This commit moves the parsing of linker optimization hints into
`ARM64::applyOptimizationHints`. This lets us avoid allocating memory
for holding the parsed information, and moves work out of
`ObjFile::parse`, which is not parallelized at the moment.
This change reduces the overhead of processing LOHs to 25-30 ms when
linking Chromium Framework on my M1 machine; previously it took close to
100 ms.
There's no statistically significant change in runtime for a --threads=1
link.
Performance figures with all 8 cores utilized:
N Min Max Median Avg Stddev
x 20 3.8027232 3.8760762 3.8505335 3.8454145 0.026352574
+ 20 3.7019017 3.8660538 3.7546209 3.7620371 0.032680043
Difference at 95.0% confidence
-0.0833775 +/- 0.019
-2.16823% +/- 0.494094%
(Student's t, pooled s = 0.0296854)
Differential Revision: https://reviews.llvm.org/D133439
This is what ld64 does (though it doesn't use ICF to do this; instead it
always dedups selrefs by default).
We'll want to dedup implicitly-defined selrefs as well, but I will leave
that for future work.
Additionally, I'm not *super* happy with the current LLD implementation
because I think it is rather janky and inefficient. But at least it
moves us toward the goal of closing the size gap with ld64. I've
described ideas for cleaning up our implementation here:
https://github.com/llvm/llvm-project/issues/57714
Differential Revision: https://reviews.llvm.org/D133780
So @keith observed
[here](https://reviews.llvm.org/D128108#inline-1263900) that the
StringRefs we were returning from `CStringInputSection::getStringRef()`
included the null terminator in their total length, but regular
StringRefs do not. Let's fix that so these StringRefs are less confusing
to use.
Reviewed By: #lld-macho, keith, Roger
Differential Revision: https://reviews.llvm.org/D133728
This commit removes the `relocTargets` vector, and instead makes the
code reconstruct the referent addresses from the relocated instructions.
This will allow us to move `applyOptimizationHints` from
`ConcatInputSection::writeTo` to a separate pass that parses and applies
LOHs in one step, on a per-file basis. This will improve performance, as
parsing is currently done serially in `ObjFile::parse`.
I opted to remove the sanity check that ensures that all relocations
within a LOH point to the same symbol. This completely eliminates the
need to search through relocations. It is my understanding that
mismatched relocation targets should not be present in valid object
files, so it's unlikely that the removal will lead to mislinks.
Differential Revision: https://reviews.llvm.org/D133274
Apple Clang in Xcode 14 introduced a new feature for reducing the
overhead of objc_msgSend calls by deduplicating the setup calls for each
individual selector. This works by clang adding undefined symbols for
each selector called in a translation unit, such as `_objc_msgSend$foo`
for calling the `foo` method on any `NSObject`. There are 2
different modes for this behavior, the default directly does the setup
for `_objc_msgSend` and calls it, and the smaller option does the
selector setup, and then calls the standard `_objc_msgSend` stub
function.
The general overview of how this works is:
- Undefined symbols with the given prefix are collected
- The suffix of each matching undefined symbol is added as a string to
`__objc_methname`
- A pointer is added for every method name in the `__objc_selrefs`
section
- A `got` entry is emitted for `_objc_msgSend`
- Stubs are emitting pointing to the synthesized locations
Notes:
- Both `__objc_methname` and `__objc_selrefs` can also exist from object
files, so their contents are merged with our synthesized contents
- The compiler emits method names for defined methods, but not for
undefined symbols you call, but stubs are used for both
- This only implements the default "fast" mode currently just to reduce
the diff, I also doubt many folks will care to swap modes
- This only implements this for arm64 and x86_64, we don't need to
implement this for 32 bit iOS archs, but we should implement it for
watchOS archs in a later diff
Differential Revision: https://reviews.llvm.org/D128108
If the `-demangle` flag is passed to lld, symbol names will now be
demangled in the "referenced by:" message in addition to the referenced
symbol's name, which was already demangled before this change.
Differential Revision: https://reviews.llvm.org/D130490
To do this, we need to slice away the LSDA pointer, just like we are
slicing away the functionAddress pointer.
