P4 to Git Change 1398097 by lmoriche@lmoriche_opencl_dev2 on 2017/04/13 13:01:56
SWDEV-102733 - [OCL-LC-ROCm] Cmake build Write CMakeLists.txt to enable building with and without the DK environment - Change the coding convention of the runtime files. Use Google's Style (https://google.github.io/styleguide/cppguide.html). Affected files ... ... //depot/stg/opencl/drivers/opencl/.clang-format#1 add ... //depot/stg/opencl/drivers/opencl/api/opencl/amdocl/cl_agent_amd.h#2 edit ... //depot/stg/opencl/drivers/opencl/api/opencl/amdocl/cl_command.cpp#13 edit ... //depot/stg/opencl/drivers/opencl/api/opencl/amdocl/cl_context.cpp#53 edit ... //depot/stg/opencl/drivers/opencl/api/opencl/amdocl/cl_counter.cpp#2 edit ... //depot/stg/opencl/drivers/opencl/api/opencl/amdocl/cl_d3d10.cpp#15 edit ... //depot/stg/opencl/drivers/opencl/api/opencl/amdocl/cl_d3d11.cpp#22 edit ... //depot/stg/opencl/drivers/opencl/api/opencl/amdocl/cl_d3d9.cpp#32 edit ... //depot/stg/opencl/drivers/opencl/api/opencl/amdocl/cl_debugger_amd.cpp#8 edit ... //depot/stg/opencl/drivers/opencl/api/opencl/amdocl/cl_debugger_amd.h#7 edit ... //depot/stg/opencl/drivers/opencl/api/opencl/amdocl/cl_device.cpp#61 edit ... //depot/stg/opencl/drivers/opencl/api/opencl/amdocl/cl_event.cpp#10 edit ... //depot/stg/opencl/drivers/opencl/api/opencl/amdocl/cl_execute.cpp#23 edit ... //depot/stg/opencl/drivers/opencl/api/opencl/amdocl/cl_gl.cpp#53 edit ... //depot/stg/opencl/drivers/opencl/api/opencl/amdocl/cl_icd.cpp#27 edit ... //depot/stg/opencl/drivers/opencl/api/opencl/amdocl/cl_icd_amd.h#18 edit ... //depot/stg/opencl/drivers/opencl/api/opencl/amdocl/cl_kernel.h#24 edit ... //depot/stg/opencl/drivers/opencl/api/opencl/amdocl/cl_kernel_info_amd.cpp#3 edit ... //depot/stg/opencl/drivers/opencl/api/opencl/amdocl/cl_kernel_info_amd.h#4 edit ... //depot/stg/opencl/drivers/opencl/api/opencl/amdocl/cl_lqdflash_amd.cpp#17 edit ... //depot/stg/opencl/drivers/opencl/api/opencl/amdocl/cl_lqdflash_amd.h#6 edit ... //depot/stg/opencl/drivers/opencl/api/opencl/amdocl/cl_memobj.cpp#81 edit ... //depot/stg/opencl/drivers/opencl/api/opencl/amdocl/cl_object.cpp#3 edit ... //depot/stg/opencl/drivers/opencl/api/opencl/amdocl/cl_pipe.cpp#6 edit ... //depot/stg/opencl/drivers/opencl/api/opencl/amdocl/cl_platform_amd.cpp#2 edit ... //depot/stg/opencl/drivers/opencl/api/opencl/amdocl/cl_platform_amd.h#2 edit ... //depot/stg/opencl/drivers/opencl/api/opencl/amdocl/cl_profile_amd.cpp#3 edit ... //depot/stg/opencl/drivers/opencl/api/opencl/amdocl/cl_profile_amd.h#2 edit ... //depot/stg/opencl/drivers/opencl/api/opencl/amdocl/cl_program.cpp#41 edit ... //depot/stg/opencl/drivers/opencl/api/opencl/amdocl/cl_sampler.cpp#6 edit ... //depot/stg/opencl/drivers/opencl/api/opencl/amdocl/cl_sdi_amd.cpp#3 edit ... //depot/stg/opencl/drivers/opencl/api/opencl/amdocl/cl_sdi_amd.h#2 edit ... //depot/stg/opencl/drivers/opencl/api/opencl/amdocl/cl_semaphore_amd.h#3 edit ... //depot/stg/opencl/drivers/opencl/api/opencl/amdocl/cl_svm.cpp#20 edit ... //depot/stg/opencl/drivers/opencl/api/opencl/amdocl/cl_thread_trace_amd.cpp#8 edit ... //depot/stg/opencl/drivers/opencl/api/opencl/amdocl/cl_thread_trace_amd.h#4 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/appprofile.cpp#17 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/appprofile.hpp#12 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/blit.cpp#4 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/blit.hpp#4 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/blitcl.cpp#11 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/cpu/cpubinary.cpp#11 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/cpu/cpubinary.hpp#4 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/cpu/cpubuiltins.cpp#13 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/cpu/cpubuiltins.hpp#4 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/cpu/cpucommand.cpp#66 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/cpu/cpucommand.hpp#40 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/cpu/cpudevice.cpp#280 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/cpu/cpudevice.hpp#96 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/cpu/cpufeat.hpp#3 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/cpu/cpukernel.hpp#8 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/cpu/cpumapping.cpp#6 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/cpu/cpumapping.hpp#4 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/cpu/cpuprogram.cpp#70 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/cpu/cpuprogram.hpp#14 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/cpu/cpusettings.cpp#33 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/cpu/cpusettings.hpp#2 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/cpu/cputables.hpp#5 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/cpu/cpuvirtual.cpp#26 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/cpu/cpuvirtual.hpp#13 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/device.cpp#209 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/device.hpp#284 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpuappprofile.cpp#12 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpuappprofile.hpp#7 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpubinary.cpp#58 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpubinary.hpp#27 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpublit.cpp#126 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpublit.hpp#41 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpucompiler.cpp#156 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpuconstbuf.cpp#10 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpuconstbuf.hpp#7 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpucounters.cpp#12 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpucounters.hpp#9 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpudebugger.hpp#7 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpudebugmanager.cpp#10 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpudebugmanager.hpp#6 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpudefs.hpp#147 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpudevice.cpp#567 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpudevice.hpp#163 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpukernel.cpp#318 