In sporadic cases it is possible that a SPI exchange returns stale RX data from the DCACHE.
This occurs when:
- DCACHE is enabled
- DMA is used
The impact of this can be hard to debug and vanishes when the timing even changes minimally.
This is caused by the DCACHE being invalidated before the actual DMA transaction starts which violates
the recommendations from AN4839 and also does not match the implementation of other drivers on the STM32H7/STM32F7.
Fixed by invalidating the DCACHE right before the actual read, which matches the implementation
of the STM32F7 SPI driver.
Signed-off-by: alexcekay <alexander@auterion.com>
The PR #18654 removed the Kconfig option `RMT_LOOP_TEST_MODE` used
by Espressif's RMT peripheral. This commit reintroduces it as a
lower-half driver interface used to enable internal loopback tests.
Signed-off-by: Tiago Medicci Serrano <tiago.medicci@espressif.com>
Enable USART3 support for STM32F412 and use it as the default NUCLEO-F412ZG nsh console path.
The ST-LINK virtual COM port on this board is wired to USART3 on PD8/PD9. Restore USART2 to its header-pin routing, switch the nsh defconfig to USART3, and update the board documentation to match the shipped solder-bridge configuration described in issue #17722.
Signed-off-by: aviralgarg05 <gargaviral99@gmail.com>
- Moved the creation of the `nxtmpdir` folder for third-party packages to the root `CMakeLists.txt` file.
cmake/nuttx_3rdparty.cmake
- Add the nuttx_remove_nxtmpdir function to remove the third-party cache directory under nuttx/../nxtmpdir
Signed-off-by: simbit18 <simbit18@gmail.com>
- fix a "a label can only be part of a statement and
a declaration is not a statement" compile error.
- toolchain xtensa-esp32s3-elf-gcc
(crosstool-NG esp-2021r2) 8.4.0
Signed-off-by: Shunchao Hu <ankohuu@gmail.com>
copy logic from arm64 to fix SMP bugs in the signal trampoline
caused signals to be lost or mis-delivered in ostest signest_test
on qemu-intel64/jumbo.
Also update comments so this logic is easier to follow
Signed-off-by: raiden00pl <raiden00@railab.me>
During startup we should avoid using _info since the system is not
fully initialized. Default to use ets_printf which is safe for
this scenario.
Signed-off-by: Filipe Cavalcanti <filipe.cavalcanti@espressif.com>
Fix lowercase -wl to correct -Wl for GCC linker pass-through
option in LTO full configuration. The lowercase form is not
recognized by the linker, causing --print-memory-usage to be
silently ignored.
Signed-off-by: v-yanxingyu <v-yanxingyu@xiaomi.com>
Increase image slot size from ~1MB to ~1.4MB to allow more features by default.
Solves issues when enabling debug features would extrapolate the slot size.
Signed-off-by: Filipe Cavalcanti <filipe.cavalcanti@espressif.com>
Fixed a few bugs in the initial stm32_usbdrdhost.c implementation
when using a USB hub. Fixed fault when freeing control endpoints.
Fixed crash when disconnecting devices from hubs due ot use after free.
Signed-off-by: daniellizewski <daniellizewski@geotab.com>
fix compilation error for CONFIG_SERIAL_TERMIOS=y
chip/stm32_serial_v2.c:923:35: error: 'struct up_dev_s' has no member named 'rxftcfg'
923 | regval |= USART_CR3_RXFTCFG(priv->rxftcfg);
Signed-off-by: raiden00pl <raiden00@railab.me>
Added stm32_usbdrdhost.c which adds USB FS host mode support using the
embedded PHY. Added Kconfig to select Device or Host.
Example config nucleo-h563zi:usbmsc added to test functionality
Signed-off-by: daniellizewski <daniellizewski@geotab.com>
This change sets three new compilation flags for avoiding cache miss
when using optimized builds and cache is manipulated.
Signed-off-by: Filipe Cavalcanti <filipe.cavalcanti@espressif.com>
This changes enables the use of the preprocessor function (nuttx_generate_preprocess_target) so it can properly
process the linker scripts, instead of using the original ones (must process some `#ifdefs`).
Signed-off-by: Filipe Cavalcanti <filipe.cavalcanti@espressif.com>
When the macro SDADC_HAVE_TIMER is enabled, a compilation
error will occur due to the missing semicolon after the variable.
