Re: Serial programming patch

Re: Serial programming patch

[J. J. Farrell]

From: Jacek Trzcinski <jacek@certum.pl>
>
> Here is my patch to serial device. I implemented ioctl function which
> makes possible
> to get modem lines state (CTS,DSR,RI,CD) and set(get) modem control
> lines (RTS and DTR)
>
> ...
>
> +
> +/* ioctl: POSIX  */
> +int
> +fhandler_serial::ioctl (unsigned int cmd,void * buffer)

Just a very minor nit-pick: it might reduce future confusion if the word
"POSIX" were removed from that comment. POSIX doesn't specify an ioctl()
function or any means to access the modem control signals on serial lines.




--
Want to unsubscribe from this list?
Check out: http://cygwin.com/ml/#unsubscribe-simple

Serial programming patch

[briseida deyanira maytorena sanchez]

Hello,

I've been searching through the mailing lists trying to find a painless
way of toggling DTR/RTS lines for some test code which I am writing. I was
trying to use ioctl() but TIOCMSET, TIOCM_DTR and all the other
definitions needed are currently unsupported. 

I also found in the archive a patch written by Jacek Trzcinski 
(see http://sources.redhat.com/ml/cygwin/2001-05/msg01163.html) and
wondered what the current status of this is, i.e. is it still under
development?  If the patch has been abandoned what can one do to help, how
would I go about testing it? Not knowing anything about tweaking cygwin
packages could you point me to appropriate documentation. Forgive my
ignorance but all I'm doing at the moment is a bit of C so looking at the
Trzcinski's posting of the patch is a bit like reading Arabic.

All I really want is to be able to toggle those lines so if there is an
easier way that you know of, sharing your knowledge would be much
appreciated.

Thanks in advance,

Deya



--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

Re: Serial programming patch

[Christopher Faylor]

On Thu, Apr 11, 2002 at 05:46:39PM +0100, briseida deyanira maytorena sanchez wrote:
>Hello,
>
>I've been searching through the mailing lists trying to find a painless
>way of toggling DTR/RTS lines for some test code which I am writing. I was
>trying to use ioctl() but TIOCMSET, TIOCM_DTR and all the other
>definitions needed are currently unsupported. 
>
>I also found in the archive a patch written by Jacek Trzcinski 
>(see http://sources.redhat.com/ml/cygwin/2001-05/msg01163.html) and
>wondered what the current status of this is, i.e. is it still under
>development?

As far as I know, we never received an assignment for the patch so it
was never applied.

>If the patch has been abandoned what can one do to help, how
>would I go about testing it?

No idea.  I guess you'd have to contact the original author.

cgf

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

Serial programming patch

[Jacek Trzcinski]

Hi !,
Here is my patch to serial device. I implemented ioctl function which
makes possible
to get modem lines state (CTS,DSR,RI,CD) and set(get) modem control
lines (RTS and DTR)
It does not get(set) other signals from MCR register(there is no need).
It will be usefull for people (like me) who have to connect devices
requiring mentioned signals f.e. chip card readers.
I made changes in fhandler_serial.cc, fhandler.h and in termios.h where
I placed
proper constants taken directly from linux headers. I tested it in NT
and W2K environment
where there is nice way to get RTS and DTR state by DeviceIoControl
function. In
Windows 9x it is no possible and so far I do not know how to do it
without big gimnastics.
In consequence for 9x I keep mirror of this signals set by functions
which influence on state
of this signals (open, ioctl,tcsetattr) and in my ioctl simply return
it.
This patch was created on base of cygwin sources version 1.3.1-1
I have to send it as attachment because my post client is netscape and
inserted patch looks
horrible
Jacek
--- fhandler_serial.old.cc	Tue Apr 24 03:19:15 2001
+++ fhandler_serial.cc	Tue May 22 11:01:29 2001
@@ -32,6 +32,9 @@ fhandler_serial::fhandler_serial (const 
   vtime_ = 0;
   pgrp_ = myself->pgid;
   set_need_fork_fixup ();
+  osVersion=GetVersion();
+  dtr=0;
+  rts=0;
 }
 
