Network Extension

RSS for tag

Customize and extend the core networking features of iOS, iPad OS, and macOS using Network Extension.

Posts under Network Extension tag

200 Posts

Post

Replies

Boosts

Views

Activity

Network Extension Resources
General: Forums subtopic: App & System Services > Networking DevForums tag: Network Extension Network Extension framework documentation Routing your VPN network traffic article Filtering Network Traffic sample code TN3120 Expected use cases for Network Extension packet tunnel providers technote TN3134 Network Extension provider deployment technote TN3165 Packet Filter is not API technote Network Extension and VPN Glossary forums post Debugging a Network Extension Provider forums post Exporting a Developer ID Network Extension forums post Network Extension Framework Entitlements forums post Network Extension vs ad hoc techniques on macOS forums post Network Extension Provider Packaging forums post NWEndpoint History and Advice forums post Extra-ordinary Networking forums post URL filter: WWDC 2025 Session 234 Filter and tunnel network traffic with NetworkExtension URL filters documentation Filtering traffic by URL sample code Setting up a PIR server for URL filtering sample code Using the Bloom filter tool to configure a URL filter sample code PIR Service Example open source server sample and specifically its documentation Wi-Fi management: Understanding NEHotspotConfigurationErrorInternal forums post See also Networking Resources for general networking resources, including information about Wi-Fi. Share and Enjoy — Quinn “The Eskimo!” @ Developer Technical Support @ Apple let myEmail = "eskimo" + "1" + "@" + "apple.com"
0
0
4.3k
Aug ’26
macOS: supported absolute deadline at final network egress, including queued bytes?
Hi, I’m assessing feasibility before implementing a small macOS HTTPS snapshot client. The requirement is strict: every request-bearing byte must cross the final controlled network-egress boundary before an absolute deadline. A byte crossing at or after the deadline fails. For context, the intended deadline is 07:41:00 Europe/London on a specified date, converted to UTC. This concerns request transmission, not response completion. An application precheck or a deadline to begin sending would not satisfy it. Is there a supported macOS API or mechanism with a documented guarantee covering this requirement? Specifically: Where is the enforcement boundary? Does expiry cover request bytes already queued in TLS/TCP, kernel, driver or NIC buffers, partially transmitted requests, and retransmitted request payload? Does enforcement remain fail-closed if the client or enforcement process is suspended, hangs or crashes, or the Mac sleeps and wakes? Can queued bytes resume transmission after expiry? What clock and deadline semantics apply, including equality, uncertainty, clock steps/regression and unavailable or unreliable time? Can connection reuse or buffering bypass expiry? Application retries, redirects and additional requests would be prohibited. How could an implementation observe and qualify the actual controlled egress boundary, rather than an earlier callback or queue? I have reviewed documentation for: NWConnection.ContentContext.expirationMilliseconds, which describes expiry for beginning transmission; NWConnection.SendCompletion.contentProcessed; Network Extension packet filtering, delayed-packet verdicts and provider lifecycle; TN3134 and TN3165. I have not found a documented end-to-end guarantee covering the cases above. I am not assuming that filters, cancellation, timers or socket timeouts provide it. Could an Apple engineer point to the applicable supported contract and platform restrictions, or clarify that this strict guarantee is not provided? If a mechanism is applicable, please identify supported macOS versions, required entitlements and any relevant interface or hardware limitations. This is a documentation/feasibility question only. No external service, account, credentials or capture is involved. An authoritative reference—or an explicit statement of unsupported cases—would be helpful. Many thanks in advance.
0
0
7
1h
Network Extension Issue - No cfil verdict callback registered - (27.3 beta 2)
I've filed FB25612343 in case others see it as well where 27.3b2 seems to cause issues with our network extension transparent proxy. The sysdiagnose shows the failure inside Network Extension, in the redirector process. Those lines (flow is closed for writes, Failed to write flow Data, No cfil verdict callback registered) are Apple's. There is no crash.Has anyone else seen issues on the latest beta?
6
0
132
1d
NEURLFilter: neagent keeps every prefilter in memory and is killed at ~90 MiB when a VPN shares it
Hello, We are building a URL filter with NEURLFilterControlProvider and PIR on iOS 27.2. Our blocklist needs to scale to 50 million URLs at a 0.1% false-positive rate. Following the documented Bloom filter format, that is a prefilter of 718,879,379 bits with 10 hash functions (89,859,923 bytes), which we return from fetchPrefilter(existingPrefilterTag:) as NEURLFilterPrefilter.PrefilterData.temporaryFilepath. On an iPhone 17 running iOS 27.2, we see two behaviours that make a prefilter of this size unreliable for users who also run a VPN. neagent keeps each prefilter it receives in its own memory and does not release the previous one After the first handover, neagent's memory footprint was 89.6 MiB, almost all of it anonymous memory, even though lookups are served from the mmap'd file under /private/var/db/urlPrefilter/. Each time we returned a new prefilter (with a new tag), the footprint grew by the size of the new bitmap, in the same neagent process: 89.6 MiB, then 150.8 MiB after a 64 MB prefilter, then 236.9 MiB after another 90 MB prefilter. It stayed at that level for the more than 25 minutes we observed. With no other network extension installed, neagent was not terminated at 236.9 MiB. When a packet tunnel VPN shares the same neagent, neagent is terminated at about 90 MiB With a packet tunnel VPN enabled (Tailscale), which iOS hosts in the same neagent process as our URL filter, neagent was terminated with JETSAM_REASON_MEMORY_HIGHWATER while holding a single prefilter at 89.4–89.7 MiB. This happened 11 and 35 minutes after start in our latest runs, with no PIR lookups and no new prefilter, and repeatedly in earlier runs (between 3 seconds and 52 minutes after start). Each termination stops both sessions: the URL filter with "Plugin failed" and the VPN with "Plugin initiated". Both restart a few seconds later, and the cycle repeats. Because a new prefilter adds a full copy (finding 1), any prefilter update also takes neagent over this limit straight away. Our questions: What is the largest NEURLFilterPrefilter that iOS supports, and does the limit depend on the device? Is neagent meant to keep the whole bitmap in its own memory, rather than read it from the mmap'd file? Should the previous prefilter's memory be released when a new one is installed? If not, how often should a control provider return a new prefilter? Is neagent's memory limit lower when a VPN shares the process, and is that intended? Can a URL filter run in its own neagent process? What approach do you recommend for a blocklist of tens of millions of URLs at a 0.1% false-positive rate? Thank you, David. Apple Developer Team ID: SSYCLSN43U
0
0
43
2d
Network Extension URL Filter registration pending review for more than a month
Hello, I am writing about a Network Extension URL Filter configuration submitted through the Identity & Trust section of the CloudKit Console on 5 September 2026 Status shown: In Review / Waiting for Approval Last Modified: 5 September 2026 (unchanged since submission) It has now been 37 days with no response — no approval, no rejection, and no request for changes. The PIR server , Privacy Pass issuer and HTTP/2 OHTTP gateway have been live and serving since the day of submission, configured per the Onboarding.md guidance in Apple's PIR service example, including the apple-url-filter DNS TXT record on the service URL and the url-filter-test validation entry. Could anyone please let me know: Whether this configuration is actually in a review queue, and the expected timeframe; Whether anything further is required from our side — if so, we will act on it the same day. The URL filter is a core protection feature of our app, which is currently in App Store review; the feature stays disabled for users until this registration is approved, so the delay is holding up what we can ship. Thank you for your help.
5
0
319
2d
Local Push Connectivity entitlement request — no response after a month (again)
Hi all, I’m hoping someone from Apple can point me in the right direction. We submitted a request for the Local Push Connectivity entitlement about a month ago and haven’t heard anything back. No approval, no rejection, just silence. Unfortunately this is the second time; we tried back in 2023 and the same thing happened. After requesting the entitlement multiple times, we gave up on the project. A bit about what we’re building: we make a nurse call system for long-term care facilities. When a resident needs help, they press a wireless call device, our on-prem server raises an alert, and our iOS app would notify the caregivers on shift. The catch is that a most of these facilities put their life safety systems on air-gapped networks with no internet access at all, so APNs just isn’t an option. Even where there is internet, the system has to keep working if the connection goes down. We understand other vendors in our industry use Local Push Connectivity for this same scenario, so we’re hopeful this is just a matter of our request getting lost somewhere. As far as I can tell, this is exactly what Local Push Connectivity was made for. The WWDC20 session that introduced it (“Build local push connectivity for restricted networks”) actually names hospitals as one of the example environments where APNs may be unreachable, and talks about keeping notifications working in constrained networks “when it matters most to the users.” A resident calling for help is about as close to that as it gets. Our App ID already has the Critical Alerts entitlement, but there’s no “Network Extensions (additional values)” option under Additional Capabilities, so Xcode, understandably, complains that the profile doesn’t match the com.apple.developer.networking.networkextension value in our entitlements file. Is there a way to check on the status of a request like this, or a better place to follow up? I’m happy to share our team and app details privately. Any help would be really appreciated. We’ve been trying to get this going for a while now. Thanks!
0
0
41
2d
NEURLFilterManager: inconsistent per-domain blocking (some sites never block, one flaps) - confirmed PIR database is correct
Hello, Some context: DropBet is an app that helps people with gambling addiction stay off gambling sites and apps, using NEURLFilterManager/PIR to block a large list of known gambling domains system-wide. For a lot of our users this filter is a core part of their recovery > reliability really matters here. Since fixing an unrelated HPKE encryption bug (separate thread, now closed), the filter works end-to-end for the first time, but blocking is inconsistent per domain in a way I can't explain from my side: winamax.fr and the Winamax app: blocked reliably, every single time > clean "connection failed" style block, confirmed to load fine with protection off. unibet.fr and betclic.fr: never blocked, consistently > confirmed even after a full device reboot (rules out stale caching), and confirmed identical across both Safari and Chrome on iOS tested side-by-side (so not browser-specific). olybet.fr: flaps > the page reloads every 2-3 seconds with protection on, also identical across both browsers. Confirmed this only happens with the filter enabled (tested toggling protection off/on against the same page). unibet.fr auto-redirects into the native Unibet app via a Universal Link before the page loads, unlike betclic.fr which stays in the browser > different code path, possibly relevant. To rule out any issue on our end, I wrote a standalone test that queries our deployed PIR database shards directly - bypassing OHTTP, the app, and the device entirely - for all of the domains above plus their www. variants. All of them correctly return the blocked value, no exceptions, including unibet.fr and betclic.fr which never block on the actual device. So the database, the PIR protocol implementation, and the server are confirmed correct for every domain I've tested. This points at something in how NEURLFilterManager/ciphermld applies a correctly-received PIR verdict on-device - possibly related to query batching, an evaluation-key edge case, or Universal Link handling bypassing the filter before a web request is even made. Is there anything you can share about how the client applies results once decrypted, or known issues around any of this?
