[ECOS] UART communication

[ECOS] UART communication

[Andreas.Karlsson]

Hi,

While receiving several bytes on the UART0 port on my AEB-board the
eCos-serialdriver says it only received 1 byte with the value of 0.

The incoming data looks very nice on an oscilloscope. 

Just to be sure that I hadn't messed something up in my receiver-thread I
connected the TX and the RX pin and sent 4 bytes, the packet was received
just as it should.



I compiled the ser0 with a baud rate of 57600 since that's the only one I
can use. The other settings are 1 stop bit, no parity, wordsize=8 and
CTS/RTS enabled.

I noticed that the baudrate of 57600 wasn't supported in aeb_serial.h but
what happens when I make this selection in the configuration?

To be able to enable CTS/RTS I had to modify the select_baud-array in
aeb_serial.h. This because I should be able to set all fields in the
cyg_serial_info_t struct. I added the value of 26 on the bauddivisor for
57600, maybe this is the problem.

I really don't know what could be wrong so if anybody has a clue...

/Andreas

Re: [ECOS] UART communication

[Dave Airlie]

With eCos on the ARM AEB you cannot choose 57600, I up came against
something similiar a while back.. check the return from
cyg_io_set_config I found it by tracing the code ...

I think you can use a different clock multiplier to get 57600 but it
involves changing eCos code...

Dave.


On Tue, 5 Sep 2000 Andreas.Karlsson@combitechsystems.com wrote:

> Hi,
> 
> While receiving several bytes on the UART0 port on my AEB-board the
> eCos-serialdriver says it only received 1 byte with the value of 0.
> 
> The incoming data looks very nice on an oscilloscope. 
> 
> Just to be sure that I hadn't messed something up in my receiver-thread I
> connected the TX and the RX pin and sent 4 bytes, the packet was received
> just as it should.
> 
> 
> 
> I compiled the ser0 with a baud rate of 57600 since that's the only one I
> can use. The other settings are 1 stop bit, no parity, wordsize=8 and
> CTS/RTS enabled.
> 
> I noticed that the baudrate of 57600 wasn't supported in aeb_serial.h but
> what happens when I make this selection in the configuration?
> 
> To be able to enable CTS/RTS I had to modify the select_baud-array in
> aeb_serial.h. This because I should be able to set all fields in the
> cyg_serial_info_t struct. I added the value of 26 on the bauddivisor for
> 57600, maybe this is the problem.
> 
> I really don't know what could be wrong so if anybody has a clue...
> 
> /Andreas
> 

-- 
      David Airlie, Software Engineer, Parthus Technologies plc.,
       Mary Rosse Centre, National Tech Park, Limerick, Ireland.
   t: +353-61-508116 / f: +353-61-508101 / David.Airlie@parthus.com

Re: [ECOS] UART communication

[Gary Thomas]

You might try the [untested] attached patch.  No guarantees, but it
will give you an idea of where to look and what to try.

Of course if you get it working, we'd appreciate any feedback/patches.

On 05-Sep-2000 Dave Airlie wrote:
> 
> With eCos on the ARM AEB you cannot choose 57600, I up came against
> something similiar a while back.. check the return from
> cyg_io_set_config I found it by tracing the code ...
> 
> I think you can use a different clock multiplier to get 57600 but it
> involves changing eCos code...
> 
> Dave.
> 
> 
> On Tue, 5 Sep 2000 Andreas.Karlsson@combitechsystems.com wrote:
> 
>> Hi,
>> 
>> While receiving several bytes on the UART0 port on my AEB-board the
>> eCos-serialdriver says it only received 1 byte with the value of 0.
>> 
>> The incoming data looks very nice on an oscilloscope. 
>> 
>> Just to be sure that I hadn't messed something up in my receiver-thread I
>> connected the TX and the RX pin and sent 4 bytes, the packet was received
>> just as it should.
>> 
>> 
>> 
>> I compiled the ser0 with a baud rate of 57600 since that's the only one I
>> can use. The other settings are 1 stop bit, no parity, wordsize=8 and
>> CTS/RTS enabled.
>> 
>> I noticed that the baudrate of 57600 wasn't supported in aeb_serial.h but
>> what happens when I make this selection in the configuration?
>> 
>> To be able to enable CTS/RTS I had to modify the select_baud-array in
>> aeb_serial.h. This because I should be able to set all fields in the
>> cyg_serial_info_t struct. I added the value of 26 on the bauddivisor for
>> 57600, maybe this is the problem.
>> 
>> I really don't know what could be wrong so if anybody has a clue...
>> 
>> /Andreas
>> 
> 
> -- 
>       David Airlie, Software Engineer, Parthus Technologies plc.,
>        Mary Rosse Centre, National Tech Park, Limerick, Ireland.
>    t: +353-61-508116 / f: +353-61-508101 / David.Airlie@parthus.com
> 

