changelog

A real-time serial port device driver, works with RT-Linux and RTAI.

2002-01-21  Jochen K黳per  <[e-mail]>

        * release 0.5.5

2001-05-08  Jochen K黳per  <[e-mail]>

	* rt_com.c: improvements submitted by James H. Puttick: 
	- In init_module, call rt_com_free_irq(), rather than free_RTirq() (the
	  latter symbol doesn't exist in RTAI version 24.x.x)
	- In rt_com_set_param(), do not set p->used until after check_region()
	  has been called. Otherwise, if check_region fails, and then the module
	  is unloaded, the region is released, even though the module did not
	  allocate it.
	- Changed rt_com_buff_free() to return the number of bytes of usable
	  space (an empty queue can only hold 255 bytes, not 256)
	- Added the ability to specify a call-back function, which is called
	  whenever characters are received and added to the queue

2000-11-15  Jochen K黳per  <[e-mail]>

	* rt_com.c (rt_com_isr): bug-fix submitted by Giuseppe Renoldi: 
	I'm using rt_com 0.5.3 with RTAI, and analyzing the code I think to
 	have found a bug and a "not optimal behaviour". The bug is that at the
 	end of the interrupt routine there is a wrong check to see if there are
 	other interrupts pending.  In fact the original code checks if the low
 	nibble of IIR is zero, but when there are no interrupts pending, the
 	low nibble of IIR is 0x01 and not zero!!!. And if there are no
 	interrupt, the default is to execute the "disable Transmitter Holding
 	Register Empty Interrupt" and you can see that it can be executed 4
 	times (4 useless and expensive IOs).
	The "not optimal behaviour" is that the rt_com use implicitly only 4
 	characters of the TX FIFO. So you have a TX interrupt every 4
 	characters instead of 16. I think that the "magical" number 4 that
 	initialize the loop variable is related to the "4 characters time" of
 	the RX timeout (or the 4 possible interrupt sources ??) but anyway I
 	think that it is not correct to do so.  The right way should be to put
 	characters in the FIFO only when you are sure that it's empty so you
 	can put in 16 characters.


2000-05-05  Jochen K黳per  <[e-mail]>

	* release 0.5.3

	* README: Updated ml info, removed note regarding setserial, it wasn磘
	correct any more and should be in the docs.

	* rt_com.c (rt_com_setup): Assure correct parity flags by bitwise and
 	with the parity bitmask ( 0x38 ) instead of the wrong if-clause.

2000-05-03  Jochen K黳per  <[e-mail]>

	* rt_com.c (rt_com_set_param): If port is negative, deinitialize
 	hardware setup. If used-meber is changed, request or release the memory
 	region and the IRQ line accordingly.

2000-04-12  Jochen K黳per  <[e-mail]>

	* release: 0.5.2

	* rt_com.h (RT_COM_BUF_SIZ): Typo - set to 0x0100 !

2000-03-10  Jochen K黳per  <[e-mail]>

	* release: 0.5.1

	* Makefile (rt_com.tar.gz): 

	* PROBLEMS: Add this file.

	* rt_com.c, rt_com.h, rt_comP.h: Cleanup.

	* Makefile: Modified to match the packaging changes: Different
	documentation, licensing files. Preserve relative paths when creating
	distribution. 

1999-11-16  Jochen K黳per  <[e-mail]>

	* rt_com.c (rt_com_enable_fifo): Put "namespace prefix" to enable_fifo.

1999-11-11  Jochen K黳per  <[e-mail]>

	* rt_com.c, rt_com.h, rt_comP.h:
 	Put in modifications by Roberto	Finazzi <[e-mail]>:
	- Setting port and irq addresses for a com from user space. In this
	  case the port region request isn't performed by init_module but at
	  the end of this function.
	- You can set many functioning mode of the hand shake signals. It is
	  possible to make a communication without any hand shake signals
 	  ( only RXD, TXD and GND connected ) or you can also have an
	  automatic hardware flow control by RTS-CTS signals. 
	- It is possible to clear input and output buffer
 	  ( the characters stored in the fifos also ).
	- You can read the remaining free space in the output buffer.
	- You can read or write all the hand-shake signals 
	  ( DTR, RTS, CTS, DSR, RI and DCD ).
	- You can have the last error detected by UART ( Receive data overrun,
	  Parity error, Framing error, Break detected) and by software ( input
 	  buffer full ).

