CWE-662

In FreeBSD 12.1-STABLE before r359565, 12.1-RELEASE before p7, 11.4-STABLE before r362975, 11.4-RELEASE before p1, and 11.3-RELEASE before p11, missing synchronization in the IPV6_2292PKTOPTIONS socket option set handler contained a race condition allowing a malicious application to modify memory after being freed, possibly resulting in code execution.

An issue was discovered in Squid 5.x before 5.0.3. Due to an Incorrect Synchronization, a Denial of Service can occur when processing objects in an SMP cache because of an Ipc::Mem::PageStack::pop ABA problem during access to the memory page/slot management list.

On Telus Actiontec WEB6000Q v1.1.02.22 devices, an attacker can login with root level access with the user "root" and password "admin" by using the enabled onboard UART headers.

An issue was discovered in Mitsubishi Electric Automation MELSEC-Q series Ethernet interface modules QJ71E71-100, all versions, QJ71E71-B5, all versions, and QJ71E71-B2, all versions. The affected Ethernet interface module is connected to a MELSEC-Q PLC, which may allow a remote attacker to connect to the PLC via Port 5002/TCP and cause a denial of service, requiring the PLC to be reset to resume operation. This is caused by an Unrestricted Externally Accessible Lock.

In FreeRADIUS 3.0.x before 3.0.20, the EAP-pwd module used a global OpenSSL BN_CTX instance to handle all handshakes. This mean multiple threads use the same BN_CTX instance concurrently, resulting in crashes when concurrent EAP-pwd handshakes are initiated. This can be abused by an adversary as a Denial-of-Service (DoS) attack.

Incorrect Synchronization in GitHub repository polonel/trudesk prior to 1.2.3.

An exploitable denial-of-service vulnerability exists in the XML_UploadFile Wi-Fi command of the NT9665X Chipset firmware, running on the Anker Roav A1 Dashcam, version RoavA1SWV1.9. A specially crafted packet can cause a semaphore deadlock, which prevents the device from receiving any physical or network inputs. An attacker can send a specially crafted packet to trigger this vulnerability.

A vulnerability in Cisco Webex Meetings and Cisco Webex Meetings Server could allow an unauthenticated, remote attacker to maintain bidirectional audio despite being expelled from an active Webex session. The vulnerability is due to a synchronization issue between meeting and media services on a vulnerable Webex site. An attacker could exploit this vulnerability by sending crafted requests to a vulnerable Cisco Webex Meetings or Cisco Webex Meetings Server site. A successful exploit could allow the attacker to maintain the audio connection of a Webex session despite being expelled.

rust-vmm vm-memory before 0.1.1 and 0.2.x before 0.2.1 allows attackers to cause a denial of service (loss of IP networking) because read_obj and write_obj do not properly access memory. This affects aarch64 (with musl or glibc) and x86_64 (with musl).

Missing synchronization vulnerability in GOT2000 series GT27 model communication driver versions 01.19.000 through 01.39.010, GT25 model communication driver versions 01.19.000 through 01.39.010 and GT23 model communication driver versions 01.19.000 through 01.39.010 and GT SoftGOT2000 versions 1.170C through 1.256S allows a remote unauthenticated attacker to cause DoS condition on the MODBUS/TCP slave communication function of the products by rapidly and repeatedly connecting and disconnecting to and from the MODBUS/TCP communication port on a target. Restart or reset is required to recover.

An issue was discovered in the spin crate before 0.5.2 for Rust, when RwLock is used. Because memory ordering is mishandled, two writers can acquire the lock at the same time, violating mutual exclusion.

An issue was discovered in the hashconsing crate before 1.1.0 for Rust. Because HConsed does not have bounds on its Send trait or Sync trait, memory corruption can occur.

The login verification can be bypassed by using the problem that the time is not synchronized after the router restarts. This affects Xiaomi router AX1800rom version < 1.0.336 and Xiaomi route RM1800 root version < 1.0.26.

