Michael Li

7 tips about building smart contract on NEO

Michael Lee2 — Wed, 20 Jun 2018 07:59:28 GMT

Build your own NEO private chain

The operation of NEO smart contracts is inseparable from the NEO blockchain support. In most cases, smart contracts need to access the underlying NEO storage services through the NEO blockchain. At the same time, when the smart contract is invoked, it needs to make a valid judgment on the information of the call initiator when it is invoked, such as verifying the signature and so on. All of these need to be completed in the NEO blockchain to complete the test. The cost of deploying smart contracts on the NEO blockchain is very expensive. A simple smart contract deployment costs 500 gas. If you need to access NEO storage services and support dynamic calls for smart contracts, the cost of deploying a smart contract is 1000Gas. Although NEO provides a test network, NEO and Gas of the test network are not easily available. Developers can apply to the NEO organization to get some test networks Gas and NEO for testing. But given the deployment of NEO's smart contracts up to 1000Gas (actually, really functional smart contracts such as ICO contracts must access NEO's underlying storage services). Even if NEO issues 20,000 Gases to developers, developers can only complete the deployment of 20 smart contracts. Therefore, it is very necessary for developers to deploy a NEO private service on their own. This reliance on 100 million NEO coins and 100 million NEO coins derived from Gas. There is no need to worry about the cost of deploying the smart contract when debugging. Another advantage of building a NEO private chain for development is that it allows developers to more easily query the transactions that have taken place. If the NEO test chain is used for debugging, each block contains not only the developer's own transaction but also the developer's transaction. Because the NEO test chain is open, all transactions that occur in the test chain are mixed together. As a result, developers find their own transaction information in the block, which increases the difficulty invisibly. Will reduce developers' development efficiency to varying degrees. At the same time, it will also expose the developer's own immature code to the public's field of vision prematurely, which will bring security risks. NEO has provided documentation to teach users how to deploy a NEO private service. Please refer to: http://docs.neo.org/en-us/network/private-chain.html After the deployment is completed, it is also necessary to expose a node that supports the RPC interface according to the exchange docking process document to provide services to developers. NEO adapting exchange documents please refer to: http://docs.neo.org/en-us/exchange/v2.7.4.html

Avoid using floating point data in smart contracts

NEO blockchain natively does not support floating point data. In order to represent a floating-point number, such as NEO's minimum unit of 0.00000001, the actual approach is that NEO will process this data by 10 million times as an integer. In the NEO virtual machine itself, it does not have a floating point data conversion byte array/byte array conversion floating point data and similar APIs. Therefore, if a floating-point data needs to be saved and read, it is very troublesome to handle it in a smart contract. The correct approach is to deal with the floating-point data of incoming/outgoing smart contracts, first become an integer to access and calculate, and so on for display, and then back to a floating-point data outside the smart contract.

Avoid string manipulation in smart contract

Since the virtual machine that runs the NEO smart contract does not have an API for handling strings, even for case conversions, the ToString() method is not supported for some other types. If a developer wants to process a string in a smart contract, he or she must read and verify it byte by byte. Even if the string processing is completed, the consumption of Gas will increase and the volume of the smart contract will also increase. The correct approach is that data needs to handle string data outside smart contracts. For some flags, you can use integers instead of strings. For some smart contract output results, it should also be replaced with an error code. It is better not to transfer strings directly.

NEO VM does not support exception handling mechanism

The execution environment of the NEO smart contract, that is, the NEO virtual machine itself does not support the exception handling mechanism. This means that try-catch segments cannot be used when writing smart contract code. The NEON compiler does not allow such code to generate AVM files by compiling. Therefore, when writing NEO smart contract code, we must pay special attention to the robustness of the code, do a good job of checking the parameters, and check the return value of the function, and avoid using the exception data to perform the next operation to avoid exceptions. Developers of NEO smart contracts should also not expect to determine the execution path of the code by catching exceptions. This does not work. The developer of the NEO smart contract should make a parameter check inside the function when writing a function and clearly define the meaning of the return value. In fact, the other two C# keywords ref and out are not supported. So don't expect a function in a smart contract to return multiple values at once.

