Overview

I'm encountering a null pointer dereference crash pointing to the internals of CFNetwork library code on iOS. I'm spawning URLSessionTasks at a decently fast rate (~1-5 per second), with the goal being to generate application layer network traffic. I can reliably encounter this crash pointing to NetConnection::dequeue right after a new task has been spawned and had the resume method called. I suspect that this is perhaps a race condition or some delegate/session object lifecycle bug. The crash appears to be more easily reproduced with a higher rate of spawning URLSessionTasks. I've included the JSON crash file, the lldb stack trace, and the source code of my URLSession(Task) usage. urlsession_stuff_stacktrace.txt urlsession_stuff_source.txt urlsession_crash_report.txt

We probably triggered a bug within Rosetta: https://github.com/docker/desktop-feedback/issues/230

I am facing an intermittent problem where iPhones are failing to pair/connect with Xcode under Xcode -> Windows -> Devices and Simulators. This happens when more than one web content filters are present, for instance, I have my web content filter (FilterSockets true, FilterGrade Firewall) and there is also Sentinel One web content filter with same configuration. Note: We are not blocking any flow from remoted / remotepairingd / core device service / MDRemoteServiceSupport etc processes. But they do get paused and resumed at times for our internal traffic verification logic. So, we are trying to understand what impact our content filter may be having on this iPhone Pairing?? If we stop either one of the filters the problem goes away. I have tracked the network traffic to the phone, and it seems to be using a ethernet interface (en5/en10) over the USB-C cable. I can see endpoints like this: localEndpoint = fe80::7:afff:fea1:edb8%en5.54442 remoteEndpoint = fe80::7:afff:fea1:ed47%en5.49813 I also see remoted process has the below ports open : sudo lsof -nP -iTCP -iUDP | grep remoted remoted 376 root 4u IPv6 0xce4a89bddba37bce 0t0 TCP [fe80:15::7:afff:fea1:edb8]:57395->[fe80:15::7:afff:fea1:ed47]:58783 (ESTABLISHED) remoted 376 root 6u IPv6 0xf20811f6922613c7 0t0 TCP [fe80:15::7:afff:fea1:edb8]:57396 (LISTEN) remoted 376 root 7u IPv6 0x2c393a52251fcc56 0t0 TCP [fe80:15::7:afff:fea1:edb8]:57397 (LISTEN) remoted 376 root 8u IPv6 0xcb9c311b0ec1d6a0 0t0 TCP [fd6e:8a96:a57d::2]:57398 (LISTEN) remoted 376 root 9u IPv6 0xc582859e0623fe4e 0t0 TCP [fd6e:8a96:a57d::2]:57399 (LISTEN) remoted 376 root 10u IPv6 0x2f7d9cee24a44c5b 0t0 TCP [fd6e:8a96:a57d::2]:57400->[fd6e:8a96:a57d::1]:60448 (ESTABLISHED) remoted 376 root 11u IPv6 0xbdb7003643659de 0t0 TCP [fd07:2e7e:2a83::2]:57419 (LISTEN) remoted 376 root 12u IPv6 0x569a5b649ff8f957 0t0 TCP [fd07:2e7e:2a83::2]:57420 (LISTEN) remoted 376 root 13u IPv6 0xa034657978a7da29 0t0 TCP [fd07:2e7e:2a83::2]:57421->[fd07:2e7e:2a83::1]:61729 (ESTABLISHED) But due to the dynamic nature of port and IPs used we are not able to decide on an effective early bypass NEFilterRule. We don't want to use a very broad bypass criteria like all link local IPs etc. Any help will be greatly appreciated.

Just wondering if anyone ran into this issue. I use NetFSMountURLSync for my application with which I allow the user to use a custom path as a mount point (instead of "/Volumes"). This has worked just fine for at least a decade now, but ... since the Tahoe 26.4 "update" mounting to a custom path only generates errors. Note: Mounting to "/Volumes" works correctly (mountpoint = NIL). Since I'm unaware of any changes; is this a bug introduced by Tahoe 26.4, or should I be using a different function to mount a network share?