An issue was discovered in the multiqueue2 crate before 0.1.7 for Rust. Because a non-Send type can be sent to a different thread, a data race can occur.

An issue was discovered in the may_queue crate through 2020-11-10 for Rust. Because Queue does not have bounds on its Send trait or Sync trait, memory corruption can occur.

An issue was discovered in the buttplug crate before 1.0.4 for Rust. ButtplugFutureStateShared does not properly consider (!Send|!Sync) objects, leading to a data race.

An issue was discovered in the atomic-option crate through 2020-10-31 for Rust. Because AtomicOption<T> implements Sync unconditionally, a data race can occur.

An issue was discovered in Input<R> in the eventio crate before 0.5.1 for Rust. Because a non-Send type can be sent to a different thread, a data race and memory corruption can occur.

In the standard library in Rust before 1.29.0, there is weak synchronization in the Arc::get_mut method. This synchronization issue can be lead to memory safety issues through race conditions.

A flaw was found in the QEMU NBD Server. This vulnerability allows a denial of service (DoS) attack via improper synchronization during socket closure when a client keeps a socket open as the server is taken offline.

A flaw was found in Linux Kernel because access to the global variable fg_console is not properly synchronized leading to a use after free in con_font_op.

Evmos is a scalable, high-throughput Proof-of-Stake EVM blockchain that is fully compatible and interoperable with Ethereum. Prior to 17.0.0, there is a way to mint arbitrary tokens due to the possibility to have two different states not in sync during the execution of a transaction. The exploit is based on the fact that to sync the Cosmos SDK state and the EVM one, we rely on the `stateDB.Commit()` method. When we call this method, we iterate though all the `dirtyStorage` and, **if and only if** it is different than the `originStorage`, we set the new state. Setting the new state means we update the Cosmos SDK KVStore. If a contract storage state that is the same before and after a transaction, but is changed during the transaction and can call an external contract after the change, it can be exploited to make the transaction similar to non-atomic. The vulnerability is **critical** since this could lead to drain of funds through creative SC interactions. The issue has been patched in versions >=V17.0.0.

Western Digital has identified a weakness in the UFS standard that could result in a security vulnerability. This vulnerability may exist in some systems where the Host boot ROM code implements the UFS Boot feature to boot from UFS compliant storage devices. The UFS Boot feature, as specified in the UFS standard, is provided by UFS devices to support platforms that need to download the system boot loader from external non-volatile storage locations. Several scenarios have been identified in which adversaries may disable the boot capability, or revert to an old boot loader code, if the host boot ROM code is improperly implemented. UFS Host Boot ROM implementers may be impacted by this vulnerability. UFS devices are only impacted when connected to a vulnerable UFS Host and are not independently impacted by this vulnerability. When present, the vulnerability is in the UFS Host implementation and is not a vulnerability in Western Digital UFS Devices. Western Digital has provided details of the vulnerability to the JEDEC standards body, multiple vendors of host processors, and software solutions providers.

Dell PowerScale OneFS contains an Unsynchronized Access to Shared Data in a Multithreaded Context in SMB CA handling. An authenticated user of SMB on a cluster with CA could potentially exploit this vulnerability, leading to a denial of service over SMB.

Jenkins Convertigo Mobile Platform Plugin 1.1 and earlier uses static fields to store job configuration information, allowing attackers with Item/Configure permission to capture passwords of the jobs that will be configured.

A sync issue was addressed with improved state validation. This issue is fixed in macOS Monterey 12.0.1. A user's messages may continue to sync after the user has signed out of iMessage.

An issue was discovered in the va-ts crate before 0.0.4 for Rust. Because Demuxer<T> omits a required T: Send bound, a data race and memory corruption can occur.

An improper synchronization logic in Samsung Email prior to version 6.1.41.0 can leak messages in certain mailbox in plain text when STARTTLS negotiation is failed.