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpukernel.hpp#126 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpumemory.cpp#131 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpumemory.hpp#50 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpuprintf.cpp#44 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpuprintf.hpp#15 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpuprogram.cpp#232 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpuprogram.hpp#69 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpuresource.cpp#238 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpuresource.hpp#87 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpusched.hpp#19 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpuschedcl.cpp#35 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpuscsi.cpp#37 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpusettings.cpp#350 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpusettings.hpp#98 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gputhreadtrace.cpp#9 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gputhreadtrace.hpp#7 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gputimestamp.cpp#27 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gputimestamp.hpp#16 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gputrap.hpp#4 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpuvirtual.cpp#410 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpuvirtual.hpp#140 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpuwavelimiter.cpp#13 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpuwavelimiter.hpp#9 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/hwdebug.cpp#7 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/hwdebug.hpp#8 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palappprofile.cpp#2 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palappprofile.hpp#3 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palbinary.cpp#2 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palbinary.hpp#3 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palblit.cpp#13 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palblit.hpp#5 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palcompiler.cpp#15 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palconstbuf.cpp#2 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palconstbuf.hpp#3 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palcounters.cpp#11 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palcounters.hpp#9 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/pal/paldebugger.hpp#3 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/pal/paldebugmanager.cpp#2 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/pal/paldebugmanager.hpp#3 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/pal/paldefs.hpp#16 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/pal/paldevice.cpp#45 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/pal/paldevice.hpp#16 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/pal/paldeviced3d10.cpp#2 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/pal/paldeviced3d11.cpp#2 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/pal/paldeviced3d9.cpp#2 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/pal/paldevicegl.cpp#4 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palkernel.cpp#34 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palkernel.hpp#11 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palmemory.cpp#13 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palmemory.hpp#3 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palprintf.cpp#5 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palprintf.hpp#3 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palprogram.cpp#39 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palprogram.hpp#17 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palresource.cpp#28 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palresource.hpp#12 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palsched.hpp#3 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palschedcl.cpp#4 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palsettings.cpp#24 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palsettings.hpp#10 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palthreadtrace.cpp#3 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palthreadtrace.hpp#5 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/pal/paltimestamp.cpp#2 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/pal/paltimestamp.hpp#3 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/pal/paltrap.hpp#2 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palvirtual.cpp#48 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palvirtual.hpp#21 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palwavelimiter.cpp#3 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palwavelimiter.hpp#4 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/mesa_glinterop.h#4 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/rocappprofile.cpp#6 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/rocappprofile.hpp#4 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/rocbinary.hpp#4 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/rocblit.cpp#17 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/rocblit.hpp#8 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/roccompiler.cpp#32 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/roccompilerlib.cpp#6 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/roccompilerlib.hpp#5 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/rocdefs.hpp#10 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/rocdevice.cpp#48 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/rocdevice.hpp#20 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/rocglinterop.cpp#4 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/rocglinterop.hpp#4 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/rockernel.cpp#22 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/rockernel.hpp#16 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/rocmemory.cpp#15 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/rocmemory.hpp#8 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/rocprintf.cpp#7 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/rocprintf.hpp#5 