The change broke some scenarios where debug information is still
needed, although MarkLive cannot see it, including the
Chromium/Android build. Reverting to unbreak that build.
llvm-svn: 360955
Summary:
We access Live and OutputOff (which may share the same memory location)
concurrently in 2 parallelForEachN loops. Separating them avoids subtle
data races like D41884/PR35788. This patch places Live and Hash
together.
2 reasons this is appealing:
1) Hash is immutable. Live is almost read-only - only written once in MarkLive.cpp where
Hash is not accessed
2) we already discard low bits of Hash to decide ShardID. It doesn't
matter much if we make 32-bit Hash to 31-bit.
For a huge internal clang -O3 executable (1.6GiB),
`Strings` in StringTableBuilder::finalizeStringTable contains at most 310253 elements.
The expected number of pair-wise collisions 2^(-31) * C(310253,2) ~= 22.41 is too small to have a negative impact on performance.
Actually, my benchmark shows there is actually a minor performance improvement.
Differential Revision: https://reviews.llvm.org/D60765
llvm-svn: 358645
Patch by Robert O'Callahan.
Rust projects tend to link in all object files from all dependent
libraries and rely on --gc-sections to strip unused code and data.
Unfortunately --gc-sections doesn't currently strip any debuginfo
associated with GC'ed sections, so lld links in the full debuginfo from
all dependencies even if almost all that code has been discarded. See
https://github.com/rust-lang/rust/issues/56068 for some details.
Properly stripping debuginfo for discarded sections would be difficult,
but a simple approach that helps significantly is to mark debuginfo
sections as live only if their associated object file has at least one
live code/data section. This patch does that. In a (contrived but not
totally artificial) Rust testcase linked above, it reduces the final
binary size from 46MB to 5.1MB.
Differential Revision: https://reviews.llvm.org/D54747
llvm-svn: 358069
This shaves another word off SectionBase and makes it possible to clone a
section using the implicit copy constructor.
This basically reverts r311056, which removed the mutex in order to
make the code easier to understand. On balance I think it's probably more
straightforward to have a mutex here than to have an unusual copy constructor
in SectionBase.
Differential Revision: https://reviews.llvm.org/D59269
llvm-svn: 355966
- The Assigned bit was previously taking a word on its own. Move
it into the bit fields in SectionBase.
- NumRelocations and AreRelocsRela were previously also taking up
a word despite only using half of it. Move them into the alignment gap
after SectionBase's fields.
Differential Revision: https://reviews.llvm.org/D59044
llvm-svn: 355622
to reflect the new license.
We understand that people may be surprised that we're moving the header
entirely to discuss the new license. We checked this carefully with the
Foundation's lawyer and we believe this is the correct approach.
Essentially, all code in the project is now made available by the LLVM
project under our new license, so you will see that the license headers
include that license only. Some of our contributors have contributed
code under our old license, and accordingly, we have retained a copy of
our old license notice in the top-level files in each project and
repository.
llvm-svn: 351636
Previously, we have a hash table containing strings and their offsets
to manage mergeable strings. Technically we can live without that, because
we can do binary search on a vector of mergeable strings to find a mergeable
strings.
We did have both the hash table and the binary search because we thought
that that is faster.
We recently observed that lld tend to consume more memory than gold when
building an output with debug info. A few percent of memory is consumed by
the hash table. So, we needed to reevaluate whether or not having the extra
hash table is a good CPU/memory tradeoff. I run a few benchmarks with and
without the hash table.
I got a mixed result for the benchmark. We observed a regression for some
programs by removing the hash table (that's what we expected), but we also
observed that performance imrpovements for some programs. This is perhaps
due to reduced memory usage.
Differential Revision: https://reviews.llvm.org/D55234
llvm-svn: 348401
Previously, we uncompress all compressed sections before doing anything.
That works, and that is conceptually simple, but that could results in
a waste of CPU time and memory if uncompressed sections are then
discarded or just copied to the output buffer.
In particular, if .debug_gnu_pub{names,types} are compressed and if no
-gdb-index option is given, we wasted CPU and memory because we
uncompress them into newly allocated bufers and then memcpy the buffers
to the output buffer. That temporary buffer was redundant.
This patch changes how to uncompress sections. Now, compressed sections
are uncompressed lazily. To do that, `Data` member of `InputSectionBase`
is now hidden from outside, and `data()` accessor automatically expands
an compressed buffer if necessary.
If no one calls `data()`, then `writeTo()` directly uncompresses
compressed data into the output buffer. That eliminates the redundant
memory allocation and redundant memcpy.