Grafana is an open-source platform for monitoring and observability. Using public dashboards users can query multiple distinct data sources using mixed queries. However such query has a possibility of crashing a Grafana instance. The only feature that uses mixed queries at the moment is public dashboards, but it's also possible to cause this by calling the query API directly. This might enable malicious users to crash Grafana instances through that endpoint. Users may upgrade to version 9.4.12 and 9.5.3 to receive a fix.

Axios is a promise based HTTP client for the browser and Node.js. Starting in version 1.13.0 and prior to 1.13.2, Axios HTTP/2 session cleanup logic contains a state corruption bug that allows a malicious server to crash the client process through concurrent session closures. The vulnerability exists in the Http2Sessions.getSession() method in lib/adapters/http.js. The session cleanup logic contains a control flow error when removing sessions from the sessions array. This vulnerability is fixed in 1.13.2.

An issue was discovered in the Linux kernel before 5.4.17. drivers/spi/spi-dw.c allows attackers to cause a panic via concurrent calls to dw_spi_irq and dw_spi_transfer_one, aka CID-19b61392c5a8.

In the Linux kernel through 5.2.14 on the powerpc platform, a local user can read vector registers of other users' processes via an interrupt. To exploit the venerability, a local user starts a transaction (via the hardware transactional memory instruction tbegin) and then accesses vector registers. At some point, the vector registers will be corrupted with the values from a different local Linux process, because MSR_TM_ACTIVE is misused in arch/powerpc/kernel/process.c.

An issue was discovered in Xen through 4.12.x allowing x86 AMD HVM guest OS users to cause a denial of service or possibly gain privileges by triggering data-structure access during pagetable-height updates. When running on AMD systems with an IOMMU, Xen attempted to dynamically adapt the number of levels of pagetables (the pagetable height) in the IOMMU according to the guest's address space size. The code to select and update the height had several bugs. Notably, the update was done without taking a lock which is necessary for safe operation. A malicious guest administrator can cause Xen to access data structures while they are being modified, causing Xen to crash. Privilege escalation is thought to be very difficult but cannot be ruled out. Additionally, there is a potential memory leak of 4kb per guest boot, under memory pressure. Only Xen on AMD CPUs is vulnerable. Xen running on Intel CPUs is not vulnerable. ARM systems are not vulnerable. Only systems where guests are given direct access to physical devices are vulnerable. Systems which do not use PCI pass-through are not vulnerable. Only HVM guests can exploit the vulnerability. PV and PVH guests cannot. All versions of Xen with IOMMU support are vulnerable.

An issue was discovered in the conquer-once crate before 0.3.2 for Rust. Thread crossing can occur for a non-Send but Sync type, leading to memory corruption.

vyper is a Pythonic Smart Contract Language for the EVM. Multiple evaluation of a single expression is possible in the iterator target of a for loop. While the iterator expression cannot produce multiple writes, it can consume side effects produced in the loop body (e.g. read a storage variable updated in the loop body) and thus lead to unexpected program behavior. Specifically, reads in iterators which contain an ifexp (e.g. `for s: uint256 in ([read(), read()] if True else [])`) may interleave reads with writes in the loop body. Vyper for loops allow two kinds of iterator targets, namely the `range()` builtin and an iterable type, like SArray and DArray. During codegen, iterable lists are required to not produce any side-effects (in the following code, `range_scope` forces `iter_list` to be parsed in a constant context, which is checked against `is_constant`). However, this does not prevent the iterator from consuming side effects provided by the body of the loop. For SArrays on the other hand, `iter_list` is instantiated in the body of a `repeat` ir, so it can be evaluated several times. This issue is being addressed and is expected to be available in version 0.4.1. Users are advised to upgrade as soon as the patched release is available. There are no known workarounds for this vulnerability.

A DMA reentrancy issue was found in the Tulip device emulation in QEMU. When Tulip reads or writes to the rx/tx descriptor or copies the rx/tx frame, it doesn't check whether the destination address is its own MMIO address. This can cause the device to trigger MMIO handlers multiple times, possibly leading to a stack or heap overflow. A malicious guest could use this flaw to crash the QEMU process on the host, resulting in a denial of service condition.