Handle byte stream operations carefully

In the NEO virtual machine, the byte[] data is treated as a special data type and is immutable. Once you need to modify the value of a byte in the byte array, you need to regenerate a new byte array object. In the actual running process, once the code similar to the following is executed, this smart contract call will crash and return directly to "FAULT BREAK".

byte[] byteArray = new byte[]{0x0a, 0x0b, 0x0c, 0x0d}; byteArray[2] = 0x01;

The correct way is to use the Range function to truncate the previously unmodified part of the code. Add the 0x01 to the array through the Concat function, then dock the parts of the original array that do not need to be modified. code show as below:

byte[] byteArray = new byte[]{0x0a, 0x0b, 0x0c, 0x0d}; byte[] temp = byteArray.Range(0, 2); temp = temp.Concat(new byte[]{0x01}); temp = temp.Concat(byteArray.Range(3, 1));

In addition to this, there is no better way to implement changes to the byte[] element.

Need to push the stack from right to left when calling multi-parameter smart contract function

In the NEO test article (http://docs.neo.org/en-us/sc/quickstart/test.html), examples of how to invoke smart contracts with parameters are given. But that is limited to smart contract functions with a single input parameter. If you need to call smart contract functions with multiple parameters, you need to pay attention to the stacking order of function parameters. For example, the following smart contract function:

bool CallMe(string param1, int param2, bool param3, byte[] param4)

Then when the code is called, the order in which the functions are stacked should be param4, param3, param2, param1. And before the actual call, pass in the packed instruction sb.Emit(Neo.VM.OpCode.PACK); Package all the parameters. Otherwise, the smart contract function call will fail to return "FAULT BREAK" because of the mismatch of the parameter types.

Using Map class which is in devpack

NEO provides developers with a generic container object of type Map in the devpack. Developers can use Map-type objects to store a series of complex data in key-value pairs in the NEO store. Advanced packaging of Storage.Get, Storage.Put, and Storage.Delete methods for the Map type. It is easier and more convenient for developers to use Map objects to manipulate NEO underlying storage than three static methods that use Storage directly.

Edge Extension: Create a custom Edge extension

Michael Lee2 — Sat, 03 Sep 2016 12:15:40 GMT

Since Windows 10 Anniversary Update has been released, Microsoft Edge has officially support extension feature. In this blog post, I want to share you my experience about creating custom extension for Edge.

Objective: Create a custom Edge extension to convert URL to QR code for mobile app sharing.

In china everyone has a mobile IM application WeChat (looks like WhatsApp), If someone wants to share a web page to others, he usually uses WeChat scan QRCode and then send a message that includes web page link.

Tools: Visual Studio Code.

It can be download at: https://code.visualstudio.com/. It supports OS X, Linux and Windows. So it means that you can create a custom extension for Edge even you are using Linux. Now, let getting start......

Implementation Edge extension

Step 1. Create a folder on you disk eg. EdgeQRCode, and then open it by Visual Studio Code File->Open Folder...... Step 2. Create a configuration file manifest.json in EdgeQRCode folder. You can get a sample file from https://developer.microsoft.com/en-us/microsoft-edge/platform/documentation/extensions/api-support/supported-manifest-keys/json-manifest-example/ Yes, it very looks like Mozilla Web extension's configuration file. Actually Edge does not support full JSON schema as Mozilla. The key properties please check https://developer.microsoft.com/en-us/microsoft-edge/platform/documentation/extensions/api-support/supported-manifest-keys/ . In this case, I removed some keys from sample file like this: Since this extension doesn't need running at background I removed background key from sample file. All paths in configuration file are relative path. The root is EdgeQRCode folder. Step 3. Create a js folder to implement functional of Edge extension. The jquery.min.js and jquery.qrcode.min.js are downloaded from their official sites. Only index.js is custom code. For security reason, extension doesn't allow javascript embedded in html file. So we have to set event binding by addEventLinstener(). We can suppose there is a

element which id is 'qrcode' used for display QR code. And other div element with 'title' id used for display page title. Step 4. Create a index.html file for popup UI that already defined in manifest.json file. It has to claim two

elements title and qrcode in HTML body. Step 5. Complement the other elements like CSS file and icons.

