The patch adds SPIRV-specific MC layer implementation, SPIRV object
file support and SPIRVInstPrinter.
Differential Revision: https://reviews.llvm.org/D116462
Authors: Aleksandr Bezzubikov, Lewis Crawford, Ilia Diachkov,
Michal Paszkowski, Andrey Tretyakov, Konrad Trifunovic
Co-authored-by: Aleksandr Bezzubikov <zuban32s@gmail.com>
Co-authored-by: Ilia Diachkov <iliya.diyachkov@intel.com>
Co-authored-by: Michal Paszkowski <michal.paszkowski@outlook.com>
Co-authored-by: Andrey Tretyakov <andrey1.tretyakov@intel.com>
Co-authored-by: Konrad Trifunovic <konrad.trifunovic@intel.com>
specifying DW_AT_trampoline as a string. Also update the signature
of DIBuilder::createFunction to reflect this addition.
Differential Revision: https://reviews.llvm.org/D123697
IRLinker builds a work list of functions to materialize, then moves them
from a source module to a destination module one at a time.
This is a problem for blockaddress Constants, since they need not refer
to the function they are used in; IPSCCP is quite good at sinking these
constants deep into other functions when passed as arguments.
This would lead to curious errors during LTO:
ld.lld: error: Never resolved function from blockaddress ...
based on the ordering of function definitions in IR.
The problem was that IRLinker would basically do:
for function f in worklist:
materialize f
splice f from source module to destination module
in one pass, with Functions being lazily added to the running worklist.
This confuses BitcodeReader, which cannot disambiguate whether a
blockaddress is referring to a function which has not yet been parsed
("materialized") or is simply empty because its body was spliced out.
This causes BitcodeReader to insert Functions into its BasicBlockFwdRefs
list incorrectly, as it will never re-materialize an already
materialized (but spliced out) function.
Because of the possibility that blockaddress Constants may appear in
Functions other than the ones they reference, this patch adds a new
bitcode function code FUNC_CODE_BLOCKADDR_USERS that is a simple list of
Functions that contain BlockAddress Constants that refer back to this
Function, rather then the Function they are scoped in. We then
materialize those functions when materializing `f` from the example loop
above. This might over-materialize Functions should the user of
BitcodeReader ultimately decide not to link those Functions, but we can
at least now we can avoid this ordering related issue with blockaddresses.
Fixes: https://github.com/llvm/llvm-project/issues/52787
Fixes: https://github.com/ClangBuiltLinux/linux/issues/1215
Reviewed By: dexonsmith
Differential Revision: https://reviews.llvm.org/D120781
Returning `std::array<uint8_t, N>` is better ergonomics for the hashing functions usage, instead of a `StringRef`:
* When returning `StringRef`, client code is "jumping through hoops" to do string manipulations instead of dealing with fixed array of bytes directly, which is more natural
* Returning `std::array<uint8_t, N>` avoids the need for the hasher classes to keep a field just for the purpose of wrapping it and returning it as a `StringRef`
As part of this patch also:
* Introduce `TruncatedBLAKE3` which is useful for using BLAKE3 as the hasher type for `HashBuilder` with non-default hash sizes.
* Make `MD5Result` inherit from `std::array<uint8_t, 16>` which improves & simplifies its API.
Differential Revision: https://reviews.llvm.org/D123100
Necessary when importing class template specializations that have
simplified template names (may otherwise be necessary - eg: Sony
requires template parameter DIEs even with unsimplified names, but short
of always importing names this is the best I can do for now) - long term
this probably needs a flag for the DICompositeType to specify whether it
needs template parameters on declarations & that flag could power this
behavior, rather than inspecting the name.
This allows both explicitly enabling and explicitly disabling
opaque pointers, in anticipation of the default switching at some
point.
This also slightly changes the rules by allowing calls if either
the opaque pointer mode has not yet been set (explicitly or
implicitly) or if the value remains unchanged.
DXIL is wrapped in a container format defined by the DirectX 11
specification. Codebases differ in calling this format either DXBC or
DXILContainer.
Since eventually we want to add support for DXBC as a target
architecture and the format is used by DXBC and DXIL, I've termed it
DXContainer here.
Most of the changes in this patch are just adding cases to switch
statements to address warnings.
