Summary:
This patch eliminates some redundancy in iterating across Listeners for the
Instruction and Stall HWEvents, by introducing a template onEvent routine.
This change was suggested by @courbet in https://reviews.llvm.org/D48576. I
hope that this patch addresses that suggestion appropriately. I do like this
change better than what we had previously.
Reviewers: andreadb, courbet, RKSimon
Reviewed By: andreadb, courbet
Subscribers: javed.absar, tschuett, gbedwell, llvm-commits, courbet
Differential Revision: https://reviews.llvm.org/D48672
llvm-svn: 336916
Some programs (e.g. Linux) aren't able to handle function/data sections when
LTO is used. Thus they need a way to disable it. That can be done with these
plugin options:
-plugin-opt=-function-sections=0
-plugin-opt=-data-sections=0
llvm-svn: 336838
This makes easier to identify changes in the instruction info flags. It also
helps spotting potential regressions similar to the one recently introduced at
r336728.
Using the same character to mark MayLoad/MayStore/HasSideEffects is problematic
for llvm-lit. When pattern matching substrings, llvm-lit consumes tabs and
spaces. A change in position of the flag marker may not trigger a test failure.
This patch only changes the character used for flag `hasSideEffects`. The reason
why I didn't touch other flags is because I want to avoid spamming the mailing
because of the massive diff due to the numerous tests affected by this change.
In future, each instruction flag should be associated with a different character
in the Instruction Info View.
llvm-svn: 336797
When manually finishing the object writer in dsymutil, it's possible
that there are pending labels that haven't been resolved. This results
in an assertion when the assembler tries to fixup a label that doesn't
have an address yet.
Differential revision: https://reviews.llvm.org/D49131
llvm-svn: 336688
debug compilation dir when compiling assembly files with -g.
Part of PR38050.
Patch by Siddhartha Bagaria!
Differential Revision: https://reviews.llvm.org/D48988
llvm-svn: 336680
When implementing the DWARF accelerator tables in dsymutil I ran into an
assertion in the assembler. Debugging these kind of issues is a lot
easier when looking at the assembly instead of debugging the assembler
itself. Since it's only a matter of creating an AsmStreamer instead of a
MCObjectStreamer it made sense to turn this into a (hidden) dsymutil
feature.
Differential revision: https://reviews.llvm.org/D49079
llvm-svn: 336561
This is a short-term fix for PR38093.
For now, we llvm::report_fatal_error if the instruction builder finds an
unsupported instruction in the instruction stream.
We need to revisit this fix once we start addressing PR38101.
Essentially, we need a better framework for error handling.
llvm-svn: 336543
The reference implementation uses a case-insensitive string
comparison for strings of equal length. This will cause the
string "tEo" to compare less than "VUo". However we were using
a case sensitive comparison, which would generate the opposite
outcome. Switch to a case insensitive comparison. Also, when
one of the strings contains non-ascii characters, fallback to
a straight memcmp.
The only way to really test this is with a DIA test. Before this
patch, the test will fail (but succeed if link.exe is used instead
of lld-link). After the patch, it succeeds even with lld-link.
llvm-svn: 336464
This patch moves the construction of the default backend from llvm-mca.cpp and
into mca::Context. The Context class is responsible for holding ownership of
the simulated hardware components. These components are subclasses of
HardwareUnit. Right now the HardwareUnit is pretty bare-bones, but eventually
we might want to add some common functionality across all hardware components,
such as isReady() or something similar.
I have a feeling this patch will probably need some updates, but it's a start.
One thing I am not particularly fond of is the rather large interface for
createDefaultPipeline. That convenience routine takes a rather large set of
inputs from the llvm-mca driver, where many of those inputs are generated via
command line options.
One item I think we might want to change is the separating of ownership of
hardware components (owned by the context) and the pipeline (which owns
Stages). In short, a Pipeline owns Stages, a Context (currently) owns hardware.
The Pipeline's Stages make use of the components, and thus there is a lifetime
dependency generated. The components must outlive the pipeline. We could solve
this by having the Context also own the Pipeline, and not return a
unique_ptr<Pipeline>. Now that I think about it, I like that idea more.
Differential Revision: https://reviews.llvm.org/D48691
llvm-svn: 336456
This diff adds support for handling static libraries
to llvm-objcopy and llvm-strip.
Test plan: make check-all
Differential revision: https://reviews.llvm.org/D48413
llvm-svn: 336455
Suppress the diagnostic for mis-sized dbg.values when a value operand is
narrower than the unsigned variable it describes. Assume that a debugger
would implicitly zero-extend these values.
llvm-svn: 336452
When emitting a CU, store the MCSymbol pointing to the beginning of the
CU. We'll need this information later when emitting the .debug_names
section (DWARF5 accelerator table).
llvm-svn: 336433
This patch modifies the Scheduler heuristic used to select the next instruction
to issue to the pipelines.
The motivating example is test X86/BtVer2/add-sequence.s, for which llvm-mca
wrongly reported an estimated IPC of 1.50. According to perf, the actual IPC for
that test should have been ~2.00.
It turns out that an IPC of 2.00 for test add-sequence.s cannot possibly be
predicted by a Scheduler that only prioritizes instructions based on their
"age". A similar issue also affected test X86/BtVer2/dependent-pmuld-paddd.s,
for which llvm-mca wrongly estimated an IPC of 0.84 instead of an IPC of 1.00.
Instructions in the ReadyQueue are now ranked based on two factors:
- The "age" of an instruction.
- The number of unique users of writes associated with an instruction.
The new logic still prioritizes older instructions over younger instructions to
minimize the pressure on the reorder buffer. However, the number of users of an
instruction now also affects the overall rank. This potentially increases the
ability of the Scheduler to extract instruction level parallelism. This patch
fixes the problem with the wrong IPC reported for test add-sequence.s and test
dependent-pmuld-paddd.s.
llvm-svn: 336420
Summary:
Add support for two additional ObjC image info flags: `IS_SIMULATED` and
`HAS_CATEGORY_CLASS_PROPERTIES`.
`IS_SIMULATED` indicates a Mach-O binary built for iOS simulator.
`HAS_CATEGORY_CLASS_PROPERTIES` indicates a Mach-O binary built by a compiler
that supports class properties in categories.
Reviewers: enderby, compnerd
Reviewed By: compnerd
Subscribers: keith, llvm-commits
Differential Revision: https://reviews.llvm.org/D48568
llvm-svn: 336411
We add an option to dump the entire global / public symbol record
stream. Previously we would dump globals or publics, but not both.
And when we did dump them, we would always dump them in the order
they were referenced by the corresponding hash streams, not in
the order they were serialized in. This patch adds a lower level
mode that just dumps the whole stream in serialization order.
Additionally, when dumping global-extras, we now dump the hash
bitmap as well as the record offset instead of dumping all zeros
for the offsets.
llvm-svn: 336407
Summary:
Add support for two additional ObjC image info flags: `IS_SIMULATED` and
`HAS_CATEGORY_CLASS_PROPERTIES`.