OliveTin gives access to predefined shell commands from a web interface. Prior to version 3000.10.3, an unauthenticated denial-of-service vulnerability exists in OliveTin’s OAuth2 login flow. Concurrent requests to /oauth/login can trigger unsynchronized access to a shared registeredStates map, causing a Go runtime panic (fatal error: concurrent map writes) and process termination. This allows remote attackers to crash the service when OAuth2 is enabled. This issue has been patched in version 3000.10.3.

NVIDIA Windows GPU Display driver software for Windows (all versions) contains a vulnerability in the kernel mode layer (nvlddmkm.sys) handler for DxgkDdiEscape where the product does not properly synchronize shared data, such as static variables across threads, which can lead to undefined behavior and unpredictable data changes, which may lead to denial of service, escalation of privileges, or information disclosure.

An issue was discovered in the late-static crate before 0.4.0 for Rust. Because Sync is implemented for LateStatic with T: Send, a data race can occur.

An issue was discovered in Xen through 4.11.x allowing x86 PV guest OS users to cause a denial of service by leveraging a long-running operation that exists to support restartability of PTE updates.

An issue was discovered in the gfwx crate before 0.3.0 for Rust. Because ImageChunkMut does not have bounds on its Send trait or Sync trait, a data race and memory corruption can occur.

An issue was discovered in the im crate through 2020-11-09 for Rust. Because TreeFocus does not have bounds on its Send trait or Sync trait, a data race can occur.

A vulnerability, which was classified as critical, has been found in Linux Kernel. Affected by this issue is the function del_timer of the file drivers/isdn/mISDN/l1oip_core.c of the component Bluetooth. The manipulation leads to use after free. It is recommended to apply a patch to fix this issue. The identifier of this vulnerability is VDB-211088.

An issue was discovered in the thex crate through 2020-12-08 for Rust. Thex<T> allows cross-thread data races of non-Send types.

An issue was discovered in the rusb crate before 0.7.0 for Rust. Because of a lack of Send and Sync bounds, a data race and memory corruption can occur.

An issue was discovered in the aovec crate through 2020-12-10 for Rust. Because Aovec<T> does not have bounds on its Send trait or Sync trait, a data race and memory corruption can occur.

In the Linux kernel, the following vulnerability has been resolved: tracing: Restructure trace_clock_global() to never block It was reported that a fix to the ring buffer recursion detection would cause a hung machine when performing suspend / resume testing. The following backtrace was extracted from debugging that case: Call Trace: trace_clock_global+0x91/0xa0 __rb_reserve_next+0x237/0x460 ring_buffer_lock_reserve+0x12a/0x3f0 trace_buffer_lock_reserve+0x10/0x50 __trace_graph_return+0x1f/0x80 trace_graph_return+0xb7/0xf0 ? trace_clock_global+0x91/0xa0 ftrace_return_to_handler+0x8b/0xf0 ? pv_hash+0xa0/0xa0 return_to_handler+0x15/0x30 ? ftrace_graph_caller+0xa0/0xa0 ? trace_clock_global+0x91/0xa0 ? __rb_reserve_next+0x237/0x460 ? ring_buffer_lock_reserve+0x12a/0x3f0 ? trace_event_buffer_lock_reserve+0x3c/0x120 ? trace_event_buffer_reserve+0x6b/0xc0 ? trace_event_raw_event_device_pm_callback_start+0x125/0x2d0 ? dpm_run_callback+0x3b/0xc0 ? pm_ops_is_empty+0x50/0x50 ? platform_get_irq_byname_optional+0x90/0x90 ? trace_device_pm_callback_start+0x82/0xd0 ? dpm_run_callback+0x49/0xc0 With the following RIP: RIP: 0010:native_queued_spin_lock_slowpath+0x69/0x200 Since the fix to the recursion detection would allow a single recursion to happen while tracing, this lead to the trace_clock_global() taking a spin lock and then trying to take it again: ring_buffer_lock_reserve() { trace_clock_global() { arch_spin_lock() { queued_spin_lock_slowpath() { /* lock taken */ (something else gets traced by function graph tracer) ring_buffer_lock_reserve() { trace_clock_global() { arch_spin_lock() { queued_spin_lock_slowpath() { /* DEAD LOCK! */ Tracing should *never* block, as it can lead to strange lockups like the above. Restructure the trace_clock_global() code to instead of simply taking a lock to update the recorded "prev_time" simply use it, as two events happening on two different CPUs that calls this at the same time, really doesn't matter which one goes first. Use a trylock to grab the lock for updating the prev_time, and if it fails, simply try again the next time. If it failed to be taken, that means something else is already updating it. Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=212761