 void
@@ -275,7 +278,8 @@ fhandler_serial::open (const char *name,
       if (!SetCommState (get_handle (), &state))
 	system_printf ("couldn't set initial state for %s, %E", get_name ());
     }
-
+  rts=TIOCM_RTS;
+  dtr=TIOCM_DTR;
   SetCommMask (get_handle (), EV_RXCHAR);
   set_open_status ();
   syscall_printf ("%p = fhandler_serial::open (%s, %p, %p)",
@@ -366,6 +370,93 @@ fhandler_serial::tcflow (int action)
   return 0;
 }
 
+
+/* ioctl: POSIX  */
+int
+fhandler_serial::ioctl (unsigned int cmd,void * buffer)
+{
+ 
+  DCB dcb;
+  DWORD ev;
+  COMSTAT st;
+  DWORD action;
+  DWORD modemLines;
+  DWORD mcr;
+  DWORD cbReturned;
+  bool  result;
+  int   modemStatus;
+  int request;
+
+  request=*(int*)buffer;
+  action=0;
+  modemStatus=0;
+  if (!ClearCommError (get_handle (), &ev, &st)) return -1;
+  switch (cmd)
+    {
+      case TIOCMGET:
+        if(GetCommModemStatus(get_handle(),&modemLines)==0) return -1;
+	if(modemLines&MS_CTS_ON) modemStatus|=TIOCM_CTS;
+	if(modemLines&MS_DSR_ON) modemStatus|=TIOCM_DSR;
+	if(modemLines&MS_RING_ON) modemStatus|=TIOCM_RNG | TIOCM_RI;
+	if(modemLines&MS_RLSD_ON) modemStatus|=TIOCM_CAR | TIOCM_CD;
+	if(!(osVersion&0x80000000))
+	{
+	// here is Windows NT or Windows 2000
+	  result=DeviceIoControl(
+	                          get_handle(),
+				  0x001B0078,
+				  NULL,
+				  0,
+				  &mcr,
+				  4,
+				  &cbReturned,
+				  0
+				  );
+	  if(!result) return -1;
+	  if(cbReturned!=4) return -1;
+	  if(mcr&2) modemStatus|=TIOCM_RTS;
+	  if(mcr&1) modemStatus|=TIOCM_DTR;
+	  
+	} else
+	{
+	
+	// here is Windows 9x
+	  modemStatus|=rts|dtr;
+	  
+	}
+	*(int*)buffer=modemStatus;
+	return 0;
+      case TIOCMSET:
+        //if(request & ~(TIOCM_RTS | TIOCM_DTR)) return -1;
+	if (request&TIOCM_RTS)
+	{
+	    if(EscapeCommFunction(get_handle(),SETRTS)==0) return -1; else
+	    rts=TIOCM_RTS;
+	} else
+	{
+	    if(EscapeCommFunction(get_handle(),CLRRTS)==0) return -1;else
+	    rts=0;
+	}
+	if (request&TIOCM_DTR)
+	{
+	    if(EscapeCommFunction(get_handle(),SETDTR)==0) return -1;else
+	    dtr=TIOCM_DTR;
+	} else
+	{
+	    if(EscapeCommFunction(get_handle(),CLRDTR)==0) return -1;else
+	    dtr=0;
+	}
+	return 0;
+      case TIOCINQ:
+        if(ev&CE_FRAME | ev&CE_IOE | ev&CE_OVERRUN | \
+	ev&CE_RXOVER | ev&CE_RXPARITY) return -1;
+	*(int*)buffer=st.cbInQue;
+	return 0;
+      default:
+	return -1;
+    }
+}
+
 /* tcflush: POSIX 7.2.2.1 */
 int
 fhandler_serial::tcflush (int queue)
@@ -373,7 +464,7 @@ fhandler_serial::tcflush (int queue)
   if (queue == TCOFLUSH || queue == TCIOFLUSH)
     PurgeComm (get_handle (), PURGE_TXABORT | PURGE_TXCLEAR);
 