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/rocprogram.cpp#64 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/rocprogram.hpp#23 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/rocregisters.hpp#4 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/rocsettings.cpp#17 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/rocsettings.hpp#8 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/rocvirtual.cpp#34 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/rocvirtual.hpp#10 edit ... //depot/stg/opencl/drivers/opencl/runtime/os/alloc.cpp#7 edit ... //depot/stg/opencl/drivers/opencl/runtime/os/alloc.hpp#4 edit ... //depot/stg/opencl/drivers/opencl/runtime/os/os.cpp#8 edit ... //depot/stg/opencl/drivers/opencl/runtime/os/os.hpp#30 edit ... //depot/stg/opencl/drivers/opencl/runtime/os/os_posix.cpp#42 edit ... //depot/stg/opencl/drivers/opencl/runtime/os/os_win32.cpp#47 edit ... //depot/stg/opencl/drivers/opencl/runtime/platform/agent.cpp#8 edit ... //depot/stg/opencl/drivers/opencl/runtime/platform/agent.hpp#6 edit ... //depot/stg/opencl/drivers/opencl/runtime/platform/command.cpp#78 edit ... //depot/stg/opencl/drivers/opencl/runtime/platform/command.hpp#83 edit ... //depot/stg/opencl/drivers/opencl/runtime/platform/commandqueue.cpp#23 edit ... //depot/stg/opencl/drivers/opencl/runtime/platform/commandqueue.hpp#18 edit ... //depot/stg/opencl/drivers/opencl/runtime/platform/context.cpp#42 edit ... //depot/stg/opencl/drivers/opencl/runtime/platform/context.hpp#26 edit ... //depot/stg/opencl/drivers/opencl/runtime/platform/counter.hpp#4 edit ... //depot/stg/opencl/drivers/opencl/runtime/platform/interop.hpp#12 edit ... //depot/stg/opencl/drivers/opencl/runtime/platform/kernel.cpp#23 edit ... //depot/stg/opencl/drivers/opencl/runtime/platform/kernel.hpp#18 edit ... //depot/stg/opencl/drivers/opencl/runtime/platform/memory.cpp#127 edit ... //depot/stg/opencl/drivers/opencl/runtime/platform/memory.hpp#100 edit ... //depot/stg/opencl/drivers/opencl/runtime/platform/ndrange.cpp#8 edit ... //depot/stg/opencl/drivers/opencl/runtime/platform/ndrange.hpp#9 edit ... //depot/stg/opencl/drivers/opencl/runtime/platform/object.cpp#2 edit ... //depot/stg/opencl/drivers/opencl/runtime/platform/object.hpp#17 edit ... //depot/stg/opencl/drivers/opencl/runtime/platform/perfctr.hpp#5 edit ... //depot/stg/opencl/drivers/opencl/runtime/platform/program.cpp#86 edit ... //depot/stg/opencl/drivers/opencl/runtime/platform/program.hpp#41 edit ... //depot/stg/opencl/drivers/opencl/runtime/platform/runtime.cpp#35 edit ... //depot/stg/opencl/drivers/opencl/runtime/platform/runtime.hpp#4 edit ... //depot/stg/opencl/drivers/opencl/runtime/platform/sampler.hpp#8 edit ... //depot/stg/opencl/drivers/opencl/runtime/platform/threadtrace.hpp#6 edit ... //depot/stg/opencl/drivers/opencl/runtime/thread/atomic.hpp#7 edit ... //depot/stg/opencl/drivers/opencl/runtime/thread/monitor.cpp#7 edit ... //depot/stg/opencl/drivers/opencl/runtime/thread/monitor.hpp#8 edit ... //depot/stg/opencl/drivers/opencl/runtime/thread/semaphore.cpp#10 edit ... //depot/stg/opencl/drivers/opencl/runtime/thread/semaphore.hpp#7 edit ... //depot/stg/opencl/drivers/opencl/runtime/thread/thread.cpp#14 edit ... //depot/stg/opencl/drivers/opencl/runtime/thread/thread.hpp#15 edit ... //depot/stg/opencl/drivers/opencl/runtime/top.hpp#26 edit ... //depot/stg/opencl/drivers/opencl/runtime/utils/concurrent.hpp#8 edit ... //depot/stg/opencl/drivers/opencl/runtime/utils/debug.cpp#5 edit ... //depot/stg/opencl/drivers/opencl/runtime/utils/debug.hpp#7 edit ... //depot/stg/opencl/drivers/opencl/runtime/utils/flags.cpp#16 edit ... //depot/stg/opencl/drivers/opencl/runtime/utils/flags.hpp#271 edit ... //depot/stg/opencl/drivers/opencl/runtime/utils/macros.hpp#8 edit ... //depot/stg/opencl/drivers/opencl/runtime/utils/util.hpp#12 edit ... //depot/stg/opencl/drivers/opencl/runtime/utils/versions.hpp#2150 edit
This commit is contained in:
foreman
@@ -33,376 +33,313 @@ class Memory;
|
||||
*/
|
||||
|
||||
GpuDebugManager::GpuDebugManager(amd::Device* device)
|
||||
: HwDebugManager(device)
|
||||
, vGpu_(NULL)
|
||||
, debugMessages_(0)
|
||||
, addressWatch_(NULL)
|
||||
, addressWatchSize_(0)
|
||||
, oclEventHandle_(NULL)
|
||||
{
|
||||
// Initialize the exception info and the kernel execution mode
|
||||
excpPolicy_.exceptionMask = 0x0;
|
||||
excpPolicy_.waveAction = CL_DBG_WAVES_RESUME;
|
||||
excpPolicy_.hostAction = CL_DBG_HOST_IGNORE;
|
||||
excpPolicy_.waveMode = CL_DBG_WAVEMODE_BROADCAST;
|
||||
: HwDebugManager(device),
|
||||
vGpu_(NULL),
|
||||
debugMessages_(0),
|
||||
addressWatch_(NULL),
|
||||
addressWatchSize_(0),
|
||||
oclEventHandle_(NULL) {
|
||||
// Initialize the exception info and the kernel execution mode
|
||||
excpPolicy_.exceptionMask = 0x0;
|
||||
excpPolicy_.waveAction = CL_DBG_WAVES_RESUME;
|
||||
excpPolicy_.hostAction = CL_DBG_HOST_IGNORE;
|
||||
excpPolicy_.waveMode = CL_DBG_WAVEMODE_BROADCAST;
|
||||
|
||||
execMode_.ui32All = 0;
|
||||
execMode_.ui32All = 0;
|
||||
|
||||
rtTrapHandlerInfo_.trap_.trapHandler_ = NULL;
|
||||
rtTrapHandlerInfo_.trap_.trapBuffer_ = NULL;
|
||||
rtTrapHandlerInfo_.trap_.trapHandler_ = NULL;
|
||||
rtTrapHandlerInfo_.trap_.trapBuffer_ = NULL;
|
||||
|
||||
aqlPacket_ = (hsa_kernel_dispatch_packet_t *) NULL;
|
||||
aqlPacket_ = (hsa_kernel_dispatch_packet_t*)NULL;
|
||||
|
||||
return;
|
||||
return;
|
||||
}
|
||||
|
||||
GpuDebugManager::~GpuDebugManager()
|
||||
{
|
||||
if (NULL != addressWatch_) {
|
||||
delete [] addressWatch_;
|
||||
}
|
||||
GpuDebugManager::~GpuDebugManager() {
|
||||
if (NULL != addressWatch_) {
|
||||
delete[] addressWatch_;
|
||||
}
|
||||
}
|
||||
|
||||
void
|
||||
GpuDebugManager::executePreDispatchCallBack(void* aqlPacket,
|
||||
void* toolInfo)
|
||||
{
|
||||
DebugToolInfo* info = reinterpret_cast<DebugToolInfo*>(toolInfo);
|
||||
void GpuDebugManager::executePreDispatchCallBack(void* aqlPacket, void* toolInfo) {
|
||||
DebugToolInfo* info = reinterpret_cast<DebugToolInfo*>(toolInfo);
|
||||
|
||||
aqlPacket_ = reinterpret_cast<hsa_kernel_dispatch_packet_t*>(aqlPacket);
|
||||
aqlPacket_ = reinterpret_cast<hsa_kernel_dispatch_packet_t*>(aqlPacket);
|
||||
|
||||
// Only if the pre-dispatch callback is set, will we update cache
|
||||
// flush configuration and build the memory descriptor.