TensorFlow is an open source platform for machine learning. In affected versions the code behind `tf.function` API can be made to deadlock when two `tf.function` decorated Python functions are mutually recursive. This occurs due to using a non-reentrant `Lock` Python object. Loading any model which contains mutually recursive functions is vulnerable. An attacker can cause denial of service by causing users to load such models and calling a recursive `tf.function`, although this is not a frequent scenario. The fix will be included in TensorFlow 2.7.0. We will also cherrypick this commit on TensorFlow 2.6.1, TensorFlow 2.5.2, and TensorFlow 2.4.4, as these are also affected and still in supported range.

A bug in QEMU could cause a guest I/O operation otherwise addressed to an arbitrary disk offset to be targeted to offset 0 instead (potentially overwriting the VM's boot code). This could be used, for example, by L2 guests with a virtual disk (vdiskL2) stored on a virtual disk of an L1 (vdiskL1) hypervisor to read and/or write data to LBA 0 of vdiskL1, potentially gaining control of L1 at its next reboot.

FISCO BCOS 3.11.0 has an issue with synchronization of the transaction pool that can, for example, be observed when a malicious node (that has modified the codebase to allow a large min_seal_time value) joins a blockchain network.

An issue exists in SoftIron HyperCloud where drive caddy removal and reinsertion without a reboot may erroneously cause the system to recognize the caddy as new media and wipe all data on the drives due to a missing synchronization flaw, which impacts data availability and integrity. This issue only impacts SoftIron HyperCloud "density" storage nodes running HyperCloud software versions 1.0 to before 2.0.3.

A Missing Synchronization vulnerability in the Packet Forwarding Engine (PFE) of Juniper Networks Junos OS on ACX5448 and ACX710 allows an unauthenticated, adjacent attacker to cause a Denial-of-Service (DoS). If an interface flaps while the system gathers statistics on that interface, two processes simultaneously access a shared resource which leads to a PFE crash and restart. This issue affects Junos OS: * All versions before 20.4R3-S9, * 21.2 versions before 21.2R3-S5,  * 21.3 versions before 21.3R3-S5,  * 21.4 versions before 21.4R3-S4, * 22.1 versions before 22.1R3-S2, * 22.2 versions before 22.2R3-S2, * 22.3 versions before 22.3R2-S2, 22.3R3, * 22.4 versions before 22.4R2.

Improper synchronization in the firmware for some Intel(R) TDX may allow a privileged user to potentially enable escalation of privilege via local access.

In ccd, there is a possible use after free due to a race condition. This could lead to local escalation of privilege with System execution privileges needed. User interaction is not needed for exploitation. Patch ID: ALPS07341261; Issue ID: ALPS07341261.

In adsp, there is a possible double free due to a race condition. This could lead to local escalation of privilege with System execution privileges needed. User interaction is not needed for exploitation. Patch ID: ALPS07628532; Issue ID: ALPS07628532.

In gpu, there is a possible use after free due to a race condition. This could lead to local escalation of privilege with System execution privileges needed. User interaction is not needed for exploitation. Patch ID: ALPS07588678; Issue ID: ALPS07588678.

