Commit Graph

3418 Commits

Author SHA1 Message Date
Peter Smith 1ec42d9a76 Finalize content of synthetic sections prior to Thunk Creation
This change moves the calls to finalizeContent() for each synthetic section
before createThunks(). This will allow us to assign addresses prior to
calling createThunks(). As addition of thunks may add to the static
symbol table and may affect the size of the mips got section we introduce a
couple of additional member functions to update these values.
    
Differential revision: https://reviews.llvm.org/D29983

llvm-svn: 297277
2017-03-08 14:06:24 +00:00
Rafael Espindola 8bb4087f88 Use 32 bits for alignment.
This is consistent with what we do for input sections.

llvm-svn: 297152
2017-03-07 15:51:09 +00:00
Rafael Espindola c8de15bdda Rename remaining uses of Addralign to Alignment.
llvm-svn: 297146
2017-03-07 15:11:21 +00:00
Rafael Espindola 3770763cda Rename Addralign to Alignment.
It now matches the name used in InputSectionBase.

llvm-svn: 297144
2017-03-07 14:55:52 +00:00
Peter Smith 24fb250cad Fix Mips LA25 Thunks where Caller and Callee in different OS
This change fixes a bug in which the Mips LA25 Thunks are always assigned
to the same Output section as the caller and not the callee as expected.

Differential Revision: https://reviews.llvm.org/D30637

llvm-svn: 297135
2017-03-07 09:45:04 +00:00
Rui Ueyama af6198d96d Remove Config->Rela and define Config->isRela() instead.
llvm-svn: 297108
2017-03-07 00:43:53 +00:00
Rui Ueyama df8eb17d21 Remove Config->Mips64EL and define Config->isMips64EL() instead.
llvm-svn: 297107
2017-03-07 00:43:33 +00:00
Rafael Espindola 2a80e1180e Make Discarded a InputSectionBase.
NFC, just a bit simpler.

llvm-svn: 297087
2017-03-06 22:36:19 +00:00
Rafael Espindola 5c02b741eb Detemplate EhInputSection. NFC.
llvm-svn: 297077
2017-03-06 21:17:18 +00:00
Evgeniy Stepanov 5ac5542544 Fix -Werror build error.
tools/lld/ELF/Symbols.cpp:215:13: error: unused variable 'S'
[-Werror,-Wunused-variable]
  if (auto *S = dyn_cast<SharedSymbol>(this)

llvm-svn: 297063
2017-03-06 20:27:38 +00:00
Rafael Espindola 6119b865ec Detemplate merge (input and synthetic) sections. NFC.
llvm-svn: 297061
2017-03-06 20:23:56 +00:00
Rafael Espindola 77408b067c Convert two uses of uintX_t. NFC.
llvm-svn: 297059
2017-03-06 20:04:47 +00:00
Rafael Espindola 692b2f88d3 Fully precise gc handling of __start and __stop symbols.
This puts us at parity with bfd, which could already gc this case.

I noticed the sections not being gced when linking a modified freebsd
kernel. A section that was not gced and not mentioned in the linker
script would end up breaking the expected layout. Since fixing the gc
is relatively simple and an improvement, that seems better than trying
to hack the orphan placement code.

There are 173 input section in the entire link whose names are valid C
identifiers, so this is probably not too performance critical.

llvm-svn: 297049
2017-03-06 18:48:18 +00:00
Rui Ueyama 967fc1cef4 Fix -Wc++11-narrowing.
llvm-svn: 297012
2017-03-06 15:16:18 +00:00
George Rimar c215a2ac40 [ELF] - Make Bss and BssRelRo sections to be synthetic (#2).
In compare with D30458, this makes Bss/BssRelRo to be pure
synthetic sections.

That removes CopyRelSection class completely, making
Bss/BssRelRo to be just regular synthetics.

SharedSymbols involved in creating copy relocations are
converted to DefinedRegular, what also simplifies things.