`IS_SIMULATED` indicates a Mach-O binary built for iOS simulator.
`HAS_CATEGORY_CLASS_PROPERTIES` indicates a Mach-O binary built by a compiler
that supports class properties in categories.
Reviewers: enderby, compnerd
Reviewed By: compnerd
Subscribers: keith, llvm-commits
Differential Revision: https://reviews.llvm.org/D48568
llvm-svn: 336399
The target does just enough to be able to run llvm-exegesis in latency mode for
at least some opcodes.
Differential Revision: https://reviews.llvm.org/D48780
llvm-svn: 336187
The verifier identified several modules that were broken due to incorrect
linkage on declarations. To fix this, CompileOnDemandLayer2::extractFunction
has been updated to change decls to external linkage.
llvm-svn: 336150
Different CodeBlocks don't overlap. The same MCInst cannot appear in more than
one code block because all blocks are instantiated before the simulation is run.
We should always clear the content of map VariantDescriptors before every
simulation, since VariantDescriptors cannot possibly store useful information
for the next blocks. It is also "safer" to clear its content because `MCInst*`
is used as the key type for map VariantDescriptors.
llvm-svn: 336142
On darwin, all virtual sections have zerofill type, and having a
.zerofill directive in a non-virtual section is not allowed. Instead of
asserting, show a nicer error.
In order to use the equivalent of .zerofill in a non-virtual section,
the usage of .zero of .space is required.
This patch replaces the assert with an error.
Differential Revision: https://reviews.llvm.org/D48517
llvm-svn: 336127
Summary:
This adds a new -no-weak flag to nm to hide weak symbols in its output.
This also adds a -W alias for this which is analogous to -U.
Patch by Keith Smiley
Reviewers: kastiglione, enderby, compnerd
Reviewed By: kastiglione
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D48751
llvm-svn: 336126
Currently the llvm-exegesis native architecture is determined by comparing the
llvm native architecture with X86, so to add a new target would mean adding a
new check. Change this to building up a list of the targets llvm-exegesis
supports then using that, as this means that when adding a new target you just
add the target to the list of supported targets.
Differential Revision: https://reviews.llvm.org/D48778
llvm-svn: 336105
Currently the cycle counter is taken from the subtarget schedule model, which
isn't any use if the subtarget doesn't have one. Delegate the decision to the
target benchmark runner, as it may know better what to do in that case, with
the default being the current behaviour.
Differential Revision: https://reviews.llvm.org/D48779
llvm-svn: 336099
We were printing every character, even those that weren't printable. It
doesn't really make sense for this option.
The string content was sticked to its address, added two spaces in
between.
Differential Revision: https://reviews.llvm.org/D48271
llvm-svn: 336058
The original binary holder has an optimization where it caches a static
library (archive) between consecutive calls to GetObjects. However, the
actual memory buffer wasn't cached between calls.
This made sense when dsymutil was processing objects one after each
other, but when processing them in parallel, several binaries have to be
in memory at the same time. For this reason, every link context
contained a binary holder.
Having one binary holder per context is problematic, because the same
static archive was cached for every object file. Luckily, when the file
is mmap'ed, this was only costing us virtual memory.
This patch introduces a new BinaryHolder variant that is fully cached,
for all the object files it load, as well as the static archives. This
way, we don't have to give up on this optimization of bypassing the
file system.
Differential revision: https://reviews.llvm.org/D48501
llvm-svn: 335990
This simplifies the logic that updates RAW dependencies in the DispatchStage.
There is no advantage in storing that flag in the ReadDescriptor; we should
simply rely on the call to `STI.getReadAdvanceCycles()` to obtain the
ReadAdvance cycles. If there are no read-advance entries, then method
`getReadAdvanceCycles()` quickly returns 0.
No functional change intended.
llvm-svn: 335977
This change adds experimental support for SHT_RELR sections, proposed
here: https://groups.google.com/forum/#!topic/generic-abi/bX460iggiKg
Definitions for the new ELF section type and dynamic array tags, as well
as the encoding used in the new section are all under discussion and are
subject to change. Use with caution!
Author: rahulchaudhry
Differential Revision: https://reviews.llvm.org/D47919
llvm-svn: 335922
The checking logic should not treat artificial locations as being
somehow problematic. Producing these locations can be the desired
behavior of some passes.
See llvm.org/PR37961.
llvm-svn: 335897
This patch introduces a new class named WriteRef. A WriteRef is used by the
RegisterFile to keep track of register definitions. Internally it wraps a
WriteState, as well as the source index of the defining instruction.
This patch allows the tool to propagate additional information to support future
analysis on data dependencies.
llvm-svn: 335867
Rather than calling std::find in a loop, just sort the vector and remove
duplicate entries at the end of the function.
Also, move the debug print at the end of the function, and query the
MCRegisterInfo to print register names rather than physreg IDs.
No functional change intended.
llvm-svn: 335837
Summary:
This enables the X86-specific X86FloatingPointStackifierPass, and allow
llvm-exegesis to generate and measure X87 latency/uops for some FP ops.
Reviewers: gchatelet
Subscribers: tschuett, llvm-commits
Differential Revision: https://reviews.llvm.org/D48592
llvm-svn: 335815
This patch splits off some abstractions used by dsymutil's dwarf linker
and moves them into separate header and implementation files. This
almost halves the number of LOC in DwarfLinker.cpp and makes it a lot
easier to understand what functionality lives where.
Differential revision: https://reviews.llvm.org/D48647
llvm-svn: 335749
Summary:
This patch removes a few callbacks from Pipeline. It comes at the cost of
registering Listeners with all Stages. Not all stages need listeners or issue
callbacks, this registration is a bit redundant. However, as we build-out the
API, this redundancy can disappear.
The main purpose here is to move callback code from the Pipeline and into the
stages that actually issue those callbacks. This removes the back-pointer to
the Pipeline that was put into a few Stage subclasses.
Reviewers: andreadb, courbet, RKSimon
Reviewed By: andreadb, courbet
Subscribers: tschuett, gbedwell, llvm-commits
Differential Revision: https://reviews.llvm.org/D48576
llvm-svn: 335748
When promoting instructions from the wait queue to the ready queue, we should
check if an instruction has already reached the IS_READY state before
calling method update().
llvm-svn: 335722
It's not possible to get the fragment size of some dbg.values. Teach the
mis-sized dbg.value diagnostic to detect this scenario and bail out.
Tested with:
$ find test/Transforms -print -exec opt -debugify-each -instcombine {} \;
llvm-svn: 335695
Report an error in -check-debugify when the size of a dbg.value operand
doesn't match up with the size of the variable it describes.
Eventually this check should be moved into the IR verifier. For the
moment, it's useful to include the check in -check-debugify as a means
of catching regressions and finding existing bugs.