In ccd, there is a possible memory corruption due to a race condition. This could lead to local escalation of privilege with System execution privileges needed. User interaction is not needed for exploitation. Patch ID: ALPS07326547; Issue ID: ALPS07326547.

In ccu, there is a possible memory corruption due to a race condition. This could lead to local escalation of privilege with System execution privileges needed. User interaction is not needed for exploitation. Patch ID: ALPS07512839; Issue ID: ALPS07512839.

In display drm, there is a possible memory corruption due to a race condition. This could lead to local escalation of privilege with System execution privileges needed. User interaction is not needed for exploitation. Patch ID: ALPS07363469; Issue ID: ALPS07363469.

In vow, there is a possible use after free due to a race condition. This could lead to local escalation of privilege with System execution privileges needed. User interaction is not needed for exploitation. Patch ID: ALPS07494473; Issue ID: ALPS07494473.

In disp, there is a possible use after free due to a race condition. This could lead to local escalation of privilege with System execution privileges needed. User interaction is not needed for exploitation. Patch ID: ALPS06535964; Issue ID: ALPS06535964.

In vcu, there is a possible use after free due to a race condition. This could lead to local escalation of privilege with System execution privileges needed. User interaction is not needed for exploitation. Patch ID: ALPS07203476; Issue ID: ALPS07203476.

In vcu, there is a possible use after free due to a race condition. This could lead to local escalation of privilege with System execution privileges needed. User interaction is not needed for exploitation. Patch ID: ALPS07203410; Issue ID: ALPS07203410.

In vow, there is a possible information disclosure due to a race condition. This could lead to local information disclosure with System execution privileges needed. User interaction is not needed for exploitation. Patch ID: ALPS07494477; Issue ID: ALPS07494477.

