I want to draw latency information for each struct bio that passes through the block layer. I have a module that overrides make_request_fn. I would want to find out how long did that bio took from there to reach request queue and from there to driver and so on.
I tried to attach a custom struct to the bio I receive at make_request_fn but since I did not create those, I cant use the bi_private field. Is there any way to work around this?
One option I have is to make a bio wrapper structure and copy bio structs into it before passing it to the lower functions so that I could use container_of to record times.
I have read about tools like blktrace and btt but I need that information inside my module. Is there any way to achieve this?
Thank you.
The solution I used seemed like a common workaround once I found something similar in the source of drbd block driver. The bi_private field can be used only by the function that allocates it. So I used bio_clone in the following way
bio_copy = bio_clone(bio_source, GFP_NOIO);
struct something *instance = kmalloc(sizeof(struct something), GFP_KERNEL);
instance->bio_original = bio_source;
//update timestamps for latency inside this struct instance
bio_copy->bi_private = instance;
bio_copy->bi_end_io = my_end_io_function;
bio_copy->bi_dev = bio_source->bi_dev;
...
...
make_request_fn(queue, bio_copy);
You'll have to write a bi_end_io function. Do remember to call bio_endio for original bio inside this function. You might need to copy bi_error field into bio_source's bi_error before calling bio_endio(bio_source).
Hope this helps someone.
Related
Im new in Go-EVM, and some time ago I got a source code to get transaction tracking. But some functions in the source codes have been updated and changed, here are some questions I wanna ask:
The first one is: How to get *snapshot.Tree?
stateDB, err := state.New(block.Root(), state.NewDatabase(db))
Now this statements need three parameter and the lost parameter's type is *sanpshot.Tree. It is a struct, here is the link to its source code, in line 164.
The second one is: What are AsseccList and GasTipFee?
message := types.NewMessage(from, tx.To(), 0, tx.Value(), tx.Gas(), from Address, to Address, nonce, amount, gasLimit, tx.GasPrice(), GasTopfee, GasTipFee, tx.Data(), accesslist AccessList, false)
AccessList is also a struct. You can see its struct from here. What should I input into AccessList and GasTipFee?
Really appreciate it if you can help me solve these questions.
In your case you do not need to pass a tree snapshot if you do not have one. The purpose of the tree snapshot is to, if the snapshot matches with the given root of the block's trie, run what they call a pre-fetcher routine, that is in charge of preloading nodes in memory so that when the state reaches the commit phase, it is more performant because it already has most of the required nodes in memory. So in your case, you should be perfectly fine passing nil to that constructor.
As for the AccessList and GasTipFee parameters:
AccessList is an EIP-2930 access list. It's once again something optional that transactions can provide to specify the addresses and storage keys that they need access to. Once again you can provide a nil slice.
What you have called GasTipFee is called GasTipCap on master and is basically the limit value of a gas tip, as far as I understand. You can find more information on gas fees in the official documentation.
When I create a function like this:
v8::Function::New(<Isolate>, <C_Function>, <Data_Value>);
The Data_Value that I supply is useful for many things and I can access that when the function is called, with something like FunctionCallbackInfo->GetData().
But I have found no way to get back this data in a different scenario. Let's say I store that Function in a Persistent object, and then I would like to read which data is currently bound to it. Any ideas?
I don't think it's exposed via the API.
But there's an alternative:
manually construct a v8::FunctionTemplate
set its ->InstanceTemplate()->SetInternalFieldCount(num_fields)
get the v8::Function from the template with template->GetFunction(context),
now you should have function->InternalFieldCount() == num_fields
you can use function->SetInternalField(index, value) and function->GetInternalField(index) to store any data you want.
For complete examples, search for "SetInternalFieldCount" in V8's test-api.cc.
I'm relatively new to GraphQL so please bear with me ...
That said, I'm writing an app in node.js to push/pull data from two disparate systems, one of which has an API written in GraphQL.
For the Graph system, I have, something like, the following types defined for me:
Time {
TimeId: Int
TaskId: Int
ProjectId: Int
Project: [Project]
TimeInSeconds: Int
Timestamp: Date
}
and
Task {
TaskId: Int
TaskName: String
TaskDescription: String
}
Where Project is another type whose definition isn't important, only that it is included in the type definition as a field...
What I would like to know is if there is a way to write a query for Time in such a way that I can include the Task type's values in my results in a similar way as the values for the Project type are included in the definition?
I am using someone else's API and do not have the ability to define my own custom types. I can write my own limited queries, but I don't know if the limits are set by the devs that wrote the API or my limited ability with GraphQL.
