Vulkan(1)用apispec生成Vulkan库

我的Vulkan.net库已在(https://github.com/bitzhuwei/Vulkan.net)开源,欢迎交流。

apispec.html

在Vulkan SDK的安装文件夹里,有一个Documentation\apispec.html文件。这是一个由代码生成的对Vulkan API的说明。它包含了Vulkan API的枚举类型、结构体、函数声明以及这一切的详细注释。

由于它是自动生成的,所以其格式非常规则。只需将少数几处<br>改为<br />,几处<col .. >改为<col .. />,就可以直接用 XElement 来加载和解析它。

由于它包含了每个枚举类型及其成员的注释,包含了每个结构体及其成员的注释,包含了每个函数声明及其参数的注释,我就想,如果我能将它转换为C#代码,那会是多么美妙的一个Vulkan库啊!

我在网上找到的几个Vulkan库,基本上都没有什么注释,这让我使用起来很不方便,严重妨碍了学习速度。很多结构体的成员类型都是粗糙的 IntPtr ,而不是具体类型的指针,这也使得用起来很麻烦。

那么就动手做自己的Vulkan库吧!

分类

首先,要将巨大的apispec.html文件里的内容分为几个类别,即C宏定义、Command(函数声明)、Enum、Extension、Flag、Handle、PFN、Scalar Type和Struct。其中的C宏定义和Extension暂时用不到,就不管了,Scalar Type数量很少,又不包含实质内容,直接手工编写即可。

我们按照Enum、Handle、Flag、PFN、Struct和Command的顺序依次分析,因为后者可能依赖前者。

Enum

我们来观察apispec.html中对Enum的描述:

<h4 id="_name_798">Name</h4>
<div class="paragraph">
<p>VkAccelerationStructureMemoryRequirementsTypeNV - Acceleration structure memory requirement type</p>
</div>
</div>
<div class="sect3">
<h4 id="_c_specification_798">C Specification</h4>
<div class="paragraph">
<p>Possible values of <code>type</code> in
<code>VkAccelerationStructureMemoryRequirementsInfoNV</code> are:,</p>
</div>
<div id="VkAccelerationStructureMemoryRequirementsTypeNV" class="listingblock">
<div class="content">
<pre class="highlight"><code class="language-c++" data-lang="c++">typedef enum VkAccelerationStructureMemoryRequirementsTypeNV {
VK_ACCELERATION_STRUCTURE_MEMORY_REQUIREMENTS_TYPE_OBJECT_NV = ,
VK_ACCELERATION_STRUCTURE_MEMORY_REQUIREMENTS_TYPE_BUILD_SCRATCH_NV = ,
VK_ACCELERATION_STRUCTURE_MEMORY_REQUIREMENTS_TYPE_UPDATE_SCRATCH_NV = ,
VK_ACCELERATION_STRUCTURE_MEMORY_REQUIREMENTS_TYPE_MAX_ENUM_NV = 0x7FFFFFFF
} VkAccelerationStructureMemoryRequirementsTypeNV;</code></pre>
</div>
</div>
</div>
<div class="sect3">
<h4 id="_description_798">Description</h4>
<div class="ulist">
<ul>
<li>
<p><code>VK_ACCELERATION_STRUCTURE_MEMORY_REQUIREMENTS_TYPE_OBJECT_NV</code>
requests the memory requirement for the <code>VkAccelerationStructureNV</code>
backing store.</p>
</li>
<li>
<p><code>VK_ACCELERATION_STRUCTURE_MEMORY_REQUIREMENTS_TYPE_BUILD_SCRATCH_NV</code>
requests the memory requirement for scratch space during the initial
build.</p>
</li>
<li>
<p><code>VK_ACCELERATION_STRUCTURE_MEMORY_REQUIREMENTS_TYPE_UPDATE_SCRATCH_NV</code>
requests the memory requirement for scratch space during an update.</p>
</li>
</ul>
</div>
</div>
<div class="sect3">
<h4 id="_see_also_798">See Also</h4>

我们将发现,对于每个Enum类型,apispec都有这样的规律:从一个<h4>Name</h4>标签开始,接下来的<p></p>标签是对这个Enum的注释,接下来的<code class="language-c++"></code>标签是这个Enum的定义;然后,从<h4>Descriptor</h4>开始到<h4>See Also</h4>结束,这两个标签之间的所有<p></p>标签,分别是Enum的某个成员的注释,而且,这个注释都是以<code>此成员的名字</code>开头(这可以用于识别此注释属于哪个成员)。