Inquiry) Inquiry Regarding USB Network Connectivity Between an iPad (Wi‑Fi Model) and an Embedded Linux Device An embedded device (OS: Linux) is connected to an iPad (Wi‑Fi model) using a USB‑C cable. The ipheth driver is installed on the embedded device, and the iPad is recognized correctly. A web server is running on the embedded device. To launch a browser on the iPad and access the web server running on the embedded device via a USB network connection. Based on our verification, the iPad is not assigned an IP address, and therefore communication with the web server on the embedded device is not possible. We would appreciate it if you could provide guidance on the following questions. We would like to assign an IP address to the iPad (Wi‑Fi Model) so that it can communicate with the embedded device over a USB network connection. Is there a way to achieve this through the standard settings on the iPad? If this cannot be achieved through settings alone, are there any existing applications that provide this functionality? If no such application currently exists, is it technically possible to develop an application that enables this capability on iPadOS? Information) The USB‑C port on the embedded device is fixed in HOST mode. The embedded device operates as the USB host, and the iPad operates as a USB device. When a cellular model iPad is connected and “Personal Hotspot” is enabled, an IP address is assigned via DHCP, and we have confirmed that the web server can be accessed from the iPad’s browser. We are investigating whether a similar solution is possible with a Wi‑Fi model iPad.

Hello, How does macOS handle coexistence between multiple Transparent Proxy providers from different vendors if their network rules overlap and one provider modifies the traffic? Thank you in advance!

Hello, Is it possible for a Network Extension (running in its sandbox) to act as a client for an XPC service hosted by a Launch Daemon (e.g., to offload data processing)? Are there any specific sandbox restrictions or entitlement requirements for this type of XPC communication? Thank you in advance!

I am trying to find a datasheet containing information such as "Key Exchange / Key Agreement / Key Establishment Protocols Used", "Digital Signature Algorithms Used", "Hash Algorithms Used", etc. Any information would greatly appreciated.

Team ID: MB9VR977ND We have changed apple developer account for the our application, after changing account we have submitted app for notarization multiple times but all showing in-progress without any logs. do we need to wait until this passed from apple side ? does submitting many request has any issue ? Submission id: 8c5ac51d-bcd3-4fc4-9b38-671e5ea2bf14

I've been working on an app. It uses iCloud syncing. 48 hours ago everything was working 100%. Make a change on the iPhone it immediately changed on the Mac. Change on the Mac, it immediately changed on the iPhone. I didn't work on it yesterday. I updated to iOS26.4 on the iPhone and 26.4 on the Mac yesterday instead. Today, I pull up the project again. I made NO changes to the code or settings. Make a change on the iPhone it immediately updates on the Mac. Make a change on the Mac, nothing happens on the iPhone. I've waited an hour, and the change never happens. If you leave the iPhone app, then return, it updates as it should. It appears that iCloud's silent notification is to being received by the iPhone. Anyone else having the issue? Is there something new with iOS 26.4 that needs to be adjusted to get this to work? Again, works flawlessly with the Mac, just not with the iPhone.

I'm building a freediving app for Apple Watch Ultra using the shallow depth entitlement (com.apple.developer.submerged-shallow-depth-and-pressure). My app uses WKExtendedRuntimeSession with the underwater-depth background mode, and it works correctly — the session starts, Water Lock activates automatically, and Crown hold water ejection ends the session as expected. However, the app does not appear in Settings > General > Auto-Launch > When Submerged on the watch. Other third-party apps (including one that hasn't been updated in ~2 years and presumably only has the shallow entitlement) do appear in this list. My configuration: WKBackgroundModes: ["underwater-depth", "workout-processing"] WKSupportsAutomaticDepthLaunch: true (Boolean, in watch app Info.plist) Entitlement verified in both the signed binary and provisioning profile watchOS 26.3, Apple Watch Ultra 2 Tested with: development build, TestFlight, and direct Xcode deploy. Watch restarted after each. The app does not appear in any case. The documentation at https://developer.apple.com/documentation/coremotion/accessing-submersion-data states: "Adding the underwater-depth Background Mode capability also adds your app to the list of apps that the system can autolaunch when the wearer submerges the watch." Does auto-depth-launch require the full depth entitlement (com.apple.developer.submerged-depth-and-pressure), or should the shallow entitlement be sufficient? Is there an additional step required for the app to appear in the When Submerged list? Any guidance appreciated.