Differential revision: https://reviews.llvm.org/D30541

llvm-svn: 297008
2017-03-06 14:37:45 +00:00
Rafael Espindola 066d3d2141 Pass a callback to resolveReloc. NFC.
This is consistent with rest of the file and opens the way for a
relocation keeping multiple sections alive.

llvm-svn: 296788
2017-03-02 19:29:28 +00:00
Rui Ueyama 8139eab3c3 Simplify. NFC.
llvm-svn: 296773
2017-03-02 17:55:29 +00:00
Peter Collingbourne ab76a19afb LTO: When creating a local cache, create the cache directory if it does not already exist.
Differential Revision: https://reviews.llvm.org/D30519

llvm-svn: 296726
2017-03-02 02:02:38 +00:00
Rafael Espindola 4368bdb270 Make gc a bit more aggressive.
We were not gcing any section whose name was a C identifier. Both gold
and bfd only keep those if they are used.

To avoid having to create the __start/__stop symbols early or doing
string lookups in resolvedReloc, this patch just looks for undefined
symbols __start/__stop to decide if a section is needed or not.

llvm-svn: 296723
2017-03-02 01:50:34 +00:00
Peter Collingbourne e02775f068 ELF: Add ThinLTO caching support.
This patch adds an option named --thinlto-cache-dir, which specifies the
path to a directory in which to cache native object files for ThinLTO
incremental builds.

Differential Revision: https://reviews.llvm.org/D30509

llvm-svn: 296702
2017-03-01 23:00:10 +00:00
Rui Ueyama ac2d815a2a Split GdbIndexBuilder class into non-member functions.
That class had three member functions, and all of them are just reader
methods that did not depend on class members, so they can be just non-
member functions.

Probably we should reorganize the functions themselves because their
return types doesn't make much sense to me, but for now I just moved
these functions out of the class.

llvm-svn: 296700
2017-03-01 22:54:50 +00:00
Rui Ueyama aab18c0cfc Remove useless variables and declarations.
llvm-svn: 296695
2017-03-01 22:24:46 +00:00
Rui Ueyama 97d7a74b35 Replace `auto` with its real type.
llvm-svn: 296694
2017-03-01 22:24:25 +00:00
Rui Ueyama e5e320de06 Style fix.
llvm-svn: 296689
2017-03-01 22:03:17 +00:00
Rui Ueyama a4d9a22ba2 Reduce nesting. NFC.
llvm-svn: 296688
2017-03-01 22:02:57 +00:00
Rui Ueyama e02cc60c72 Do not inherit LoadedObjectInfo.
GdbIndexBuilder class inherited LoadedObjectInfo, but that's not necessary.

llvm-svn: 296687
2017-03-01 22:02:37 +00:00
Rui Ueyama 475b1ddfe2 Inline a function that is too short to be an independent function.
llvm-svn: 296686
2017-03-01 22:02:17 +00:00
Rui Ueyama 962a83fbe9 Unbreak Windows bots.
llvm-svn: 296680
2017-03-01 21:28:41 +00:00
Rui Ueyama d0e07b919c Don't implement the gdb hash table as a generic in-memory hash table.
Looks like .gdb.index and its support classes do things that they don't
have to or shouldn't do do. This patch addresses one of these issues.

GdbHashTab class is a hash table class. Just like other in-memory hash
tables, that incrementally updates its internal data and resizes buckets
as new elements are added so that key lookup is always fast.
But that is completely not necessary.

Unlike debuggers, we only produce hash tables for .gdb.index and
never read them. So incrementally updating a hash table in memory is
just a waste of resource and complicates the code. What we should
do is to accumulate symbols and then create the final hash table
at once.

llvm-svn: 296678
2017-03-01 21:08:21 +00:00
Rui Ueyama 7986b4544b Add a reference to a Ulrich Drepper's paper.
llvm-svn: 296650
2017-03-01 18:09:09 +00:00
Rafael Espindola bedccb5e78 Simplify. NFC.
llvm-svn: 296620
2017-03-01 14:21:31 +00:00
Rafael Espindola bd12e2a0ce Simplify. NFC.
llvm-svn: 296619
2017-03-01 14:12:21 +00:00
George Rimar efc31dd9bb [ELF] - Reset output section size when assigning offsets.
In many places we reset Size to 0 before calling assignOffsets()
manually. Sometimes we don't do that. 
It looks we can just always do that inside.

Previous code had:
template <class ELFT> void OutputSection::assignOffsets() {
  uint64_t Off = Size;

And tests feels fine with Off = 0. 
I think Off = Size make no sence.

Differential revision: https://reviews.llvm.org/D30463

llvm-svn: 296609
2017-03-01 11:10:53 +00:00
George Rimar c84ca256d3 [ELF] - Fixed crash in GnuHashTableSection.
r296570 intorduced crash revealed by 
gnu-hash-table.s which crashed for me.

