At an interrupt barrier, the host must be able to ensure that no
interrupt generated before reaching the barrier is still pending and
will be delivered after crossing the barrier.
For this, we introduce the following concept:
- interrupts have a serial number. This number is sent to the host
on EP 1 (currently bulk) to signal the interrupt, instead of the
zero byte we used previously.
- the new request ATUSB_SPI_WRITE2_SYNC returns the interrupt
serial number from after the register write (the register write
itself is the interrupt barrier).
- the host can now check if the serial indicated from bulk and the
serial from ATUSB_SPI_WRITE2_SYNC are the same. If yes, interrupts
are synchronized. If not, it has to wait for the interrupt to be
signaled on EP 1.
We should also consider the case that the interrupt serial has gotten
ahead of ATUSB_SPI_WRITE2_SYNC. But that seems to happen rarely. In
any case, it's something for the host driver to worry about, not for
the firmware.
- board.h (irq_serial), board_app.c (irq_serial, INT0_vect): count
the interrupt serial number and return it when signaling the
interrupt
- include/atusb/ep0.h (ATUSB_SPI_WRITE2_SYNC), ep0.c (my_setup):
new request ATUSB_SPI_WRITE2_SYNC that does a register write, then
returns the interrupt serial
This reverts commit 39e3d4e61a.
The log isn't so useful in its present state because we may have to
unplug/replug to get out of a troublesome condition.
- usb/atu2.c (NUM_EPS), board.h (NUM_EPS): moved definition to
board.h
- board.h (NUM_EPS): use one EP for the boot loader, two for the
application
- usb/atu2.c (ep_init): initialize EP 1 only in an application build
- Makefile (USB_OBJS): moved into OBJS and BOOT_OBJS
- Makefile (OBJS): atu2.o is now app-atu2.o
- Makefile (BOOT_OBJS): atu2.o is now boot-atu2.o
- Makefile (MKDEP, %.o): moved dependency generation to macro $(MKDEP)
- Makefile (app-%.o): build from usb/%.c
- Makefile (boot -%.o): build from usb/%.c and set -DBOOT_LOADER
- board_app.c (INT0_vect): always read IRQ_STATUS, even if an
interrupt is already enqueued
- board_app.c (INT0_vect): if an interrupt is enqueued, accumulate
the interrupts signaled since
- board_app.c (irqs_more): when a pending transfer completes, send
interrupts accumulated since (if any)
- board.h (gpio_cleanup), board_app.c (gpio_cleanup): restore INT0
when done manipulating GPIOs
- board.c (gpio): updated comment explaining how to restore INT0
- include/atusb/ep0.h (enum atspi_requests), ep0.c
(ATUSB_GPIO_CLEANUP): new request to return to normal operation
after ATUSB_GPIO
Since GPIO manipulations may cause stray INT0 activity, we turn off
INT0 handing on ATUSB_GPIO. The MCU must be reset to restore access
to INT0.
There are still issues to resolve with the GPIO test. It may turn
out that there's a less invasive solution than just turning off
INT0 completely.
- board_app.c (gpio): mask INT0 before manipulating GPIOs
- board_app.c (board_app_init): make it clearer that EIMSK is a bit
mask
- board.h (board_app_init), board_app.c (board_app_init): new function
for application-specific board initialization
- atusb.c (main): call board_app_init
- board_app.c (INT0_vect): on RF interrupt, read REG_IRQ_STATUS and
send the status byte on EP1
- board_app.c (board_app_init): set up interrupt on rising edge of
INT0 (INT_RF)
- descr.c (config_descriptor): added EP1 as bulk IN
The code is quite special-cased. Some more changes will be necessary
for interrupt or OUT EPs.
- atu2.c (NUM_EPS): increased from 1 to 2
- atu2.c (ep_rx): send zero-length packet also for EPs != 0
- atu2.c (handle_ep): clear FIFOCON when sending data on EP != 0
- atu2.c (ep_init): configure EP 1 as bulk IN with size 64
- usb.c (NUM_EPS): we don't need to define NUM_EPS here
- ep0.c (setup_request): store combined request code in "req", for
later reuse
- include/ep0.h (ATUSB_SPI_WRITE, ATUSB_SPI_READ1, ATUSB_SPI_READ2),
ep0.c (setup_request): new requests that translate easily into
general SPI operations
With warnings treated as errors I get this:
In file included from flash.c:19:0:
usb/dfu.h:107:35: error: 'struct setup_request' declared inside parameter list
usb/dfu.h:107:35: error: its scope is only this definition or declaration, which
is probably not what you want
make: *** [flash.o] Error 1
Struct setup_request is declared in usb.h so include it.
