Commit Graph

165459 Commits

Author SHA1 Message Date
Jakob Stoklund Olesen 497332c05f SPARCv9 implements long double as an IEEE quad.
llvm-svn: 199399
2014-01-16 16:43:19 +00:00
Alp Toker 5a1a81841a www: link to the planet clang newsfeed
Also shuffle the Communication section so the bug tracker comes first.

(The sidebar isn't scrollable at the moment so this gives a better chance of
the bug tracker being seen. The links further down are basically invisible --
we should look into that.)

llvm-svn: 199398
2014-01-16 16:29:02 +00:00
Rafael Espindola 0b694814a8 Add an emitRawComment function and use it to simplify some uses of EmitRawText.
llvm-svn: 199397
2014-01-16 16:28:37 +00:00
Daniel Sanders 4d20f0c00f [mips][sched] Split IIseb into II_SEB and II_SEH
No functional change since there are no InstrItinData's.

llvm-svn: 199396
2014-01-16 16:19:38 +00:00
Daniel Sanders 306ef07b8d [mips][sched] Split IILogic into II_AND, II_OR, II_XOR, II_ANDI, II_ORI, II_XORI
This is necessary because the classes are shared between all implementations.

No functional change since the InstrItinData's have been duplicated.

llvm-svn: 199394
2014-01-16 15:57:05 +00:00
Amara Emerson dba59eb3f4 Move the xscale build attribute test to the proper place and remove the old one.
The encoding of build attributes is already tested in CodeGen/ARM/build-attributes-encoding.s

llvm-svn: 199393
2014-01-16 15:11:54 +00:00
Aaron Ballman 4755304fd1 Fixing a warning that causes the sanitizer build disliked about mixing && and ||. Since this is generated code, the && has been removed from the expression entirely.
llvm-svn: 199392
2014-01-16 14:32:03 +00:00
Daniel Sanders 980589a803 [mips][sched] Split IIArith in preparation for the first scheduler targeting a specific MIPS CPU.
IIArith -> II_ADD, II_ADDU, II_AND, II_CL[ZO], II_DADDIU, II_DADDU,
  II_DROTR, II_DROTR32, II_DROTRV, II_DSLL, II_DSLL32, II_DSLLV,
  II_DSR[AL], II_DSR[AL]32, II_DSR[AL]V, II_DSUBU, II_LUI, II_MOV[ZFNT],
  II_NOR, II_OR, II_RDHWR, II_ROTR, II_ROTRV, II_SLL, II_SLLV, II_SR[AL],
  II_SR[AL]V, II_SUBU, II_XOR

No functional change since the InstrItinData's have been duplicated.

This is necessary because the classes are shared between all schedulers.

Once this patch series is committed there will be an InstrItinClass for
each mnemonic with minimal grouping. This does increase the size of the
itinerary tables for each MIPS scheduler but we have a few options for dealing
with that later. These options include reducing the number of classes once
we see the best way to simplify them, or by extending tablegen to be able
to compress the table by eliminating duplicates entries, etc.

llvm-svn: 199391
2014-01-16 14:27:20 +00:00
Evgeniy Stepanov 877bb8ba59 [asan] Fix a build warning on windows.
llvm-svn: 199390
2014-01-16 14:21:17 +00:00
Daniel Sanders bfe1830ab9 [mips] Correct itin class for MULT_MM and MULTu_MM to IIImult.
This matches the itin class used by the non-microMIPS equivalents of these
instructions.

llvm-svn: 199389
2014-01-16 14:02:48 +00:00
Alexander Potapenko f47e667c05 [ASan] Enough to have llvm-symbolizer in $PATH.
llvm-svn: 199388
2014-01-16 14:01:39 +00:00
Aaron Ballman 12b9f6535b Factored some function-like type reasoning out of SemaDeclAttr and onto Decl itself. This allows for more declarative subjects in attribute tablegen where the attribute appertains to something function-like, but not strictly a FunctionDecl.
llvm-svn: 199387
2014-01-16 13:55:42 +00:00
Joerg Sonnenberger da93c0f37c Add missing __aeabi_fcmpun alias to match C version.
llvm-svn: 199386
2014-01-16 13:55:24 +00:00
Alexander Potapenko 108bc0311f [ASan] Typo fix (symbolize=0 disables online symbolization)
llvm-svn: 199385
2014-01-16 13:48:16 +00:00
Alexey Samsonov 4d075df406 [Sanitizer] Add basic support for using libbacktrace in symbolizer.
This change allows to compile sanitizer sources so that *san runtime
will attempt to use libbacktrace and/or libiberty for symbolization
(instead of communicating with llvm-symbolizer).