有了这些规律,就可以将其解析为C#代码了。解析代码很简单,就不解释了。

 using System;
using System.Collections.Generic;
using System.Xml.Linq; namespace ApiSpec {
class EnumsParser { static readonly char[] inLineSeparator = new char[] { ' ', '\t', '\r', '\n', };
static readonly char[] lineSeparator = new char[] { '\r', '\n' };
const string leftBrace = "{";
const string rightBrace = "}"; const string filename = "Enums.content.xml";
const string strName = "Name";
const string strCSpecification = "C Specification";
const string strDescription = "Description";
const string strSeeAlso = "See Also";
const string strDocNotes = "Document Notes"; class EnumDefinetion {
/*typedef enum VkAccelerationStructureMemoryRequirementsTypeNV {
VK_ACCELERATION_STRUCTURE_MEMORY_REQUIREMENTS_TYPE_OBJECT_NV = 0,
VK_ACCELERATION_STRUCTURE_MEMORY_REQUIREMENTS_TYPE_BUILD_SCRATCH_NV = 1,
VK_ACCELERATION_STRUCTURE_MEMORY_REQUIREMENTS_TYPE_UPDATE_SCRATCH_NV = 2,
VK_ACCELERATION_STRUCTURE_MEMORY_REQUIREMENTS_TYPE_MAX_ENUM_NV = 0x7FFFFFFF
} VkAccelerationStructureMemoryRequirementsTypeNV;
*/
public string raw; public string[] Dump() {
string[] lines = this.raw.Split(lineSeparator, StringSplitOptions.RemoveEmptyEntries);
if (lines == null || lines.Length < ) { return lines; } {
string[] parts = lines[].Split(inLineSeparator, StringSplitOptions.RemoveEmptyEntries);
lines[] = $"public enum {parts[2]} {leftBrace}";
}
{
int last = lines.Length - ;
lines[last] = $"{rightBrace}";
} return lines;
}
} class EnumItemComment {
public List<string> lstComment = new List<string>(); public Dictionary<string, string> Dump() {
Dictionary<string, string> dict = new Dictionary<string, string>();
foreach (var item in lstComment) {
int left = item.IndexOf("<code>");
int right = item.IndexOf("</code>");
if (left != - && right != -) {
string key = item.Substring(left + "<code>".Length, right - (left + "<code>".Length));
if (!dict.ContainsKey(key)) {
dict.Add(key, item);
}
}
} return dict;
}
} public static void DumpEnums() {
XElement root = XElement.Load(filename);
var lstDefinition = new List<EnumDefinetion>(); bool inside = false;
TraverseNodesEnumDefinitions(root, lstDefinition, ref inside);
var listEnumItemComment = new List<EnumItemComment>(); inside = false;
TraverseNodesEnumItemComments(root, listEnumItemComment, ref inside);
var lstEnumComment = new List<string>(); inside = false;
TraverseNodesEnumComments(root, lstEnumComment, ref inside); using (var sw = new System.IO.StreamWriter("Enums.gen.cs")) {
for (int i = ; i < lstDefinition.Count; i++) {
EnumDefinetion definition = lstDefinition[i];
//sw.WriteLine(definition.raw);
string[] definitionLines = definition.Dump();
EnumItemComment itemComment = listEnumItemComment[i];
Dictionary<string, string> item2Comment = itemComment.Dump(); sw.WriteLine($"// Enum: {i}");
string enumComment = lstEnumComment[i];
sw.WriteLine($"/// <summary>{enumComment}</summary>");
{
string line = definitionLines[];
if (line.Contains("FlagBits")) { sw.WriteLine("[Flags]"); }
sw.WriteLine(line);
}
for (int j = ; j < definitionLines.Length - ; j++) {
string line = definitionLines[j];
if (item2Comment != null) {
string strComment = ParseItemComment(line, item2Comment);
if (strComment != string.Empty) {
strComment = strComment.Replace("\r\n", "\n");
strComment = strComment.Replace("\r", "\n");
strComment = strComment.Replace("\n", $"{Environment.NewLine} /// ");
sw.WriteLine($" /// <summary>{strComment}</summary>");
}
}
sw.WriteLine(line);
}
{
string line = definitionLines[definitionLines.Length - ];
sw.WriteLine(line); // }
}
}
}
Console.WriteLine("Done");
} /*<h4 id="_name_800">Name</h4>
<div class="paragraph">
<p>VkAccessFlagBits - Bitmask specifying memory access types that will participate in a memory dependency</p>
</div>*/
private static void TraverseNodesEnumComments(XElement node, List<string> list, ref bool inside) {
if (node.Name == "h4") {
if (node.Value == "Name") {
inside = true;
}
}
else if (node.Name == "p") {
if (inside) {
string text = node.ToString();
text = text.Substring("<p>".Length, text.Length - "<p></p>".Length);
text = text.Trim();
list.Add(text);
inside = false;
}
} foreach (XElement item in node.Elements()) {
TraverseNodesEnumComments(item, list, ref inside);
}
} /* line: VK_ACCELERATION_STRUCTURE_TYPE_TOP_LEVEL_NV = 0,
*
comment: <code>VK_ACCELERATION_STRUCTURE_TYPE_TOP_LEVEL_NV</code> is a top-level
acceleration structure containing instance data referring to
bottom-level level acceleration structures.
<code>VK_ACCELERATION_STRUCTURE_TYPE_BOTTOM_LEVEL_NV</code> is a bottom-level
acceleration structure containing the AABBs or geometry to be
intersected.
*/
static readonly char[] equalSeparator = new char[] { '=', ' ', '\t', '\r', '\n', };
private static string ParseItemComment(string line, Dictionary<string, string> dict) {
string result = string.Empty;
string[] parts = line.Split(equalSeparator, StringSplitOptions.RemoveEmptyEntries);
if (parts.Length == ) {
string key = parts[];
if (dict.ContainsKey(key)) {
result = dict[key];
}
} return result;
} /// <summary>
///
/// </summary>
/// <param name="node"></param>
/// <param name="list"></param>
/// <param name="inside"></param>
private static void TraverseNodesEnumItemComments(XElement node, List<EnumItemComment> list, ref bool inside) {
if (node.Name == "h4") {
if (node.Value == "Description") {
inside = true;
var comment = new EnumItemComment();
list.Add(comment);
}
else if (node.Value == "See Also") {
inside = false;
}
}
else if (node.Name == "p") {
if (inside) {
EnumItemComment comment = list[list.Count - ];
string text = node.ToString();
text = text.Substring("<p>".Length, text.Length - "<p></p>".Length);
text = text.Trim();
comment.lstComment.Add(text);
}
} foreach (XElement item in node.Elements()) {
TraverseNodesEnumItemComments(item, list, ref inside);
}
} private static void TraverseNodesEnumDefinitions(XElement node, List<EnumDefinetion> list, ref bool inside) {
if (node.Name == "h4") {
if (node.Value == "C Specification") {
inside = true;
}
}
else if (node.Name == "code") {
if (inside) {
XAttribute attrClass = node.Attribute("class");
if (attrClass != null && attrClass.Value == "language-c++") {
string v = node.Value;
var item = new EnumDefinetion() { raw = v, };
list.Add(item);
inside = false;
}
}
} foreach (XElement item in node.Elements()) {
TraverseNodesEnumDefinitions(item, list, ref inside);
}
}
}
}