Reason was use of uninitialized NBuckets variable.

llvm-svn: 296597
2017-03-01 10:12:49 +00:00
Rui Ueyama 6e50fd53d4 Instead of passing an empty ArrayRef as contents, pass actual contents. NFC.
llvm-svn: 296586
2017-03-01 07:39:06 +00:00
Sean Silva d4e606273b Improve comment.
Thanks to Rafael for helping to dig down into what we're really trying
to communicate here.

llvm-svn: 296580
2017-03-01 04:44:04 +00:00
Rui Ueyama 2787664db7 Set output section's st_entsize based on input section's st_entsize.
Each input section knows its sh_entsize value, so we can set output
section's sh_entsize based on input sections values.

llvm-svn: 296577
2017-03-01 04:04:23 +00:00
Rui Ueyama e13373b5b9 Partially rewrite .gnu.hash synthetic section and add comments.
This implementation is probably slightly inefficient than before,
but that should be negligible because this is not a performance-
critical pass.

llvm-svn: 296570
2017-03-01 02:51:42 +00:00
Martell Malone 6b43b7ad49 [ELF] - Allow the Code Model flag when using LTO
Differential Revision: https://reviews.llvm.org/D29445

llvm-svn: 296542
2017-02-28 23:43:26 +00:00
Rui Ueyama 13ed0b691e ELF ICF: Merge only functions.
Previously, LLD merged all read-only sections. So the following
program prints out "true" if -icf=all is specified.

  static const int foo = 1;
  static const int bar = 1;
  int main() { printf("%s\n", &foo == &bar ? "true" : "false"); }

This is somewhat counter-intuitive, and it actually caused nasty issues.
One example is https://bugs.chromium.org/p/chromium/issues/detail?id=682773#c24.

This patch changes the way how it works. Now ICF merges only functions
(i.e. executable sections).

Differential Revision: https://reviews.llvm.org/D30365

llvm-svn: 296534
2017-02-28 22:42:49 +00:00
Rui Ueyama e412063397 Add comments about .hash and .gnu.hash.
llvm-svn: 296529
2017-02-28 22:05:13 +00:00
Rui Ueyama 73d29ab3a7 Minor refactoring. NFC.
llvm-svn: 296511
2017-02-28 19:43:54 +00:00
Rui Ueyama 175e81cf3a Use make<> instead of new (BAlloc). NFC.
We converted them before, but there were a few remaining occurrences.

llvm-svn: 296510
2017-02-28 19:36:30 +00:00
Rui Ueyama 80474a26b9 De-template DefinedRegular.
Differential Revision: https://reviews.llvm.org/D30348

llvm-svn: 296508
2017-02-28 19:29:55 +00:00
Rui Ueyama 1b00318371 Make it obvious that st_value for common symbols has a different meaning.
llvm-svn: 296507
2017-02-28 19:22:09 +00:00
Rui Ueyama 4d574b324a Fix wrong TLS symbol values.
I do not fully understand why we had these values in the tests, but
the new value matches what ld.bfd and ld.gold set, so I guess that
was just a mistake.

Differential Revision: https://reviews.llvm.org/D30441

llvm-svn: 296505
2017-02-28 19:06:32 +00:00
Rafael Espindola b691ccf0a5 Revert "Add terminator to .eh_frame sections"
This reverts commit r296378.

I am pretty sure this is incorrect. In particular, for just

        .cfi_startproc
        nop
        .cfi_endproc

We now add an extra 4 zeros that neither bfd nor gold add.

llvm-svn: 296503
2017-02-28 18:55:08 +00:00
George Rimar 9f57a4f14a [ELF] - Do not set flags for CopyRelSection. NFC.
It does not make sense. Them added either to
Out::BssRelRo or Out::Bss, which are always RW.

llvm-svn: 296475
2017-02-28 14:22:15 +00:00
George Rimar 6159f1290e [ELF] - Remove ElfSym::EhdrStart member. NFC.
We do not use it later, so don't have to store.

llvm-svn: 296466
2017-02-28 10:34:56 +00:00
Sean Silva 6ab3926ee6 Comment on the typical/simple case of VA calculation
There are many special cases and a layer of abstraction or two in the
way, but the VA calculation in the typical case is actually very simple
and probably makes perfect sense even to somebody new to linkers.