I've tested this patch by manually defining SANITIZER_LIBBACKTRACE and/or
SANITIZER_CP_DEMANGLE, linking with necessary libraries and verifying that
all tests from ASan test suite work.

Based on patches by Jakub Jelinek!

llvm-svn: 199384
2014-01-16 13:48:15 +00:00
Alexander Potapenko c7549007c9 [ASan] Describe online/offline symbolization of reports, mention dsymutil.
llvm-svn: 199383
2014-01-16 13:46:29 +00:00
Aaron Ballman 95755d6d05 Giving the asm attribute some keyword spellings based off of GCC's documentation. No functional changes.
llvm-svn: 199382
2014-01-16 13:45:57 +00:00
Daniel Sanders 818058b2ab [mips] IIImult should have an InstrItinData in the generic scheduler. Used the same one as for IIImul.
Affects:
  DMULT, DMULTu, MADD, MADD_MM, MADDU, MADDU_MM, MSUB, MSUB_MM, MSUBU,
  MSUBU_MM, MULT, MULTu

Does not affect MULT_MM, MULTu_MM since they are currently miscategorised
as IIImul.

llvm-svn: 199381
2014-01-16 13:45:53 +00:00
Evgeniy Stepanov 364bfdf4c9 [asan] Fix a bunch of style issues.
llvm-svn: 199380
2014-01-16 13:17:13 +00:00
Evgeniy Stepanov fe906a1df6 [asan] Restore lost -pie in Android tests.
llvm-svn: 199379
2014-01-16 13:11:37 +00:00
Aaron Ballman 36a5350e51 Distinguish between attributes explicitly written at the request of the user, and attributes implicitly generated to assist in bookkeeping by the compiler. This is done so by table generating a CreateImplicit method for each attribute.
Additionally, remove the optional nature of the spelling list index when creating attributes. This is supported by table generating a Spelling enumeration when the spellings for an attribute are distinct enough to warrant it.

llvm-svn: 199378
2014-01-16 13:03:14 +00:00
Evgeniy Stepanov c61623b170 [asan] Implement delayed activation of AddressSanitizer
This change adds ASAN_OPTIONS=start_deactivated=1 flag. When present, ASan will
start in "deactivated" mode, with no heap poisoning, no quarantine, no stack
trace gathering, and minimal redzones. All this features come back when
__asan_init is called for the constructor of an instrumented library.

The primary use case for this feature is Android. Code itself is not
Android-specific, and this patch includes a Linux test for it.

llvm-svn: 199377
2014-01-16 12:31:50 +00:00
Tim Northover 3657cb0350 ReMat: fix overly cavalier attitude to sub-register indices
There are two attempted optimisations in reMaterializeTrivialDef, trying to
avoid promoting the size of a register too much when rematerializing.
Unfortunately, both appear to be flawed. First, we see if the original register
would have worked, but this is inadequate. Consider:

    v1 = SOMETHING (v1 is QQ)
    v2:Q0 = COPY v1:Q1 (v1, v2 are QQ)
    ...
    uses of v2

In this case even though v2 *could* be used directly as the output of
SOMETHING, this would set the wrong bits of the QQ register involved. The
correct rematerialization must be:

    v2:Q0_Q1 = SOMETHING (v2 promoted to QQQ)
    ...
    uses of v2:Q1_Q2

For the second optimisation, if the correct remat is "v2:idx = SOMETHING" then
we can't necessarily expect v2 itself to be valid for SOMETHING, but we do try
to hunt for a class between v1 and v2 that works. Unfortunately, this is also
wrong:

    v1 = SOMETHING (v1 is QQ)
    v2:Q0_Q1 = COPY v1 (v1 is QQ, v2 is QQQ)
    ...
    uses of v2 as a QQQ

The canonical rematerialization here is "v2:Q0_Q1 = SOMETHING". However current
logic would decide that v2 could be a QQ (no interest is taken in later uses).