Signed-off-by: Stefan Schmidt <stefan@datenfreihafen.org>
This saves about 2 mA, leaving about 8 mA when idle. The transceiver
should consume 0.5 mA in TRX_OFF, CLKM up to 4 mA, and the idle MCU
core 1 mA. USB current is unknown.
- boot.c (MS_TO_LOOPS): increased loop count now that we no longer
have to poll
- boot.c (main): increased delay from 2.0 s to 2.5 s, because we were
racing with a sleep(2) in dfu-util
- usb/usb.h (usb_ep_change), usb/atu2.c (usb_ep_change): new function
called by the USB stack to notify the hardware-specific driver of
an endpoint state change (EP_TX or EP_RX)
- usb/usb.c (usb_io): call usb_ep_change
- usb/atu2.c (handle_ep): mask TXINI if we have nothing to send
Note: this change surprisingly _increases_ the DFU wait in the boot
loader. Not yet sure why.
- boot.c (main): move the interrupt vectors to the boot loader
section
- atusb.c (main): move the interrupt vectors to the application
section
- boot.c (main): enable global interrupts while looping (disable
them before jumping to the application)
- board_app.c (__timer_read, timer_read): removed wrapped since
we're now always called with interrupts disabled
- usb/atu2.c (ep_init): enable endpoint interrupts
- usb/atu2.c (usb_init): enable device interrupts
- usb/atu2.c (usb_poll, USB_GEN_vect, USB_COM_vect): moved poll
loop code into separate handlers for device and endpoint
interrupts
- boot.c (main), atusb.c (main): removed call to usb_poll
The old code could accidently acknowledge interrupts that appeared
between the read and the write in UxINTx &= mask.
- atu2.c (handle_ep): writing to UEINTX implies a race-free "and"
- atu2.c (usb_poll): writing to UDINT implies a race-free "and"
- board.c (board_init), board_app.h (timer_init), board_app.c
(timer_init): moved timer initialization from board_init to
new function timer_init
- atusb.c (main): call timer_init
This allows us to do other things as well, e.g., change state.
- ep0.c (ep0_init), usb/dfu.c (dfu_init): use user_setup
- usb/atu2.c (usb_poll): no need to reset user_setup - the user's
reset function can do that
- usb/usb.h (user_setups), usb/usb.c (handle_setup): removed
user_setups
- usb/usb.h (user_set_interface), usb/usb.c (user_set_interface,
handle_setup): callback for SET_INTERFACE
The measured values are:
board idle, looping 9 mA
LED on + 5 mA
TX CW (+3 dBm) +14 mA
------
28 mA
======
Adding a margin of 12 mA, we get 40 mA. DFU used 30 mA so far, the
application 50 mA.
- board.h (BOARD_MAX_mA): define maximum USB bus current
- descr.c (config_descriptor), usb/dfu.c (config_descriptor):
use BOARD_MAX_mA for bMaxPower instead of hard-coding values
- ep0.c (ep0_init), usb/dfu.c (dfu_init): set user_setups[0] instead
of user_setup
- usb/atu2.c (usb_poll): reset user_setup on bus reset
- usb/usb.h (user_setups), usb/usb.c (user_setups): array of
interface-specific setup functions
- usb/usb.c (handle_setup): in SET_INTERFACE, select setup function
from user_setups according to interface
- usb/usb.c (handle_setup): if user_setup is not set (e.g., the
optional SET_INTERFACE was never issued), fall back to user_setups[0]
- spi.h (spi_off), spi.c: disable the UART
- spi.c (spi_initialized, spi_begin, spi_init): initialize the UART
if necessary
- board_app.c (gpio): call spi_off to disable the UART instead of
open-coding the operation
- board_app.c (gpio): don't explicitly re-enable the UART but defer it
to the next communication
reset_rf now resets all GPIOs to their default state. This way, we
can easily recover from an incomplete or incorrect use of ATUSB_GPIO.