Here are some instances of bugs the new check finds in the -O2 pipeline
(all in InstCombine):
1) A float is used where a double is expected:
ERROR: dbg.value operand has size 32, but its variable has size 64:
call void @llvm.dbg.value(metadata float %expf, metadata !12, metadata
!DIExpression()), !dbg !15
2) An i8 is used where an i32 is expected:
ERROR: dbg.value operand has size 8, but its variable has size 32:
call void @llvm.dbg.value(metadata i8 %t4, metadata !14, metadata
!DIExpression()), !dbg !24
3) A <4 x i32> is used where something twice as large is expected
(perhaps a <4 x i64>, I haven't double-checked):
ERROR: dbg.value operand has size 128, but its variable has size 256:
call void @llvm.dbg.value(metadata <4 x i32> %4, metadata !40, metadata
!DIExpression()), !dbg !95
Differential Revision: https://reviews.llvm.org/D48408
llvm-svn: 335682
LLJIT is a prefabricated ORC based JIT class that is meant to be the go-to
replacement for MCJIT. Unlike OrcMCJITReplacement (which will continue to be
supported) it is not API or bug-for-bug compatible, but targets the same
use cases: Simple, non-lazy compilation and execution of LLVM IR.
LLLazyJIT extends LLJIT with support for function-at-a-time lazy compilation,
similar to what was provided by LLVM's original (now long deprecated) JIT APIs.
This commit also contains some simple utility classes (CtorDtorRunner2,
LocalCXXRuntimeOverrides2, JITTargetMachineBuilder) to support LLJIT and
LLLazyJIT.
Both of these classes are works in progress. Feedback from JIT clients is very
welcome!
llvm-svn: 335670
When checking the debug info in a module, don't treat a missing
dbg.value as an error. The dbg.value may simply have been DCE'd, in
which case the debugger has enough information to display the variable
as <optimized out>.
llvm-svn: 335647
Summary:
Adds assembly parsing support for the module summary index (follow on
to r333335 which added the assembly writing support).
I added support to llvm-as to invoke the index parsing, so that it can
create either a bitcode file with a Module and a per-module index, or
a combined index without a Module.
I will send follow on patches soon to do the following:
- add support to tools such as llvm-lto2 to parse the per-module indexes
from assembly instead of bitcode when testing the thin link.
- verification support.
Depends on D47844 and D47842.
Reviewers: pcc, dexonsmith, mehdi_amini
Subscribers: inglorion, eraman, steven_wu, llvm-commits
Differential Revision: https://reviews.llvm.org/D47905
llvm-svn: 335602
Summary:
This allows targets to override code generation for some instructions.
As an example of override, this also moves ad-hoc instruction filtering
for X86 into the X86 ExegesisTarget.
Reviewers: gchatelet
Subscribers: mgorny, tschuett, llvm-commits
Differential Revision: https://reviews.llvm.org/D48587
llvm-svn: 335582
Summary:
This change renames the Backend and BackendPrinter to Pipeline and PipelinePrinter respectively.
Variables and comments have also been updated to reflect this change.
The reason for this rename, is to be slightly more correct about what MCA is modeling. MCA models a Pipeline, which implies some logical sequence of stages.
Reviewers: andreadb, courbet, RKSimon
Reviewed By: andreadb, courbet
Subscribers: mgorny, javed.absar, tschuett, gbedwell, llvm-commits
Differential Revision: https://reviews.llvm.org/D48496
llvm-svn: 335496
Summary:
This ensures that the snippet always sees the same values for registers,
making measurements reproducible.
This will also allow exploring different values.
Reviewers: gchatelet
Subscribers: tschuett, llvm-commits
Differential Revision: https://reviews.llvm.org/D48542
llvm-svn: 335465
The DispatchUnit is no longer a dependency of RCU, so this patch removes a
stale include and forward decl. This patch also cleans up some comments.
llvm-svn: 335392
Summary:
Remove explicit stages and introduce a list of stages.
A pipeline should be composed of an arbitrary list of stages, and not any
predefined list of stages in the Backend. The Backend should not know of any
particular stage, rather it should only be concerned that it has a list of
stages, and that those stages will fulfill the contract of what it means to be
a Stage (namely pre/post/execute a given instruction).
For now, we leave the original set of stages defined in the Backend ctor;
however, I imagine these will be moved out at a later time.
This patch makes an adjustment to the semantics of Stage::isReady.
Specifically, what the Backend really needs to know is if a Stage has
unfinished work. With that said, it is more appropriately renamed
Stage::hasWorkToComplete(). This change will clean up the check in
Backend::run(), allowing us to query each stage to see if there is unfinished
work, regardless of what subclass a stage might be. I feel that this change
simplifies the semantics too, but that's a subjective statement.
Given how RetireStage and ExecuteStage handle data in their preExecute(), I've
had to change the order of Retire and Execute in our stage list. Retire must
complete any of its preExecute actions before ExecuteStage's preExecute can
take control. This is mainly because both stages utilize the RCU. In the
meantime, I want to see if I can adjust that or remove that coupling.
Reviewers: andreadb, RKSimon, courbet
Reviewed By: andreadb
Subscribers: tschuett, gbedwell, llvm-commits
Differential Revision: https://reviews.llvm.org/D46907
llvm-svn: 335361
After the recent refactoring that introduced parallel handling of
different object, the binary holder became unique per object file. This
defeats its optimization of caching archives, leading to an archive
being opened for every binary it contains. This is obviously unfortunate
and will need to be refactored soon.
Luckily in practice, the impact of this is limited as most files are
mmap'ed instead of memcopy'd. There's a caveat however: when the memory
buffer requires a null terminator and it's a multiple of the page size,
we allocate instead of mmap'ing. If this happens for a static archive,
we end up with N copies of it in memory, where N is the number of
objects in the archive, leading to exuberant memory usage. This provided
a stopgap solution to ensure that all the files it loads are mmap in
memory by removing the requirement for a terminating null byte.
Differential revision: https://reviews.llvm.org/D48397
llvm-svn: 335293
Summary: Pretty much everything we need is in llvm::TargetMachine.
Reviewers: gchatelet
Subscribers: llvm-commits, tschuett
Differential Revision: https://reviews.llvm.org/D48428
llvm-svn: 335237
Errors found processing the DW_AT_ranges attribute are propagated by lower level
routines and reported by their callers.
Reviewer: JDevlieghere
Differential Revision: https://reviews.llvm.org/D48344
llvm-svn: 335188
This patch teaches llvm-mca how to identify register writes that implicitly zero
the upper portion of a super-register.
On X86-64, a general purpose register is implemented in hardware as a 64-bit
register. Quoting the Intel 64 Software Developer's Manual: "an update to the
lower 32 bits of a 64 bit integer register is architecturally defined to zero
extend the upper 32 bits". Also, a write to an XMM register performed by an AVX
instruction implicitly zeroes the upper 128 bits of the aliasing YMM register.
This patch adds a new method named clearsSuperRegisters to the MCInstrAnalysis
interface to help identify instructions that implicitly clear the upper portion
of a super-register. The rest of the patch teaches llvm-mca how to use that new
method to obtain the information, and update the register dependencies
accordingly.