This patch, therefore, always accepts the widened register class without trying
to be clever. Generally there is no penalty to this (e.g. in the common GR32 <
GR64 case, expanding the width doesn't matter because it's not like you were
going to do anything else with the high bits of a GR32 register). It can
increase register pressure in cases like the ARM VFP regs though (multiple
non-overlapping but equivalent subregisters). This situation can be
spotted by the fact that both source and destination in the
not-quite-coalesced pair have a sub-register index and
rematerialisation is skipped in that situation.

Unfortunately, no in-tree targets actually expose this as far as I can tell
(there are so few isAsCheapAsAMove instructions for it to trigger on) so I've
been unable to produce a test. It was exposed in our ARM64 SPEC tests though,
and I will be adding a test there that we should be able to contribute
soon(TM).

rdar://problem/15775279

llvm-svn: 199376
2014-01-16 12:29:55 +00:00
David Majnemer f5b93794b7 Sema: Fix crash during member pointer conversion involving incomplete classes
We would attempt to determine the inheritance relationship between
classes 'A' and 'B' during static_cast if we tried to convert from 'int
A::*' to 'int B::*'.  However, the question "does A derive from B" is
not meaningful when 'A' isn't defined.

Handle this case by requiring that 'A' be defined.

This fixes PR18506.

llvm-svn: 199374
2014-01-16 12:02:55 +00:00
Evgeniy Stepanov 780d85a8ef [asan] Remove leftover use of -fsanitize-address-zero-base-shadow.
llvm-svn: 199373
2014-01-16 11:42:40 +00:00
Evgeniy Stepanov d04b861e69 [asan] Remove -fsanitize-address-zero-base-shadow command line
flag from clang, and disable zero-base shadow support on all platforms
where it is not the default behavior.

- It is completely unused, as far as we know.
- It is ABI-incompatible with non-zero-base shadow, which means all
objects in a process must be built with the same setting. Failing to
do so results in a segmentation fault at runtime.
- It introduces a backward dependency of compiler-rt on user code,
which is uncommon and complicates testing.

This is the Clang part of a larger change.

llvm-svn: 199372
2014-01-16 10:19:31 +00:00
Evgeniy Stepanov 13665367a0 [asan] Remove -fsanitize-address-zero-base-shadow command line
flag from clang, and disable zero-base shadow support on all platforms
where it is not the default behavior.

- It is completely unused, as far as we know.
- It is ABI-incompatible with non-zero-base shadow, which means all
objects in a process must be built with the same setting. Failing to
do so results in a segmentation fault at runtime.
- It introduces a backward dependency of compiler-rt on user code,
which is uncommon and complicates testing.

This is the LLVM part of a larger change.

llvm-svn: 199371
2014-01-16 10:19:12 +00:00
Evgeniy Stepanov e259b658e4 [asan] Nuke ASAN_FLEXIBLE_MAPPING_AND_OFFSET.
llvm-svn: 199370
2014-01-16 10:16:19 +00:00
Jiangning Liu 4df2363a23 For ARM, fix assertuib failures for some ld/st 3/4 instruction with wirteback.
llvm-svn: 199369
2014-01-16 09:16:13 +00:00
Daniel Jasper cb51cf409b clang-format: Enable formatting of lambdas with explicit return type.
So clang-format can now format:

  int c = []()->int { return 2; }();
  int c = []()->vector<int> { return { 2 }; }();

llvm-svn: 199368
2014-01-16 09:11:55 +00:00
Tim Northover 157d911b42 MachO: use *-*-*-macho for MachO embedded targets.
Previously we had bodged together some hacks mapping MachO embedded
targets (i.e. mainly ARM v6M and v7M) to the "*-*-darwin-eabi" triple.
This is incorrect in both details (they don't run Darwin and they're
not EABI in any real sense).