No observable difference in perf on chromium_framework:
base diff difference (95% CI)
sys_time 1.769 ± 0.068 1.761 ± 0.065 [ -2.7% .. +1.8%]
user_time 9.517 ± 0.110 9.528 ± 0.116 [ -0.6% .. +0.8%]
wall_time 8.291 ± 0.174 8.307 ± 0.183 [ -1.1% .. +1.5%]
samples 21 25
Reviewed By: #lld-macho, thakis
Differential Revision: https://reviews.llvm.org/D129830
This fixes https://github.com/llvm/llvm-project/issues/56238. ld64.lld currently does not generate __dof section in Mach-O, and -no_dtrace_dof option is on by default. However when there are user-defined dtrace symbols, ld64.lld will treat them as undefined symbols, which causes the linking to fail because lld cannot find their definitions. This patch allows ld64.lld to rewrite the instructions calling dtrace symbols to instructions like nop as what ld64 does; therefore, when encountered with user-provided dtrace probes, the linking can still succeed.
I'm not sure whether support for dtrace is expected in lld, so for now I didn't add codes to make lld emit __dof section like ld64, and only made it possible to link with dtrace symbols provided. If this feature is needed, I can add that part in Dtrace.cpp & Dtrace.h.
Reviewed By: int3, #lld-macho
Differential Revision: https://reviews.llvm.org/D129062
Linker optimization hints mark a sequence of instructions used for
synthesizing an address, like ADRP+ADD. If the referenced symbol ends up
close enough, it can be replaced by a faster sequence of instructions
like ADR+NOP.
This commit adds support for 2 of the 7 defined ARM64 optimization
hints:
- LOH_ARM64_ADRP_ADD, which transforms a pair of ADRP+ADD into ADR+NOP
if the referenced address is within +/- 1 MiB
- LOH_ARM64_ADRP_ADRP, which transforms two ADRP instructions into
ADR+NOP if they reference the same page
These two kinds already cover more than 50% of all LOHs in
chromium_framework.
Differential Review: https://reviews.llvm.org/D128093
Similarly to how undefined symbol diagnostics were changed in D128184,
we now show where in the source file duplicate symbols are defined at:
ld64.lld: error: duplicate symbol: _foo
>> defined in bar.c:42
>> /path/to/bar.o
>> defined in baz.c:1
>> /path/to/libbaz.a(baz.o)
For objects that don't contain DWARF data, the format is unchanged.
A slight difference to undefined symbol diagnostics is that we don't
print the name of the symbol on the third line, as it's already
contained on the first line.
Differential Revision: https://reviews.llvm.org/D128425
The error used to look like this:
ld64.lld: error: undefined symbol: _foo
>>> referenced by /path/to/bar.o:(symbol _baz+0x4)
If DWARF line information is available, we now show where in the source
the references are coming from:
ld64.lld: error: unreferenced symbol: _foo
>>> referenced by: bar.cpp:42 (/path/to/bar.cpp:42)
>>> /path/to/bar.o:(symbol _baz+0x4)
The reland is identical to the first time this landed. The fix was in D128294.
This reverts commit 0cc7ad4175.
Differential Revision: https://reviews.llvm.org/D128184
The error used to look like this:
ld64.lld: error: undefined symbol: _foo
>>> referenced by /path/to/bar.o:(symbol _baz+0x4)
If DWARF line information is available, we now show where in the source
the references are coming from:
ld64.lld: error: unreferenced symbol: _foo
>>> referenced by: bar.cpp:42 (/path/to/bar.cpp:42)
>>> /path/to/bar.o:(symbol _baz+0x4)
Differential Revision: https://reviews.llvm.org/D128184
This commit fixes the issue that getLocation always printed the name of
the first symbol in the section.
For clarity, upper_bound is used instead of a linear search for finding
the closest symbol name. Note that this change does not affect
performance: this function is only called when printing errors and
`symbols` typically contains a single symbol because of
.subsections_via_symbols.
Differential Revision: https://reviews.llvm.org/D127670
This reverts commit 942f4e3a7c.
The additional change required to avoid the assertion errors seen
previously is:
--- a/lld/MachO/ICF.cpp
+++ b/lld/MachO/ICF.cpp
@@ -443,7 +443,9 @@ void macho::foldIdenticalSections() {
/*relocVA=*/0);
isec->data = copy;
}
- } else {
+ } else if (!isEhFrameSection(isec)) {
+ // EH frames are gathered as hashables from unwindEntry above; give a
+ // unique ID to everything else.
isec->icfEqClass[0] = ++icfUniqueID;
}
}
Differential Revision: https://reviews.llvm.org/D123435
== Background ==
`llvm-mc` generates unwind info in both compact unwind and DWARF
formats. LLD already handles the compact unwind format; this diff gets
us close to handling the DWARF format properly.
== Caveats ==
It's not quite done yet, but I figure it's worth getting this reviewed
and landed first as it's shaping up to be a fairly large code change.