Debugging

Before debugging extension, you have to enable sideloading feature on your Edge. To achieve this please follow section two of below link https://developer.microsoft.com/en-us/microsoft-edge/platform/documentation/extensions/troubleshooting/ And then open Edge menu->Extensions->Load Extension, locate EdgeQRCode folder to load. After loading finished, you have to set "Edge QRCode" on and "Show button next to the address bar" on. like this: Debugging popup UI is a little bit specific. You can use ms-browser-extension://[Extension_Id]/index.html to let Edge load your popup UI. The extension id is shown in extension properties UI. In this case the URL should be: ms-browser-extension://EdgeQRCode_08b2vmtxn4krp/index.html When html and javascript has been loaded. You can enable Edge debugging tool by press F12.

Appendix

The source code can be found here: https://github.com/micli/EdgeQRCode-Extension For more information about web extension, please visit: https://developer.mozilla.org/en-US/Add-ons/WebExtensions Edge team wants to compatible Mozilla standard, but there is a gap by now. If you want to convert an existing Chrome/Firefox extension to Edge, please check: http://bav0.com/converting-any-chrome-extension-to-edge-in-theory/ The extension running like this:

Is it a security issue on SQL Azure firewall ?

Michael Lee2 — Wed, 02 Mar 2016 09:15:37 GMT

Today I found a terrible fact: Anyone can use Azure PowerShell manipulation SQL Azure firewall without Azure login. Today, I wanted to create a firewall exception by PowerShell and I found this blog: http://blogs.msdn.com/b/umits/archive/2012/11/20/windows-azure-sql-database-how-to-manage-your-firewall-rules-with-powershell.aspx It works fine. But when PowerShell finished, I realized I never did Add-AzureAccount or Login-AzureAccount before. God, It means anyone who knows my SQL Azure Server name can freely CRUD my SQL Azure firewall rules. I can't belive my eyes. And I signed off my Azure account and did it again. Then I reproduced it. You see I executed Get-AzureSqlDatabaseServerFirewallRule after PowerShell initialized. NO login required. So I just wondering is it a security issue on SQL Azure firewall?

Something I want to share you about beta version UWP exam

Michael Lee2 — Thu, 14 Jan 2016 14:09:54 GMT

I have already taken the beta version of UWP development exams. They are: 70-354 Universal Windows Platform – App Architecture and UX/UI (beta) 70-355 Universal Windows Platform – App Data, Services, and Coding Patterns (beta) Due to its very less information for prepare examinations, I want to share some key points about these two exams. Summary: 70-354 has 56 questions, 70-355 has 59 questions. Each exam has 3 case studies. Each case study has 6~7 questions. The max time of taking each exam is 160 mins. Key points of knowledge: 1. Services in Azure: Mobile Service, OneDrive, API Service 2. Data Binding: Binding path, DataContext, ItemsSource 3. Design Patterns: Dependency Injection, Singleton, Factory, Decorate,Observation 4. OOP: public, private, protected,internal 5.Controls: AppBar,Page,Frame,List control,Splash Screen,TextBox 6.Control Events: LostFocus,PointerExited,PointerMoved,PointerPressed,PointerReleased,Item_Selected,Item_SelectedChanged 7.Data Sync: pull/push data from/to Mobile Service 8.XAML: namespace,VisualStateManager and sub elements,Control Template,Data Template,MediaElement 9.Project structure: App.xaml.cs, package.manifest 10.Source Control: VSO,Git client, branch and merge strategy 11.Sign app or assembly: sn.exe makecert.exe, pfx 12..NET basic: assembly versioning,GAC,assembly binding 13.Security: Role-based security, full trust, PasswordVault, Credential class 14.Testing:Unit Test,Test Manager,Performance Wizard 15.Store: pricing strategy,App validate strategy 16.Background Task: IBackgroundTask,Triggers,Schedule Job(Mobile Service) 17.Notifications: Toast,Tile 18.Cortana: voice commands definition 19. Authentication: Azure AD integration 20. Ink: InkCanvas, How to detect pen device? That's all my remembered. Enjoy your testing experience!

