void AMDGPUTargetELFStreamer::EmitAmdhsaKernelDescriptor(
const MCSubtargetInfo &STI, StringRef KernelName,
const amdhsa::kernel_descriptor_t &KernelDescriptor, uint64_t NextVGPR,
uint64_t NextSGPR, bool ReserveVCC, bool ReserveFlatScr,
bool ReserveXNACK) {
auto &Streamer = getStreamer();
auto &Context = Streamer.getContext();
MCSymbolELF *KernelCodeSymbol = cast<MCSymbolELF>(
Context.getOrCreateSymbol(Twine(KernelName)));
MCSymbolELF *KernelDescriptorSymbol = cast<MCSymbolELF>(
Context.getOrCreateSymbol(Twine(KernelName) + Twine(".kd")));
// Copy kernel descriptor symbol's binding, other and visibility from the
// kernel code symbol.
KernelDescriptorSymbol->setBinding(KernelCodeSymbol->getBinding());
KernelDescriptorSymbol->setOther(KernelCodeSymbol->getOther());
KernelDescriptorSymbol->setVisibility(KernelCodeSymbol->getVisibility());
// Kernel descriptor symbol's type and size are fixed.
KernelDescriptorSymbol->setType(ELF::STT_OBJECT);
KernelDescriptorSymbol->setSize(
MCConstantExpr::create(sizeof(KernelDescriptor), Context));
// The visibility of the kernel code symbol must be protected or less to allow
// static relocations from the kernel descriptor to be used.
if (KernelCodeSymbol->getVisibility() == ELF::STV_DEFAULT)
KernelCodeSymbol->setVisibility(ELF::STV_PROTECTED);
Streamer.EmitLabel(KernelDescriptorSymbol);
Streamer.EmitBytes(StringRef(
(const char*)&(KernelDescriptor),
offsetof(amdhsa::kernel_descriptor_t, kernel_code_entry_byte_offset)));
// FIXME: Remove the use of VK_AMDGPU_REL64 in the expression below. The
// expression being created is:
// (start of kernel code) - (start of kernel descriptor)
// It implies R_AMDGPU_REL64, but ends up being R_AMDGPU_ABS64.
Streamer.EmitValue(MCBinaryExpr::createSub(
MCSymbolRefExpr::create(
KernelCodeSymbol, MCSymbolRefExpr::VK_AMDGPU_REL64, Context),
MCSymbolRefExpr::create(
KernelDescriptorSymbol, MCSymbolRefExpr::VK_None, Context),
Context),
sizeof(KernelDescriptor.kernel_code_entry_byte_offset));
Streamer.EmitBytes(StringRef(
(const char*)&(KernelDescriptor) +
offsetof(amdhsa::kernel_descriptor_t, kernel_code_entry_byte_offset) +
sizeof(KernelDescriptor.kernel_code_entry_byte_offset),
sizeof(KernelDescriptor) -
offsetof(amdhsa::kernel_descriptor_t, kernel_code_entry_byte_offset) -
sizeof(KernelDescriptor.kernel_code_entry_byte_offset)));
}
void AMDGPUTargetELFStreamer::EmitAmdhsaKernelDescriptor(
const MCSubtargetInfo &STI, StringRef KernelName,
const amdhsa::kernel_descriptor_t &KernelDescriptor, uint64_t NextVGPR,
uint64_t NextSGPR, bool ReserveVCC, bool ReserveFlatScr,
bool ReserveXNACK) {
auto &Streamer = getStreamer();
auto &Context = Streamer.getContext();
MCSymbolELF *KernelDescriptorSymbol = cast<MCSymbolELF>(
Context.getOrCreateSymbol(Twine(KernelName) + Twine(".kd")));
KernelDescriptorSymbol->setBinding(ELF::STB_GLOBAL);
KernelDescriptorSymbol->setType(ELF::STT_OBJECT);
KernelDescriptorSymbol->setSize(
MCConstantExpr::create(sizeof(KernelDescriptor), Context));
MCSymbolELF *KernelCodeSymbol = cast<MCSymbolELF>(
Context.getOrCreateSymbol(Twine(KernelName)));
KernelCodeSymbol->setBinding(ELF::STB_LOCAL);
Streamer.EmitLabel(KernelDescriptorSymbol);
Streamer.EmitBytes(StringRef(
(const char*)&(KernelDescriptor),
offsetof(amdhsa::kernel_descriptor_t, kernel_code_entry_byte_offset)));
// FIXME: Remove the use of VK_AMDGPU_REL64 in the expression below. The
// expression being created is:
// (start of kernel code) - (start of kernel descriptor)
// It implies R_AMDGPU_REL64, but ends up being R_AMDGPU_ABS64.
Streamer.EmitValue(MCBinaryExpr::createSub(
MCSymbolRefExpr::create(
KernelCodeSymbol, MCSymbolRefExpr::VK_AMDGPU_REL64, Context),
MCSymbolRefExpr::create(
KernelDescriptorSymbol, MCSymbolRefExpr::VK_None, Context),
Context),
sizeof(KernelDescriptor.kernel_code_entry_byte_offset));
Streamer.EmitBytes(StringRef(
(const char*)&(KernelDescriptor) +
offsetof(amdhsa::kernel_descriptor_t, kernel_code_entry_byte_offset) +
sizeof(KernelDescriptor.kernel_code_entry_byte_offset),
sizeof(KernelDescriptor) -
offsetof(amdhsa::kernel_descriptor_t, kernel_code_entry_byte_offset) -
sizeof(KernelDescriptor.kernel_code_entry_byte_offset)));
}