Trezor firmware

This commit is a routine developer tooling change. It turns on stricter Python type checking (Pyright) for the emulator script and some test helper scripts, and makes small code adjustments so those files pass the new checks. There is no change to the security-critical Trezor firmware itself, wallet operations, or cryptographic handling.

This commit changes how the Trezor firmware build and test tooling handles the Tropic secure-element emulator. Previously, Tropic support was disabled by default in Unix emulator builds and had to be explicitly enabled; now it is enabled by default and the test harness starts the Tropic model automatically. It also refactors port allocation so each emulator worker reserves a wider block of ports. The changes are confined to build scripts, CI workflows, and test/development tooling. There is no direct fix for a runtime vulnerability in shipped firmware, but enabling a hardware-security emulator by default could theoretically expose test-only code paths or secrets if the emulator is misused outside of testing.

This commit is a routine code reorganization: it moves existing test-only helper code from the project's internal test directory into the public `trezorlib` Python package as a new `testing` module. The moved code includes utilities for handling emulated devices during tests, translation blob helpers, and Bitcoin compact-size encoding. There is no change to the firmware that runs on Trezor devices, no change to production cryptography, and no security fix or vulnerability introduced.

This commit adds support for a new Trezor hardware model, the T3T2. It is a product enablement change that introduces board definitions, memory layouts, bootloader hashes, vendor headers, and Python tooling entries for the new device. There is no indication of a security vulnerability or a fix for one.

This commit adds support for a new Trezor development board called D003 (based on an STM32U5 Nucleo evaluation kit). It is purely an enablement change for internal hardware bring-up: new configuration files, memory layout, bootloader hash, vendor headers, and build-system entries. There is no indication it fixes a security bug or introduces a vulnerability in shipping products.

This commit adds support for a new microcontroller chip (the STM32U5A5) to the Trezor firmware build system. It is a hardware-enablement change: it wires up the right compiler flags, linker scripts, memory layouts, and trustzone settings so the existing code can compile and run on the new chip. There is no user-facing behavior change, no bug fix, and no security patch in this diff.

This commit hardens how Trezor compares secret values (passwords, PINs, cryptographic tags, and checksums) so that an attacker cannot learn information by measuring how long the comparison takes. It also adds a fault-injection check: if a glitch skips the comparison loop, the device is supposed to detect that and halt. The change is defensive and reduces the risk of side-channel and fault attacks, but it does not by itself fix a known, directly exploitable bug.

This commit is a small test-only UI adjustment for the Trezor hardware wallet's Caesar layout. It changes when a menu icon is reported as visible in automated UI snapshots, limiting it to the last page of a multi-page flow. The code change only affects test fixtures and snapshot hashes, not user-facing security behavior. There is no indication this fixes a security vulnerability.

This is a minor user-interface polish change for the Trezor hardware wallet's Caesar layout. It removes an unnecessary blank line from transaction summary screens when either the amount label or the amount value is empty. There is no security relevance: it does not change what the device signs, how it signs, or any cryptographic behavior. The large diff is mostly updated test screenshot hashes reflecting the slightly different screen layout.

This commit changes how Trezor's Ethereum 'clear signing' helper data is loaded in memory. Previously, all definitions were stored in a permanent module-level list that MicroPython's garbage collector could not reclaim. The patch wraps them in a generator function so they are created on demand and can be freed afterward. The stated goal is to avoid memory pressure, not to fix a security bug. There is no direct evidence of an exploitable vulnerability in the diff itself.

This commit replaces the Python implementation of Trezor's THP (Trezor Host Protocol) with a Rust-based one. It is a large refactoring that moves channel state management, encryption, packet handling, and handshake logic from Python into a Rust module exposed as `trezorthp`. The change also adjusts how credentials are validated, how sessions are cached, and how the event loop handles reads, writes, and retransmissions. There is no explicit security bug in the diff, but the scope of the rewrite and the removal of several safety checks (for example around buffer allocation and unexpected-message handling) create a non-trivial risk of introducing memory-management, concurrency, or protocol-edge-case bugs. The vendor does not describe this as a security fix.

This commit is a code cleanup in the Trezor firmware's Rust cryptography code. It replaces hand-written memory-clearing code with a standard library feature called ZeroizeOnDrop. The goal is the same: erase sensitive cryptographic data from memory when objects are destroyed. There is no evidence this commit fixes an active security bug; it is a defensive hardening and refactoring change.