The new Visual Studio 2015 extension for Linux Debug

Michael Lee2 — Mon, 26 Oct 2015 23:15:00 GMT

Recently, Microsoft Visual Studio team has released a amazing VC++ extension to support remote debugging Linux application by Visual Studio 2015. Using this extension you will have unified debugging experience on Windows, Linux even Android. By this feature, we can debugging a Linux application looks like debugging native app on Windows. The extension can offer developers Quick/Add Watch, Run to cursor, Go into and so on.

Please follow this steps to setup debugging environment:

1. Please install VSLinuxDebugger.visx extension. It can be download from:

https://visualstudiogallery.msdn.microsoft.com/35dbae07-8c1a-4f9d-94b7-bac16cad9c01

2. You need create a Linux physical or virtual machine. I will create a Ubuntu Linux virtual machine on Azure.

3. After Linux virtual machine provisioning, you'd better get PuTTY tool set (PuTTY.zip) from: http://www.chiark.greenend.org.uk/~sgtatham/putty/download.html

It used to connect Linux console via SSH and copy source code file to Linux.

4. On PuTTY folder, please run PUTTYGEN.EXE to generate key for SSH connection.

Please copy key raw data from TextBox and paste to notepad. And then save public key to file via Save private key button.

5. Please use PuTTY to connect Linux machine by user name and password.

6. Please run sudo apt-get install gcc g++ nano to install gcc and g++ compiler.

7. Run nano ~/.ssh/authorized_keys to edit SSH connection key file. And then copy key raw data from notepad and paste key raw data to nano editor in one line. At last, press Ctrl + X, Y and Enter to save file.

8. Run mkdir /home/dbg to create a folder named dbg.

9. Run logout to disconnect linux

10. Open Visual Sudio 2015, Select Visual C++ as language, Cross-platform category, Linux Makefile project

11. Open project properties window, Select Debug Tab. And then set below properties:

Host Name: $(LinuxHostName)
User Name: $(LinuxUsername)
Private Key: $(PrivateKey)             The ppk file you saved on step 4.
PuTTY Home: $(PuttyHome)        Please DO NOT use a path with whitespace.
Remote Working Directory: $(WorkingDirectory)      /home/dbg
Remote Executable Name: $(RemoteExecutable)   Whatever you want.

12. After you completed above steps, you can debug your Linux app remotely like a normal VC++ Win32 app on Windows.

Enjoy!

How fast is Microsoft Edge?

