We added the com.apple.developer.background-tasks.continued-processing.gpu key to the entitlement file and set it to true, but BGTaskScheduler.supportedResources does not include gpu. How can we configure it to obtain permission for GPU access in the background? Test device: iPhone 16 Pro Max, iOS 26 release version.

Hello im creating an expo module using this new API, but the problem i found currently testing this functionality is that when the task fails, the notification error doesn't go away and is always showing the failed task notification even if i start a new task and complete that one. I want to implement this module into the production app but i feel like having always the notification error might confuse our users or find it a bit bothersome. Is there a way for the users to remove this notification? Best regards!

I work for a large medical device company. We have a 1st party BLE enabled medical device that must be very battery efficient. To this end, if a connection is lost, the BLE radio is powered down after 60 seconds and will only turn back on when a physical button on the device is pressed. I've been tasked with connecting to the device, staying connected to the device, and being able to retrieve data from the device upon a timed action. For instance, this could include a data read and transmission while they sleep. The key part of this is staying reliably connected for extended periods of time. This is a BYOD setup, and we cannot control power profiles. I would very much appreciate any information, recommendations, and/or insights into solving this problem. Thanks!

Testing Environment: iOS: 26.0 Beta 7 Xcode: Beta 6 Description: We are implementing the new BGContinuedProcessingTask API introduced in iOS 26. We have followed the official documentation and WWDC session guidance to configure our project. The Background Modes (processing) and Background GPU Access capabilities have been added in Xcode. The com.apple.developer.background-tasks.continued-processing.gpu entitlement is present and set to in the .entitlements file. The provisioning profile details viewed within Xcode explicitly show that the "Background GPU Access" capability and the corresponding entitlement are included. Despite this correct configuration, when running the app on supported hardware (iPhone 16 Pro), a call to BGTaskScheduler.supportedResources.contains(.gpu) consistently returns false. This prevents us from setting request.requiredResources = .gpu. As a result, when the BGContinuedProcessingTask starts without the GPU resource flag, our internal Metal-based exporter attempts to access the GPU and is terminated by the system, throwing an IOGPUMetalError: Insufficient Permission (to submit GPU work from background). We have performed extensive debugging, including a full reset of the provisioning profile (removing/re-adding capabilities, toggling automatic signing, cleaning build folders, and reinstalling the app), but the issue persists. This strongly suggests a bug in the iOS 26 beta where the runtime is failing to correctly validate a valid entitlement. Additionally, we've observed inconsistent behavior across devices. On an A16-based iPad, the task submits successfully (BGTaskScheduler.submit does not throw an error), but the launch handler is never invoked by the system. On the iPhone 16 Pro, the handler is invoked, but we encounter the supportedResources issue described above. This leads us to ask for clarification on the exact hardware requirements for this feature. We hypothesize that it may be limited to devices that support Apple Intelligence (A17 Pro and newer). Could you please confirm this and provide official documentation on the device support criteria? Steps to Reproduce: Create a new Xcode project. In Signing & Capabilities, add "Background Modes" (with "Background processing" checked) and "Background GPU Access". Add a permitted identifier (e.g., "com.company.test.*") to BGTaskSchedulerPermittedIdentifiers in Info.plist. In application(_:didFinishLaunchingWithOptions:) or a ViewController's viewDidLoad, log the result of BGTaskScheduler.shared.supportedResources.contains(.gpu). Build and run on a physical, supported device (e.g., iPhone 16 Pro). Expected Results: The log should indicate that BGTaskScheduler.shared.supportedResources.contains(.gpu) returns true. Actual Results: The log shows that BGTaskScheduler.shared.supportedResources.contains(.gpu) returns false.