EnumsParser

解析得到了143个Enum类型,其中前2个如下:

     // Enum: 0
/// <summary>VkAccelerationStructureMemoryRequirementsTypeNV - Acceleration structure memory requirement type</summary>
public enum VkAccelerationStructureMemoryRequirementsTypeNV {
/// <summary><code>VK_ACCELERATION_STRUCTURE_MEMORY_REQUIREMENTS_TYPE_OBJECT_NV</code>
/// requests the memory requirement for the <code>VkAccelerationStructureNV</code>
/// backing store.</summary>
VK_ACCELERATION_STRUCTURE_MEMORY_REQUIREMENTS_TYPE_OBJECT_NV = ,
/// <summary><code>VK_ACCELERATION_STRUCTURE_MEMORY_REQUIREMENTS_TYPE_BUILD_SCRATCH_NV</code>
/// requests the memory requirement for scratch space during the initial
/// build.</summary>
VK_ACCELERATION_STRUCTURE_MEMORY_REQUIREMENTS_TYPE_BUILD_SCRATCH_NV = ,
/// <summary><code>VK_ACCELERATION_STRUCTURE_MEMORY_REQUIREMENTS_TYPE_UPDATE_SCRATCH_NV</code>
/// requests the memory requirement for scratch space during an update.</summary>
VK_ACCELERATION_STRUCTURE_MEMORY_REQUIREMENTS_TYPE_UPDATE_SCRATCH_NV = ,
VK_ACCELERATION_STRUCTURE_MEMORY_REQUIREMENTS_TYPE_MAX_ENUM_NV = 0x7FFFFFFF
}
// Enum: 1
/// <summary>VkAccelerationStructureTypeNV - Type of acceleration structure</summary>
public enum VkAccelerationStructureTypeNV {
/// <summary><code>VK_ACCELERATION_STRUCTURE_TYPE_TOP_LEVEL_NV</code> is a top-level
/// acceleration structure containing instance data referring to
/// bottom-level level acceleration structures.</summary>
VK_ACCELERATION_STRUCTURE_TYPE_TOP_LEVEL_NV = ,
/// <summary><code>VK_ACCELERATION_STRUCTURE_TYPE_BOTTOM_LEVEL_NV</code> is a bottom-level
/// acceleration structure containing the AABBs or geometry to be
/// intersected.</summary>
VK_ACCELERATION_STRUCTURE_TYPE_BOTTOM_LEVEL_NV = ,
VK_ACCELERATION_STRUCTURE_TYPE_MAX_ENUM_NV = 0x7FFFFFFF
}