7
0
347
3d
macOS Transparent Proxy fails to start before login on Wi-Fi: "No network available"
I'm investigating an issue when deploying my VPN agent via Jamf during initial setup on clean macOS machines. The Network Extension process starts immediately after installation, but its Transparent Proxy provider starts only after a user logs in. I tried to reboot, and still observed the transparent proxy start fails. 10:00:52 === system boot 10:00:57.535 NESMTransparentProxySession[Primary Tunnel:Prisma Access Agent:68CAC11B-...]: Received a start command from PASrv[303] 10:00:57.538 NESMTransparentProxySession[...]: No network available <<<<<<<<< 10:00:57.541 NESMTransparentProxySession[...]: status changed to disconnected, last stop reason No network available <<<<<<<<< 10:01:21.541 NESMTransparentProxySession[...]: handleChangeEventForRankedInterfaces - stopped 0 status disconnected includeAllNetworks 0 10:01:26.373 NESMTransparentProxySession[...]: handleChangeEventForRankedInterfaces - stopped 0 status disconnected includeAllNetworks 0 10:06:01.293 NESMTransparentProxySession[...]: Received a start command from user_context[1547] 10:06:01.615 NESMTransparentProxySession[...]: status changed to connectedOur product requires pre-login functionality. Is this behavior expected, and how can we enable the Transparent Proxy provider before user login? I suspect a race between network availability and the attempt to start the Transparent Proxy. The issue does not occur with Ethernet, possibly because connectivity becomes available earlier than with Wi-Fi. Should we call startVPNTunnel again once connectivity becomes available? If so, what is the recommended way to detect this network transition?
5
0
156
4d
Egress addresses of the URL filter OHTTP relay
Hello, I am wondering about the egress addresses of the Oblivious HTTP relay that Apple runs for Network Extension URL filters. I am building an iOS app that uses NEURLFilterManager. The app builds a Bloom filter from the block lists a user picks. The system only asks our server about a site when that filter has a hit. Each lookup is encrypted with keyword PIR and carries a Privacy Pass token instead of anything tied to the user. It reaches our server through your relay, which strips the device's address. We run our own PIR server, Privacy Pass issuer and OHTTP gateway instead of a hosted service. The gateway is the one part of the setup that should only ever hear from your relay. Since every real lookup comes through the relay, I would really like to drop all other traffic to it at the firewall. Right now I can't do that. I also can't rate limit by address because every legitimate request shares the relay's addresses. As a result the gateway just has to stay open to everyone and rely on global limits, which is a far cry from what I'd want for a service that only one client is meant to reach. I have three questions about this. 1) Are the relay's egress address ranges published somewhere, or can they be shared with developers who have been onboarded? 2) If they are shared, how would we find out when they change and how much notice there is? 3) If the addresses can't be shared, is there another way to tell that a request came from the relay, such as a header or a client certificate? Please let me know if any of this is unclear or if there is a better place to ask. Thank you in advance, Alexander
1
0
92
5d
Kernel tag check fault after wake on Mac17 hardware, Google Chrome Helper, macOS 26.5–27.0.1 — related to r.172793638?
We're seeing a recurring kernel panic across six Mac17-family Macs (Mac17,2 / Mac17,4 / Mac17,8; T8142 and T6050) in a managed fleet. 34 panic logs collected 30 Sept – 9 Oct, all with the same signature: panic(cpu N caller ...): Kernel tag check fault (expected tagged address: 0x...028), esr 0x96000011 Faulting address is always at offset 0x28 of the object Same crash PC per build: kernel text exec base + 0x48da38 on 27.0/27.0.1, + 0x46a784 on 26.7, + 0x466ee0 on 26.5 Panicked task is "Google Chrome Helper" in all 34 cases, thread Chrome_ChildIOThread (Chrome's IPC/network I/O thread) Always 3–60 s after wake from sleep, usually a dark wake with the lid closed (overnight or after a 3–18 min sleep) No third-party kexts loaded; "Post-boot loaded kexts" lists only Apple drivers Affected builds: 26.5 (25F71), 26.7 (25G229), 27.0 (26A428), 27.0.1 (26A434). Not a regression from one update: a machine still on 26.5 panics identically. Only Mac17-family hardware is affected. Older Apple silicon Macs with the same software stack are fine. Present on every affected machine: Chrome PWAs (Gmail/Chat/Meet) running in the background, keeping Chrome's network service alive across sleep. On one machine Microsoft Edge is the primary browser and never panics; only Chrome Helper does. Microsoft Defender for Endpoint with its network filter (content filter) extension, MDM-managed. This looks related to thread 821372, where DTS described a kernel networking panic on M5-family hardware with a Network Extension present (r.172793638 / FB22401652), fixed in 26.5b2. We appear to be seeing a tag-check-fault variant on 26.5 release and later. Questions: Is this a known issue, and is it the same root cause as r.172793638? Is there a build where it's fixed? Any recommended mitigation beyond preventing sleep and quitting Chrome? Feedback Assistant: FB25128015 with panic logs and sysdiagnose. Happy to collect more data or test a build. Backtrace from a 27.0.1 panic (kernel text exec base 0xfffffe003aa54000), panic PC first: lr: 0xfffffe003aee1a38 lr: 0xfffffe003aeec580 lr: 0xfffffe003b12cf34 lr: 0xfffffe003b12dc50 lr: 0xfffffe003b07b1f8 lr: 0xfffffe003b07c758 lr: 0xfffffe003b237ff4 lr: 0xfffffe003ac787f0 lr: 0xfffffe003aa576ac lr: 0x000000018747a57c (userspace)
4
0
393
1w
URL Filter (Oblivious HTTP) configuration In Review for almost a week with no feedback
Hello, I submitted a Network Extension URL Filter (Oblivious HTTP) configuration in CloudKit Console almost a week ago. It still says In Review / Waiting for Approval and I haven't heard anything back since the submission notice. I wanted to check whether anything is missing on my end. Just in case, the configuration ID is 60802ed4-abfa-4b20-a64e-0d35230f478a. The app is an open-source ad and tracker blocker from the developers of an existing content blocker with a large user base. On iPhone and iPad its system-wide blocking runs entirely through the URL filter. Until the configuration is approved the iOS app can't block anything at all, and our testers are left with a Mac-only product in the meantime. Because lookups go through Apple's relay to our own PIR server, we never learn which sites anyone visits. All four endpoints have been live since the day I submitted and the TXT records are in place. Every database we build also includes the www.apple.com/url-filter-test validation entry. Is there anything else I need to provide, or a rough timeline I should expect? Thank you in advance!
1
0
86
1w
Apple thought it was impossible
Apple thought it was impossible. I made it possible. No server. Just an iPhone. LUMEN-JEV OS v3.0 — On-device OS using Pyto + iCloud Drive + Obsidian Architecture: Pyto runtime = server: Persistent loop, 922 tickers scanned, 502.6 ev/s, 12ms latency, sub-2ms cache iCloud Drive = bus: Writes JEV-BUS.md every 15s, shared working directory Obsidian = OS window manager: 00-JEV-DASHBOARD-main 48 nodes live Vault: 11/11 LOCKED, Factory Complete Anchor: 572b9df5e8e1f101365f8316ba7b34da4aa54ff89a2dcdb68ed20b126e8cabed Chain: REAL-EXCHANGE-PUBLIC-WS -> 8-MODULE-DEMO -> DEMO-TERMINATED -> NO EXECUTION Feed: real-public-readonly / public_ticker_only Modules (8-MODULE-v1px): WARREN-WATCHDOG, KEEPER-BOX, NODE-MODE-SYNC, VAULT-MULT, SCOUT-SCAN, SIREN-FILTER, FOCUS-SIGNAL-20K, ORACLE-PRICE-FEED, RELAY-ROUTER, AKE-ARBITRAGE, YIELD-BACK-4, MEDUSAs: 03 DYNAMIC OBSERVER-SPACE Context: Built in 33hrs over 11 nights, day job, single dad, solo dev, on iPhone only. Pyto working directory set to iCloud Drive. No Mac backend, no CloudKit function, no external server. Why this matters for iOS: Apple's sandbox says you can't run a persistent OS-like loop with live file sync on iOS. Pyto + iCloud Drive persistence proves you can. iPhone becomes the server. Proof: 48 NODES LIVE canvas PROOF CHAIN 11/11 FLAT VIEW VAULT LOCKED Jev.py V2.6 CLEAN BUS log (checked 50/100/150/200 -> 922/922 passes) Anchor hash matches zip JEV-OS-v3.0-572b9df5-40files.zip Requesting: Guidance on showcasing this as an iOS innovation + Ask Apple office hours. X proof pinned: @Activepython007
23
1
1.5k
1w
[macOS 27.2]: System-Wide DNS Resolution Failures After Sleep/Wake – mDNSResponder and NETransparentProxyProvider
Summary Recent macOS release observed intermittent system-wide DNS failures after sleep/wake on multiple devices running Global Secure Access (GSA) with NETransparentProxyProvider. Previously successful DNS lookups time out after wake across applications, including browsers and GSA. Two captures contain failure-time evidence of the mDNSResponder resolver queue predominantly waiting in close()/kernel. Other captures show similar DNS failures, but a shared root cause has not been established. Call chain, from thread entry to kernel wait: start_wqthread + 8 _pthread_wqthread + 292 _dispatch_workloop_worker_thread + 720 _dispatch_root_queue_drain_deferred_wlh + 284 _dispatch_lane_invoke + 392 _dispatch_lane_serial_drain + 332 _dispatch_source_invoke + 920 _dispatch_source_cancel_callout + 208 _dispatch_continuation_pop + 596 _dispatch_client_callout + 16 close + 8 fleh_synchronous + 72 sleh_synchronous (sleh.c:1610) unix_syscall (systemcalls.c:181) fp_close_and_unlock_with_action (kern_descrip.c:1874) fg_drop (kern_descrip.c:287) soclose (uipc_socket.c:1345) soclose_locked (uipc_socket.c:1138) cfil_sock_close_wait (content_filter.c:5743) soflow_db_apply (socket_flows.c; line unavailable) cfil_filters_udp_attached_per_flow (content_filter.c:6675) _sleep (kern_synch.c:213) lck_mtx_sleep_deadline (locks.c:1583) thread_block_reason (sched_prim.c:4060) thread_invoke (sched_prim.c:3443) Steps to reproduce The issue is intermittent. These steps describe the observed sequence, not a deterministic minimal reproduction. Run the GSA client using NETransparentProxyProvider. The affected configurations also have Microsoft Defender network extensions active. Verify that a system DNS lookup succeeds for a known hostname. Allow the Mac to sleep, including maintenance DarkWake cycles. Wake the Mac and repeat system DNS lookups or use applications requiring hostname resolution. Observe repeated approximately 30-second lookup timeouts, sometimes