CUDA requires that static variables be visible to the host when
offloading. However, The standard semantics of a stiatc variable dictate
that it should not be visible outside of the current file. In order to
access it from the host we need to perform "externalization" on the
static variable on the device. This requires generating a semi-unique
name that can be affixed to the variable as to not cause linker errors.
This is currently done using the CUID functionality, an MD5 hash value
set up by the clang driver. This allows us to achieve is mostly unique
ID that is unique even between multiple compilations of the same file.
However, this is not always availible. Instead, this patch uses the
unique ID from the file to generate a unique symbol name. This will
create a unique name that is consistent between the host and device side
compilations without requiring the CUID to be entered by the driver. The
one downside to this is that we are no longer stable under multiple
compilations of the same file. However, this is a very niche use-case
and is not supported by Nvidia's CUDA compiler so it likely to be good
enough.
Reviewed By: tra
Differential Revision: https://reviews.llvm.org/D125904
Since function parameters and return values are passed via param space, we
can force special alignment for values hold in it which will add vectorization
options. This change may be done if the function has private or internal
linkage. Special alignment is forced during 2 phases.
1) Instruction selection lowering. Here we use special alignment for function
prototypes (changing both own return value and parameters alignment), call
lowering (changing both callee's return value and parameters alignment).
2) IR pass nvptx-lower-args. Here we change alignment of byval parameters that
belong to param space (or are casted to it). We only handle cases when all
uses of such parameters are loads from it. For such loads, we can change the
alignment according to special type alignment and the load offset. Then,
load-store-vectorizer IR pass will perform vectorization where alignment
allows it.
Special alignment calculated as maximum from default ABI type alignment and
alignment 16. Alignment 16 is chosen because it's the maximum size of
vectorized ld.param & st.param.
Before specifying such special alignment, we should check if it is a multiple
of the alignment that the type already has. For example, if a value has an
enforced alignment of 64, default ABI alignment of 4 and special alignment
of 16, we should preserve 64.
This patch will be followed by a refactoring patch that removes duplicating
code in handling byval and non-byval arguments.
Differential Revision: https://reviews.llvm.org/D120129
Since function parameters and return values are passed via param space, we
can force special alignment for values hold in it which will add vectorization
options. This change may be done if the function has private or internal
linkage. Special alignment is forced during 2 phases.
1) Instruction selection lowering. Here we use special alignment for function
prototypes (changing both own return value and parameters alignment), call
lowering (changing both callee's return value and parameters alignment).
2) IR pass nvptx-lower-args. Here we change alignment of byval parameters that
belong to param space (or are casted to it). We only handle cases when all
uses of such parameters are loads from it. For such loads, we can change the
alignment according to special type alignment and the load offset. Then,
load-store-vectorizer IR pass will perform vectorization where alignment
allows it.
Special alignment calculated as maximum from default ABI type alignment and
alignment 16. Alignment 16 is chosen because it's the maximum size of
vectorized ld.param & st.param.
Before specifying such special alignment, we should check if it is a multiple
of the alignment that the type already has. For example, if a value has an
enforced alignment of 64, default ABI alignment of 4 and special alignment
of 16, we should preserve 64.
This patch will be followed by a refactoring patch that removes duplicating
code in handling byval and non-byval arguments.
Differential Revision: https://reviews.llvm.org/D121549
For a default visibility external linkage definition, dso_local is set for ELF
-fno-pic/-fpie and COFF and Mach-O. Since default clang -cc1 for ELF is similar
to -fpic ("PIC Level" is not set), this nuance causes unneeded binary format differences.
To make emitted IR similar, ELF -cc1 -fpic will default to -fno-semantic-interposition,
which sets dso_local for default visibility external linkage definitions.
To make this flip smooth and enable future (dso_local as definition default),
this patch replaces (function) `define ` with `define{{.*}} `,
(variable/constant/alias) `= ` with `={{.*}} `, or inserts appropriate `{{.*}} `.
Currently for explicit template function instantiation in CUDA/HIP device
compilation clang emits instantiated kernel with external linkage
and instantiated device function with internal linkage.
This is fine for -fno-gpu-rdc since there is only one TU.
However this causes duplicate symbols for kernels for -fgpu-rdc if
the same instantiation happen in multiple TU. Or missing symbols
if a device function calls an explicitly instantiated template function
in a different TU.
To make explicit template function instantiation work for
-fgpu-rdc we need to follow the C++ linkage paradigm, i.e.
use weak_odr linkage.
Differential Revision: https://reviews.llvm.org/D90311
Joseph Huber