**Known limitations of the current code:**
* Only works for x86_64, for which `llvm-mc` emits "abs-ified"
relocations as described in 618def651b.
`llvm-mc` emits regular relocations for ARM EH frames, which we do not
yet handle correctly.
Since the feature is not ready for real use yet, I've gated it behind a
flag that only gets toggled on during test suite runs. With most of the
new code disabled, we see just a hint of perf regression, so I don't
think it'd be remiss to land this as-is:
base diff difference (95% CI)
sys_time 1.926 ± 0.168 1.979 ± 0.117 [ -1.2% .. +6.6%]
user_time 3.590 ± 0.033 3.606 ± 0.028 [ +0.0% .. +0.9%]
wall_time 7.104 ± 0.184 7.179 ± 0.151 [ -0.2% .. +2.3%]
samples 30 31
== Design ==
Like compact unwind entries, EH frames are also represented as regular
ConcatInputSections that get pointed to via `Defined::unwindEntry`. This
allows them to be handled generically by e.g. the MarkLive and ICF
code. (But note that unlike compact unwind subsections, EH frame
subsections do end up in the final binary.)
In order to make EH frames "look like" a regular ConcatInputSection,
some processing is required. First, we need to split the `__eh_frame`
section along EH frame boundaries rather than along symbol boundaries.
We do this by decoding the length field of each EH frame. Second, the
abs-ified relocations need to be turned into regular Relocs.
== Next Steps ==
In order to support EH frames on ARM targets, we will either have to
teach LLD how to handle EH frames with explicit relocs, or we can try to
make `llvm-mc` emit abs-ified relocs for ARM as well. I'm hoping to do
the latter as I think it will make the LLD implementation both simpler
and faster to execute.
== Misc ==
The `obj-file-with-stabs.s` test had to be updated as the previous
version would trip assertion errors in the code. It appears that in our
attempt to produce a minimal YAML test input, we created a file with
invalid EH frame data. I've fixed this by re-generating the YAML and not
doing any hand-pruning of it.
Reviewed By: #lld-macho, Roger
Differential Revision: https://reviews.llvm.org/D123435
This matches ld64, and makes dsymutil work better with lld's output.
Fixes PR54783, see there for details.
Reduces time needed to run dsymutil on Chromium Framework from 8m30s
(which is already down from 26 min with D123218) to 6m30s and removes
many lines of "could not find object file symbol for symbol" from dsymutil output
(previously: several MB of those messages, now dsymutil is completely silent).
Differential Revision: https://reviews.llvm.org/D123252
ld64 breaks down `__objc_classrefs` on a per-word level and deduplicates
them. This greatly reduces the number of bind entries emitted (and
therefore the amount of work `dyld` has to do at runtime). For
chromium_framework, this change to LLD cuts the number of (non-lazy)
binds from 912 to 190, getting us to parity with ld64 in this aspect.
Reviewed By: #lld-macho, thakis
Differential Revision: https://reviews.llvm.org/D121053
... from a `uint64_t` to a `uint32_t`. (LLD-ELF uses a `uint32_t` too.)
About a 1.7% reduction in peak RSS when linking chromium_framework on my
3.2 GHz 16-Core Intel Xeon W Mac Pro, and no stat sig change in wall
time.
</Users/jezng/test2.sh ["before"]> </Users/jezng/test2.sh ["after"]> difference (95% CI)
RSS 1003036672.000 ± 9891065.259 985539505.231 ± 10272748.749 [ -2.3% .. -1.2%]
samples 27 26
base diff difference (95% CI)
sys_time 1.277 ± 0.023 1.277 ± 0.024 [ -0.9% .. +0.9%]
user_time 6.682 ± 0.046 6.598 ± 0.043 [ -1.6% .. -0.9%]
wall_time 5.904 ± 0.062 5.895 ± 0.063 [ -0.7% .. +0.4%]
samples 46 28
No appreciable change (~0.01%) in number of `equals` comparisons either:
Before:
ld64.lld: ICF needed 8 iterations
ld64.lld: equalsConstant() called 701643 times
ld64.lld: equalsVariable() called 3438526 times
After:
ld64.lld: ICF needed 8 iterations
ld64.lld: equalsConstant() called 701729 times
ld64.lld: equalsVariable() called 3438526 times
Reviewed By: #lld-macho, MaskRay, thakis
Differential Revision: https://reviews.llvm.org/D121052
This makes it easier to pinpoint the source of the problem.
TODO: Have more relocation error messages make use of this
functionality.