This commit fixes a bug in how the Trezor hardware wallet calculates the width of on-screen text. Previously, very long text could cause an internal counter to overflow from a large positive number to a negative number, which might make text appear incorrectly sized or positioned. The fix uses saturating arithmetic so the counter stops at the maximum safe value instead of wrapping around.

This commit is a straightforward internal code refactor in the Trezor hardware wallet's Rust transport-handshake protocol (THP) library. It replaces a simple pairing-state field with a richer 'phase' enum that tracks whether the channel is still in the pairing/credential setup stage or has moved to encrypted application messaging. The change adds helper methods so the application can explicitly mark pairing as finished. There is no direct security fix here; it is a structural improvement that makes state tracking clearer and less error-prone for callers.

This commit adds a new Rust implementation of the Trezor Host Protocol (THP), which is the secure communication layer between a Trezor hardware wallet and a host computer/phone. It introduces encryption, channel management, pairing, and credential verification. Because this is a large new cryptographic and network-facing subsystem, any bugs here could affect device security, but the commit itself is a feature addition rather than a documented fix for a known vulnerability.

This commit refactors the cryptographic glue code that connects Trezor's embedded firmware to a Rust-based Noise protocol implementation used for secure device communication. The most notable security-relevant change is the addition of a constant-time comparison function for checking AES-GCM authentication tags, which helps prevent timing-based attacks that could leak information about the tag. The commit also splits AES-GCM into separate encrypt and decrypt types, adds a non-pinned SHA-256 wrapper, and introduces wrappers for Curve25519 and AES-GCM to satisfy the Noise protocol backend interface. There is no explicit vendor statement that this fixes a security vulnerability, and no independent researcher is credited.

This commit fixes a spelling mistake in the legal license text at the top of six source files. The typo 'HMAC_SHALL' is corrected to 'SHALL'. It does not change any program logic, security behavior, or compiled code.

This commit only adds new automated tests for an existing cryptographic handshake protocol called noise_xxpsk3. It does not change the actual security code, fix a bug, or introduce a vulnerability. The tests verify that the protocol works correctly, that tampered messages are rejected, and that keys are wiped after use.

This commit is a simple rename of the cryptographic module from 'noise' to 'noise_kk1' across file names, function names, constants, and build scripts. No security behavior changes are visible in the diff. It is a code-maintenance refactor, not a security fix.

This commit fixes several minor issues in a cryptographic module used for secure device communication. The changes include preventing undefined behavior when copying from NULL pointers, zeroing out sensitive key material after use, and properly returning errors when encryption fails. These are defensive hardening fixes rather than a clear, exploitable vulnerability.

This commit adds a brand-new cryptographic module implementing the Noise_XXpsk3_25519_AESGCM_SHA256 protocol for secure encrypted communication. It is a feature addition with no changelog entry and no accompanying security advisory. The code itself is a clean implementation of a well-known cryptographic handshake, but because it is new code handling private keys, pre-shared keys, and AES-GCM encryption, it is security-relevant by nature. There is no direct evidence in the commit or supplied references that this fixes a known vulnerability or introduces a bug.

This commit fixes a build problem in Trezor's firmware. The code used a very common filename, version.h, which could accidentally pull in the wrong file from another software package during compilation. The fix creates a symlink with a more specific name, firmware_version.h, and uses that instead. There is no direct evidence this is a security vulnerability; it appears to be a build reliability fix.

This is a minor build cleanup. A developer added a compile-time branch so that when neither the Optiga nor Tropic security chips are used, the 'strong' random-number wrapper simply calls the normal random-number function and returns. The only reason for the change is to silence a compiler warning about an unused variable. There is no functional security change.

This commit changes how Trezor devices display Ethereum transactions when someone sends ERC-20 token approval or transfer calls that also include native ETH. Previously, the device might not clearly show the attached ETH, so a user could unknowingly approve sending tokens plus real ETH. The patch makes the device surface that native ETH amount on the confirmation screen. It is a defensive fix that improves transparency rather than introducing a vulnerability.

This commit is a routine feature update for Trezor's Ethereum 'clear signing' feature. It adds human-readable names for more DeFi and staking providers (like Lido, Morpho, Kiln) and makes the address lookup aware of which blockchain network (chain ID) an address belongs to. There is no security vulnerability or bug fix here.