I'm developing a macOS console application that uses ODBC to connect to PostgreSQL. The application works fine when run normally, but fails to load the ODBC driver when debugging with LLDB(under root works fine as well). Error Details When running the application through LLDB, I get this sandbox denial in the system log (via log stream): Error 0x0 0 0 kernel: (Sandbox) Sandbox: logd_helper(587) deny(1) file-read-data /opt/homebrew/lib/psqlodbcw.so The application cannot access the PostgreSQL ODBC driver located at /opt/homebrew/lib/psqlodbcw.so(also tried copy to /usr/local/lib/...). Environment macOS Version: Latest Sequoia LLDB: Using LLDB from Xcode 16.3 (/Applications/Xcode16.3.app/Contents/Developer/usr/bin/lldb) ODBC Driver: PostgreSQL ODBC driver installed via Homebrew Code Signing: Application is signed with Apple Development certificate What is the recommended approach for debugging applications that need to load dynamic libraries? Are there specific entitlements or configurations that would allow LLDB to access ODBC drivers during debugging sessions? Any guidance would be greatly appreciated. Thank you for any assistance!

Problem summary I have a macOS helper app that is launched from a sandboxed main app. The helper: has com.apple.security.app-sandbox = true and com.apple.security.inherit = true in its entitlements, is signed and embedded inside the main app bundle (placed next to the main executable in Contents/MacOS), reports entitlement_check = 1 (code signature contains sandbox entitlement, implemented via SecStaticCode… check), sandbox_check(getpid(), NULL, 0) returns 1 (runtime sandbox enforcement present), APP_SANDBOX_CONTAINER_ID environment variable is not present (0). Despite that, Cocoa APIs return non-container home paths: NSHomeDirectory() returns /Users/<me>/ (the real home). [[NSFileManager defaultManager] URLsForDirectory:inDomains:] and URLForDirectory:inDomain:appropriateForURL:create:error: return paths rooted at /Users/<me>/ (not under ~/Library/Containers/<app_id>/Data/...) — i.e. they look like non-sandboxed locations. However, one important exception: URLForDirectory:... for NSItemReplacementDirectory (temporary/replacement items) does return a path under the helper's container (example: ~/Library/Containers/<app_id>/Data/tmp/TemporaryItems/NSIRD_<helper_name>_hfc1bZ). This proves the sandbox is active for some FileManager APIs, yet standard directory lookups (Application Support, Documents, Caches, and NSHomeDirectory()) are not being redirected to the container. What I expect The helper (which inherits the sandbox and is clearly sandboxed) should get container-scoped paths from Cocoa’s FileManager APIs (Application Support, Documents, Caches), i.e. paths under the helper’s container: /Users/<me>/Library/Containers/<app_id>/Data/.... What I tried / diagnostics already gathered Entitlements & code signature codesign -d --entitlements :- /path/to/Helper.app/Contents/MacOS/Helper # shows com.apple.security.app-sandbox = true and com.apple.security.inherit = true Runtime checks (Objective-C++ inside helper): extern "C" int sandbox_check(pid_t pid, const char *op, int flags); NSLog(@"entitlement_check = %d", entitlement_check()); // SecStaticCode check NSLog(@"env_variable_check = %d", (getenv("APP_SANDBOX_CONTAINER_ID") != NULL)); NSLog(@"runtime_sandbox_check = %d", sandbox_check(getpid(), nullptr, 0)); NSLog(@"NSHomeDirectory = %s", NSHomeDirectory()); NSArray *urls = [[NSFileManager defaultManager] URLsForDirectory:NSApplicationSupportDirectory inDomains:NSUserDomainMask]; NSLog(@"URLsForDirectory: %@", urls); Observed output: entitlement_check = 1 env_variable_check = 0 runtime_sandbox_check = 1 NSHomeDirectory = /Users/<me> URLsForDirectory: ( "file:///Users/<me>/Library/Application%20Support/..." ) Temporary/replacement directory (evidence sandbox active for some APIs): NSURL *tmpReplacement = [[NSFileManager defaultManager] URLForDirectory:NSItemReplacementDirectory inDomain:NSUserDomainMask appropriateForURL:nil create:YES error:&err]; NSLog(@"NSItemReplacementDirectory: %@", tmpReplacement.path); Observed output (example): /Users/<me>/Library/Containers/<app_id>/Data/tmp/TemporaryItems/NSIRD_<helper_name>_hfc1bZ Other facts Calls to NSHomeDirectory() and URLsForDirectory: are made after main() to avoid "too early" initialization problems. Helper is placed in Contents/MacOS (not Contents/Library/LoginItems). Helper is a non-GUI helper binary launched by the main app (not an XPC service). macOS version: Sequoia 15.6 Questions Why do NSHomeDirectory() and URLsForDirectory: return the real /Users/<me>/... paths in a helper process that is clearly sandboxed (entitlement + runtime enforcement), while NSItemReplacementDirectory returns a container-scoped temporary path? Is this behavior related to how the helper is packaged or launched (e.g., placement in Contents/MacOS vs Contents/Library/LoginItems, or whether it is launched with posix_spawn/fork+exec vs other APIs)? Are there additional entitlements or packaging rules required for a helper that inherits sandbox to have Cocoa directory APIs redirected to the container (for Application Support, Documents, Caches)? *Thanks in advance — I can add any requested logs