|
||||
if (NULL != preDispatchCallBackFunc_) {
|
||||
// Build the scratch memory descriptor
|
||||
device()->gslCtx()->BuildScratchBufferResource(debugInfo_.scratchMemoryDescriptor_,
|
||||
info->scratchAddress_,
|
||||
info->scratchSize_);
|
||||
// Only if the pre-dispatch callback is set, will we update cache
|
||||
// flush configuration and build the memory descriptor.
|
||||
if (NULL != preDispatchCallBackFunc_) {
|
||||
// Build the scratch memory descriptor
|
||||
device()->gslCtx()->BuildScratchBufferResource(debugInfo_.scratchMemoryDescriptor_,
|
||||
info->scratchAddress_, info->scratchSize_);
|
||||
|
||||
// Build the global memory descriptor
|
||||
device()->gslCtx()->BuildHeapBufferResource(debugInfo_.globalMemoryDescriptor_,
|
||||
info->globalAddress_);
|
||||
// Build the global memory descriptor
|
||||
device()->gslCtx()->BuildHeapBufferResource(debugInfo_.globalMemoryDescriptor_,
|
||||
info->globalAddress_);
|
||||
|
||||
// // for invalidate cache (BuildEndOfKernelNotifyCommands)
|
||||
// aqlPacket->release_fence_scope = 2;
|
||||
// // for invalidate cache (BuildEndOfKernelNotifyCommands)
|
||||
// aqlPacket->release_fence_scope = 2;
|
||||
|
||||
aclBinary_ = reinterpret_cast<void*>(info->aclBinary_);
|
||||
oclEventHandle_ = reinterpret_cast<void*>(as_cl(info->event_));
|
||||
aclBinary_ = reinterpret_cast<void*>(info->aclBinary_);
|
||||
oclEventHandle_ = reinterpret_cast<void*>(as_cl(info->event_));
|
||||
|
||||
cl_device_id clDeviceId = as_cl(device_);
|
||||
preDispatchCallBackFunc_(clDeviceId,
|
||||
oclEventHandle_,
|
||||
aqlPacket_,
|
||||
aclBinary_,
|
||||
preDispatchCallBackArgs_);
|
||||
}
|
||||
cl_device_id clDeviceId = as_cl(device_);
|
||||
preDispatchCallBackFunc_(clDeviceId, oclEventHandle_, aqlPacket_, aclBinary_,
|
||||
preDispatchCallBackArgs_);
|
||||
}
|
||||
|
||||
// setup the trap handler information only if the debugger has been registered
|
||||
if (isRegistered()) {
|
||||
// Copy the various info set by the debugger/profiler to the tool info structure
|
||||
setupTrapInformation(info);
|
||||
}
|
||||
// setup the trap handler information only if the debugger has been registered
|
||||
if (isRegistered()) {
|
||||
// Copy the various info set by the debugger/profiler to the tool info structure
|
||||
setupTrapInformation(info);
|
||||
}
|
||||
}
|
||||
|
||||
void
|
||||
GpuDebugManager::executePostDispatchCallBack()
|
||||
{
|
||||
if (NULL != postDispatchCallBackFunc_) {
|
||||
cl_device_id clDeviceId = as_cl(device_);
|
||||
postDispatchCallBackFunc_(clDeviceId,
|
||||
aqlPacket_->completion_signal.handle,
|
||||
postDispatchCallBackArgs_);
|
||||
}
|
||||
void GpuDebugManager::executePostDispatchCallBack() {
|
||||
if (NULL != postDispatchCallBackFunc_) {
|
||||
cl_device_id clDeviceId = as_cl(device_);
|
||||
postDispatchCallBackFunc_(clDeviceId, aqlPacket_->completion_signal.handle,
|
||||
postDispatchCallBackArgs_);
|
||||
}
|
||||
}
|
||||
|
||||
//! Map the kernel code for host access
|
||||
void
|
||||
GpuDebugManager::mapKernelCode(void* aqlCodeInfo) const
|
||||
{
|
||||
AqlCodeInfo* codeInfo = reinterpret_cast<AqlCodeInfo*>(aqlCodeInfo);
|
||||
void GpuDebugManager::mapKernelCode(void* aqlCodeInfo) const {
|
||||
AqlCodeInfo* codeInfo = reinterpret_cast<AqlCodeInfo*>(aqlCodeInfo);
|
||||
|
||||
codeInfo->aqlCode_ = reinterpret_cast<amd_kernel_code_t*>(aqlCodeAddr_);
|
||||
codeInfo->aqlCodeSize_ = aqlCodeSize_;
|
||||
codeInfo->aqlCode_ = reinterpret_cast<amd_kernel_code_t*>(aqlCodeAddr_);
|
||||
codeInfo->aqlCodeSize_ = aqlCodeSize_;
|
||||
}
|
||||
|
||||
cl_int
|
||||
GpuDebugManager::registerDebugger(amd::Context* context, uintptr_t messageStorage)
|
||||
cl_int GpuDebugManager::registerDebugger(amd::Context* context, uintptr_t messageStorage) {
|
||||
if (!device()->settings().enableHwDebug_) {
|
||||
LogError("debugmanager: Register debugger error - HW DEBUG is not enable");
|
||||
return CL_DEBUGGER_REGISTER_FAILURE_AMD;
|
||||
}
|
||||
|
||||
// first time register - set the message storage, flush queue and enable hw debug
|
||||
if (!isRegistered()) {
|
||||
debugMessages_ = messageStorage;
|
||||
if (!device()->gslCtx()->registerHwDebugger(debugMessages_)) {
|
||||
LogError("debugmanager: Register debugger failed");
|
||||
return CL_OUT_OF_RESOURCES;
|
||||
}
|
||||
|
||||
isRegistered_ = true;
|
||||
|
||||
if (CL_SUCCESS != createRuntimeTrapHandler()) {
|
||||
LogError("debugmanager: Create runtime trap handler failed");
|
||||
return CL_OUT_OF_RESOURCES;
|
||||
}
|
||||
}
|
||||
|
||||
context_ = context;
|
||||
|
||||
return CL_SUCCESS;
|
||||
}
|
||||
|
||||
void GpuDebugManager::unregisterDebugger() {
|
||||
if (isRegistered()) {
|
||||
// reset the debugger registration flag
|
||||
isRegistered_ = false;
|
||||
context_ = NULL;
|
||||
}
|
||||
}
|
||||
|
||||
void GpuDebugManager::flushCache(uint32_t mask) {
|
||||
HwDbgGpuCacheMask cacheMask(mask);
|
||||
device()->xferQueue()->flushCuCaches(cacheMask);
|
||||
}
|
||||
|
||||
|
||||
void GpuDebugManager::setupTrapInformation(DebugToolInfo* toolInfo) {
|
||||
toolInfo->scratchAddress_ = 0;
|
||||
toolInfo->scratchSize_ = 0;
|
||||
toolInfo->globalAddress_ = 0;
|
||||
toolInfo->sqPerfcounterEnable_ = false;
|
||||
|
||||
// Set up trap related info in the kernel info structure to be
|
||||
// used in the kernel dispatch.