I compared the kernels from tests clear-super-register-1.s and
clear-super-register-2.s against the output from perf on btver2. Previously
there was a large discrepancy between the estimated IPC and the measured IPC.
Now the differences are mostly in the noise.
Differential Revision: https://reviews.llvm.org/D48225
llvm-svn: 335113
Summary: Introducing a Prototype object to capture Variables that must be set but keeps degrees of freedom as Invalid. This allows exploring non constraint variables later on.
Reviewers: courbet
Subscribers: tschuett, llvm-commits
Differential Revision: https://reviews.llvm.org/D48316
llvm-svn: 335105
Summary: This is a step towards implementing memory operands and X87.
Reviewers: gchatelet
Subscribers: mgorny, tschuett, llvm-commits
Differential Revision: https://reviews.llvm.org/D48210
llvm-svn: 335038
Summary:
While that is indeed a quite interesting summary stat,
there are cases where it does not really add anything
other than consuming extra lines.
Declutters the output of D48190.
Reviewers: RKSimon, andreadb, courbet, craig.topper
Reviewed By: andreadb
Subscribers: javed.absar, gbedwell, llvm-commits
Differential Revision: https://reviews.llvm.org/D48209
llvm-svn: 334833
Summary:
On hover, the whole asm snippet is displayed, including operands.
This requires the actual assembly output instead of just the MCInsts:
This is because some pseudo-instructions get lowered to actual target
instructions during codegen (e.g. ABS_Fp32 -> SSE or X87).
Reviewers: gchatelet
Subscribers: mgorny, tschuett, llvm-commits
Differential Revision: https://reviews.llvm.org/D48164
llvm-svn: 334805
Summary:
Get rid of OpcodeName.
To remove the opcode name from an old file:
```
cat old_file | sed '/opcode_name.*/d'
```
Reviewers: gchatelet
Subscribers: tschuett, llvm-commits
Differential Revision: https://reviews.llvm.org/D48121
llvm-svn: 334691
Summary: This patch transforms the Scheduler class into the ExecuteStage. Most of the logic remains.
Reviewers: andreadb, RKSimon, courbet
Reviewed By: andreadb
Subscribers: mgorny, javed.absar, tschuett, gbedwell, llvm-commits
Differential Revision: https://reviews.llvm.org/D47246
llvm-svn: 334679
Fixes PR37790.
In some (very rare) cases, the LSUnit (Load/Store unit) was wrongly marking a
load (or store) as "ready to execute" effectively bypassing older memory barrier
instructions.
To reproduce this bug, the memory barrier must be the first instruction in the
input assembly sequence, and it doesn't have to perform any register writes.
llvm-svn: 334633
Not sure why, but it breaks buildbot clang-cmake-armv8-full.
It causes a failure in TEST 'Xray-armhf-linux :: TestCases/Posix/profiling-single-threaded.cc'.
llvm-svn: 334617
Summary: Previous design was relying on the 'mutate' keyword and was quite confusing. This version separate mutable from immutable data and makes it clearer what changes and what doesn't.
Reviewers: courbet
Subscribers: tschuett, llvm-commits
Differential Revision: https://reviews.llvm.org/D48020
llvm-svn: 334596
Summary:
This method was not correct for entries in DWO files as it assumed it
could just add up the CU and DIE offsets to get the absolute DIE offset.
This is not correct for the DWO files, as here the CU offset will
reference the skeleton unit, whereas the DIE offset will be the offset
in the full unit in the DWO file.
Unfortunately, this means that we are not able to determine the absolute
DIE offset using the information in the .debug_names section alone,
which means we have to offload some of this work to the users of this
class.
To demonstrate how this can be done, I've added/fixed the ability to
lookup entries using accelerator tables in DWO files in llvm-dwarfdump.
To make this happen, I've needed to make two extra changes in other
classes:
- made the DWARFContext method to lookup a CU based on the section
offset public. I've needed this functionality to lookup a CU, and this
seems like a useful thing in general.
- made DWARFUnit::getDWOId call extractDIEsIfNeeded. Before this, the
DWOId was filled in only if the root DIE happened to be parsed
before we called the accessor. Since the lazy parsing is supposed to
happen under the hood, calling extractDIEsIfNeeded seems appropriate.
Reviewers: JDevlieghere, aprantl, dblaikie
Subscribers: mgrang, llvm-commits
Differential Revision: https://reviews.llvm.org/D48009
llvm-svn: 334578
This simplifies some code which had StringRefs to begin with, and
makes other code more complicated which had const char* to begin
with.
In the end, I think this makes for a more idiomatic and platform
agnostic API. Not all platforms launch process with null terminated
c-string arrays for the environment pointer and argv, but the api
was designed that way because it allowed easy pass-through for
posix-based platforms. There's a little additional overhead now
since on posix based platforms we'll be takign StringRefs which
were constructed from null terminated strings and then copying
them to null terminate them again, but from a readability and
usability standpoint of the API user, I think this API signature
is strictly better.
llvm-svn: 334518
Don't provide the assembler source as the "root file" unless the user
asked to have debug info for the assembler source (with -g).
If the source doesn't provide an explicit ".file 0" then (a) use the
compilation directory as directory #0, and (b) use the file #1 info
for file #0 also.
Differential Revision: https://reviews.llvm.org/D48055
llvm-svn: 334512
Summary: This patch moves linking of libpfm from different places to a single one.
Reviewers: courbet
Subscribers: mgorny, tschuett, llvm-commits
Differential Revision: https://reviews.llvm.org/D48075
llvm-svn: 334499
Name table occupies a big chunk of size in current binary format sample profile.
In order to reduce its size, the patch changes the sample writer/reader to
save/restore MD5Hash of names in the name table. Sample annotation phase will
also use MD5Hash of name to query samples accordingly.
Experiment shows compact binary format can reduce the size of sample profile by
2/3 compared with binary format generally.
Differential Revision: https://reviews.llvm.org/D47955
llvm-svn: 334447
NFC here, this just raises some platform specific ifdef hackery
out of a class and creates proper platform-independent typedefs
for the relevant things. This allows these typedefs to be
reused in other places without having to reinvent this preprocessor
logic.
llvm-svn: 334294
I don't know how to build this code, but based on the failing
buildbot error message it looks like this change should get
the buildbot up and running again.
llvm-svn: 334231
This breaks the OpenFlags enumeration into two separate
enumerations: OpenFlags and CreationDisposition. The first
controls the behavior of the API depending on whether or not
the target file already exists, and is not a flags-based
enum. The second controls more flags-like values.
This yields a more easy to understand API, while also allowing
flags to be passed to the openForRead api, where most of the
values didn't make sense before. This also makes the apis more
testable as it becomes easy to enumerate all the configurations
which make sense, so I've added many new tests to exercise all
the different values.
llvm-svn: 334221
The class Object contains std::shared_ptr<MemoryBuffer> OwnedData
which is not used anywhere. Besides avoiding two stage initialization
the motivation to remove it comes from the plan to add (currently missing) support
for static libraries.