Also, this line brings together many components and is a good place to
start understanding the linker (or improve one's existing
understanding).

llvm-svn: 296451
2017-02-28 09:01:58 +00:00
Sean Silva a9ba450c52 Add a comment.
Naively it seemed at first like getVA had the responsibility of adding
the addend, and getSymVA had the responsibility of getting the symbol
VA.
So it was not obvious to me at first why getVA passes Addend to
getSymVA. In fact, it passes it as a mutable reference.

It turns out that it only matters for SHF_MERGE sections, and in
particular only for STT_SECTION symbols that are used as a hack for
reducing the number of local symbols (e.g. to avoid a local symbol for
each string in the string table).

llvm-svn: 296448
2017-02-28 08:32:56 +00:00
Rui Ueyama 6b776ad985 Refactor {Gnu,}HashTableSection classes.
In other places in LLD, we use write32<E> instead of Elf_Word.
This patch uses the same technique in the hash table classes.
The hash table classes needs improving as they have almost no
comments. We at least need to describe the hash table structure
and why we have to support two different on-disk hash tables
for the same purpose. I'll do that later.

llvm-svn: 296439
2017-02-28 05:53:47 +00:00
Rui Ueyama ab0a85ef4d Remove unused typedefs.
llvm-svn: 296437
2017-02-28 05:25:37 +00:00
Rui Ueyama c3726f8263 Remove useless assignments.
These assignments seem meaningful, but actually all tests pass without them.

llvm-svn: 296436
2017-02-28 04:41:20 +00:00
Rui Ueyama b8dcdb5900 Merge SymbolTableSection::add{Global,Local} into one function.
Previously, these two functions put their symbols in different queues.
Now that the queues have been merged. So there's no point to keep two
separate functions.

llvm-svn: 296435
2017-02-28 04:20:16 +00:00
Rui Ueyama 406331ebfd Return early. NFC.
llvm-svn: 296434
2017-02-28 04:11:01 +00:00
Rui Ueyama 968db48cee Move SymbolTableSection::getOutputSection to SymbolBody::getOutputSection.
That function doesn't use any member of SymbolTableSection, so I
couldn't see a reason to make it a member of that class. The function
takes a SymbolBody, so it is more natural to make it a member of
SymbolBody.

llvm-svn: 296433
2017-02-28 04:02:42 +00:00
Rui Ueyama 6e96734dd1 Simplify SymbolTableSection<ELFT>::finalizeContents().
Previously, there were three conditions: .symtab, .dynsym or we are
producing a relocatable output. Turned out that the third condition is
the same as the first one. This patch removes that third condition and
simplify code.

llvm-svn: 296431
2017-02-28 03:29:12 +00:00
Sean Silva 1d96185f57 Add Triple::thumb to getBitcodeMachineKind
We really need to find a way to get this info from a single point of
truth in the LLVM backend, but it seems that the EM_* constants are
buried deep inside the constructors of the MCAsmBackend's.

For now, just fill in entries as we run into cases. AFAIK these mappings
are largely immutable, so we get a 75% discount on the technical debt
(code is duplicated, but little chance of divergence).

llvm-svn: 296429
2017-02-28 03:00:48 +00:00
Rui Ueyama 1f03253767 Improve SymbolTableSection synthetic section.
The previous code was a bit hard to understand because it unnecessarily
distinguished local and non-local symbols. It had NumLocals member
variable, but that variable didn't have a number of local symbols but
had some value that I cannot describe easily.

This patch rewrites SynbolTableSection::finalizeContents and
SymbolTableSection::writeTo to make it easy to understand. NumLocals
member variable has been removed, and writeGlobalSymbols and
writeLocalSymbols have been merged into one function.

There's still a piece of code that I think unnecessary. I'm not removing
that code in this patch, but will do in a follow-up patch.

llvm-svn: 296423
2017-02-28 01:56:36 +00:00
Rui Ueyama 3bc39016cd Refactor write{Global,Local}Symbols.
This part of code is hard to understand because NumLocals does not
actually mean the number of local symbols but something else (!).
We need to rewrite. But before that we need to clean it up.

llvm-svn: 296400
2017-02-27 22:39:50 +00:00
Rui Ueyama 1720ef1343 Add terminator to .eh_frame sections
Patch by Mark Kettenis.