Michael Lee2 — Mon, 20 Jul 2015 22:47:00 GMT

I did some testing between the newest version of Chrome, firefox and Microsoft Edge by Sunspider tool. The Sunspider is a tool focus on testing JavaScript performance for modern web browsers that was created by apple WebKit project. The less time consumption means the faster Web Browser you have. Here is the testing result: Microsoft Edge(version:20.10240.16384.0): Total: 256.8ms+/- 5.8% Chrome(version: 43.0.2357.134): Total: 635.1ms +/- 5.6% Firefox(39.0): Total: 589.3ms +/- 9.8% By now, Microsoft Edge is the fastest Web browser! Compare with Chrome, Edge has a significant performance improvement that faster 2.7 times than Chrome: TEST COMPARISON FROM TO DETAILS =============================================================================== ** TOTAL **: 2.47x as fast 635.1ms +/- 5.6% 256.8ms +/- 5.8% significant =============================================================================== 3d: 3.21x as fast 98.5ms +/- 11.9% 30.7ms +/- 20.3% significant cube: 2.97x as fast 26.7ms +/- 27.1% 9.0ms +/- 17.6% significant morph: 15.9x as fast 38.2ms +/- 15.0% 2.4ms +/- 25.1% significant raytrace: 1.74x as fast 33.6ms +/- 11.6% 19.3ms +/- 27.6% significant access: 2.45x as fast 39.5ms +/- 16.3% 16.1ms +/- 6.1% significant binary-trees: - 5.3ms +/- 38.2% 4.2ms +/- 10.8% fannkuch: 3.55x as fast 18.8ms +/- 14.4% 5.3ms +/- 14.3% significant nbody: 2.33x as fast 9.8ms +/- 35.9% 4.2ms +/- 10.8% significant nsieve: 2.33x as fast 5.6ms +/- 23.5% 2.4ms +/- 20.8% significant bitops: 2.01x as fast 33.9ms +/- 14.6% 16.9ms +/- 13.9% significant 3bit-bits-in-byte: 4.67x as fast 4.2ms +/- 15.6% 0.9ms +/- 45.1% significant bits-in-byte: 4.32x as fast 10.8ms +/- 28.6% 2.5ms +/- 33.7% significant bitwise-and: ?? 6.9ms +/- 17.9% 8.4ms +/- 21.3% not conclusive: might be *1.22x as slow* nsieve-bits: 2.35x as fast 12.0ms +/- 20.8% 5.1ms +/- 19.2% significant controlflow: 2.50x as fast 8.5ms +/- 31.0% 3.4ms +/- 10.9% significant recursive: 2.50x as fast 8.5ms +/- 31.0% 3.4ms +/- 10.9% significant crypto: 3.55x as fast 55.0ms +/- 18.5% 15.5ms +/- 26.0% significant aes: 2.39x as fast 19.1ms +/- 27.2% 8.0ms +/- 25.6% significant md5: 2.67x as fast 17.1ms +/- 28.9% 6.4ms +/- 39.1% significant sha1: 17.1x as fast 18.8ms +/- 22.7% 1.1ms +/- 36.9% significant date: - 52.2ms +/- 17.8% 43.8ms +/- 5.4% format-tofte: 1.70x as fast 26.2ms +/- 25.9% 15.4ms +/- 9.6% significant format-xparb: ?? 26.0ms +/- 14.5% 28.4ms +/- 5.3% not conclusive: might be *1.092x as slow* math: 4.10x as fast 68.1ms +/- 14.1% 16.6ms +/- 5.8% significant cordic: 4.44x as fast 16.0ms +/- 26.1% 3.6ms +/- 19.2% significant partial-sums: 3.43x as fast 41.2ms +/- 14.0% 12.0ms +/- 7.9% significant spectral-norm: 10.9x as fast 10.9ms +/- 30.2% 1.0ms +/- 0.0% significant regexp: 1.63x as fast 18.8ms +/- 15.3% 11.5ms +/- 20.6% significant dna: 1.63x as fast 18.8ms +/- 15.3% 11.5ms +/- 20.6% significant string: 2.55x as fast 260.6ms +/- 8.0% 102.3ms +/- 5.5% significant base64: 3.86x as fast 16.6ms +/- 14.2% 4.3ms +/- 34.2% significant fasta: 3.57x as fast 53.2ms +/- 10.9% 14.9ms +/- 9.4% significant tagcloud: 1.91x as fast 74.5ms +/- 13.2% 39.1ms +/- 5.4% significant unpack-code: 2.37x as fast 74.5ms +/- 9.8% 31.5ms +/- 11.2% significant validate-input: 3.34x as fast 41.8ms +/- 17.8% 12.5ms +/- 6.2% significant

DevOps for Azure Cloud Service

Michael Lee2 — Fri, 10 Jul 2015 00:56:00 GMT

Wanna implement DevOps process on Azure Cloud Service? I created a document can help you resolve related issues on Cloud Service. I really hope it can help more and more Azure IT Professionals. The download link is HERE.

1. Overview
1.1 What’s DevOps?
1.2 Why Choose Azure Cloud Service?
2. Deploy Web application to Cloud Service
2.1 Create a Cloud Service Project in Visual Studio
2.2 Build deploy package
3. Implement Scale computing on Cloud Service
3.1 Scale computing on Cloud Service
3.2 Planning scale computing under Cloud Service
3.3 Cloud Service scaling settings
4. Implement Staging/Production Switch
4.1 What’s is Staging Environment?
4.2 Switch Staging deployment to Production
5. Monitoring Web Application
5.1 The real time monitoring
5.2 The Diagnostics data
5.3 Configure Cloud Service to output log files
5.4 Collecting log files from Cloud Service
5.5 Add the AzureLogCollector as a VM Extension
5.6 Analysis Cloud Service log files with tools
5.6.1 Retrieve unhandled exceptions from Windows application log
5.6.2 Retrieve the slowest web pages from IIS log
5.6.3 Analysis crash dump by DebugDiag
6. Conclusions

The AzureLogCollector extension cannot be deleted?

