Pass 2D array from Go to C using runtime.Pinner

Question content
I am trying to pass a 2D array from Go to some C functionvoid foo(in **float, out *double). Since I want a wrapper for this C function, I want the Go function to have a definition like func FooWrapper([][]float32) []float64. The simplest but not very efficient implementation is to allocate all memory via C listed below:
<code>func FooWrapper(values [][]float32) []float64 {
totalObj := len(values)
totalParams := len(values[0])
results := make([]float64, totalObj)
ptrArrLength := uintptr(totalObj) * cPointerSize
paramArrLength := uintptr(totalParams) * cFloatSize
ptr := C.malloc(C.size_t(ptrArrLength + paramArrLength*uintptr(totalObj)))
defer C.free(ptr)
ptrSlice := unsafe.Slice((**C.float)(ptr), totalObj)
for i, obj := range values {
paramArrPtr := (*C.float)(unsafe.Add(ptr, ptrArrLength+uintptr(i)*paramArrLength))
ptrSlice[i] = paramArrPtr
paramSlice := unsafe.Slice(paramArrPtr, totalParams)
for j, param := range obj {
paramSlice[j] = (C.float)(param)
}
}
C.foo((**C.float)(ptr), (*C.double)(&results[0]))
return results
}
</code>
Is such an implementation safe? Can I pass a pointer to the result data? As far as I know, this pointer will be pinned because it is passed to the C function.
But I want to allocate less memory and just reuse Go memory, I have learned about runtime.Pinner which keeps the pointer pinned until runtime.Pinner.Unpin() is called . I tried writing another implementation using pinner:
<code>func FooWrapper(values [][]float32) []float64 {
length := len(values)
results := make([]float64, length)
pinner := runtime.Pinner{}
defer pinner.Unpin()
arr := (**C.float)(C.malloc(C.size_t(uintptr(length) * cPointerSize)))
defer C.free(unsafe.Pointer(arr))
slice := unsafe.Slice(arr, length)
for i, v := range values {
pinner.Pin(&v[0])
slice[i] = (*C.float)(&v[0])
}
C.foo(arr, (*C.double)(&results[0]))
return results
}
</code>
But, unfortunately, this code doesn't work
runtime: pointer 0xc016ecbfc0 to unused region of span span.base()=0xc016eca000 span.limit=0xc016ecbfa0 span.state=1 fatal error: found bad pointer in Go heap (incorrect use of unsafe or cgo?)
Am I wrong to use runtime.Pinner (as far as I know I can pin slice data)? Or there is another bug in this code. Is there some implementation of passing 3d (4d etc.) arrays to C functions besides allocating and copying all the data to C memory?
Solution
I found the answer to my problem. Using malloc is dangerous because when I try to write a Go pointer to C memory, the GC can identify uninitialized memory as "bad" memory, so if I change malloc to calloc there will be no problem. Of course this is not the best solution, it is better to allocate memory in Go as it is faster and safer.
All discussionshere一个>.
The above is the detailed content of Pass 2D array from Go to C using runtime.Pinner. For more information, please follow other related articles on the PHP Chinese website!
Hot AI Tools
Undress AI Tool
Undress images for free
Undresser.AI Undress
AI-powered app for creating realistic nude photos
AI Clothes Remover
Online AI tool for removing clothes from photos.
Clothoff.io
AI clothes remover
Video Face Swap
Swap faces in any video effortlessly with our completely free AI face swap tool!
Hot Article
Hot Tools
Notepad++7.3.1
Easy-to-use and free code editor
SublimeText3 Chinese version
Chinese version, very easy to use
Zend Studio 13.0.1
Powerful PHP integrated development environment
Dreamweaver CS6
Visual web development tools
SublimeText3 Mac version
God-level code editing software (SublimeText3)
Hot Topics
How do I call a method on a struct instance in Go?