为了保持Vulkan API的原汁原味(也为了我自己省事),Enum的成员名字就保持这么长的大写+下划线版本好了。

Handle

这里的Handle指的是Vulkan中的不透明对象提供给程序员的句柄,例如一个VkInstance类型的对象,在程序员这里看到的只是一个UInt32的句柄,它的实际内容由Vulkan内部来管理。因此这里只需找到各个Handle的名字,将其改写为一个struct即可。

在apispec.html中对Handle的描述如下:

<h3 id="_vkaccelerationstructurenv3">VkAccelerationStructureNV()</h3>

只需找到各个<h3></h3>标签,就可以找到各个Handle的名字了。解析后得到37个Handle,其中的2个Handle如下:

     // Object Handles: 1
/// <summary>VkBuffer - Opaque handle to a buffer object
/// <para>Buffers represent linear arrays of data which are used for various purposesby binding them to a graphics or compute pipeline via descriptor sets or viacertain commands, or by directly specifying them as parameters to certaincommands.</para>
/// <para>Buffers are represented by VkBuffer handles:</para>
/// </summary>
public struct VkBuffer {
public UInt64 handle;
} // Object Handles: 21
/// <summary>VkInstance - Opaque handle to an instance object
/// <para>There is no global state in Vulkan and all per-application state is storedin a VkInstance object.Creating a VkInstance object initializes the Vulkan library and allowsthe application to pass information about itself to the implementation.</para>
/// <para>Instances are represented by VkInstance handles:</para>
/// </summary>
public struct VkInstance {
public UInt32 handle;
}

对于上述这样的struct,其长度等于内部成员的长度。因此,实际上VkInstance只是UInt32的一个别名,这样的别名大大强化了类型的作用,加快了编程速度。

要注意的是,有的Handle使用UInt64,有的使用UInt32,这是不可以随意改变的,否则Vulkan会卡住不动。当然,只要字节长度相同,就可以代替,例如可以用IntPtr代替UInt32,因为两者都是4字节的。

Flag

在apispec.html中,Flag实际上是一个别名,即C语言中用 typedef 定义的一个名字。2个例子如下:

 <p>VkAccessFlags - Bitmask of VkAccessFlagBits</p>
<p>VkBufferViewCreateFlags - Reserved for future use</p>

这是目前的apispec中仅有的2种Flag的说明形式。对于它们,我们分别可以用下面的代码代替:

 using VkAccessFlags = ApiSpec.Generated.VkAccessFlagBits;
// VkBufferViewCreateFlags - Reserved for future use

解析方法也很简单,用 string.Split() 拆分一下即可。

最后得到的这些using代码,将用于后面解析的Struct和Command中。

PFN

这里的PFN是函数指针的意思,也就是C#里的delegate那一套。其解析方式与Enum十分相似,不再赘述。解析后得到了8个函数指针的定义,其中几个如下:

     // PFN: 0
/// <summary>PFN_vkAllocationFunction - Application-defined memory allocation function</summary>
public unsafe delegate void* PFN_vkAllocationFunction(
/// <summary>pUserData is the value specified for
/// VkAllocationCallbacks::pUserData in the allocator specified
/// by the application.</summary>
void* pUserData,
/// <summary>size is the size in bytes of the requested allocation.</summary>
Int32 size,
/// <summary>alignment is the requested alignment of the allocation in bytes
/// and must be a power of two.</summary>
Int32 alignment,
/// <summary>allocationScope is a VkSystemAllocationScope value
/// specifying the allocation scope of the lifetime of the allocation, as
/// described here.</summary>
VkSystemAllocationScope allocationScope);
// PFN: 1
/// <summary>PFN_vkDebugReportCallbackEXT - Application-defined debug report callback function</summary>
public unsafe delegate VkBool32 PFN_vkDebugReportCallbackEXT(
/// <summary>flags specifies the VkDebugReportFlagBitsEXT that triggered
/// this callback.</summary>
VkDebugReportFlagBitsEXT flags,
/// <summary>objectType is a VkDebugReportObjectTypeEXT value specifying
/// the type of object being used or created at the time the event was
/// triggered.</summary>
VkDebugReportObjectTypeEXT _objectType,
/// <summary>object is the object where the issue was detected.
/// If objectType is VK_DEBUG_REPORT_OBJECT_TYPE_UNKNOWN_EXT,
/// object is undefined.</summary>
UInt64 _object,
/// <summary>location is a component (layer, driver, loader) defined value that
/// specifies the location of the trigger.
/// This is an optional value.</summary>
Int32 location,
/// <summary>messageCode is a layer-defined value indicating what test
/// triggered this callback.</summary>
Int32 messageCode,
/// <summary>pLayerPrefix is a null-terminated string that is an abbreviation
/// of the name of the component making the callback.
/// pLayerPrefix is only valid for the duration of the callback.</summary>
IntPtr pLayerPrefix,
/// <summary>pMessage is a null-terminated string detailing the trigger
/// conditions.
/// pMessage is only valid for the duration of the callback.</summary>
IntPtr pMessage,
/// <summary>pUserData is the user data given when the
/// VkDebugReportCallbackEXT was created.</summary>
void* pUserData);