Michael Lee2 — Wed, 08 Jul 2015 04:44:18 GMT

Recently I got a interesting case when I want to collect log files from a Azure Cloud Service. As you know, I use Set-AzureServiceExtension Cmdlet to enable AzureLogCollector retrieve diagnostics log files from VM instance. For details you can visit:

http://azure.microsoft.com/blog/2015/03/09/simplifying-virtual-machine-troubleshooting-using-azure-log-collector/

And then I want to disable AzureLogCollector extension after I had got log files from Azure Storage. Unfortunatly, It seems I cannot disable this extension. PowerShell like this:

# First time, I list all service extension in specified Cloud Service
 PS > Get-AzureServiceExtension -ServiceName "devopssample" -Slot Staging
 
 Role : DevOpsWebSite
 Extension : AzureLogCollector
 ProviderNameSpace : Microsoft.WindowsAzure.Compute
 Version : 1.0
 Id : Default-AzureLogCollector-Staging-Ext-0
 PublicConfiguration : {
 "Instances": "*",
 "Mode": "Full",
 "SasUri": "################################"
 }
 OperationDescription : Get-AzureServiceExtension
 OperationId : b7c4c5e1-4f90-b07c-9524-97e74d1a06d5
 OperationStatus : Succeeded
 
 #Now I am going to remove AzureLogCollector by Remove-AzureServiceExtension. And it seems succeed.
 PS > Remove-AzureServiceExtension -ServiceName "DevOpsSample" -Role "DevOpsWebSite" -Slot Staging -ProviderNamespace "Microsoft.WindowsAzure.Compute" -ExtensionName "AzureLogCollector"
 Warning: Removing AzureLogCollector configuration for DevOpsWebSite from Cloud Service DevOpsSample.
 
 OperationDescription OperationId OperationStatus 
 -------------------- ----------- --------------- 
 Remove-AzureServiceExtension 99d1c589-3b60-b88b-b390-ee90a0459d43 Succeeded 
 
 
 #After Remove extension succeed. The extension still working.
 PS > Get-AzureServiceExtension -ServiceName "devopssample" -Slot Staging
 
 Role : DevOpsWebSite
 Extension : AzureLogCollector
 ProviderNameSpace : Microsoft.WindowsAzure.Compute
 Version : 1.0
 Id : Default-AzureLogCollector-Staging-Ext-0
 PublicConfiguration : {
 "Instances": "*",
 "Mode": "Full",
 "SasUri": "####################################"
 }
 OperationDescription : Get-AzureServiceExtension
 OperationId : 488b5e87-a7c1-b9b5-a44b-2b7449a9ade0
 OperationStatus : Succeeded

Does any guy know why?

A way to retrieve SQL statement from SharePoint native call stack

Michael Lee2 — Thu, 13 Nov 2014 23:27:01 GMT

In some situations of SharePoint slow is caused by SQL SERVER or poor SQL statements. To resolve issue, we usually catch a hang dump and figure out what exactly SharePoint doing at that hanging moment. If the performance issue caused by customized code. There are too much easy way to display SharePoint callstack like !clrstack/!dumpstack etc in windbg. Another way is that you can analysis automatically hang dump by DebugDiag 2.0 If you want to enum all SQL statements from hang dump, you can also execute windbg command like this:

.foreach(s {!dumpheap -type System.Data.SqlClient.SqlCommand
 -short}){r $t1 = poi(${s}+20);!do -nofields $t1;}

All of above ways are dedicated on managed callstack in SharePoint. As we know, SharePoint service has some COM objects. That's why we have to dispose some SharePoint objects after using. Inappropriate using SharePoint objects would cause memory leak and let your SharePoint site slowly down. For more information please visit: http://msdn.microsoft.com/en-us/library/office/ee557362(v=office.14).aspx

Meanwhile these COM object in SharePoint also access database by OLE DB. We usually can get callstack from SharePoint hang dump like this:


