From 1af35a604453768eb33e99466cd29dacd37eaa78 Mon Sep 17 00:00:00 2001
From: vsytch <vsytch@hotmail.com>
Date: Mon, 27 Jan 2020 19:37:00 -0500
Subject: [PATCH] Add missing texturePitchAlignment member to the
 hipDeviceProp_t struct. (#1802)

* Add missing texturePitchAlignment member to the hipDeviceProp_t struct.

* Add missing hipDeviceAttributeTexturePitchAlignment enumerator to the hipDeviceAttribute_t enum.

* Initialize texturePitchAlignment to 256. This works for gfx9+, but is technically overaligned in most cases for pre-gfx9.

* Add the texturePitchAlignment property to the NVCC path.


[ROCm/hip commit: f72a669487dd352e45321c4b3038f8fe2365c236]
---
 projects/hip/include/hip/hip_runtime_api.h             | 2 ++
 projects/hip/include/hip/nvcc_detail/hip_runtime_api.h | 4 ++++
 projects/hip/src/hip_device.cpp                        | 5 ++++-
 projects/hip/src/hip_hcc.cpp                           | 1 +
 4 files changed, 11 insertions(+), 1 deletion(-)

diff --git a/projects/hip/include/hip/hip_runtime_api.h b/projects/hip/include/hip/hip_runtime_api.h
index e725846cbd..64b2a85d8a 100644
--- a/projects/hip/include/hip/hip_runtime_api.h
+++ b/projects/hip/include/hip/hip_runtime_api.h
@@ -124,6 +124,7 @@ typedef struct hipDeviceProp_t {
     unsigned int* hdpRegFlushCntl;      ///< Addres of HDP_REG_COHERENCY_FLUSH_CNTL register
     size_t memPitch;                 ///<Maximum pitch in bytes allowed by memory copies
     size_t textureAlignment;         ///<Alignment requirement for textures
+    size_t texturePitchAlignment;    ///<Pitch alignment requirement for texture references bound to pitched memory
     int kernelExecTimeoutEnabled;    ///<Run time limit for kernels executed on the device
     int ECCEnabled;                  ///<Device has ECC support enabled
     int tccDriver;                   ///< 1:If device is Tesla device using TCC driver, else 0
@@ -321,6 +322,7 @@ typedef enum hipDeviceAttribute_t {
 
     hipDeviceAttributeMaxPitch,             ///< Maximum pitch in bytes allowed by memory copies
     hipDeviceAttributeTextureAlignment,     ///<Alignment requirement for textures
+    hipDeviceAttributeTexturePitchAlignment, ///<Pitch alignment requirement for 2D texture references bound to pitched memory;
     hipDeviceAttributeKernelExecTimeout,    ///<Run time limit for kernels executed on the device
     hipDeviceAttributeCanMapHostMemory,     ///<Device can map host memory into device address space
     hipDeviceAttributeEccEnabled            ///<Device has ECC support enabled
diff --git a/projects/hip/include/hip/nvcc_detail/hip_runtime_api.h b/projects/hip/include/hip/nvcc_detail/hip_runtime_api.h
index 0e7c936fc7..05a162478a 100644
--- a/projects/hip/include/hip/nvcc_detail/hip_runtime_api.h
+++ b/projects/hip/include/hip/nvcc_detail/hip_runtime_api.h
@@ -1130,6 +1130,7 @@ inline static hipError_t hipGetDeviceProperties(hipDeviceProp_t* p_prop, int dev
 
     p_prop->memPitch                 = cdprop.memPitch;
     p_prop->textureAlignment         = cdprop.textureAlignment;
+    p_prop->texturePitchAlignment    = cdprop.texturePitchAlignment;
     p_prop->kernelExecTimeoutEnabled = cdprop.kernelExecTimeoutEnabled;
     p_prop->ECCEnabled               = cdprop.ECCEnabled;
     p_prop->tccDriver                = cdprop.tccDriver;
@@ -1244,6 +1245,9 @@ inline static hipError_t hipDeviceGetAttribute(int* pi, hipDeviceAttribute_t att
         case hipDeviceAttributeTextureAlignment:
             cdattr = cudaDevAttrTextureAlignment;
             break;
+        case hipDeviceAttributeTexturePitchAlignment:
+            cdattr = cudaDevAttrTexturePitchAlignment;
+            break;
         case hipDeviceAttributeKernelExecTimeout:
             cdattr = cudaDevAttrKernelExecTimeout;
             break;
diff --git a/projects/hip/src/hip_device.cpp b/projects/hip/src/hip_device.cpp
index 403194483a..aa89e62271 100644
--- a/projects/hip/src/hip_device.cpp
+++ b/projects/hip/src/hip_device.cpp
@@ -312,9 +312,12 @@ hipError_t ihipDeviceGetAttribute(int* pi, hipDeviceAttribute_t attr, int device
             case hipDeviceAttributeMaxPitch:
                 *pi = prop->memPitch;
                 break;
-	    case hipDeviceAttributeTextureAlignment:
+            case hipDeviceAttributeTextureAlignment:
                 *pi = prop->textureAlignment;
                 break;
+            case hipDeviceAttributeTexturePitchAlignment:
+                *pi = prop->texturePitchAlignment;
+                 break;
             case hipDeviceAttributeKernelExecTimeout:
                 *pi = prop->kernelExecTimeoutEnabled;
                 break;
diff --git a/projects/hip/src/hip_hcc.cpp b/projects/hip/src/hip_hcc.cpp
index 63bc8fe14f..175d301ee1 100644
--- a/projects/hip/src/hip_hcc.cpp
+++ b/projects/hip/src/hip_hcc.cpp
@@ -925,6 +925,7 @@ hipError_t ihipDevice_t::initProperties(hipDeviceProp_t* prop) {
 
     prop->memPitch = INT_MAX; //Maximum pitch in bytes allowed by memory copies (hardcoded 128 bytes in hipMallocPitch)
     prop->textureAlignment = 0; //Alignment requirement for textures
+    prop->texturePitchAlignment = IMAGE_PITCH_ALIGNMENT; //Alignment requirment for texture pitch
     prop->kernelExecTimeoutEnabled = 0; //no run time limit for running kernels on device
 
     hsa_isa_t isa;