Reviewed By: #lld-macho, oontvoo
Differential Revision: https://reviews.llvm.org/D118798
Earlier in LLD's evolution, I tried to create the illusion that
subsections were indistinguishable from "top-level" sections. Thus, even
though the subsections shared many common field values, I hid those
common values away in a private Shared struct (see D105305). More
recently, however, @gkm added a public `Section` struct in D113241 that
served as an explicit way to store values that are common to an entire
set of subsections (aka InputSections). Now that we have another "common
value" struct, `Shared` has been rendered redundant. All its fields can
be moved into `Section` instead, and the pointer to `Shared` can be replaced
with a pointer to `Section`.
This `Section` pointer also has the advantage of letting us inspect other
subsections easily, simplifying the implementation of {D118798}.
P.S. I do think that having both `Section` and `InputSection` makes for
a slightly confusing naming scheme. I considered renaming `InputSection`
to `Subsection`, but that would break the symmetry with `OutputSection`.
It would also make us deviate from LLD-ELF's naming scheme.
This change is perf-neutral on my 3.2 GHz 16-Core Intel Xeon W machine:
base diff difference (95% CI)
sys_time 1.258 ± 0.031 1.248 ± 0.023 [ -1.6% .. +0.1%]
user_time 3.659 ± 0.047 3.658 ± 0.041 [ -0.5% .. +0.4%]
wall_time 4.640 ± 0.085 4.625 ± 0.063 [ -1.0% .. +0.3%]
samples 49 61
There's also no stat sig change in RSS (as measured by `time -l`):
base diff difference (95% CI)
time 998038627.097 ± 13567305.958 1003327715.556 ± 15210451.236 [ -0.2% .. +1.2%]
samples 31 36
Reviewed By: #lld-macho, oontvoo
Differential Revision: https://reviews.llvm.org/D118797
std::vector can have different sizes depending on the STL's debug level,
so account for its size separately. (You could argue that we should be
accounting for all the other members separately as well, but that would
be very unergonomic, and std::vector is the only one that's caused
problems so far.)
In order to keep signal:noise high for the `__eh_frame` diff, I have teased-out the NFC changes and put them here.
Differential Revision: https://reviews.llvm.org/D114017
As discussed in https://reviews.llvm.org/D113809#3128636. It's a bit
unfortunate to move the asserts away from the structs whose sizes
they're checking, but it's a far better developer experience when one of
the asserts is violated, because you get a single error instead of every
single source file including the header erroring out.
WordLiteralSection dedupes literals by content.
WordLiteralInputSection::getOffset() used to read a literal at the passed-in
offset and look up this value in the deduping map to find the offset of the
deduped value.
But it's possible that (e.g.) a 16-byte literal's value is accessed 4 bytes in.
To get the offset at that address, we have to get the deduped value at offset 0
and then apply the offset 4 to the result.
(See also WordLiteralSection::finalizeContents() which fills in those maps.)
Only a problem on arm64 because in x86_64 the offset is part of the instruction
instead of a separate ARM64_RELOC_ADDEND relocation. (See bug for more details.)
Fixes PR51999.
Differential Revision: https://reviews.llvm.org/D112584
Compact unwind entries (CUEs) contain pointers to their respective
function symbols. However, during the link process, it's far more useful
to have pointers from the function symbol to the CUE than vice versa.
This diff adds that pointer in the form of `Defined::compactUnwind`.
In particular, when doing dead-stripping, we want to mark CUEs live when
their function symbol is live; and when doing ICF, we want to dedup
sections iff the symbols in that section have identical CUEs. In both
cases, we want to be able to locate the symbols within a given section,
as well as locate the CUEs belonging to those symbols. So this diff also
adds `InputSection::symbols`.
The ultimate goal of this refactor is to have ICF support dedup'ing
functions with unwind info, but that will be handled in subsequent
diffs. This diff focuses on simplifying `-dead_strip` --
`findFunctionsWithUnwindInfo` is no longer necessary, and
`Defined::isLive()` is now a lot simpler. Moreover, UnwindInfoSection no
longer has to check for dead CUEs -- we simply avoid adding them in the
first place.
Additionally, we now support stripping of dead LSDAs, which follows
quite naturally since `markLive()` can now reach them via the CUEs.
Reviewed By: #lld-macho, gkm
Differential Revision: https://reviews.llvm.org/D109944
ICF previously operated only within a given OutputSection. We would
merge all CFStrings first, then merge all regular code sections in a
second phase. This worked fine since CFStrings would never reference
regular `__text` sections. However, I would like to expand ICF to merge
functions that reference unwind info. Unwind info references the LSDA
section, which can in turn reference the `__text` section, so we cannot
perform ICF in phases.