1999-11-04  Jochen K黳per  <[e-mail]>

	* release: 0.4

	* test/: Created this directory and put examples in here.

1999-10-28  Jochen K黳per  <[e-mail]>

	* rt_comP.h (rt_com_request_irq): On RTAI enable the irq after
	requesting it.

1999-10-27  Jochen K黳per  <[e-mail]>

	* rt_comP.h: Use spinlock to mark critical sections in RT-Linux v2.

1999-10-22  Jochen K黳per  <[e-mail]>

	* Makefile.standalone: Updated for the new files.
	Provide dist target.

	* rt_com_posix.c, testing.c: Added.

	* TODO: Created this file. Put in Michael's suggestions.

	* README: Merged with Michael's README.RTL.

1999-10-21  Jochen K黳per  <[e-mail]>

	* Makefile.standalone: Include rtl.mk. Provide target clean.

	* rt_com.c: Change include statements to support RTLinux 2.0pre1.
	Do not define MODULE, __KERNEL__, __RT__ twice.

1999-10-01  Jochen K黳per  <[e-mail]>

	* rt_com.c: Make this module RTAI aware. It should work for NMT
	RT-Linux v1 and v2 and for RTAI ( v0.6 ).

	* rt_comP.h: Generate defines to hide the differnt interrupt handling
	of different NMT major versions and RTAI.

1999-07-28  Jochen K黳per  <[e-mail]>

	* Makefile.am (package): Added this rule to package up the stuff.

	* License: Moved our old COPYING file here.
	
	* COPYING: Changed this to the GPL file.

	* rt_com.c (init_module): Put rt_com_setup out of ifdefs, its needed in 
	any case.

1999-07-27  Jochen K黳per  <[e-mail]>

	* README: Added. This file shall contain some information on the
	purpose and use of the package.

	* Makefile.standalone: Created this file to allow simple compilation
	without use of automake/autoconf.

	* rt_comP.h: ifdef'ed declaration of interrupt handlers for RT-Linux
	version.

	* rt_com.h, rt_comP.h: Put Hua Mao into Copyright notice.

1999-07-26  Jochen K黳per  <[e-mail]>

	* rt_com.c (rt_com0_isr): Return the value we get from rt_com_isr, not
	just "0" all time.

	* rtlinux-2 conformity: Incorporate changes by Hua Mao from Victor
	Yodaikens group to make the module RT-Linux v2 conform.
	To get a RT-Linux v1 conforming version, define RTLINUX_V1. The small
	differences regarding the interrupt handler stuff are ifdef'ed betweend
	the two versions.

1999-07-23  Jochen K黳per  <[e-mail]>

	* rt_com.c: Changed back to the old static-field approach.
	Use rtl_sync functions instead of r_cli/r_sti.
	( This is copied from rt_com in the the khidaq-1_1 branch. )

1999-05-07  Jochen K黳per  <[e-mail]>

	* rt_com.c: Changed all functions to use the buffersize field, esp.
	rt_com_setup needs to allocate and free the buffer memory.

	* rt_com.h: Change rt_com_setup interface to take an additional parameter
	for the buffersize of the FISOs.

1999-03-30  Jochen K黳per  <[e-mail]>

	* rt_comP.h: Remove define R_COM_BUF_SIZ and add a size-field to
	rt_buf_struct.

1999-03-29  Jochen K黳per  <[e-mail]>

	* rt_com.c: Determine the number of rt_com_table entries automatically.
	(rt_com_read), (rt_com_write), (rt_com_irq_get), (rt_com_irq_put):
 	Wrap FIFO head/tail by explicit check for the maximum element instead
 	of the tricky binary and. So we can use any size now.
	(init_module): On failure really free all allocated resources, not just
	the last one several times.

1998-11-11  Jochen K黳per  <[e-mail]>

	* rt_com.h: Take out all unneeded stuff and put it into rt_comP.h.

	* rt_comP.h: Put module name into define RT_COM_NAME and use it
	wherever approbiate.

1998-10-09  Jochen K黳per  <[e-mail]>

	* rt_com.c (cleanup_module): Unregister ISRs before cleaning up the
	line settings.

	* general: Use priority descriptors with all printk statements.
	Document all functions using javadoc / kdoc style documents. For each
 	function a short description and the meaning of all parameter and the
 	return value is given.

1998-03-25  Jochen K黳per  <[e-mail]>

	* general: Started this ChangeLog file.


	
	* Local Variables:
	* mode: auto-fill
	* fill-column: 79
	* End: