Documentation/ReleaseNotes/NuttX-7.9

NuttX-7.9
---------

The 109th release of NuttX, Version 7.9, was made on April 14, 2015,
and is available for download from the SourceForge website. Note
that release consists of two tarballs: nuttx-7.9.tar.gz and
apps-7.9.tar.gz. Both may be needed (see the top-level nuttx/README.txt
file for build information).

Additional new features and extended functionality:

* Core OS:

- Assertions: Add support for dumping board-specific information on
assertions. From David Sidrane.
- boardctl(): : Add a new non-standard OS interface called boardctl().
This is similar to a driver IOCTL call. But this is an IOCTL call
directly on the board logic. This function will eventually replace
all of the ad hoc OS interfaces that are current used to perform
application specific initialization and application driver test
configuration. It essentially formalizes and institutionalizes
these rogue interface in to at least a single crazy call.

* Graphics:

- Line Joins: Line drawing graphic functions now take an argument that
will draw a circular "cap" on the ends of lines to support cleaner
joining of lines segments..

* Common Drivers:

- SPI iInterface: Add an SPIDEV_USER so that there can be board-
specific definitions of SPI devices.
- I2C Interface: Add a user provided void * argument to the I2C
callback method for the I2C slave interface. Suggested by Nate
Weibley.
- AT24 EEPROM Driver: Add support for (1) the byte-oriented read()
method, (2) devices that have extended memory regions, and (3)
devices that use 8-bit addressing.
- Freescale MPL115A barometer: Add driver for the Freescale MPL115A
barometer. From Alan Carvalho de Assis.
- CS2100-CP: Add a driver for the CS2100-CP Fractional-N Multiplier
chip.
- RTC IOCTls: Added the definitions of a driver based RTC interfaces
and RTC upper half driver.

* File Systems/Block Drivers/MTD:

- umount2(): Added to support un-mount flags. umount() is now a macro
that just calls umount2() with flags = 0.
- File system interface: Pass the umount2() flags to every unbind()
implementation. That is where the decision to umount or not and how
to umount will be made.
- FAT unbind: Add support for umount2(target, MNT_FORCE) in the FAT
file system.

* Networking:

- netdev_register(): Improve the network device registration logic.
When multiple link types are used, modify how each interface number
is assigned to the device name by incrementing the device number
individually for each interface link type. From Max Neklyudov.
- Add TUN network device. Used to support the Netutils PPPD
implementation. From Max Neklyudov.

* ARMv7-M

- ARM Cortex-M7: Add basic support for the Cortex-M7 architecture.
- ARMv7 Common Lazy Exception handling. Add common support for lazy
FPU state saving. MCU-specific lazy FPU saving was already in place.
This change just makes the code share-able.

* Atmel SAMA5D3/4 Drivers:

- SAMA5 Serial: Backport support for flowcontrol and termios from
SAM3/4.

* Atmel SAMV7:

- Atmel SAMV7: Add architecture support for the Atmel SAMV7 family
(Cortex-M7)

* Atmel SAMV7 Boards:

- Atmel SAMV71 Xplained Ultra: Add board support for this SAMV7 board.
This initialize release includes support for: SDRAM, TWI/I2C, SPI,
Ethernet, HSMCI, DMA, GPIOs, UARTs, AT24 EEPROM and the maXTouch
Xplained Pro LCD. Drivers are available for SSC/I2C, the WM8904,
the maXTouch trouchscreen, and USB device but have not yet been
fully integrated in this release.

* SiLabs EFM32 Drivers:

- EFM32 USB Device is now functional with the efforts of From Pierre-
noel Bouteville.
- EFM32 I2C: Add I2C driver. From Pierre-noel Bouteville.

* STMicro STM32:

- STM32F372/F373: Adds architecture support for this STM32 parts.
Tested on STM32F373CC. Contributed by Marten Svanfeldt.

* STMicro STM32 Drivers:

- STM32 RTCC: Break out a new internal interface, stm32_rtc_setdatetime().
This eliminates some un-necessary time conversions. From Freddie Chopin.
- STM32 RTCC: Add an implementation of the RTC lower half interface
for the STM32.
- STM32 RTCC: Extend the RTC interface to support reading subseconds.
From Jussi Kivilinna.
- STM32 BBSRAM: Adds the ability to use the STM32F2 and STM32F4 Battery
Backed SRAM in the file system. With an option to Save Panic context
to one of the files. From David Sidrane.