Hi all, I’ve built an Electron application that uses two child processes: An Express.js server A Python executable (packaged .exe/binary) During the development phase, everything works fine — the Electron app launches, both child processes start, and the app functions as expected. But when I create a production build for macOS, the child processes don’t run. Here’s a simplified snippet from my electron.mjs: import { app, BrowserWindow } from "electron"; import { spawn } from "child_process"; import path from "path"; let mainWindow; const createWindow = () => { mainWindow = new BrowserWindow({ width: 1200, height: 800, webPreferences: { nodeIntegration: true, }, }); mainWindow.loadFile("index.html"); // Start Express server const serverPath = path.join(process.resourcesPath, "app.asar.unpacked", "server", "index.js"); const serverProcess = spawn(process.execPath, [serverPath], { stdio: "inherit", }); // Start Python process const pythonPath = path.join(process.resourcesPath, "app.asar.unpacked", "python", "myapp"); const pythonProcess = spawn(pythonPath, [], { stdio: "inherit", }); serverProcess.on("error", (err) => console.error("Server process error:", err)); pythonProcess.on("error", (err) => console.error("Python process error:", err)); }; app.whenReady().then(createWindow); I’ve already done the following: Configured package.json with the right build settings Set up extraResources / asarUnpack to include the server and Python files Verified both child processes work standalone Questions: What’s the correct way to package and spawn these child processes for macOS production builds? Do I need to move them into a specific location (like Contents/Resources/app.asar.unpacked) and reference them differently? Is there a more reliable pattern for handling Express + Python child processes inside an Electron app bundle? Any insights or working examples would be really appreciated!

I'm working on a Mac app that receives a process ID via NSXPCConnection, and I'm trying to figure out the best way to determine whether that process is a native macOS app like Safari—with bundles and all—or just a script launched by something like Node or Python. The executable is signed with a Team ID using codesign. I was thinking about getting the executable's path as one way to handle it, but I’m wondering if there’s a more reliable method than relying on the folder structure.