In order to have ICF operate on InputSections spanning multiple
OutputSections, we need a way to distinguish InputSections that are
destined for different OutputSections, so that we don't fold across
section boundaries. We achieve this by creating OutputSections early,
and setting `InputSection::parent` to point to them. This is what
LLD-ELF does. (This change should also make it easier to implement the
`section$start$` symbols.)
This diff also folds InputSections w/o checking their flags, which I
think is the right behavior -- if they are destined for the same
OutputSection, they will have the same flags in the output (even if
their input flags differ). I.e. the `parent` pointer check subsumes the
`flags` check. In practice this has nearly no effect (ICF did not become
any more effective on chromium_framework).
I've also updated ICF.cpp's block comment to better reflect its current
status.
Reviewed By: #lld-macho, smeenai
Differential Revision: https://reviews.llvm.org/D105641
We have been creating many ConcatInputSections with identical values due
to .subsections_via_symbols. This diff factors out the identical values
into a Shared struct, to reduce memory consumption and make copying
cheaper.
I also changed `callSiteCount` from a uint32_t to a 31-bit field to save an
extra word.
All in all, this takes InputSection from 120 to 72 bytes (and
ConcatInputSection from 160 to 112 bytes), i.e. 30% size reduction in
ConcatInputSection.
Numbers for linking chromium_framework on my 3.2 GHz 16-Core Intel Xeon W:
N Min Max Median Avg Stddev
x 20 4.14 4.24 4.18 4.183 0.027548999
+ 20 4.04 4.11 4.075 4.0775 0.018027756
Difference at 95.0% confidence
-0.1055 +/- 0.0149005
-2.52211% +/- 0.356215%
(Student's t, pooled s = 0.0232803)
Reviewed By: #lld-macho, thakis
Differential Revision: https://reviews.llvm.org/D105305
`__cfstring` is a special literal section, so instead of breaking it up
at symbol boundaries, we break it up at fixed-width boundaries (since
each literal is the same size). Symbols can only occur at one of those
boundaries, so this is strictly more powerful than
`.subsections_via_symbols`.
With that in place, we then run the section through ICF.
This change is about perf-neutral when linking chromium_framework.
Reviewed By: #lld-macho, gkm
Differential Revision: https://reviews.llvm.org/D105045
This is a pretty big refactoring diff, so here are the motivations:
Previously, ICF ran after scanRelocations(), where we emitting
bind/rebase opcodes etc. So we had a bunch of redundant leftovers after
ICF. Having ICF run before Writer seems like a better design, and is
what LLD-ELF does, so this diff refactors it accordingly.
However, ICF had two dependencies on things occurring in Writer: 1) it
needs literals to be deduplicated beforehand and 2) it needs to know
which functions have unwind info, which was being handled by
`UnwindInfoSection::prepareRelocations()`.
In order to do literal deduplication earlier, we need to add literal
input sections to their corresponding output sections. So instead of
putting all input sections into the big `inputSections` vector, and then
filtering them by type later on, I've changed things so that literal
sections get added directly to their output sections during the 'gather'
phase. Likewise for compact unwind sections -- they get added directly
to the UnwindInfoSection now. This latter change is not strictly
necessary, but makes it easier for ICF to determine which functions have
unwind info.
Adding literal sections directly to their output sections means that we
can no longer determine `inputOrder` from iterating over
`inputSections`. Instead, we store that order explicitly on
InputSection. Bloating the size of InputSection for this purpose would
be unfortunate -- but LLD-ELF has already solved this problem: it reuses
`outSecOff` to store this order value.
One downside of this refactor is that we now make an additional pass
over the unwind info relocations to figure out which functions have
unwind info, since want to know that before `processRelocations()`. I've
made sure to run that extra loop only if ICF is enabled, so there should
be no overhead in non-optimizing runs of the linker.
The upside of all this is that the `inputSections` vector now contains
only ConcatInputSections that are destined for ConcatOutputSections, so
we can clean up a bunch of code that just existed to filter out other
elements from that vector.
I will test for the lack of redundant binds/rebases in the upcoming
cfstring deduplication diff. While binds/rebases can also happen in the
regular `.text` section, they're more common in `.data` sections, so it
seems more natural to test it that way.
This change is perf-neutral when linking chromium_framework.
Reviewed By: oontvoo
Differential Revision: https://reviews.llvm.org/D105044
Fixes PR50637.