Jun 24, 2025 pm 03:17 PM
In Go language, calling a structure method requires first defining the structure and the method that binds the receiver, and accessing it using a point number. After defining the structure Rectangle, the method can be declared through the value receiver or the pointer receiver; 1. Use the value receiver such as func(rRectangle)Area()int and directly call it through rect.Area(); 2. If you need to modify the structure, use the pointer receiver such as func(r*Rectangle)SetWidth(...), and Go will automatically handle the conversion of pointers and values; 3. When embedding the structure, the method of embedded structure will be improved, and it can be called directly through the outer structure; 4. Go does not need to force use getter/setter,
What are interfaces in Go, and how do I define them?
Jun 22, 2025 pm 03:41 PM
In Go, an interface is a type that defines behavior without specifying implementation. An interface consists of method signatures, and any type that implements these methods automatically satisfy the interface. For example, if you define a Speaker interface that contains the Speak() method, all types that implement the method can be considered Speaker. Interfaces are suitable for writing common functions, abstract implementation details, and using mock objects in testing. Defining an interface uses the interface keyword and lists method signatures, without explicitly declaring the type to implement the interface. Common use cases include logs, formatting, abstractions of different databases or services, and notification systems. For example, both Dog and Robot types can implement Speak methods and pass them to the same Anno
Strategies for Integrating Golang Services with Existing Python Infrastructure
Jul 02, 2025 pm 04:39 PM
TointegrateGolangserviceswithexistingPythoninfrastructure,useRESTAPIsorgRPCforinter-servicecommunication,allowingGoandPythonappstointeractseamlesslythroughstandardizedprotocols.1.UseRESTAPIs(viaframeworkslikeGininGoandFlaskinPython)orgRPC(withProtoco
How do I use the time package to work with time and durations in Go?
Jun 23, 2025 pm 11:21 PM
Go's time package provides functions for processing time and duration, including obtaining the current time, formatting date, calculating time difference, processing time zone, scheduling and sleeping operations. To get the current time, use time.Now() to get the Time structure, and you can extract specific time information through Year(), Month(), Day() and other methods; use Format("2006-01-0215:04:05") to format the time string; when calculating the time difference, use Sub() or Since() to obtain the Duration object, and then convert it into the corresponding unit through Seconds(), Minutes(), and Hours();
How do I use if statements to execute code based on conditions in Go?
Jun 23, 2025 pm 07:02 PM
InGo,ifstatementsexecutecodebasedonconditions.1.Basicstructurerunsablockifaconditionistrue,e.g.,ifx>10{...}.2.Elseclausehandlesfalseconditions,e.g.,else{...}.3.Elseifchainsmultipleconditions,e.g.,elseifx==10{...}.4.Variableinitializationinsideif,l
Understanding the Performance Differences Between Golang and Python for Web APIs
Jul 03, 2025 am 02:40 AM
Golangofferssuperiorperformance,nativeconcurrencyviagoroutines,andefficientresourceusage,makingitidealforhigh-traffic,low-latencyAPIs;2.Python,whileslowerduetointerpretationandtheGIL,provideseasierdevelopment,arichecosystem,andisbettersuitedforI/O-bo
How does Go support concurrency?
Jun 23, 2025 pm 12:37 PM
Gohandlesconcurrencyusinggoroutinesandchannels.1.GoroutinesarelightweightfunctionsmanagedbytheGoruntime,enablingthousandstorunconcurrentlywithminimalresourceuse.2.Channelsprovidesafecommunicationbetweengoroutines,allowingvaluestobesentandreceivedinas
How do I use the Lock() and Unlock() methods to protect a critical section of code in Go?
Jun 23, 2025 pm 08:37 PM
The standard way to protect critical areas in Go is to use the Lock() and Unlock() methods of sync.Mutex. 1. Declare a mutex and use it with the data to be protected; 2. Call Lock() before entering the critical area to ensure that only one goroutine can access the shared resources; 3. Use deferUnlock() to ensure that the lock is always released to avoid deadlocks; 4. Try to shorten operations in the critical area to improve performance; 5. For scenarios where more reads and less writes, sync.RWMutex should be used, read operations through RLock()/RUnlock(), and write operations through Lock()/Unlock() to improve concurrency efficiency.