NFC.
Test plan: make check-all
Differential revision: https://reviews.llvm.org/D47855
llvm-svn: 334217
Summary: BenchmarkRunner subclasses can now create many configurations - although this patch still generates one.
Reviewers: courbet
Subscribers: tschuett, llvm-commits
Differential Revision: https://reviews.llvm.org/D47877
llvm-svn: 334197
Summary: This is the first step to have the BenchmarkRunner create and measure many different configurations (different initial values for instance).
Reviewers: courbet
Subscribers: tschuett, llvm-commits
Differential Revision: https://reviews.llvm.org/D47826
llvm-svn: 334169
Summary: BenchmarkResult IO functions now return an Error or Expected so caller can deal take proper action.
Reviewers: courbet
Subscribers: tschuett, llvm-commits
Differential Revision: https://reviews.llvm.org/D47868
llvm-svn: 334167
With '-elf-output-style=GNU -relocations', a header containing the number
of entries is printed before all the relocation entries in the section.
For Android packed format, we need to perform the unpacking first before
we can get the actual number of relocations in the section.
Patch by Rahul Chaudhry!
Differential Revision: https://reviews.llvm.org/D47800
llvm-svn: 334147
With the upcoming patch to add summary parsing support, IsAnalysis would
be true in contexts where we are not performing module summary analysis.
Rename to the more specific and approprate HaveGVs, which is essentially
what this flag is indicating.
llvm-svn: 334140
Before this patch, debugify would insert debug value intrinsics before the
terminating instruction in a block. This had the advantage of being simple,
but was a bit too simple/unrealistic.
This patch teaches debugify to insert debug values immediately after their
operand defs. This enables better testing of the compiler.
For example, with this patch, `opt -debugify-each` is able to identify a
vectorizer DI-invariance bug fixed in llvm.org/PR32761. In this bug, the
vectorizer produced different output with/without debug info present.
Reverting Davide's bugfix locally, I see:
$ ~/scripts/opt-check-dbg-invar.sh ./bin/opt \
.../SLPVectorizer/AArch64/spillcost-di.ll -slp-vectorizer
Comparing: -slp-vectorizer .../SLPVectorizer/AArch64/spillcost-di.ll
Baseline: /var/folders/j8/t4w0bp8j6x1g6fpghkcb4sjm0000gp/T/tmp.iYYeL1kf
With DI : /var/folders/j8/t4w0bp8j6x1g6fpghkcb4sjm0000gp/T/tmp.sQtQSeet
9,11c9,11
< %5 = getelementptr inbounds %0, %0* %2, i64 %0, i32 1
< %6 = bitcast i64* %4 to <2 x i64>*
< %7 = load <2 x i64>, <2 x i64>* %6, align 8, !tbaa !0
---
> %5 = load i64, i64* %4, align 8, !tbaa !0
> %6 = getelementptr inbounds %0, %0* %2, i64 %0, i32 1
> %7 = load i64, i64* %6, align 8, !tbaa !5
12a13
> store i64 %5, i64* %8, align 8, !tbaa !0
14,15c15
< %10 = bitcast i64* %8 to <2 x i64>*
< store <2 x i64> %7, <2 x i64>* %10, align 8, !tbaa !0
---
> store i64 %7, i64* %9, align 8, !tbaa !5
:: Found a test case ^
Running this over the *.ll files in tree, I found four additional examples
which compile differently with/without DI present. I plan on filing bugs for
these.
llvm-svn: 334118
Moves the Mode field out of the Key. The existing yaml benchmark results can be fixed with the following script:
```
readonly FILE=$1
readonly MODE=latency # Change to uops to fix a uops benchmark.
cat $FILE | \
sed "/^\ \+mode:\ \+$MODE$/d" | \
sed "/^cpu_name.*$/i mode: $MODE"
```
Differential Revision: https://reviews.llvm.org/D47813
Authored by: Guillaume Chatelet
llvm-svn: 334079
This patch fixe the logic in ReadState::cycleEvent(). That method was not
correctly updating field `TotalCycles`.
Added extra code comments in class ReadState to better describe each field.
llvm-svn: 334028
We want llvm-exegesis to explore instructions (effect of initial register values, effect of operand selection). To enable this a BenchmarkResult muststore all the relevant data in its key. This patch starts adding such data. Here we simply allow to store the generated instructions, following patches will add operands and initial values for registers.
https://reviews.llvm.org/D47764
Authored by: Guilluame Chatelet
llvm-svn: 334008
The -check-debugify pass should preserve all analyses. Otherwise, it may
invalidate an optional analysis and inadvertently alter codegen.
The test case is reduced from deopt-bundle.ll. The result of `opt -O1`
on this file would differ when -debugify-each was toggled. That happened
because CheckDebugify failed to preserve GlobalsAA.
Thanks to Davide Italiano for his help chasing this down!
llvm-svn: 333959
Review feedback from r328165. Split out just the one function from the
file that's used by Analysis. (As chandlerc pointed out, the original
change only moved the header and not the implementation anyway - which
was fine for the one function that was used (since it's a
template/inlined in the header) but not in general)
llvm-svn: 333954
Summary:
These tools failed for a very large bitcode file produced by LTO due to
64-bit values being assigned to 32-bit types. For the BitstreamReader.h
fix, the value initially fit into the 32-bit unsigned, but there was an
overflow when multiplying by 32 furter below to compute the bit offset.
No test case in the patch as this requires a huge bitcode file.
Reviewers: pcc, george.karpenkov
Subscribers: mehdi_amini, a.sidorin, llvm-commits
Differential Revision: https://reviews.llvm.org/D47731
llvm-svn: 333942
This patch is the last of a sequence of three patches related to LLVM-dev RFC
"MC support for variant scheduling classes".
http://lists.llvm.org/pipermail/llvm-dev/2018-May/123181.html
This fixes PR36672.
The main goal of this patch is to teach llvm-mca how to solve variant scheduling
classes. This patch does that, plus it adds new variant scheduling classes to
the BtVer2 scheduling model to identify so-called zero-idioms (i.e. so-called
dependency breaking instructions that are known to generate zero, and that are
optimized out in hardware at register renaming stage).
Without the BtVer2 change, this patch would not have had any meaningful tests.
This patch is effectively the union of two changes:
1) a change that teaches llvm-mca how to resolve variant scheduling classes.
2) a change to the BtVer2 scheduling model that allows us to special-case
packed XOR zero-idioms (this partially fixes PR36671).
Differential Revision: https://reviews.llvm.org/D47374
llvm-svn: 333909
Resubmit of r333424. This version contains the fix for fails found by buildbots
on some targets.