longer. Observed results November 5, 2026 - UTC+05:30 capture User-reported impact: Recurrence of the same sleep/wake connectivity symptoms. OS: macOS 27.2 (26B123). Incident window: 20:38-20:41 IST / 15:08-15:11 UTC. The same Chrome client/name A lookup succeeds earlier in 12.984 ms but times out after wake in 29,999.705 ms. The DNS service/server route changed, so this is not a controlled identical-network comparison. The resolver records 1,007 timeout callbacks during the incident window, across browsers, Teams, GSA, Defender and other processes. The failure-time spindump at 20:40:21.009-20:40:31.013 IST shows the original mDNSResponder PID 78534 with 715/715 resolver-queue samples under dispatch-source cancellation → close() → kernel → lck_mtx_sleep_deadline. Source: spindump.txt:75998-76093. This establishes a sampled wait, not a permanent deadlock or proof of the same root cause as November 2. The file descriptor and kernel/filter wait owner remain unidentified. A later matching Chrome lookup returns an address at 20:43:18.191816 IST on the same resolver PID, without a recorded resolver restart. This establishes some recovery, not that every application or tunnel recovered. November 2, 2026 - UTC-08:00 capture User-reported impact: No network upon wake. This eventually resolves without a reboot, after approximately 5 to 15 minutes of waiting. OS: macOS 27.2 (26B123). Incident window: 09:40-09:45 PST / 17:40-17:45 UTC. The same Edge A lookup succeeds before sleep in 19.346 ms but times out after wake in 29,999.826 ms. The resolver records 1,403 timeout callbacks during the incident window. The failure-time spindump shows 770/777 resolver-queue samples under dispatch-source cancellation → close() → kernel; 769 include lck_mtx_sleep_deadline. This establishes a dominant sampled wait, not a permanent deadlock. The file descriptor and kernel/filter wait owner remain unidentified. November 3, 2026 - UTC-08:00 capture (Mihai) Evidence: mihai_gsa_error.tar.gz The following example records a failure resolving GSA’s own control-plane endpoint, whose DNS lookup is not intercepted by the GSA tunnel. Times in these excerpts are local (UTC-08:00). GSA error [2026-11-03 10:13:58.463] Network connection to the Global Secure Access Edge could not be established StreamId: ControlStream Hostname: <tenant>.auth.client.globalsecureaccess.microsoft.com:443 Error: hostname lookup error: Address lookup failed os_error: nodename nor servname provided, or not known syscall: getaddrinfo mDNSResponder records 10:13:19.663661 [R125367->DupQ50990->Q38408] Duplicate question <private> dn:04wpeA (Addr) 10:13:58.468519 [R125367] DNSServiceQueryRecord STOP -- name hash: dn:04wpeA, duration: 39s 10:13:59.078847 [R125340] QueryRecordOpCallback: Question <private> dn:04wpeA (Addr) timing out, InterfaceID 0x0 Additional observations The original resolver records 11 timeout callbacks across GSA, Defender, Spotify and the system time service. A GSA A-query returns a positive result in 1.282 ms earlier (cache-compatible, not a cold upstream baseline), then the same client/name/qtype/flags times out after 41.865784 wall-clock seconds; that interval includes sleep. The DNS service/server set changes, so this is not a controlled same-network comparison. At 09:40 PST the Wi-Fi interface becomes inactive and loses its IPv4 address, followed by no-dns-service results. The first native lookup errors precede this link loss; it does not explain the entire case. Four GSA A/AAAA timeout transactions have matching late UDP replies. For one A-query, the reply is logged 25.095999 seconds after its matching send and 7.197434 seconds after the API timeout, during shutdown. 48 acceptable UDP response records appear within 1.467 ms. The location of the delay is not identified.
5
1
465
1w
Is it safe to strip with All Symbols for a content filter?
We would like to minimise the amount of code exposed in the product we have developed and an obvious option is to configure our build setting to use the All Symbols for Strip Style. Some investigation with an AI tool reports there are possible problems doing so. ContentFilter: non-global → all — risky. The NetworkExtension and XPC subsystems load your bundle and look up classes by name through the ObjC runtime, which uses __objc_classlist and friends (not the Mach-O symbol table), so in principle all works. In practice it occasionally breaks bundle loading in edge cases. If you try this, test specifically that systemextensionsctl activates the bundle and the filter actually traps packets. Is it safe to use All Symbols with a content filter or do we need to accept Non-Global symbols? Thanks, Dave
1
0
176
1w
NETransparentProxyManager: saving isOnDemandEnabled = true disables On-Demand on other proxies
On macOS, multiple NETransparentProxyManager (NEAppProxyProvider) configurations from different vendors can be enabled (isEnabled = true) and run simultaneously. However, we discovered that when one NETransparentProxyManager configuration calls saveToPreferencesWithCompletionHandler: with isOnDemandEnabled = true, /usr/libexec/nehelper (NEConfigurationManager) silently modifies any other enabled NETransparentProxyManager configuration in /Library/Preferences/com.apple.networkextension.plist and flips its OnDemandEnabled to false (and nesessionmanager logs Resetting VPN On Demand and marks the other session not-kernel-activated). Unlike NEVPNManager.isEnabled (which documents mutual exclusion for personal VPNs), NETransparentProxyManager is designed so multiple transparent proxies can coexist simultaneously, and neither NETransparentProxyManager.h nor NEVPNManager.h documents that isOnDemandEnabled is mutually exclusive across coexisting transparent proxy configurations. Root Cause Observed in NetworkExtension.framework When inspecting -[NEConfigurationManager saveConfigurationToDisk:updateSCPreferences:currentSignature:userUUID:notifyNow:isUpgrade:], when iterating over existing NEConfiguration objects of the same grade: If [[newConfig VPN] isEnabled] && [[otherConfig VPN] isEnabled] && [[newConfig VPN] isOnDemandEnabled] && [[otherConfig VPN] isOnDemandEnabled], NEConfigurationManager invokes: +[NEConfigurationManager disableConfiguration:otherConfig onDemandOnly:YES] disableConfiguration:onDemandOnly:YES calls [[otherConfig VPN] setOnDemandEnabled:NO], even when both configurations are NETransparentProxyManager (TunnelType == 2). Real-World Impact When two enterprise security products that both use NETransparentProxyManager are installed on the same Mac: Both configure isEnabled = true, onDemandRules = [NEOnDemandRuleConnect()], and isOnDemandEnabled = true. Whenever Proxy A saves its preferences, nehelper silently flips Proxy B's isOnDemandEnabled to false (and vice versa). After a reboot or network transition where Proxy B's initial startVPNTunnel fails while Wi-Fi is still associating and Proxy B relies on VPN On Demand (NEOnDemandRuleConnect) to start its NEAppProxyProvider once network connectivity comes up, macOS never starts Proxy B because its isOnDemandEnabled flag was cleared. Minimal Reproducer (Swift) Compile this minimal CLI inside an app bundle signed with the com.apple.developer.networking.networkextension = ["app-proxy-provider"] entitlement: import Foundation import NetworkExtension let args = CommandLine.arguments let name = args.count > 1 ? args[1] : "PocProxyA" let action = args.count > 2 ? args[2] : "status" DispatchQueue.main.async { NETransparentProxyManager.loadAllFromPreferences { managers, error in if let error = error { print("[\(name)] loadAllFromPreferences error: \(error)") exit(1) } let all = managers ?? [] let existing = all.first(where: { $0.localizedDescription == name }) if action == "status" { if let mgr = existing { print("[\(name)] -> isEnabled=\(mgr.isEnabled), isOnDemandEnabled=\(mgr.isOnDemandEnabled), rulesCount=\(mgr.onDemandRules?.count ?? 0)") } else { print("[\(name)] -> not found") } exit(0) } if action == "remove" { guard let mgr = existing else { exit(0) } mgr.removeFromPreferences { err in print("[\(name)] removed (err=\(String(describing: err)))") exit(0) } return } // action == "save" let mgr = existing ?? NETransparentProxyManager() let proto = (mgr.protocolConfiguration as? NETunnelProviderProtocol) ?? NETunnelProviderProtocol() proto.providerBundleIdentifier = "com.example.poc.extension" proto.serverAddress = "127.0.0.1" mgr.protocolConfiguration = proto mgr.localizedDescription = name mgr.isEnabled = true mgr.onDemandRules = [NEOnDemandRuleConnect()] mgr.isOnDemandEnabled = true print("[\(name)] Saving NETransparentProxyManager with isEnabled=true, isOnDemandEnabled=true...") mgr.saveToPreferences { err in if let err = err { print("[\(name)] saveToPreferences error: \(err)") exit(1) } print("[\(name)] saveToPreferences succeeded! isEnabled=\(mgr.isEnabled), isOnDemandEnabled=\(mgr.isOnDemandEnabled)") exit(0) } } } RunLoop.main.run()Reproduction Output Running save on PocProxyA, then save on PocProxyB, and checking status on both: === Step 1: PocProxyA saves isOnDemandEnabled = true === [PocProxyA] Saving NETransparentProxyManager with isEnabled=true, isOnDemandEnabled=true... [PocProxyA] saveToPreferences succeeded! isEnabled=true, isOnDemandEnabled=true [PocProxyA] -> isEnabled=true, isOnDemandEnabled=true, rulesCount=1 === Step 2: PocProxyB saves isOnDemandEnabled = true === [PocProxyB] Saving NETransparentProxyManager with isEnabled=true, isOnDemandEnabled=true... [PocProxyB] saveToPreferences succeeded! isEnabled=true, isOnDemandEnabled=true === Step 3: Check BOTH configurations after PocProxyB saved === [PocProxyA] -> isEnabled=true, isOnDemandEnabled=false, rulesCount=1 [PocProxyB] -> isEnabled=true, isOnDemandEnabled=true, rulesCount=1 === Step 4: PocProxyA saves isOnDemandEnabled = true again === [PocProxyA] Saving NETransparentProxyManager with isEnabled=true, isOnDemandEnabled=true... [PocProxyA] saveToPreferences succeeded! isEnabled=true, isOnDemandEnabled=true === Step 5: Check BOTH configurations after PocProxyA saved === [PocProxyA] -> isEnabled=true, isOnDemandEnabled=true, rulesCount=1 [PocProxyB] -> isEnabled=true, isOnDemandEnabled=false, rulesCount=1Questions Is disableConfiguration:onDemandOnly:YES stripping isOnDemandEnabled across coexisting NETransparentProxyManager (TunnelType = 2) configurations a bug in NEConfigurationManager, or an intentional limitation? Is there a recommended way for two coexisting NETransparentProxyManager configurations from different vendors to both use On-Demand without disabling each other's isOnDemandEnabled flag on saveToPreferences? Filed in Feedback Assistant as FB25463155.