Signed-off-by: zhanxiaoqi <zhanxiaoqi@bytedance.com>
macOS does not provide full POSIX timer support for the host-side
simulator build, so references to timer_t and timer_create can fail when
building sim:fb on Apple hosts. Replace the POSIX timer path with a
Grand Central Dispatch timer on macOS while keeping the existing POSIX
implementation for other hosts.
The dispatch timer raises SIGALRM from its event handler so the host
side keeps the same timer-driven behavior expected by the simulator.
This preserves the existing timing model and fixes the macOS host build
failure for sim:fb.
Signed-off-by: Peter Bee <bijunda@bytedance.com>
macOS does not ship X11 headers and libraries in the default system
search paths, so sim:fb fails to build unless the XQuartz installation
prefix is detected explicitly. Probe /opt/X11 first and fall back to
/usr/X11 so the simulator can find the required X11 headers and libs on
common macOS setups.
This fixes the sim:fb build failure on macOS hosts that rely on XQuartz
for X11 development files.
Signed-off-by: Peter Bee <bijunda@bytedance.com>
When ESP32's flat build mode is selected with the external PSRAM
on USER_HEAP mode, the ROM strdup internally calls malloc which may
be mapped to a different heap than lib_free, causing cross-heap
allocation issues and system crashes.
This commit adds a new Kconfig option ESP32_DONT_USE_ROM_LIBC to
allow using NuttX libc instead of ROM newlib functions. When
USER_HEAP mode is selected, this option is automatically enabled.
Signed-off-by: Tiago Medicci Serrano <tiago.medicci@espressif.com>
In ESP32-S3 flat build with USER_HEAP mode, ROM strdup internally calls
malloc which may be mapped to a different heap than lib_free, causing
cross-heap allocation issues and system crashes.
This commit adds a new Kconfig option ESP32S3_DONT_USE_ROM_LIBC to
allow using NuttX libc instead of ROM newlib functions. When USER_HEAP
mode is selected, this option is automatically enabled.
Fixes: 60ca804b56 ("esp32s3: Fix bug related to the PSRAM-allocated task stack")
Signed-off-by: 461911662 <461911662@qq.com>
Reviewed-by: Tiago Medicci Serrano <tiago.medicci@espressif.com>
Reviewed-by: Almir Okato <almir.okato@espressif.com>
Reviewed-by: Eren Terzioglu <eren.terzioglu@espressif.com>
Wi-Fi used a locally allocated vector_desc, so those CPU interrupt lines
were never recorded in the HAL non_iram_int_mask.
During SPI flash, esp_intr_noniram_disable() therefore did not mask them,
and Wi-Fi ISRs could still run with the cache off.
Now, Wi-Fi IRQs are treated as non-IRAM and get masked while flash/cache is suspended.
Signed-off-by: Filipe Cavalcanti <filipe.cavalcanti@espressif.com>
Wi-Fi used a locally allocated vector_desc, so those CPU interrupt lines
were never recorded in the HAL non_iram_int_mask.
During SPI flash, esp_intr_noniram_disable() therefore did not mask them,
and Wi-Fi ISRs could still run with the cache off.
Now, Wi-Fi IRQs are treated as non-IRAM and get masked while flash/cache is suspended.
Signed-off-by: Filipe Cavalcanti <filipe.cavalcanti@espressif.com>
Replace direct putreg32/getreg32 register accesses in esp32s3_cam.c
with cam_ll_* inline functions from Espressif's esp_hal_cam component.
This reduces maintenance burden by using the vendor-provided HAL
abstraction instead of raw register manipulation.
Changes:
- Add lcd_cam_dev_t *hw pointer to driver struct
- Use cam_ll_start/stop/reset/fifo_reset for CAM control
- Use cam_ll_get_interrupt_status/clear_interrupt_status for ISR
- Use cam_ll_set_recv_data_bytelen for DMA buffer length
- Use cam_ll_select_clk_src/set_group_clock_coeff for clock config
- Use cam_ll_enable_vsync_filter/set_vsync_filter_thres
- Use cam_ll_enable_vsync_generate_eof/enable_rgb_yuv_convert
- Use struct access for interrupt enable (cam_ll_enable_interrupt
requires __DECLARE_RCC_ATOMIC_ENV not available in NuttX)
- Add esp_hal_cam include paths to hal.mk
Signed-off-by: wangjianyu3 <wangjianyu3@xiaomi.com>
This commit introduces the necessary changes to enable low power
mode for ESP32-P4. Please check the corresponding documentation.