Currenlty ld.lld does not add a terminator (a CIE with its length field
set to zero) to the .eh_frame sections it generates.  While the relevant
standards (the AMD64 SysV ABI and the Linux LSB) are not explicit about
this, such a terminator is expected by some unwinder implementations and
seems to be always emitted by ld.bfd. In addition to that, the Linux LSB

  https://refspecs.linuxfoundation.org/LSB_5.0.0/LSB-Core-generic/LSB-Core-generic/ehframechpt.html#EHFRAME

explicitly says that

  The .eh_frame section shall contain 1 or more Call Frame Information
  (CFI) records.

Currently, if the .eh_frame sections of the input files only contain
terminators, ld.lld emits a zero=sized .eh_frame section
which clearly doesn't meet that requirement.

The diff makes sure a terminator gets added to each .eh_frame section
and adjusts all the relevant tests to account for that.  An additional
test isn't needed as these adjustments mean that the existence of the
terminator is tested for by several tests already.

Differential Revision: https://reviews.llvm.org/D30335

llvm-svn: 296378
2017-02-27 20:44:59 +00:00
Rui Ueyama 3ead2e73ee De-template ResolvedReloc as it doesn't actually depends on ELFT.
Pointed out by Bob Haarman.

llvm-svn: 296377
2017-02-27 20:32:34 +00:00
Rui Ueyama aa5c52744b Simplify sortMipsSymbols function a bit.
llvm-svn: 296319
2017-02-27 03:31:38 +00:00
Rui Ueyama bb07d10601 Add comments to SymbolTableSection.
llvm-svn: 296318
2017-02-27 03:31:19 +00:00
Rui Ueyama 0c001d9407 Remove dead function.
llvm-svn: 296316
2017-02-27 03:13:36 +00:00
Rui Ueyama 0fb284aa35 Remove unnecessary "this->".
llvm-svn: 296315
2017-02-27 03:10:06 +00:00
Rui Ueyama 945055aabf Rename SyntheticSection::finalize -> finalizeContents.
In LLD source code, too many functions are called "finalize", although
what they do are different. This patch gives it a better name.

llvm-svn: 296314
2017-02-27 03:07:41 +00:00
Rui Ueyama 9320cb0719 De-template SyntheticSection.
This class didn't use ELFT.

llvm-svn: 296313
2017-02-27 02:56:02 +00:00
Rui Ueyama 3255a52200 De-template InterpSection because it doesn't use ELFT.
llvm-svn: 296311
2017-02-27 02:32:49 +00:00
Rui Ueyama f8d47fda43 Update comments. Now thunks are synthetic sections.
llvm-svn: 296310
2017-02-27 02:32:29 +00:00
Rui Ueyama 536a26706f Move SymbolTable<ELFT>::Sections out of the class.
The list of all input sections was defined in SymbolTable class for a
historical reason. The list itself is not a template. However, because
SymbolTable class is a template, we needed to pass around ELFT to access
the list. This patch moves the list out of the class so that it doesn't
need ELFT.

llvm-svn: 296309
2017-02-27 02:32:08 +00:00
Rui Ueyama 02a036f2e6 De-template OutputSectionFactory.
Since OutputSection is no longer a template, it doesn't make much
sense to tempalte its factory class.

llvm-svn: 296308
2017-02-27 02:31:48 +00:00
Rui Ueyama 9d1bacb1b4 Remove useless template so that Out<ELFT> becomes just Out.
llvm-svn: 296307
2017-02-27 02:31:26 +00:00
Rui Ueyama 4076fa1e21 De-template SharedSymbol.
Differential Revision: https://reviews.llvm.org/D30351

llvm-svn: 296303
2017-02-26 23:35:34 +00:00
Rui Ueyama 28eada105f Factor out more code. NFC.
llvm-svn: 296232
2017-02-25 02:43:01 +00:00
Rui Ueyama 47926abfa6 Remove a default value that is always overwritten.
llvm-svn: 296230
2017-02-25 02:27:58 +00:00
Rui Ueyama 10571cd306 Make function names consistent.
Now a function to parse -foo-bar is always named getFooBar.