This patch allows parsing GNU_PROPERTY_X86_FEATURE_1_AND
notes in .note.gnu.property sections. These notes
indicate that the object file is built to support Intel CET.
patch by mike.dvoretsky
Differential Revision: https://reviews.llvm.org/D47473
llvm-svn: 333908
This is required if we want to correctly match the behavior of method
SubtargetEmitter::ExpandProcResource() in Tablegen. When computing the set of
"consumed" processor resources and resource cycles, the logic in
ExpandProcResource() doesn't update the number of resource cycles contributed by
a "Super" resource to a group. We need to take this into account when a model
declares a processor resource which is part of a 'processor resource group', and
it is also used as the "Super" of other resources.
llvm-svn: 333892
Summary:
We now highlight any sched classes whose measurements do not match the
LLVM SchedModel. "bad" clusters are marked in red.
Screenshot in phabricator diff.
Reviewers: gchatelet
Subscribers: tschuett, mgrang, RKSimon, llvm-commits
Differential Revision: https://reviews.llvm.org/D47639
llvm-svn: 333884
After r333856, opt -debugify would just stop emitting debug value
intrinsics after encountering a musttail call. This wasn't sufficient to
avoid verifier failures.
Debug value intrinicss for all instructions preceding a musttail call
must also be emitted before the musttail call.
llvm-svn: 333866
Applying synthetic debug info before the bitcode writer pass has no
testing-related purpose. This commit prevents that from happening.
It also adds tests which check that IR produced with/without
-debugify-each enabled is identical after stripping. This makes it
possible to check that individual passes (or full pipelines) are
invariant to debug info.
llvm-svn: 333861
The -strip-module-flags option strips llvm.module.flags metadata from a
module at the beginning of the opt pipeline.
This will be used to test whether the output of a pass is debug info
(DI) invariant.
E.g, after applying synthetic debug info to a test case, we'd like to
strip out all DI-related metadata and check that the final IR is
identical to a baseline file without any DI applied, to check that
optimizations aren't inhibited by debug info.
llvm-svn: 333860
Object FIle Representation
At codegen time this is emitted into the ELF file a pair of symbol indices and a weight. In assembly it looks like:
.cg_profile a, b, 32
.cg_profile freq, a, 11
.cg_profile freq, b, 20
When writing an ELF file these are put into a SHT_LLVM_CALL_GRAPH_PROFILE (0x6fff4c02) section as (uint32_t, uint32_t, uint64_t) tuples as (from symbol index, to symbol index, weight).
Differential Revision: https://reviews.llvm.org/D44965
llvm-svn: 333823
This fixes the bug where strip-all option was
leading to a malformed outputted ELF file.
Differential Revision: https://reviews.llvm.org/D47414
llvm-svn: 333772
Summary:
"Unknown" for platforms that were not manually added into the switch
did not make sense at all. Now it prints Target + addend for all
elf-machines that were not explicitly mentioned.
Addresses PR21059 and PR25124.
Original author: fedor.sergeev
Reviewers: jyknight, espindola, fedor.sergeev
Reviewed By: jyknight
Subscribers: eraman, dcederman, jfb, dschuff, aheejin, llvm-commits
Differential Revision: https://reviews.llvm.org/D36464
llvm-svn: 333726
This diff implements the option -o
for specifying a file to write the output to.
Test plan: make check-all
Differential revision: https://reviews.llvm.org/D47505
llvm-svn: 333693
The lambda functions used by method ResourceManager::mustIssueImmediately() was
incorrectly truncating masks of buffered processor resources to 32-bit quantities.
The invalid mask values were then used to access a map of processor
resource descriptors.
Fixes PR37643.
llvm-svn: 333692
As noted by Adrian on llvm-commits, PrintHTMLEscaped and PrintEscaped in
StringExtras did not conform to the LLVM coding guidelines. This commit
rectifies that.
llvm-svn: 333669
Summary:
Both (Apple and DWARF5) implementations of the iterators had bugs which
resulted in crashes if one attempted to iterate through the accelerator
tables all the way.
For the Apple tables, the issue was that we did not clear the DataOffset
field when we reached the end, which made our iterator compare unequal
to the "end" iterator. For the Dwarf5 tables, the problem was that we
incremented the CurrentIndex pointer and then used the incremented
(possibly invalid) pointer to check whether we have reached the end of
the index list.
The reason these bugs went undetected is because their only user
(dwarfdump) only ever searched for the first match. Besides allowing us
to test this fix, changing llvm-dwarfdump --find to display all matches
seems like a good improvement (it makes the behavior consistent with the
--name option), so I change llvm-dwarfdump to do that.
The existing tests would be sufficient to test this fix with the new
llvm-dwarfdump behavior, but I add a special test that demonstrates that
the tool indeed displays multiple results. The find.test test needed to
be tweaked a bit as the tool now does not print the ".debug_info
contents" header (also consistent with how --name works).
Reviewers: JDevlieghere, aprantl, dblaikie
Subscribers: mgrang, llvm-commits
Differential Revision: https://reviews.llvm.org/D47543
llvm-svn: 333635
Summary:
I'm slowly looking into a new X86 scheduler model,
for AMD Bulldozer CPU, model 2 (bdver2, Piledriver).
And naturally, i have hit that assert :)
I happened to know what it meant, and how to fix it,
but that is not too common knowledge.
Reviewers: courbet, RKSimon
Reviewed By: courbet
Subscribers: tschuett, llvm-commits, craig.topper
Differential Revision: https://reviews.llvm.org/D47572
llvm-svn: 333632
Per discussion on the generic-abi mailing list:
https://groups.google.com/forum/#!topic/generic-abi/MPr8TVtnVn4
An object file manipulation tool must either write out a symbol
table with the same number of entries as the original symbol table
and in the same order, or if this is impossible, refuse to operate
on the object file if it has unrecognized sections that are linked
to the symtab section. However, existing tools (namely GNU strip,
GNU objcopy and ld.{bfd,gold,lld} -r) do not comply with this at
present: they change symbol table indexes and set sh_link to 0 on
the unrecognized symtab-linked sections.
We intend to use the latter as a (temporary) signal that a tool has
operated on a proposed new symtab-linked section and invalidated the
symbol table indexes. However, llvm-objcopy currently keeps sh_link
pointing to the new symtab section. This patch changes llvm-objcopy
to set sh_link to 0 to match the behaviour of the other tools.
Differential Revision: https://reviews.llvm.org/D47404
llvm-svn: 333581
When printing string in the Plist, we weren't escaping the characters
which lead to invalid XML. This patch adds the escape logic to
StringExtras.
rdar://39785334
llvm-svn: 333565
Previously JITCompileCallbackManager only supported single threaded code. This
patch embeds a VSO (see include/llvm/ExecutionEngine/Orc/Core.h) in the callback
manager. The VSO ensures that the compile callback is only executed once and that
the resulting address cached for use by subsequent re-entries.
llvm-svn: 333490
This patch allows parsing GNU_PROPERTY_X86_FEATURE_1_AND
notes in .note.gnu.property sections. These notes
indicate that the object file is built to support Intel CET.
patch by mike.dvoretsky
Differential Revision: https://reviews.llvm.org/D47473
llvm-svn: 333424
Summary:
Implements AsmWriter support for printing the module summary index to
assembly with the format discussed in the RFC "LLVM Assembly format for
ThinLTO Summary".