3
2
376
1w
NEAppPushProvider times out connecting to a server on a routed subnet, while the containing app connects fine
We use Local Push Connectivity (NEAppPushProvider, matchSSIDs set) to receive notifications from an on-premise MQTT broker over WebSocket (wss://host:port/mqtt). iPhone on Wi-Fi, VLAN A (SSID matches matchSSIDs) Server on wired VLAN B, routing and firewall rules allow VLAN A → VLAN B on that port Result: the containing app, in the foreground, connects successfully to wss://<VLAN B IP>:<port>/mqtt. At the same moment, the app push provider (URLSessionWebSocketTask to the exact same URL) fails with "The request timed out". The provider starts normally (start() is called). When the server has an address on the Wi-Fi subnet (VLAN A), the provider connects without any issue. Is the app push provider restricted to servers on the local subnet of the matched Wi-Fi network? If so, is this documented anywhere? If not, what could prevent the provider from reaching a routed address that the app itself can reach? iOS version: 27.0.1 device : iPhone 13
1
0
103
1w
NEURLFilter configuration still "Queued" six weeks after submission
Our Network Extension URL Filter configurations in CloudKit Identity & Trust are not moving through onboarding. STG: configuration 6e17673d-dbaa-4b35-98c5-ff9e7a357577, bundle ca.itellyou.stg. Submitted 2026-09-25, status Approved, Onboarding Status "Queued". PRD: configuration 06fa2558-281b-4b55-b59f-402ecf359d18, bundle ca.itellyou. Submitted 2026-10-20, status In Review. Evidence (iPhone 17 Pro Max, 2026-11-06, same commit, same server, same user token): Development-signed build: the filter reaches "running" and blocks our test domains. Ad-hoc (distribution-signed) build: it cycles starting -> stopped with NEURLFilterManager.Error 9. The PIR server, Privacy Pass issuer and OHTTP gateway are live, and the test entry www.apple.com/url-filter-test is present. An Apple engineer confirmed on GitHub (apple/pir-service-example#185) that nothing is pending on our side. We also opened support case 30000007408474. Has anyone seen how long onboarding takes these days, or found a way to get it moving?
1
0
97
1w
Is a `flowClosed` report guaranteed when the owning process exits or crashes?
Our macOS NEFilterDataProvider sets shouldReport = true on flow verdicts and receives NEFilterReport.Event.flowClosed in handle(_:). In testing, we have observed close reports after the owning process exits or crashes, sometimes with a short delay. Is a flowClosed report guaranteed for every TCP and UDP flow in these cases? Does the behavior differ by protocol, and is there a documented upper bound on delivery time? We also set statisticsReportFrequency = .low. Does that setting affect only periodic .statistics reports, with no effect on flowClosed delivery?
2
0
134
2w
NEURLFilter PIR status check fails with CipherML errors (Code 1100/1800)
Hello ! following up from a related thread (account transfer issue, now resolved: https://developer.apple.com/forums/thread/843546?answerId=911820022), but this is a distinct, new issue so starting fresh as suggested. The extension launches fine, XPC handshake succeeds, and PIR registration completes successfully. But the subsequent NEPIRChecker status check fails every time (retried repeatedly over several minutes, same result each time): PIR status returned error <Error Domain=com.apple.CipherML Code=1100 "Unable to query status due to errors: Error details were logged and redacted." UserInfo={NSLocalizedDescription=Unable to query status due to errors: Error details were logged and redacted., NSUnderlyingError=0x677037cf0 {Error Domain=com.apple.CipherML Code=1800 "Error details were logged and redacted." UserInfo={NSLocalizedDescription=Error details were logged and redacted.}}}>Config being registered (confirmed from the logs): pirServerURL = https://pir.dropbet.app/ pirPrivacyPassIssuerURL = https://pir.dropbet.app/issueBoth endpoints are live and were already validated as part of the original NEURL Filter configuration approval. I've captured a sysdiagnose from the device right when this was happening, so I have that ready to attach to a bug report if useful — just let me know and I'll file one. Any idea what CipherML error 1100/1800 typically indicates?
8
0
515
2w
Cannot open Chrome UDP flows in Transparent Proxy Provider
We are implementing a Transparent Proxy for HTTPS (via TCP and QUIC). The following rules are set in startProxy: settings.includedNetworkRules = [ NENetworkRule(destinationNetwork: NWHostEndpoint(hostname: "0.0.0.0", port: "443"), prefix: 0, protocol: .TCP), NENetworkRule(destinationNetwork: NWHostEndpoint(hostname: "::", port: "443"), prefix: 0, protocol: .TCP), NENetworkRule(destinationNetwork: NWHostEndpoint(hostname: "0.0.0.0", port: "443"), prefix: 0, protocol: .UDP), NENetworkRule(destinationNetwork: NWHostEndpoint(hostname: "::", port: "443"), prefix: 0, protocol: .UDP) ] Handling TCP connections seems to work fine. But opening UDP flows from Chrome (or Brave) always fails with Error Domain=NEAppProxyFlowErrorDomain Code=2 "The peer closed the flow" (Doing the same for Firefox works!) BTW: We first create a remote UDP connection (using the Network framework) and when it is in the ready state, we use connection?.currentPath?.localEndpoint as the localEndpoint parameter in the open method of the flow. Is it a known issue that QUIC connections from Chrome cannot be handled by a Transparent Proxy Provider?
11
0
1.6k
2w
iPadOS: Intermittent disconnection after NEHotspotConfigurationManager automatic hotspot join
Background We are developing Share&Share, an offline device-to-device content transfer application. An iPad automatically joins a nearby Samsung Android LocalOnlyHotspot. The hotspot credentials are delivered through an authenticated, encrypted Bluetooth Low Energy (BLE) session. On iPadOS, our application decrypts the credentials natively in Swift and uses NEHotspotConfigurationManager.apply() to join the hotspot. We are investigating an intermittent connection problem during the initial automatic join. We have also submitted a private report through Apple Feedback Assistant: Feedback ID: FB25408475 Test environment Receiver: Physical iPad with a Home button. Sender/hotspot: Samsung SM-E055F. Android version: Android 16. Hotspot security: WPA2/WPA3 transition configuration. iOS implementation: Swift and NetworkExtension. joinOnce = false. The hidden property is not explicitly set. Observed behaviour Two independently instrumented automatic-join sessions showed a similar sequence: A1: The iPad associates with the Samsung hotspot and completes the WPA3 authentication exchange recorded by the Samsung. Approximately 3.35 seconds after the iPad-side join completion, iOS records a beacon-loss/roam declaration and the connection drops. A2: iOS attempts to reconnect, but the authentication exchange does not complete. A3: iOS automatically reconnects successfully, obtains a DHCP address and remains connected during the observation period. The same encrypted credentials work during both the initial and subsequent successful attempts. The apply() completion handler returns without an error. We understand that this does not establish that the device has joined successfully or that local IP connectivity is ready. What we have investigated The first join used cached beacon information approximately eight seconds old; the subsequent successful join used fresh scan information. AWDL activity was observed, but we have not established a causal relationship. Location Services scanning does not explain the first failure in our second instrumented session. The diagnostic label "Beacons Lost" is not being treated as proof of actual missed beacon frames. We have not independently established which WPA mode the iPad negotiated. We have not yet measured TCP readiness during the failed and successful attempts. We do not currently have a standalone minimal Xcode reproduction project. The behaviour was observed using our Flutter-based iOS probe and physical devices. Questions for developers familiar with NetworkExtension Has anyone observed intermittent first-join instability with NEHotspotConfigurationManager when joining a newly created Android LocalOnlyHotspot using WPA2/WPA3 transition mode? Are there documented limitations or known issues involving the first automatic join to a newly created hotspot profile? What is the recommended supported approach for determining that the intended Wi-Fi network is genuinely connected after apply() completes? What is the most reliable way to establish that local TCP communication is ready before beginning an application-level transfer? Are there additional public diagnostic facilities or measurements that could help distinguish Wi-Fi association instability from DHCP or subsequent network-path problems? We are looking for a documented, App Store-compatible approach. We are not assuming a particular root cause and do not want to rely on private APIs or undocumented system behaviour. Any practical experience or pointers to relevant Apple documentation would be appreciated.