Upgraded to Tahoe 26.4 and Xcode 26.4 and realized Xcode is missing the documentation for Foundation. Occasionally, if other (not yet missing) documentation references a Foundation entity, such as a type, clicking on it may open in a web browser pointing to the externally hosted Apple documentation. Anyone else experiencing this or have a workaround (perhaps the documentation files are on disk but were somehow lost / disconnected from the Xcode doc browser)? I checked the release notes for 26.4 and for 26.3 but did not find a mention of anything like that. Filed an issue with Apple, please do too if you are affected so hopefully it will see some love and lead to restoring the documentation. Edited: found some possibly related files in my ~/Library and the newer version v302 (26.4 related, I think, is considerably smaller): du -sh ~/Library/Developer/Xcode/DocumentationCache/* | awk -F/ '{print $NF "\t" $1}' v296 1.0G v302 507M

I've been able to run this sample project with the PIRServer. But the urls are still not blocked. https://developer.apple.com/documentation/networkextension/filtering-traffic-by-url https://github.com/apple/pir-service-example I got this on the log Received filter status change: <FilterStatus: 'running'>

Hi, We are trying to integrate Metric Kit into our MacOS Application. Our application is a background process. We are interested in getting CPU and Memory metrics for our process. MXMetricPayload is the one we are looking at. We tried to integrate metric Kit and left the background app for 24 hours, we did not get any callback. So, does metric kit work for background app in MacOS? Also does it for Network Extension?

Does the App Review process have access to Apple-Hosted Asset Packs during review? My app uses Asset Packs to offer a library of data to the end-user (with a workaround, if unavailable), but I am frequently seeing the workaround screen in App Review with errors I haven't seen elsewhere. The latest error I encountered (via the App Review team's feedback) was: "A server with the specified hostname could not be found." thrown from (to my belief) AssetPackManager.shared.ensureLocalAvailability. This is unexpected to me, as both this code as well as the asset packs have already been released and are working reliably in production. Has anyone else experienced these issues?

After updating to Xcode 26.4 the pasteboard sharing feature in the simulators doesn't work.