|
||||
toolInfo->exceptionMask_ = excpPolicy_.exceptionMask;
|
||||
toolInfo->gpuSingleStepMode_ = execMode_.gpuSingleStepMode;
|
||||
toolInfo->monitorMode_ = execMode_.monitorMode;
|
||||
|
||||
// The order of these three bits is determined by the definition
|
||||
// of the register COMPUTE_DISPATCH_INITIATOR
|
||||
toolInfo->cacheDisableMask_ = ((execMode_.disableL1Scalar << 2) |
|
||||
(execMode_.disableL2Cache << 1) | (execMode_.disableL1Vector));
|
||||
|
||||
toolInfo->reservedCuNum_ = execMode_.reservedCuNum;
|
||||
|
||||
toolInfo->trapHandler_ = rtTrapInfo_[kDebugTrapHandlerLocation];
|
||||
toolInfo->trapBuffer_ = rtTrapInfo_[kDebugTrapBufferLocation];
|
||||
}
|
||||
|
||||
void GpuDebugManager::getPacketAmdInfo(const void* aqlCodeInfo, void* packetInfo) const
|
||||
|
||||
{
|
||||
if (!device()->settings().enableHwDebug_) {
|
||||
LogError("debugmanager: Register debugger error - HW DEBUG is not enable");
|
||||
return CL_DEBUGGER_REGISTER_FAILURE_AMD;
|
||||
const AqlCodeInfo* codeInfo = reinterpret_cast<const AqlCodeInfo*>(aqlCodeInfo);
|
||||
|
||||
const amd_kernel_code_t* hostAqlCode = codeInfo->aqlCode_;
|
||||
|
||||
PacketAmdInfo* packet = reinterpret_cast<PacketAmdInfo*>(packetInfo);
|
||||
|
||||
const amd_kernel_code_t* akc = hostAqlCode;
|
||||
|
||||
packet->numberOfSgprs_ = akc->wavefront_sgpr_count;
|
||||
packet->numberOfVgprs_ = akc->workitem_vgpr_count;
|
||||
|
||||
// use mapped kernel_object_address for host accessing of ISA buffer
|
||||
packet->pointerToIsaBuffer_ = (char*)(hostAqlCode) + akc->kernel_code_entry_byte_offset;
|
||||
|
||||
packet->scratchBufferWaveOffset_ = akc->debug_wavefront_private_segment_offset_sgpr;
|
||||
|
||||
packet->sizeOfIsaBuffer_ = codeInfo->aqlCodeSize_;
|
||||
|
||||
packet->sizeOfStaticGroupMemory_ = akc->workgroup_group_segment_byte_size;
|
||||
|
||||
// The trap_reserved_vgpr_index will be 4 less the original
|
||||
// This value must be used only by the debugger
|
||||
packet->trapReservedVgprIndex_ = akc->workitem_vgpr_count - NumberReserveVgprs;
|
||||
}
|
||||
|
||||
DebugEvent GpuDebugManager::createDebugEvent(const bool autoReset) {
|
||||
// create the event object
|
||||
osEventHandle shaderEvent = osEventCreate(!autoReset);
|
||||
|
||||
// event object has been created, set the initial state
|
||||
if (shaderEvent != 0) {
|
||||
osEventReset(shaderEvent); // initial state is non-signaled
|
||||
|
||||
if (device()->gslCtx()->exceptionNotification(shaderEvent)) {
|
||||
return shaderEvent;
|
||||
}
|
||||
}
|
||||
|
||||
// first time register - set the message storage, flush queue and enable hw debug
|
||||
if (!isRegistered()) {
|
||||
debugMessages_ = messageStorage;
|
||||
if (!device()->gslCtx()->registerHwDebugger(debugMessages_)) {
|
||||
LogError("debugmanager: Register debugger failed");
|
||||
return CL_OUT_OF_RESOURCES;
|
||||
}
|
||||
|
||||
isRegistered_ = true;
|
||||
|
||||
if (CL_SUCCESS != createRuntimeTrapHandler()) {
|
||||
LogError("debugmanager: Create runtime trap handler failed");
|
||||
return CL_OUT_OF_RESOURCES;
|
||||
}
|
||||
}
|
||||
|
||||
context_ = context;
|
||||
return 0;
|
||||
}
|
||||
|
||||
cl_int GpuDebugManager::waitDebugEvent(DebugEvent pEvent, uint32_t timeOut) const {
|
||||
if (osEventTimedWait(pEvent, timeOut)) {
|
||||
return CL_SUCCESS;
|
||||
} else {
|
||||
return CL_EVENT_TIMEOUT_AMD;
|
||||
}
|
||||
}
|
||||
|
||||
void
|
||||
GpuDebugManager::unregisterDebugger()
|
||||
{
|
||||
if (isRegistered()) {
|
||||
// reset the debugger registration flag
|
||||
isRegistered_ = false;
|
||||
context_ = NULL;
|
||||
}
|
||||
void GpuDebugManager::destroyDebugEvent(DebugEvent* pEvent) {
|
||||
osEventDestroy(*pEvent);
|
||||
*pEvent = 0;
|
||||
|
||||
device()->gslCtx()->exceptionNotification(0);
|
||||
}
|
||||
|
||||
void
|
||||
GpuDebugManager::flushCache(uint32_t mask)
|
||||
{
|
||||
HwDbgGpuCacheMask cacheMask(mask);
|
||||
device()->xferQueue()->flushCuCaches(cacheMask);
|