This commit appropriates the existing "MachO" environment for the
purpose instead.

llvm-svn: 199367
2014-01-16 08:48:16 +00:00
Elena Demikhovsky d1487261a0 AVX-512: fixed a compare pattern
llvm-svn: 199366
2014-01-16 08:45:54 +00:00
Craig Topper a9d2c67cc2 Copy segment register when optimizing to MOV8ao8/MOV16ao16/MOV32ao32.
llvm-svn: 199365
2014-01-16 07:57:45 +00:00
Craig Topper 35da3d190a Allow x86 mov instructions to/from memory with absolute address to be encoded and disassembled with a segment override prefix. Fixes PR16962.
llvm-svn: 199364
2014-01-16 07:36:58 +00:00
Rafael Espindola 74c3e63193 Use a slightly smaller hack.
llvm-svn: 199363
2014-01-16 07:36:00 +00:00
Quentin Colombet 73a3b2209e Revert r199361: Now, the sanitizer got the change
llvm-svn: 199362
2014-01-16 07:29:07 +00:00
Quentin Colombet a823e8c266 [LTO] Modify lto.exports to force the sanitizer to rebuilt LTO.exports
llvm-svn: 199361
2014-01-16 07:14:01 +00:00
Sylvestre Ledru 660712570a llvm/IR/Writer.h has been removed in llvm commit r198836 and seems useless in polly
llvm-svn: 199360
2014-01-16 07:10:09 +00:00
Bill Wendling 2f98bac1f4 Fix typo: : not ;
llvm-svn: 199359
2014-01-16 07:08:22 +00:00
Rui Ueyama ad882ba896 llmv-objdump/COFF: Print export table contents.
This patch adds the capability to dump export table contents. An example
output is this:

  Export Table:
   Ordinal      RVA  Name
         5   0x2008  exportfn1
         6   0x2010  exportfn2

By adding this feature to llvm-objdump, we will be able to use it to check
export table contents in LLD's tests. Currently we are doing binary
comparison in the tests, which is fragile and not readable to humans.

llvm-svn: 199358
2014-01-16 07:05:49 +00:00
Rafael Espindola f69b850d60 CommentColumn is always 40. Simplify.
llvm-svn: 199357
2014-01-16 07:04:11 +00:00
Quentin Colombet cb2ae8d8f7 [cmake] Attempt to fix sanitizer buildbot.
The generation of the native_export_file end up in
several different makefiles. All those makefiles
write the same file, but can be executed concurrently...
and bad things happen!

llvm-svn: 199356
2014-01-16 06:43:55 +00:00
Rafael Espindola 306e2b019f Convert test to FileCheck.
llvm-svn: 199355
2014-01-16 06:31:20 +00:00
Bill Wendling 91686d6dee Reapply r194218 with fix:
Move copying of global initializers below the cloning of functions.

The BlockAddress doesn't have access to the correct basic blocks until the
functions have been cloned. This causes the BlockAddress to point to the old
values. Just wait until the functions have been cloned before copying the
initializers.
PR13163

llvm-svn: 199354
2014-01-16 06:29:36 +00:00
Craig Topper 8a60fff260 Remove use of OpSize for populating VEX_PP field. A prefix encoding is now used instead. Simplify some other code. No functional changes intended.
llvm-svn: 199353
2014-01-16 06:14:45 +00:00
Rafael Espindola 098000eb72 Attempt to fix the MSVC build.
llvm-svn: 199352
2014-01-16 05:09:32 +00:00
Arnold Schwaighofer e3ac099726 BasicAA: We need to check both access sizes when comparing a gep and an
underlying object of unknown size.

Fixes PR18460.

llvm-svn: 199351
2014-01-16 04:53:18 +00:00
Rafael Espindola c3d687667d Prevent calls to __jit_debug_register_code from being optimized out.
Patch by Andrew MacPherson. I just tweaked the comment.

llvm-svn: 199350
2014-01-16 04:50:58 +00:00
Rui Ueyama a045b73a96 Don't use DataRefImpl to implement ImportDirectoryEntryRef.
DataRefImpl (a union of two integers and a pointer) is not the ideal data type
to represent a reference to an import directory entity. We should just use the
pointer to the import table and an offset instead to simplify. No functionality
change.

llvm-svn: 199349
2014-01-16 03:13:19 +00:00
Alp Toker 11a71128b7 Forbid driver use in Sema tests
This ports the last Sema tests over to use the frontend directly, and adds a
local lit substitution to disable inappropriate %clang usage under this
directory.

llvm-svn: 199348
2014-01-16 02:37:08 +00:00