llvm-svn: 296229
2017-02-25 02:23:28 +00:00
Rui Ueyama 524d44c6fa Do not handle -strip in a special way. NFC.
llvm-svn: 296227
2017-02-25 02:12:37 +00:00
Rui Ueyama dfa3ad1313 Factor out code to parse -hash-style.
llvm-svn: 296226
2017-02-25 02:07:27 +00:00
Rui Ueyama a37ace8d89 Rename OMagic -> Omagic for consistency.
llvm-svn: 296225
2017-02-25 01:52:03 +00:00
Rui Ueyama 7edde294aa Factor out code to parse -L and -rpath.
llvm-svn: 296224
2017-02-25 01:51:44 +00:00
Rui Ueyama a4a643c6b9 Sort options. NFC.
llvm-svn: 296223
2017-02-25 01:51:25 +00:00
Rui Ueyama 9cc84384ad Use uint64_t instead of uintX_t where the larger type just works.
llvm-svn: 296162
2017-02-24 19:52:52 +00:00
Rafael Espindola 24e6f363c5 Merge OutputSectionBase and OutputSection. NFC.
Now that all special sections are SyntheticSections, we only need one
OutputSection class.

llvm-svn: 296127
2017-02-24 15:07:30 +00:00
Rafael Espindola 29c1afb880 Delete trivial setter.
llvm-svn: 296124
2017-02-24 14:34:12 +00:00
Rafael Espindola 40849419e2 Delete trivial getter.
llvm-svn: 296123
2017-02-24 14:28:00 +00:00
Rafael Espindola d224c08efc Delete unused enum values.
llvm-svn: 296118
2017-02-24 13:26:52 +00:00
Rafael Espindola a2e3fee072 Remove unnecessary template. NFC.
llvm-svn: 296117
2017-02-24 13:19:33 +00:00
Rafael Espindola 798ad9a1e8 Expand a comment. NFC.
llvm-svn: 296114
2017-02-24 13:06:59 +00:00
George Rimar 87b0d68633 [ELF] - Implemented --no-dynamic-linker option
Feature is used for producing static-linked PIE executables
(https://gcc.gnu.org/ml/gcc/2015-06/msg00008.html)
And was implemented in GNU ld https://gcc.gnu.org/ml/gcc/2015-08/msg00099.html

I also found it in linux kernel build system,
though I think that x86/x64 bootloader does not really rely on it.
Seems it used for PPC though.

Differential revision: https://reviews.llvm.org/D30258

llvm-svn: 296097
2017-02-24 08:26:18 +00:00
Petr Hosek f087e312b6 [ELF] Make __ehdr_start point to ELF file headers
__ehdr_start should be pointing to ELF file headers, not program
headers.

This is a reland of D30319.

Differential Revision: https://reviews.llvm.org/D30323

llvm-svn: 296085
2017-02-24 04:11:02 +00:00
Petr Hosek 9a2bba7251 Revert "[ELF] Make __ehdr_start point to ELF file headers"
This reverts commit r296079.

llvm-svn: 296083
2017-02-24 03:18:59 +00:00
Petr Hosek 701ad3fbf9 [ELF] Make __ehdr_start point to ELF file headers
__ehdr_start should be pointing to ELF file headers, not program
headers.

Differential Revision: https://reviews.llvm.org/D30319

llvm-svn: 296079
2017-02-24 03:00:52 +00:00
Rafael Espindola 66b4e21534 Convert EhOutputSection to be a synthetic section.
With this we complete the transition out of special output sections,
and with the previous patches it should be possible to merge
OutputSectionBase and OuputSection.

llvm-svn: 296023
2017-02-23 22:06:28 +00:00
Rafael Espindola 774ea7d0a9 Make InputSection a class. NFC.
With the current design an InputSection is basically anything that
goes directly in a OutputSection. That includes plain input section
but also synthetic sections, so this should probably not be a
template.

llvm-svn: 295993
2017-02-23 16:49:07 +00:00
Rui Ueyama a7e87252ce Always add PT_GNU_STACK.
If -z stack-size is given, we need to add PT_GNU_STACK even if
-z execstack is not given.

llvm-svn: 295945
2017-02-23 08:09:51 +00:00
George Rimar 2146787609 [ELF] - Refactoring of LMA offset handling code. NFC.
Thanks to Rui Ueyama for suggestion.

llvm-svn: 295943
2017-02-23 07:57:55 +00:00
Rui Ueyama b7f39b08f4 Slightly improve an error message.
llvm-svn: 295942
2017-02-23 07:35:30 +00:00
Rui Ueyama c1a0ac2c4a Add a comment about .note.GNU-split-stack.
llvm-svn: 295941
2017-02-23 07:35:11 +00:00