1
0
175
2w
Network Extension Resources
General: Forums subtopic: App & System Services > Networking DevForums tag: Network Extension Network Extension framework documentation Routing your VPN network traffic article Filtering Network Traffic sample code TN3120 Expected use cases for Network Extension packet tunnel providers technote TN3134 Network Extension provider deployment technote TN3165 Packet Filter is not API technote Network Extension and VPN Glossary forums post Debugging a Network Extension Provider forums post Exporting a Developer ID Network Extension forums post Network Extension Framework Entitlements forums post Network Extension vs ad hoc techniques on macOS forums post Network Extension Provider Packaging forums post NWEndpoint History and Advice forums post Extra-ordinary Networking forums post URL filter: WWDC 2025 Session 234 Filter and tunnel network traffic with NetworkExtension URL filters documentation Filtering traffic by URL sample code Setting up a PIR server for URL filtering sample code Using the Bloom filter tool to configure a URL filter sample code PIR Service Example open source server sample and specifically its documentation Wi-Fi management: Understanding NEHotspotConfigurationErrorInternal forums post See also Networking Resources for general networking resources, including information about Wi-Fi. Share and Enjoy — Quinn “The Eskimo!” @ Developer Technical Support @ Apple let myEmail = "eskimo" + "1" + "@" + "apple.com"
Replies
0
Boosts
0
Views
4.3k
Activity
Aug ’26
macOS: supported absolute deadline at final network egress, including queued bytes?
Hi, I’m assessing feasibility before implementing a small macOS HTTPS snapshot client. The requirement is strict: every request-bearing byte must cross the final controlled network-egress boundary before an absolute deadline. A byte crossing at or after the deadline fails. For context, the intended deadline is 07:41:00 Europe/London on a specified date, converted to UTC. This concerns request transmission, not response completion. An application precheck or a deadline to begin sending would not satisfy it. Is there a supported macOS API or mechanism with a documented guarantee covering this requirement? Specifically: Where is the enforcement boundary? Does expiry cover request bytes already queued in TLS/TCP, kernel, driver or NIC buffers, partially transmitted requests, and retransmitted request payload? Does enforcement remain fail-closed if the client or enforcement process is suspended, hangs or crashes, or the Mac sleeps and wakes? Can queued bytes resume transmission after expiry? What clock and deadline semantics apply, including equality, uncertainty, clock steps/regression and unavailable or unreliable time? Can connection reuse or buffering bypass expiry? Application retries, redirects and additional requests would be prohibited. How could an implementation observe and qualify the actual controlled egress boundary, rather than an earlier callback or queue? I have reviewed documentation for: NWConnection.ContentContext.expirationMilliseconds, which describes expiry for beginning transmission; NWConnection.SendCompletion.contentProcessed; Network Extension packet filtering, delayed-packet verdicts and provider lifecycle; TN3134 and TN3165. I have not found a documented end-to-end guarantee covering the cases above. I am not assuming that filters, cancellation, timers or socket timeouts provide it. Could an Apple engineer point to the applicable supported contract and platform restrictions, or clarify that this strict guarantee is not provided? If a mechanism is applicable, please identify supported macOS versions, required entitlements and any relevant interface or hardware limitations. This is a documentation/feasibility question only. No external service, account, credentials or capture is involved. An authoritative reference—or an explicit statement of unsupported cases—would be helpful. Many thanks in advance.
Replies
0
Boosts
0
Views
7
Activity
1h
Network Extension Issue - No cfil verdict callback registered - (27.3 beta 2)
I've filed FB25612343 in case others see it as well where 27.3b2 seems to cause issues with our network extension transparent proxy. The sysdiagnose shows the failure inside Network Extension, in the redirector process. Those lines (flow is closed for writes, Failed to write flow Data, No cfil verdict callback registered) are Apple's. There is no crash.Has anyone else seen issues on the latest beta?
Replies
6
Boosts
0
Views
132
Activity
1d
NEURLFilter: neagent keeps every prefilter in memory and is killed at ~90 MiB when a VPN shares it
Hello, We are building a URL filter with NEURLFilterControlProvider and PIR on iOS 27.2. Our blocklist needs to scale to 50 million URLs at a 0.1% false-positive rate. Following the documented Bloom filter format, that is a prefilter of 718,879,379 bits with 10 hash functions (89,859,923 bytes), which we return from fetchPrefilter(existingPrefilterTag:) as NEURLFilterPrefilter.PrefilterData.temporaryFilepath. On an iPhone 17 running iOS 27.2, we see two behaviours that make a prefilter of this size unreliable for users who also run a VPN. neagent keeps each prefilter it receives in its own memory and does not release the previous one After the first handover, neagent's memory footprint was 89.6 MiB, almost all of it anonymous memory, even though lookups are served from the mmap'd file under /private/var/db/urlPrefilter/. Each time we returned a new prefilter (with a new tag), the footprint grew by the size of the new bitmap, in the same neagent process: 89.6 MiB, then 150.8 MiB after a 64 MB prefilter, then 236.9 MiB after another 90 MB prefilter. It stayed at that level for the more than 25 minutes we observed. With no other network extension installed, neagent was not terminated at 236.9 MiB. When a packet tunnel VPN shares the same neagent, neagent is terminated at about 90 MiB With a packet tunnel VPN enabled (Tailscale), which iOS hosts in the same neagent process as our URL filter, neagent was terminated with JETSAM_REASON_MEMORY_HIGHWATER while holding a single prefilter at 89.4–89.7 MiB. This happened 11 and 35 minutes after start in our latest runs, with no PIR lookups and no new prefilter, and repeatedly in earlier runs (between 3 seconds and 52 minutes after start). Each termination stops both sessions: the URL filter with "Plugin failed" and the VPN with "Plugin initiated". Both restart a few seconds later, and the cycle repeats. Because a new prefilter adds a full copy (finding 1), any prefilter update also takes neagent over this limit straight away. Our questions: What is the largest NEURLFilterPrefilter that iOS supports, and does the limit depend on the device? Is neagent meant to keep the whole bitmap in its own memory, rather than read it from the mmap'd file? Should the previous prefilter's memory be released when a new one is installed? If not, how often should a control provider return a new prefilter? Is neagent's memory limit lower when a VPN shares the process, and is that intended? Can a URL filter run in its own neagent process? What approach do you recommend for a blocklist of tens of millions of URLs at a 0.1% false-positive rate? Thank you, David. Apple Developer Team ID: SSYCLSN43U
Replies
0
Boosts
0
Views
43
Activity
2d
Network Extension URL Filter registration pending review for more than a month
Hello, I am writing about a Network Extension URL Filter configuration submitted through the Identity & Trust section of the CloudKit Console on 5 September 2026 Status shown: In Review / Waiting for Approval Last Modified: 5 September 2026 (unchanged since submission) It has now been 37 days with no response — no approval, no rejection, and no request for changes. The PIR server , Privacy Pass issuer and HTTP/2 OHTTP gateway have been live and serving since the day of submission, configured per the Onboarding.md guidance in Apple's PIR service example, including the apple-url-filter DNS TXT record on the service URL and the url-filter-test validation entry. Could anyone please let me know: Whether this configuration is actually in a review queue, and the expected timeframe; Whether anything further is required from our side — if so, we will act on it the same day. The URL filter is a core protection feature of our app, which is currently in App Store review; the feature stays disabled for users until this registration is approved, so the delay is holding up what we can ship. Thank you for your help.
Replies
5
Boosts
0
Views
319
Activity
2d
Local Push Connectivity entitlement request — no response after a month (again)
Hi all, I’m hoping someone from Apple can point me in the right direction. We submitted a request for the Local Push Connectivity entitlement about a month ago and haven’t heard anything back. No approval, no rejection, just silence. Unfortunately this is the second time; we tried back in 2023 and the same thing happened. After requesting the entitlement multiple times, we gave up on the project. A bit about what we’re building: we make a nurse call system for long-term care facilities. When a resident needs help, they press a wireless call device, our on-prem server raises an alert, and our iOS app would notify the caregivers on shift. The catch is that a most of these facilities put their life safety systems on air-gapped networks with no internet access at all, so APNs just isn’t an option. Even where there is internet, the system has to keep working if the connection goes down. We understand other vendors in our industry use Local Push Connectivity for this same scenario, so we’re hopeful this is just a matter of our request getting lost somewhere. As far as I can tell, this is exactly what Local Push Connectivity was made for. The WWDC20 session that introduced it (“Build local push connectivity for restricted networks”) actually names hospitals as one of the example environments where APNs may be unreachable, and talks about keeping notifications working in constrained networks “when it matters most to the users.” A resident calling for help is about as close to that as it gets. Our App ID already has the Critical Alerts entitlement, but there’s no “Network Extensions (additional values)” option under Additional Capabilities, so Xcode, understandably, complains that the profile doesn’t match the com.apple.developer.networking.networkextension value in our entitlements file. Is there a way to check on the status of a request like this, or a better place to follow up? I’m happy to share our team and app details privately. Any help would be really appreciated. We’ve been trying to get this going for a while now. Thanks!
Replies
0
Boosts
0
Views
41
Activity
2d
NEURLFilterManager: inconsistent per-domain blocking (some sites never block, one flaps) - confirmed PIR database is correct
Hello, Some context: DropBet is an app that helps people with gambling addiction stay off gambling sites and apps, using NEURLFilterManager/PIR to block a large list of known gambling domains system-wide. For a lot of our users this filter is a core part of their recovery > reliability really matters here. Since fixing an unrelated HPKE encryption bug (separate thread, now closed), the filter works end-to-end for the first time, but blocking is inconsistent per domain in a way I can't explain from my side: winamax.fr and the Winamax app: blocked reliably, every single time > clean "connection failed" style block, confirmed to load fine with protection off. unibet.fr and betclic.fr: never blocked, consistently > confirmed even after a full device reboot (rules out stale caching), and confirmed identical across both Safari and Chrome on iOS tested side-by-side (so not browser-specific). olybet.fr: flaps > the page reloads every 2-3 seconds with protection on, also identical across both browsers. Confirmed this only happens with the filter enabled (tested toggling protection off/on against the same page). unibet.fr auto-redirects into the native Unibet app via a Universal Link before the page loads, unlike betclic.fr which stays in the browser > different code path, possibly relevant. To rule out any issue on our end, I wrote a standalone test that queries our deployed PIR database shards directly - bypassing OHTTP, the app, and the device entirely - for all of the domains above plus their www. variants. All of them correctly return the blocked value, no exceptions, including unibet.fr and betclic.fr which never block on the actual device. So the database, the PIR protocol implementation, and the server are confirmed correct for every domain I've tested. This points at something in how NEURLFilterManager/ciphermld applies a correctly-received PIR verdict on-device - possibly related to query batching, an evaluation-key edge case, or Universal Link handling bypassing the filter before a web request is even made. Is there anything you can share about how the client applies results once decrypted, or known issues around any of this?