可以看到,函数注释和参数注释都十分详尽,看了就开心。

Struct

对于Struct的解析也与Enum类似,不再赘述。解析后得到434个结构体。其中几个如下:

     // Struct: 4
/// <summary>VkAllocationCallbacks - Structure containing callback function pointers for memory allocation
/// </summary>
public unsafe struct VkAllocationCallbacks {
/// <summary> pUserData is a value to be interpreted by the implementation of
/// the callbacks.
/// When any of the callbacks in VkAllocationCallbacks are called, the
/// Vulkan implementation will pass this value as the first parameter to the
/// callback.
/// This value can vary each time an allocator is passed into a command,
/// even when the same object takes an allocator in multiple commands.</summary>
public void* pUserData;
/// <summary> pfnAllocation is a pointer to an application-defined memory
/// allocation function of type PFN_vkAllocationFunction.</summary>
public /*PFN_vkAllocationFunction*/IntPtr pfnAllocation;
/// <summary> pfnReallocation is a pointer to an application-defined memory
/// reallocation function of type PFN_vkReallocationFunction.</summary>
public /*PFN_vkReallocationFunction*/IntPtr pfnReallocation;
/// <summary> pfnFree is a pointer to an application-defined memory free
/// function of type PFN_vkFreeFunction.</summary>
public /*PFN_vkFreeFunction*/IntPtr pfnFree;
/// <summary> pfnInternalAllocation is a pointer to an application-defined
/// function that is called by the implementation when the implementation
/// makes internal allocations, and it is of type
/// PFN_vkInternalAllocationNotification.</summary>
public /*PFN_vkInternalAllocationNotification*/IntPtr pfnInternalAllocation;
/// <summary> pfnInternalFree is a pointer to an application-defined function
/// that is called by the implementation when the implementation frees
/// internal allocations, and it is of type
/// PFN_vkInternalFreeNotification.</summary>
public /*PFN_vkInternalFreeNotification*/IntPtr pfnInternalFree;
}
// Struct: 9
/// <summary>VkApplicationInfo - Structure specifying application info
/// </summary>
public unsafe struct VkApplicationInfo {
/// <summary> sType is the type of this structure.</summary>
public VkStructureType sType;
/// <summary> pNext is NULL or a pointer to an extension-specific structure.</summary>
public /*-const-*/ void* pNext;
/// <summary> pApplicationName is NULL or is a pointer to a null-terminated
/// UTF-8 string containing the name of the application.</summary>
public IntPtr pApplicationName;
/// <summary> applicationVersion is an unsigned integer variable containing the
/// developer-supplied version number of the application.</summary>
public UInt32 applicationVersion;
/// <summary> pEngineName is NULL or is a pointer to a null-terminated UTF-8
/// string containing the name of the engine (if any) used to create the
/// application.</summary>
public IntPtr pEngineName;
/// <summary> engineVersion is an unsigned integer variable containing the
/// developer-supplied version number of the engine used to create the
/// application.</summary>
public UInt32 engineVersion;
/// <summary> apiVersion
/// must be the highest version of Vulkan that the
/// application is designed to use, encoded as described in
/// html/vkspec.html#extendingvulkan-coreversions-versionnumbers.
/// The patch version number specified in apiVersion is ignored when
/// creating an instance object.
/// Only the major and minor versions of the instance must match those
/// requested in apiVersion.</summary>
public UInt32 apiVersion;
}
// Struct: 193
/// <summary>VkInstanceCreateInfo - Structure specifying parameters of a newly created instance
/// </summary>
public unsafe struct VkInstanceCreateInfo {
/// <summary> sType is the type of this structure.</summary>
public VkStructureType sType;
/// <summary> pNext is NULL or a pointer to an extension-specific structure.</summary>
public /*-const-*/ void* pNext;
/// <summary> flags is reserved for future use.</summary>
public VkInstanceCreateFlags flags;
/// <summary> pApplicationInfo is NULL or a pointer to an instance of
/// VkApplicationInfo.
/// If not NULL, this information helps implementations recognize behavior
/// inherent to classes of applications.
/// VkApplicationInfo is defined in detail below.</summary>
public /*-const-*/ VkApplicationInfo* pApplicationInfo;
/// <summary> enabledLayerCount is the number of global layers to enable.</summary>
public UInt32 enabledLayerCount;
/// <summary> ppEnabledLayerNames is a pointer to an array of
/// enabledLayerCount null-terminated UTF-8 strings containing the
/// names of layers to enable for the created instance.
/// See the html/vkspec.html#extendingvulkan-layers section for further details.</summary>
public IntPtr /*-const-*/ * ppEnabledLayerNames;
/// <summary> enabledExtensionCount is the number of global extensions to
/// enable.</summary>
public UInt32 enabledExtensionCount;
/// <summary> ppEnabledExtensionNames is a pointer to an array of
/// enabledExtensionCount null-terminated UTF-8 strings containing the
/// names of extensions to enable.</summary>
public IntPtr /*-const-*/ * ppEnabledExtensionNames;
}