||||
void GpuDebugManager::wavefrontControl(uint32_t waveAction, uint32_t waveMode, uint32_t trapId,
|
||||
void* waveAddr) const {
|
||||
device()->gslCtx()->executeSqCommand(waveAction, waveMode, trapId, waveAddr);
|
||||
}
|
||||
|
||||
void GpuDebugManager::setAddressWatch(uint32_t numWatchPoints, void** watchAddress,
|
||||
uint64_t* watchMask, uint64_t* watchMode, DebugEvent* event) {
|
||||
size_t requiredSize = numWatchPoints * sizeof(HwDbgAddressWatch);
|
||||
|
||||
void
|
||||
GpuDebugManager::setupTrapInformation(DebugToolInfo* toolInfo)
|
||||
{
|
||||
toolInfo->scratchAddress_ = 0;
|
||||
toolInfo->scratchSize_ = 0;
|
||||
toolInfo->globalAddress_ = 0;
|
||||
toolInfo->sqPerfcounterEnable_ = false;
|
||||
// previously allocated size is not big enough, allocate new memory
|
||||
if (addressWatchSize_ < requiredSize) {
|
||||
if (NULL != addressWatch_) { // free the smaller address watch storage
|
||||
delete[] addressWatch_;
|
||||
}
|
||||
addressWatch_ = new HwDbgAddressWatch[numWatchPoints];
|
||||
addressWatchSize_ = requiredSize;
|
||||
}
|
||||
|
||||
// Set up trap related info in the kernel info structure to be
|
||||
// used in the kernel dispatch.
|
||||
toolInfo->exceptionMask_ = excpPolicy_.exceptionMask;
|
||||
toolInfo->gpuSingleStepMode_ = execMode_.gpuSingleStepMode;
|
||||
toolInfo->monitorMode_ = execMode_.monitorMode;
|
||||
// fill in the address watch structure
|
||||
memset(addressWatch_, 0, addressWatchSize_);
|
||||
|
||||
// The order of these three bits is determined by the definition
|
||||
// of the register COMPUTE_DISPATCH_INITIATOR
|
||||
toolInfo->cacheDisableMask_ = ((execMode_.disableL1Scalar << 2)
|
||||
| (execMode_.disableL2Cache << 1)
|
||||
| (execMode_.disableL1Vector));
|
||||
for (uint32_t i = 0; i < numWatchPoints; i++) {
|
||||
amd::Memory* watchMem = as_amd(reinterpret_cast<cl_mem>(watchAddress[i]));
|
||||
Memory* watchMemAddress = device()->getGpuMemory(watchMem);
|
||||
|
||||
toolInfo->reservedCuNum_ = execMode_.reservedCuNum;
|
||||
addressWatch_[i].watchAddress_ = reinterpret_cast<void*>(watchMemAddress->vmAddress());
|
||||
addressWatch_[i].watchMask_ = watchMask[i];
|
||||
addressWatch_[i].watchMode_ = (cl_dbg_address_watch_mode_amd)watchMode[i];
|
||||
addressWatch_[i].event_ = (0 != event) ? event[i] : 0;
|
||||
}
|
||||
|
||||
toolInfo->trapHandler_ = rtTrapInfo_[kDebugTrapHandlerLocation];
|
||||
toolInfo->trapBuffer_ = rtTrapInfo_[kDebugTrapBufferLocation];
|
||||
// setup the watch addresses
|
||||
device()->gslCtx()->setAddressWatch(numWatchPoints, (void*)addressWatch_);
|
||||
}
|
||||
|
||||
void
|
||||
GpuDebugManager::getPacketAmdInfo(
|
||||
const void* aqlCodeInfo,
|
||||
void* packetInfo) const
|
||||
void GpuDebugManager::setGlobalMemory(amd::Memory* memObj, uint32_t offset, void* srcPtr,
|
||||
uint32_t size) {
|
||||
gpu::Memory* globalMem = device()->getGpuMemory(memObj);
|
||||
|
||||
{
|
||||
const AqlCodeInfo* codeInfo =
|
||||
reinterpret_cast<const AqlCodeInfo*>(aqlCodeInfo);
|
||||
address mappedMem = static_cast<address>(globalMem->map(NULL, 0));
|
||||
assert(mappedMem != 0);
|
||||
|
||||
const amd_kernel_code_t* hostAqlCode = codeInfo->aqlCode_;
|
||||
void* dest_ptr = reinterpret_cast<void*>(mappedMem + offset);
|
||||
memcpy(dest_ptr, srcPtr, size);
|
||||
|
||||
PacketAmdInfo* packet =
|
||||
reinterpret_cast<PacketAmdInfo*>(packetInfo);
|
||||
|
||||
const amd_kernel_code_t* akc = hostAqlCode;
|
||||
|
||||
packet->numberOfSgprs_ = akc->wavefront_sgpr_count;
|
||||
packet->numberOfVgprs_ = akc->workitem_vgpr_count;
|
||||
|
||||
// use mapped kernel_object_address for host accessing of ISA buffer
|
||||
packet->pointerToIsaBuffer_ = (char*) (hostAqlCode) +
|
||||
akc->kernel_code_entry_byte_offset;
|
||||
|
||||
packet->scratchBufferWaveOffset_ =
|
||||
akc->debug_wavefront_private_segment_offset_sgpr;
|
||||
|
||||
packet->sizeOfIsaBuffer_ = codeInfo->aqlCodeSize_;
|
||||
|
||||
