AI/rocm-systems
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

P4 to Git Change 1123622 by wchau@wchau_WINDOWS7_OCL on 2015/02/19 17:52:17

Browse Source 
ECR #399840 - OpenCL Runtime HW Debug support development

	- use device to control debugger registration and exception notification so that debug event will not be tied to any particular queue.
	- use aqlCodeInfo parameter for clHwDbgMapKernelCodeAMD() to be consistent with clHwDbgGetAqlPacketInfoAMD()

Affected files ...

... //depot/stg/opencl/drivers/opencl/api/opencl/amdocl/cl_debugger_amd.cpp#6 edit
... //depot/stg/opencl/drivers/opencl/api/opencl/amdocl/cl_debugger_amd.h#6 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpudebugmanager.cpp#7 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpudebugmanager.hpp#4 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpuvirtual.cpp#352 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpuvirtual.hpp#127 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gslbe/src/rt/GSLContext.cpp#70 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gslbe/src/rt/GSLContext.h#43 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/hwdebug.cpp#5 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/hwdebug.hpp#6 edit
This commit is contained in:
foreman
2015-02-19 19:40:35 -05:00
parent 861ef524fb
commit 3e53caa02e
8 changed files with 28 additions and 89 deletions
Download Patch File Download Diff File Expand all files Collapse all files
rocclr/runtime/device/gpu/gpuvirtual.cpp
+6 -17
View File
@@ -403,7 +403,6 @@ VirtualGPU::VirtualGPU(
, schedParamIdx_(0)
, deviceQueueSize_(0)
, hsaQueueMem_(NULL)
, useHwDebug_(false)
{
memset(&cal_, 0, sizeof(CalVirtualDesc));
for (uint i = 0; i < AllEngines; ++i) {
@@ -587,15 +586,6 @@ VirtualGPU::create(
return false;
}
// Check if HW Debug is used and register the debugger if not done yet
amd::HwDebugManager * dbgManager = dev().hwDebugMgr();
if ( dbgManager && dbgManager->isMsgBufferReady() ) {
if ( dbgManager->registerDebuggerOnQueue(this) == CL_SUCCESS ) {
useHwDebug_ = true;
}
}
return true;
}
@@ -1648,6 +1638,7 @@ VirtualGPU::submitKernelInternalHSA(
uint64_t vmDefQueue = 0;
amd::DeviceQueue* defQueue = kernel.program().context().defDeviceQueue(dev());
VirtualGPU* gpuDefQueue = NULL;
amd::HwDebugManager * dbgManager = dev().hwDebugMgr();
// Get the HSA kernel object
const HSAILKernel& hsaKernel =
@@ -1735,8 +1726,8 @@ VirtualGPU::submitKernelInternalHSA(
// setup the storage for the memory pointers of the kernel parameters
uint numParams = kernel.signature().numParameters();
if (useHwDebug_) {
dev().hwDebugMgr()->allocParamMemList(numParams);
if (dbgManager) {
dbgManager->allocParamMemList(numParams);
}
// Program the kernel arguments for the GPU execution
@@ -1768,7 +1759,7 @@ VirtualGPU::submitKernelInternalHSA(
HwDbgKernelInfo kernelInfo;
HwDbgKernelInfo *pKernelInfo = NULL;
if (useHwDebug_) {
if (dbgManager) {
buildKernelInfo(hsaKernel, aqlPkt, kernelInfo, enqueueEvent);
pKernelInfo = &kernelInfo;
}
@@ -1778,10 +1769,8 @@ VirtualGPU::submitKernelInternalHSA(
runAqlDispatch(gpuEvent, aqlPkt, vmMems(), cal_.memCount_,
scratch, scratchOffset, hsaKernel.cpuAqlCode(), hsaQueueMem_->vmAddress(), pKernelInfo);
if (useHwDebug_) {
if (NULL != dev().hwDebugMgr()->postDispatchCallBackFunc()) {
dev().hwDebugMgr()->executePostDispatchCallBack();
}
if (dbgManager && (NULL != dbgManager->postDispatchCallBackFunc())) {
dbgManager->executePostDispatchCallBack();
}
if (hsaKernel.dynamicParallelism()) {