This PR is motivated by a mismatch we discovered between compilation
results with vs. without `-g3`. We noticed this when compiling SPEC2017
testcases. The specific instance we saw is fixed in this PR by modifying
a guard (see below), but it is likely similar instances exist elsewhere
in the codebase.
The specific case fixed in this PR manifests itself in the `SimplifyCFG`
pass doing different things depending on whether DebugInfo is generated
or not. At the end of this comment, there is reduced example code that
shows the behavior in question.
The differing behavior has two root causes:
1. Commit https://github.com/llvm/llvm-project/commit/c07e19b adds loop
metadata including debug locations to loops that otherwise would not
have loop metadata
2. Commit https://github.com/llvm/llvm-project/commit/ac28efa6c100 adds
a guard to a simplification action in `SImplifyCFG` that prevents it
from simplifying away loop metadata
So, the change in 2. does not consider that when compiling with debug
symbols, loops that otherwise would not have metadata that needs
preserving, now have debug locations in their loop metadata. Thus, with
`-g3`, `SimplifyCFG` behaves differently than without it.
The larger issue is that while debug info is not supposed to influence
the final compilation result, commits like 1. blur the line between what
is and is not debug info, and not all optimization passes account for
this.
This PR does not address that and rather just modifies this particular
guard in order to restore equivalent behavior between debug and
non-debug builds in this one instance.
---
Here is a reduced version of a file from `f526.blender_r` that showcases
the behavior in question:
```C
struct LinkNode;
typedef struct LinkNode {
struct LinkNode *next;
void *link;
} LinkNode;
void do_projectpaint_thread_ph_v_state() {
int *ps = do_projectpaint_thread_ph_v_state;
LinkNode *node;
while (do_projectpaint_thread_ph_v_state)
for (node = ps; node; node = node->next)
;
}
```
Compiling this with and without DebugInfo, and then disassembling the
results, leads to different outcomes (tested on SystemZ and X86). The
reason for this is that the `SimplifyCFG` pass does different things in
either case.
This patch updates the `VFABIDemangler` to support vector functions that
return struct types. For example, a vector variant of `sincos` that
returns a vector of sine values and a vector of cosine values within a
struct.
This patch also adds some helpers for vectorizing types (including
struct types). Some of these are used in the `VFABIDemangler`, and
others will be used in subsequent patches, so this patch simply adds
tests for them.
This makes sure no optimizations are applied that assume the
bigger alignment or size, which could be incorrect if we link
together with non-instrumented code.
Improve the non-fatal cases to use DiagnosticInfo, which will now
provide a location. The allocators attempt to report different errors
if it happens to see inline assembly is involved (this detection is
quite unreliable) using srcloc instead of dbgloc. For now, leave this
behavior unchanged. I think reporting the full location and context
function would be more useful.
Strip hash_value() for CmpPredicate, as different callers have different
hashing use-cases. In this case, there is just one caller, namely
EarlyCSE, which calls hash_combine() on a CmpPredicate, which used to
call hash_combine() on a CmpInst::Predicate prior to 4a0d53a
(PatternMatch: migrate to CmpPredicate). This has uncovered a bug where
two icmp instructions differing in just the fact that one of them has
the samesign flag on it are hashed differently, leading to divergent
hashing, and a crash. Fix this crash by dropping samesign information on
icmp instructions before hashing them, preserving the former behavior.
Fixes#119893.
With the introduction of CmpPredicate in 51a895a (IR: introduce struct
with CmpInst::Predicate and samesign), PatternMatch is one of the first
key pieces of infrastructure that must be updated to match a CmpInst
respecting samesign information. Implement this change to Cmp-matchers.
This is a preparatory step in migrating the codebase over to
CmpPredicate. Since we no functional changes are desired at this stage,
we have chosen not to migrate CmpPredicate::operator==(CmpPredicate)
calls to use CmpPredicate::getMatching(), as that would have visible
impact on tests that are not yet written: instead, we call
CmpPredicate::operator==(Predicate), preserving the old behavior, while
also inserting a few FIXME comments for follow-ups.
To avoid changing the behavior in the general case, only do this
for anonymous functions. Otherwise, we'll end up with a leading
'@' on the name, which may not be meaningful to end users.
Currently LLVMContext::emitError emits any error as an "inline asm"
error which does not make any sense. InlineAsm appears to be special,
in that it uses a "LocCookie" from srcloc metadata, which looks like
a parallel mechanism to ordinary source line locations. This meant
that other types of failures had degraded source information reported
when available.
Introduce some new generic error types, and only use inline asm
in the appropriate contexts. The DiagnosticInfo types are still
a bit of a mess, and I'm not sure why DiagnosticInfoWithLocationBase
exists instead of just having an optional DiagnosticLocation in the
base class.