Implements just enough of the parsing support to recognize and ignore
the summary entries. As agreed in the RFC thread, this will be the
behavior when assembling the IR. A follow on change will implement
parsing/assembling of the summary entries for use by tools that
currently build the summary index from bitcode.
Reviewers: dexonsmith, pcc
Subscribers: inglorion, eraman, steven_wu, dblaikie, llvm-commits
Differential Revision: https://reviews.llvm.org/D46699
llvm-svn: 333335
Summary:
This class maintains the same logic as the original RetireControlUnit.
This is just an intermediate patch to make the RCU a Stage. Future patches will remove the dependency on the DispatchStage, and then more properly populate the pre/execute/post Stage interface.
Reviewers: andreadb, RKSimon, courbet
Reviewed By: andreadb, courbet
Subscribers: javed.absar, mgorny, tschuett, gbedwell, llvm-commits
Differential Revision: https://reviews.llvm.org/D47244
llvm-svn: 333292
Setting the "Debug Info Version" module flag makes it possible to pipe
synthetic debug info into llc, which is useful for testing backends.
llvm-svn: 333237
Before printing the block reciprocal throughput, ensure that the floating point
number is always rounded the same way on every target.
No functional change intended.
llvm-svn: 333210
This commit adds a color category so tools can document this option and
enables it for dwarfdump and dsymuttil.
rdar://problem/40498996
llvm-svn: 333176
This is a small follow-up to the revisions r333117 and r331663.
1. Avoid the name conflicts of the generated variables for prefixes.
2. Apply clang-format -i -style=llvm to llvm-objcopy.cpp once again.
3. Add a test for the flag with double dash.
Test plan: make check-all
llvm-svn: 333120
This patch implements the "block reciprocal throughput" computation in the
SummaryView.
The block reciprocal throughput is computed as the MAX of:
- NumMicroOps / DispatchWidth
- Resource Cycles / #Units (for every resource consumed).
The block throughput is bounded from above by the hardware dispatch throughput.
That is because the DispatchWidth is an upper bound on how many opcodes can be part
of a single dispatch group.
The block throughput is also limited by the amount of hardware parallelism. The
number of available resource units affects how the resource pressure is
distributed, and also how many blocks can be delivered every cycle.
llvm-svn: 333095
Summary:
This is an intermediate change, it moves the non-notification logic from
Backend::notifyCycleBegin to runCycle().
Once the scheduler becomes part of the Execution stage
the explicit call to Scheduler::cycleEvent will disappear.
The logic for Dispatch::cycleEvent() can be in
the preExecute phase, which this patch addresses.
Reviewers: andreadb, RKSimon, courbet
Reviewed By: andreadb
Subscribers: tschuett, gbedwell, llvm-commits
Differential Revision: https://reviews.llvm.org/D47213
llvm-svn: 333029
If one runs llvm-objcopy --strip-all --keep-symbol foo
and the symbol table indeed contains the symbol "foo"
then it should not be removed.
Test plan: make check-all
Differential revision: https://reviews.llvm.org/D47052
llvm-svn: 333008
Summary: This makes the report much more readable.
Reviewers: gchatelet
Subscribers: tschuett, mgrang, craig.topper, RKSimon, llvm-commits
Differential Revision: https://reviews.llvm.org/D47189
llvm-svn: 332979
Also clean up a couple of hacks where we were writing the section
contents to another stream by setting the object writer's stream,
writing and setting it back.
Part of PR37466.
Differential Revision: https://reviews.llvm.org/D47038
llvm-svn: 332858
Summary: Add LLVMDIBuilderCreateObjCIVar, LLVMDIBuilderCreateObjCProperty, and LLVMDIBuilderCreateInheritance to allow declaring metadata for Objective-C class hierarchies and their associated properties and instance variables.
Reviewers: whitequark, deadalnix
Reviewed By: whitequark
Subscribers: harlanhaskins, llvm-commits
Differential Revision: https://reviews.llvm.org/D47123
llvm-svn: 332850
Change the "recoverable" error callback to take an Error instaed of a
string.
Reviewed by: JDevlieghere
Differential Revision: https://reviews.llvm.org/D46831
llvm-svn: 332845
Summary: Add wrappers for a module's alias iterators and a getter and setter for the aliasee value.
Reviewers: whitequark, deadalnix
Reviewed By: whitequark
Subscribers: llvm-commits, harlanhaskins
Differential Revision: https://reviews.llvm.org/D46808
llvm-svn: 332826
Summary:
- Provide LLVMGetValueName2 and LLVMSetValueName2 that return and take the length of the provided C string respectively
- Deprecate LLVMGetValueName and LLVMSetValueName
Reviewers: whitequark, deadalnix
Reviewed By: whitequark
Subscribers: llvm-commits, harlanhaskins
Differential Revision: https://reviews.llvm.org/D46890
llvm-svn: 332810
Provide some free functions to reduce verbosity of endian-writing
a single value, and replace the endianness template parameter with
a field.
Part of PR37466.
Differential Revision: https://reviews.llvm.org/D47032
llvm-svn: 332757
The idea is that a client that wants split dwarf would create a
specific kind of object writer that creates two files, and use it to
create the streamer.
Part of PR37466.
Differential Revision: https://reviews.llvm.org/D47050
llvm-svn: 332749
Summary:
The logic of dispatch remains the same, but now DispatchUnit is a Stage (DispatchStage).
This change has the benefit of simplifying the backend runCycle() code.
The same logic applies, but it belongs to different components now. This is just a start,
eventually we will need to remove the call to the DispatchStage in Scheduler.cpp, but
that will be a separate patch. This change is mostly a renaming and moving of existing logic.
This change also encouraged me to remove the Subtarget (STI) member from the
Backend class. That member was used to initialize the other members of Backend
and to eventually call DispatchUnit::dispatch(). Now that we have Stages, we
can eliminate this by instantiating the DispatchStage with everything it needs
at the time of construction (e.g., Subtarget). That change allows us to call
DispatchStage::execute(IR) as we expect to call execute() for all other stages.
Once we add the Stage list (D46907) we can more cleanly call preExecute() on
all of the stages, DispatchStage, will probably wrap cycleEvent() in that
case.
Made some formatting and minor cleanups to README.txt. Some of the text
was re-flowed to stay within 80 cols.
Reviewers: andreadb, courbet, RKSimon
Reviewed By: andreadb, courbet
Subscribers: mgorny, javed.absar, tschuett, gbedwell, llvm-commits
Differential Revision: https://reviews.llvm.org/D46983
llvm-svn: 332652
Currently debugify prints it's output to stdout,
with this patch all the output generated goes to stderr.
This change lets us use debugify without taking away
the ability to pipe the output to other llvm tools.
llvm-svn: 332642
Summary:
Warn on instructions that should have the same performance
characteristics according to the sched model but actually
differ in their benchmarks.
Next step: Make the display nicer to browse, I was thinking maybe html.