Testing Environment: iOS Version: 26.0 Beta 7 Xcode Version: 17.0 Beta 6 Device: iPhone 16 Pro Description: We are implementing the new BGContinuedProcessingTask API and are using the wildcard identifier notation as described in the official documentation. Our Info.plist is correctly configured with a permitted identifier pattern, such as com.our-bundle.export.*. We then register a single launch handler for this exact wildcard pattern. We are performing this registration within a UIViewController, which is a supported pattern as BGContinuedProcessingTask is explicitly exempt from the "register before applicationDidFinishLaunching" requirement, according to the BGTaskScheduler.h header file. The register method correctly returns true, indicating the registration was successful. However, when we then try to submit a task with a unique identifier that matches this pattern (e.g., com.our-bundle.export.UUID), the BGTaskScheduler.shared.submit() call throws an NSInternalInconsistencyException and terminates the app. The error reason is: 'No launch handler registered for task with identifier com.our-bundle.export.UUID'. This indicates that the system is not correctly matching the specific, unique identifier from the submit call to the registered wildcard pattern handler. This behavior contradicts the official documentation. Steps to Reproduce: Create a new Xcode project. In Signing & Capabilities, add "Background Modes" (with "Background processing" checked) and "Background GPU Access". Add a permitted identifier (e.g., "com.company.test.*") to BGTaskSchedulerPermittedIdentifiers in Info.plist. In a UIViewController's viewDidLoad, register a handler for the wildcard pattern. Check that the register method returns true. Immediately after, try to submit a BGContinuedProcessingTaskRequest with a unique identifier that matches the pattern. Expected Results: The submit call should succeed without crashing, and the task should be scheduled. Actual Results: The app crashes immediately upon calling submit(). The console shows an uncaught NSInternalInconsistencyException with the reason: 'No launch handler registered for task with identifier com.company.test.UUID'. Workaround: The issue can be bypassed if we register a new handler for each unique identifier immediately before submitting a request with that same unique identifier. This strongly suggests the bug is in the system's wildcard pattern-matching logic.

when we use raise in GCD, the signal handler is executed asynchronously, whereas in pthread, it is executed synchronously as expected. example: #include <Foundation/Foundation.h> #include <pthread/pthread.h> static void HandleSignal(int sigNum, siginfo_t* signalInfo, void* userContext) { printf("handle signal %d\n", sigNum); printf("begin sleep\n"); sleep(3); printf("end sleep\n"); } void InstallSignal(void) { static const int g_fatalSignals[] = { SIGABRT, SIGBUS, SIGFPE, SIGILL, SIGPIPE, SIGSEGV, SIGSYS, SIGTRAP, }; int fatalSignalsCount = sizeof(g_fatalSignals) / sizeof(int); struct sigaction action = {{0}}; action.sa_flags = SA_SIGINFO | SA_ONSTACK; #if defined(__LP64__) action.sa_flags |= SA_64REGSET; #endif sigemptyset(&action.sa_mask); action.sa_sigaction = &HandleSignal; struct sigaction pre_sa; for(int i = 0; i < fatalSignalsCount; i++) { int sigResult = sigaction(g_fatalSignals[i], &action, &pre_sa); } } void* RaiseAbort(void *userdata) { raise(SIGABRT); printf("signal handler has finished\n"); return NULL; } int main(int argc, const char * argv[]) { InstallSignal(); dispatch_async(dispatch_get_global_queue(0, 0), ^{ raise(SIGABRT); // abort(); // abort() is ok RaiseAbort(nullptr); }); // pthread is ok // pthread_t tid; // int ret = pthread_create(&tid, NULL, RaiseAbort, NULL); // if (ret != 0) { // fprintf(stderr, "create thread failed\n"); // return EXIT_FAILURE; // } [[NSRunLoop mainRunLoop] run]; return 0; } console log: signal handler has finished handle signal 6 begin sleep end sleep

I'm looking into a newer XPC API available starting with macOS 14. Although it's declared as a low-level API I can't figure it how to specify code signing requirement using XPCListener and XPCSession. How do I connect it with xpc_listener_set_peer_code_signing_requirement and xpc_connection_set_peer_code_signing_requirement which require xpc_listener_t and xpc_connection_t respectively? Foundation XPC is declared as a high-level API and provides easy ways to specify code signing requirements on both ends of xpc. I'm confused with all these XPC APIs and their future: Newer really high-level XPCListener and XPCSession API (in low-level framework???) Low-level xpc_listener_t & xpc_connection_t -like API. Is it being replaced by newer XPCListener and XPCSession? How is it related to High-level Foundation XPC? Are NSXPCListener and NSXPCConnection going to be deprecated and replaced by XPCListener and XPCSession??