## Summary This report covers the client-triggered DoQ forwarding path in: - `dnsproxy` `v0.81.2` (`adguard/dnsproxy:v0.81.2`) - `AdGuard Home` `v0.107.74` (`adguard/adguardhome:latest`, image version label `v0.107.74`) The issue was reproduced on `2026-04-25` with the products configured through their documented DoQ listener and plain UDP upstream surfaces. The scope is the internal backend UDP hop created when a DoQ query is forwarded to a `udp://` upstream. On that path, the backend DNS `ID` is not preserved as an independent source of entropy. For both products, the backend observer saw `dns_id=0` for every sampled client-triggered query on the tested path. Repeated reruns then showed the same `txid=0` behavior and the same positive source-port oracle on every sampled run. A separate quoted-port ICMP oracle distinguished the correct backend UDP source port from a wrong one with a stable, client-visible behavior change. Attached evidence: - `dnsproxy` oracle path on `v0.81.2`: [attachments/artifacts/g03/20260425T141500Z-g03-v0812/summary.txt](attachments/artifacts/g03/20260425T141500Z-g03-v0812/summary.txt) - `dnsproxy` `v0.81.2` repeatability: [attachments/artifacts/g03/repeatability-v0812.md](attachments/artifacts/g03/repeatability-v0812.md) - `dnsproxy` steering follow-up on `v0.81.2`: [attachments/artifacts/g04/20260425T141900Z-g04-v0812/summary.txt](attachments/artifacts/g04/20260425T141900Z-g04-v0812/summary.txt) - `AdGuard Home` oracle path: [attachments/artifacts/g05/20260425T113000Z-g05/summary.txt](attachments/artifacts/g05/20260425T113000Z-g05/summary.txt) ## Root Cause Analysis The observable behavior is consistent across both products: 1. A DoQ client query is accepted on the frontend listener. 2. The query is forwarded over a backend UDP leg. 3. On that backend leg, the forwarded DNS `ID` collapses to `0` on the client-triggered path instead of remaining a fresh per-query variable. 4. The backend UDP source port is still allocated per query. 5. When an ICMP error quotes the actual backend source port, the forwarding path flips behavior in a way that does not occur for a wrong quoted port. That combination removes `txid` from the backend tuple on the tested path and leaves the UDP source port as the main remaining variable. In practical terms, the backend hop stops behaving like a fresh `(txid, source-port)` pair per forwarded query and instead becomes a one-variable state exposure. For `dnsproxy`, the correct quoted port does more than produce a failure signal: it can push resolution away from the primary UDP upstream and into the fallback upstream. For `AdGuard Home`, the same condition produces a fast `SERVFAIL`. ## Reproduce Prerequisites: - Docker and Docker Compose - OpenSSL - build the lab helper image used by the attached harness and observer The attached reproducer bundle contains only the files needed for this report: - scripts: `attachments/scripts/` - helper image build files: `attachments/docker/unbound-doq-attacker/` - compose files: `attachments/docker-compose.g03.yml`, `attachments/docker-compose.g04.yml`, `attachments/docker-compose.g05.yml` - shipped evidence: `attachments/artifacts/...` Build the helper image first: 1. `cd attachments` 2. `docker build -t unbound-doq-attacker:latest -f docker/unbound-doq-attacker/Dockerfile docker/unbound-doq-attacker` To rerun `dnsproxy`: 1. `cd attachments` 2. `bash scripts/repro-g03-dnsproxy-oracle.sh` 3. Inspect `artifacts/g03/<RUN_ID>/summary.txt` 4. Inspect `artifacts/g03/<RUN_ID>/entropy-backend.jsonl`, `txid_correct-backend.jsonl`, and `port_correct-backend.jsonl` To rerun the `dnsproxy` fallback-steering case: 1. `cd attachments` 2. `bash scripts/repro-g04-dnsproxy-steering.sh` 3. Inspect `artifacts/g04/<RUN_ID>/summary.txt` 4. Inspect `steering_correct-main.jsonl` and `steering_correct-fallback.jsonl` To rerun `AdGuard Home`: 1. `cd attachments` 2. `bash scripts/repro-g05-adguardhome-oracle.sh` 3. Inspect `artifacts/g05/<RUN_ID>/summary.txt` 4. Inspect `entropy-backend.jsonl`, `txid_correct-backend.jsonl`, and `port_correct-backend.jsonl` The attached evidence includes fresh `dnsproxy v0.81.2` reruns, one official- profile `AdGuard Home` run, and the minimal reproducer bundle used by both. ## Impact For both products, the tested DoQ-to-UDP path is no longer a full `(txid, source-port)` search surface: - `dnsproxy`: four of four sampled runs showed `txid=0` on the backend hop and a positive source-port oracle on `v0.81.2`. The remaining unknown is `port_only`. Median wrong/correct port latency was `327.99 ms / 40.93 ms`. - `AdGuard Home`: four of four sampled runs showed `txid=0` on the backend hop and a positive source-port oracle. The aggregate again classifies the remaining unknown as `port_only`. Median wrong/correct port latency was `319.14 ms / 37.02 ms`. Product-specific effects: - `dnsproxy`: a correct port guess produced an empty client-visible answer on the base oracle path, and in the fallback profile it steered all eight tested queries away from the main upstream and into the fallback upstream. - `AdGuard Home`: a correct port guess produced fast `SERVFAIL` and an extra backend query. This is the security-relevant point. On the tested official profiles, the backend hop no longer forces an off-path attacker to deal with two fresh random fields per forwarded DNS race. The DNS ID is already known: it is deterministically `0` on the client-triggered DoQ-to-UDP path. The only remaining backend tuple variable is the UDP source port, and the attached evidence shows a repeatable oracle for that remaining variable. That places the path in the same threat-model class as oracle-assisted DNS forgery work such as SAD DNS and TUdoor: the attack first uses an oracle to learn or validate the tuple state that protects an off-path response race, and only then attempts the forged response. This report stops short of a forgery demo, but the evidence already shows the crucial precondition on the tested backend hop: the tuple is not high-entropy anymore. It has been reduced from `(txid, source-port)` to `source-port` only. --- **Attachments** [attachments.zip](https://github.com/user-attachments/files/27227054/attachments.zip)