Reviewed By: pete
Differential Revision: https://reviews.llvm.org/D122062
The one use check here is very misleading: At this point we should
actually have no uses, because the only possible use in llvm.used
was already dropped. But because the use in llvm.used is generally
bitcasted, we end up still having one dead use here.
What we actually want to check is that there are no live uses, for
which a helper has recently been added.
This allows us to more easily test opaque pointers e.g. in the case of
ThinLTO where we only have to pass -opaque-pointers to the frontend.
Reviewed By: #opaque-pointers, nikic
Differential Revision: https://reviews.llvm.org/D122048
Includes verifier changes checking the elementtype, clang codegen
changes to emit the elementtype, and ISel changes using the elementtype.
Basically the same as D120527.
Reviewed By: #opaque-pointers, nikic
Differential Revision: https://reviews.llvm.org/D121847
Includes verifier changes checking the elementtype, clang codegen
changes to emit the elementtype, and ISel changes using the elementtype.
Reviewed By: #opaque-pointers, nikic
Differential Revision: https://reviews.llvm.org/D120527
We should not be using APIs here that try to fetch the attribute
from both the call attributes and the function attributes. Otherwise
we'll try to upgrade a non-existent sret attribute on the call using
the attribute on the function.
The constexpr source element type was enumerated if the GEP was
used as part of an instruction. However, things like global
initializers go through a different code path, and we need to
enumerate the type there as well.
Since D101045, allocas are no longer required to be part of the
default alloca address space. There may be allocas in multiple
different address spaces. However, the bitcode reader would
simply assume the default alloca address space, resulting in
either an error or incorrect IR.
Add an optional record for allocas which encodes the address
space.
As these errors are detected after the instruction has already been
created (but before it has been inserted into the function), we
also need to delete it.
This will let us start moving away from hard-coded attributes in
MemoryBuiltins.cpp and put the knowledge about various attribute
functions in the compilers that emit those calls where it probably
belongs.
Differential Revision: https://reviews.llvm.org/D117921
This completes the propagation of type IDs through bitcode reading,
and switches remaining uses of getPointerElementType() to use
contained type IDs.
The main new thing here is that sometimes we need to create a type
ID for a type that was not explicitly encoded in bitcode (or we
don't know its ID at the current point). For such types we create a
"virtual" type ID, which is cached based on the type and the
contained type IDs. Luckily, we generally only need zero or one
contained type IDs, and in the one case where we need two, we can
get away with not including it in the cache key.
With this change, we pass the entirety of llvm-test-suite at O3
with opaque pointers.
Differential Revision: https://reviews.llvm.org/D120471
Currently adding attribute no_sanitize("bounds") isn't disabling
-fsanitize=local-bounds (also enabled in -fsanitize=bounds). The Clang
frontend handles fsanitize=array-bounds which can already be disabled by
no_sanitize("bounds"). However, instrumentation added by the
BoundsChecking pass in the middle-end cannot be disabled by the
attribute.
The fix is very similar to D102772 that added the ability to selectively
disable sanitizer pass on certain functions.
In this patch, if no_sanitize("bounds") is provided, an additional
function attribute (NoSanitizeBounds) is attached to IR to let the
BoundsChecking pass know we want to disable local-bounds checking. In
order to support this feature, the IR is extended (similar to D102772)
to make Clang able to preserve the information and let BoundsChecking
pass know bounds checking is disabled for certain function.
Reviewed By: melver
Differential Revision: https://reviews.llvm.org/D119816
This is the next step towards supporting bitcode auto upgrade with
opaque pointers. The ValueList now stores the Value* together with
its associated type ID, which allows inspecting the original pointer
element type of arbitrary values.
This is a largely mechanical change threading the type ID through
various places. I've left TODOTypeID placeholders in a number of
places where determining the type ID is either non-trivial or
requires allocating a new type ID not present in the original
bitcode. For this reason, the new type IDs are also not used for
anything yet (apart from propagation). They will get used once the
TODOs are resolved.
Differential Revision: https://reviews.llvm.org/D119821
This is step two of supporting autoupgrade of old bitcode to opaque
pointers. Rather than tracking the element type ID of pointers in
particular, track all type IDs that a type contains. This allows us
to recover the element type in more complex situations, e.g. when
we need to determine the pointer element type of a vector element
or function type parameter.