DK_Generic is for any error that derives from an IR level instruction,
and thus can pull debug locations directly from it. DK_GenericWithLoc
is functionally the generic codegen error, since it does not depend
on the IR and instead can construct a DiagnosticLocation from the
MI debug location.
This PR changes how target extension types are printed when they are
emitted as IR.
This prevents repetitive phrases like "struct = type {...}" from being
repeated over and over in the outputted IR.
Additionally, it should allow opt to not crash when parsing the DXIL
output.
Fixes [#114131](https://github.com/llvm/llvm-project/issues/114131)
Clang [defaults to aligning `__int128_t` to 16 bytes], while LLVM
`datalayout` strings [default to aligning `i128` to 8 bytes]. Wasm is
currently using the defaults for both, so it's inconsistent. Fix this by
adding `-i128:128` to Wasm's `datalayout` string so that it aligns
`i128` to 16 bytes too.
This is similar to
[llvm/llvm-project@dbad963](dbad963a69)
for SPARC.
This fixesrust-lang/rust#133991; see that issue for further discussion.
[defaults to aligning `__int128_t` to 16 bytes]:
f8b4182f07/clang/lib/Basic/TargetInfo.cpp (L77)
[default to aligning `i128` to 8 bytes]:
https://llvm.org/docs/LangRef.html#langref-datalayout
This avoids the need to dynamically relocate each pointer in the table.
To make this work, this PR also moves the binary search of intrinsic
names to an internal function with an adjusted signature, and switches
the unittesting to test against actual intrinsics.
Because it was implemented in terms of getMaxIndexSize, it was always
rounding the values up to a multiple of 8. Additionally, it was using
the PointerSpec's BitWidth rather than its IndexBitWidth, which was
self-evidently incorrect.
Since getMaxIndexSize was only used by getMaxIndexSizeInBits, and its
name and function seem niche and somewhat confusing, go ahead and remove
it until a concrete need for it arises.
It doesn't make sense to add a new generic ISD to handle riscv tuple
type. Instead we use `SPLAT_VECTOR` for ISD and further lower to
`VMV_V_X`.
Note: If there's `visitSPLAT_VECTOR` in generic DAG combiner, it needs
to skip riscv vector tuple type.
Stack on https://github.com/llvm/llvm-project/pull/114329
This update enhances the implementation of structural hashing for global
variables, using their initial contents. Private global variables or
constants are often used for metadata, where their names are not unique.
This can lead to the creation of different hash results although they
could be merged by the linker as they are effectively identical.
- Refine the hashing of GlobalVariables for strings or certain
Objective-C metadata cases that have section names. This can be further
extended to other scenarios.
- Expose StructuralHash for GlobalVariable so that this API can be
utilized by MachineStableHashing, which is also employed in the global
function outliner.
This change significantly improves size reduction by an additional 1% on
the LLD binary when the global function outliner and merger are enabled
together. As discussed in the RFC
https://discourse.llvm.org/t/loh-conflicting-with-machineoutliner/83279/8?u=kyulee-com,
if we disable or relocate the LOH pass, the size impact could increase
to 4%.
Introduce llvm::CmpPredicate, an abstraction over a floating-point
predicate, and a pack of an integer predicate with samesign information,
in order to ease extending large portions of the codebase that take a
CmpInst::Predicate to respect the samesign flag.
We have chosen to demonstrate the utility of this new abstraction by
migrating parts of ValueTracking, InstructionSimplify, and InstCombine
from CmpInst::Predicate to llvm::CmpPredicate. There should be no
functional changes, as we don't perform any extra optimizations with
samesign in this patch, or use CmpPredicate::getMatching.
The design approach taken by this patch allows for unaudited callers of
APIs that take a llvm::CmpPredicate to silently drop the samesign
information; it does not pose a correctness issue, and allows us to
migrate the codebase piece-wise.
When `-import-declaration` option is enabled, declaration import is
supported for functions. https://github.com/llvm/llvm-project/pull/88024
has the context for this option.
This patch supports declaration import for global variables in
distributed ThinLTO. The motivating use case is to propagate `dso_local`
attribute of global variables across modules, to optimize global
variable access when a binary is built with
`-fno-direct-access-external-data`.
* With `-fdirect-access-external-data`, non thread-local global
variables will [have `dso_local`
attributes](fe3c23b439/clang/lib/CodeGen/CodeGenModule.cpp (L1730-L1746)).
This optimizes the global variable access as shown by
https://gcc.godbolt.org/z/vMzWcKdh3
This should act like range.
Previously ConstantRangeList assumed a 64-bit range. Now query from the
actual entries. This also means that the empty range has no bitwidth, so
move asserts to avoid checking the bitwidth of empty ranges.
SLEEF math vector library now supports RISC-V target.
Commit: https://github.com/shibatch/sleef/pull/477
This patch enables the use of auto-vectorization with
subsequent replacement by the corresponding SLEEF function.