Replies
7
Boosts
0
Views
347
Activity
3d
macOS Transparent Proxy fails to start before login on Wi-Fi: "No network available"
I'm investigating an issue when deploying my VPN agent via Jamf during initial setup on clean macOS machines. The Network Extension process starts immediately after installation, but its Transparent Proxy provider starts only after a user logs in. I tried to reboot, and still observed the transparent proxy start fails. 10:00:52 === system boot 10:00:57.535 NESMTransparentProxySession[Primary Tunnel:Prisma Access Agent:68CAC11B-...]: Received a start command from PASrv[303] 10:00:57.538 NESMTransparentProxySession[...]: No network available <<<<<<<<< 10:00:57.541 NESMTransparentProxySession[...]: status changed to disconnected, last stop reason No network available <<<<<<<<< 10:01:21.541 NESMTransparentProxySession[...]: handleChangeEventForRankedInterfaces - stopped 0 status disconnected includeAllNetworks 0 10:01:26.373 NESMTransparentProxySession[...]: handleChangeEventForRankedInterfaces - stopped 0 status disconnected includeAllNetworks 0 10:06:01.293 NESMTransparentProxySession[...]: Received a start command from user_context[1547] 10:06:01.615 NESMTransparentProxySession[...]: status changed to connectedOur product requires pre-login functionality. Is this behavior expected, and how can we enable the Transparent Proxy provider before user login? I suspect a race between network availability and the attempt to start the Transparent Proxy. The issue does not occur with Ethernet, possibly because connectivity becomes available earlier than with Wi-Fi. Should we call startVPNTunnel again once connectivity becomes available? If so, what is the recommended way to detect this network transition?
Replies
5
Boosts
0
Views
156
Activity
4d
Egress addresses of the URL filter OHTTP relay
Hello, I am wondering about the egress addresses of the Oblivious HTTP relay that Apple runs for Network Extension URL filters. I am building an iOS app that uses NEURLFilterManager. The app builds a Bloom filter from the block lists a user picks. The system only asks our server about a site when that filter has a hit. Each lookup is encrypted with keyword PIR and carries a Privacy Pass token instead of anything tied to the user. It reaches our server through your relay, which strips the device's address. We run our own PIR server, Privacy Pass issuer and OHTTP gateway instead of a hosted service. The gateway is the one part of the setup that should only ever hear from your relay. Since every real lookup comes through the relay, I would really like to drop all other traffic to it at the firewall. Right now I can't do that. I also can't rate limit by address because every legitimate request shares the relay's addresses. As a result the gateway just has to stay open to everyone and rely on global limits, which is a far cry from what I'd want for a service that only one client is meant to reach. I have three questions about this. 1) Are the relay's egress address ranges published somewhere, or can they be shared with developers who have been onboarded? 2) If they are shared, how would we find out when they change and how much notice there is? 3) If the addresses can't be shared, is there another way to tell that a request came from the relay, such as a header or a client certificate? Please let me know if any of this is unclear or if there is a better place to ask. Thank you in advance, Alexander
Replies
1
Boosts
0
Views
92
Activity
5d
Kernel tag check fault after wake on Mac17 hardware, Google Chrome Helper, macOS 26.5–27.0.1 — related to r.172793638?
We're seeing a recurring kernel panic across six Mac17-family Macs (Mac17,2 / Mac17,4 / Mac17,8; T8142 and T6050) in a managed fleet. 34 panic logs collected 30 Sept – 9 Oct, all with the same signature: panic(cpu N caller ...): Kernel tag check fault (expected tagged address: 0x...028), esr 0x96000011 Faulting address is always at offset 0x28 of the object Same crash PC per build: kernel text exec base + 0x48da38 on 27.0/27.0.1, + 0x46a784 on 26.7, + 0x466ee0 on 26.5 Panicked task is "Google Chrome Helper" in all 34 cases, thread Chrome_ChildIOThread (Chrome's IPC/network I/O thread) Always 3–60 s after wake from sleep, usually a dark wake with the lid closed (overnight or after a 3–18 min sleep) No third-party kexts loaded; "Post-boot loaded kexts" lists only Apple drivers Affected builds: 26.5 (25F71), 26.7 (25G229), 27.0 (26A428), 27.0.1 (26A434). Not a regression from one update: a machine still on 26.5 panics identically. Only Mac17-family hardware is affected. Older Apple silicon Macs with the same software stack are fine. Present on every affected machine: Chrome PWAs (Gmail/Chat/Meet) running in the background, keeping Chrome's network service alive across sleep. On one machine Microsoft Edge is the primary browser and never panics; only Chrome Helper does. Microsoft Defender for Endpoint with its network filter (content filter) extension, MDM-managed. This looks related to thread 821372, where DTS described a kernel networking panic on M5-family hardware with a Network Extension present (r.172793638 / FB22401652), fixed in 26.5b2. We appear to be seeing a tag-check-fault variant on 26.5 release and later. Questions: Is this a known issue, and is it the same root cause as r.172793638? Is there a build where it's fixed? Any recommended mitigation beyond preventing sleep and quitting Chrome? Feedback Assistant: FB25128015 with panic logs and sysdiagnose. Happy to collect more data or test a build. Backtrace from a 27.0.1 panic (kernel text exec base 0xfffffe003aa54000), panic PC first: lr: 0xfffffe003aee1a38 lr: 0xfffffe003aeec580 lr: 0xfffffe003b12cf34 lr: 0xfffffe003b12dc50 lr: 0xfffffe003b07b1f8 lr: 0xfffffe003b07c758 lr: 0xfffffe003b237ff4 lr: 0xfffffe003ac787f0 lr: 0xfffffe003aa576ac lr: 0x000000018747a57c (userspace)
Replies
4
Boosts
0
Views
393
Activity
1w
URL Filter (Oblivious HTTP) configuration In Review for almost a week with no feedback
Hello, I submitted a Network Extension URL Filter (Oblivious HTTP) configuration in CloudKit Console almost a week ago. It still says In Review / Waiting for Approval and I haven't heard anything back since the submission notice. I wanted to check whether anything is missing on my end. Just in case, the configuration ID is 60802ed4-abfa-4b20-a64e-0d35230f478a. The app is an open-source ad and tracker blocker from the developers of an existing content blocker with a large user base. On iPhone and iPad its system-wide blocking runs entirely through the URL filter. Until the configuration is approved the iOS app can't block anything at all, and our testers are left with a Mac-only product in the meantime. Because lookups go through Apple's relay to our own PIR server, we never learn which sites anyone visits. All four endpoints have been live since the day I submitted and the TXT records are in place. Every database we build also includes the www.apple.com/url-filter-test validation entry. Is there anything else I need to provide, or a rough timeline I should expect? Thank you in advance!
Replies
1
Boosts
0
Views
86
Activity
1w
Apple thought it was impossible
Apple thought it was impossible. I made it possible. No server. Just an iPhone. LUMEN-JEV OS v3.0 — On-device OS using Pyto + iCloud Drive + Obsidian Architecture: Pyto runtime = server: Persistent loop, 922 tickers scanned, 502.6 ev/s, 12ms latency, sub-2ms cache iCloud Drive = bus: Writes JEV-BUS.md every 15s, shared working directory Obsidian = OS window manager: 00-JEV-DASHBOARD-main 48 nodes live Vault: 11/11 LOCKED, Factory Complete Anchor: 572b9df5e8e1f101365f8316ba7b34da4aa54ff89a2dcdb68ed20b126e8cabed Chain: REAL-EXCHANGE-PUBLIC-WS -> 8-MODULE-DEMO -> DEMO-TERMINATED -> NO EXECUTION Feed: real-public-readonly / public_ticker_only Modules (8-MODULE-v1px): WARREN-WATCHDOG, KEEPER-BOX, NODE-MODE-SYNC, VAULT-MULT, SCOUT-SCAN, SIREN-FILTER, FOCUS-SIGNAL-20K, ORACLE-PRICE-FEED, RELAY-ROUTER, AKE-ARBITRAGE, YIELD-BACK-4, MEDUSAs: 03 DYNAMIC OBSERVER-SPACE Context: Built in 33hrs over 11 nights, day job, single dad, solo dev, on iPhone only. Pyto working directory set to iCloud Drive. No Mac backend, no CloudKit function, no external server. Why this matters for iOS: Apple's sandbox says you can't run a persistent OS-like loop with live file sync on iOS. Pyto + iCloud Drive persistence proves you can. iPhone becomes the server. Proof: 48 NODES LIVE canvas PROOF CHAIN 11/11 FLAT VIEW VAULT LOCKED Jev.py V2.6 CLEAN BUS log (checked 50/100/150/200 -> 922/922 passes) Anchor hash matches zip JEV-OS-v3.0-572b9df5-40files.zip Requesting: Guidance on showcasing this as an iOS innovation + Ask Apple office hours. X proof pinned: @Activepython007
Replies
23
Boosts
1
Views
1.5k
Activity
1w
[macOS 27.2]: System-Wide DNS Resolution Failures After Sleep/Wake – mDNSResponder and NETransparentProxyProvider
Summary Recent macOS release observed intermittent system-wide DNS failures after sleep/wake on multiple devices running Global Secure Access (GSA) with NETransparentProxyProvider. Previously successful DNS lookups time out after wake across applications, including browsers and GSA. Two captures contain failure-time evidence of the mDNSResponder resolver queue predominantly waiting in close()/kernel. Other captures show similar DNS failures, but a shared root cause has not been established. Call chain, from thread entry to kernel wait: start_wqthread + 8 _pthread_wqthread + 292 _dispatch_workloop_worker_thread + 720 _dispatch_root_queue_drain_deferred_wlh + 284 _dispatch_lane_invoke + 392 _dispatch_lane_serial_drain + 332 _dispatch_source_invoke + 920 _dispatch_source_cancel_callout + 208 _dispatch_continuation_pop + 596 _dispatch_client_callout + 16 close + 8 fleh_synchronous + 72 sleh_synchronous (sleh.c:1610) unix_syscall (systemcalls.c:181) fp_close_and_unlock_with_action (kern_descrip.c:1874) fg_drop (kern_descrip.c:287) soclose (uipc_socket.c:1345) soclose_locked (uipc_socket.c:1138) cfil_sock_close_wait (content_filter.c:5743) soflow_db_apply (socket_flows.c; line unavailable) cfil_filters_udp_attached_per_flow (content_filter.c:6675) _sleep (kern_synch.c:213) lck_mtx_sleep_deadline (locks.c:1583) thread_block_reason (sched_prim.c:4060) thread_invoke (sched_prim.c:3443) Steps to reproduce The issue is intermittent. These steps describe the observed sequence, not a deterministic minimal reproduction. Run the GSA client using NETransparentProxyProvider. The affected configurations also have Microsoft Defender network extensions active. Verify that a system DNS lookup succeeds for a known hostname. Allow the Mac to sleep, including maintenance DarkWake cycles. Wake the Mac and repeat system DNS lookups or use applications requiring hostname resolution. Observe