This patch significantly reduces memory consumption (20 GiB max RSS to
15 Gib) for an executable whose .debug_gnu_pub{names,types} are in total
5 GiB in an uncompressed form.
Differential Revision: https://reviews.llvm.org/D52917
llvm-svn: 343979
Once we create .gdb_index contents, .zdebug_gnu_pub{names,types}
are useless, so there's no need to keep their uncompressed data
in memory.
I observed that for a test case in which lld creates a 3GB .gdb_index
section, the maximum resident set size reduced from 43GB to 29GB after
this patch.
Differential Revision: https://reviews.llvm.org/D52126
llvm-svn: 342297
Patch by PkmX.
This patch makes lld recognize RISC-V target and implements basic
relocation for RV32/RV64 (and RVC). This should be necessary for static
linking ELF applications.
The ABI documentation for RISC-V can be found at:
https://github.com/riscv/riscv-elf-psabi-doc/blob/master/riscv-elf.md.
Note that the documentation is far from complete so we had to figure out
some details from bfd.
The patch should be pretty straightforward. Some highlights:
- A new relocation Expr R_RISCV_PC_INDIRECT is added. This is needed as
the low part of a PC-relative relocation is linked to the corresponding
high part (auipc), see:
https://github.com/riscv/riscv-elf-psabi-doc/blob/master/riscv-elf.md#pc-relative-symbol-addresses
- LLVM's MC support for RISC-V is very incomplete (we are working on
this), so tests are given in objectyaml format with the original
assembly included in the comments. Once we have complete support for
RISC-V in MC, we can switch to llvm-as/llvm-objdump.
- We don't support linker relaxation for now as it requires greater
changes to lld that is beyond the scope of this patch. Once this is
accepted we can start to work on adding relaxation to lld.
Differential Revision: https://reviews.llvm.org/D39322
llvm-svn: 339364
Note that this doesn't do the right thing in the case where there is
a linker script. We probably need to move output section assignment
before ICF to get the correct behaviour here.
Differential Revision: https://reviews.llvm.org/D47241
llvm-svn: 333052
The --keep-unique <symbol> option is taken from gold. The intention is that
<symbol> will be prevented from being folded by ICF. Although not
specifically mentioned in the documentation <symbol> only matches
global symbols, with a warning if the symbol is not found.
The implementation finds the Section defining <symbol> and removes it from
the set of sections considered for ICF.
Differential Revision: https://reviews.llvm.org/D46755
llvm-svn: 332332
Now that getSectionPiece is fast (uses a hash) it is probably OK to
split merge sections early.
The reason I want to do this is to split eh_frame sections in the same
place.
This does mean that we have to decompress early. Given that the only
compressed sections are debug info, I don't think we are missing much.
It is a small improvement: 0.5% on the geometric mean.
llvm-svn: 331058
In the lld perf builder r328686 had a negative impact in
stalled-cycles-frontend. Somehow that stat is not showing on my
machine, but the attached patch shows an improvement on cache-misses,
which is probably a reasonable proxy.
My working theory is that given a large input the pieces vector is out
of cache by the time initOffsetMap runs.
Both finalizeContents implementation have a convenient location for
initializing the OffsetMap, so this seems the best solution.
llvm-svn: 329117
We normally avoid "switch (Config->EKind)", but in this case I think
it is worth it.
It is only executed when there is an error and it allows detemplating
a lot of code.
llvm-svn: 321404
It is currently in InputSectionBase. Only InputSections are used in
ICF, so Repl should be move to InputSection to clear the class
hierarchy or, like this patch does, to SectionBase for convenience.
The convenience of having it on the base class is that we can just
access the replacement without having to first check if it is an
InputSection. It is a bit less code and a bit faster as some of this
code is very hot.
I got up to 1.77% improvement in clang-gdb-index and no regressions
according to lnt.
llvm-svn: 320654
The size of an OutputSection is calculated early, to aid handling of compressed
debug sections. However, subsequent to this point, unused synthetic sections are
removed. In the event that an OutputSection, from which such an InputSection is
removed, is still required (e.g. because it has a symbol assignment), and no longer
has any InputSections, dot assignments, or BYTE()-family directives, the size
member is never updated when processing the commands. If the removed InputSection
had a non-zero size (such as a .got.plt section), the section ends up with the
wrong size in the output.
The fix is to reset the OutputSection size prior to processing the linker script
commands relating to that OutputSection. This ensures that the size is correct even
in the above situation.