-  if (queue == TCIFLUSH | queue == TCIOFLUSH)
+  if (queue == TCIFLUSH || queue == TCIOFLUSH)
     /* Input flushing by polling until nothing turns up
        (we stop after 1000 chars anyway) */
     for (int max = 1000; max > 0; max--)
@@ -404,6 +495,8 @@ fhandler_serial::tcsetattr (int action, 
   COMMTIMEOUTS to;
   DCB ostate, state;
   unsigned int ovtime = vtime_, ovmin = vmin_;
+  int tmpDtr,tmpRts;
+  tmpDtr=tmpRts=0;
 
   termios_printf ("action %d", action);
   if ((action == TCSADRAIN) || (action == TCSAFLUSH))
@@ -569,6 +662,7 @@ fhandler_serial::tcsetattr (int action, 
     {							/* disable */
       state.fRtsControl = RTS_CONTROL_ENABLE;
       state.fOutxCtsFlow = FALSE;
+      tmpRts=TIOCM_RTS;
     }
 
   if (t->c_cflag & CRTSXOFF)
@@ -601,7 +695,10 @@ fhandler_serial::tcsetattr (int action, 
   set_w_binary ((t->c_oflag & ONLCR) ? 0 : 1);
 
   if (dropDTR == TRUE)
+  {
     EscapeCommFunction (get_handle (), CLRDTR);
+    tmpDtr=0;
+  }
   else
     {
       /* FIXME: Sometimes when CLRDTR is set, setting
@@ -610,7 +707,11 @@ fhandler_serial::tcsetattr (int action, 
       parameters while DTR is still down. */
 
       EscapeCommFunction (get_handle (), SETDTR);
+      tmpDtr=TIOCM_DTR;
     }
+    
+  rts=tmpRts;
+  dtr=tmpDtr;
 
   /*
   The following documentation on was taken from "Linux Serial Programming
--- fhandler.old.h	Tue Apr 24 03:19:14 2001
+++ fhandler.h	Tue May 22 10:34:58 2001
@@ -529,6 +529,9 @@ private:
   unsigned int vmin_;			/* from termios */
   unsigned int vtime_;			/* from termios */
   pid_t pgrp_;
+  unsigned int osVersion;
+  int	rts;
+  int	dtr;
 
 public:
   int overlapped_armed;
@@ -547,6 +550,7 @@ public:
   int tcsendbreak (int);
   int tcdrain ();
   int tcflow (int);
+  int ioctl (unsigned int cmd, void *);
   int tcsetattr (int a, const struct termios *t);
   int tcgetattr (struct termios *t);
   off_t lseek (off_t, int) { return 0; }
--- ./include/sys/termios.old.h	Sun Mar 25 21:09:52 2001
+++ ./include/sys/termios.h	Tue May 22 12:03:16 2001
@@ -13,6 +13,22 @@ details. */
 #ifndef	_SYS_TERMIOS_H
 #define _SYS_TERMIOS_H
 
+
+#define	TIOCMGET	0x5415
+#define TIOCMSET	0x5418
+#define TIOCINQ		0x541B
+
+#define	TIOCM_LE	0x001
+#define	TIOCM_DTR	0x002
+#define	TIOCM_RTS	0x004
+#define	TIOCM_CTS	0x020
+#define	TIOCM_CAR	0x040
+#define	TIOCM_RNG	0x080
+#define	TIOCM_DSR	0x100
+#define	TIOCM_CD	TIOCM_CAR
+#define	TIOCM_RI	TIOCM_RNG
+
+
 #define TCOOFF		0
 #define TCOON		1
 #define TCIOFF		2

Re: Serial programming patch

[Corinna Vinschen]

Thanks for your work. Unfortunately, we can't use it for three
reasons:

- The patch is too long to use it without you signing a copyright
  assignment form. Look on http://cygwin.com/contrib.html for details.

