idf.py mcp-server printed startup, shutdown, and error messages on
stdout. On the stdio MCP transport, stdout is the JSON-RPC channel, so
those non-JSON lines can confuse or break strict clients. Route them to
stderr like the other diagnostics in mcp_ext.py.
Co-authored-by: Cursor <cursoragent@cursor.com>
idf.py mcp-server tool handlers were spawning idf.py without redirecting
stdin, so the child inherited the long-lived MCP JSON-RPC transport and
could hang indefinitely on tools/call. Pass stdin=subprocess.DEVNULL on
every spawn of idf.py.
Closes https://github.com/espressif/esp-idf/issues/18961
Extend the CTR test data length to 6433 bytes so the trailing partial
block is exercised with external RAM buffers (which stalls the ESP32-S2
Crypto DMA on an unfixed driver), and add AES-GCM PSRAM tests verified
against internal RAM references.
The ESP32-S2 Crypto DMA in-channel stalls silently when a receive
descriptor list transitions from external to internal RAM. The AES
driver hits this when a PSRAM-output operation has a trailing partial
block, as the internal stream descriptor is linked after the external
RAM data descriptors.
- esp_aes_process_dma(): process the block-aligned part and the partial
block as two separate DMA operations, keeping each descriptor list
uniform
- crypto_dma_ll_reset(): also reset the in-channel (per the TRM receive
reset sequence), otherwise stale state from a preceding external-RAM
operation corrupts the next operation's output
The GCM DMA path is unaffected; it never operates on PSRAM buffers.
Bool symbols in nimble, BTDM, and two examples used invalid default
literals. The parser already falls back to 'n'; set that explicitly.
Co-authored-by: Cursor <cursoragent@cursor.com>
esp_crt_check_bundle() read the 4-byte certificate header (name_len,
key_len) via esp_crt_get_len() after only checking that the cert's
start offset lies inside the bundle, so a crafted bundle whose first
or last certificate starts within the final 3 bytes caused a transient
out-of-bounds read of up to 3 bytes before the extent check rejected
it. Require the whole header to lie inside the bundle before reading
it.
The Key Manager hardware peripheral in its current form needs further
design changes before it can be offered as a production feature.
Until a revised peripheral design is available, withdraw ESP-IDF
support for it on all Key Manager capable targets.
The cxx component re-injects its own archive late on the link line so its
stack-unwind wrappers take precedence over libgcc's (when C++ exceptions
are disabled). It does this through an INTERFACE helper, libgcc_cxx, that
carries `$<TARGET_FILE:${cxx}>`. That helper was linked back into the cxx
target with PUBLIC, which places it in the cxx archive's own
LINK_LIBRARIES, so the cxx target ends up transitively referencing its
own output file.
When CMake computes the cxx target's sources it walks the target's own
link implementation, encounters `$<TARGET_FILE:${cxx}>`, and to resolve the
file it needs the target's link language, which in turn needs its
sources. This loop makes generation fail with:
The SOURCES of "..." use a generator expression that depends on the
SOURCES themselves.
CMake only hits this when it reaches the cxx target standalone before the
target has been pulled into a link by a consumer, and whether that
happens depends on the project layout and target evaluation order. That
is why it surfaces only in some projects, while the examples build fine.
Link libgcc_cxx as INTERFACE instead of PUBLIC. A static archive is not
linked, so its private LINK_LIBRARIES are inert; consumers still receive
libgcc_cxx through INTERFACE_LINK_LIBRARIES, so the intended link order is
preserved. Removing the self-reference makes generation independent of
evaluation order.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
Wire esp_blockdev, comp_a, and comp_b into the test app dependency graph
so ioctl def files are registered and the POST_BUILD overlap checker runs.
Co-authored-by: Cursor <cursoragent@cursor.com>
(cherry picked from commit ddc9c5fc40)
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
FastMCP was renamed to MCPServer in mcp 2.0, which made idf.py mcp-server
report a false "MCP dependencies not available" error. Prefer MCPServer
and keep a FastMCP fallback for mcp 1.x.
Closes https://github.com/espressif/esp-idf/issues/18904
Co-authored-by: Cursor <cursoragent@cursor.com>
Add an OPTIONAL flag to idf_component_include so callers can ask
"include this component if it exists" without aborting the build
when the component is unknown.
Without OPTIONAL, the function behaves exactly as before -- it
calls __get_component_interface_or_die and the build aborts on a
miss. With OPTIONAL, the function performs a non-fatal interface
lookup and returns silently if the component is not known. When
combined with INTERFACE <variable>, the variable is set to the
empty string on miss and to the component's interface target on
hit, so callers can write:
idf_component_include(button OPTIONAL INTERFACE button_iface)
if(button_iface)
target_sources(${COMPONENT_TARGET} PRIVATE button_glue.c)
target_link_libraries(${COMPONENT_TARGET} PRIVATE ${button_iface})
endif()
This gives consumers a public, non-fatal way to wire up integrations
with components that may or may not be in the build (managed
dependencies pulled in only by some board configurations, optional
feature glue, etc.) without reaching into the private
__get_component_interface helper or doing dual-namespace
COMPONENTS_DISCOVERED checks against both <name> and <ns>__<name>
forms.
Backward compatible: existing call sites do not pass OPTIONAL and
continue to fail loudly on miss.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
Shu Chen
Mahavir Jain
Roland Dobai
Ivan Grokhotkov
Marek FialaandCursor
Martin Vychodil
harshal.patil
Jan BeranandCursor
Jiang Jiang Jian
Shreyas Sheth
wuzhenghui
Ashish Sharma
Sajia
Xu Si Yu
morris
Armando (Dou Yiwen)
Frantisek Hrbata
Laukik Hase