Signed-off-by: Tiago Medicci Serrano <tiago.medicci@espressif.com>
This adds support for the PBKDF2 algorithm in the ESP32 crypto
driver. Here, it will take advantage of the hardware-accelerated
implementation of SHA.
Signed-off-by: Vlad Pruteanu <pruteanuvlad1611@yahoo.com>
Add an RP23xx hardware crypto driver for /dev/crypto with SHA-256
support backed by the RP2350 SHA accelerator.
Wire the new driver into the RP23xx ARM build and register
CRYPTO_SHA2_256 through cryptodev.
Also fix the RP23xx SHA256 CSR BSWAP bit definition in both the ARM and
RISC-V RP23xx headers. The previous value used bit 18, but the RP2350
hardware uses bit 12. Without this fix, the accelerator hashed input
words with the wrong byte order and produced incorrect digests.
Signed-off-by: karaketir16 <osmankaraketir@gmail.com>
Implement imgdata_ops alloc/free callbacks so the V4L2 framework
can use MMAP mode to allocate frame buffers with proper GDMA
alignment. This lets applications avoid hardcoding platform-
specific alignment values.
- Add esp32s3_cam_alloc() using kmm_memalign with the alignment
derived from the GDMA external memory block size setting.
- Add esp32s3_cam_free() wrapper around kmm_free.
- Introduce ESP32S3_CAM_EXT_MEMBLK / ESP32S3_CAM_DMA_ALIGN macros
so the block size enum and byte alignment are defined in one
place and stay in sync automatically.
Signed-off-by: wangjianyu3 <wangjianyu3@xiaomi.com>
Fix several issues in the ESP32-S3 CAM driver:
- stop_capture: reset DMA channel, CAM module and AFIFO under
spinlock to fully quiesce hardware before returning. Clear
pending VSYNC interrupt to prevent stale ISR firing.
- uninit: reset CAM/AFIFO before releasing DMA to prevent
in-flight transfers after channel detach. Use esp_teardown_irq
with correct peripheral ID (ESP32S3_PERIPH_LCD_CAM) instead of
irq_detach which corrupts the shared IRQ mapping table. Mask
interrupts and clear pending flags under spinlock before
detaching handler.
- uninit: preserve XCLK output so the sensor remains accessible
via I2C for subsequent re-initialization.
- set_buf/uninit: track driver-allocated vs user-provided frame
buffers with fb_allocated flag to prevent double-free when
using V4L2 USERPTR mode.
Signed-off-by: wangjianyu3 <wangjianyu3@xiaomi.com>
arch/arm/include/imx6/irq.h looks to originally be for i.MX6 Quad
and includes multiple (re)definitions of IRQ names, some of which are
different or don't exist on i.MX6 SoloLite/Solo/Dual/DualLite.
Update irq.h to specify/reserve IRQ names/numbers as defined by the
IM6SDLRM, IMX6DLRM, and IMX6DQRM reference manuals.
Signed-off-by Peter Barada <peter.barada@gmail.com>
The struct stm32_mdio_bus_s aka g_stm32_mdio_bus
has not been referenced at all, because mdio_register()
allocates its own structure. Then lower half drivers
receive only their respective part in mdio_read()
and mdio_write() because MDIO_WRITE/MDIO_READ provides
only d->lower to the specific lower level drivers.
Signed-off-by: Pavel Pisa <pisa@fel.cvut.cz>
Protect the common Espressif Wi-Fi netdev RX queue with a spinlock. The Wi-Fi RX callback enqueues packets while the netdev upper-half RX thread dequeues them, and the IOB queue helpers are not internally serialized.
Under sustained receive traffic that race can corrupt the queue state, strand IOBs, and degrade throughput over time as reported in issue #16915.
Initialize the lock during device setup and use it when clearing the queue on ifup, enqueueing received packets, and dequeuing them for the upper half.
Signed-off-by: aviralgarg05 <gargaviral99@gmail.com>