============= Patch to support 57600 and 115200 baud on AEB-1 ================
Index: devs/serial/arm/aeb/current/src/aeb_serial.h
===================================================================
RCS file: /local/cvsfiles/ecc/ecc/devs/serial/arm/aeb/current/src/aeb_serial.h,v
retrieving revision 1.1
diff -u -5 -p -r1.1 aeb_serial.h
--- devs/serial/arm/aeb/current/src/aeb_serial.h        2000/04/10 14:44:46     1.1
+++ devs/serial/arm/aeb/current/src/aeb_serial.h        2000/09/05 15:59:29
@@ -144,11 +144,11 @@ static unsigned short select_baud[] = {
         69*3,  // 7200
         52*3,  // 9600
     (69*3)/2,  // 14400
         26*3,  // 19200
         13*3,  // 38400
-           0,  // 57600
-           0,  // 115200
+          26,  // 57600
+          13,  // 115200
            0,  // 230400
 };
 
 #endif // CYGONCE_ARM_AEB_SERIAL_H

Re: [ECOS] UART communication

[Jonathan Larmour]

Andreas.Karlsson@combitechsystems.com wrote:
> 
> While receiving several bytes on the UART0 port on my AEB-board the
> eCos-serialdriver says it only received 1 byte with the value of 0.
> 
> The incoming data looks very nice on an oscilloscope.
> 
> Just to be sure that I hadn't messed something up in my receiver-thread I
> connected the TX and the RX pin and sent 4 bytes, the packet was received
> just as it should.
> 
> I compiled the ser0 with a baud rate of 57600 since that's the only one I
> can use. The other settings are 1 stop bit, no parity, wordsize=8 and
> CTS/RTS enabled.

Can you clarify what you mean by CTS/RTS enabled? How did you enable this?
 
> I noticed that the baudrate of 57600 wasn't supported in aeb_serial.h but
> what happens when I make this selection in the configuration?
> 
> To be able to enable CTS/RTS I had to modify the select_baud-array in
> aeb_serial.h. This because I should be able to set all fields in the
> cyg_serial_info_t struct. I added the value of 26 on the bauddivisor for
> 57600, maybe this is the problem.

Does 57600 bps communication work normally?

Jifl
-- 
Red Hat, 35 Cambridge Place, Cambridge, UK. CB2 1NS  Tel: +44 (1223) 728762
"Plan to be spontaneous tomorrow."  ||  These opinions are all my own fault

Re: [ECOS] UART communication

[Jesper Skov]

>>>>> "Jonathan" == Jonathan Larmour <jlarmour@redhat.com> writes:

Jonathan> Andreas.Karlsson@combitechsystems.com wrote:
>> > Can you clarify what you mean by CTS/RTS enabled? How did you >
>> enable this?
>> 
>> I enabled it with the cyg_io_set_config-function like this:
>> 
Jonathan> [snip]
>> settings_buf.flags=CYG_SERIAL_FLAGS_RTSCTS; //enable CTS/RTS

Jonathan> I think you should find that setting this has no effect,
Jonathan> which is why you are seeing overruns. If you update to eCos
Jonathan> from CVS, you will find that "real" support for hardware
Jonathan> flow control has been added to the generic layer, although
Jonathan> the only device driver that uses it currently is the ARM PID
Jonathan> serial driver.