这里有几点要注意。

函数委托用在struct中后,这个struct无法使用指针形式(SomeStruct*),所以这里不得不用IntPtr代替了具体的函数委托。

在 IntPtr pApplicationName 中应当用 Marshal.StringToHGlobalAnsi(string s) 为其赋值。函数 Marshal.StringToHGlobalAnsi(string s) 会在非托管内存中为s创建一个副本,然后返回此副本的指针。这样pApplicationName才会指向一个固定位置的字符串。当然,用完后,这个副本应当用 Marshal.FreeHGlobal(IntPtr hglobal) 释放掉。为了简化这一过程,我提供一个扩展函数:

         /// <summary>
/// Set a string to specified <paramref name="target"/>.
/// </summary>
/// <param name="value"></param>
/// <param name="target">address of string.</param>
public static void Set(this string value, ref IntPtr target) {
{ // free unmanaged memory.
if (target != IntPtr.Zero) {
Marshal.FreeHGlobal(target);
target = IntPtr.Zero;
}
}
{
if (value != null && value.Length > ) {
target = Marshal.StringToHGlobalAnsi(value);
}
else {
target = IntPtr.Zero;
}
}
}

这个扩展函数会将上一次 Marshal.StringToHGlobalAnsi() 的内存释放,但是无法保证这次的。也就是说,它可以保证,最多还只需调用1次内存释放函数Marshal.FreeHGlobal(IntPtr hglobal)。

在 public IntPtr /*-const-*/ * ppEnabledLayerNames; 中也有类似的问题,这个成员指向一个IntPtr数组,这个数组的每个成员都是一个IntPtr,每个IntPtr都指向一个由 Marshal.StringToHGlobalAnsi(string s) 提供的返回值。所以这需要另一个扩展函数来简化之:

         /// <summary>
/// Set an array of structs to specified <paramref name="target"/> and <paramref name="count"/>.
/// <para>Enumeration types are not allowed to use this method.
/// If you have to, convert them to byte/short/ushort/int/uint according to their underlying types first.</para>
/// </summary>
/// <param name="value"></param>
/// <param name="target">address of first element/array.</param>
/// <param name="count">How many elements?</param>
public static void Set<T>(this T[] value, ref IntPtr target, ref UInt32 count) where T : struct {
{ // free unmanaged memory.
if (target != IntPtr.Zero) {
Marshal.FreeHGlobal(target);
target = IntPtr.Zero;
count = ;
}
}
{
count = (UInt32)value.Length; int elementSize = Marshal.SizeOf<T>();
int byteLength = (int)(count * elementSize);
IntPtr array = Marshal.AllocHGlobal(byteLength);
var dst = (byte*)array;
GCHandle pin = GCHandle.Alloc(value, GCHandleType.Pinned);
IntPtr address = Marshal.UnsafeAddrOfPinnedArrayElement(value, );
var src = (byte*)address;
for (int i = ; i < byteLength; i++) {
dst[i] = src[i];
}
pin.Free(); target = array;
}
}

在此函数参数中,我使用 ref IntPtr target ,而不是 ref T* target ,是因为C#不允许这样。编译器说,无法获取托管类型(”T”)的大小,或声明指向它的指针。那么在调用此扩展函数时,就得先创建一个临时变量 IntPtr ptr = IntPtr.Zero ,调用完扩展函数后,再将ptr赋予具体类型的指针。例如:

         var deviceInfo = new VkDeviceCreateInfo();
IntPtr ptr = IntPtr.Zero;
new VkDeviceQueueCreateInfo[] { queueInfo }.Set(ref ptr, ref deviceInfo.queueCreateInfoCount);
deviceInfo.pQueueCreateInfos = (VkDeviceQueueCreateInfo*)ptr;