- atusb.c (main), board.c (reset_rf): moved call to spi_init into
reset_rf
- board.c (reset_rf, board_init): moved GPIO setup to reset_rf
- board.c (reset_rf): set GPIOs to reset defaults
- board.h (gpio), board.c: read/modify/write all settings of GPIO
ports
- include/atusb/ep0.h (enum atspi_requests): added new request
ATUSB_GPIO
- ep0.c (my_setup): ATUSB_GPIO reads/reconfigures a GPIO port
- ep0.c (my_setup): ATUSB_SRAM_READ is FROM_DEV, not TO_DEV
- ep0.c (my_setup): in ATUSB_SRAM_READ, write setup->wLength bytes,
not "size" (which is uninitialized)
- atusb.c (main): poll the 16 bit timer for overflows
- board.h (timer_poll, timer_read), (board.c (timer_h, timer_poll,
timer_read, board_init): added support for a free-running 48 bit timer
(16 bits in hardware, 32 bits in software)
- include/atusb/ep0.h (ATUSB_TIMER), ep0.c (my_setup): new request
ATUSB_TIMER to retrieve the value of the 8 MHz counter
- include/atusb/ep0.h (enum atspi_requests): describe what the groups
of requests do
- usb/usb.h (USB_LANGID_ENGLISH_US): added USB LANGID for US-English
- board.h (board_sernum), board.c (board_sernum, hex, get_sernum,
board_init): provide the board's serial number in "board_sernum"
(UTF-encoded)
- sernum.h (sernum_get_descr), sernum.c (sernum_get_descr): return
string descriptors for the serial number
- descr.c (device_descriptor), usb/dfu.c (device_descriptor):
set iSerialNumber if serial number is available
- atusb.c (main), usb/dfu.c (my_descr): call sernum_get_descr for
unknown descriptors
- Makefile (OBJS, BOOT_OBJS): added sernum.o
What caused the error that looked like a problem with the functional
descriptor was in fact the boot loader resetting between the bus scan
and retrieval of the descriptor.
- board.c (board_init): disable the watchdog timer (tricky !)
- board.h (reset_cpu), board.c: enable the watchdog timer to cause a
CPU reset
- ep0.c (my_setup): uncommented and updated ATUSB_RESET handler
We currently don't provide the DFU Functional Descriptor, which modern
versions of dfu-util request to determine the transfer size. Luckily,
they don't do this if the transfer size is given on the command line.
- usb/atu2.c (ep_init): moved before usb_poll
- usb/atu2.c (usb_poll): register bit was used as mask, not as shift
- usb/atu2.c (usb_poll): call ep_init on USB bus reset
- usb/atu2.c (usb_reset): don't make USB bus reset force a hardware reset
- Makefile (.PHONY, dfu): new target to upload the application with DFU
- usb/usb.h, usb/atu2.c (usb_reset): reset the USB bus by detaching and
re-attaching the device
- boot.c (main): force a USB reset before running the payload
- usb/atu2.c (usb_poll): test for USB reset from the host (in progress)
- usb/dfu.c (my_reset, dfu_init): register user USB reset handler
- dfu.h (flash_end_write), flash.c (flash_end_write): write any incomplete
buffer
- flash.c (flash_write, flash_end_write): call boot_rww_enable only at the
very end, it won't erase our buffer in mid-page
- usb/dfu.c (my_setup): call flash_end_write at the end of a download
- flash.c (flash_write): we don't use eeprom_busy_wait, removed it
- flash.c (flash_write): write the page only at one place
- flash.c (flash_write): corrected the address of the page write
- flash.c (flash_write): value being added to word wasn't shifted
- Makefile (BOOT_ADDR, boot.elf): use variable instead of hiding the
address in a command
- Makefile (CFLAGS): pass BOOT_ADDR as a macro
- Makefile: removed commented-out application flashing code
- boot.c: basic boot loader that runs DFU for 2 s, then starts the payload
- board.h (DFU_USB_VENDOR, DFU_USB_PRODUCT): added USB IDs for DFU
- flash.c: stubs for board-specific Flash functions
- Makefile: build boot.hex for the boot loader
- Makefile (prog): load the boot loader at its rightful place
- Makefile (prog): also set hfuse and the lock fuse
- dfu.c: updated includes
- dfu.c (device_descriptor): renamed USB IDs from USB_VENDOR/PRODUCT to
DFU_USB_VENDOR/PRODUCT to allow differentiation
- dfu.c: changed all __bit to "int"
- dfu.c: removed all __xdata and __reentrant
- dfu.c: changed "ep0" to "eps[0]"
- dfu.c (payload, flash_erase_page, flash_write_byte, block_write,
block_receive, block_transmit, my_setup): abstracted Flash interface
and removed target-specific operations
- dfu.h: added prototypes for target-specific Flash operations
- dfu,c (my_setup, my_descr): removed SDCC-specific hacks
- dfu.c (my_reset): commented out - did we actually use this ?