Jonathan> However, it would not be difficult to make the same changes
Jonathan> to the AEB serial driver. Perhaps you should consider that?

I'm plotting on restructuring the 16450/1655x serial drivers just as I
have just done with the SH SCI/SCIF serial drivers.

I haven't looked into all the details yet, but I believe I should be
able to make a generic 16450/1655x serial driver which is shared
between all the platforms that presently have a dedicated driver for
such a beast. This would affect at least PID, Cogent and AEB -
possibly other platforms as well.

I would base that reorg on the PID driver, thus extending the current
PID BREAK+CTS/RTS support to the other affected platforms.

By the looks of things, however, I wouldn't get around to that for
some weeks, so changing the AEB driver would probably be a good (and
necessary) short term fix.

Jesper

Re: [ECOS] UART communication

[Jonathan Larmour]

Andreas.Karlsson@combitechsystems.com wrote:
> 
> I've been trying today to connect to a terminal on a Laptop just to check
> that the 57600kbps is working and it works very well. But maybe you are
> right that CTS/RTS doesn't work. I hope I don't have to rebuild any of the
> tools just eCos, right?

You shouldn't need a new toolchain, although you may need new versions of
ecosconfig/eCos configuration tool if you want to switch to the eCos CVS
sources (see http://sources.redhat.com/ecos/anoncvs.html )

Jifl
-- 
Red Hat, 35 Cambridge Place, Cambridge, UK. CB2 1NS  Tel: +44 (1223) 728762
"Plan to be spontaneous tomorrow."  ||  These opinions are all my own fault

RE: [ECOS] UART communication

[Andreas.Karlsson]

Ok,

I've been trying today to connect to a terminal on a Laptop just to check
that the 57600kbps is working and it works very well. But maybe you are
right that CTS/RTS doesn't work. I hope I don't have to rebuild any of the
tools just eCos, right?

/Andreas

Re: [ECOS] UART communication

[Jonathan Larmour]

Andreas.Karlsson@combitechsystems.com wrote:
> > Can you clarify what you mean by CTS/RTS enabled? How did you
> > enable this?
> 
> I enabled it with the cyg_io_set_config-function like this:
> 
[snip] 
> settings_buf.flags=CYG_SERIAL_FLAGS_RTSCTS; //enable CTS/RTS

I think you should find that setting this has no effect, which is why you
are seeing overruns. If you update to eCos from CVS, you will find that
"real" support for hardware flow control has been added to the generic
layer, although the only device driver that uses it currently is the ARM
PID serial driver.

However, it would not be difficult to make the same changes to the AEB
serial driver. Perhaps you should consider that?

Or if you update to CVS, you could try the software (XON/XOFF) flow control
which should "just work".

Jifl
-- 
Red Hat, 35 Cambridge Place, Cambridge, UK. CB2 1NS  Tel: +44 (1223) 728762
"Plan to be spontaneous tomorrow."  ||  These opinions are all my own fault

RE: [ECOS] UART communication

[Andreas.Karlsson]

> 

> Can you clarify what you mean by CTS/RTS enabled? How did you 

> enable this?

> Jifl



I enabled it with the cyg_io_set_config-function like this:



cyg_serial_info_t settings_buf;



settings_buf.baud=CYGNUM_SERIAL_BAUD_57600; //had to modify the sources in
aeb_serial.h

settings_buf.stop=CYGNUM_SERIAL_STOP_1;

settings_buf.parity=CYGNUM_SERIAL_PARITY_NONE;

settings_buf.word_length=CYGNUM_SERIAL_WORD_LENGTH_8;

settings_buf.flags=CYG_SERIAL_FLAGS_RTSCTS; //enable CTS/RTS



err = cyg_io_lookup( "/dev/ser0", &ser0_hdl );

if(ENOERR != err)

{

	printf("Couldn't find /dev/ser0 !!\n");

}

len=sizeof(cyg_serial_info_t);

err=cyg_io_set_config(ser0_hdl, CYG_IO_SET_CONFIG_SERIAL_INFO,
&settings_buf, &len);



When checking the Line Status Register after receiving the zero-byte there
is an overrun error. 

The value of LSR is 0x62.

/Andreas