My suspicion is that I cannot and that I will have to query both separately and combine them after the fact, but I wanted to check here just in case.
Unfortunately, unless the Time type exposes some kind of field to fetch the relevant Task, you won't be able to query for it within the same request. You can include multiple queries within a single GraphQL request; however, they are ran in parallel, which means you won't be able to use the TaskId value returned by one query as a variable used in another query.
This sort of problem is best solved by modifying the schema, but if that's not an option then unfortunately the only other option is to make each request sequentially and then combine the results client-side.
I am working on developing a debugger plugin for visual studio using VSIX. My problem is I have an array of addresses but I cannot set the IDebugMemoryBytes2 to a particular address. I use DEBUG_PROPERTY_INFO and get the array of addresses, and I also am able to set the context to the particular addresses in the array using the Add function in IDebugMemoryContext2. However, I need to use the ReadAt function to retrieve n bytes from a specified address (from IDebugMemoryBytes2).
Does anyone have any idea how to retrieve data from arbitrary addresses from memory?
I am adding more information on the same:
I am using the Microsoft Visual Studio Extensibility package to build my debugger plugin. In the application I am trying to debug using this plugin, there is a double pointer and I need to read those values to process them further in my plugin. For this, there is no way to display all the pointer variables in the watch window and hence, I am not able to get the DEBUG_PROPERTY_INFO for all the block of arrays which the pointer variable is pointing to. This is my problem which I am trying to address. There is no way for me to read the memory pointed to by this double pointer.
Now as for the events in the debuggee process, since the plugin is for debugging variables, I put a breakpoint at a place where I know this pointer is populated and then come back to the plugin for further evaluation.
As a start, I was somehow able to get the starting addresses of each of the array. But still, I am not able to read x bytes of memory from each of these starting addresses.
ie., for example, if I have int **ptr = // pointing to something
I have the addresses present in ptr[0], ptr[1], ptr[2], etc. But I need to go to each of these addresses and fetch the memory block they are pointing to.
For this, after much search, I found this link: https://macropolygon.wordpress.com/2012/12/16/evaluating-debugged-process-memory-in-a-visual-studio-extension/ which seems to address exactly my issue.
So to use expression evaluator functions, I need an IDebugStackFrame2 object to get the ExpressionContext. To get this object, I need to register to events in the debuggee process which is for breakpoint. As said in the post, I did:
public int Event(IDebugEngine2 engine, IDebugProcess2 process,
IDebugProgram2 program, IDebugThread2 thread, IDebugEvent2
debugEvent, ref Guid riidEvent, uint attributes)
{
if (debugEvent is IDebugBreakpointEvent2)
{
this.thread = thread;
}
return VSConstants.S_OK;
}
And my registration is like:
private void GetCurrentThread()
{
uint cookie;
DBGMODE[] modeArray = new DBGMODE[1];
// Get the Debugger service.
debugService = Package.GetGlobalService(typeof(SVsShellDebugger)) as
IVsDebugger;
if (debugService != null)
{
// Register for debug events.
// Assumes the current class implements IDebugEventCallback2.
debugService.AdviseDebuggerEvents(this, out cookie);
debugService.AdviseDebugEventCallback(this);
debugService.GetMode(modeArray);
modeArray[0] = modeArray[0] & ~DBGMODE.DBGMODE_EncMask;
if (modeArray[0] == DBGMODE.DBGMODE_Break)
{
GetCurrentStackFrame();
}
}
}
But this doesn't seem to invoke the Event function at all and hence, I am not sure how to get the IDebugThread2 object.
I also tried the other way suggested in the same post:
namespace Microsoft.VisualStudio.Debugger.Interop.Internal
{
[InterfaceType(ComInterfaceType.InterfaceIsIUnknown), Guid("1DA40549-8CCC-48CF-B99B-FC22FE3AFEDF")]
public interface IDebuggerInternal11 {
[DispId(0x6001001f)]
IDebugThread2 CurrentThread { [return:
MarshalAs(UnmanagedType.Interface)]
[MethodImpl(MethodImplOptions.InternalCall, MethodCodeType =
MethodCodeType.Runtime)]
get; [param: In, MarshalAs(UnmanagedType.Interface)]
[MethodImpl(MethodImplOptions.InternalCall, MethodCodeType =
MethodCodeType.Runtime)] set; }
}
}
private void GetCurrentThread()
{
debugService = Package.GetGlobalService(typeof(SVsShellDebugger)) as IVsDebugger;
if (debugService != null)
{
IDebuggerInternal11 debuggerServiceInternal =
(IDebuggerInternal11)debugService;
thread = debuggerServiceInternal.CurrentThread;
GetCurrentStackFrame();
}
}
But in this method, I think I am missing something but I am not sure what, because after the execution of the line
IDebuggerInternal11 debuggerServiceInternal =
(IDebuggerInternal11)debugService;
when I check the values of the debuggerServiceInternal variable, I see there is a System.Security.SecurityException for CurrentThread, CurrentStackFrame (and so obviously the next line causes a crash). For this, I googled the error and found I was missing the ComImport attribute to the class. So I added that and now, I get a System.AccessViolationException : Attempted to read or write protected memory. This is often an indication that other memory is corrupt.