Area: CoreML / Machine Learning Describe the issue: On iOS 26.4, calling MLModel(contentsOf:configuration:) to load an .mlpackage model hangs indefinitely and eventually kills the app via watchdog. The same model loads and runs inference successfully in under 1 second on iOS 26.3.1. The hang occurs inside eort_eo_compiler_compile_from_ir_program (espresso) during on-device AOT recompilation triggered by MLE5ProgramLibraryOnDeviceAOTCompilationImpl createProgramLibraryHandleWithRespecialization:error:. A C++ exception (__cxa_throw) is thrown inside libBNNS.dylib during the exception unwind, which then hangs inside __cxxabiv1::dyn_cast_slow and __class_type_info::search_below_dst. Swift's try/catch does not catch this — the exception originates in C++ and the process hangs rather than terminating cleanly. Setting config.computeUnits = .cpuOnly does not resolve the issue. MLE5ProgramLibrary initialises as shared infrastructure regardless of compute units. Steps to reproduce: Create an app with an .mlpackage CoreML model using the MLE5/espresso backend Call MLModel(contentsOf: modelURL, configuration: config) at runtime Run on a device on iOS 26.3.1 — loads successfully in <1 second Update device to iOS 26.4 — hangs indefinitely, app killed by watchdog after 60–745 seconds Expected behaviour: Model loads successfully, or throws a catchable Swift error on failure. Actual behaviour: Process hangs in MLE5ProgramLibrary.lazyInitQueue. App killed by watchdog. No Swift error thrown. Full stack trace at point of hang: Thread 1 Queue: com.apple.coreml.MLE5ProgramLibrary.lazyInitQueue (serial) frame 0: __cxxabiv1::__class_type_info::search_below_dst libc++abi.dylib frame 1: __cxxabiv1::(anonymous namespace)::dyn_cast_slow libc++abi.dylib frame 2: ___lldb_unnamed_symbol_23ab44dd4 libBNNS.dylib frame 23: eort_eo_compiler_compile_from_ir_program espresso frame 24: -[MLE5ProgramLibraryOnDeviceAOTCompilationImpl createProgramLibraryHandleWithRespecialization:error:] CoreML frame 25: -[MLE5ProgramLibrary _programLibraryHandleWithForceRespecialization:error:] CoreML frame 26: __44-[MLE5ProgramLibrary prepareAndReturnError:]_block_invoke CoreML frame 27: _dispatch_client_callout libdispatch.dylib frame 28: _dispatch_lane_barrier_sync_invoke_and_complete libdispatch.dylib frame 29: -[MLE5ProgramLibrary prepareAndReturnError:] CoreML frame 30: -[MLE5Engine initWithContainer:configuration:error:] CoreML frame 31: +[MLE5Engine loadModelFromCompiledArchive:modelVersionInfo:compilerVersionInfo:configuration:error:] CoreML frame 32: +[MLLoader _loadModelWithClass:fromArchive:modelVersionInfo:compilerVersionInfo:configuration:error:] CoreML frame 45: +[MLModel modelWithContentsOfURL:configuration:error:] CoreML frame 46: @nonobjc MLModel.__allocating_init(contentsOf:configuration:) GKPersonalV2 frame 47: MDNA_GaitEncoder_v1_3.__allocating_init(contentsOf:configuration:) frame 48: MDNA_GaitEncoder_v1_3.__allocating_init(configuration:) frame 50: GaitModelInference.loadModel() frame 51: GaitModelInference.init() iOS version: Reproduced on iOS 26.4. Works correctly on iOS 26.3.1. Xcode version: 26.2 Device: iPhone (model used in testing) Model format: .mlpackage

We have an application that capture audio and video. App captures audio PCM on internal or external microphone and displays audio level on the screen. App was working fine for many years but after iOS 26.4 upgrade, averagePowerLevel and peakHoldLevel are stuck to -120 values. Any suggestion?

I’m encountering a persistent issue with my Network Extension (specifically NEFilterDataProvider) and would really appreciate any insights. The extension generally works as expected, but after some time — especially after sleep/wake cycles or network changes — a global network outage occurs. During this state, no network traffic works: pings fail, browsers can’t load pages, etc. As soon as I stop the extension (by disabling it in System Preferences), the network immediately recovers. If I re-enable it, the outage returns instantly. I’ve also noticed that once this happens, the extension stops receiving callbacks like handleNewFlow(), and reinstalling the app or restarting the extension doesn’t help. The only thing that resolves the issue is rebooting the system. After reboot, the extension works fine again — until the problem reoccurs later. I asked AI about this behavior, and it suggested the possibility that the kernel might have marked the extension as untrusted, causing the system to intentionally block all network traffic as a safety mechanism. Has anyone experienced similar behavior with NEFilterDataProvider? Could there be a way to detect or prevent this state without rebooting? Is there any logging or diagnostic data I should collect when it happens again? Any guidance or pointers would be greatly appreciated. Thanks in advance!

Hi Apple Community, We are using Automted Device Enrollment to Enroll macOS Devices and we used to Create AutoAdmin, PrimaryAccount using the Command Account Configuration . As a Part of Primary Account Creation while testing we see that BootStrap Token is Escrowed to MDM, and SecureToken is Created to Primary Account. The Primary Account user will enable FileVault as part of our process. As Tested internally, we seen that SecureToken is escrowed to AutoAdmin only when BootStrapToken is escrowed to MDM By device and AutoAdmin logs in then. That too After FileVault Unlock Since we Sendout the Laptop to users to setup themselves there are no chances of AutoAdmin Login to occur. And it defeats the purpose of having the AutoAdmin Account in emergency situation to login into it from Login Window. Can someone confirm if this behavior is expected and what are the expectation and recommendations from Apple on when to use AutoAdmin Account. Is there any other ways to use AutoAdmin directly from LoginWindow Before To FileVault Disk Unlock