packet->sizeOfStaticGroupMemory_ = akc->workgroup_group_segment_byte_size;
|
||||
|
||||
// The trap_reserved_vgpr_index will be 4 less the original
|
||||
// This value must be used only by the debugger
|
||||
packet->trapReservedVgprIndex_ = akc->workitem_vgpr_count - NumberReserveVgprs;
|
||||
globalMem->unmap(NULL);
|
||||
}
|
||||
|
||||
DebugEvent
|
||||
GpuDebugManager::createDebugEvent(
|
||||
const bool autoReset)
|
||||
{
|
||||
// create the event object
|
||||
osEventHandle shaderEvent = osEventCreate(!autoReset);
|
||||
cl_int GpuDebugManager::createRuntimeTrapHandler() {
|
||||
size_t codeSize = 0;
|
||||
const uint32_t* rtTrapCode = NULL;
|
||||
|
||||
// event object has been created, set the initial state
|
||||
if (shaderEvent != 0) {
|
||||
if (device()->settings().viPlus_) {
|
||||
codeSize = sizeof(RuntimeTrapCodeVi);
|
||||
rtTrapCode = RuntimeTrapCodeVi;
|
||||
} else {
|
||||
codeSize = sizeof(RuntimeTrapCode);
|
||||
rtTrapCode = RuntimeTrapCode;
|
||||
}
|
||||
|
||||
osEventReset(shaderEvent); // initial state is non-signaled
|
||||
uint32_t numCodes = codeSize / sizeof(uint32_t);
|
||||
|
||||
if (device()->gslCtx()->exceptionNotification(shaderEvent)) {
|
||||
return shaderEvent;
|
||||
}
|
||||
}
|
||||
// Handle TMA corruption hw bug workaround -
|
||||
// The trap handler buffer has extra 256 bytes allocated, the TMA address
|
||||
// is stored in the first two DWORDs and the actual trap handler code
|
||||
// is stored starting at the location of 256 bytes (TbaStartOffset).
|
||||
//
|
||||
// allocate memory for the runtime trap handler (TBA) + TMA address
|
||||
uint32_t allocSize = codeSize + TbaStartOffset;
|
||||
|
||||
return 0;
|
||||
}
|
||||
Memory* rtTBA = new Memory(*device(), allocSize);
|
||||
runtimeTBA_ = rtTBA;
|
||||
|
||||
cl_int
|
||||
GpuDebugManager::waitDebugEvent(
|
||||
DebugEvent pEvent,
|
||||
uint32_t timeOut) const
|
||||
{
|
||||
if (osEventTimedWait(pEvent, timeOut)) {
|
||||
return CL_SUCCESS;
|
||||
}
|
||||
else {
|
||||
return CL_EVENT_TIMEOUT_AMD;
|
||||
}
|
||||
}
|
||||
if ((rtTBA == NULL) || !rtTBA->create(Resource::RemoteUSWC)) {
|
||||
return CL_OUT_OF_RESOURCES;
|
||||
}
|
||||
address tbaAddress = reinterpret_cast<address>(rtTBA->map(NULL));
|
||||
|
||||
void
|
||||
GpuDebugManager::destroyDebugEvent(DebugEvent* pEvent)
|
||||
{
|
||||
osEventDestroy(*pEvent);
|
||||
*pEvent = 0;
|
||||
// allocate buffer for the runtime trap handler buffer (TMA)
|
||||
uint32_t tmaSize = 0x100;
|
||||
Memory* rtTMA = new Memory(*device(), tmaSize);
|
||||
runtimeTMA_ = rtTMA;
|
||||
|
||||
device()->gslCtx()->exceptionNotification(0);
|
||||
if ((rtTMA == NULL) || !rtTMA->create(Resource::RemoteUSWC)) {
|
||||
return CL_OUT_OF_RESOURCES;
|
||||
}
|
||||
|
||||
}
|
||||
uint64_t rtTmaAddress = rtTMA->vmAddress();
|
||||
if ((rtTBA->vmAddress() & 0xFF) != 0 || (rtTmaAddress & 0xFF) != 0) {
|
||||
LogError("debugmanager: Trap handler/buffer is not 256-byte aligned");
|
||||
return CL_INVALID_VALUE;
|
||||
}
|
||||
|
||||
void
|
||||
GpuDebugManager::wavefrontControl(
|
||||
uint32_t waveAction,
|
||||
uint32_t waveMode,
|
||||
uint32_t trapId,
|
||||
void* waveAddr) const
|
||||
{
|
||||
device()->gslCtx()->executeSqCommand(waveAction, waveMode, trapId, waveAddr);
|
||||
}
|
||||
// store the TMA address at the beginning of trap handler buffer
|
||||
uint64_t* tbaStorage = reinterpret_cast<uint64_t*>(tbaAddress);
|
||||
tbaStorage[0] = rtTmaAddress;
|
||||
|
||||
void
|
||||
GpuDebugManager::setAddressWatch(
|
||||
uint32_t numWatchPoints,
|
||||
void** watchAddress,
|
||||
uint64_t* watchMask,
|
||||
uint64_t* watchMode,
|
||||
DebugEvent* event)
|
||||
{
|
||||
size_t requiredSize = numWatchPoints * sizeof(HwDbgAddressWatch);
|
||||
// save the trap handler code
|
||||
uint32_t* trapHandlerPtr = (uint32_t*)(tbaAddress + TbaStartOffset);
|
||||
for (uint32_t i = 0; i < numCodes; i++) {
|
||||
trapHandlerPtr[i] = rtTrapCode[i];
|
||||
}
|
||||
|
||||