好消息是,对于字符串数组string[]和(

bool、byte、short、int、long、char、sbyte、ushort、uint、ulong、float、double

)这12种特殊基础类型的数组,可以直接使用Set扩展函数。因为我专门为它们编写了特定的扩展函数。

Command

对于Command的解析也与Struct类似,不再赘述。解析后得到326个command,几个例子如下:

         // Command: 4
/// <summary>vkAllocateCommandBuffers - Allocate command buffers from an existing command pool
/// </summary>
/// <param name="device"> device is the logical device that owns the command pool.</param>
/// <param name="pAllocateInfo"> pAllocateInfo is a pointer to an instance of the
/// VkCommandBufferAllocateInfo structure describing parameters of the
/// allocation.</param>
/// <param name="pCommandBuffers"> pCommandBuffers is a pointer to an array of VkCommandBuffer
/// handles in which the resulting command buffer objects are returned.
/// The array must be at least the length specified by the
/// commandBufferCount member of pAllocateInfo.
/// Each allocated command buffer begins in the initial state.</param>
[DllImport(VulkanLibrary, CallingConvention = CallingConvention.Winapi)]
public static extern VkResult vkAllocateCommandBuffers(
VkDevice device,
/*-const-*/ VkCommandBufferAllocateInfo* pAllocateInfo,
VkCommandBuffer* pCommandBuffers);
// Command: 324
/// <summary>vkUpdateDescriptorSets - Update the contents of a descriptor set object
/// </summary>
/// <param name="device"> device is the logical device that updates the descriptor sets.</param>
/// <param name="descriptorWriteCount"> descriptorWriteCount is the number of elements in the
/// pDescriptorWrites array.</param>
/// <param name="pDescriptorWrites"> pDescriptorWrites is a pointer to an array of
/// VkWriteDescriptorSet structures describing the descriptor sets to
/// write to.</param>
/// <param name="descriptorCopyCount"> descriptorCopyCount is the number of elements in the
/// pDescriptorCopies array.</param>
/// <param name="pDescriptorCopies"> pDescriptorCopies is a pointer to an array of
/// VkCopyDescriptorSet structures describing the descriptor sets to
/// copy between.</param>
[DllImport(VulkanLibrary, CallingConvention = CallingConvention.Winapi)]
public static extern void vkUpdateDescriptorSets(
VkDevice device,
UInt32 descriptorWriteCount,
/*-const-*/ VkWriteDescriptorSet* pDescriptorWrites,
UInt32 descriptorCopyCount,
/*-const-*/ VkCopyDescriptorSet* pDescriptorCopies);

其中有一个函数使用了 void** 这个二级指针,我觉得实在难看又难用,就用 IntPtr* 代替了。

对非托管内存的管理(释放)问题

每个struct都应该自己负责自己使用的非托管资源的释放问题。给这样的struct的指针成员 T* p; 赋值时,也应当为数据复制一个副本,将副本赋值给p。这样它释放资源时,就不会影响到其它地方了。实际上,在各个扩展函数 Set(..) 中,我就是用副本赋值的。

如果struct的指针成员 T* p; 实际上只需得到1个对象,也就是说,数组中的元素只有1个,那么可以直接将此元素的地址赋值给p,并且不释放资源。例如:

     UInt32 index = ;
var info = new VkSwapchainCreateInfoKHR();
{
info.sType = VkStructureType.VK_STRUCTURE_TYPE_SWAPCHAIN_CREATE_INFO_KHR;
// other stuff ..
//new UInt32[] { 0 }.Set(ref info.QueueFamilyIndices, ref info.QueueFamilyIndexCount);
info.pQueueFamilyIndices = &index; info.queueFamilyIndexCount = ;
} VkSwapchainKHR swapchain;
vkAPI.vkCreateSwapchainKHR(device, &info, null, &swapchain);

这是稳妥、可移植、无需程序员直接写 Marshal. AllocHGlobal() 的内存管理方法。

那么,如果程序员忘记释放某些struct的资源了呢?Vulkan说,程序员应当清楚自己在做什么,不然他们何必用Vulkan。我觉得呢,这些struct不会被反复使用,因此,它们最多泄漏一点点内存,不会像服务器代码那样占用越来越多的内存,所以不碍事的。

总结

有了这么带劲的注释,整个档次都不一样了。

Vulkan(1)用apispec生成Vulkan库的更多相关文章

  1. Xcode 生成静态库相关设置:

    Xcode 生成静态库相关设置: #Build Setting1. Architectures ------- Architectures -----> $(ARCHS_STANDARD) -- ...

  2. 在Linux下如何使用GCC编译程序、简单生成 静态库及动态库

      最近在编写的一个Apache  kafka 的C/C++客户端,,在看他写的 example中,他的编译是用librdkafka++.a和librdkafka.a    静态库编译的,,,而我们这 ...