* STMicro STM32 Boards:

- Olimexino-STM32: Board support for the Olimexino STM32 board from David
Sidrane.
- Viewtool-STM32F107: Add support for an optional, add-on Freescale
MPL115A baramoter. From Alan Carvalho de Assis.

* TI Tiva:

- TM4C1294: Add support for the TM4C1294. Contributed by Frank Sautter.

* TI Tiva Drivers:

- Tiva ADC: Updated files to allow for ADC and PWM triggering by a
timer. Refactoring of ADC implementation. From Calvin Maguranis.
- Tiva Timer: Refactoring by Calvin Maguranis

* TI Tiva Boards:

- TM4C123G-launchpad: Add ADC support. From Calvin Maguranis.
- TM4C123G-launchpad: Add board button interrupt support. From Calvin
Maguranis.
- TM4C1294 Launchpad: Add support for the TM4C1294 Launchpad.
Contributed by Frank Sautter.

* Microchip PIC32

- PIC32MZ: Add architectural support for the PIC32MZ (MIPS32 M14k).
Includes contributions from Kristopher Tate.

* Microchip PIC32MZ Boards

- PIC32MZ Starter Kit: Add basic board support for the PIC32MZ.
Includes contributions from David Sidrane.

* C Library/Header Files:

- gettimeofday(): This function is no longer a core OS interface. It
is only a wrapper around clock_settime().
- settimeofday(): Added. gettimeofday has been deprecated in POSIX
2008. settimeofday() was never part of POSIX, but I decided to add
it to libc as well just for symmetry.
- Day-of-Week: Add CONFIG_TIME_EXTENDED to support the last 3 members
of the tm struct and support for filling them in with the day of the
week. From David Sidrane.
- sighold(), sigrelse(), sigignore(), sigset() and sigpause(): New
signal handling functions.
- sys/select.h: Extend the fd_set type definition so that it can
handle more than 32 descriptors (if so configured). From Max
Neklyudov.
- netdb.h: Add a mostly-commented-out version of the standard netdb.h
header file.
- include/nuttx/board.h: Remove common board function prototypes from
other files and consolidate them in this new header file.

* Applications:

- OS test: Improved error checking for sem_timedwait() from Juha
Niskanen.
- OS test: Add a simple test of sigprocmask, sighold, and sigrelse.
- PPPD test: Add PPPD test case. Split out from the PPPD daemon by
Brennan Ashton.
- Netutils: Add a PPP daemon. From Max Neklyudov.
- NSH: NSH initialization now calls boardctl(BOARDIOC_INIT) instead
of nsh_archinitialize(). Remove support for CONFIG_NSH_ARCHMAC.
It is not used and there are better ways to do that operation.
- NSH: Allow NSH 'date' command with no hardware RTC. This command
is useful without an RTC too. Also, this permits testing on the
simulator which never has an RTC.
- apps/examples/: Several tests changed (touchscreen adc, pwm,
graphics). No longer call directly into the OS, but rather
indirectly via boardctl().
- Modbus: Includes extensions to support RTU master mode by
Armink(383016632@qq.com): https://github.com/armink/FreeModbus_Slave-Master-RTT-STM32.
Ported to NuttX by Darcy Gong.
- TZ database: All logic to build timezone database needed for
localtime() support. Does not do much else yet.

Efforts In Progress. The following are features that are partially
implemented but present in this release. They are not likely to be
completed soon.

* Processes. Much of the work in this release is focused on the
realization of Unix-style user processes in NuttX. There is more to
be done, however. The full roadmap and status is available at:
https://cwiki.apache.org/confluence/display/NUTTX/Memory+Configurations

* XMega: There are some fragments in place for an XMega port. That
port has not really started, however.

Bugfixes. Only the most critical bugfixes are listed here (see the
ChangeLog for the complete list of bugfixes and for additional, more
detailed bugfix information):

* Core OS:

- atexit(): Fix compile failure due to reference to undefined variable
in certain configurations. From Juha Niskanen.
- syscalls: Last changes to message queue prototypes did not make it
into call gate logic. Some additional errors had crept as well since
the system calls are seldom used.
- waitpid(): Do not use uninitialized boolean in waitpid(). From Juha
Niskanen.
- Timing functions: Fix some nanosecond comparisons. Comparisons should
be greater than OR EQUAL TO 1000000000 for maximum value of nanoseconds
in struct tm. From Juha Niskanen.
- Time initialization: Fix an error in time initialization when there
is no RTC and the time is initialized from a fixed configured value.
The call to clock_calendar2utc() was returning the time in units of
seconds. The initialization logic, however, was expecting to get
time in units of days. This problem has been there forever but
obviously has not significant impact to behavior in systems where
time is not visible.
- mq_timedsend(): Should not check for valid time if there is space
in the message queue. Noted by Freddie Chopin.
- select() and sem_timedwait(): Fix places where the errno value was
being overwritten by subsequent actions so that the returned errno
value was incorrect. From Max Neklyudov.
- clock_systimer64(): Fix a problem in clock_systimer64 that occurs
when (1) the 64-bit system time is enabled, and (2) the value of
CONFIG_USEC_PER_TICK is less than 1 milliseconds (such as when using
the tickless mode of operation). In that case, the conversion of
time to 64-bit millisecond value in clock_systmer64() causes some
bad times to be returned. Time was converted to milliseconds, then
to configured ticks. Precision was lost in the millisecond
conversion. The fix is to first convert time to a 64-bit microsecond
value, then to the configured tick value. Noted by David Sidrane.
- clock_systimespec(): When an RTC is used clock_systimespec() must
subtract the basetime from the returned time. It needs to return
the time since power up, no the current time. Issue noted by Max
Neklyudov.
- AIO: correct some backward tests in aio_read() and aio_write():
Socket descriptors have higher numbers than file descriptors. In
aio_contain(), Copy u.ptr (void *) when initializing aioc and not
the case specific u.aioc_filep. From Michal Ulianko.

* File Systems/Block Drivers/MTD:

- procfs: Fix potential NULL pointer dereference of 'node'. From
Juha Niskanen.
- FAT: fat_unbind() was accessing 'fs->fs_sem' after freeing the 'fs'
struct. From Juha Niskanen.
- mkfatfs: Use DMA memory for mkfatfs when needed. This makes mkfatfs
use fat_dma_alloc() when CONFIG_FAT_DMAMEMORY is set. This is needed
to ensure mkfatfs operates with boards that use DMA for microSD.
From Andrew Tridgell via the PX4/NuttX repository.
- FAT unmount: Fix a bug in the FAT unbind() logic. There were
problems with the way certain internal list handling was
implemented. The end result is that newly opened or cloned file
structures were never being added to the list of open files. So
when the unmount() happens, it always looked like there were not
open files and a crash could ensue.
- FAT mounting: Fix misthink when CONFIG_ENDIAN_BIG=y. From Lwazi Dube.

* Common Drivers:

- CAN upper half driver: Some lines had to be reordered in the can_open
function otherwise the open count will only be incremented from 0 to
1. This caused problems when trying to open the can bus from two
different tasks. SourceForge Patch #46.

* Crypto:

- crypto: Fix issues when AES support was added for the STM32L1.
From Juha Niskanen.

* Networking:

- Network initialization: Divide net_intiialize() into net_setup() and
net_initialize() to solve a chicken-and-egg problem. net_setup()
must be called before up_initialize() is called so that networking
data structures are ready to register new network devices.
net_initialize() now does only timer related operations and is
called AFTER up_initialize() where the timers are configured.
Problem noted by Max Neklyudov.
- netdev_unregister(): Fix a list handling error. From Max Neklyudov.
- Socket dup: Do not call sockfd_socket() twice. From Juha Niskanen.

* Common Drivers:

- max11802 touchscreen: Fix possible unmatched IRQ restore (not
normally a problem, but could happen in certain configurations).
Noted by Juha Niskanen.
- MMCSD SPI Multi-block transfers: ACMD23 must be preceded with CMD55
command as any other ACMD commands. Also, after the block reading
wait loop must be inserted before writing the next block of data to
the sd card. From Dmitry Nikolaev via Jussi Kivilinna.
- MMCSD SDIO: Fix some bad logic when file system debug is turned on:
Arguments to syslog were missing so that garbage was being printed.
- Common upper-half serial driver: Report correct number of bytes free
in serial buffer with FIONWRITE. From Sami Pelkonen via Jussi Kivilinna.
- Pipes/Fifos: Fix zero-lenth writes to pipes/FIFOs. From Jussi Kivilinna.
- USB device MSC: Add NULL pointer check. From Juha Niskanen.
- BCH driver: Avoid NULL pointer dereference when bch->refs == 255.
From Juha Niskanen.
- Various drivers: The inode unlink method should not be support if
operations on the root pseudo-filesystem are disabled.