repeated approximately 30-second lookup timeouts, sometimes longer. Observed results November 5, 2026 - UTC+05:30 capture User-reported impact: Recurrence of the same sleep/wake connectivity symptoms. OS: macOS 27.2 (26B123). Incident window: 20:38-20:41 IST / 15:08-15:11 UTC. The same Chrome client/name A lookup succeeds earlier in 12.984 ms but times out after wake in 29,999.705 ms. The DNS service/server route changed, so this is not a controlled identical-network comparison. The resolver records 1,007 timeout callbacks during the incident window, across browsers, Teams, GSA, Defender and other processes. The failure-time spindump at 20:40:21.009-20:40:31.013 IST shows the original mDNSResponder PID 78534 with 715/715 resolver-queue samples under dispatch-source cancellation → close() → kernel → lck_mtx_sleep_deadline. Source: spindump.txt:75998-76093. This establishes a sampled wait, not a permanent deadlock or proof of the same root cause as November 2. The file descriptor and kernel/filter wait owner remain unidentified. A later matching Chrome lookup returns an address at 20:43:18.191816 IST on the same resolver PID, without a recorded resolver restart. This establishes some recovery, not that every application or tunnel recovered. November 2, 2026 - UTC-08:00 capture User-reported impact: No network upon wake. This eventually resolves without a reboot, after approximately 5 to 15 minutes of waiting. OS: macOS 27.2 (26B123). Incident window: 09:40-09:45 PST / 17:40-17:45 UTC. The same Edge A lookup succeeds before sleep in 19.346 ms but times out after wake in 29,999.826 ms. The resolver records 1,403 timeout callbacks during the incident window. The failure-time spindump shows 770/777 resolver-queue samples under dispatch-source cancellation → close() → kernel; 769 include lck_mtx_sleep_deadline. This establishes a dominant sampled wait, not a permanent deadlock. The file descriptor and kernel/filter wait owner remain unidentified. November 3, 2026 - UTC-08:00 capture (Mihai) Evidence: mihai_gsa_error.tar.gz The following example records a failure resolving GSA’s own control-plane endpoint, whose DNS lookup is not intercepted by the GSA tunnel. Times in these excerpts are local (UTC-08:00). GSA error [2026-11-03 10:13:58.463] Network connection to the Global Secure Access Edge could not be established StreamId: ControlStream Hostname: <tenant>.auth.client.globalsecureaccess.microsoft.com:443 Error: hostname lookup error: Address lookup failed os_error: nodename nor servname provided, or not known syscall: getaddrinfo mDNSResponder records 10:13:19.663661 [R125367->DupQ50990->Q38408] Duplicate question <private> dn:04wpeA (Addr) 10:13:58.468519 [R125367] DNSServiceQueryRecord STOP -- name hash: dn:04wpeA, duration: 39s 10:13:59.078847 [R125340] QueryRecordOpCallback: Question <private> dn:04wpeA (Addr) timing out, InterfaceID 0x0 Additional observations The original resolver records 11 timeout callbacks across GSA, Defender, Spotify and the system time service. A GSA A-query returns a positive result in 1.282 ms earlier (cache-compatible, not a cold upstream baseline), then the same client/name/qtype/flags times out after 41.865784 wall-clock seconds; that interval includes sleep. The DNS service/server set changes, so this is not a controlled same-network comparison. At 09:40 PST the Wi-Fi interface becomes inactive and loses its IPv4 address, followed by no-dns-service results. The first native lookup errors precede this link loss; it does not explain the entire case. Four GSA A/AAAA timeout transactions have matching late UDP replies. For one A-query, the reply is logged 25.095999 seconds after its matching send and 7.197434 seconds after the API timeout, during shutdown. 48 acceptable UDP response records appear within 1.467 ms. The location of the delay is not identified.
Replies
5
Boosts
1
Views
465
Activity
1w
Is it safe to strip with All Symbols for a content filter?
We would like to minimise the amount of code exposed in the product we have developed and an obvious option is to configure our build setting to use the All Symbols for Strip Style. Some investigation with an AI tool reports there are possible problems doing so. ContentFilter: non-global → all — risky. The NetworkExtension and XPC subsystems load your bundle and look up classes by name through the ObjC runtime, which uses __objc_classlist and friends (not the Mach-O symbol table), so in principle all works. In practice it occasionally breaks bundle loading in edge cases. If you try this, test specifically that systemextensionsctl activates the bundle and the filter actually traps packets. Is it safe to use All Symbols with a content filter or do we need to accept Non-Global symbols? Thanks, Dave
Replies
1
Boosts
0
Views
176
Activity
1w
NETransparentProxyManager: saving isOnDemandEnabled = true disables On-Demand on other proxies
On macOS, multiple NETransparentProxyManager (NEAppProxyProvider) configurations from different vendors can be enabled (isEnabled = true) and run simultaneously. However, we discovered that when one NETransparentProxyManager configuration calls saveToPreferencesWithCompletionHandler: with isOnDemandEnabled = true, /usr/libexec/nehelper (NEConfigurationManager) silently modifies any other enabled NETransparentProxyManager configuration in /Library/Preferences/com.apple.networkextension.plist and flips its OnDemandEnabled to false (and nesessionmanager logs Resetting VPN On Demand and marks the other session not-kernel-activated). Unlike NEVPNManager.isEnabled (which documents mutual exclusion for personal VPNs), NETransparentProxyManager is designed so multiple transparent proxies can coexist simultaneously, and neither NETransparentProxyManager.h nor NEVPNManager.h documents that isOnDemandEnabled is mutually exclusive across coexisting transparent proxy configurations. Root Cause Observed in NetworkExtension.framework When inspecting -[NEConfigurationManager saveConfigurationToDisk:updateSCPreferences:currentSignature:userUUID:notifyNow:isUpgrade:], when iterating over existing NEConfiguration objects of the same grade: If [[newConfig VPN] isEnabled] && [[otherConfig VPN] isEnabled] && [[newConfig VPN] isOnDemandEnabled] && [[otherConfig VPN] isOnDemandEnabled], NEConfigurationManager invokes: +[NEConfigurationManager disableConfiguration:otherConfig onDemandOnly:YES] disableConfiguration:onDemandOnly:YES calls [[otherConfig VPN] setOnDemandEnabled:NO], even when both configurations are NETransparentProxyManager (TunnelType == 2). Real-World Impact When two enterprise security products that both use NETransparentProxyManager are installed on the same Mac: Both configure isEnabled = true, onDemandRules = [NEOnDemandRuleConnect()], and isOnDemandEnabled = true. Whenever Proxy A saves its preferences, nehelper silently flips Proxy B's isOnDemandEnabled to false (and vice versa). After a reboot or network transition where Proxy B's initial startVPNTunnel fails while Wi-Fi is still associating and Proxy B relies on VPN On Demand (NEOnDemandRuleConnect) to start its NEAppProxyProvider once network connectivity comes up, macOS never starts Proxy B because its isOnDemandEnabled flag was cleared. Minimal Reproducer (Swift) Compile this minimal CLI inside an app bundle signed with the com.apple.developer.networking.networkextension = ["app-proxy-provider"] entitlement: import Foundation import NetworkExtension let args = CommandLine.arguments let name = args.count > 1 ? args[1] : "PocProxyA" let action = args.count > 2 ? args[2] : "status" DispatchQueue.main.async { NETransparentProxyManager.loadAllFromPreferences { managers, error in if let error = error { print("[\(name)] loadAllFromPreferences error: \(error)") exit(1) } let all = managers ?? [] let existing = all.first(where: { $0.localizedDescription == name }) if action == "status" { if let mgr = existing { print("[\(name)] -> isEnabled=\(mgr.isEnabled), isOnDemandEnabled=\(mgr.isOnDemandEnabled), rulesCount=\(mgr.onDemandRules?.count ?? 0)") } else { print("[\(name)] -> not found") } exit(0) } if action == "remove" { guard let mgr = existing else { exit(0) } mgr.removeFromPreferences { err in print("[\(name)] removed (err=\(String(describing: err)))") exit(0) } return } // action == "save" let mgr = existing ?? NETransparentProxyManager() let proto = (mgr.protocolConfiguration as? NETunnelProviderProtocol) ?? NETunnelProviderProtocol() proto.providerBundleIdentifier = "com.example.poc.extension" proto.serverAddress = "127.0.0.1" mgr.protocolConfiguration = proto mgr.localizedDescription = name mgr.isEnabled = true mgr.onDemandRules = [NEOnDemandRuleConnect()] mgr.isOnDemandEnabled = true print("[\(name)] Saving NETransparentProxyManager with isEnabled=true, isOnDemandEnabled=true...") mgr.saveToPreferences { err in if let err = err { print("[\(name)] saveToPreferences error: \(err)") exit(1) } print("[\(name)] saveToPreferences succeeded! isEnabled=\(mgr.isEnabled), isOnDemandEnabled=\(mgr.isOnDemandEnabled)") exit(0) } } } RunLoop.main.run()Reproduction Output Running save on PocProxyA, then save on PocProxyB, and checking status on both: === Step 1: PocProxyA saves isOnDemandEnabled = true === [PocProxyA] Saving NETransparentProxyManager with isEnabled=true, isOnDemandEnabled=true... [PocProxyA] saveToPreferences succeeded! isEnabled=true, isOnDemandEnabled=true [PocProxyA] -> isEnabled=true, isOnDemandEnabled=true, rulesCount=1 === Step 2: PocProxyB saves isOnDemandEnabled = true === [PocProxyB] Saving NETransparentProxyManager with isEnabled=true, isOnDemandEnabled=true... [PocProxyB] saveToPreferences succeeded! isEnabled=true, isOnDemandEnabled=true === Step 3: Check BOTH configurations after PocProxyB saved === [PocProxyA] -> isEnabled=true, isOnDemandEnabled=false, rulesCount=1 [PocProxyB] -> isEnabled=true, isOnDemandEnabled=true, rulesCount=1 === Step 4: PocProxyA saves isOnDemandEnabled = true again === [PocProxyA] Saving NETransparentProxyManager with isEnabled=true, isOnDemandEnabled=true... [PocProxyA] saveToPreferences succeeded! isEnabled=true, isOnDemandEnabled=true === Step 5: Check BOTH configurations after PocProxyA saved === [PocProxyA] -> isEnabled=true, isOnDemandEnabled=true, rulesCount=1 [PocProxyB] -> isEnabled=true, isOnDemandEnabled=false, rulesCount=1Questions Is disableConfiguration:onDemandOnly:YES stripping isOnDemandEnabled across coexisting NETransparentProxyManager (TunnelType = 2) configurations a bug in NEConfigurationManager, or an intentional limitation? Is there a recommended way for two coexisting NETransparentProxyManager configurations from different vendors to both use On-Demand without disabling each other's isOnDemandEnabled flag on saveToPreferences? Filed in Feedback Assistant as FB25463155.
