syslog:  There is yet another place where the output can get split.  That is in syslog_dev_write():  It will break up the stream to insert a CR before the LF.  This can that can be avoid be generating the CR-LF sequence in the buffer and then detecting and ignoring valid CR-LF sequences, rather than expecting syslog_dev_write() to insert the CR in this case.

I don't like the idea that syslog_dev_write() still scans the entire output buffer to expand CR-LF sequence.  This seems really wasteful, especially in this case where we can be sure that the is no CR or LF without a matching LF or CR.  Bu, I think, the existing behavior in syslog_dev_write() must be retained because it is needed in other contexts.

syslog: Avoid flushing syslog_stream buffer, if possible, until lib_vsprintf() completely parses the format.  This assures that the flush will flush the entire output, even data that may potentially follow the linefeed.  And, in that case, it cannot be interleaved with other devug output.  Suggested by Jussi Kivilinna.

mtd: fix some unallocated and NULL pointer issues.  rwb->wrflush and rwb->wrmaxblocks in rwbuffer could get unallocated values from ftl_initialize() in some configurations.  Also fixes related assert:

  up_assert: Assertion failed at file:rwbuffer.c line: 643

that can happen with the following configuration:

  CONFIG_FTL_WRITEBUFFER=y
  CONFIG_DRVR_WRITEBUFFER=y
  # CONFIG_FS_WRITABLE is not set

These problems are caused by CONFIG variable differences between the buffer layers. TODO: This is not a perfect solution. readahead support has similar issues.

When syslog message has addition characters after last new-line. With buffering those now get lost as vsyslog does not flush output after lib_sprintf. Additional trailing characters could be ANSI escape sequence to reset state that message setups. For example, macro here uses colors and resets state after actual message (including '\n'):

With flushing added to vsyslog, then there is problem that next syslog line might come from other task before reset sequence, causing wrong line getting color. This could be avoided by not flushing on '\n' but only if IOB is full and/or at end of vsyslog. Would this make sense?

STM32L4: Review of last PR + Move separate stm32l4_flash.h; move hardware-specific definitions to chip/stm32l4_flash.h

STM32L4: add internal flash write support

Approved-by: Gregory Nutt <gnutt@nuttx.org>

Moved LIS3DSH from the I2C-dependent block to the SPI-block to make Make.defs consistent with the driver (SPI only) and drivers/sensors/Kconfig.

Approved-by: Gregory Nutt <gnutt@nuttx.org>

Moved LIS3DSH from the I2C-dependent block to the SPI-block to make Make.defs consistent with the driver (SPI only) and drivers/sensors/Kconfig.

I discovered a problem in the file drivers/serial/serial.c concerning the function uart_close(…). In the case that a serial device is opened with the flag O_NONBLOCK the function uart_close(…) blocks until all data in the buffer is transmitted. The function close(…) called on an handle opened with O_NONBLOCK should not block. The problem occurred with a CDC/ACM device.

wireless/ieee802154: Brings closer to Nuttx coding style

Approved-by: Gregory Nutt <gnutt@nuttx.org>

STM32L4: add dbgmcu header files

Approved-by: Gregory Nutt <gnutt@nuttx.org>