| Age | Commit message (Collapse) | Author |
|---|
|
|
|
In hwff_push_n() and hwff_pull_n(), the HWFIFO_ADDR_NEXT_N call after
processing the linear part of a wrap-around read/write used the data
byte count (lin_even) as the address stride increment. On STM32 FSDEV
PMA, data_stride=2 and addr_stride=4, so the pointer must advance by
(lin_even / data_stride) * addr_stride bytes, not lin_even bytes.
Fixes: 74e59e433 ("fix hwfifo pull/push n with address stride > 0")
Signed-off-by: Siddharth Chandrasekaran <[email protected]>
|
|
|
|
|
|
|
|
|
|
implement tusb_time_millis_api() with osal_time_millis() when OS is not NONE
|
|
Signed-off-by: HiFiPhile <[email protected]>
|
|
Signed-off-by: HiFiPhile <[email protected]>
|
|
Signed-off-by: HiFiPhile <[email protected]>
|
|
enumeration delay
|
|
|
|
|
|
update vendor_device.c for direct usbd xfer when either RX/TX BUFSIZE is 0 i.e CFG_TUD_VENDOR_TXRX_BUFFERED = 0
|
|
|
|
|
|
|
|
|
|
HWFIFO access-mode refactor and stride-aware FIFO fixes across FSDEV/RUSB2/MUSB
|
|
Co-authored-by: Copilot <[email protected]>
|
|
Co-authored-by: Copilot <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
read/write
|
|
rusb use custom write
enable all hil test for ra4m1
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
duplicated packet write/read for fsdev
|
|
|
|
|
|
|
|
|
|
|
|
Enable D-Cache unless we're compiling for the M4 core of a dual-core H7 mcu by directly testing CORE_CM4
|
|
|
|
|
|
Signed-off-by: HiFiPhile <[email protected]>
|
|
Signed-off-by: Zixun LI <[email protected]>
|
|
Signed-off-by: Zhihong Chen <[email protected]>
|
|
|
|
|
|
|
