Background: BPI stores a collection of edge branch-probabilities, and
also a set of Callback value-handles for the blocks in the
edge-collection. When a block is deleted, BPI's eraseBlock method is
called to clear the edge-collection of references to that block, to
avoid dangling pointers.
However, when move-constructing or assigning a BPI object, the
edge-collection gets moved, but the value-handles are discarded. This
can lead to to stale entries in the edge-collection when blocks are
deleted without the callback -- not normally a problem, but if a new
block is allocated with the same address as an old block, spurious
branch probabilities will be recorded about it. The fix is to transfer
the handles from the source BPI object.
This was exposed by an unrelated debug-info change, it probably just
shifted around allocation orders to expose this. Detected as
nondeterminism and reduced by Zequan Wu:
f1b0a54451 (commitcomment-136737090)
(No test because IMHO testing for a behaviour that varies with memory
allocators is likely futile; I can add the reproducer with a CHECK for
the relevant branch weights if it's desired though)
The files and directories that Clang accesses are uniqued by their
inode. For each inode `FileManager` will create exactly one `FileEntry`
or `DirectoryEntry` object, which makes answering the question _"Are
these two files/directories the same?"_ a simple pointer equality check.
However, since the same inode can be accessed through multiple different
paths, asking the `FileEntry` or `DirectoryEntry` object _"What is your
name?"_ doesn't have clear semantics. In c0ff9908 we started reporting
the most recent name used to access the entry, which turned out to be
necessary for Clang modules. However, the long-term solution has always
been to explicitly track the as-requested name. This has been
implemented in 4dc5573a as `FileEntryRef` and `DirectoryEntryRef`.
The `DirectoryEntry::getName()` interface has been deprecated since the
Clang 17 release and `FileEntry::getName()` since Clang 18. We have
replaced uses of these deprecated APIs in `main` with
`DirectoryEntryRef::getName()` and `FileEntryRef::getName()`
respectively.
This makes it possible to remove `{File,Directory}Entry::getName()` for
good along with the `FileManager` code that implements them.
This commit introduces a helper function to DWARFAcceleratorTable::Entry
which follows DW_IDX_Parent attributes to returns the corresponding
parent Entry in the table.
It is tested by enhancing dwarfdump so that it now prints:
1. When data is corrupt.
2. When parent information is present, but the parent is not indexed.
3. The parent entry offset, when the parent is present and indexed. This
is printed in terms a real entry offset (the same that gets printed at
the start of each entry: "Entry @ 0x..."), instead of the encoded number
in the table (which is an offset from the start off the Entry list).
This makes it easy to visually inspect the dwarfdump and check what the
parent is.
Implement PhiLoweringHelper for GlobalISel in DivergenceLoweringHelper.
Use machine uniformity analysis to find divergent i1 phis and select
them as lane mask phis in same way SILowerI1Copies select VReg_1 phis.
Note that divergent i1 phis include phis created by LCSSA and all cases
of uses outside of cycle are actually covered by "lowering LCSSA phis".
GlobalISel lane masks are registers with sgpr register class and S1 LLT.
TODO: General goal is that instructions created in this pass are fully
instruction-selected so that selection of lane mask phis is not split
across multiple passes.
patch 3 from: https://github.com/llvm/llvm-project/pull/73337
This patch extends HWASAN to support maintenance of debug-info that
isn't stored as intrinsics, but is instead in a DPValue object. This is
straight-forwards: we collect any such objects in StackInfoBuilder, and
apply the same operations to them as we would to dbg.value and similar
intrinsics.
I've also replaced some calls to getNextNode with debug-info skipping
next calls, and use iterators for instruction insertion rather than
instruction pointers. This avoids any difference in output between
intrinsic / non-intrinsic debug-info, but also means that any debug-info
comes before code inserted by HWAsan, rather than afterwards. See the
test modifications, where the variable assignment (presented as a
dbg.value) jumps up over all the code inserted by HWAsan. Seeing how the
code inserted by HWAsan is always (AFAIUI) given the source-location of
the instruction being instrumented, I don't believe this will have any
effect on which lines variable assignments become visible on; it may
extend the number of instructions covered by the assignments though.
`CodeGenPassBuilder` is not very tightly coupled to CodeGen, it may need
to reference some method in pass builder in future, so move
`CodeGenPassBuilder.h` to Passes.
Add new pass manager support to `llc`. Users can use
`--passes=pass1,pass2...` to run mir passes, and use `--enable-new-pm`
to run default codegen pipeline.