- A ChangeLog entry is missing. Please add a plain ChangeLog entry to
  your posting like the ones you can see in the Cygwin ChangeLog file.

- Your code is not conforming to the GNU coding standard.
  Look on http://www.gnu.org/prep/standards_toc.html for details
  (or just compare with the other code in Cygwin).

Then you're adding a member to fhandler_serial to check for the OS
that way:

	osVersion=GetVersion();
	[...]
	if(!(osVersion&0x80000000))

while Cygwin already has a way to check that without the need to
add a member:

	if (os_being_run == winNT)  // or win95, win98, winME

Btw, there's a special mailing list cygwin-patches@cygwin.com to send
and discuss patches to Cygwin. If you want to contribute to the project
it would have some advantages to subscribe to that list and to the list
cygwin-developers@cygwin.com as well.

I hope I didn't discourage you too much. Each contributor (and Chris and
I began as contributors, too) had to go through this.

Corinna

On Tue, May 22, 2001 at 02:27:52PM +0200, Jacek Trzcinski wrote:
> Hi !,
> Here is my patch to serial device. I implemented ioctl function which
> makes possible
> [...]

-- 
Corinna Vinschen                  Please, send mails regarding Cygwin to
Cygwin Developer                                mailto:cygwin@cygwin.com
Red Hat, Inc.

--
Want to unsubscribe from this list?
Check out: http://cygwin.com/ml/#unsubscribe-simple

Re: Serial programming patch

[Jacek Trzcinski]

Hi Corinna,
I see it will be cost a little pain.
What concerns Your 3-rd remark (code not conforming GNU standard) is it
enough to
transform my sources utilizing indent.exe ?. If so, what switches should
I use ?
I have two remarks I forgot mention about in my previous e-mail.

To read input queue there should be use constant FIONREAD (information
taken from different sources) but cygwin occupy it for socket software
!(it is macro). Constant TIOCINQ I have
taken from linux where it is equal FIONREAD(FIONREAD is not occupied in
linux like in cygwin).
Next matter concerns device name convention. In cygwin one may use
/dev/com1 or /dev/ttyS1
and so on. (what for is utilizing /dev/comx I do not know - it is not
portable to linux).
I think also that number of devices should start from 0 not from 1
(/dev/ttyS0 and so on) like in linux because again it is not portable.
What is Your opinion ?
Jacek


Corinna Vinschen wrote:
> 
> Thanks for your work. Unfortunately, we can't use it for three
> reasons:
> 
> - The patch is too long to use it without you signing a copyright
>   assignment form. Look on http://cygwin.com/contrib.html for details.
> 
> - A ChangeLog entry is missing. Please add a plain ChangeLog entry to
>   your posting like the ones you can see in the Cygwin ChangeLog file.
> 
> - Your code is not conforming to the GNU coding standard.
>   Look on http://www.gnu.org/prep/standards_toc.html for details
>   (or just compare with the other code in Cygwin).
> 
> Then you're adding a member to fhandler_serial to check for the OS
> that way:
> 
>         osVersion=GetVersion();
>         [...]
>         if(!(osVersion&0x80000000))
> 
> while Cygwin already has a way to check that without the need to
> add a member:
> 
>         if (os_being_run == winNT)  // or win95, win98, winME
> 
> Btw, there's a special mailing list cygwin-patches@cygwin.com to send
> and discuss patches to Cygwin. If you want to contribute to the project
> it would have some advantages to subscribe to that list and to the list
> cygwin-developers@cygwin.com as well.
> 
> I hope I didn't discourage you too much. Each contributor (and Chris and
> I began as contributors, too) had to go through this.
> 
> Corinna
> 
> On Tue, May 22, 2001 at 02:27:52PM +0200, Jacek Trzcinski wrote:
> > Hi !,
> > Here is my patch to serial device. I implemented ioctl function which
> > makes possible
> > [...]
> 
> --
> Corinna Vinschen                  Please, send mails regarding Cygwin to
> Cygwin Developer                                mailto:cygwin@cygwin.com
> Red Hat, Inc.
> 
> --
> Want to unsubscribe from this list?
> Check out: http://cygwin.com/ml/#unsubscribe-simple