Additionally, to reduce the risk of developers misusing OutputSection Size and
InputSection OutSecOff, they are set to simply the number of InputSections in an
OutputSection, and the corresponding index respectively. We cannot completely
stop using them, due to SHF_LINK_ORDER sections requiring them.
Compressed debug sections also require the full size. This is now calculated in
maybeCompress for these kinds of sections.
Reviewers: ruiu, rafael
Differential Revision: https://reviews.llvm.org/D38361
llvm-svn: 320472
Now that DefinedRegular is the only remaining derived class of
Defined, we can merge the two classes.
Differential Revision: https://reviews.llvm.org/D39667
llvm-svn: 317448
Common symbols are now represented with a DefinedRegular that points
to a BssSection, even during symbol resolution.
Differential Revision: https://reviews.llvm.org/D39666
llvm-svn: 317447
Now that we have only SymbolBody as the symbol class. So, "SymbolBody"
is a bit strange name now. This is a mechanical change generated by
perl -i -pe s/SymbolBody/Symbol/g $(git grep -l SymbolBody lld/ELF lld/COFF)
nd clang-format-diff.
Differential Revision: https://reviews.llvm.org/D39459
llvm-svn: 317370
This is PR34826.
Currently LLD is unable to report line number when reporting
duplicate declaration of some variable.
That happens because for extracting line information we always use
.debug_line section content which describes mapping from machine
instructions to source file locations, what does not help for
variables as does not describe them.
In this patch I am taking the approproate information about
variables locations from the .debug_info section.
Differential revision: https://reviews.llvm.org/D38721
llvm-svn: 317080
This bit is to manage whether an input section has already been assigned
to some output section by linker scripts or not. So it logically belongs
to InputSectionBase. SectionBase is a common base class for input and
output sections, so that wasn't the right place to define the bit.
llvm-svn: 316879
This is "Bug 34836 - --gc-sections remove relocations from --emit-relocs",
When --emit-relocs is used, LLD currently always drops SHT_REL[A] sections from
output if --gc-sections is present. Patch fixes the issue.
Differential revision: https://reviews.llvm.org/D38724
llvm-svn: 316418
We used to have a map from section piece offsets to section pieces
as a cache for binary search. But I found that the map took quite a
large amount of memory and didn't make linking faster. So, in this
patch, I removed the map.
This patch saves 566 MiB of RAM (2.019 GiB -> 1.453 GiB) when linking
clang with debug info, and the link time is 4% faster in that test case.
Thanks for Sean Silva for pointing this out.
llvm-svn: 316305
By assuming that mergeable input sections are smaller than 4 GiB,
lld's memory usage when linking clang with debug info drops from
2.788 GiB to 2.019 GiB (measured by valgrind, and that does not include
memory space for mmap'ed files). I think that's a reasonable assumption
given such a large RAM savings, so this patch.
According to valgrind, gold needs 3.54 GiB of RAM to do the same thing.
NB: This patch does not introduce a limitation on the size of
output sections. You can still create sections larger than 4 GiB.
llvm-svn: 316280
When reporting a symbol conflict, LLD parses the debug info to report
source location information. Sections have not been decompressed at this
point, so if an object file contains zlib compressed debug info, LLD
ends up passing this compressed debug info to the DWARF parser, which
causes debug info parsing failures and can trigger assertions in the
parser (as the test case demonstrates).
Decompress debug sections when constructing the LLDDwarfObj to avoid
this issue. This doesn't handle GNU-style compressed debug info sections
(.zdebug_*), which at present are simply ignored by LLDDwarfObj; those
can be done in a follow-up.
Differential Revision: https://reviews.llvm.org/D38491
llvm-svn: 314866
New lld's files are spread under lib subdirectory, and it isn't easy
to find which files are actually maintained. This patch moves maintained
files to Common subdirectory.
Differential Revision: https://reviews.llvm.org/D37645
llvm-svn: 314719
Reads from `Live` and writes to `OutputOff` in the following code race
even though they are logically independent because they are bitfields
sharing the same word.
for (size_t I = 0, E = Sec->Pieces.size(); I != E; ++I) {
if (!Sec->Pieces[I].Live)
continue;
CachedHashStringRef Str = Sec->getData(I);
size_t ShardId = getShardId(Str.hash());
if ((ShardId & (Concurrency - 1)) == ThreadId)
Sec->Pieces[I].OutputOff = Shards[ShardId].add(Str);
}
llvm-svn: 314711
EhSectionPiece used to have a pointer to a section, but that pointer was
mostly redundant because we almost always know what the section is without
using that pointer. This patch removes the pointer from the struct.
