From mboxrd@z Thu Jan  1 00:00:00 1970
X-Spam-Checker-Version: SpamAssassin 3.4.4 (2020-01-24) on polar.synack.me
X-Spam-Level: *
X-Spam-Status: No, score=1.3 required=5.0 tests=BAYES_00,INVALID_MSGID,
	MSGID_RANDY autolearn=no autolearn_force=no version=3.4.4
X-Google-Language: ENGLISH,ASCII-7-bit
X-Google-Thread: 103376,6551e8ab1fe1fab7
X-Google-Attributes: gid103376,public
From: David C. Hoos, Sr. <david.c.hoos.sr@ada95.com>
Subject: Re: how to access a (serial) device (linux)
Date: 2000/01/27
Message-ID: <86pat1$2pi$1@nnrp1.deja.com>#1/1
X-Deja-AN: 578242440
References: <86op66$lu9$1@nnrp1.deja.com>
To: tilmanglotzner@my-deja.com
X-Http-User-Agent: Mozilla/4.0 (compatible; MSIE 5.0; Windows NT; DigExt)
X-Http-Proxy: 1.1 x37.deja.com:80 (Squid/1.1.22) for client 205.149.60.17
Organization: Deja.com - Before you buy.
X-Article-Creation-Date: Thu Jan 27 11:41:21 2000 GMT
X-MyDeja-Info: XMYDJUIDdhoossr
Newsgroups: comp.lang.ada
Date: 2000-01-27T00:00:00+00:00
List-Id: <comp.lang.ada>

In article <86op66$lu9$1@nnrp1.deja.com>,
  tilmanglotzner@my-deja.com wrote:
> Hello,
>
> I wrote a program to access serial port via the device file. One
> task reads on the port, another task writes to the port. The
> tasks somehow seem to block each other, and an antenna controller
> connected to the serial port does not receive anything.
>
> 1) Is accessing the serial port via a device file the wrong approach ?
> Is it better to do this by an interrupt service routine ? A sample
> which I could reuse would be very, very helpful here :-)
>
> 2) Maybe the processes are in a dead lock because 2 processes try to
> access the same file. Is it feasible to access a device file from 2
> task  ?
>
> 3) Maybe it is better to access the device file from only 1 task.
> I would need to open the file as read/write. Text_io seems not be
> capable of this.Are there other packages which allow opening
> (device) file as read/write ?
I have implemented a network physical layer via a serial port in
Linux, using a custom PCMCI card modem.  This channel runs reliably
at 512000 bps, because the hardware supports that kind of rate.

The underlying implementation uses the POSIX.IO and
POSIX.Terminal_Characteristics packages from the FLORIST implementation
of the IEEE Ada95 POSIX bindings.

The /dev/ttySN device is opened with POSIX.IO.Open with Read_Write
mode.

Reading is done with a task that puts the incoming octets in a
protected queue from which the link layer reads using an Ada.Streams
interface.  Reading is done with POSIX.IO.Read.

Writing is done without a task directly from the Stream's Write
procedure by calling POSIX.IO.Write.

To build a robust serial port package requires attention to quite
a number of details. For example, my port Open procedure first obtains
a lock on the port, using the lock file feature.

The link layer has a task (which I call a frame server), that reads the
incoming stream, identifying frames, and calling a procedure that has
registered with the task to receive frames.  This procedure is called
passing a validated frame as an Ada.Streams.Stream_Element_Array.


Sent via Deja.com http://www.deja.com/
Before you buy.