--
Want to unsubscribe from this list?
Check out: http://cygwin.com/ml/#unsubscribe-simple

Re: Serial programming patch

[Corinna Vinschen]

On Wed, May 23, 2001 at 09:46:59AM +0200, Jacek Trzcinski wrote:
> Hi Corinna,
> I see it will be cost a little pain.
> What concerns Your 3-rd remark (code not conforming GNU standard) is it
> enough to
> transform my sources utilizing indent.exe ?. If so, what switches should
> I use ?

GNU indent does what it should do without switches. The GNU coding
standard is the default. However, I suggest to change your patched
sources by hand since there are two problems with indent. On one
hand it's sometimes not that clever when indenting C++ sources and
- most important - if you indent all concerned files they will be
reformated everywhere and your patch will include parts which you
never touched.

Corinna

> I have two remarks I forgot mention about in my previous e-mail.
> 
> To read input queue there should be use constant FIONREAD (information
> taken from different sources) but cygwin occupy it for socket software
> !(it is macro). Constant TIOCINQ I have
> taken from linux where it is equal FIONREAD(FIONREAD is not occupied in
> linux like in cygwin).

I see. You should be able to use it anyway by including asm/socket.h.
The macro expands to a constant expression. I understand that including
asm/socket.h seems to make not much sense. Probably we should create
a asm/ioctls.h file like in Linux.

Anyway, may I suggest to discuss that in cygwin-patches@cygwin.com?

> Next matter concerns device name convention. In cygwin one may use
> /dev/com1 or /dev/ttyS1
> and so on. (what for is utilizing /dev/comx I do not know - it is not
> portable to linux).

Good ol' COM...

> I think also that number of devices should start from 0 not from 1
> (/dev/ttyS0 and so on) like in linux because again it is not portable.
> What is Your opinion ?

As above: Let's discuss that in cygwin-patches@cygwin.com.

Corinna

-- 
Corinna Vinschen                  Please, send mails regarding Cygwin to
Cygwin Developer                                mailto:cygwin@cygwin.com
Red Hat, Inc.

--
Want to unsubscribe from this list?
Check out: http://cygwin.com/ml/#unsubscribe-simple

Re: Serial programming patch

[Jacek Trzcinski]

Hi Corrina,
I would not like bother You too much so I hope it is my last e-mail
today.
I created sources (with Your little help - thanks, I utilized
os_being_run) at last
are conforming to GNU coding. I also created my ChangeLog file.
What concerns signing a copyright assignment form - is it really
necessary ?
As I have read I have to send letter via normal mail, so I guess a lot
of time
may pass by I will get any response. Am I understand that until I will
not get
response from Red Hat I shouldn't send my patch ?
Jacek
P.S.
I have just subscribed to cygwin-patches list so I will send our
conversation concerning
ioctl codes and /dev... there.