This patch also use uint32_t/int32_t instead of size_t to represent
offsets that are hardly be larger than 4 GiB. At the moment, I think it is
OK even if we cannot handle .eh_frame sections larger than 4 GiB.
Differential Revision: https://reviews.llvm.org/D38012
llvm-svn: 313697
This patch removes lot of static Instances arrays from different input file
classes and introduces global arrays for access instead. Similar to arrays we
have for InputSections/OutputSectionCommands.
It allows to iterate over input files in a non-templated code.
Differential revision: https://reviews.llvm.org/D35987
llvm-svn: 313619
EhSectionPiece inherited from SectionPiece, but we did not actually use
EhSectionPiece objects as SectionPiece ojbects. They were handled as
distinct types. So it didn't make much sense to use inheritance.
llvm-svn: 313587
We had a lock to guard BAlloc from being used concurrently, but that
is not very easy to understand. This patch replaces it with a
std::unique_ptr.
llvm-svn: 311056
This is PR33889,
Patch adds support of combination of linkerscript and
-symbol-ordering-file option.
If no sorting commands are present in script inside section declaration
and no --sort-section option specified, code uses sorting from ordering
file if any exist.
Differential revision: https://reviews.llvm.org/D35843
llvm-svn: 310045
Before InputSectionBase had an OutputSection pointer, but that was not
always valid. For example, if it was a merge section one actually had
to look at MergeSec->OutSec.
This was brittle and caused bugs like the one fixed by r304260.
We now have a single Parent pointer that points to an OutputSection
for InputSection, but to a SyntheticSection for merge sections and
.eh_frame. This makes it impossible to accidentally access an invalid
OutSec.
llvm-svn: 304338
We would crash if a SHF_LINK_ORDER section pointed to a non
InputSection section. Since those sections are not merged in order,
SHF_LINK_ORDER is pretty meaningless and we can error on that case.
llvm-svn: 304327
This is PR33052, "Bug 33052 - -r eats comdats ".
To fix it I stop removing group section from out when -r is given
and fixing SHT_GROUP content when writing it just like we do some
other fixup, e.g. for Rel[a]. (it needs fix for section indices that
are in group).
Differential revision: https://reviews.llvm.org/D33485
llvm-svn: 304140
Summary:
This is required on some platforms, as GNU libstdc++ std::call_once is known to be buggy.
This fixes operation of LLD on at least NetBSD and perhaps OpenBSD and Linux PowerPC.
The same change has been introduced to LLVM and LLDB.
Reviewers: ruiu
Reviewed By: ruiu
Subscribers: emaste, #lld
Tags: #lld
Differential Revision: https://reviews.llvm.org/D33508
llvm-svn: 303788
GetSection is a template because write calls relocate.
relocate has two parts. The non alloc code really has to be a
template, as it is looking a raw input file data.
The alloc part is only a template because of getSize.
This patch folds the value of getSize early, detemplates
getRelocTargetVA and splits relocate into a templated non alloc case
and a regular function for the alloc case. This has the nice advantage
of making sure we collect all the information we need for relocations
before getting to InputSection::relocateNonAlloc.
Since we know got is alloc, it can just call the function directly and
avoid the template.
llvm-svn: 303355
Previously, undefined symbol errors are one line like this
and wasn't easy to read.
/ssd/clang/bin/ld.lld: error: /ssd/llvm-project/lld/ELF/Writer.cpp:207: undefined symbol 'lld:🧝:EhFrameSection<llvm::object::ELFType<(llvm::support::endianness)0, true> >::addSection(lld:🧝:InputSectionBase*)'
This patch make it more structured like this.
bin/ld.lld: error: undefined symbol: lld:🧝:EhFrameSection<llvm::object::ELFType<(llvm::support::endianness)0, true>
>>> Referenced by Writer.cpp:207 (/ssd/llvm-project/lld/ELF/Writer.cpp:207)
>>> Writer.cpp.o in archive lib/liblldELF.a
Discussion thread:
http://lists.llvm.org/pipermail/llvm-dev/2017-March/111459.html
Differential Revision: https://reviews.llvm.org/D31481
llvm-svn: 299097
With this we have a single section hierarchy. It is a bit less code,
but the main advantage will be in a future patch being able to handle
foo = symbol_in_obj;
in a linker script. Currently that fails since we try to find the
output section of symbol_in_obj. With this we should be able to just
return an InputSection from the expression.
llvm-svn: 297313
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
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
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
Bob Haarman