Unfortunately there's no upstream frontend for Metal but since the id's
are now assigned by the DWARF standard I think it makes sense to have
the enums upstream to enable tools like llvm-dwarfdump. This patch
therefore uses an AArch64 test with artificially modified debug info to
verify that the Metal language id can be used.
https://dwarfstd.org/issues/241111.1.html
Verify the format is valid and the type is one of the expected
i32 vectors. Verify the used vector types at least cover the
requirements of the corresponding format operand.
Add a property to allow marking target extension types that cannot be
used in an alloca instruction or byval argument, similar to CanBeGlobal
for global variables.
---------
Co-authored-by: Nikita Popov <github@npopov.com>
If the size is larger than the index width, truncate it instead
of asserting.
Longer-term we should consider rejecting types larger than the
index size in the verifier, though this is probably tricky in
practice (it's address space dependent, and types are owned by
the context, not the module).
Fixes https://github.com/llvm/llvm-project/issues/116960.
De-duplicate the functions getSignedPredicate and getUnsignedPredicate,
nearly identical versions of which were present in CmpInst and ICmpInst,
creating less confusion.
Bit shift operations with a shift operand greater than or equal to the bit width
of the (promoted) value type result in undefined behavior according to C++
[expr.shift]p1. This change adds checking for this situation and avoids attempts
to constant fold DIExpressions that would otherwise provoke such behavior.
An existing test that presumably intended to exercise shifts at the UB boundary
has been updated; it now checks for shifts of 64 bits instead of 65. This issue
was reported by a static analysis tool; no actual cases of shift operations that
would result in undefined behavior in practice have been identified.
Improve the information printed when -memprof-report-hinted-sizes is
enabled. Now print the full context hash computed from the original
profile, similar to what we do when reporting matching statistics. This
will make it easier to correlate with the profile.
Note that the full context hash must be computed at profile match time
and saved in the metadata and summary, because we may trim the context
during matching when it isn't needed for distinguishing hotness.
Similarly, due to the context trimming, we may have more than one full
context id and total size pair per MIB in the metadata and summary,
which now get a list of these pairs.
Remove the old aggregate size from the metadata and summary support.
One other change from the prior support is that we no longer write the
size information into the combined index for the LTO backends, which
don't use this information, which reduces unnecessary bloat in
distributed index files.
Relands 7ff3a9acd84654c9ec2939f45ba27f162ae7fbc3 after regenerating the
test case.
Supersedes the draft PR #94992, taking a different approach following
feedback:
* Lower in PreISelIntrinsicLowering
* Don't require that the number of bytes to set is a compile-time
constant
* Define llvm.memset_pattern rather than llvm.memset_pattern.inline
As discussed in the [RFC
thread](https://discourse.llvm.org/t/rfc-introducing-an-llvm-memset-pattern-inline-intrinsic/79496),
the intent is that the intrinsic will be lowered to loops, a sequence of
stores, or libcalls depending on the expected cost and availability of
libcalls on the target. Right now, there's just a single lowering path
that aims to handle all cases. My intent would be to follow up with
additional PRs that add additional optimisations when possible (e.g.
when libcalls are available, when arguments are known to be constant
etc).
This reverts commit 7ff3a9acd84654c9ec2939f45ba27f162ae7fbc3.
Recent scheduling changes means tests need to be re-generated. Reverting
to green while I do that.
The logic for marking runtime libcalls unavailable currently duplicates
essentially the same logic for some random subset of targets, where
someone reported an issue and then someone went and fixed the issue for
that specific target only. However, the availability for most of these
is completely target independent. In particular:
* MULO_I128 is never available in libgcc
* Various I128 libcalls are not available for 32-bit targets in libgcc
* powi is never available in MSVCRT
Unify the logic for these, so we don't miss any targets. This fixes
https://github.com/llvm/llvm-project/issues/16778 on AArch64, which is
one of the targets that was previously missed in this logic.
Supersedes the draft PR #94992, taking a different approach following
feedback:
* Lower in PreISelIntrinsicLowering
* Don't require that the number of bytes to set is a compile-time
constant
* Define llvm.memset_pattern rather than llvm.memset_pattern.inline
As discussed in the [RFC
thread](https://discourse.llvm.org/t/rfc-introducing-an-llvm-memset-pattern-inline-intrinsic/79496),
the intent is that the intrinsic will be lowered to loops, a sequence of
stores, or libcalls depending on the expected cost and availability of
libcalls on the target. Right now, there's just a single lowering path
that aims to handle all cases. My intent would be to follow up with
additional PRs that add additional optimisations when possible (e.g.
when libcalls are available, when arguments are known to be constant
etc).
As discussed in #112738, it may be better to have an intrinsic to represent vector element extracts based on mask bits. This intrinsic is for the case of extracting the last active element, if any, or a default value if the mask is all-false.
The target-agnostic SelectionDAG lowering is similar to the IR in #106560.