Downstream bug: https://crbug.com/1218958
Currently, we split __cstring along symbol boundaries with .subsections_via_symbols
when not deduplicating, and along null bytes when deduplicating. This change splits
along null bytes unconditionally, and preserves original alignment in the non-
deduplicated case.
Removing subsections-section-relocs.s because with this change, __cstring
is never reordered based on the order file.
Differential Revision: https://reviews.llvm.org/D104919
Literal sections can be deduplicated before running ICF. That makes it
easy to compare them during ICF: we can tell if two literals are
constant-equal by comparing their offsets in their OutputSection.
LLD-ELF takes a similar approach.
Reviewed By: #lld-macho, gkm
Differential Revision: https://reviews.llvm.org/D104671
`icfEqClass` only makes sense on ConcatInputSections since (in contrast
to literal sections) they are deduplicated as an atomic unit.
Similarly, `hasPersonality` and `replacement` don't make sense on
literal sections.
This mirrors LLD-ELF, which stores `icfEqClass` only on non-mergeable
sections.
Reviewed By: #lld-macho, gkm
Differential Revision: https://reviews.llvm.org/D104670
ICF = Identical C(ode|OMDAT) Folding
This is the LLD ELF/COFF algorithm, adapted for MachO. So far, only `-icf all` is supported. In order to support `-icf safe`, we will need to port address-significance tables (`.addrsig` directives) to MachO, which will come in later diffs.
`check-{llvm,clang,lld}` have 0 regressions for `lld -icf all` vs. baseline ld64.
We only run ICF on `__TEXT,__text` for reasons explained in the block comment in `ConcatOutputSection.cpp`.
Here is the perf impact for linking `chromium_framekwork` on a Mac Pro (16-core Xeon W) for the non-ICF case vs. pre-ICF:
```
N Min Max Median Avg Stddev
x 20 4.27 4.44 4.34 4.349 0.043029977
+ 20 4.37 4.46 4.405 4.4115 0.025188761
Difference at 95.0% confidence
0.0625 +/- 0.0225658
1.43711% +/- 0.518873%
(Student's t, pooled s = 0.0352566)
```
Reviewed By: #lld-macho, int3
Differential Revision: https://reviews.llvm.org/D103292
I removed them in rG5de7467e982 but @thakis pointed out that
they were useful to keep, so here they are again. I've also converted
the `!isCoalescedWeak()` asserts into `!shouldOmitFromOutput()` asserts,
since the latter check subsumes the former.
Reviewed By: #lld-macho, thakis
Differential Revision: https://reviews.llvm.org/D104169
`outSecFileOff` and the associated `getFileOffset()` accessors were
unnecessary.
For all the cases we care about, `outSecFileOff` is the same as
`outSecOff`. The only time they deviate is if there are zerofill
sections within a given segment. But since zerofill sections are always
at the end of a segment, the only sections where the two values deviate
are zerofill sections themselves. And we never actually query the
outSecFileOff of zerofill sections.
As for `getFileOffset()`, the only place it was being used was to
calculate the offset of the entry symbol. However, we can compute that
value by just taking the difference between the address of the entry
symbol and the address of the Mach-O header. In fact, this appears to be
what ld64 itself does. This difference is the same as the file offset as
long as there are no intervening zerofill sections, but since `__text`
is the first section in `__TEXT`, this never happens, so our previous
use of `getFileOffset()` was not wrong -- just inefficient.
Reviewed By: #lld-macho, thakis
Differential Revision: https://reviews.llvm.org/D104177
D103977 broke a bunch of stuff as I had only tested the release build
which eliminated asserts.
I've retained the asserts where possible, but I also removed a bunch
instead of adding a whole lot of verbose ConcatInputSection casts.
Literal sections are not atomically live or dead. Rather,
liveness is tracked for each individual literal they contain. CStrings
have their liveness tracked via a `live` bit in StringPiece, and
fixed-width literals have theirs tracked via a BitVector.
The live-marking code now needs to track the offset within each section
that is to be marked live, in order to identify the literal at that
particular offset.
Numbers for linking chromium_framework on my 3.2 GHz 16-Core Intel Xeon W
with both `-dead_strip` and `--deduplicate-literals`, with and without this diff
applied:
```
N Min Max Median Avg Stddev
x 20 4.32 4.44 4.375 4.372 0.03105174
+ 20 4.3 4.39 4.36 4.3595 0.023277502
No difference proven at 95.0% confidence
```
This gives us size savings of about 0.4%.