This patch is taken from [D83612](https://reviews.llvm.org/D83612), the
original author is @yuanfang-chen.
---------
Co-authored-by: Yuanfang Chen <455423+yuanfang-chen@users.noreply.github.com>
Now that the work embedding PGO information in SHT_LLVM_BB_ADDR_MAP ELF
sections has landed, there is no longer a need to keep around the
mbb-profile-dump flag.
This patch adds basic TLSDESC support in the RISC-V backend.
Specifically, we add new relocation types for TLSDESC, as prescribed in
https://github.com/riscv-non-isa/riscv-elf-psabi-doc/pull/373, and add a
new pseudo instruction to simplify code generation.
This patch does not try to optimize the local dynamic case, which can be
improved in separate patches.
Linker side changes will also be handled separately.
The current implementation is only enabled when passing the new
`-enable-tlsdesc` codegen flag.
Here's a raft of minor fixes for the RemoveDIs project that's replacing
dbg.value intrinsics with DPValue objects, all IMO trivial:
* When inserting functions or blocks and calling setIsNewDbgInfoFormat,
do that after setting the Parent pointer, just in case conversion from
(or to) dbg.value mode is triggered.
* When transferring DPValues from an empty range in a splice call, don't
transfer if there are no DPValues attached to the source block at all.
* stripNonLineTableDebugInfo should drop DPValues.
* In insertBefore, don't try to transfer DPValues if there aren't any.
Currently, the UnifiedLTO pipeline seems to have trouble with several
LTO features, like SplitLTO units, which means we cannot use important
optimizations like Whole Program Devirtualization or security hardening
instrumentation like CFI.
This patch reverts FatLTO to using distinct pipelines for Full LTO and
ThinLTO. It still avoids module cloning, since that was error prone.
`--function=<regex>` Include functions matching regex in the output
`--no-function=<regex>` Exclude functions matching regex from the output
If both are specified, `--no-function` has a higher precedence if a
function name matches both filters
This commit unpoisons memory before its reuse (with reinterpret_cast).
Required by https://github.com/llvm/llvm-project/pull/79049
Notice that it's a temporary solution to prevent buildbots from failing.
Read FIXME for details.
With this commit, scribbling under AddressSanitizer (ASan) is disabled to prevent overwriting poisoned objects (e.g., annotated short strings).
Needed by https://github.com/llvm/llvm-project/pull/79049
CacheOptForSize to remove quadratic behavior.
For each constant analyzed, ConstantHoising calls
`shouldOptimizeForSize(F)`, which calls `PSI.getTotalCallCount(F)`.
PSI.getTotalCallCount(F) goes through all the instructions in all basic
blocks, and checks if each is a call, to count them up.
This reduces `llc` time for a very large IR from ~10min to under 3min.
Reproducer testcase is much too large to share.
C++20 accepted two papers, [P0668](https://wg21.link/P0668) and
[P0982](https://wg21.link/P0982), which changed the atomics memory model
slightly in order to reflect the realities of the existing
implementations.
The rationale for these changes applies as well to the LLVM IR atomics
model. No code changes are expected to be required from this change: it
is primarily a matter of more-correctly-documenting the existing state
of the world.
There's three changes: two of them weaken guarantees, and one
strengthens them:
1. The memory ordering guaranteed by some backends/CPUs when seq_cst
operations are mixed with acquire/release operations on the same
location was weaker than the spec guaranteed. Therefore, the
specification is changed to remove the requirement that seq_cst ordering
is consistent with happens-before, and replaces it with a slightly
weaker requirement of consistency with a new relation named
strongly-happens-before.
2. The rules for a "release sequence" were weakened. Previously, an
acquire synchronizes with an release even if it observes a later
monotonic store from the same thread as the release store. That has now
been removed: now, only read-modify-write operations can extend a
release sequence.
3. The model for a a seq_cst fence is strengthened, such that placing a
seq_cst between monotonic accesses now _is_ sufficient to guarantee
sequential consistency in the model (as it always has been on existing
implementations.)
Note that I've directly referenced the C++ standard's atomics.order
section for the precise semantics of seq_cst, instead of fully
describing them. They are quite complex, and a lot of work has gone into
refining the words in the standard. I'm afraid if I attempt to reiterate
them, I would only introduce errors.
This stub type loads an absolute address directly into the PC register.
It's the simplest and most compatible way to implement a branch
indirection across the entire address space (and probably the slowest as
well). It's the ideal fallback for all targets for which we did not
(yet) implement a more performant solution.