Reviewers: gchatelet
Subscribers: tschuett, llvm-commits
Differential Revision: https://reviews.llvm.org/D46945
llvm-svn: 332601
Restructuring the code to measure latency and uops.
The end goal is to have this program spawn another process to deal with SIGILL and other malformed programs. It is not yet the case in this redesign, it is still the main program that runs the code (and may crash).
It now uses BitVector instead of Graph for performance reasons.
https://reviews.llvm.org/D46821
(with fixed ARM tests)
Authored by Guillaume Chatelet
llvm-svn: 332592
Restructuring the code to measure latency and uops.
The end goal is to have this program spawn another process to deal with SIGILL and other malformed programs. It is not yet the case in this redesign, it is still the main program that runs the code (and may crash).
It now uses BitVector instead of Graph for performance reasons.
https://reviews.llvm.org/D46821
Authored by Guillaume Chatelet
llvm-svn: 332579
Fuchsia uses ELF as a file format and LLD as the linker so we can
use the same implementation as other ELF based platforms.
Differential Revision: https://reviews.llvm.org/D46991
llvm-svn: 332570
VSOs now track dependencies for materializing symbols. Each symbol must have its
dependencies registered with the VSO prior to finalization. Usually this will
involve registering the dependencies returned in
AsynchronousSymbolQuery::ResolutionResults for queries made while linking the
symbols being materialized.
Queries against symbols are notified that a symbol is ready once it and all of
its transitive dependencies are finalized, allowing compilation work to be
broken up and moved between threads without queries returning until their
symbols fully safe to access / execute.
Related utilities (VSO, MaterializationUnit, MaterializationResponsibility) are
updated to support dependence tracking and more explicitly track responsibility
for symbols from the point of definition until they are finalized.
llvm-svn: 332541
Summary: This change will help us turn the DispatchUnit into its own stage.
Reviewers: andreadb, RKSimon, courbet
Reviewed By: andreadb, courbet
Subscribers: mgorny, tschuett, gbedwell, llvm-commits
Differential Revision: https://reviews.llvm.org/D46916
llvm-svn: 332493
The module ID numbering typically starts at 0 (in both the new and old
LTO APIs, used by linkers). Make llvm-lto consistent with that.
Split out of D46699.
llvm-svn: 332476
Revision 332390 introduced a FetchStage class in llvm-mca.
By design, FetchStage owns all the instructions in-flight in the OoO Backend.
Before this change, new instructions were added to a DenseMap indexed by
instruction id. The problem with using a DenseMap is that elements are not
ordered by key. This was causing a massive slow down in method
FetchStage::postExecute(), which searches for instructions retired that can be
deleted.
This patch replaces the DenseMap with a std::map ordered by instruction index.
At the end of every cycle, we search for the first instruction which is not
marked as "retired", and we remove all the previous instructions before it.
This works well because instructions are retired in-order.
Before this patch, a debug build of llvm-mca (on my Ryzen linux machine) took
~8.0 seconds to simulate 3000 iterations of a x86 dot-product (a `vmulps,
vpermilps, vaddps, vpermilps, vaddps` sequence). With this patch, it now takes
~0.8s to run all the 3000 iterations.
llvm-svn: 332461
This option just keeps being a problem and really needs to be implemented
in some fashion. Implementing it properly requires some kind of
"replaceSectionReference" method because all the existing links need to be
maintained. The desired behavior is just for allocated sections to become
NOBITS but actually implementing that is rather tricky due to the current
design of llvm-objcopy. However converting allocated sections to NOBITS is
just an optimization and not something debuggers need. Debuggers can debug
a stripped executable and take an unstripped executable for that stripped
executable as input. Additionally allocated sections account for a very
small part of debug binaries so this optimization is quite small. I propose
that for the time being we implement this as a NOP so that people can use
llvm-objcopy where they need to, just in a sub-optimal way.
This option has already blocked a lot of people and its currently blocking me.
llvm-svn: 332396
Summary:
This is just an idea, really two ideas. I expect some push-back,
but I realize that posting a diff is the most comprehensive way to express
these concepts.
This patch introduces a Stage class which represents the
various stages of an instruction pipeline. As a start, I have created a simple
FetchStage that is based on existing logic for how MCA produces
instructions, but now encapsulated in a Stage. The idea should become more concrete
once we introduce additional stages. The idea being, that when a stage completes,
the next stage in the pipeline will be executed. Stages are chained together
as a singly linked list to closely model a real pipeline. For now there is only one stage,
so the stage-to-stage flow of instructions isn't immediately obvious.
Eventually, Stage will also handle event notifications, but that functionality
is not complete, and not destined for this patch. Ideally, an interested party
can register for notifications from a particular stage. Callbacks will be issued to
these listeners at various points in the execution of the stage.
For now, eventing functionality remains similar to what it has been in mca::Backend.
We will be building-up the Stage class as we move on, such as adding debug output.
This patch also removes the unique_ptr<Instruction> return value from
InstrBuilder::createInstruction. An Instruction pointer is still produced,
but now it's up to the caller to decide how that item should be managed post-allocation
(e.g., smart pointer). This allows the Fetch stage to create instructions and
manage the lifetime of those instructions as it wishes, and not have to be bound to any
specific managed pointer type. Other callers of createInstruction might have different
requirements, and thus can manage the pointer to fit their needs. Another idea would be to push the
ownership to the RCU.
Currently, the FetchStage will wrap the Instruction
pointer in a shared_ptr. This allows us to remove the Instruction container in
Backend, which was probably going to disappear, or move, at some point anyways.
Note that I did run these changes through valgrind, to make sure we are not leaking
memory. While the shared_ptr comes with some additional overhead it relieves us
from having to manage a list of generated instructions, and/or make lookup calls
to remove the instructions.
I realize that both the Stage class and the Instruction pointer management
(mentioned directly above) are separate but related ideas, and probably should
land as separate patches; I am happy to do that if either idea is decent.
The main reason these two ideas are together is that
Stage::execute() can mutate an InstRef. For the fetch stage, the InstRef is populated
as the primary action of that stage (execute()). I didn't want to change the Stage interface
to support the idea of generating an instruction. Ideally, instructions are to
be pushed through the pipeline. I didn't want to draw too much of a
specialization just for the fetch stage. Excuse the word-salad.
Reviewers: andreadb, courbet, RKSimon
Reviewed By: andreadb
Subscribers: llvm-commits, mgorny, javed.absar, tschuett, gbedwell
Differential Revision: https://reviews.llvm.org/D46741
llvm-svn: 332390
This option permits to explicitly keep the specified
symbol so that it doesn't get removed.
Differential Revision: https://reviews.llvm.org/D46819
llvm-svn: 332356
Summary:
AsmTemplate becomes IntructionBenchmarkKey, which has three components.
This allows retreiving the opcode for analysis.
Reviewers: gchatelet
Subscribers: tschuett, llvm-commits
Differential Revision: https://reviews.llvm.org/D46873
llvm-svn: 332348
Matt Davis