Differential Revision: https://reviews.llvm.org/D119339
We have the `clang -cc1` command-line option `-funwind-tables=1|2` and
the codegen option `VALUE_CODEGENOPT(UnwindTables, 2, 0) ///< Unwind
tables (1) or asynchronous unwind tables (2)`. However, this is
encoded in LLVM IR by the presence or the absence of the `uwtable`
attribute, i.e. we lose the information whether to generate want just
some unwind tables or asynchronous unwind tables.
Asynchronous unwind tables take more space in the runtime image, I'd
estimate something like 80-90% more, as the difference is adding
roughly the same number of CFI directives as for prologues, only a bit
simpler (e.g. `.cfi_offset reg, off` vs. `.cfi_restore reg`). Or even
more, if you consider tail duplication of epilogue blocks.
Asynchronous unwind tables could also restrict code generation to
having only a finite number of frame pointer adjustments (an example
of *not* having a finite number of `SP` adjustments is on AArch64 when
untagging the stack (MTE) in some cases the compiler can modify `SP`
in a loop).
Having the CFI precise up to an instruction generally also means one
cannot bundle together CFI instructions once the prologue is done,
they need to be interspersed with ordinary instructions, which means
extra `DW_CFA_advance_loc` commands, further increasing the unwind
tables size.
That is to say, async unwind tables impose a non-negligible overhead,
yet for the most common use cases (like C++ exceptions), they are not
even needed.
This patch extends the `uwtable` attribute with an optional
value:
- `uwtable` (default to `async`)
- `uwtable(sync)`, synchronous unwind tables
- `uwtable(async)`, asynchronous (instruction precise) unwind tables
Reviewed By: MaskRay
Differential Revision: https://reviews.llvm.org/D114543
Make it clearer that this method is specifically for pointer
element types, and not other element types. This distinction will
be relevant in the future.
The somewhat unusual spelling is to make sure this does not show
up when grepping for getPointerElementType.
Auto-upgrades that rely on the pointer element type do not work in
opaque pointer mode. The idea behind this patch is that we can
instead work with type IDs, for which we can retain the pointer
element type. For typed pointer bitcode, we will have a distinct
type ID for pointers with distinct element type, even if there will
only be a single corresponding opaque pointer type.
The disclaimer here is that this is only the first step of the change,
and there are still more getPointerElementType() calls to remove.
I expect that two more patches will be needed:
1. Track all "contained" type IDs, which will allow us to handle
function params (which are contained in the function type) and GEPs
(which may use vectors of pointers)
2. Track type IDs for values, which is e.g. necessary to handle loads.
Differential Revision: https://reviews.llvm.org/D118694
This header is very large (3M Lines once expended) and was included in location
where dwarf-specific information were not needed.
More specifically, this commit suppresses the dependencies on
llvm/BinaryFormat/Dwarf.h in two headers: llvm/IR/IRBuilder.h and
llvm/IR/DebugInfoMetadata.h. As these headers (esp. the former) are widely used,
this has a decent impact on number of preprocessed lines generated during
compilation of LLVM, as showcased below.
This is achieved by moving some definitions back to the .cpp file, no
performance impact implied[0].
As a consequence of that patch, downstream user may need to manually some extra
files:
llvm/IR/IRBuilder.h no longer includes llvm/BinaryFormat/Dwarf.h
llvm/IR/DebugInfoMetadata.h no longer includes llvm/BinaryFormat/Dwarf.h
In some situations, codes maybe relying on the fact that
llvm/BinaryFormat/Dwarf.h was including llvm/ADT/Triple.h, this hidden
dependency now needs to be explicit.
$ clang++ -E -Iinclude -I../llvm/include ../llvm/lib/Transforms/Scalar/*.cpp -std=c++14 -fno-rtti -fno-exceptions | wc -l
after: 10978519
before: 11245451
Related Discourse thread: https://llvm.discourse.group/t/include-what-you-use-include-cleanup
[0] https://llvm-compile-time-tracker.com/compare.php?from=fa7145dfbf94cb93b1c3e610582c495cb806569b&to=995d3e326ee1d9489145e20762c65465a9caeab4&stat=instructions
Differential Revision: https://reviews.llvm.org/D118781
Ilia Diachkov