Reviewed By: #lld-macho, thakis
Differential Revision: https://reviews.llvm.org/D103979
This is motivated by an upcoming diff in which the
WordLiteralInputSection ctor sets itself up based on the value of its
section flags. As such, it needs to be passed the `flags` value as part
of its ctor parameters, instead of having them assigned after the fact
in `parseSection()`. While refactoring code to make that possible, I
figured it would make sense for the other InputSections to also take
their initial values as ctor parameters.
Reviewed By: #lld-macho, thakis
Differential Revision: https://reviews.llvm.org/D103978
Conceptually, the implementation is pretty straightforward: we put each
literal value into a hashtable, and then write out the keys of that
hashtable at the end.
In contrast with ELF, the Mach-O format does not support variable-length
literals that aren't strings. Its literals are either 4, 8, or 16 bytes
in length. LLD-ELF dedups its literals via sorting + uniq'ing, but since
we don't need to worry about overly-long values, we should be able to do
a faster job by just hashing.
That said, the implementation right now is far from optimal, because we
add to those hashtables serially. To parallelize this, we'll need a
basic concurrent hashtable (only needs to support concurrent writes w/o
interleave reads), which shouldn't be to hard to implement, but I'd like
to punt on it for now.
Numbers for linking chromium_framework on my 3.2 GHz 16-Core Intel Xeon W:
N Min Max Median Avg Stddev
x 20 4.27 4.39 4.315 4.3225 0.033225703
+ 20 4.36 4.82 4.44 4.4845 0.13152846
Difference at 95.0% confidence
0.162 +/- 0.0613971
3.74783% +/- 1.42041%
(Student's t, pooled s = 0.0959262)
This corresponds to binary size savings of 2MB out of 335MB, or 0.6%.
It's not a great tradeoff as-is, but as mentioned our implementation can
be signficantly optimized, and literal dedup will unlock more
opportunities for ICF to identify identical structures that reference
the same literals.
Reviewed By: #lld-macho, gkm
Differential Revision: https://reviews.llvm.org/D103113
Not sure sectionType() carries its weight, but while we have it
we should use it consistently.
No behavior change.
Differential Revision: https://reviews.llvm.org/D104027
Our implementation draws heavily from LLD-ELF's, which in turn delegates
its string deduplication to llvm-mc's StringTableBuilder. The messiness of
this diff is largely due to the fact that we've previously assumed that
all InputSections get concatenated together to form the output. This is
no longer true with CStringInputSections, which split their contents into
StringPieces. StringPieces are much more lightweight than InputSections,
which is important as we create a lot of them. They may also overlap in
the output, which makes it possible for strings to be tail-merged. In
fact, the initial version of this diff implemented tail merging, but
I've dropped it for reasons I'll explain later.
**Alignment Issues**
Mergeable cstring literals are found under the `__TEXT,__cstring`
section. In contrast to ELF, which puts strings that need different
alignments into different sections, clang's Mach-O backend puts them all
in one section. Strings that need to be aligned have the `.p2align`
directive emitted before them, which simply translates into zero padding
in the object file.
I *think* ld64 extracts the desired per-string alignment from this data
by preserving each string's offset from the last section-aligned
address. I'm not entirely certain since it doesn't seem consistent about
doing this; but perhaps this can be chalked up to cases where ld64 has
to deduplicate strings with different offset/alignment combos -- it
seems to pick one of their alignments to preserve. This doesn't seem
correct in general; we can in fact can induce ld64 to produce a crashing
binary just by linking in an additional object file that only contains
cstrings and no code. See PR50563 for details.
Moreover, this scheme seems rather inefficient: since unaligned and
aligned strings are all put in the same section, which has a single
alignment value, it doesn't seem possible to tell whether a given string
doesn't have any alignment requirements. Preserving offset+alignments
for strings that don't need it is wasteful.
In practice, the crashes seen so far seem to stem from x86_64 SIMD
operations on cstrings. X86_64 requires SIMD accesses to be
16-byte-aligned. So for now, I'm thinking of just aligning all strings
to 16 bytes on x86_64. This is indeed wasteful, but implementation-wise
it's simpler than preserving per-string alignment+offsets. It also
avoids the aforementioned crash after deduplication of
differently-aligned strings. Finally, the overhead is not huge: using
16-byte alignment (vs no alignment) is only a 0.5% size overhead when
linking chromium_framework.
With these alignment requirements, it doesn't make sense to attempt tail
merging -- most strings will not be eligible since their overlaps aren't
likely to start at a 16-byte boundary. Tail-merging (with alignment) for
chromium_framework only improves size by 0.3%.