Hello, I am trying to implement a subscriber which specifies its own demand for how many elements it wants to receive from a publisher. My code is below: import Combine var array = [1, 2, 3, 4, 5, 6, 7] struct ArraySubscriber<T>: Subscriber { typealias Input = T typealias Failure = Never let combineIdentifier = CombineIdentifier() func receive(subscription: any Subscription) { subscription.request(.max(4)) } func receive(_ input: T) -> Subscribers.Demand { print("input,", input) return .max(4) } func receive(completion: Subscribers.Completion<Never>) { switch completion { case .finished: print("publisher finished normally") case .failure(let failure): print("publisher failed due to, ", failure) } } } let subscriber = ArraySubscriber<Int>() array.publisher.subscribe(subscriber) According to Apple's documentation, I specify the demand inside the receive(subscription: any Subscription) method, see link. But when I run this code I get the following output: input, 1 input, 2 input, 3 input, 4 input, 5 input, 6 input, 7 publisher finished normally Instead, I expect the subscriber to only "receive" elements 1, 2, 3, 4 from the array. How can I accomplish this?

The application is placed into the idle state. Subsequently, the device enters a sleep state. While the device is in sleep, App start background task within the application successfully receives its expirationHandler callback. App received the expiration callback and App called the end BGtask OS did not released the Assertion. Resulting in App getting terminated by the OS for exceeding the BG task Apple Feedback- FB19192371

My app does really large uploads. Like several GB. We use the AWS SDK to upload to S3. It seemed like using BGContinuedProcessingTask to complete a set of uploads for a particular item may improve UX as well as performance and reliability. When I tried to get BGContinuedProcessingTask working with the AWS SDK I found that the task would fail after maybe 30 seconds. It looked like this was because the app stopped receiving updates from the AWS upload and the task wants consistent updates. The AWS SDK always uses a background URLSession and this is not configurable. I understand the background URLSession runs in a separate process from the app and maybe that is why progress updates did not continue when the app was in the background. Is it expected that BGContinuedProcessingTask and background URLSession are not really compatible? It would not be shocking since they are 2 separate background APIs. Would the Apple recommendation be to use a normal URLSession for this, in which case AWS would need to change their SDK? Or does Apple think that BGContinuedProcessingTask should just not be used with uploads? In other words use an upload specific API. Thanks!

I'm working on an enterprise product that's mainly a daemon (with Endpoint Security) without any GUI component. I'm looking into the update process for daemons/agents that was introduced with Ventura (Link), but I have to say that the entire process is just deeply unfun. Really can't stress this enough how unfun. Anyway... The product bundle now contains a dedicated Swift executable that calls SMAppService.register for both the daemon and agent. It registers the app in the system preferences login items menu, but I also get an error. Error registering daemon: Error Domain=SMAppServiceErrorDomain Code=1 "Operation not permitted" UserInfo={NSLocalizedFailureReason=Operation not permitted} What could be the reason? I wouldn't need to activate the items, I just need them to be added to the list, so that I can control them via launchctl. Which leads me to my next question, how can I control bundled daemons/agents via launchctl? I tried to use launchctl enable and bootstrap, just like I do with daemons under /Library/LaunchDaemons, but all I get is sudo launchctl enable system/com.identifier.daemon sudo launchctl bootstrap /Path/to/daemon/launchdplist/inside/bundle/Library/LaunchDaemons/com.blub.plist Bootstrap failed: 5: Input/output error (not super helpful error message) I'm really frustrated by the complexity of this process and all of its pitfalls.