* ARMv7-M

- ARMv7M MPU: Bad syntax in ARMv7-M MPU logic would cause failure to
write the correct value to the MPU_RASR register. From Juha Niskanen.

* Atmel SAM3/4 Drivers:

- SAM3/4 Serial TERMIOS: A side-effect of changing serial settings
via TERMIOS (such as tcsetattr) is that serial interrupts were being
left disabled. This is not a problem if the serial configuration is
changed when there are no open references to the serial device. In
that case, serial interrupts are disabled and will not be enabled
enabled until the serial device is first opened. But it is fatal if
the serial device is already opened and if there is a task waiting
to receive data. In that case, the side-effect of disabling
interrupts is fatal: That task is then left hanging with interrupts
disabled.
- SAM3/4 Serial: The IMR register is read-only. This means that
sam_restoreints() did not actually re-enable UART interrupts
(Fortunately, that function was not being used up to this point).
- SAMA5 HSMCI: Fix a bug in SAMA5 HSMCI. The bitfield mask and shift
values were reversed resulting in a trashed value for the number of
blocks in the BLOCKR register. This was sufficient to prevent DMA
writes from working.

* Atmel SAMA5 Drivers:

- SAMA5 Timer/Counter: Fix typos in timer/counter header file. From
Bob Doiron.

* STMicro STM32 Drivers:

- STM32 PWR and RTC: Changes to support fully write protecting the
backup domain. N.B. stm32_pwr_enablebkp did not account for the delay
from enable to the domain being writable. The KISS solution is a
up_udelay. A more complex solution would be a negated write test
with restore. From David Sidrane.
- STM32 SPI: Add missing SPI callback functions to the STM32 SPI
driver. From Freddie Chopin.
- STM32 I2C: Add missing NULL check. From Juha Niskanen.
- STM32 USB device: Fix stale initialization invalidating later NULL
check. From Juha Niskanen.
- STM32 UART: Fix RX DMA setup for UART5. From Jussi Kivilinna.
- STM32 UART: Make input hardware flow-control work with RX DMA. From
Jussi Kivilinna.
- STM32 RTC and clock control: The STM32F4Discovery board doesn't
come with a Low speed external oscillator so the default LSE source
for the RTC doesn't work. In stm32_rtcc.c the up_rtc_initialize()
logic doesn't work with the LSI. The check on RTC_MAGIC on the BK0R
register lead to rtc_setup() call that rightfully enables the LSI
clock; but the next times, when the rtc is already setup, the
rtc_resume() call does NOT start the LSI clock! The right place to
put LSE/LSI initialization is inside stm32_stdclockconfig() in
stm32fxxxxx_rcc.c. Doing this I checked the possible uses of the
LSI and the LSE sources: the LSI can be used for RTC and/or the
IWDG, while the LSE only for the RTC (and to output the MCO1 pin).
This change is not verified for any other platforms. From Leo
Aloe3132

* Tiva Drivers:

- Tiva GPIO driver needed small fixes in the configuration routines and
discovered false-positive bugs in interrupt testing: interrupts are
now verified to actually be working reliably. From Calvin Maguranis .

* MIPS32:

- mips32/Toolchain.mk: Current Pinguino toolchain uses the prefix p32-
instead of mips-.

* C Library/Header Files:

- sigaddset() and sigdelset(): Need to set errno if a bad signal
number is received.
- sfrtime(): Missing implementation of %C in sfrtime() C (was being
treated as %y). From Freddie Chopin.
- pthread.h: Fix PTHREAD_COND_INITIALIZER.
- ungetc.c: Fix NULL pointer dereference in ungetc() if 'stream' was
NULL, 'stream->fs_oflags' was evaluated. From Juha Niskanen.
- Fixes to asinh(), atanh(), and sinh(): The 'basic' expansions all
exhibited bad cancellation errors near zero (<= 1E-10). This can be
easily seen e.g. with x = 1E-30, the results are all zero though they
should be extremely close to x. The cutoff values (1E-5, 1E-9) are
chosen so that the next term in the Taylor series is negligible (for
double). Functions could maybe be optimized to use only first term (x)
and a smaller cutoff, just bigger than where the cancellation occurs
- localtime(): Inconsistent configuration name: CONFIG_LIBC_TZDIR vs
CONFIG_LIBC_TZ_TZDIR. Former wins.

* Applications:

- Modbus: Fix some compile problems when TCP is enabled.