The patch adds support for FEAT_MOPS (Memory Copy and Memory Set
instructions) in Function Multi Versioning. The bits [19:16] of the
system register ID_AA64ISAR2_EL1 indicate whether FEAT_MOPS is
implemented in AArch64 state. This information is accessible via ELF
hwcaps.
The RemoveDIs project is aiming to eliminate debug intrinsics like
dbg.value and dbg.declare from LLVM, and replace them with DPValue objects
attached to instructions. ISel is one of the "terminals" where that
information needs to be converted into MIR format: this patch implements
support for that in GlobalISel. We aim for the output of LLVM to be
identical with/without RemoveDIs debug-info.
This patch should be NFC, as we're handling the same data about variables
stored in a different format -- it now appears in a DPValue object rather
than as an intrinsic. To that end, I've refactored the handling of
dbg.values into a dedicated function, and call it whenever a dbg.value or a
DPValue is encountered. dbg.declare is handled in a similar way.
Testing: adding the --try-experimental-debuginfo-iterators switch to llc
causes it to try and convert to the "new" debug-info format if it's built
in (LLVM_EXPERIMENTAL_DEBUGINFO_ITERATORS=On), and it'll be covered by our
buildbot. One test has a few extra wildcard-regexes added: this is because
there's some extra data printed about attached debug-info, which is safe to
ignore.
Make Candidate's front() and back() functions return references to
MachineInstr and introduce begin() and end() returning iterators, the
same way it is usually done in other container-like classes.
This makes possible to iterate over the instructions contained in
Candidate the same way one can iterate over MachineBasicBlock (note that
begin() and end() return bundled iterators, just like MachineBasicBlock
does, but no instr_begin() and instr_end() are defined yet).
SROA needs to update llvm.dbg.assign intrinsics when it migrates debug
info in response to alloca splitting; this patch updates the debug info
migration code to handle DPVAssigns as well, making use of generic code
to avoid duplication as much as possible.
IntrArgMemOnly means the intrinsic only accesses memory that its
pointer-typed argument(s) points to. I think RVV load/store intrinsics
meets it. Add IntrArgMemOnly would help in some passes, by example, it
could add `alais.scope` to intrinsics callee when try to inline a
function that has noalais parameter(s).
This patch adds support for additional types of validation counters and
also adds mappings between these new validation counter types and
physical counters on the hardware for microarchitectures that I have the
ability to test on.
This combines the previously posted patches with some additional work
I've done to more closely match MSVC output.
Most of the important logic here is implemented in
AArch64Arm64ECCallLowering. The purpose of the
AArch64Arm64ECCallLowering is to take "normal" IR we'd generate for
other targets, and generate most of the Arm64EC-specific bits:
generating thunks, mangling symbols, generating aliases, and generating
the .hybmp$x table. This is all done late for a few reasons: to
consolidate the logic as much as possible, and to ensure the IR exposed
to optimization passes doesn't contain complex arm64ec-specific
constructs.
The other changes are supporting changes, to handle the new constructs
generated by that pass.
There's a global llvm.arm64ec.symbolmap representing the .hybmp$x
entries for the thunks. This gets handled directly by the AsmPrinter
because it needs symbol indexes that aren't available before that.
There are two new calling conventions used to represent calls to and
from thunks: ARM64EC_Thunk_X64 and ARM64EC_Thunk_Native. There are a few
changes to handle the associated exception-handling info,
SEH_SaveAnyRegQP and SEH_SaveAnyRegQPX.
I've intentionally left out handling for structs with small
non-power-of-two sizes, because that's easily separated out. The rest of
my current work is here. I squashed my current patches because they were
split in ways that didn't really make sense. Maybe I could split out
some bits, but it's hard to meaningfully test most of the parts
independently.
Thanks to @dpaoliello for extensive testing and suggestions.
(Originally posted as https://reviews.llvm.org/D157547 .)
`R_ARM_PREL31` is a 31-bits relative data relocation where the
most-significant bit is preserved. It's used primarily in `.ARM.exidx`
sections, which we skipped processing until now, because we didn't
support the relocation type. This was implemented in RuntimeDyld with
https://reviews.llvm.org/D25069 and I implemented it in a similar way in
JITLink in order to reach feature parity.
We want to emit stubs that match the instruction set state of the
relocation site. This is important for branches that have no built-in
switch for the instruction set state. It's the case for Jump24
relocations. Relocations on instructions that support switching on
the fly will be rewritten in a relaxation step in the future. This
affects Call relocations on `BL`/`BLX` instructions.