Every time macOS goes to sleep the processes get suspended which is expected. But during the sleep period, all processes keep coming back and they all get a small execution window where they make some n/w requests. Regardless of what power settings i have. It also does not matter whether my app is a daemon or not Is there any way that i can disable this so that when system is in sleep, it stays in suspended, no intermittent execution window? I have tried disabling Wake for network access setting but processes still keep getting intermittent execution window. Is there any way that i can prevent my app from coming back while in sleep. I don't want my app to get execution window, perform some executions and then get suspended not knowing when it will get execution window again?

Hello, We're seeing some strange crashes and noticed the following. It's unclear if related or not. The contract for xpc_main, which internally calls dispatch_main, is This function never returns. and they are appropriately peppered with __attribute__((__noreturn__)). Documentation states: This function “parks” the main thread and waits for blocks to be submitted to the main queue. However, internally, dispatch_main calls pthread_exit. pthread_exit's documentation states that: After a thread has terminated, the result of access to local (auto) variables of the thread is undefined. Thus, references to local variables of the exiting thread should not be used for the pthread_exit() value_ptr parameter value. I'd say the two contracts of This function never returns. and thread exiting with its storage released are diametrically opposed and can create nuanced issues. Consider the following code: struct asd { int a; }; struct asd* ptr; void fff(void* ctx) { while(true) { printf("%d\n", ptr->a); ptr->a = (ptr->a + 1); usleep(100000); } } int main(int argc, const char * argv[]) { struct asd zxc; zxc.a = 1; ptr = &zxc; dispatch_async_f(dispatch_get_global_queue(DISPATCH_QUEUE_PRIORITY_HIGH, 0), NULL, fff); dispatch_main(); return 0; } This is a gross over-simplification of the code we have, but in the same "spirit". We have a C++ object that is created on the stack and exposes one of its members as a global pointer, with the assumption that it would never release. What I understand from This function never returns is that the calling thread remains dormant and its stack remains alive. What I understand from pthread_exit is that the thread is killed (this is verified with a debugger attached) and its stack storage is released. Another thing that is throwing me off is that no sanitizer that is provided by clang/Xcode catches this issue. I don't see any special handling of the internal pthread_t in libdispatch to keep the stack storage alive. Our code is more complex, but can be solved by allocating the initial object on the heap, rather than on the stack. But still I would like to understand if this is the expected behavior. Perhaps my preconception of __attribute__((__noreturn__)) is wrong, and accessing stack variables post call to a __attribute__((__noreturn__)) function is UB? Thanks

I'm developing a safety-critical monitoring app that needs to fetch data from government APIs every 30 minutes and trigger emergency audio alerts for threshold violations. The app must work reliably in background since users depend on it for safety alerts even while sleeping. Main Challenge: iOS background limitations seem to prevent consistent 30-minute intervals. Standard BGTaskScheduler and timers get suspended after a few minutes in background. Question: What's the most reliable approach to ensure consistent 30-minute background monitoring for a safety-critical app where missed alerts could have serious consequences? Are there special entitlements or frameworks for emergency/safety applications? The app needs to function like an alarm clock - working reliably even when backgrounded with emergency audio override capabilities.

Hi, I'm working on an application on MacOS. It contains a port-forward feature on TCP protocol. This application has no UI, but a local HTTP server where user can access to configure this application. I found that my application always exit for unknown purpose after running in backgruond for minutes. I think this is about MacOS's background process controlling. Source codes and PKG installers are here: https://github.com/burningtnt/Terracotta/actions/runs/16494390417

I've discovered that a system network extension can communicate with a LaunchDaemon (loaded using SMAppService) over XPC, provided that the XPC service name begins with the team ID. If I move the launchd daemon plist to Contents/Library/LaunchAgents and swap the SMAppService.daemon calls to SMAppService.agent calls, and remove the .privileged option to NSXPCConnection, the system extension receives "Couldn't communicate with a helper application" as an error when trying to reach the LaunchAgent advertised service. Is this limitation by design? I imagine it is, but wanted to check before I spent any more time on it.