  3. [转]Linux下用gcc/g++生成静态库和动态库(Z)

    Linux下用gcc/g++生成静态库和动态库(Z) 2012-07-24 16:45:10|  分类: linux |  标签:链接库  linux  g++  gcc  |举报|字号 订阅     ...

  4. Android studio 下JNI编程实例并生成so库

    Android studio 下JNI编程实例并生成so库 因为公司需要为Android相机做美颜等图像后期处理,需要使用JNI编程,最近学了下JNI,并且在Android Studio下实现了一个小 ...

  5. 由动态库文件dll生成lib库文件(手动生成.def文件,然后使用lib命令编译,非常牛),同理可使用dll生成.a库文件

    本文基于OpenBlas的编译和安装,来说明如何从一个dll文件生成lib库文件. 参考OpenBlas的说明“Howto generate import library for MingW”,和Mi ...

  6. Android NDK入门实例 计算斐波那契数列二生成.so库文件

    上一篇文章输生成了jni头文件,里面包含了本地C代码的信息,提供我们引用的C头文件.下面实现本地代码,再用ndk-build编译生成.so库文件.由于编译时要用到make和gcc,这里很多人是通过安装 ...

  7. Linux生成动态库系统

    Linux生成动态库系统 一个.说明 Linux下动态库文件的扩展名为 ".so"(Shared Object). 依照约定,全部动态库文件名称的形式是libname.so(可能在 ...

  8. Ubuntu下通过makefile生成静态库和动态库简单实例

    本文转自http://blog.csdn.net/fengbingchun/article/details/17994489 Ubuntu环境:14.04 首先创建一个test_makefile_gc ...

  9. 如何使用 VS生成动态库?

    如何使用 VS生成动态库? //.cpp 文件默认定义了 __cplusplus 宏 #ifdef __cplusplus extern "C"{ #endif xxx #ifde ...

随机推荐

  1. 佳木斯集训Day8

    本来能AK的啊啊啊啊啊,唯一一天可以AK,却被Champion误导了(好吧实际上是我理解有问题) T1我写了俩小时,就是一道数列题,推公式的,可以二分解,我觉得二分麻烦,就直接想O(1)了 #incl ...

  2. 《SpringCloud docker》读书笔记

    yml配置意义 当Ribbon和Eureka配合使用时,会自动将虚拟主机名映射成微服务的网络地址. yml中info可以展示一些信息 server: port: 8000 # 指定端口 spring: ...

  3. java随笔之接口

    /* * 接口大致上可以分为:哑接口,抽象接口,接口类 * 哑接口:就是public,protected(注意protect有包权限,只有本包才开放接口)方法 * 抽象接口:就是哑接口变为抽象方法,在 ...

  4. 分享一个非常好用又好看的终端工具--Hyper (支持windows、MacOS、Linux)

    分享一个非常好用又好看的终端工具--Hyper 官网地址: https://hyper.is/ 打开官网,选择对应版本安装即可:(可能网络原因,无法下载, 可以从我分享的链接下载 链接: https: ...

  5. 30213Java_接口

    1.简介 接口作用 为什么需要接口?接口和抽象类的区别? 接口就是比“抽象类”还“抽象”的“抽象类”,可以更加规范的对子类进行约束.全面地专业地实现了:规范和具体实现的分离. 抽象类还提供某些具体实现 ...

  6. ZooKeeper系列(二)—— Zookeeper 单机环境和集群环境搭建

    一.单机环境搭建 1.1 下载 下载对应版本 Zookeeper,这里我下载的版本 3.4.14.官方下载地址:https://archive.apache.org/dist/zookeeper/ # ...

  7. 重读《学习JavaScript数据结构与算法-第三版》- 第4章 栈

    定场诗 金山竹影几千秋,云索高飞水自流: 万里长江飘玉带,一轮银月滚金球. 远自湖北三千里,近到江南十六州: 美景一时观不透,天缘有分画中游. 前言 本章是重读<学习JavaScript数据结构 ...

  8. RedHat 6.5换源

    https://wenku.baidu.com/view/5b87fb42c77da26924c5b03b.html

  9. 淘宝自动登录2.0,新增Cookies序列化

    前段时间时间为大家讲解了如何使用requests库模拟登录淘宝,而今天我们将对该功能进行丰富.所以我们把之前的那个版本定为1.0,而今天修改的版本定为2.0.版本的地跌意味着功能的升级,那今天的2.0 ...

  10. 关于Function和Object之间先后问题的相关研究

    文章说明,博主是一个前端小白,本片文章是博主在学习的过程中碰到的疑惑,根据查找的资料,之后得出的个人结论,文中如果出现错误,欢迎指正. -------路漫漫其修远兮吾将上下而求索,与诸君共勉----- ...