// previously allocated size is not big enough, allocate new memory
|
||||
if (addressWatchSize_ < requiredSize) {
|
||||
if (NULL != addressWatch_) { // free the smaller address watch storage
|
||||
delete [] addressWatch_;
|
||||
}
|
||||
addressWatch_ = new HwDbgAddressWatch[numWatchPoints];
|
||||
addressWatchSize_ = requiredSize;
|
||||
}
|
||||
rtTBA->unmap(NULL);
|
||||
|
||||
// fill in the address watch structure
|
||||
memset(addressWatch_, 0, addressWatchSize_);
|
||||
|
||||
for (uint32_t i = 0; i < numWatchPoints; i++)
|
||||
{
|
||||
amd::Memory* watchMem = as_amd(reinterpret_cast<cl_mem>(watchAddress[i]));
|
||||
Memory* watchMemAddress = device()->getGpuMemory(watchMem);
|
||||
|
||||
addressWatch_[i].watchAddress_ = reinterpret_cast<void*>(watchMemAddress->vmAddress());
|
||||
addressWatch_[i].watchMask_ = watchMask[i];
|
||||
addressWatch_[i].watchMode_ = (cl_dbg_address_watch_mode_amd) watchMode[i];
|
||||
addressWatch_[i].event_ = (0 != event) ? event[i] : 0;
|
||||
}
|
||||
|
||||
// setup the watch addresses
|
||||
device()->gslCtx()->setAddressWatch(numWatchPoints, (void*) addressWatch_);
|
||||
|
||||
}
|
||||
|
||||
void
|
||||
GpuDebugManager::setGlobalMemory(
|
||||
amd::Memory* memObj,
|
||||
uint32_t offset,
|
||||
void* srcPtr,
|
||||
uint32_t size)
|
||||
{
|
||||
gpu::Memory* globalMem = device()->getGpuMemory(memObj);
|
||||
|
||||
address mappedMem = static_cast<address>(globalMem->map(NULL,0));
|
||||
assert(mappedMem != 0);
|
||||
|
||||
void* dest_ptr = reinterpret_cast<void*>(mappedMem + offset);
|
||||
memcpy(dest_ptr, srcPtr, size);
|
||||
|
||||
globalMem->unmap(NULL);
|
||||
}
|
||||
|
||||
cl_int
|
||||
GpuDebugManager::createRuntimeTrapHandler()
|
||||
{
|
||||
size_t codeSize = 0;
|
||||
const uint32_t* rtTrapCode = NULL;
|
||||
|
||||
if (device()->settings().viPlus_) {
|
||||
codeSize = sizeof(RuntimeTrapCodeVi);
|
||||
rtTrapCode = RuntimeTrapCodeVi;
|
||||
}
|
||||
else {
|
||||
codeSize = sizeof(RuntimeTrapCode);
|
||||
rtTrapCode = RuntimeTrapCode;
|
||||
}
|
||||
|
||||
uint32_t numCodes = codeSize / sizeof(uint32_t);
|
||||
|
||||
// Handle TMA corruption hw bug workaround -
|
||||
// The trap handler buffer has extra 256 bytes allocated, the TMA address
|
||||
// is stored in the first two DWORDs and the actual trap handler code
|
||||
// is stored starting at the location of 256 bytes (TbaStartOffset).
|
||||
//
|
||||
// allocate memory for the runtime trap handler (TBA) + TMA address
|
||||
uint32_t allocSize = codeSize + TbaStartOffset;
|
||||
|
||||
Memory* rtTBA = new Memory(*device(), allocSize);
|
||||
runtimeTBA_ = rtTBA;
|
||||
|
||||
if ((rtTBA == NULL) || !rtTBA->create(Resource::RemoteUSWC)) {
|
||||
return CL_OUT_OF_RESOURCES;
|
||||
}
|
||||
address tbaAddress = reinterpret_cast<address>(rtTBA->map(NULL));
|
||||
|
||||
// allocate buffer for the runtime trap handler buffer (TMA)
|
||||
uint32_t tmaSize = 0x100;
|
||||
Memory* rtTMA = new Memory(*device(), tmaSize);
|
||||
runtimeTMA_ = rtTMA;
|
||||
|
||||
if ((rtTMA == NULL) || !rtTMA->create(Resource::RemoteUSWC)) {
|
||||
return CL_OUT_OF_RESOURCES;
|
||||
}
|
||||
|
||||
uint64_t rtTmaAddress = rtTMA->vmAddress();
|
||||
if ((rtTBA->vmAddress() & 0xFF) != 0 || (rtTmaAddress & 0xFF) != 0) {
|
||||
LogError("debugmanager: Trap handler/buffer is not 256-byte aligned");
|
||||
return CL_INVALID_VALUE;
|
||||
}
|
||||
|
||||
// store the TMA address at the beginning of trap handler buffer
|
||||
uint64_t* tbaStorage = reinterpret_cast<uint64_t*>(tbaAddress);
|
||||
tbaStorage[0] = rtTmaAddress;
|
||||
|
||||
// save the trap handler code
|
||||
uint32_t* trapHandlerPtr = (uint32_t*)(tbaAddress + TbaStartOffset);
|
||||
for (uint32_t i = 0; i < numCodes; i++) {
|
||||
trapHandlerPtr[i] = rtTrapCode[i];
|
||||
}
|
||||
|
||||
rtTBA->unmap(NULL);
|
||||
|
||||
return CL_SUCCESS;
|
||||
return CL_SUCCESS;
|
||||
}
|
||||
|
||||
} // namespace gpu
|
||||
|
||||
Fai riferimento in un nuovo problema
Block a user