 0:035> kpM
 # Child-SP          RetAddr           Call Site
00 00000000`0e8a9f28 000007ff`76f15280 ntdll!NtWaitForSingleObject+0xa
01 00000000`0e8a9f30 000007ff`7705173e mswsock!WSPRecv+0x66b
02 00000000`0e8aa070 000007ff`76ff3518 ws2_32!WSARecv+0x166
03 00000000`0e8aa120 000007ff`72729c0e wsock32!recv+0x38
04 00000000`0e8aa180 000007ff`60761ae7 dbnetlib!ConnectionRead+0x4fe
05 00000000`0e8aa260 000007ff`60762f2d sqloledb!CDataSource::ConnectionTransact+0xf7
06 00000000`0e8aa2e0 000007ff`607bac34 sqloledb!CDBConnection::SendPacket+0x25d
07 00000000`0e8aa3a0 000007ff`60754b69 sqloledb!CStmt::SQLExecRPC+0x4d4
08 00000000`0e8aa4b0 000007ff`60755df3 sqloledb!CCommand::ExecuteHelper+0x2a9
09 00000000`0e8aa550 000007ff`60756cc7 sqloledb!CCommand::Execute+0xc73
0a 00000000`0e8aa640 000007ff`63f3e4cc sqloledb!CImpICommandText::Execute+0x187
0b 00000000`0e8aa6c0 000007ff`63f3f65b oledb32!CCommandText::DoExecute+0x4fc
0c 00000000`0e8aa900 00000000`108159ca oledb32!CCommandText::Execute+0x8ab
0d 00000000`0e8aabb0 00000000`3561ccb1 STSWEL!Voledb::ExecQuery+0x37e
0e 00000000`0e8aad10 00000000`3561dbee OWSSVR!GetExtensionVersion+0x62c5
0f 00000000`0e8ad5c0 00000000`35622540 OWSSVR!GetExtensionVersion+0x7202
10 00000000`0e8ad650 00000000`35650e0a OWSSVR!GetExtensionVersion+0xbb54
11 00000000`0e8adda0 00000642`7f600887 OWSSVR!GetExtensionVersion+0x3a41e

In this case you should watch this line: sqloledb!CStmt::SQLExecRPC.

0:035> .frame /r 7 
 07 00000000`0e8aa3a0 000007ff`60754b69 sqloledb!CStmt::SQLExecRPC+0x4d4
 rax=0000000000000105 rbx=0000000010a311f0 rcx=0000000000000000
 rdx=000000000ddcac78 rsi=0000000000000000 rdi=00000000173c5be0
 rip=000007ff607bac34 rsp=000000000e8aa3a0 rbp=0000000010a22e30
 r8=000000000ddcac7e r9=000000ffffffffff r10=0000000000000000
 r11=0000000000000139 r12=00000000ffffffff r13=000000000e8aa5b8
 r14=0000000000000045 r15=00000000173cd390
 iopl=0 nv up ei ng nz na pe cy
 cs=0033 ss=002b ds=002b es=002b fs=0053 gs=002b efl=00000283
 sqloledb!CStmt::SQLExecRPC+0x4d4:
 000007ff`607bac34 83f8ff cmp eax,0FFFFFFFFh

The address of SQL statement is located in register 15.

0:035> du 00000000173cd390
 00000000`173cd390 "{?=call proc_GetTpWebMetaDataAnd"
 00000000`173cd3d0 "ListMetaData(?,NULL,?,NULL,1,?,?"
 00000000`173cd410 ",18)}."

Hope it would be useful :)

Migrating IIS 6.0 Web Application to New Version Guidance

Michael Lee2 — Thu, 17 Jul 2014 06:47:00 GMT

The date of ending support Windows Server 2003 is approaching. Recently many customers want to migrate their old web application to new version of Windows Server. There is a significant difference between IIS 6.0 and IIS 7.x and 8.x. Migrate web application to new Windows not only means upgrade Windows but also means migrate web app from 32 bit to 64 bit, account type changes, change classic pipeline to integrated pipeline and so on. It seems a little bit complicated. I created a IIS migrate guidance that based on my work experience of recent years. Hopefully it can help engineers who want to migrate their web application to new version of IIS.

You can download from HERE