In this patch, the StubManager gains a second stub symbol slot for each
target and selects which one to use based on the relocation type. For
testing, we select the appropriate slot with a stub-kind filter, i.e.
`arm` or `thumb`. With that we can implement Armv7 stubs and test
that we can have both kinds of stubs for a single external symbol.
Although there are predicated versions of minnum/maxnum, the ones for
minimum/maximum are currently missing. This patch introduces these
intrinsics and implements their lowering to RISC-V.
Some operations behave like selects. For example `or(zext(c), y)` is the
same as select(c, y|1, y)` and instcombine can canonicalize the select
to the or form. These operations can still be worthwhile converting to
branch as opposed to keeping as a select or or instruction.
This patch attempts to add some basic handling for them, creating a
SelectLike abstraction in the select optimization pass. The backend can
opt into handling `or(zext(c),x)` as a select if it could be profitable,
and the select optimization pass attempts to handle them in much the
same way as a `select(c, x|1, x)`. The Or(x, 1) may need to be added as
a new instruction, generated as the or is converted to branches.
This helps fix a regression from selects being converted to or's
recently.
The life of `MCDCRecordProcessor`'s instance is short. It may accept
`const` objects to process.
On the other hand, the life of `MCDCBranches` is shorter than `Record`.
It may be rewritten with reference, rather than copying.
Previously, tail jump pseudo-opcodes were skipped by the
`encodeInstruction()` call inside `X86AsmPrinter::LowerPATCHABLE_OP`.
This caused emission of a 2-byte NOP and dropping of the tail jump.
With this PR, we change `PATCHABLE_OP` to not wrap the first
`MachineInstr` anymore, but inserting itself before,
leaving the instruction unaltered. At lowering time in `X86AsmPrinter`,
we now "look ahead" for the next non-pseudo `MachineInstr` and
lower+encode it, to inspect its size. If the size is below what
`PATCHABLE_OP` expects, it inserts NOPs; otherwise it does nothing. That
way, now the first `MachineInstr` is always lowered as usual even if
`"patchable-function"="prologue-short-redirect"` is used.
Fixes https://github.com/llvm/llvm-project/issues/76879,
https://github.com/llvm/llvm-project/issues/76958 and
https://github.com/llvm/llvm-project/issues/59039
I've been working on more/better configuration for improving DEBUGINFOD
support. This is the first (and easiest) slice of the work.
I've added `timeout` and `cache-path` settings that can override the
DEBUGINFOD library defaults (and environment variables.) I also renamed
the `plugin.symbol-locator.debuginfod.server_urls` setting to
`server-urls` to be more consistent with the rest of LLDB's settings
(the underscore switch is switched to a hyphen)
I've got a few tests that validate the cache-path setting (as a
side-effect), but they've exposed a few bugs that I'll be putting up a
separate PR for (which will include the tests).
---------
Co-authored-by: Kevin Frei <freik@meta.com>
- With PGO, indirect call edges are constructed using value profiles, and the profile address is mapped to a function's PGO name. The PGO name is computed using a functions linkage before LTO internalization or global promotion.
- With ThinLTO, local functions [could be
promoted](2663d2cb9c/llvm/lib/Transforms/Utils/FunctionImportUtils.cpp (L288)) to have external linkage; and with
[full](2663d2cb9c/llvm/lib/LTO/LTO.cpp (L1328))
or
[thin](2663d2cb9c/llvm/lib/LTO/LTO.cpp (L448))
LTO, global functions could be internalized. Edge construction should use a function's PGO name before its linkage is updated.
DPValues are already supported by most of the utilities that remove
redundant debug info after certain passes; the exception to this is
`removeUndefDbgAssignsFromEntryBlock`, which applies only to
llvm.dbg.assigns which were previously unimplemented for DPValues. Now
that DPVAssigns exist, we have to support removing redundant instances
in the same way, which this patch implements.
Rename intrinsics for fcvtu to fcvtzu and fcvts to fcvtzs.
Use llvm_anyvector_ty for both multi vector returns and operands,
therefore the return and operands can be specified in the intrinsic
call, e.g.
@llvm.aarch64.sve.scvtf.x4.nxv4f32.nxv4i32
This patch adds the preliminary changes for handling DPValues in
AssignmentTrackingAnalysis - very few functional changes are included,
but internal data structures have been changed to operate with DPValues
as well as Instructions, allowing future patches to process DPValues
correctly.
For inline asm with memory operands, we can merge the offset into
the second operand of memory constraint operands.
Differential Revision: https://reviews.llvm.org/D158062