Replies
3
Boosts
2
Views
376
Activity
1w
NEAppPushProvider times out connecting to a server on a routed subnet, while the containing app connects fine
We use Local Push Connectivity (NEAppPushProvider, matchSSIDs set) to receive notifications from an on-premise MQTT broker over WebSocket (wss://host:port/mqtt). iPhone on Wi-Fi, VLAN A (SSID matches matchSSIDs) Server on wired VLAN B, routing and firewall rules allow VLAN A → VLAN B on that port Result: the containing app, in the foreground, connects successfully to wss://<VLAN B IP>:<port>/mqtt. At the same moment, the app push provider (URLSessionWebSocketTask to the exact same URL) fails with "The request timed out". The provider starts normally (start() is called). When the server has an address on the Wi-Fi subnet (VLAN A), the provider connects without any issue. Is the app push provider restricted to servers on the local subnet of the matched Wi-Fi network? If so, is this documented anywhere? If not, what could prevent the provider from reaching a routed address that the app itself can reach? iOS version: 27.0.1 device : iPhone 13
Replies
1
Boosts
0
Views
103
Activity
1w
NEURLFilter configuration still "Queued" six weeks after submission
Our Network Extension URL Filter configurations in CloudKit Identity & Trust are not moving through onboarding. STG: configuration 6e17673d-dbaa-4b35-98c5-ff9e7a357577, bundle ca.itellyou.stg. Submitted 2026-09-25, status Approved, Onboarding Status "Queued". PRD: configuration 06fa2558-281b-4b55-b59f-402ecf359d18, bundle ca.itellyou. Submitted 2026-10-20, status In Review. Evidence (iPhone 17 Pro Max, 2026-11-06, same commit, same server, same user token): Development-signed build: the filter reaches "running" and blocks our test domains. Ad-hoc (distribution-signed) build: it cycles starting -> stopped with NEURLFilterManager.Error 9. The PIR server, Privacy Pass issuer and OHTTP gateway are live, and the test entry www.apple.com/url-filter-test is present. An Apple engineer confirmed on GitHub (apple/pir-service-example#185) that nothing is pending on our side. We also opened support case 30000007408474. Has anyone seen how long onboarding takes these days, or found a way to get it moving?
Replies
1
Boosts
0
Views
97
Activity
1w
Is a `flowClosed` report guaranteed when the owning process exits or crashes?
Our macOS NEFilterDataProvider sets shouldReport = true on flow verdicts and receives NEFilterReport.Event.flowClosed in handle(_:). In testing, we have observed close reports after the owning process exits or crashes, sometimes with a short delay. Is a flowClosed report guaranteed for every TCP and UDP flow in these cases? Does the behavior differ by protocol, and is there a documented upper bound on delivery time? We also set statisticsReportFrequency = .low. Does that setting affect only periodic .statistics reports, with no effect on flowClosed delivery?
Replies
2
Boosts
0
Views
134
Activity
2w
NEURLFilter PIR status check fails with CipherML errors (Code 1100/1800)
Hello ! following up from a related thread (account transfer issue, now resolved: https://developer.apple.com/forums/thread/843546?answerId=911820022), but this is a distinct, new issue so starting fresh as suggested. The extension launches fine, XPC handshake succeeds, and PIR registration completes successfully. But the subsequent NEPIRChecker status check fails every time (retried repeatedly over several minutes, same result each time): PIR status returned error <Error Domain=com.apple.CipherML Code=1100 "Unable to query status due to errors: Error details were logged and redacted." UserInfo={NSLocalizedDescription=Unable to query status due to errors: Error details were logged and redacted., NSUnderlyingError=0x677037cf0 {Error Domain=com.apple.CipherML Code=1800 "Error details were logged and redacted." UserInfo={NSLocalizedDescription=Error details were logged and redacted.}}}>Config being registered (confirmed from the logs): pirServerURL = https://pir.dropbet.app/ pirPrivacyPassIssuerURL = https://pir.dropbet.app/issueBoth endpoints are live and were already validated as part of the original NEURL Filter configuration approval. I've captured a sysdiagnose from the device right when this was happening, so I have that ready to attach to a bug report if useful — just let me know and I'll file one. Any idea what CipherML error 1100/1800 typically indicates?
Replies
8
Boosts
0
Views
515
Activity
2w
Cannot open Chrome UDP flows in Transparent Proxy Provider
We are implementing a Transparent Proxy for HTTPS (via TCP and QUIC). The following rules are set in startProxy: settings.includedNetworkRules = [ NENetworkRule(destinationNetwork: NWHostEndpoint(hostname: "0.0.0.0", port: "443"), prefix: 0, protocol: .TCP), NENetworkRule(destinationNetwork: NWHostEndpoint(hostname: "::", port: "443"), prefix: 0, protocol: .TCP), NENetworkRule(destinationNetwork: NWHostEndpoint(hostname: "0.0.0.0", port: "443"), prefix: 0, protocol: .UDP), NENetworkRule(destinationNetwork: NWHostEndpoint(hostname: "::", port: "443"), prefix: 0, protocol: .UDP) ] Handling TCP connections seems to work fine. But opening UDP flows from Chrome (or Brave) always fails with Error Domain=NEAppProxyFlowErrorDomain Code=2 "The peer closed the flow" (Doing the same for Firefox works!) BTW: We first create a remote UDP connection (using the Network framework) and when it is in the ready state, we use connection?.currentPath?.localEndpoint as the localEndpoint parameter in the open method of the flow. Is it a known issue that QUIC connections from Chrome cannot be handled by a Transparent Proxy Provider?
Replies
11
Boosts
0
Views
1.6k
Activity
2w
iPadOS: Intermittent disconnection after NEHotspotConfigurationManager automatic hotspot join
Background We are developing Share&Share, an offline device-to-device content transfer application. An iPad automatically joins a nearby Samsung Android LocalOnlyHotspot. The hotspot credentials are delivered through an authenticated, encrypted Bluetooth Low Energy (BLE) session. On iPadOS, our application decrypts the credentials natively in Swift and uses NEHotspotConfigurationManager.apply() to join the hotspot. We are investigating an intermittent connection problem during the initial automatic join. We have also submitted a private report through Apple Feedback Assistant: Feedback ID: FB25408475 Test environment Receiver: Physical iPad with a Home button. Sender/hotspot: Samsung SM-E055F. Android version: Android 16. Hotspot security: WPA2/WPA3 transition configuration. iOS implementation: Swift and NetworkExtension. joinOnce = false. The hidden property is not explicitly set. Observed behaviour Two independently instrumented automatic-join sessions showed a similar sequence: A1: The iPad associates with the Samsung hotspot and completes the WPA3 authentication exchange recorded by the Samsung. Approximately 3.35 seconds after the iPad-side join completion, iOS records a beacon-loss/roam declaration and the connection drops. A2: iOS attempts to reconnect, but the authentication exchange does not complete. A3: iOS automatically reconnects successfully, obtains a DHCP address and remains connected during the observation period. The same encrypted credentials work during both the initial and subsequent successful attempts. The apply() completion handler returns without an error. We understand that this does not establish that the device has joined successfully or that local IP connectivity is ready. What we have investigated The first join used cached beacon information approximately eight seconds old; the subsequent successful join used fresh scan information. AWDL activity was observed, but we have not established a causal relationship. Location Services scanning does not explain the first failure in our second instrumented session. The diagnostic label "Beacons Lost" is not being treated as proof of actual missed beacon frames. We have not independently established which WPA mode the iPad negotiated. We have not yet measured TCP readiness during the failed and successful attempts. We do not currently have a standalone minimal Xcode reproduction project. The behaviour was observed using our Flutter-based iOS probe and physical devices. Questions for developers familiar with NetworkExtension Has anyone observed intermittent first-join instability with NEHotspotConfigurationManager when joining a newly created Android LocalOnlyHotspot using WPA2/WPA3 transition mode? Are there documented limitations or known issues involving the first automatic join to a newly created hotspot profile? What is the recommended supported approach for determining that the intended Wi-Fi network is genuinely connected after apply() completes? What is the most reliable way to establish that local TCP communication is ready before beginning an application-level transfer? Are there additional public diagnostic facilities or measurements that could help distinguish Wi-Fi association instability from DHCP or subsequent network-path problems? We are looking for a documented, App Store-compatible approach. We are not assuming a particular root cause and do not want to rely on private APIs or undocumented system behaviour. Any practical experience or pointers to relevant Apple documentation would be appreciated.
Replies
1
Boosts
0
Views
175
Activity
2w