Corinna Vinschen wrote:
> 
> On Wed, May 23, 2001 at 09:46:59AM +0200, Jacek Trzcinski wrote:
> > Hi Corinna,
> > I see it will be cost a little pain.
> > What concerns Your 3-rd remark (code not conforming GNU standard) is it
> > enough to
> > transform my sources utilizing indent.exe ?. If so, what switches should
> > I use ?
> 
> GNU indent does what it should do without switches. The GNU coding
> standard is the default. However, I suggest to change your patched
> sources by hand since there are two problems with indent. On one
> hand it's sometimes not that clever when indenting C++ sources and
> - most important - if you indent all concerned files they will be
> reformated everywhere and your patch will include parts which you
> never touched.
> 
> Corinna
> 
> > I have two remarks I forgot mention about in my previous e-mail.
> >
> > To read input queue there should be use constant FIONREAD (information
> > taken from different sources) but cygwin occupy it for socket software
> > !(it is macro). Constant TIOCINQ I have
> > taken from linux where it is equal FIONREAD(FIONREAD is not occupied in
> > linux like in cygwin).
> 
> I see. You should be able to use it anyway by including asm/socket.h.
> The macro expands to a constant expression. I understand that including
> asm/socket.h seems to make not much sense. Probably we should create
> a asm/ioctls.h file like in Linux.
> 
> Anyway, may I suggest to discuss that in cygwin-patches@cygwin.com?
> 
> > Next matter concerns device name convention. In cygwin one may use
> > /dev/com1 or /dev/ttyS1
> > and so on. (what for is utilizing /dev/comx I do not know - it is not
> > portable to linux).
> 
> Good ol' COM...
> 
> > I think also that number of devices should start from 0 not from 1
> > (/dev/ttyS0 and so on) like in linux because again it is not portable.
> > What is Your opinion ?
> 
> As above: Let's discuss that in cygwin-patches@cygwin.com.
> 
> Corinna
> 
> --
> Corinna Vinschen                  Please, send mails regarding Cygwin to
> Cygwin Developer                                mailto:cygwin@cygwin.com
> Red Hat, Inc.
> 
> --
> Want to unsubscribe from this list?
> Check out: http://cygwin.com/ml/#unsubscribe-simple

--
Want to unsubscribe from this list?
Check out: http://cygwin.com/ml/#unsubscribe-simple

Re: Serial programming patch

[Corinna Vinschen]

On Wed, May 23, 2001 at 02:15:26PM +0200, Jacek Trzcinski wrote:
> Hi Corrina,
> I would not like bother You too much so I hope it is my last e-mail
> today.
> I created sources (with Your little help - thanks, I utilized
> os_being_run) at last
> are conforming to GNU coding. I also created my ChangeLog file.
> What concerns signing a copyright assignment form - is it really
> necessary ?

Unfortunately, yes. I can't explain that really good. It's one of
the weirdnesses of the US copyright law. Even if it's not applicable
to the situation of us non-US folks, Red Hat as a US company has to
care for that to have no trouble. If your patch would just change two
or three lines of the code, it wouldn't be needed but your patch has
a not negligible size.

> As I have read I have to send letter via normal mail, so I guess a lot
> of time
> may pass by I will get any response. Am I understand that until I will
> not get
> response from Red Hat I shouldn't send my patch ?

You can send the patch but we have to wait until your copyright
assignment arrives.

> Jacek
> P.S.
> I have just subscribed to cygwin-patches list so I will send our
> conversation concerning
> ioctl codes and /dev... there.

Thanks,
Corinna

-- 
Corinna Vinschen                  Please, send mails regarding Cygwin to
Cygwin Developer                                mailto:cygwin@cygwin.com
Red Hat, Inc.

--
Want to unsubscribe from this list?
Check out: http://cygwin.com/ml/#unsubscribe-simple

Re: Serial programming patch

[Christopher Faylor]

On Wed, May 23, 2001 at 04:29:29PM +0200, Corinna Vinschen wrote:
>On Wed, May 23, 2001 at 02:15:26PM +0200, Jacek Trzcinski wrote:
>> Hi Corrina,
>> I would not like bother You too much so I hope it is my last e-mail
>> today.
>> I created sources (with Your little help - thanks, I utilized
>> os_being_run) at last
>> are conforming to GNU coding. I also created my ChangeLog file.
>> What concerns signing a copyright assignment form - is it really
>> necessary ?
>
>Unfortunately, yes. I can't explain that really good. It's one of
>the weirdnesses of the US copyright law.

Actually, it's really simple.  Red Hat needs to own all of the work
in Cygwin.  So, any contributions to cygwin have to be assigned to us.

This is similar to the requirement that any work done to gcc or gdb
need to be assigned to the FSF.

cgf

--
Want to unsubscribe from this list?
Check out: http://cygwin.com/ml/#unsubscribe-simple