I am new to C# programming as well and hence, it is a bit difficult to grasp many things in short duration. I am lost as to how to proceed further now.
Any help in the same or suggestions to try another way to achieve my objective will be greatly appreciated.
Thanks a lot,
Esash
After much search, since I am short of time, I need a quick solution and hence, for now, it seems like the quickest way to solve this problem is to hack the .natvis files by making it display all the elements of the pointer and then using the same old way by using IDebug* interface methods to access and retrieve the memory context for each of the pointer elements. But, after posting the same question in msdn forums, I think the proper answer to this problem is as mentioned by Greggs:
"For reading memory, if you want a fast way to do this, you just want the raw memory, and the debug engine of the target is the normal Visual Studio native engine (in other words, you aren't creating your own debug engine), I would recommend referencing Microsoft.VisualStudio.Debugger.Engine. You can then use DkmStackFrame.ExtractFromDTEObject to get the DkmStackFrame object. This will give you the DkmProcess object and you can call DkmProcess.ReadMemory to read memory from the target."
Now, after trying a lot to understand how to implement this, I found that you could just accomplish this using :
DkmProcess.GetProcesses() and doing a ReadMemory on the process returned.
There is a question now, what if more than one processes are returned. Well, I tried attaching many processes to the current debugging process and tried attaching many processes to the debuggee process as well, but found that the DkmProcess.GetProcesses() gets only the one from which I regained the control from, and not the other processes I am attached to. I am not sure if this will work in all cases but for me, it worked this way and for anyone who has similar requirements, this might work as well.
Using the .natvis files to accomplish this means, using IndexListItems for VS2013 and prior versions, and using CustomListItems for VS2015 and greater versions, and to make it look prettier, use the "no-derived" attribute. There is no way to make the Synthetic tag display only the base address of each variable and hence, the above attribute is the best way to go about, but this is not available in VS2013 and prior versions (The base address might get displayed but for people who want to go beyond just displaying contents and also access the memory context of the pointer element, Synthetic tag is not the right thing).
I hope this helps some developer who struggled like me using IDebug* interfaces. For reference, I am also giving the link to the msdn forum where my question was answered.
https://social.msdn.microsoft.com/Forums/en-US/030cef1c-ee79-46e9-8e40-bfc59f14cc34/how-can-i-send-a-custom-debug-event-to-my-idebugeventcallback2-handler?forum=vsdebug
Thanks.
I am quite new to Qt and am in a situation where I want to use a model for my needs:
I have a dynamic number of instances of a subclass that need to be handled differently (different UI controls for each if it is selected). I want to get a list view where I can add new elements or delete old ones, as well as disabling/enabling existing ones.
Of course I want to rewrite as least of the code as possible, so I thought of utilizing the Listwidget and a ListModel to give some controls to the user. But how to link these (or better the items) to instances of the classes?
Do you know any tutorials on this?
I already looked in QtDemo and Google but I do not know the right words to search for
so I had no good results.
Basically what I think I need is a model item that accepts Collider* for its data.
But when I plug this into QStandardItem.setData() it says error: ‘QVariant::QVariant(void*)’ is private
So I found the solution to this problem.
As QStandardItems are capable of storing QVariants as data I wanted to store a pointer to my data in a QVariant. To achieve this I had to use Q_DECLARE_METATYPE(MyType*).
With this I was able to
MyType *MyInstance = new MyType;
QVariant data;
data.setValue(MyInstance);
QStandardItem *item = new QStandardItem("My Item");
item->setData(data);
standardModel->appendRow(item);
And the best is you can add as many types you want and let QVariant do the work to decide if it contains the type you wanted:
if(v.canConvert<MyType*>())
//Yes it is MyType
else if( v.canConvert<MyOtherType*>())
//Oh it is the other one
So finally this only requires to declare the meta type so you do not have to subclass any items.
Also you should read on the limitations of this here:
Q_DECLARE_METATYPE
qRegisterMetaType
Does this page answer your questions? There's an example of deriving a StringListModel item that you might be able to use as a template