It's worth noting that LLD-ELF only does tail merging at `-O2`. By
default (at `-O1`), it just deduplicates w/o tail merging. @thakis has
also mentioned that they saw it regress compressed size in some cases
and therefore turned it off. `ld64` does not seem to do tail merging at
all.
**Performance Numbers**
CString deduplication reduces chromium_framework from 250MB to 242MB, or
about a 3.2% reduction.
Numbers for linking chromium_framework on my 3.2 GHz 16-Core Intel Xeon W:
N Min Max Median Avg Stddev
x 20 3.91 4.03 3.935 3.95 0.034641016
+ 20 3.99 4.14 4.015 4.0365 0.0492336
Difference at 95.0% confidence
0.0865 +/- 0.027245
2.18987% +/- 0.689746%
(Student's t, pooled s = 0.0425673)
As expected, cstring merging incurs some non-trivial overhead.
When passing `--no-literal-merge`, it seems that performance is the
same, i.e. the refactoring in this diff didn't cost us.
N Min Max Median Avg Stddev
x 20 3.91 4.03 3.935 3.95 0.034641016
+ 20 3.89 4.02 3.935 3.9435 0.043197831
No difference proven at 95.0% confidence
Reviewed By: #lld-macho, gkm
Differential Revision: https://reviews.llvm.org/D102964
This diff changes the type of the argument of isCodeSection to const InputSection *.
NFC.
Test plan: make check-lld-macho
Differential revision: https://reviews.llvm.org/D102664
Extend the range of calls beyond an architecture's limited branch range by first calling a thunk, which loads the far address into a scratch register (x16 on ARM64) and branches through it.
Other ports (COFF, ELF) use multiple passes with successively-refined guesses regarding the expansion of text-space imposed by thunk-space overhead. This MachO algorithm places thunks during MergedOutputSection::finalize() in a single pass using exact thunk-space overheads. Thunks are kept in a separate vector to avoid the overhead of inserting into the `inputs` vector of `MergedOutputSection`.
FIXME:
* arm64-stubs.s test is broken
* add thunk tests
* Handle thunks to DylibSymbol in MergedOutputSection::finalize()
Differential Revision: https://reviews.llvm.org/D100818
Before this, if an inline function was defined in several input files,
lld would write each copy of the inline function the output. With this
patch, it only writes one copy.
Reduces the size of Chromium Framework from 378MB to 345MB (compared
to 290MB linked with ld64, which also does dead-stripping, which we
don't do yet), and makes linking it faster:
N Min Max Median Avg Stddev
x 10 3.9957051 4.3496981 4.1411121 4.156837 0.10092097
+ 10 3.908154 4.169318 3.9712729 3.9846753 0.075773012
Difference at 95.0% confidence
-0.172162 +/- 0.083847
-4.14165% +/- 2.01709%
(Student's t, pooled s = 0.0892373)
Implementation-wise, when merging two weak symbols, this sets a
"canOmitFromOutput" on the InputSection belonging to the weak symbol not put in
the symbol table. We then don't write InputSections that have this set, as long
as they are not referenced from other symbols. (This happens e.g. for object
files that don't set .subsections_via_symbols or that use .alt_entry.)
Some restrictions:
- not yet done for bitcode inputs
- no "comdat" handling (`kindNoneGroupSubordinate*` in ld64) --
Frame Descriptor Entries (FDEs), Language Specific Data Areas (LSDAs)
(that is, catch block unwind information) and Personality Routines
associated with weak functions still not stripped. This is wasteful,
but harmless.
- However, this does strip weaks from __unwind_info (which is needed for
correctness and not just for size)
- This nopes out on InputSections that are referenced form more than
one symbol (eg from .alt_entry) for now
Things that work based on symbols Just Work:
- map files (change in MapFile.cpp is no-op and not needed; I just
found it a bit more explicit)
- exports
Things that work with inputSections need to explicitly check if
an inputSection is written (e.g. unwind info).
This patch is useful in itself, but it's also likely also a useful foundation
for dead_strip.
I used to have a "canoncialRepresentative" pointer on InputSection instead of
just the bool, which would be handy for ICF too. But I ended up not needing it
for this patch, so I removed that again for now.
Differential Revision: https://reviews.llvm.org/D102076
As preparation for a subsequent diff that implements builtin section renaming, define more `constexpr` strings in namespaces `lld::macho::segment_names` and `lld::macho::section_names`, and use them to replace string literals.
Differential Revision: https://reviews.llvm.org/D101393
Daniel Bertalan