUtilSBValue::GetChildValueCString
# Improve MiGdbSetShowTestCase.test_lldbmi_gdb_set_show_print_char_array_as_string test to check char16_t/char32_t (MI)
# Fix CMICmnLLDBUtilSBValue::GetSimpleValueCStringArray for Unicode (MI)
## Fix handling of char16_t and char32_t
## Improve CMICmnLLDBUtilSBValue::ReadCStringFromHostMemory to read
variables of type char[]: add vnMaxLen argument
## Turn on few cases to check char16_t[] and char32_t[] output in
MiGdbSetShowTestCase.test_lldbmi_gdb_set_show_print_char_array_as_string
# Remove unnecessary checks in CMICmnLLDBUtilSBValue (MI)
llvm-svn: 236827
This patch incldues the following:
# Add from/to arguments in -var-list-children command (MI)
## Handle from and to args
## Don't print children=[] if numchild is 0
## A bit refactoring
## Add MiVarTestCase.test_lldbmi_var_list_children test
# Refactor -var-list-children (MI)
## Remove CMICmdCmdVarListChildren::VecMIValueResult_t
## Add CMICmdCmdVarListChildren::m_miValueList
## Remove CMICmdCmdVarListChildren::m_vecMiValueResult
(use CMICmdCmdVarListChildren::m_miValueList instead)
## Print error message if value is invalid:
```
^error,msg="variable invalid"
```
## Refactor CMICmdCmdVarListChildren::Acknowledge and remove duplicated code
## Don't print children=[] if numchild is 0
## Add error checking
## Use CMICmnLLDBDebugSessionInfo::VariableInfoFormat_e for print-values
## Add few more empty-range test cases in MiVarTestCase.test_lldbmi_var_list_children
#Improve MiVarTestCase.test_lldbmi_var_list_children test (MI)
llvm-svn: 236820
Summary:
This changes lldb_assert to accept bool expressions as the parameter, this is because some
objects (such as std::shared_ptr) are convertible to bool, but are not convertible to int, which
leads to surprising errors.
Reviewers: granata.enrico, zturner
Subscribers: lldb-commits
Differential Revision: http://reviews.llvm.org/D9565
llvm-svn: 236819
The ELF data contains two different errors in some ELF files on android.
* The link field of the symbol table don't point to the plt section or
to the dynsym section even when it is present in the ELF files.
* The size of the plt entries aren't specified in the section header of
the plt section.
This CL adds some workarounds for these two issue with finding the
sections by name if the link field is empty and by using a heuristic to
calculate the size and offset of the plt entries.
Differential revision: http://reviews.llvm.org/D9560
llvm-svn: 236818
Summary:
This set of changes addresses the issue I described in [[ http://llvm.org/pr23403 | Bug 23403 ]], in short most LLDB-MI tests fail unexpectedly due to timeout on my Ubuntu 14.10 x86_64 VirtualBox VM. I have no idea why the Ubuntu buildbots don't seem to suffer from this issue. This patch just makes **MiTestCaseBase.spawnLldbMi()** block until the spawned LLDB-MI child process outputs the **(gdb)** prompt, which ensures that MI commands aren't sent to the LLDB-MI child process until it's ready.
Before these changes 52 LLDB-MI tests failed unexpectedly (all but one due to timeouts), after these changes no tests fail due to timeouts. I still get one unexpected failure in **test_lldbmi_source_option_start_script_error** in **TestMiStartupOptions.py**, but that's due to a file error, that test should be skipped on Linux just like the other two //start_script// tests until the file error is fixed.
Patch from Vadim Macagon. Thanks!
Test Plan: ./dotest.py -A x86_64 -C clang --executable $BUILDDIR/bin/lldb tools/lldb-mi/
Reviewers: abidh, domipheus, ki.stfu
Reviewed By: ki.stfu
Subscribers: brucem, lldb-commits
Differential Revision: http://reviews.llvm.org/D9595
llvm-svn: 236816
Summary:
In microMIPS, labels need to know whether they are on code or data. This is
indicated with STO_MIPS_MICROMIPS and can be inferred by being followed
by instructions. For empty basic blocks, we can ensure this by emitting the
.insn directive after the label.
Also, this fixes some failures in our out-of-tree microMIPS buildbots, for the
exception handling regression tests under: SingleSource/Regression/C++/EH
Reviewers: dsanders
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D9530
llvm-svn: 236815
Summary:
The stop callback is a remnant of the ThreadStateCoordinator. We don't need it now that TSC is
gone, as we know exactly which function to call when threads stop. This also removes some
stop-related functions, which were just forwarding calls to one another.
Test Plan: ninja check-lldb continues to pass
Reviewers: chaoren, ovyalov
Subscribers: lldb-commits
Differential Revision: http://reviews.llvm.org/D9531
llvm-svn: 236814
The misc-static-assert check will not warn on the followings:
assert(NULL == "shouldn't warn");
assert(__null == "shouldn't warn");
Where NULL is a macro defined as __null.
llvm-svn: 236812
Now caller of ELFState::writeSectionContent() methods is responsible to check
a section type and selects an appropriate writeSectionContent method.
So unexpected section type inside writeSectionContent method indicates
a wrong usage of the method and should be guarded by assert.
llvm-svn: 236808
Andrea Di Biagio