- Makefile, atusb.pro, atusb.sch, atrf.sch, usb.sch, atusb.cmp, atusb.brd:
resurrected from git history (commit
14b00823b0)
- fw/: moved all C8051F326-specific files over from atusb/fw/
- fw/include/: copied MCU-independent files over from atusb/fw/include/
- fw/include/atusb/ep0.h (HW_TYPE_110131): defined hardware type 2 for
ATmega32U2-based boards
- fw/ep0.c: define HW_TYPE as HW_TYPE_110131, not directly as 2
- fw2/atusb.c: include atusb/ep0.h instead of declaring ep0_init locally
- at86rf230.h (REG_BATMON): address is 0x11, not 0x10
- at86rf230.h (TRX_CMD_MASK): width is 5 bits, not 3
- at86rf230.h (CLKM_CTRL_MASK): width is 3 bits, not 2
- at86rf230.h (PA_EXT_EN, IRQ_2_EXT_EN, TX_AUTO_CRC_ON_231): shifted by one
bit
- at86rf230.h (RX_BL_CTRL, AACK_DIS_ACK): added missing field
- at86rf230.h: added fields PA_BUF_LT and PA_LT
- at86rf230.h (ANT_DIV): added fields of this register
- at86rf230.h (NATMON_VTH_SHIFT, NATMON_VTH_MASK): corrected typo
- fw/Makefile: replaced "make" with $(MAKE) (just for style)
- common/Makefile.common: Makefile settings shared within project. For now,
this contains only the board version, which defaults to 2010-12-16.
- fw/common/Makefile, fw/boot/Makefile, fw/atusb/Makefile: include
common/Makefile.common
- fw/common/Makefile, fw/boot/Makefile, fw/atusb/Makefile: pass board
version to cpp and gcc
- fw/atusb/atusb.c (init_io): individually set IRQ_RF to one, LED and TST
to zero
- fw/atusb/atusb.c (init_io): added macros to set all unused pins to zero
in a way that doesn't need updating if a signal moves from one pin to
another
- include/atusb/ep0.h: added hardware type 1 (2010-12-16 design)
- common/config.h: set hardware type depending on board version
- common/io.h: assign pins depending on board version
- atusb/fw/boot/Makefile: still referred to the project as ATSPI (not ATRF)
in title comment
- atusb/fw/include/atusb/usb-ids.h: still had old atspi path in title comment
- common/io.h: TST is connected to P0_7
- include/at86rf230.h: added test mode registers REG_CONT_TX_0 (0x36) and
REG_CONT_TX_1 (0x3d), with their values
- atspi/atspi.c (set_test_mode): set or reset TST and the LED
- atspi/atspi.c (reset_rf): leave test mode
- atspi/atspi.c (test_mode): enter test mode
- atspi/atspi.c (main): flash the LED when in test mode
- include/atspi/ep0.h (enum atspi_requests), atspi/ep0.c (my_setup): new
command ATSPI_TEST to enter test mode
- include/atspi/ep0.h (EP0ATSPI_MINOR): bumped protocol to version 0.1
- include/at86rf230.h: corrected comment before TRX_CMD field to indicate
that the register is called TRX_STATE
- atrf/: rename to atusb/
- atrf/wpan-atrf.pro, atrf/wpan-atrf.sch, atrf/wpan-atrf.brd,
atrf/wpan-atrf.cmp: rename to atusb.*
- atrf/Makefile: change ben-wpan to atusb
- atrf/atusb.pro (LastNetListRead): update for name change
- tools/Makefile.common (CFLAGS): change fw/ include location from atrf/
to atusb/
